1. Introduction
Mussels from
Mytilus genus represented 20% of the worldwide mollusks production in 2018 [
1].
Mytilus genus is one of the most cultivated and marketed bivalve, widely appreciated as a tasty and nutritious source of protein. Mussel aquaculture is a relevant economic activity for many coastal communities [
2,
3]. The related commercial species,
Mytilus edulis (Linnaeus, 1758),
Mytilus galloprovincialis (Lamarck, 1819),
Mytilus chilensis (Hupé, 1854) and
Mytilus trossulus (Gould, 1850) are taxonomically recognized in the World Register of Marine Species [
4] and the Integrated Taxonomic Information System [
5] together with other
Mytilus spp. Nowadays, the European regulation on the common organization of the markets in fishery and aquaculture products [
6] requests the declaration of the commercial designation of the species and its scientific name in the label. Each participant country has its official list of commercial designations and scientific names for fishery and aquaculture products (
https://ec.europa.eu/fisheries/cfp/market/consumer-information/names_en, accessed on 21 May 2021). This specific traceability requirement is aimed at confirming the authenticity of the products. Species substitution can result in an inexpensive product being labelled as high-priced. It can also affect food safety via unnoticed consumption of allergens due to undeclared species [
7,
8]. Seafood mislabeling is well documented throughout history [
9,
10]; it impacts not only food authenticity [
11] but also allows the trade in the markets of endangered species or products from illegal, unreported and unregulated (IUU) fisheries, threatening wildlife [
12], hampering conservation and negatively affecting consumers decisions [
13]. Nowadays, the breadth and depth of mislabeling are coming into sharper focus, thanks to DNA-based species verification methods [
14].
DNA testing is currently reported as the regulatory tool of choice for seafood species identification (SI) [
15,
16,
17]. Several DNA analysis techniques rely on Polymerase Chain Reaction (PCR) [
18]; however, the informative potential of the targeted genomic regions should be rigorously tested before its adoption in a standardized system. Recently, PCR-based methods have been widely used in seafood authentication [
19] and harvesting location detection [
20].
For
Mytilus SI in either fresh or processed products, different molecular markers and techniques have been developed [
21,
22,
23,
24]. However, the most common DNA marker used for this purpose targets the polyphenolic adhesive protein gene, highly conserved within, but polymorphic among species [
22,
24,
25,
26,
27,
28]. Targeting this region, Jilberto et al. [
29] developed a high-resolution melting (HRM) assay that allowed them to identify
M. chilensis, M. edulis and
M. galloprovincialis and their hybrids. This technique is simple and affordable to most molecular analytical laboratories.
SI based in a single locus (“mono-locus”) is relatively easy to perform, but this approach can produce conflicting results among markers and/or when individuals from hybrid zones are analyzed. Single nucleotide polymorphism (SNP) multi-locus panels have allowed for identifying
M. edulis, M. galloprovincialis, M. trossulus [
30,
31],
M. chilensis,
M. planulatus and
M. platensis [
32,
33,
34]. However, genotyping a high number of SNPs may not be practical in routine analysis; therefore, it is attractive to develop reduced panels selecting highly informative SNPs for SI. The identification of the most informative loci can be performed by different criteria such as
FST outliers or minor allele frequency (MAF) [
35,
36].
Before the application of an analytical method in food analysis, its performance should be evaluated through compliance with quality criteria according to international guidelines [
37,
38,
39]. Moreover, in the case of laboratories involved in import and export food testing or law enforcement, compliance with ISO/IEC Standard 17025 requirements is necessary [
40]. The fit for purpose and the performance of a qualitative real-time PCR method can be first validated in-house, assessing the specificity, sensitivity, repeatability, reproducibility and practicability. Moreover, the transference of the method to a second laboratory can be performed [
38].
To support the confidence of consumers, the food industry, business operators and regulators about seafood authenticity, and to avoid the mislabeling of mussels, it is essential to have available an affordable method whose results are internationally recognized.
In this study, we aim to develop a multi-locus SNPs method based on the PCR-HRM analysis, for the identification of species of the Mytilus genus (M. chilensis, M. edulis, M. galloprovincialis and M. trossulus), and to assess its fitness for purpose by an in-house validation process.
4. Discussion
We developed a multi-locus assay for species identification in the Mytilus genus based on the 10 most informative SNPs from 123 analyzed. Using the multi-locus genotypes, the four target species, Chilean mussel (M. chilensis), Blue mussel (M. edulis), Mediterranean mussel (M. galloprovincialis) and the Bay mussel (M. trossulus), can be confidently identified. This method allowed us to discard the presence of other mussels not belonging to the Mytilus genus but commonly found in seafood products, such as A. atra and C. chorus.
4.1. Multi-Locus PCR-HRM Method Development
The FST outlier and MAFMAX criteria were useful to select the most informative SNPs for SI that are intended to reveal variations among but not within species. The reduced 10 SNP panel was constructed including the SNPs selected by each of these two criteria.
The MAF
MAX criteria maximize the difference in allelic frequencies among species, selecting locus showing a maximum allelic frequency for one species, but a minimum for another (MAX
MAF) [
36]. However, to use this approach it is necessary to sample several locations per species and with enough number of individuals, to rule out that the observed allele frequencies are population specific. This criterion is similar to the one used by Wilson et al. [
30], these authors choose the more informative SNP using the “loading values” calculated after a discriminant analysis of principal components. Higher loading values reflects SNPs that capture more variance in the allele frequencies and therefore greater differences among species.
A locus shows high
FST values when two or more groups are homozygous for contrasting alleles. Therefore, the
FST outlier criterion has been used to select SNPs for geographic assignment because differences in their allele frequencies are probably a product of local adaptation [
12,
63]. In the
Mytilus genus, species are restricted to specific geographic areas, especially the native species from the southern hemisphere, therefore,
FST outlier SNPs are expected to be informative to differentiate them.
Wilkinson et al. [
64] have shown a low correlation between both criteria selecting informative loci, a fact corroborated by our study. Only two loci (PAPM and L7) were selected by MAF
MAX and
FST outlier at the same time. Therefore, it is important to considerer both criteria together when selecting the most informative loci for SI.
Searching for loci that can be technically genotyped, each of the potentially most informative SNPs must be revised to determine the feasibility to design primers. Moreover, to obtain clearly distinguishable melting curves, amplicons containing more than one polymorphism should be discarded. Other technical aspects of HRM analysis must be also considered [
65]. However, optimal conditions scarcely happened, and to design a practical assay it is necessary to be flexible. For example, we included one SNP A/T because was informative to separate between
M. edulis and
M. galloprovincialis although these kinds of SNPs are not recommended as they are hard to genotype. We also removed some C/T and G/A SNPs because they gave redundant information.
The performance of the reduced 10 SNP panel was similar to the obtained with 49 SNPs (Dataset A) and 90 SNPs (Dataset B). Assignment probability, sensitivity and specificity were maximum for all four species tested here, indicating that the method performs well, correctly identifying the species and excluding individuals that do not correspond to species false positive. To our knowledge, the only reduced multi-locus panel to perform SI in the three European species was developed by Wilson et al. (2018) [
30]. The performance of this 12 SNP panel, calculated from their raw data and excluding populations used for validation purposes, showed a maximum specificity for the three species. Sensitivity assigning
M. trossulus and
M. galloprovincialis was also a maximum; however, in
M. edulis, this value reaches only 0.96. The panel developed here can detect one species more with fewer SNPs (10 vs. 12) and with higher performance. As Wilson et al. (2018) [
30] have clearly shown, the most valuable aspect of multi-locus panels is the identification of putatively hybrid not detected by the mono-locus assays.
4.2. Fitness for Purpose Assessment of the Multi-Locus PCR-HRM Method
Before the commercial application of any analytical method, it must be validated following accepted quality criteria [
37,
38]. In laboratories involved in food testing, compliance with ISO/IEC Standard 17025 [
40] requirements is mandatory. However, formal validation studies of qualitative techniques that use the PCR-HRM technique to identify marine species are scarce in the current literature. In the case of qualitative PCR-based methods, their scope or applicability relies on the capacity to obtain enough DNA with proper integrity and purity, and also in the absence of PCR interferents. In our study, the CTAB method allowed us to extract enough quality DNA in all tested food matrices. However, every testing laboratory must validate its own DNA extraction method before using this multi-locus assay.
The canning process involves the application of high temperatures (≈121 °C) to the product and its packing media, hermetically sealed in an anaerobic environment [
66]. Is widely recognize that in highly processed food (canned or cooked), the DNA is fragmented below 300 bp, leading to extracting less and degraded DNA [
9,
67]. The HRM analysis use amplicons shorter than 300 bp [
11] and the sensitivity of detection is enhanced in smaller amplicons [
65]; therefore, it is ideal for genetic analysis of highly processed food. In our method, the size of the amplicons sizes were between 50 to 170 bp, appropriated to be successfully genotyped in processed food. However, the packing media may contain organic acids, ions, chelating agents and other ingredients that favor DNA fragmentation during the thermal process and/or interfere with the PCR reaction [
67,
68,
69,
70]. Although we obtained DNA of high quality from products packed in vinegar or sauces containing vinegar, amplification failed or the
Cq values were larger than the threshold defined to obtain genotypes. It has been demonstrated that using commercial silica-based columns along with chaotropic solutions allow the removal of some PCR inhibitors in comparison with non-commercial methods such as the phenol–chloroform [
71]. Our results pose the challenge to further testing to remove these interferents and/or using amplification enhancers [
72,
73], to expand the scope of our assay to include complex matrices.
The method was practical and easy to apply, and requires short training time. The HRM technology works in a closed tube, reducing human effort, errors and the risk of drag contamination [
74,
75]. In addition, it is amendable with small laboratories with low samples flow. The reagents cost are affordable (≈USD 23 per sample in Chile) and could be reduced in countries where it is not necessary to import reagents and consumables.
Although three primer pairs retrieve potential in silico amplifications during Primer Blast analysis, and two of them showed amplification during the in vitro assays in non-target species, the method was specific. The multi-locus approach means that a specimen is assigned to the species using the information of the 10 loci simultaneously. Therefore, the isolated amplification of one or two SNPs in the non-target species will not produce any species assignment. This panel avoid assigned errors derived from the analysis performed in a mono-locus way.
Although this method is qualitative, the DNA concentration that can be detected with 95% confidence (LOD
95% = 5 and 8 ng/μL) was 25–40% of the DNA concentration recommended in the SOP of analysis (20 ng/μL), showing adequate sensitivity for the intended use. The LOD of this qualitative method was appropriated on the base of the
Cq values ranges, all positive and specific amplification observed showed
Cq values below 29 [
38,
65]. The method was robust and stable in front of small variations in the eight common operational parameters tested in an orthogonal test as recommended by ISO 20813:2019 [
76]. Transferability to a second laboratory allowed the evaluation of the method under other different condition (operator, equipment, etc.). As the method is multi-locus, when one lab fails to amplify some loci, the SI of the sample cannot be obtained. Therefore, no discrepancies in species assignment arise.