High resolution melting is a useful tool to detect animal species sources of various milk types

Abstract

Accurate identification of animal species sources in milk have become quite important due to adulteration of high-priced milk types in the dairy industry. To date, milk identification methods have mainly depended on biochemical properties or physical properties detected by spectroscopic methods. The current study aimed to develop an easy to use and sensitive DNA-based High resolution melting (HRM) assay to identify animal species and detect cross-adulteration of water buffalo, bovine, goat, sheep, camel and donkey milks. HRM compatible designed primer set, targeted mitochondrial region, successfully amplified the specific targeted region for six animal species DNA and showed a high degree of specificity based on nucleotide variations. Capillary electrophoresis analysis validated the specific amplicons and determined the amplicon lengths as 114 bp for bovine, goat, sheep, and camel, 115 bp for water buffalo, and 121 bp for donkey. HRM analysis showed a clear discrimination for water buffalo-bovine, camel-bovine and donkey-bovine adulteration down to 0.5%, and goat-sheep adulteration down to 1% in the milk admixtures. The efficacy of the method was also confirmed by its standard curve with a very high correlation coefficient In conclusion, the designed HRM assay allows for the rapid, sensitive and cost-effective authentication of milk and dairy products.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13197-023-05705-3.

Introduction

The nutritional value of milk and dairy products (i.e. protein quality, fats, carbohydrates, vitamins and minerals) makes these products beneficial to human health (Paul et al. 2020). However, depending on seasonal changes, some milk types (e.g. camel, goat and water buffalo) are produced in lower amounts. Consisting of variable compositions, these milk types are economically more valuable than other widely available milk types that are produced yearlong. Therefore, in Turkey and many other countries worldwide, valuable milk types are often subjected to adulteration or imitation using less valuable milk sources (Ertaş and Topal 2009).

Milk is largely obtained from bovine animals, sheep, goats, water buffalo and camels (Boukria et al. 2020). However, due to better accessibility and affordability, bovine milk remains the most prevalent of sources. It also has similar protein properties to human milk and other animal species and, as such, manufacturers have been known to substitute bovine milk for sheep's milk (Ferreira and Caçote 2003). Similarly, as a result of its reduced levels of production, goat milk with its high digestibility, low allergenic properties and bio-functional compounds (e.g. polyunsaturated fatty acids and serum proteins) have been adulterated with bovine milk (Goswami et al. 2017). Water buffalo milk with its high amount of dry matter and fat content can be processed into products such as cheese, butter, ice cream and yoghurt (Becskei et al. 2020). Specifically, products such as Afyon cream and locally produced type of yoghurt are produced from mixtures of water buffalo milk and goat or bovine milk. Yet, since water buffalo milk is produced in lower quantities it is also more expensive and has consequently been subjected to replacement with bovine milk (Bezerra et al. 2012; Vijayendra and Gupta 2014; Kara and Demirel 2016). Lastly, camel milk, a good nutrition source for humans, has several food applications (e.g. butter, cheese and milk tea) as well as cosmetic applications (e.g. face masks and soaps) (Zhao et al. 2015). It has been recommended that children with food allergies may consume donkey milk, whereas adults may enjoy the latter as fresh or fermented products (Martini et al. 2018; Derdak et al. 2020).

Due to the many advantages of these different types of milk, the adulteration of milk and dairy products have become a common issue—resulting in food sector fraudulence and economic problems for the consumer. For example, because of the allergenic effects of bovine milk, the use of alternative kinds of milk such as goat, donkey and mare milk has become popular (Chiofalo et al. 2001). Thus, the replacement of or mixing with another type of milk (usually bovine) may pose serious health risks where the consumption of trace amounts of bovine milk may cause allergic reactions in people with sensitivity to bovine milk (Agrimonti et al. 2015). Furthermore, true lactose intolerance (a potential cause of bovine milk intolerance) can induce inflammation (Pal et al. 2015).

According to European Union laws, the origin of the milk (in terms of production and consumption) must be stated on the label of dairy products (EU 2001). Furthermore, milk ratios in cheeses produced from mixed milk sources should be indicated on product labels (in European Union standards) (Dietary Guidelines Advisory Committee 2015). By revealing milk origin (i.e. animal species) in dairy products, consumers could thus be enlightened in terms of fair trade and health risks (Genis et al. 2019) and crucial information will be provided for the traceability of foods and control of adulteration (Dalmasso et al. 2011).

Several analytical methods (both immunological and non-immunological) can be used to determine milk types. For example, multiplex polymerase chain reaction (PCR) has been used to determine goat, sheep and bovine contributions to cheese (Bottero et al. 2002); immunochromatography to detect the presence of bovine milk in sheep cheese (Colak et al. 2006); synchronous fluorescent spectroscopy (SFS) to determine bovine, water buffalo, goat and sheep milk types in fermented milk products (Genis et al. 2019); polyacrylamide gel electrophoresis to determine the presence of bovine milk in sheep cheese (Şimşek and Tümer 2008); near-infrared (NIR) spectroscopy to identify sheep, goat and bovine milk (Durna et al. 2016) and ELISA to detect the presence of bovine milk in sheep cheese (Altun and Durmaz 2017) and goat's milk (Song et al. 2011).

Since chemical and physical properties of milk may vary depending on numerous factors (e.g. animal nutrition, breed, lactation period, individual characteristics, climatic conditions, health and age), methodologies that rely on such properties may be questionable. Therefore, DNA-based methodologies are considered to be more reliable (i.e. using stable molecules unaffected by such factors). Yet, DNA-based methods such as multiplex PCRs may have other disadvantages such as self-inhibition among different sets of primers, low amplification efficiency and no identical efficiency on different templates (Traugott et al. 2013).

High resolution melting (HRM), a post-PCR method depending on melting kinetics of double-stranded DNA (dsDNA), can support the easy determination of DNA variants (without sequencing) (Druml and Cichna-Markl 2014). This closed-tube method with minimum sample handling reduces contamination risks and makes use of specific primers to amplify a target DNA region (ensuring that contamination by other organic and inorganic substances does not affect the results of the analysis). Moreover, since PCR allows researchers to amplify DNA (from a few copies to millions of copies), HRM analysis can be performed with very little starting materials. Although there are many studies that used HRM as detection method of the adulteration in the literature, there are only few studies targets milk adulteration. HRM method was successfully applied to detect bovine milk in pure water buffalo mozzarella and other buffalo dairy products (Sakaridis et al. 2013), testing the authentication of Greek protected designed origin Feta cheese (Ganopoulos et al. 2013) before.

Therefore, since HRM analyses do not require large numbers of samples, specialised background knowledge or complicated software, the current study aimed to develop a novel sensitive, accurate, reliable and easy-to-apply HRM method to authenticate bovine, water buffalo, sheep, goat, camel, and donkey milk and to detect the most common cross-adulteration events at very low rates.

Materials and methods

Sampling, preparing the admixtures, and DNA extraction

Raw milks of water buffalo, bovine, goat, sheep, camel and donkey directly obtained from local farms in Turkey (under aseptic conditions), were frozen using liquid nitrogen and stored at − 80 °C (until DNA extractions could be performed). Raw milk was mixed with an adulterant at ratios of 199:1, 99:1 and 9:1 (as genuine milk:adulterant; water buffalo:bovine, goat:sheep, camel:bovine, and donkey:bovine) to a final volume of 20 mL. Genuine milks were coded as water buffalo (W), bovine (B), Goat (G), sheep (S), camel (C) and donkey (D), and the admixture combinations were coded as water buffalo-bovine (WB), goat-sheep (GS), camel-bovine (CB), and donkey-bovine (DB).

To extract total DNA, a previously described protocol (Pokorska et al. 2016) was modified. Starting material was decreased to 2 mL and, before the Buffer 2 step, 20 μL of proteinase K solution (Qiagen, USA) was supplemented at 56 °C for 30 min to prevent protein contamination and performance inhibition on HRM. DNA concentration and integrity had been assessed with Qubit 2.0 using a dsDNA BR assay kit (Invitrogen, USA) and 2% agarose gel electrophoresis analysis. The extracted DNA solution was stored at − 20 °C.

DNA mining for designing of the primers

The complete mitochondrial genome of water buffalo (Bubalus bubalis Kerr, 1792), bovine (Bos taurus Linnaeus, 1758), goat (Capra hircus Linnaeus, 1758), sheep (Ovis aries Linnaeus, 1758), camel (Camelus dromedialus Linnaeus, 1758), and donkey (Equus asinus Linnaeus, 1758) had been retrieved from the National Center for Biotechnology Information (NCBI) GenBank database with respective accession numbers of CM034296, GU947021, MZ073671, NC_001941, NC_009849 and MZ073671. Sequences were imported and aligned using Geneious R8 software and the Geneious Alignment Tool (Kearse et al. 2012). Suitable primer binding regions and amplicons (with various sites for HRM) were determined (considering variable positions, GC content (~ 50%), expected amplicon size (100–130 bp) and melting temperature (~ 60 °C). Designed primer sets were confirmed for their specificity in-silico using the Primer-BLAST tool on NCBI and capillary electrophoresis (CE) following end-point PCR. Based on these tests, the 2579_F and 2700_R primer set (targeting 16S sequence for HRM analysis) was chosen out of six designed candidate primer sets (targeting various gene regions on the mitochondrion).

HRM-PCR amplification and analysis the results

The HRM reaction was performed on a Rotor-Gene-Q 5 Plex HRM instrument (Qiagen, USA) using a 72-well carousel and 0.1 mL assay tubes. The reaction mixture included 10 ng template DNA, 10 μmol L⁻¹ forward (2579_F: 5′- GACGAGAAGACCCTATGGAGC-3ʹ) and reverse (2700_R: 5ʹ-CTCCGAGGTCACCCCAACC-3ʹ) primers (0.5 μL each), 5 μL Luminaris Colour HRM Mastermix (Thermo Scientific, USA), and nuclease-free water to a final volume of 10 μL. Cycling conditions were as follows: an initial denaturation step at 95 °C for 10 min followed by 40 cycles of denaturation at 95 °C for 10 s, annealing at 60 °C for 30 s and elongation at 72 °C for 30 s. Fluorescence data were acquired immediately after each elongation step. Steps for heteroduplex formation (i.e. 95 °C for 30 s and 50 °C for 30 s) was added to the end of the reaction. The HRM step was performed immediately after amplification in increments of 0.1 °C s⁻¹ hold time from 75 to 95 °C and fluorescence data was acquired continuously. All reactions were performed in triplicate and a no template control (NTC) was included for each reaction.

The HRM data was analysed using Rotor-Gene-Q 2.3.5 software (Qiagen, USA) and difference plots were constructed (with genuine milk serving as positive controls). In addition, with a confidence threshold set at 90% for reliable results, genotyping confidence percentages (GCPs) were calculated for each sample.

Primers’ validation

End-point PCR and subsequent CE analysis were used to validate the specificity of the designed primers and amplicon lengths. PCR reactions included 2.5 μL of 2X Reaction Buffer (Thermo Scientific, USA), 0.5 μL of 10 mM dNTPs, 0.5 μL of 10 μM primers (each), 1 U Taq DNA polymerase (Thermo Scientific, USA), 2 μL of 25 mM Mg⁺², 10 ng template DNA and nuclease-free water to a final volume of 25 μL. Amplification was performed on a SimpliAmp Thermal Cycler (Applied Biosystems, USA) with the following conditions: 95 °C for 3 min (first denaturation), followed by 35 cycles of 95 °C for 30 s (denaturation), 60 °C for 30 s (annealing), 72 °C for 1 min (elongation) and finally 72 °C for 10 min (final extension step). The CE analysis was performed on a Qiaxcel Advanced instrument (Qiagen, USA) using a Qiaxcel DNA High-Resolution Kit (Qiagen, USA). Software settings were as follows: process profile: default High Res v2.0; Method: 0L800; a size marker run by side the samples: Gene Ruler 100 bp Plus, (Thermo Scientific, USA); alignment marker: QX15bp-3 kb (Qiagen, USA). Results were visualised and analysed using ScreenGel 1.6 software (Qiagen, USA).

Positive control amplicons were sequenced to validate the HRM results using the identical primers 2579_F and 2700_R. Sanger sequencing was performed by Macrogen Inc. (The Netherlands) using the 3500 Genetic Analyzer (Applied Biosystems, USA) and sequence files were imported to the Geneious R8 software and validated for sequencing quality. Raw reads were trimmed and assembled to generate consensus sequences that were subsequently aligned. Nucleotide variation counts and percentages were calculated.

Statistical analyses

Descriptive statistical analyses were performed for the quantification cycle (C_q), melting point (T_m) and GCPs. Values presented as mean ± standard deviation (SD) were based on triplicate data. One-way ANOVA and post hoc (LSD) analyses were used to compare mean values (at a significance level of P < 0.0001). Linear regression was also performed to calculate the correlation coefficient of the GCPs. Statistical analyses were performed using XLSTAT Basic 2021 software (Addinsoft).

Results

CE validated the primers’ specificity

We obtained DNA concentrations of 5.1–6.1 ng µL⁻¹ allowed for subsequent HRM analyses. Positive control amplicons were assessed using CE analysis (Fig. 1) and revealed that designed primers indeed amplified a specific region and that amplicon lengths of samples ranged between 114 and 121 bp (Table 1). The bovine (B), camel (C), goat (G) and sheep (S) samples produced 114 bp amplicon lengths, while the water buffalo (W) sample produced a 115 bp amplicon and the donkey (D) sample produced a 121 bp amplicon. All amplicons were inside the expected range and thus optimal for HRM.

Fig. 1.

Capillary electropherogram of the 16S target region amplicons from water buffalo (W), bovine (B), goat (G), sheep (S), camel (C), donkey (D) positive controls. The profiles are colour coded, and amplicon sizes represented on the legends

Table 1.

Details about the HRM amplicons amplified by the 2579_F and 2700_R primers

Sample	Amplicon size (bp)	Variation in bp (%)	GC Content (%)
Bovine	114	11 (9.65)	43.0
Water Buffalo	115	14 (12.17)	46.1
Goat	114	13 (11.40)	42.1
Sheep	114	8 (7.02)	43.0
Donkey	121	25 (20.66)	41.3
Camel	114	12 (10.53)	40.4

HRM discriminated the adulterant

Based on the HRM-PCR results, the T_m value differentiated between the positive controls of water buffalo-bovine and goat-sheep milk samples but failed to do so for camel-bovine and donkey-bovine milk samples (P < 0.0001). In addition, the T_m value successfully detected adulteration of goat milk (with sheep milk) at 9:1, 99:1 and 199:1 ratio. As calculated based on melting profiles, GCPs can be used to genotype and discriminate between samples and the selected standard (with values lower than 90 indicating successful discrimination whilst higher percentages indicated no distinction). For the current study, the GCPs successfully validated the positive controls and those calculated for genuine milk in each admixture also detected adulteration. However, the 93.17% GCP failed to detect adulteration of the goat:sheep 199:1 (0.5%) admixture (GS-200). The most discriminated admixtures (< 10.69%) belonged to donkey-bovine milk samples (Table 2).

Table 2.

Descriptive statistics of HRM analysis

Abbreviation	Sample name	Cq ± SD	Tm ± SD	GCP	R2
Water Buffalo–Bovine admixtures
W	Water Buffalo	26.66 ± 0.02b	81.34 ± 0.01a	100	0.9665
B	Bovine	27.88 ± 0.02a	79.50 ± 0.12b	100
WB-10	Water Buffalo:Bovine (9:1)	24.32 ± 0.14d	80.01 ± 0.02ab	38.89 ± 0.53
WB-100	Water Buffalo:Bovine (99:1)	25.05 ± 0.17c	80.06 ± 0.09ab	47.91 ± 0.88
WB-200	Water Buffalo:Bovine (199:1)	25.02 ± 0.21c	79.94 ± 0.04ab	85.11 ± 0.94
Goat–Sheep admixtures
G	Goat	23.25 ± 0.03a	79.40 ± 0.01d	100	0.9836
S	Sheep	22.63 ± 0.04b	79.18 ± 0.01e	100
GS-10	Goat:Sheep (9:1)	20.83 ± 0.08d	79.90 ± 0.04a	8.66 ± 4.21
GS-100	Goat:Sheep (99:1)	21.06 ± 0.11c	79.76 ± 0.07b	29.75 ± 11.20
GS-200	Goat:Sheep (199:1)	18.50 ± 0.11 cd	79.54 ± 0.02c	93.17 ± 3.52
Camel–Bovine admixtures
C	Camel	18.08 ± 0.03c	76.84 ± 0.01a	100	0.9994
B	Bovine	27.88 ± 0.02a	79.50 ± 0.12a	100
CB-10	Camel:Bovine (9:1)	21.14 ± 0.09b	76.16 ± 0.00a	0.07 ± 0.04
CB-100	Camel:Bovine (99:1)	17.32 ± 0.43d	78.02 ± 0.02a	34.12 ± 0.04
CB-200	Camel:Bovine (199:1)	17.19 ± 0.01d	76.90 ± 0.00a	82.41 ± 2.44
Donkey–Bovine admixtures
D	Donkey	15.05 ± 0.08e	76.10 ± 0.01a	100	0.9986
B	Bovine	27.88 ± 0.02a	79.50 ± 0.12a	100
DB-10	Donkey:Bovine (9:1)	18.86 ± 0.28c	76.18 ± 0.08a	0.47 ± 0.98
DB-100	Donkey:Bovine (99:1)	19.59 ± 0.29b	77.06 ± 0.58a	2.31 ± 0.18
DB-200	Donkey:Bovine (199:1)	18.50 ± 0.11d	79.90 ± 0.02a	10.69 ± 3.69

Cq: cycle threshold; T_m: melting temperature; GCP: genotyping confidence percentage

Superscript letters: The mean difference is significant at P < 0.0001 level

R²: Correlation Coefficient value of linear regression

The melting profile differential plots provided good visualisation of discrimination (Fig. 2). These plots, using genuine milk as a reference genotype, clearly discriminated genuine milk from the adulterant. The plots also visualised the discrimination of each admixture (except for GS-200 which was not discriminated by GCP). Whilst camel:bovine 199:1 admixture (CB-200) and the genuine camel milk plots had been drawn very close to one another, the distinction was clear with a GCP calculation of 82.41%. Similarly, the water buffalo:bovine 199:1 admixture (DB-200) and genuine water buffalo milk plots had been close but showed a clear distinction with a GCP of 85.11%. The efficacy of the method was confirmed by its standard curve with a very high correlation coefficient (Table 2). We successfully performed all the analysis under two and half hour including the steps; DNA extraction 30 m, setting up HRM reaction 20 m, HRM run in the instrument 80 m, and analysing of data 10 m. The total consumed time was less than DNA sequencing-based methods (approximately 3 h).

Fig. 2.

Difference graphs obtained by HRM analysis of the camel-bovine (A), donkey-bovine (B), goat-sheep (C) and water buffalo-bovine (D) samples. The profiles are colour coded as represented in legends (geniunie milk:adulterant)

DNA sequencing validated the HRM results

HRM results were validated by analysing target amplicon nucleotide sequences of bovine, water buffalo, goat, sheep, donkey and camel positive control samples. These alignments of target regions visualised nucleotide variations and insertions/deletions (indels) (providing discrimination resolution for the HRM method) (Fig. 3; Supplementary Data 1). The shortest amplicons of 114 bp for the bovine, goat, sheep and camel samples differed from the largest amplicon of 121 bp (donkey sample) due to indels. In addition, the amplified region had been rich in nucleotide variation with the sheep sample amplicon showing the least variability of eight variations (7.02%) (representing a good variation rate for HRM analysis). In contrast, the most variable amplicon with 25 variable sites (20.66%) belonged to the donkey sample. The GC content of amplicons varied from 40.4 to 46.1% (Table 1).

Fig. 3.

Multiple sequence alignment of PCR amplicons obtained from water buffalo, bovine, goat, sheep, camel, donkey species using the primers 2579_F and 2700_R. Primer binding regions are indicated with arrows, and variable sites highlighted with colours

Discussion

A reliable and sensitive method was necessary for authentication of milks

Due to economic issues, adulteration of raw milk and dairy products have become common. Yet, due to similar physical properties, it is almost impossible for consumers to detect when high-value raw milk with limited production (e.g. water buffalo milk) have been mixed with a low-value adulterant (e.g. bovine milk). Nevertheless, this adulteration could pose a serious health risk to sensitive consumers with allergic reactions to certain milk proteins (López-Calleja et al. 2005) and, as such, the authentication of milk and dairy products has become an important issue for consumers.

Whilst several methods (e.g. electrophoretic, chromatographic and spectroscopic techniques) have been used to identify milk origins, these methods often have certain drawbacks regarding detection sensitivity and reliability (because of targeting protein structures). For instance, the official detection method for milk authenticity suggested by the European Union is that of γ-casein isoelectric focusing (Commision Regulation 2001). Yet, since heat application degrades proteins, some researchers have voiced their concerns about the reliability of this method when used on heat-treated milk (Bottero et al. 2002). Moreover, this method does not allow for adulteration detection in fermented dairy products (e.g. yoghurt and cheese) due to the target protein character change. Finally, at low ratio adulterations, the sensitivity of this method is questionable.

Genis et al. (2019) recently described a new spectroscopic method that allows for adulteration detection of water buffalo-cow yoghurt (0.46%) and ewe-cow cheese (1.47%). However, since it is not a DNA-based method, the ‘universality’ of this method for different animal breeds remains questionable. Furthermore, by lacking a targeted approach (such as for PCRs using specific primers), bacterial, viral or non-organic contaminants may affect the results. It is also known that milk components may vary because of several factors (including animal nutrition, breed, lactation period, individual characteristics, climatic conditions, health and age) (Haygert-Velho et al. 2018; Paszczyk et al. 2022) which may ‘skew authentication tests based on physical and chemical milk characteristics.

The need for a more reliable and sensitive method to detect low ratio adulterations in harsh conditioned milk and dairy products thus arose and using DNA (a stable molecule even in harsh conditions such as heat treatments and industrial processes) may be the answer. Therefore, the current study aimed to develop a simple, rapid, cost-effective and closed-tube identification method for discriminating bovine, water buffalo, goat, sheep, donkey and camel milk based on DNA analysis.

A Novel specific primer set was designed for HRM discrimination of the adulteration

Previously described water buffalo-specific primers, targeting the 12S rRNA gene (López-Calleja et al. 2005) and the _D-loop mitochondrial genes (Pegels et al. 2011), were successfully applied for the detection of bovine milk adulterant in water buffalo products (Sakaridis et al. 2013). However, these primers and their target regions were not suitable for the selected animals of this study (failing to bind goat, sheep, donkey and camel DNA sequences in silico tests (data not shown). The universal DNA barcode region for animals (i.e. cytochrome c oxidase subunit I (CO1) have been successfully used to discriminate between species from diverse animal groups since 2003 (Hebert et al. 2003). Unfortunately, this 600–800 bp region does not meet the optimal size range necessary for HRM analysis where the optimal amplicon length should be no longer than 300 bp (Druml and Cichna-Markl 2014). As such, we designed and tested novel common primers (targeting conserved regions) in varied sites for species discrimination between water buffalo, bovine, goat, sheep, camel and donkey milk samples for HRM analysis. Both CE and sequencing results indicated that the novel primer set of 2579_F and 2700_R successfully amplified the intended region. The annealing temperature (~ 60 °C), fragment size (114–121 bp) and nucleotide variation among the different species were also suited for HRM analyses (Table 1). Sequencing results also revealed differences in amplicon GC content among the species (except for bovine-sheep).

Detection of cross-adulteration by HRM

Successfully implementation of the post-PCR HRM assay allowed for the identification of DNA variants without sequencing (Druml and Cichna-Markl 2014). The guanine-cytosine (GC) content of a DNA sequence specifies its melting temperature (T_m) and the denaturation kinetics of its double-stranded DNA (having three hydrogen bonds instead of only two as is the case for adenine–thymine (AT)). This difference can be visualised via HRM analysis and thereby provides discrimination between DNA variants. This discrimination is also related to amplicon length, although the reliability of HRM analyses strictly depends on PCR product specificity. Designing novel HRM methods thus require the sequencing of amplicons (van der Stoep et al. 2009).

In the current study, HRM results showed that the T_m values for milk from goats vs. milk from sheep were significantly different (P < 0.0001) in the goat-sheep milk samples. Contrastingly, although there were length and nucleotide variations among the samples, there were no significant differences in T_m values for the water buffalo-bovine, camel-bovine or donkey-bovine samples. As such, T_m values cannot serve as the only parameter in adulteration detection HRM assays. Fortunately, melting kinetics of denaturing DNA also represents an important parameter in the drawing of normalised and differential plots (Wittwer et al. 2003). Here, HRM results discriminated the adulterants both visually via differential plots and quantitatively via GCPs. To establish HRM assay sensitivity, adulterants were assessed down to a 0.5% (199:1) ratio. In doing so, adulteration had been successfully detected for water buffalo-bovine (WB), camel-bovine (CB) and donkey-bovine (DB) adulteration samples. However, detection of goat-sheep 199:1 (GS-200) adulteration failed (with a GCP value of 93.17%), which may have been due to similar nucleotide variations (Fig. 3) and identical amplicon sizes (114 bp).

The R² values for each adulteration scenario suggested a high correlation of GCPs values with adulteration levels (Table 2). In using this method, adulteration (genuine milk combinations with adulterant milk) could be detected down to 199:1 (0.5%) for WB, CB and DB samples and 99:1 (1%) for GS samples.

Furthermore, validation of HRM assay reliability via sequencing of amplicons (i.e. alignments) revealed nucleotide variations in bovine, water buffalo, goat, sheep, donkey and camel sequences that supported HRM results. These variations were visualised with HRM analyses and discriminated the adulteration.

Conclusion

In this study, an HRM method, coupled with a novel primer set targeting 16S rRNA, was developed and tested for the detection of cross-adulteration in bovine, water buffalo, goat, sheep, donkey and camel milk. This method supported both the detection and quantification of adulteration. Specifically, milk adulteration between water buffalo and bovine, camel and bovine as well as donkey and bovine samples could be detected down to 0.5%, while goat and sheep adulteration could be detected down to 1%. HRM results were validated with amplicon sequence analysis and this novel method proved to have a very high correlation coefficient. We, therefore, suggest that HRM analysis with 2579_F and 2700_R primers provides a rapid, sensitive and accurate method to discriminate and authenticate bovine, water buffalo, goat, sheep, donkey and camel milk. As a DNA targeted technique, it might easily be applied for dairy products, meat products and other products obtained from the animal species used in this study.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

HRM

High resolution melting

PCR

Polymerase chain reaction

NIR

Near infrared

dsDNA

Double-stranded DNA

SFS

Synchronous fluorescent spectroscopy

W

Water buffalo

B

Bovine

G

Goat

S

Sheep

C

Camel

D

Donkey

WB

Water buffalo–Bovine

GS

Goat–Sheep

CB

Camel–Bovine

DB

Donkey–Bovine

CE

Capillary electrophoresis

GCPs

Genotyping confidence percentages

C_q

Quantification cycle

C_t

Melting point

SD

Standard deviation

DNA

Deoxyribonucleic acid

bp

Base pair

Author’s contribution

KH: Investigation, Formal Analysis, writing original draft; KH&MB: Conceptualization, methodology, validation, writing—review & edit.

Funding

There is no funding available for this publication.

Data availability

All data generated or analysed during this study are included in this published article (and its supplementary information files). The nucleotide alignment data was given as supplementary file.

Code availability

Not applicable.

Declarations

Conflict of interest

The authors have no conflicts of interest to declare.

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.