Food Chemistry 169 (2015) 305–313

Contents lists available at ScienceDirect

Food Chemistry journal homepage: www.elsevier.com/locate/foodchem

Analytical Methods

A multiplex real-time PCR method for the quantification of beef and pork fractions in minced meat A. Iwobi ⇑, D. Sebah, I. Kraemer, C. Losher, G. Fischer, U. Busch, I. Huber Bavarian Health and Food Safety Authority, Veterinaerstrasse 2, 85764, Germany

a r t i c l e

i n f o

Article history: Received 24 December 2013 Received in revised form 1 July 2014 Accepted 30 July 2014 Available online 8 August 2014 Keywords: Real-time PCR Species identification and quantification Beef Pork Validation

a b s t r a c t One popular staple food in many lands is minced meat, traditionally prepared from beef and/or pork fractions. While beef is the more expensive of the two meat fractions, the possibility exists for manufacturers to fraudulently declare higher proportions of it. Additionally, the need exists to protect consumers who, out of medical or ethical reasons, reject specific meat fractions. In this work, we report on a quantitative triplex real-time PCR approach for the quantification of meat in minced meat products. With the method, beef and pork fractions are quantified employing primer and probe sequences that specifically recognise cow and pig components, against the backdrop of myostatin, a universal sequence commonly found in mammals and poultry species. The limit of detection of the qPCR method was 20 genome equivalents, while the measurement of uncertainty was determined at 1.83%. The method was validated on several commercially available minced meat products and performed well in terms of handling, reproducibility and robustness. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction An integral part of the duties of a food control agency is the routine surveillance of food products, including meat-based edibles to ensure that their actual composition correlate with declared components. Meat products comprise a significant proportion of the protein intake of millions worldwide, with global consumption of meat rising steadily. While demand for beef was at its peak in the early 60s’, accounting for 40% of the global meat consumption, its dominance has declined steadily, with consumption falling to 23% in 2007. Pork accounts for the most commonly consumed meat fractions today, partly because of its relative cheapness, abundance, and lower production costs (The Economist Online, 2012). Authentic declaration of meat products may be particularly important to several members of the community, for example individuals who as a result of religious persuasions or health reasons, reject certain types of animal fractions (Ali, Hashim, Sabar Dhahi, Mustafa, & Bin Che Man, 2012). Additionally substitution of more expensive meat with cheaper derivatives might violate consumer trust and confidence. The foregoing emphasizes the importance of the implementation of reliable analytical methods by the relevant regulatory bodies for the determination of the exact ⇑ Corresponding author. Tel.: +49 9131 6808 5158; fax: +49 9131 6808 5458. E-mail address: [email protected] (A. Iwobi). http://dx.doi.org/10.1016/j.foodchem.2014.07.139 0308-8146/Ó 2014 Elsevier Ltd. All rights reserved.

composition of meat products. Recent scandals like the horse meat scandal that spread across Europe in early 2013, show the importance of analytical tools not only for detection of the meat constellation in a particular product, but also for quantitative determination of the individual components. This is important in distinguishing inadvertent contamination from deliberate adulteration of meat products, with accompanying legal consequences. PCR-based methods, from singleplex reactions to multiplex systems (mostly real-time PCR assays) have increasingly become relevant in the analysis of food products including meat samples (Girish, Haunshi, Vaithiyanathan, Rajitha, & Ramakrishna, 2013; Mane, Mendiratta, & Tiwari, 2012; Köppel, Eugster, Ruf, & Rentsch, 2012 and Köppel, Daniels, Felderer, & Brünen-Nieweler, 2013). Multiplex PCR reactions offer the distinct advantages of lower costs and expenditures, coupled with a time-saving feature. Such methods however, typically quantify the DNA of the animal species present in the meat product (López-Andreo, Aldeguer, Guillén, Gabaldón, & Puyet, 2012; Eugster, Ruf, Rentsch, & Köppel, 2009; Drummond et al., 2013). While such results are useful, a direct correlation between DNA content and actual meat percentages is more desirable and this may not always be possible considering the complexity of tissues utilised for meat preparations, with accompanying variations in the extractable DNA. For reliable quantification of actual meat contents, reference materials suitable for each meat product under examination would be required. Production of such appropriate meat standards is

306

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

however time-consuming and laborious. Additionally, due to the complexity in the manufacture of several meat products, with accompanying variations in manufacturers’ recipes and production style, generating reference materials appropriate for each commercial meat product might not be feasible. In this work, a multiplex real-time PCR assay for the quantitative determination of beef and pork fractions in minced meat is described. The triplex assay utilizes previously published animal species - specific primers and probes, relative to the proportion of the reference gene myostatin present in most mammals and bird species (Laube, Zagon, Spiegelberg, et al., 2007; Köppel, Ruf, Zimmerli, & Breitenmoser, 2008). The meat contents of the samples are accordingly computed as percentage compositions. Results from comparison of the triplex method with two other previously described assays are presented and discussed. 2. Materials and methods 2.1. Production of reference minced meat samples For validation of the presented triplex real-time qPCR method, approximately 6 kg of analytically pure beef and pork minced meat were prepared in a professional environment at the Bavarian Health and Food Safety Authority (LGL). 300 g of minced meat fractions derived from varying proportions of beef and pork were produced to cover a dynamic range of 5–95% beef/pork (50beef/50pork, 70beef/30pork, 80beef/20pork, 45beef/55pork, 5beef/95pork) and vice versa in a first series, and a second series of beef and pork mixtures to cover the trace regions of 0.1–2% of beef and pork respectively (98beef/2pork, 99beef/1pork, 99.5beef/0.5pork, and 99.9beef/0.1pork and vice versa). Homogenisation was carried out in a dedicated thermomixer (Thermomix TM21, Vorwerk, Germany) at mode 2 for up to 5 min. Mixtures were typically stored at 20 °C until required. 2.2. Minced meat and other meat products Additional to the reference minced meat samples described above, the performance and robustness of the presented quantitative triplex real-time PCR was tested on 50 commercially available minced meat samples randomly selected by the official food monitoring and surveillance authority. More than thirty meat products with varying composition and matrices were additionally included to assess transferability of the method to other meat matrices (see Tables 3 and 5). 2.3. DNA extraction Four grams each of the examined meat samples was subjected to DNA extraction procedures, employing a modified CTAB protocol previously described (ISO 21571:2005, modified). Additionally

Table 2 The table depicts precision (relative repeatability standard deviation, RSDr), accuracy and trueness results obtained from analysis of minced meat containing defined proportions of beef and pork. Results were compiled from at least 5 different runs with an average of 21 measurement points or test results. Actual pork proportion (%)

Measured pork proportion (%)

Precision (RSDr)%

Accuracy (%)

Trueness (%)

50 30 70 20 80 5 95

51.78 33.35 65.72 21.37 75.55 5.71 93.21

2.79 4.88 6.12 5.61 2.48 11.56 0.28

2.37 6.71 7.07 7.59 4.36 13.82 1.36

2.44 7.10 4.61 4.54 4.16 11.33 1.36

a commercially available silicon-column based DNA extraction kit (Surefood Animal X Kit, Congen Biotechnology, Germany) was used to extract DNA in parallel from a subset of meat products of other composition. The two extraction methods were compared to determine the suitability and efficiency of the commercial kit against the time-intensive CTAB Extraction protocol. Following DNA extraction, the purity and concentration of the DNA samples were confirmed either by conventional photometry, employing Nanodrop technology (Nanodrop 1000, Peqlab, Germany) or by Picogreen measurement. DNA samples were typically diluted 1:200, resulting in a final template concentration of at least 10 ng pro PCR reaction. 2.4. Primers and probes The primers and probes described in this work have been previously reported and are listed in Table 1. Beef and pork fractions were quantified over dedicated primer and probe sequences against the backdrop of a universal sequence commonly found in mammals, namely the housekeeping gene myostatin. For each of the three targets the specific TaqMan probe was labelled with a different fluorescence dye (see Table 1). The primer and probe systems applied in this work all target single copy, chromosomally encoded gene sequences. The 6-FAM, HEX and ROX – labelled probes were quenched with a Blackberry quencher (BBQ, TIB Molbiol, Berlin, Germany) on their 30 -end. Preliminary titration experiments were initially carried out to determine the optimal primer and probe concentrations for the multiplex reaction, without negative impact on the sensitivity of the assay. 2.5. Specificity Specificity of the applied primer and probe constellation is an important prerequisite for any real-time PCR system. Although the primers and probes applied in this work had been previously reported by other workers, an exhaustive specificity test was carried out against the backdrop of several animal and plant species

Table 1 Primer and probe sequences used for the quantitative triplex real-time PCR assay. Name

Target gene

Sequence 50 –30

References

Bos-PDE-f Bos-PDE-r Bos-PDE-probe (ROX)

Cyclic-GMP-phosphodiesterase

ACTCCTACCCATCATGCAGAT TGTTTTTAAATATTTCAGCTAAGAAAAA AACATCAGGATTTTTGCTGCATTTGC

Laube, Zagon, Spiegelberg, et al. (2007)

Sus1-F_pork Sus1-R_pork Sus1_TMP (HEX)

Beta-actin

CGAGAGGCTGCCGTAAAGG TGCAAGGAACACGGCTAAGTG TCTGACGTGACTCCCCGACCTGG

Köppel et al. (2008)

My-f My-r My-probe (6-FAM)

Myostatin

TTGTGCAAATCCTGAGACTCAT ATACCAGTGCCTGGGTTCAT CCCATGAAAGACGGTACAAGGTATACTG

Laube, Zagon, Spiegelberg, et al. (2007)

307

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

Table 3 Quantification of 50 commercial minced meat products randomly selected for validation of the triplex qPCR (P/B stands for pork before beef meat declaration, depicting a greater proportion of pork in the meat mixture, while B/P depicts beef for pork, indicating a greater proportion of beef in the mince). The calculated measurement of uncertainty (MU) and the expanded MU are also indicated. Quantification (%)

mm12_01 mm12_02 mm12_03 mm12_04 mm12_05 mm12_06 mm12_07 mm12_08 mm12_09 mm12_010 mm12_011 mm12_012 mm12_013 mm12_014 mm12_015 mm12_016 mm12_017 mm12_018 mm12_019 mm12_020 mm12_021 mm12_022 mm12_023 mm12_024 mm12_025 mm12_026 mm12_027 mm12_028 mm12_029 mm12_030 mm12_031 mm12_032 mm12_033 mm12_034 mm12_035 mm12_036 mm12_037 mm12_038 mm12_039 mm12_040 mm12_041 mm12_042 mm12_043 mm12_044 mm12_045 mm12_046 mm12_047 mm12_048 mm12_049 mm12_050

Declaration (%)

Pork

Beef

Pork

3.50 36.09 39.79 60.75 34.89 67.60 46.90 47.48 68.36 67.40 71.64 40.89 57.20 57.24 64.85 66.47 58.45 62.51 65.68 71.32 64.01 46.36 65.79 65.18 59.43 54.41 75.83 70.15 72.17 72.86 67.65 49.61 70.27 67.09 65.86 56.19 59.13 53.59 45.17 39.79 50.45 62.29 62.95 50.95 60.24 59.70 64.45 59.59 26.46 100.00

96.50 63.91 60.21 39.25 65.11 32.40 53.10 52.52 31.64 32.60 28.36 59.11 42.80 42.76 35.15 33.53 41.55 37.49 34.32 28.68 35.99 53.64 34.12 34.82 40.57 45.59 24.17 29.85 27.83 27.14 32.35 50.39 29.73 32.91 34.14 43.81 40.87 46.41 54.83 60.21 49.55 37.71 37.05 49.05 39.76 40.30 35.55 40.41 73.54 0.00

0

100 B/P B/P

40 40 45 50 50 50 50 50 50 50 55 55 55 55 55 55 55 55 55 55 55 60 60 60 65 65 65 65 65 65

60 60 55 50 50 50 50 50 50 50 45 45 45 45 45 45 45 45 45 45 45 40 40 40 35 35 35 35 35 35 P/B P/B P/B P/B P/B P/B P/B P/B P/B P/B P/B P/B P/B P/B P/B P/B

100

(see Supplementary information) especially relevant in the food industry, employing 1 ng/ll of template DNA with the presented triplex qPCR. 2.6. Preliminary animal species screening In order to exclude the presence of other commonly consumed meat products, a preliminary PCR screening was carried out. This was important because the reference gene myostatin is present in all mammals, and most bird and poultry species. For this screening PCR, the method of choice was the Chipron LCD Animal Array Kit (Chipron, Berlin Germany), which simultaneously detects the presence of up to 24 animal species in meat products (see Supplement). An alternative approach was the application of the commercial AllMeat multiplex qPCR kit (Microsynth, Switzerland, Köppel et al., 2008), which detects the presence of beef, pork, chicken

MU (%)

Expanded MU (%)

2.12 4.62 3.81 7.64 3.95 2.15 4.76 8.22 4.57 3.16 2.77 3.30 3.39 2.32 3.93 3.26 3.50 4.26 3.51 5.22 3.46 3.97 3.47 5.40 5.10 4.46 3.41 8.51 3.16 5.88 3.63 4.01 3.23 5.27 5.86 3.09 7.17 7.07 6.18 2.38 3.17 3.99 2.61 3.34 4.18 3.33 5.12 3.47 3.37 1.83

3.95 6.45 5.64 9.47 5.78 3.98 6.59 10.05 6.40 4.99 4.60 5.13 5.38 4.15 5.76 5.09 5.33 6.09 5.34 7.05 5.29 5.80 5.30 7.23 6.93 6.29 5.24 10.34 4.99 7.71 5.46 5.84 5.06 7.10 7.69 4.92 9.00 8.90 8.01 4.21 5.00 5.82 4.44 5.17 6.01 5.16 6.95 5.30 5.20 3.66

Beef

0

and turkey. Both the Chipron biochip kit and the AllMeat multiplex real-time PCR method were employed as screening approaches, strictly according to manufacturers’ instructions. Following exclusion of the presence of other meat species apart from beef and pork, the triplex qPCR method was carried out. 2.7. Quantitative triplex real-time PCR The optimised real-time PCR assay described in this work was carried out and validated on an Mx3005P real-time PCR cycler (Agilent Technologies, USA). Typically, a 25 ll reaction volume was employed containing the following components: 2 Quantitect Multiplex PCR NoROX reagent (Qiagen, Hilden, Germany), 5 ll template DNA and primer and probe at appropriately optimised concentrations. For the beef and pork specific systems, primer concentrations were optimised at 300 nM for beef and

308

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

Table 4 Comparison of the triplex qPCR with two other real-time PCR approaches approved for use by official food monitoring agencies. The first is based on the Laube Protocol (Laube, Zagon, Spiegelberg, et al., 2007) while the AllMeat is a commercially available PCR Kit for quantification of beef, pork and poultry fractions in food. The latter was used in connection with a DNA dilution series for quantification. Both methods were compared in terms of handling for routine analysis of meat products against the triplex qPCR described in this work. B/P denotes more beef than pork in meat product, while P/B implies more pork than beef fractions in mince. Sample

mm12_01 mm12_02 mm12_04 mm12_05 mm12_08 mm12_011 mm12_012 mm12_016 mm12_018 mm12_019 mm12_020 mm12_022 mm12_025 mm12_026 mm12_027 mm12_028 mm12_030 mm12_031 mm12_033 mm12_038 mm12_050

Laube, Zagon, Spiegelberg, et al. (2007) (%)

AllMeat (%)

Beef

Declaration (%) Pork

Beef

Pork

Beef

Pork

Beef

Pork

100

0

100.00 65.70 53.50 73.60 61.10 24.30 45.50 23.50 40.70 38.90 19.60 54.20 50.30 37.50 23.90 23.10 34.80 32.30 24.90 18.20 0.10

0.00 34.30 46.50 26.40 38.90 75.70 54.50 76.50 59.30 61.10 80.40 45.80 49.70 62.50 76.10 76.90 65.20 67.7 75.10 81.80 99.90

97.00 65.30 9.00 55.20 48.40 21.90 29.30 21.80 26.20 21.90 18.70 44.30 28.20 19.70 14.50 23.00 17.10 22.10 20.20 20.10 0.00

3.00 34.70 91.00 44.80 51.60 78.10 70.70 78.20 73.80 78.10 81.30 55.70 71.80 80.30 85.50 77.00 82.90 77.90 79.80 79.90 100.00

96.50 63.90 39.30 65.10 52.50 28.40 59.10 33.50 37.50 34.30 28.70 53.60 40.60 45.60 24.20 29.90 27.10 32.40 29.70 46.40 0.00

3.50 36.10 60.80 34.90 47.50 71.60 40.90 66.50 62.50 65.70 71.30 46.40 59.40 54.40 75.80 70.20 72.90 67.70 70.30 53.60 100.00

B/P 60 60 60 45 50 45 45 45 45 45 40 40 40 35 35 35 35

40 40 40 55 50 55 55 55 55 55 60 60 60 65 65 65 65 P/B

0

100

Triplex (%)

Table 5 Application of the triplex qPCR method for the quantitative determination of various meat products. Product description

Declaration (%)

Sample ID

Beef (%)

Pork (%)

MU (%)

Expanded MU (%)

Salami Frankfurter sausages Hamburger Cevapcici Veal sausage Noodle sauce Bolognese Salami Salami with chili Salami with herbs Bavarian veal sausage Cappellaci Veal liver sausage Beef meat loaf Regensburger small sausage Beer sausage Sausage Bifteki Salami with cheese Sausage Mortadella with pistachio Cheese kransky Lyoner sausages Mortadella with paprika Lyoner sausage Lyoner sausage Ham sausages Walnut salami Pfälzer Corned beef Meat loaf Ham Rolled ham fillet

Beef Beef Beef Beef Beef 27% Beef Pork/beef Pork/beef Pork/beef 50% (Pork/beef) , bacon, rind 26% Pork, 27% Beef 35% Pork, 25% Pig liver, bacon, 15% calf 43% Pork, 30% pork liver, 15% veal 45% Pork, 7% Beef, bacon 50% Pork, 35% beef 50% Pork, 35% beef, bacon 51% Pork, 39% beef 56,2% Pork, 32,8% beef 58% Pork, 24% beef 60% Pork, 5% beef, bacon 69% Pork, 22% beef 70% Pork, 5% beef, bacon 70% Pork, 2% Rind, bacon 70% Pork, bacon, 5% beef 75% Pork, bacon 80% Pork, 5% beef, bacon 86% Pork, 7% beef 86% Pork/4% beef 87% Beef, pork rind, -gelatine 89% (Pork/beef) Pork Pork

H12-1 H12-2 H12-3 H12-4 H12-5 H12-6 H12-7 H12-8 H12-9 H12-10 H12-11 H12-12 H12-13 H12-14 H12-15 H12-16 H12-17 H12-18 H12-19 H12-20 H12-21 H12-22 H12-23 H12-24 H12-25 H12-26 H12-27 H12-28 H12-29 H12-30 H12-31 H12-32

100.00 99.98 100.00 100.00 100.00 27.00 17.02 20.63 22.86 2.58 33.03 4.40 13.75 3.67 39.27 38.08 29.56 57.49 19.33 0.06 27.55 0.02 0.04 1.89 0.00 0.00 1.72 89.40 60.11 3.94 0.00 0.00

0.00 0.02 0.00 0.00 0.00 0.00 82.98 79.37 77.14 47.42 19.94 70.60 74.25 48.33 45.73 46.92 60.44 42.51 62.68 64.94 63.45 74.98 71.96 73.13 100.00 85.00 91.28 0.60 39.89 85.06 100.00 100.00

1.83 1.84 1.83 1.83 1.83 1.83 2.93 2.94 3.46 2.58 1.97 2.66 3.68 2.18 3.02 3.49 2.87 1.94 4.05 1.92 5.89 1.85 1.88 2.67 1.83 1.83 2.17 2.12 4.34 2.60 1.83 1.83

3.66 3.67 3.66 3.66 2.66 3.66 4.76 4.77 5.29 4.41 3.80 4.49 5.51 4.01 4.85 5.32 4.70 3.77 5.88 3.75 7.72 3.68 3.71 4.50 3.66 3.66 4.00 3.95 6.17 4.43 3.66 3.66

200 nM for pork, with 200 nM for the beef specific probe and 80 nM probe concentrations for the pork detection system. For the myostatin specific system, forward and reverse primers were employed at 300 nM and an optimised probe concentration of 200 nM was employed. The PCR thermo profile consisted of an initial denaturation and activation of the polymerase at 95 °C for 15 min, followed by 45 cycles with 30 s denaturation at 95 °C, and 60 s annealing at 60 °C.

The MxPro software (Agilent Technologies, USA) was employed for data analysis on the Mx3005P cycler. The three fluorescence channels (FAM, HEX and ROX) were analysed separately. 2.7.1. Generation of standard curves for quantification of the DNAcontent of the animal species For the generation of standard curves, the DNA concentration from pure beef and pork mince (or pure genomic DNA from both

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

meat species) was measured with the Nanodrop or with the Picogreen method and the genomic DNA was diluted with nucleasefree water to yield a DNA dilution series. Two separate calibration curves were created for beef and pork respectively with the following defined genome copy equivalents: 156250, 31250, 6250, 1250, 250, and 50. 2.7.2. Quantification strategy The method described here exploits the principle of relative quantification: the generated copy numbers for beef or pork fractions are extrapolated against the calculated copy numbers for the endogenous reference gene myostatin, to give the proportion of beef or pork in percentages as exemplified below:

x% ¼

100%  bos-PDE-cp my-cp

ð1Þ

y% ¼

100%  sus-cp my-cp

ð2Þ

where x% and y% denote the proportion of beef and pork in percentages, respectively, and bos-PDE-cp and sus-cp depict the generated copy numbers of beef and pork fractions in the samples as calculated from the respective standard curves, against the generated copy numbers of the endogenous universal gene myostatin (my-cp). 2.8. Validation of the quantitative triplex real-time PCR The triplex real-time qPCR assay presented in this work was critically assessed against recommended validation guidelines proposed in national and international documents such as the MIQE Guidelines governing the publication of quantitative real-time PCR experiments and the ENGL Criteria governing the assessment of precision and LOD of an analytical method (Bustin et al., 2009; ENGL, 2008). The efficiency, robustness, reproducibility as well as limit of detection and quantification (LOD, LOQ) of the assay were extensively tested in an in-house validation process. 2.8.1. Efficiency and precision of the triplex real-time qPCR assay PCR reaction efficiency was extrapolated from the slope of the line of best fit drawn to the standard curve. The standard curve plots the log of starting template vs. PCR cycle number, which is generated by the MxPro Software (Agilent Technologies, USA). Acceptance criteria were PCR efficiencies between 90 and 110%, typically corresponding to a slope of regression between 3.1 and 3.6, and Rsq value of P0.98. 2.8.2. Limit of detection (LOD6) Four grams each of beef and pork meat mixtures derived from the following beef/pork meat constellations: 100beef/0pork and 0beef/ 100pork respectively were subjected to DNA isolation procedures. The extracted DNA was accordingly diluted to yield 20,000, 5000, 1250, 250, 50, 20, 10, 5, 2, 1 und 0.1 genome copy equivalents per PCR reaction of cow and pig respectively. Two runs were carried out under repeatability conditions for reliable generation of the LOD6 (AFNOR Standard, 2008). In an extended experiment, the LOD95% was determined which is defined as the LOD at which the analytical assay detects the presence of the analyte at least 95% of the time (thus ensuring 6 5% false negative results) (AFNOR Standard, 2008; EURL Report, 2009). 2.8.3. Limit of quantification (LOQ) In the context of this work, the LOQ is defined as the lowest amount of the analyte in a sample that can be reliably quantified within an acceptable level of precision and accuracy (ENGL Criteria, 2008). Generally, the relative standard deviation under

309

repeatability conditions should be within the 25%–30% range. To assess the LOQ of the triplex qPCR, DNA extracts from the reference minced meat prepared for the analysis of trace samples were subjected to the triplex real-time PCR reaction under repeatability conditions. 2.8.4. Precision–relative repeatability standard deviation (RSDr) To evaluate the precision of the method, DNA from reference minced meat containing the following meat combinations were subjected to analysis: 50pork and 50beef, 70pork and 30beef, 30pork and 70beef, 20pork and 80beef, 80pork and 20beef, 5pork and 95beef, 95pork and 5beef. As acceptance criterion, the relative repeatability standard deviation was at least 625% over the whole dynamic range of the assay. Estimates of repeatability were obtained on sufficient number of test results, typically 18 or greater (ENGL Criteria, 2008, ISO 5725-3). 2.8.5. Robustness and reproducibility of the quantitative triplex PCR assay The robustness of the triplex qPCR described in this work was determined by assessment of the transferability of the method on different real-time cyclers. The procedure for determination of the LOD as described above, and originally carried out on the Mx3005P (Agilent Technologies) was repeated on the RotorgeneQ-Cycler (Qiagen, Germany). Additionally, three reference meat samples (95pork5beef, 95beef5pork and 70pork30beef) were analysed in parallel on the Mx3005P (Agilent Technologies), the RotorgeneQ-Cycler, and on the CFX384 real-time PCR Cycler from BioRad (Germany). Through application of different DNA extraction procedures, the variability of the DNA extraction procedures and its impact on the extraction efficiency and amplicability in the PCR reaction was also assessed. In this regard, the modified CTAB method was compared with other DNA extraction methods such as the commercially based silica-column DNA extraction kits: Surefood Prep Animal (Congen Biotechnology, Germany) in the extraction of DNA from meat of other matrices analysed in this work. 2.8.6. Measurement of uncertainty (MU) The measurement of uncertainty was determined by analysis of the in-house generated reference minced meat products with varying beef/pork composition (see Table 1 above). The reference minced meats (with meat mixtures in the 5–95% range) were subjected to the triplex qPCR method here described in three independent runs, with each sample analysed six times. The standard deviations for all measurements points were determined and the measurement of uncertainty accordingly computed. For the commercial minced meat products quantitatively analysed in this study, the measurement of uncertainty was determined separately for each sample result and added to the doubled MU determined for the reference mince samples to generate the expanded measurement of uncertainty. 2.9. Comparison of the proficiency of the real-time qPCR with other real-time based quantification strategies In order to assess the efficiency of the triplex qPCR method, its proficiency was compared with two other real-time qPCR systems that had been previously validated, published and partly adopted as analytical methods by various Official Food Monitoring and Surveillance Authorities. The first method was the AllMeat Kit (Microsynth, Switzerland, Köppel et al., 2008), a tetraplex real-time PCR method employing a DNA dilution series for quantification. The second was the relative quantification approach (two singleplex PCRs) published by Laube, Zagon, Spiegelberg, et al. (2007).

310

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

Both methods were carried out strictly according to the prescribed mode of practice. 3. Results and discussion 3.1. Efficiency of the triplex qPCR system In order to determine the suitability of the applied real-time PCR method for quantification of beef and pork fractions in minced meat products, the amplification efficiency (AE) and correlations coefficient (R2) were compared for multiple runs, employing dilutions of DNA extracted from selected meat products (data not included). The standard curves, which were linear over all six measurement points, showed an average AE P 91%, while the mean computed R2 was 0.999. Fig. 1 shows typical standard curves generated for the beef (bos PDE), pork (sus) and myostatin (my) specific gene systems. In most multiplex systems, the challenge is demonstrating comparable run efficiencies among the different systems included in the multiplex mix. Amplicon size and the amount of starting template both play important roles in the overall efficiency of the reaction. In the work of Bai, Xu, Huang, Cao, and Luo (2009), common disadvantages of multiplex systems were cited as low amplification efficiencies and unequal PCR proficiencies on different templates, thus decreasing the commercial application of such methods. The triplex qPCR in this work was developed with these parameters in mind. Short amplicon length (80 bp–104 bp) for example ensures that the amplified products are representative of the animal species present in the meat sample, even in highly processed samples. To demonstrate that no significant loss of sensitivity occurred, the efficiency of the triplex reaction was compared with several combinations of singleplex and duplex reactions (data not shown). The PCR reaction efficiencies, precision and sensitivities were comparable in all cases. The results in this work also showed a typical linear correlation between CT and log DNA, with a PCR efficiency nearing the predicted one doubling per cycle. 3.2. Specificity With the plant species tested, exclusivity was 100% (false positive rate 0%) for all employed primers and probe systems, with no cross reactivity observed. With the beef primer and probe system, a slight cross reactivity was observed with buffalo (ct values however appeared very late) while the pork system could not accurately discriminate between pork and wild boar. In the context of this work where quantitative determination of minced meat fractions is the focus, this observed cross reactivity may not be significant. Additionally, contamination by wild boar may be unlikely because wild boar is a less common and more expensive meat source. However, when analysing meat products where traces of venison are expected, this might merit some consideration. Such cross reactivity among animal species that are closely related have been previously reported (Iwobi, Huber, Hauner, Miller, & Busch, 2011; Rentsch et al., 2013). When the analytical focus of the work is not compromised by such cross-reactivity, the applied multiplex systems can be readily implemented without bias. 3.3. Validation of the triplex qPCR method 3.3.1. Limit of detection (LOD) and limit of quantification (LOQ) The limit of detection (LOD6) of the system reached 20 genome copies for both pork and beef which meets the acceptance criterion specified in the AFNOR XP V03-020-2 guidelines (2003). The LOD95% which was carried out to validate the result of the LOD6 over 60 replicates, verified the 20 genome copies as limit of detection of the assay (EURL-GMFF Report, 2009).

In this study, the absolute LOQ was determined at 1% for pork and 2% for beef. The presence of pork in greater proportions appeared to compromise the detection and quantification of trace amounts of beef in a sample. Thus 0.5% pork against a 99.5% background of beef could be reliably quantified, while the reverse, namely, 0.5% beef in a binary mixture containing 99.5% pork was not reproducibly quantified (data not shown). This could be due to the inherent fatty nature of pork which could hamper the efficiency of DNA extraction procedures (Laube, Zagon, & Broll, 2007). Other published work regarding quantification of beef and pork binary mixtures report similar analytical sensitivities, namely 1% sensitivity for low processed meat (López-Andreo et al., 2012; Laube, Zagon, Spiegelberg, et al., 2007; Rodríguez, García, Gonzalez, Hernandez, & Martin, 2005). 3.3.2. Precision and accuracy In order to assess the precision (relative repeatability standard deviation RSDr) and accuracy (relative mean deviation in% against the true value of pork/beef) of the triplex method, various minced meat fractions with defined proportions of beef and pork, covering the dynamic range of the assay, were analysed under repeatability conditions. The results, summarised in Table 2 indicate good performance of the applied method, with the calculated precision, accuracy and trueness of the method lying well within the acceptance criterion of 625% (ENGL 2008). 3.3.3. Robustness and reproducibility For assessment of robustness, the method was validated against the background of different PCR Cyclers: CFX 384 Biorad, USA), Rotorgene-Q (Qiagen, Hilden) and Mx3005P (Agilent Technologies, USA). All real-time PCR platforms supported with 100% fidelity the applied real-time triplex approach. In a complementary approach, the DNA extraction procedure was implemented with another DNA-isolation method, namely the Surefood kit (Qiagen, Hilden) for processing of meat of other matrices as analysed in this work. While the CTAB method was the best in terms of DNA yield and enabled the most representative sampling, the Surefood kit (Qiagen, Hilden) however also generated acceptable results with good DNA yield. 3.3.4. Measurement of uncertainty (MU) The measurement of uncertainty of the method was determined at 1.83%. This was estimated by calculating the statistical mean of the measurement of uncertainty (MU) generated from three independent runs (2.18%, 1.80%, and 1.52%) (see Section 2.8.6). 3.4. Validation of the triplex qPCR assay with 50 commercial minced meat products In order to demonstrate robustness of the qPCR assay, 50 commercially available minced meat products randomly selected by the official food surveillance authority were analysed by the method. The results, which are summarised in Table 3, indicate good applicability of the method on real mince samples. Generally, the proportion of quantified pork was greater in relation to the quantifiable beef in most of the samples. This was expected as pork, being the cheapest meat source of the two, would be preferentially used in greater amounts in the production of commercial mince. In at least one of the samples, (mm12_049), the percentage of pork was surprisingly almost a third lower than the quantified beef fractions in the mince, although the product declaration indicated the presence of more pork than beef in the sample. After repeated analysis of the sample, no significant deviation in results was observed. A possible explanation for this perceived discrepancy could be an inadvertent mislabelling of the product. Alternatively, the producers might have used ‘‘normal’’ beef tissues

311

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

Myostatin (FAM)

Myostain

R²= 0.999 Eff: 101.4%

Beef (ROX)

Beef

R²=0.999 Eff: 101.1%

Pork (HEX)

Pork

R²=0.996 Eff: 91.6%

Fig. 1. Amplification plots and accompanying standard curves for the quantification of beef and pork fractions in minced meat. Three standard curves are generated for the calculation of pork and beef fractions extrapolated against the proportion of myostatin in the respective samples.

(homogeneous mix of muscle and fatty tissues) with more fatty components of pork. Because fatty tissues give a comparatively lower DNA yield, the presence of pork in the minced meat might be underestimated against the backdrop of the beef fractions present in the mix, although this would make little sense from a

commercial point of view. In the work of Laube, Zagon, and Broll (2007), various tissue types (kidney, heart, liver, sinews, muscle, brain, fatty tissue etc.) taken from a single pig were examined and an assessment of their extracted DNA concentrations was carried out. The assessment uncovered significant variations in the

312

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

extracted DNA quantities, with kidney, heart, liver and sinews yielding the highest DNA levels, while fatty tissues yielded considerably lower DNA concentrations compared with muscle. The difficulties in homogenizing fatty substances were cited as possible factors impacting the efficiency of the DNA extraction procedure, coupled with the fact that high quantities of fatty tissues might influence the separation of phases during DNA isolation employing the CTAB method. Sometimes however, the difficulty in carefully cleaning meat processing equipment between batches might result in unavoidable cross contaminations, which do not constitute economic fraud – this fact thus merits consideration in the analysis of results. 3.5. Comparison of the performance of the triplex qPCR with other real-time based quantification approaches For the comparison, 21 minced meat products were examined in parallel with the triplex qPCR method, the AllMeat Kit and the method from Laube, Zagon, Spiegelberg, et al. (2007) (see Section 2.9). Results are depicted in Table 4. The triplex qPCR compared well with the Laube method which utilised two singleplex reactions on a single reaction plate – differences in generated beef and pork percentages were on average 6.9%, with some mince samples, for example mm12_031 exhibiting no significant percentile difference between the two methods. One sample, namely mm12_038, which according to manufacturer’s declaration contains more pork than beef (P/B), however exhibited unusually high variance between the two methods (more than 25% percentile variance). The Laube protocol relies on two separate singleplex reactions for quantification of beef relative to the universal sequence myostatin present in the sample (our triplex PCR reaction is an adaptation of the Laube method which employs a multiplex rather than two singleplex reactions). While the method directly quantifies only the presence of beef in the sample, and the proportion of pork is indirectly inferred (100% – % beef), the sensitivity of our triplex reaction might be higher because the proportion of beef and pork (using beef and pork specific detection systems) are concomitantly computed against the backdrop of the universal sequence myostatin. Following normalisation of results (beef and pork fractions must yield 100% total meat content), the triplex reaction might offer a more realistic and accurate quantification strategy because of its dual analytical approach. Generally, the amount of pork in the products was considerably more than the product declaration – in most cases up to 10% over-quantification of pork fractions. This is however expected as most manufacturers of mince would most likely incorporate more pork, which is a cheaper meat source, than beef in the production process. While multiplex real-time PCR systems can be sometimes limited in efficiency and are more prone to variability of results compared with singleplex reactions, the results generated in this work indicate comparable results for both the Laube method (2 singleplex reactions) and the triplex qPCR method. No significant differences were found between the two methods, with the computed beef proportions in the samples and extended measurement of uncertainties indicating comparable results for all analysed data sets. In contrast, the AllMeat method although showing good comparability at a few measurement points, exhibited generally the greatest variability in computed meat percentages. This illustrates a common problem inherent in quantification strategies that rely purely on a DNA dilution series. In the work of Eugster et al. (2009), the accurate measurement of meat proportions using DNA-based methods is impaired when analysing samples with a variety of tissue types. Since different tissue types may exhibit variable DNA concentrations, such DNA-based analytical procedures may not always be inherently accurate.

3.6. Application of the triplex qPCR for quantification of other commercially available meat products with different matrices In order to assess the transferability of the triplex qPCR to other matrices, the method was used for quantitative assessment of meat products with other composition and texture (see Table 5). The results show good performance of the triplex qPCR, with the computed measurement of uncertainty lying between 1.83 and 5.9, indicating little dispersion of results. The commercial products examined in this part of the study exhibited great variability in texture and composition – cheese kransky (sausages like frankfurters with cheese), mortadella with paprika, beer sausages, salami and beef meat loaf. It is therefore noteworthy that regardless of these apparent differences in texture and complexity of the products, the triplex qPCR performed considerably well. It is well known that the sensitivity of PCR reactions relies heavily on the quality of the initial DNA template. In cases where food or meat products have been heavily processed, accompanying PCR reactions may not perform optimally due to decreased DNA quality and concentration (Buntjer, Lamine, Haagsma, & Lenstra, 1999; Laube, Zagon, & Broll, 2007). In the work of Laube, Zagon, and Broll (2007) the application of a real-time PCR approach allowed for quantification in low-processed foods at 10 times the efficiency for foods with a much higher processing index. López-Andero, Aldeguer, Guillén, Gabaldón, and Puyet (2012) also investigated in their work the correlation between heat treatment and the extent of DNA degradation. They reported that cooking at 65 °C followed by sterilisation at 126 °C from 10 to 30 min led to DNA rupture to approximately 100 bp-long fragments, which however still allowed for detection of 5% pork and its accurate quantification in binary mixtures. The authors thus concluded that the capability of short qPCR detectors considerably enhanced the PCR efficiency. In this work, the PCR systems employed all target genome sequences of about 100 bp or less, thus enabling the amplification of sequences in trace amounts or in highly processed foods. Accordingly, the analysis of samples with some degree of processing as with the commercial products like meat loaf, mortadella, or Pfälzer as listed in Table 5 was possible. 3.7. Use of a reference gene for meat species quantification One of the biggest challenges in the quantitative assessment of meat with DNA-based analysis lies in the generation of accurate results against the backdrop of the variability of tissues employed in the production of such meat samples. Equal amounts of beef and pork lean muscle may not contain the same number of copies of target DNA. In the study by Laube, Zagon, and Broll mentioned previously (2007), the extracted DNA quantities from various tissues showed high variations. A similar study in our laboratory also revealed striking variations in the DNA extractable from the same amount of starting material from different tissue types taken from an animal. Generally, fatty tissues yielded the least amount of DNA with higher yields observed for kidney, liver, connective tissue and muscle. In order to address this issue, the use of matrix-adapted reference material containing different proportions of meat species has been proposed. In the work of Eugster, Ruf, Rentsch, Hubner, and Köppel (2008), Eugster et al. (2009), such matrix-adapted standards yielded more accurate results than the use of just DNA dilutions to build calibration curves. However, it is not economically feasible or logistically possible to produce such matrix-adapted standards for each commercial product category. Additionally, the manufacturing process for the same product might vary widely in terms of the meat and tissue types employed. In this work, an alternative approach is presented, namely the relative quantification of meat components over the universal gene myostatin. This triplex qPCR approach was applied for the

A. Iwobi et al. / Food Chemistry 169 (2015) 305–313

quantification of minced meat fractions and other commercial products. The accuracy as revealed through validation on binary mince of known composition was between 1.36% and 13.82% on average for pork fractions in all tested samples, while the precision was between 0.3% and 11.6% (Table 2). These values compare favourably with previous reports where matrix-adapted standards were employed for quantification (Köppel et al., 2012; Rentsch et al., 2013). In the work of López-Andreo et al. (2012) where a single matrix reference material was used, the accuracy for all runs were 17% and 13% for non-treated and treated samples respectively. While the triplex qPCR relies on the principle of relative quantification over the endogenous universal myostatin, it is crucial to pre-screen the samples for the presence of other meat sources which could interfere with reliable result generation. When this basic requirement is met, the use of a reference gene for quantification could successfully circumvent the rigours associated with matrix-adapted standards. 4. Conclusion The method presented in this work offers a reliable quantification strategy for minced meat, which was successfully extended to other commercial meat products with varying matrix and composition. When compared with two other quantification strategies for minced meat analysis, the method was shown to be robust and the results were comparable in accuracy for all measurement points. The calculated measurement of uncertainties when the method was applied to other commercial meat products was surprisingly low and this shows the method can be readily transferred to other products of different matrix and composition. Finally, the triplex method with its reliance on the universal sequence myostatin and dual quantification nature (beef and pork fractions are quantified in parallel relative to the myostatin content) offers a unique alternative to the application of matrix-adapted standards. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foodchem.2014. 07.139. References AFNOR XP V03–020-2. (2003). Produits alimentaires. Détection et quantification des organismes végétaux génétiquement modifiés et produits dérivés. Partie 2: Méthodes basées sur la réaction de polymérisation en chaîne. Norme expérimentale. AFNOR Standard XP-V-03-044. (2008). Critères de validation intra-laboratoire pour les méthodes de détection et quantification de séquences d’acides nucléiques spécifiques; AFNOR: Saint-Denis La Plaine. Ali, M. E., Hashim, U., Sabar Dhahi, Th., Mustafa, S., Bin Che Man, Y., & Abdul Latif, Md. (2012). Analysis of pork adulteration in commercial burgers targeting porcine-specific mitochondrial cytochrome b gene by Taqman probe real-time polymerase chain reaction. Food Analytical Methods, 5, 784–794. Bai, W., Xu, W., Huang, Y., Cao, S., & Luo, Y. (2009). A novel common primer multiplex PCR (CP-M-PCR) method for the simultaneous detection of meat species. Food Control, 20, 366–370.

313

Buntjer, J. B., Lamine, A., Haagsma, N., & Lenstra, J. A. (1999). Species identification by oligonucleotide hybridisation: The influence of processing of meat products. Journal of the Science of Food and Agriculture, 7, 53–57. Bustin, S. A., Benes, V., Garson, J. A., Hellemans, J., Huggett, J., Kubista, M., et al. (2009). The MIQE guidelines: Minimum information for publication of quantitative real-time PCR experiments. Clinical Chemistry, 55, 611–622. Drummond, M. G., Brasil, B. S. A. F., Dalsecco, L. S., Brasil, R. S. A. F., Teixeira, L. V., & Oliveira, D. A. A. (2013). A versatile real-time PCR method to quantify bovine contamination in buffalo products. Food Control, 29, 131–137. Eugster, A., Ruf, J., Rentsch, J., Hubner, P., & Köppel, R. (2008). Quantification of beef and pork fractions by real-time PCR analysis: Results of an interlaboratory trial. European Food Research Technology, 227, 17–20. Eugster, A., Ruf, J., Rentsch, J., & Köppel, R. (2009). Quantification of beef, pork, chicken and turkey proportions in sausages: Use of matrix-adapted standards and comparison of single versus multiplex PCR in an interlaboratory trial. European Food Research Technology, 230, 55–61. European Network of Genetically Modified Organism Laboratories (ENGL) Document. (2008). Definition of minimum performance requirements for analytical methods of GMO testing. http://gmo-crl.jrc.ec.europa.eu/doc/ Min_Perf_Requirements_Analytical_methods.pdf. European Union Reference Laboratory for GM Food and Feed (EURL-GMFF). (2009) Report on the verification of the performance of a construct-specific assay for the detection of Flax CDC Triffid Event FP967 using real-time PCR. 15–10-2009. Girish, P. S., Haunshi, S., Vaithiyanathan, S., Rajitha, R., & Ramakrishna, C. (2013). A rapid method for authentication of Buffalo (Bubalus bubalis) meat by Alkaline Lysis method of DNA extraction and species specific polymerase chain reaction. Journal of Food Science Technology, 50, 141–146. International Standard (ISO) 5725-3. (1994). Accuracy (Trueness and Precision) of Measurement Methods and Results. Genève, Switzerland: International Organisation for Standardisation. International Standard (ISO) 21571:2005. Foodstuffs – Methods of analysis for the detection of genetically modified organisms and derived products – Nucleic acid extraction. Iwobi, A. N., Huber, I., Hauner, G., Miller, A., & Busch, U. (2011). Biochip technology for the detection of animal species in meat products. Food Analytical Methods, 4, 389–398. Laube, I., Zagon, J., Spiegelberg, A., Butschke, A., Kroh, L. W., & Broll, H. (2007). Development and design of a, ready-to-use’ reaction plate for a PCR-based simultaneous detection of animal species used in foods. International Journal of Food Science Technology, 42, 9–17. Laube, I., Zagon, J., & Broll, H. (2007). Quantitative determination of commercially relevant species in foods by real-time PCR. International Journal of Food Science Technology, 42, 336–341. López-Andero, M., Aldeguer, M., Guillén, I., Gabaldón, J. A., & Puyet, A. (2012). Detection and quantification of meat species by qPCR in a heat-processed food containing highly fragmented DNA. Food Chemistry, 134, 518–523. Köppel, R., Ruf, J., Zimmerli, F., & Breitenmoser, A. (2008). Multiplex real-time pCR for the detection and quantification of DNA from beef, pork, chicken and turkey. European Food Research Technology, 227, 1199–1203. Köppel, R., Eugster, A., Ruf, J., & Rentsch, J. (2012). Quantification of meat proportions by measuring DNA contents in raw and boiled sausages using matrix-adapted calibrators and multiplex real-time PCR. Journal of AOAC International, 95, 494–499. Köppel, R., Daniels, M., Felderer, N., & Brünen-Nieweler, C. (2013). Multiplex realtime PCR for the detection and quantification of DNA from duck, goose, chicken, turkey and pork. European Food Research Technology, 236, 1093–1098. Mane, B. G., Mendiratta, S. K., & Tiwari, A. K. (2012). Beef specific polymerase chain reaction assay for authentication of meat and meat products. Food Control, 28, 246–249. Rentsch, J., Weibel, S., Ruf, J., Eugster, A., Beck, K., & Köppel, R. (2013). Interlaboratory validation of two multiplex quantitative real-time PCR methods to determine species of cow, sheep and goat as a measure of milk proportions in cheese. European Food Research Technology, 236, 217–227. Rodríguez, M. A., García, T., Gonzalez, L., Hernandez, P. E., & Martin, R. (2005). Taqman real-time PCR for the detection and quantification of pork in meat mixtures. Meat Science, 70, 113–120. The Economist Online. (2012). Kings of the Carnivores. http://www.economist.com/ blogs/graphicdetail/2012/04/daily-chart-17.