The authors have declared that no competing interests exist.
Current address: College of Food Science and Technology, Agriculture University of Hebei, Baoding, Hebei, P. R. China
Food adulteration and feed contamination are significant issues in the food/feed industry, especially for meat products. Reliable techniques are needed to monitor these issues. Droplet Digital PCR (ddPCR) assays were developed and evaluated for detection and quantification of bovine, porcine, chicken and turkey DNA in food and feed samples. The ddPCR methods were designed based on mitochondrial DNA sequences and integrated with an artificial recombinant plasmid DNA to control variabilities in PCR procedures. The specificity of the ddPCR assays was confirmed by testing both target species and additional 18 non-target species. Linear regression established a detection range between 79 and 33200 copies of the target molecule from 0.26 to 176 pg of fresh animal tissue DNA with a coefficient of determination (R2) of 0.997–0.999. The quantification ranges of the methods for testing fortified heat-processed food and feed samples were 0.05–3.0% (wt/wt) for the bovine and turkey targets, and 0.01–1.0% (wt/wt) for pork and chicken targets. Our methods demonstrated acceptable repeatability and reproducibility for the analytical process for food and feed samples. Internal validation of the PCR process was monitored using a control chart for 74 consecutive ddPCR runs for quantifying bovine DNA. A matrix effect was observed while establishing calibration curves with the matrix type under testing, and the inclusion of an internal control in DNA extraction provides a useful means to overcome this effect. DNA degradation caused by heating, sonication or Taq I restriction enzyme digestion was found to reduce ddPCR readings by as much as 4.5 fold. The results illustrated the applicability of the methods to quantify meat species in food and feed samples without the need for a standard curve, and to potentially support enforcement activities for food authentication and feed control. Standard reference materials matching typical manufacturing processes are needed for future validation of ddPCR assays for absolute quantification of meat species.
Food adulteration or mislabeling remains a key challenge to the food industry [
In order to protect the food and feed industry and consumers, regulatory bodies need to monitor food authenticity and feed contamination by ruminant materials. Reliable, rapid and accurate methods are needed for use in food authentication and in feed control [
The objective of this study is to develop, optimize and validate common droplet digital PCR (ddPCR) assays integrated with an internal control for detection and quantification of bovine, porcine, chicken and turkey species in meat products and animal feed samples. Mitochondrial DNA was selected as the PCR target due to its high copy numbers present in cells, facilitating detection of trace quantities in food and feed matrices, especially when these commodities may be subject to extensive processing. To our knowledge, this is the first comprehensive report to describe ddPCR assays integrated with an internal control to quantify bovine, chicken, porcine and turkey DNA in food and feed samples.
All fresh raw animal tissue samples, raw milk, blood and plant samples were obtained either from a local grocery store or from Animal Health Laboratory Services, University of Guelph, Ontario. Dry feed samples were obtained from local feed processing plants in Ontario. Pure fresh beef, pork, chicken and turkey muscle tissues and poultry and pork meal feed samples were used as reference materials. The species identities of all materials were confirmed by DNA barcoding based on the standard CO1 gene [
Target |
Primer name | Sequence (5’-3’) |
Length (bp) | Accession number | Position on reference sequence | Reference |
---|---|---|---|---|---|---|
Cytb-F | 106 | L08376 | 690–714 | [ |
||
Cytb-R | 774–795 | [ |
||||
Cytb-Probe | 716–735 | [ |
||||
F7773 | 313 | AF034253 | 8950–8971 | [ |
||
R8064 | 9241–9262 | [ |
||||
Porcine-Probe | 9040–9069 | This study | ||||
12S-FW | 122 | KP171707 | 90–115 | [ |
||
12S-RV2 | 185–211 | [ |
||||
12S-Probe | 129–154 | This study | ||||
F8108 | 270 | KC153975 | 8108–8128 | [ |
||
R8357 | 8357–8377 | [ |
||||
Bovine-Probe | 8139–8167 | This study | ||||
IC-F | 134 | 1–25 | This study | |||
IC-R | 115–134 | |||||
IC-Probe | 27–52 | |||||
IC-Fragment |
aProbes were labeled with 6-FAM or Cal Fluor Orange (for IC) at 5’ end.
bMinor modifications were made to the published primers.
ddPCR target | ddPCR target | ||||||||
---|---|---|---|---|---|---|---|---|---|
Samples |
Porcine | Chicken | Turkey | Bovine | Samples |
Porcine | Chicken | Turkey | Bovine |
Bovine blood | - | - | - | + | Turkey burgers | - | - | + | - |
Bovine heart | - | - | - | + | Turkey hot dogs | - | - | + | - |
Beef meat (cooked) -2 | - | - | - | + | Turkey kidney | - | - | + | - |
Beef meat (raw) -2 | - | - | - | + | Turkey liver | - | - | + | - |
Bovine milk | - | - | - | + | Turkey meat (cooked) | - | - | + | - |
Beef salami | - | - | - | + | Turkey meat balls | - | - | + | - |
Beef wiener | - | - | - | + | Turkey meat tissue -2 | - | - | + | - |
Bovine liver | - | - | - | + | Turkey sausage | - | - | + | - |
Chicken breast (cooked) | - | + | - | ||||||
Chicken breast fillers | - | + | - | - | Dog meat tissue | - | - | - | - |
Chicken heart | - | + | - | - | Duck meat tissue | - | - | - | - |
Chicken hot dogs | - | + | - | - | Goat meat tissue | - | - | - | - |
Chicken liver | - | + | - | - | Horse meat tissue | - | - | - | - |
Chicken meat (raw) -3 | - | + | - | - | Mouse meat tissue | - | - | - | - |
Chicken wieners | - | + | - | - | Rabbit meat tissue | - | - | - | - |
Chicken-pork bologna | + | + | - | - | Rat meat tissue | - | - | - | - |
Pork ham | + | - | - | - | Sheep meat tissue | - | - | - | - |
Porcine kidney | + | - | - | ||||||
Porcine liver | + | - | - | - | Pollock fish | - | - | - | - |
Porcine liver sausage | + | - | - | - | Salmon | - | - | - | - |
Pork meat (cooked) | + | - | - | - | Sole fish | - | - | - | - |
Pork meat (raw) -2 | + | - | - | ||||||
Porcine spleen | + | - | - | - | Corn | - | - | - | - |
Pork summer sausage | + | - | - | - | Rice | - | - | - | - |
Pork-beef salami | + | - | - | - | Wheat | - | - | - | - |
Pork-chicken bologna | + | + | - | - | Soybean | - | - | - | - |
Turkey bacon | - | - | + | - |
aThe number (n) after an animal or food sample indicates that multiple (n) samples from different individual animals were tested.
bSpecies ID was confirmed for representative target species and all non-target animal and fish species by DNA barcoding.
Fresh meat samples were first trimmed of skin and excess fat, and deboned (if applicable), and then cut into ~1.0–1.5 cm3 pieces. The meat pieces were placed in a Cuisinart grinder and homogenized for 3–5 minutes. The homogenized samples were then used for DNA extraction. Cooked meat was prepared by autoclaving the meat pieces for 15 minutes at 121°C, and 17.5 psig pressure. The meat samples were allowed to cool, homogenized using a Cuisinart grinder, and then air dried for 72 hours at room temperature (approximately 22°C) or dried in an oven at 70°C for 24 hours to a moisture level of 4–5%. After drying, the samples were ground using a mortar and pestle and passed through a sieve (mesh no. 100) to obtain fine powder. The moisture content was measured using an air-oven method following AOAC 983.18 [
Fortified food or feed samples were prepared in a mortar by mixing/homogenizing the appropriate mass of the heat-processed and dried pure meat species powder (500 mg dry wt) with a pure meat, food or feed sample of a different species (4500 mg or 9500 mg dry wt) to obtain 10% and 5% of target species samples initially. Fortified samples at lower concentrations were prepared by mixing 1000–2000 mg (dry wt) of a homogenate at a higher concentration with 1000–4000 mg (dry wt) pure meat, food or feed powder samples accordingly. The samples were portioned into appropriate numbers of 100 mg sub-samples in 1.5 mL micro-centrifuge tubes for testing, or frozen in a -80°C freezer for testing at a later date.
Total genomic DNA was extracted from a sub-sample (100 mg) of a homogenized representative food or feed specimen using DNeasy Blood and Tissue® Kit (Qiagen, Mississauga, ON, Canada) following the manufacturer’s protocol. DNA concentrations and quality, including A260 and A280, were measured using a NanoDrop ND-2000 UV Vis ND-2000 Spectrophotometer (Thermo Fisher Scientific, Ottawa, ON, Canada) and a Qubit Fluorometer and Qubit dsDNA BR Assay Kit (Thermo Fisher Scientific). Extracted DNA samples were diluted to a concentration of 10 ng/μL prior to use, or frozen in a -20°C freezer for use at a later date.
Primers and probes (
To optimize internal control use in ddPCR assays, 10-fold serial dilutions of the internal control plasmid DNA, corresponding to 0.06–60,000 fg/μL, were prepared, mixed with Bovine DNA (2.0 ng/μL) in a 1:4 ratio, and tested using the ddPCR in duplicates. A higher level of the IC (>4500 copies/PCR) caused competitive amplification with the target DNA (
The ddPCR reaction conditions were optimized using varied amounts of target DNA in the presence of other non-target species. Optimization experiments included optimizing annealing temperatures (53–63°C) using a Gradient T100 Thermal Cycler (Bio-Rad, Mississauga, ON, Canada), and varied concentrations of primers (400–1000 nM), and probes (200 nM—500 nM). The optimized PCR reaction mixture (25 μL/reaction) contained 1x ddPCR Supermix for Probe (Bio-Rad), 96 nM each of the primers and 64 nM probe for the animal target, 40 nM each of the primers and 32 nM probe for the internal control, 1700 copies of internal control plasmid DNA and 40–50 ng of template DNA. From each PCR reaction mixture, 20 μL were mixed with 70 μL of Droplet Generation oil for Probes (Bio-Rad) in a DG8 Cartridge (Bio-Rad). The cartridge was covered with a DG8 gasket for ddPCR and loaded into the QX200 Droplet Generator (Bio-Rad) to generate PCR droplets. From each droplet mix, 20 μL were then transferred to a 96-well PCR plate (Bio-Rad). The plate was sealed with a foil heat seal using PX1™ PCR plate Sealer (Bio-Rad). PCR thermal cycling was conducted using a GeneAmp™ PCR System 9700 (Applied Biosystems), following optimized cycling conditions: an initial incubation at 95°C for 10 min, 48 cycles of 20 s at 95°C and 40 s at 59–60°C, followed by a final incubation at 98°C for 10 min and holding at 10°C until reading time. The amplification signals were read using the QX200™ Droplet Reader and analyzed using its associated QuantaSoft software (Bio-Rad) and recorded as copies/μL with confidence intervals of 95%. The results from 13,000 or more droplets were accepted and converted into % by weight or by DNA mass for reporting. A ddPCR result was considered acceptable only if the IC gave the expected output with a ≤20% variation.
To evaluate the ddPCR assays for testing food and feed samples, the assays were tested for their specificity, quantification range, repeatability, reproducibility, matrix effect, robustness and effect of DNA degradation. All experiments were conducted in duplicate unless otherwise indicated. The specificity of the ddPCR assays, including the IC ddPCR, was confirmed
To test the robustness of the assays, the methods were evaluated under different conditions that can be variable under laboratory practices, including DNA storage (fresh versus frozen for 3 weeks), prolonged PCR reagent shelf life (fresh versus old PCR master mix with a few days of shelf life prior to expiry date) and different PCR machines (two systems with different ages but same brand). The samples used were bovine, porcine or chicken-fortified feed or food at 5 or more concentration levels within the quantitative ranges of the assays (
Statistical analyses were performed using chi-square test or paired sample t-test with SAS 9.4 software program to reveal quantification variation significance against different matrices and experimental parameters. GraphPad Prism 6 was used to determine linearity and draw graphs.
Specificity of the primers and probes was tested first
The linearity of the optimized ddPCR assays was determined using purified DNA from fresh meat tissue at concentrations of 0.26–532 pg/PCR. The linear regression was established within the concentrations of 1.3–106.8 pg/PCR for beef and chicken, 2.2–176.0 pg/PCR for pork, and 0.26–64 pg/PCR for turkey. These ranges of concentrations were equivalent to 79–33200 copies/PCR, with coefficients of determination (R2) ranging from 0.997–0.999 (p-value < 0.0001) (
DNA was extracted from fresh raw (A) beef, (B) pork, (C) chicken, and (D) turkey. Shown are results from two replicates in green and blue colors.
Limit and range of quantification of the complete analytical processes were determined based on data from fortified heat-processed food/feed samples at different concentrations (0.005% to 10.00%, wt/wt) of beef in poultry meal, pork in chicken, chicken in pork or turkey in pork. The quantification range was found to be 0.05–3.00% (wt/wt) for heat-processed beef and turkey and 0.01–1.0% (wt/wt) for heat-processed pork and chicken with the coefficient of determination (R2) of 0.979–0.998 (p-value < 0.0001 for beef, pork and turkey, and p-value = 0.003 for chicken) (
(A) beef in poultry meal, (B) pork in chicken, (C) chicken in pork, and (D) turkey in pork. Shown are results from four replicates in green, blue, purple and black.
The ddPCR methods were tested for repeatability using replicated DNA extractions, PCRs and by using fortified heat-processed meat or feed samples. With few exceptions, the relative standard deviations (RSD) calculated for repeatability between PCR replicates and between DNA extractions were below 20% (
Relative standard deviations (RSD%) are presented for fortified heat-processed (A) beef in poultry meal, (B) pork in chicken, (C) chicken in pork and (D) turkey in pork. Lines represent average RSD% for repeatability between DNA extractions and reproducibility between different operators. The data points cover all five spiking levels as labeled for each of the target species tested.
The results were obtained from 74 consecutive runs conducted on different days. The variation (RSD %) was 8.6.
Matrix effect was evaluated using a heat-processed animal tissue sample spiked into food or feed matrices. A matrix effect was observed in some of the fortified samples. For example, there was an over 45% difference in ddPCR output when spiking chicken into beef and pork salami as compared to chicken spiked into beef hot dogs (
A cooked chicken sample was spiked in pork summer sausage, beef hot dog or beef and pork salami matrices.
The ddPCR assays were evaluated using DNA prepared from fortified samples of beef, pork and chicken within their quantification ranges under different DNA storage conditions, prolonged PCR reagent shelf life and using different PCR machines. RSD (%) ranges obtained under the different conditions were 2.0–22.6, 1.5–14.9 and 1.7–13.1 for the bovine, porcine and chicken ddPCR assays respectively (
The effect of DNA degradation on ddPCR output was tested on both raw and cooked beef DNA after 3 different degradation treatments: Taq I restriction enzyme digestion, 2 min sonication, and 10 min sonication. The ddPCR output from degradation-treated DNA was compared to untreated DNA. Taq I digestion was found to significantly reduce ddPCR output for both raw and cooked beef DNA. Sonication for 10 min resulted in over 4.5 fold reduction in ddPCR output for raw beef DNA, and approximately 20% reduction for cooked beef DNA while sonication for 2 min showed little effect on ddPCR output for either raw or cooked meat DNA (
Sample type | Copies/PCR upon different treatments | ||||
---|---|---|---|---|---|
(0.032 ng DNA in AE Buffer) | Replicate | No treatment | Taq I digestion | 2 min sonication | 10 min sonication |
Raw beef DNA | 1 | 6800 | 4400 | 6000 | 1380 |
2 | 6760 | 4360 | 6520 | 1480 | |
Cooked beef DNA | 1 | 2040 | 1360 | 1800 | 1520 |
2 | 1820 | 1300 | 1680 | 1600 |
The ddPCR provides an absolute quantification of target DNA without relying on a standard curve. Specifically, the ddPCR output is in copies of input DNA. The correlation between the amount of DNA from fresh tissue and their copy numbers from the ddPCR was y = 291235x+103.18 for bovine, y = 188491x + 205.78 for porcine, y = 181118x−467.71 for chicken, and y = 398422x−240.49 for turkey (where y is the copy number and x is ng of DNA) (
The IC readings were used to normalize PCR outputs affected by variabilities in the PCR procedures. However, normalization of the ddPCR outputs to the IC readings did not affect the result if the IC readings were within 20% variation as compared to the expected values.
Food authentication continues to be of interest in an era of globalization as more reports and studies demonstrate a high incidence of adulteration and/or mislabeling. Here we describe and validate ddPCR- based assays for quantitative analysis of bovine, porcine, chicken and turkey DNA in food and feed. The ddPCR assays were evaluated systematically for their specificity, limit/range of quantification, repeatability and reproducibility, matrix effect and robustness. Other researchers have reported ddPCR assays for meat species quantification previously, including ddPCR assays for quantifying pork and chicken species [
The ddPCR assays described here were demonstrated to be specific upon testing 10–11 samples containing the target species and 45–46 samples belonging to 18 non-target species. The linear quantification range of these methods was 0.26–176 pg/PCR for fresh meat tissue DNA and 0.01–1.0% (wt/wt) for porcine and chicken ddPCR, and 0.05–3.0% (wt/wt) for bovine and turkey ddPCR for fortified heat-processed food and feed. Floren et al [
The target sequences used for meat species detection in this study were mitochondrial DNA (mtDNA), which are widely used in animal species detection in complex food and feed samples [
The ddPCR assays developed in this study contained an internal control (IC) to ensure reliability of the results, to normalize variabilities in the PCR procedures and to safeguard against false negatives due to factors such as PCR inhibition or reaction failure. The internal control plasmid DNA was added at a level to generate approximately 1700±20% copies per ddPCR reaction. At this IC concentration, no amplification competition was observed between the IC and the target DNA. Competitive amplification between the detection target and the IC was observed in ddPCR when more copies of IC were used and the target copy number was lower than that of IC in a reaction (
Matrix effect was observed when testing fortified samples, or heat-processed animal tissue spiked into different food and feed sample matrices, such as sausages and hot dogs. However, the ddPCR output remained in linear correlation with target concentration. The matrix effect observed here may be explained by different components in the samples such as fats, presence of microbial population and also by the heterogeneous nature of the food and feed samples. Systematic matrix effect can be corrected by creating calibration curves or normalizing the data to the internal control when the IC is included in the sample before DNA extraction. Validation must be conducted in order to accurately quantify a target species in a matrix of different nature.
It is desirable to use “reference materials” in a validation study that are prepared under conditions close to production of the food or feed under testing. Meat materials were prepared in-house in this study due to lack of available reference materials representative of typical industrial meat and feed processing. Variations are expected from fortified materials prepared in different labs. Certified reference materials are needed to ascertain comparable results among different methods or different labs. These reference materials will help overcome limitations from using different food and feed materials in validation studies and extend the ability of ddPCR for use as a reliable quantitative technique, facilitating the establishment of consensus methods for food and feed testing. They will also ensure equivalency of results and support laboratory proficiency testing needs.
Food and feed production processes can cause severe DNA degradation due to heat or physical damage, resulting in several fold reduction in ddPCR outputs as compared to those from fresh tissues, as observed in this study. The effect of DNA degradation on ddPCR results was also observed with procedures that may be used in sample analysis, such as heating and sonication of a sample and restriction enzyme digestion of a DNA template. We found that heat treatment of a beef tissue sample by autoclaving for 15 min resulted in over 3 fold reduction in the ddPCR readings. Prolonged (e.g. 10 min) sonication of the template DNA resulted in underestimation (e.g. 4.5 fold reduction) of the target. Restriction enzyme digestion of genomic DNA templates has been recommended in the QX 200 experiment protocol for ddPCR in the manufacturer’s manual; however, Taq I digestion of the template in this study was found to significantly reduce ddPCR output although there was no Taq I restriction site within the amplicon. The prolonged incubation of the template during Taq I digestion (65°C for 1 hr) may have resulted in degradation within the amplicon. These findings emphasize the importance of minimising DNA degradation in analytical processes to ensure that the quantification numbers reflect the true nature of the samples. As shown, food processing can potentially result in underestimation of target species.
The ddPCR assays described in this report met the accepted performance criteria of the ddPCR platform. The advantages of the methods include their high sensitivity, and ability to reliably quantify low concentration of DNA in a high background DNA without using standard curves. The internal control developed in this study can be used to monitor the PCR procedures and is recommended to be included in ddPCR assays to assess recovery and correct matrix effect. The methods can be used for quantitative analysis of bovine, porcine, chicken and turkey DNA in food and feed in validated matrices, particularly for products that are deeply processed or degraded and in which trace amount of foreign meat species is not tolerated. Standard reference material should be developed in collaboration with industry to mirror common production processes. The ddPCR methods can be implemented in routine testing to identify food fraud and to monitor the prohibited animal species in feed chain with enhanced sensitivity, accuracy and precision without reliance on standard curves.
(PDF)
(PDF)
IC was tested 16 times. The RSD% was 5.53.
(TIFF)
(A) ddPCR results for fortified beef in poultry meal at 0, 0.005, 0.01, 0.05, 0.1, 0.5, 1.0, 3.0, 5.0, and 10.0% (wt/wt). The curve exhibited a plateau when beef content was over 3.0%. (B-C) are subset data of (A) where (B) shows beef in poultry meal at 0, 0.005, 0.01, 0.05, 0.1, 0.5, 1.0, and 3.0% (wt/wt). After removing the 5.0 and 10.0% data points, the curve was linear. The upper limit was thus determined to be 3.0%. (C) shows beef in poultry meal at 0, 0.005, 0.01, 0.05, and 0.1% (wt/wt). The assay was unable to differentiate among 0, 0.005, and 0.01% of beef. The lower limit was thus determined to be 0.05%. The linear relationship was established between 0.05 and 3.0% beef as shown in
(TIFF)
ddPCR results were obtained from testing fortified heat-processed beef in poultry meal without normalization to IC (A) and after normalization to IC (B).
(TIFF)