Sensitivity of Loop-Mediated Isothermal Amplification in Comparison to Digital Droplet PCR for Identification of Yersinia pseudotuberculosis in Raw Goat Milk

TL;DR

This study compares two methods for detecting Yersinia pseudotuberculosis in raw goat milk, finding that LAMP is a fast and sensitive alternative to ddPCR.

Contribution

The first evaluation of LAMP sensitivity compared to ddPCR for detecting Y. pseudotuberculosis in raw goat milk.

Findings

LAMP with violet dye detected Y. pseudotuberculosis at 3.1 × 10⁴/mL and blue dye at 3.4 × 10³/mL.

ddPCR detected the pathogen at a lower limit of 2.0 × 10¹/mL.

LAMP shows potential for on-site detection of Y. pseudotuberculosis in raw milk.

Abstract

According to the EFSA Report on Zoonoses (2024), yersiniosis was classified as the fourth most commonly reported zoonosis in humans in 2023, with a 13.5% increase in yersiniosis infections compared to 2022. In 2024, the findings were consistent with the 2020–2023 trend. Isolation and identification of enteropathogenic Yersinia is difficult and time consuming, especially when examining food and environmental samples. Among them, Y. pseudoturbeculosis poses a challenge due to the lack of a single selective medium for all bioserotypes. Therefore, faster methods for the detection of Yersinia spp. need to be implemented into the praxis. Rapid identification of pathogens in food or at the time and location of the epidemiological outbreak (point-of-care testing) enables either prevention of the outbreak or early stage diagnosis and prompt decisions. The loop-mediated isothermal amplification (LAMP) is increasingly coming to scientists’ attention as a robust and rapid methodology for pathogen detection in laboratories with limited resources and equipment. The aim of current study is to evaluate, for the first time, the sensitivity of the LAMP protocol based on colorimetric detection in the visible spectrum in comparison with that of the digital droplet PCR (ddPCR). For this aim, a series of decimal logarithmic dilutions of the pathogen Y. pseudotuberculosis in artificially contaminated raw goat milk was used. One commercial LAMP kit with two different dyes (one dsDNA-binding and one Mg2+-sensitive) was compared to the sensitivity of the detection to ddPCR. The results obtained revealed a high sensitivity of the kit for detection of DNA isolated from artificially contaminated milk samples in the following range: visible detection based on visible color change—3.1 × 104 mL (violet dye) and 3.4 × 103/mL (blue dye); detection with gel electrophoresis—2.0 × 101/mL (violet dye) and 3.4 × 102/mL (blue dye). The enumeration of the DNA copies in the same samples was performed with ddPCR, with a detection limit of 2.0 × 101/mL. Our results indicate the potential and the possible applicability of the LAMP method for rapid and sensitive visual detection of Y. pseudotuberculosis in raw goat milk. The presented ddPCR protocol can be used for highly sensitive identification and enumeration of Y. pseudtuberculosis in raw goat milk. In conclusion, the conducted comparison is of importance for future implementation of LAMP protocols for on-field analysis near the sampling site and point-of-care or laboratory diagnostics of Y. pseudtuberculosis after the successful validation procedure of an appropriate LAMP protocol.

Figures

Full text

1. Introduction

Infections caused by the Gram-negative foodborne pathogen Yersinia pseudotuberculosis are common worldwide, although at a lower frequency than the other two human Yersinia pathogens—Y. enterocolitica and Y. pestis. The infection affects children more often than adults and according to some epidemiological studies, is invasive in more than 70% of patients, thereby requiring hospitalization [1]. During recent years, yersiniosis has been the third or fourth most common reported zoonosis in the European Union [2,3,4]. The surveillance evaluation shows a 16.3% increase (910 more cases) in yersiniosis infection in 2022 compared to 2021 and a 13.5% increase in 2023 compared to 2022. In 2024, the findings were consistent with 2020–2023, wherein Yersinia was more prevalent in non-ready-to-eat foods and fresh meat [4]. It is important to note that many laboratories do not perform routine tests for the diagnosis of yersiniosis, and specific tests are performed only on recommendation of the physicians in the respective healthcare facilities. Isolation of Y. pseudotuberculosis from food and environmental samples is relatively difficult and it is not possible to use a single selective medium that can be used for all bioserotypes of this species. Classical microbiological methods for the proof of Y. pseudotuberculosis, which involve enrichment of the sample for at least 48 h, followed by inoculation on selective nutrient media and cultivation for another 48 h, are laborious and time-consuming (ISO 18867:2015). In some cases, sample enrichment requires a longer incubation time—up to two weeks—to be able to isolate cultivable bacteria [5,6]. This was followed by proof of lineage and virulence genes in the suspect colonies by conventional PCR. Recent studies have used real-time PCR methods based on non-specific fluorescent dyes or specific double-labeled probes to screen for the presence of enteropathogenic yersinia in animal and food samples [7,8,9,10]. Samples from wild boars [11], fleas [12], small rodents [13], carrot samples [14], etc., have been tested for Y. pseudotuberculosis by qPCR. Thus, microbiological diagnosis in combination with classical or real-time PCR may take at least a week or more, during which time the patient’s infection progresses and may lead to complications such as erythema nodosum and reactive arthritis [15]. In this sense, the early etiological diagnosis of yersiniosis is of particular importance. Even more, the on-time detection of yersinia in food as main source of epidemiological outbreaks will help prevent infections and the subsequent further complications for the patients. Reports on the surveillance of pathogenic Yersinia spp., including Y. pseudotuberculosis, are not mandatory for the EU member states and only some EU countries report data on yersiniosis outbreaks on an annual basis. The primary reservoirs of Y. pseudotuberculosis are wild animals, rodents, and birds, with all strains potentially pathogenic to many animal species and humans [16]. According to the EFSA reports, Y. pseudotuberculosis is more rarely reported than Y. enterocolitica and is isolated mainly from pigs [17,18], cattle [19] and poultry [20]. As early as 1988, in a study conducted on the North Island of New Zealand, yersiniosis was identified as a leading cause of mortality among goats [21]. Later, according to a retrospective study on the temporospatial distribution of the disease among these small ruminants covering a period of 23 years from 1 January 1990 to 31 December 2012, Yersinia pseudotuberculosis was found to be the most reported causative agent of yersiniosis in goats to the California Animal Health and Food Safety Laboratory System [22]. The analyzed data revealed 42 cases of goat yersiniosis (manifested mainly with enterocolitis, typhlocolitis, abortion, hepatitis, conjunctivitis, etc.) in 21 countries. Therefrom, eleven cases occurred in dairy breeds (26.2%), which is the second highest after the meet breeds (30.9%). Milk has been reported to be one of the main food products associated with human Y. pseudotuberculosis infections [23] as well. The spread of pathogenic Yersinia via milk is possible if the animal is infected with Yersinia mastitis [24]. Mastitis caused by Y. pseudotuberculosis in a cow has been reported in Italy [25]. The growth of Y. enterocolitica and Y. pseudotuberculosis in milk at different temperatures was studied. It was found that the growth of both pathogens increased significantly at 4 °C and 24 °C in pasteurized and autoclaved milk, while at 37 °C, growth was observed only in autoclaved milk [26]. In 2014, an outbreak of Y. pseudotuberculosis with 55 confirmed human cases was reported in southern Finland. The outbreak was caused by raw drinking milk, packaged in 3 L cans that were distributed for sale in supermarkets [27]. The lack of regular reports on Y. pseudotuberculosis outbreaks from many countries makes it difficult to track the epidemiology of this pathogen. The low isolation rates from food (processed and non-processed) are also due to the low sensitivity of the culture methods (e.g., indistinguishable phenotype from the closely related Yersinia similis and Yersinia pekkanenii [28,29,30] and the need for long incubation periods for enrichment). Therefore, fast and inexpensive PCR-based molecular genetic techniques, such as loop-mediated isothermal amplification (LAMP), which is suitable not only for point-of-care diagnosis in patients but also for rapid detection of the pathogen in food samples, are especially warranted and deserve special attention.

LAMP was developed by Notomi et al. for qualitative identification of target nucleic acids [31]. The method represents autocyclic strand-displacement DNA synthesis in isothermal conditions using a unique DNA polymerase with high strand-displacement activity (Bst DNA polymerase). The reaction is characterized by high specificity, speed and economic efficiency. It proceeds under isothermal conditions (60–65 °C throughout the amplification process) and can be carried out in a water bath or heat block, unlike other PCR-based methods that require the presence of PCR machines. LAMP shows better resistance to PCR inhibitors as compared to the conventional and real-time PCR method [32,33]. A large amount of product accumulates in the samples, leading to a higher sensitivity—less than 10 copies of the nucleic acid of interest can be specifically detected in the tested sample in a very short time of 30–60 min [31]. The increased yield of DNA amplicons (up to 10^9^ DNA copies in 60 min) allows the use of a variety of visual detection methods, such as turbidity and colorimetric or fluorescent indicators [34,35,36,37]. In comparison, the PCR reaction produced up to 1000-fold fewer DNA copies [38]. The advantages of the LAMP method listed so far make it accessible to laboratories with limited resources or those performing “point-of-care” diagnostics. For these reasons, in the last 10–15 years, intensive work has been done on the development of LAMP protocols for the identification of pathogens in samples of different origins.

The sensitivity of the diagnostic method chosen is crucial for making a correct diagnosis. Fastidious bacteria responsible for human gastrointestinal disorders are difficult to detect with conventional microbiological methods and there is a risk of missing the causative agent. PCR-based methods for direct identification of Y. pseudotuberculosis in samples of different origins (food, stool, water, etc.) are especially needed for rapid, specific and sensitive diagnostics during outbreaks and epidemics in order to improve patient outcomes. Up to now, a few PCR methods have been developed for direct detection of Y. pseudotuberculosis in food, clinical samples and water [7,8,10,12,39,40]. In 2008, Lambertz et al. published a real-time PCR protocol for the identification of Y. psedotuberculosis in milk samples based on the chromosomally located ail gene. On the other hand, the detection limit of the conventional PCR methods [12,39,41] was found to be comparatively high (10^1^–10^4^/g sample or even per reaction tube). Therefore, the availability of sufficient data about the sensitivity of new approaches for direct detection of pathogens in samples is especially important for the development of new diagnostic tools.

The LAMP method has been applied for the identification of Y. pseudotuberculosis in clinical samples from liver [42], milk powder [37], feces [43,44] and blood [43]. A set of primers for the chromosomal virulence gene inv was originally developed by Horisaka et al. [42] and in subsequent years has been successfully applied to other types of samples. The authors proved the selectivity of the primers on a large panel of 31 Yersinia and 10 other bacterial species. The results obtained by Horisaka et al. [42] showed that the LAMP reaction was not inhibited by blood serum or plasma heparin (which are known PCR inhibitors), being highly specific and more sensitive than the conventional PCR used by the authors for comparison. These results were confirmed by Kato et al. [44] in feces from patients with yersiniosis caused by Y. pseudotuberculosis. Their studies showed 100% sensitivity of the method and 67% specificity for culture-positive cases, while serology was characterized by 75% sensitivity and 17% specificity. Remarkably, the team was able to confirm three cases using LAMP that were reported as negative with serological methods. The authors’ conclusion is that LAMP can be used as an additional method for serological testing of Y. pseudotuberculosis-positive patients. Kamura et al. [43] used this method to diagnose the pathogen in a ten-month-old infant with bacteremia by examining blood and feces. Yersinia pseudotuberculosis was found in both samples, allowing refinement of antibacterial therapy and control of sepsis.

When determining the sensitivity of a relatively new pathogen detection approach, it is important to apply a sufficiently sensitive method for comparison. The digital droplet PCR (ddPCR) method is one of the most sensitive current techniques for DNA detection and quantification. The aim of the current study was to evaluate for the first time the sensitivity of an LAMP kit with two different dyes (one dsDNA-binding and one metal-sensitive) in comparison with that of ddPCR for detection of Y. pseudotuberculosis in raw goat milk after direct isolation of DNA from eight artificially contaminated logarithmic serial dilutions in order to simulate a wider range of possible contamination levels in conditions close to real ones. For this aim we used the above-described primers sets for LAMP of Horisaka et al. [42] and for qPCR of Lambertz et al. because their selectivity was proven on a large panel of Yersinia and other bacterial species and also on real samples. The protocol of Lambertz at al. [8] was successfully applied in our study for the first time in a ddPCR protocol after necessary modifications regarding the probe’s fluorophore and quencher required for the QX200^TM^ Droplet Reader used in the current study. To our knowledge, such a comparison has not been previously reported in the scientific literature. It will definitely improve the prospects for implementation of the LAMP method into laboratory practice and will facilitate its application.

2. Materials and Methods

2.1. Bacterial Strain and Cultivation

The reference strain Y. pseudotuberculosis IP32918 (Collection Institut Pasteur Paris) was used for all experiments in this study. The strain was stored in Trizma^®^ base (#T1503-500G, Merck, Darmstadt, Germany) with 30% glycerin (#G7893-1L, Merck, Darmstadt, Germany) at −80 °C. After defrosting, the strain was revived first in brain–heart infusion broth (BHI, #GM210, HiMedia, Mumbai, India) for 48 h at 26 °C and thereafter was plated on cefsulodin–irgasan–novobiocin (CIN) agar (#M843, #FD034, HiMedia, Mumbai, India) under the same culture conditions for 24 h to obtain single colonies. For all other experiments, the strain was cultured on CIN agar at 26 °C.

2.2. Milk Sample and Bacteriological Examination

The milk sample was obtained from a private farm located in the Eastern Rhodopes Mountain (Bulgaria). The milk sample was enriched (at 4 °C for 2 weeks). In order to determine the microbial purity of the milk used, the enriched milk sample was cultured at 37 °C for 36 h (incubator IN30, Memmert GmbH & Co.KG, Schwabach, Germany) on BHI agar and selective culture media for Yersinia spp. (CIN agar and MacConkey Sorbitol Agar—#M843-500G and #M1340-500G, HiMedia, Mumbai, India), Salmonella spp. (Xylose Lysin Desoxycholat- gar, Modified, HiMedia, M031I-500G), Listeria spp. (Chromogenic Listeria Agar (ISO) Base #CM1084 with OCLA (ISO) Selective Supplement #R0226E, Oxoid, Basingstoke, United Kingdom) and Escherichia coli (Endo Agar, HiMedia, M029-500G). The isolated single colonies were examined by Gram staining and potassium hydroxide and further subjected to full biochemical characterization with an automatic BD Phoenix^TM^ M50 system (443624, Becton, Dickinson and Company, Franklin Lakes, NJ, USA).

2.3. Identification of Bacterial Isolates with BD Phoenix M50

The BD Phoenix^TM^ M50 system (443624, Becton, Dickinson and Company, Franklin Lakes, NJ, USA) was used for the biochemical identification of the milk isolates, as mentioned above. The colonies were inoculated on Petri dishes with blood agar (Columbia Blood Agar—BD^TM^ Columbia Agar with 5% Sheep Blood, PA-254005.06) and cultured at 37 °C for 18–24 h. Separate panels for G+ (PMIC/ID, 448796, Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and G− (NMIC/ID, 448103, Becton, Dickinson and Company, Franklin Lakes, NJ, USA) microorganisms were used for the previously Gram-characterized microorganisms. A 0.5–0.6 McFarland bacterial suspension was made directly into a vial of ID broth buffer (246001, Becton, Dickinson and Company, Franklin Lakes, NJ, USA) by dissolving material from 5 single colonies from the 18–24 h culture by using a long disposable syringe or a sterile cotton swab. The suspension was mixed for 5 s on a Vortex-type shaker, left for 10 s for the bubbles to come out, and the bacterial density was measured on a nephelometer (PhoenixSpec, Dickinson and Company, Franklin Lakes, NJ, USA). The antimicrobial susceptibility testing (AST) was performed in the same panels by placing 1 drop of indicator solution (246004, Becton, Dickinson and Company, Franklin Lakes, NJ, USA) warmed to room temperature into a bottle of AST broth buffer (246003, Becton, Dickinson and Company, Franklin Lakes, NJ, USA). The bottle was mixed 1–2 times by inversion (without forming bubbles), and 25 µL of the ID broth buffer solution was added and then stirred again by inversion. Both vials (AST and ID) were poured into the relative panels for identification of Gram-positive and Gram-negative bacteria, capped and placed in the instrument where biochemical identification was performed at 35 °C for 24 ± 4 h. The data obtained were analyzed by EpiCentre™ software (V7.45A/V6.71A).

2.4. Artificial Contamination of Goat Milk with Yersinia pseudotuberculosis

Artificial contamination was performed with a 48 h culture of Y. pseudotuberculosis. The bacterial suspension was prepared in sterile phosphate-buffered saline (PBS, #P3813-10PAK, Merck, Darmastadt, Germany) with a cell density 6 × 10^8^ CFU/mL by using a densitometer DEN-1 (Biosan, Lishui, Zhejiang Province, China). From this suspension, serial dilutions in logarithmic order from 10^−1^ down to 10^−7^ were prepared in 1 mL sterile phosphate buffer pre-distributed in sterile 2 mL tubes with safe lock caps (Figure 1).

The stock solution and the dilutions were used for the contamination of raw goat’s milk distributed in equal parts (900 µL) in sterile tubes of 2 mL with safe lock caps. The reason for this approach lies in the more uniform re-suspension of the bacteria in PBS, which ensures the increase in contamination accuracy and efficacy of the milk samples thanks to the better dissemination of bacteria on the sample. Also, aliquots from the PBS dilutions were plated on BHI agar plates for evaluation of the CFU/mL as described below. Briefly, 100 µL from each tube was added to the tubes with raw milk to obtain eight serial logarithmic dilutions from 10^−1^ down to 10^−8^ (Figure 1). The bacterial density in the stock solution used for the artificial contamination was determined by a triplicate inoculation of 100 µL with dilutions of 10^−4^, 10^−5^ and 10^−6^ onto BHI agar plates using a sterile Drigalski spatula. All contaminated milk dilutions were subjected to DNA isolation.

2.5. DNA Isolation from Yersinia pseudotuberculosis Strain IP32918

DNA from a pure Y. pseudotuberculosis culture was isolated with a GenElute™ Bacterial Genomic DNA Kit (#NA2120, Sigma-Aldrich—Merck, Germany). Briefly, one colony from a 48 h bacterial culture incubated on CIN agar was lysed with lysis buffer containing 0.05M NaOH and 0.125% SDS (sodium dodecyl sulfate). Proteins in the sample were digested with proteinase K at 55 °C, and RNA was removed by treating the sample with RNase A solution. Ethanol (96%) was added to the lysate to precipitate the DNA, and the solution was purified through a silica gel column with alcohol-containing washes. The purified DNA was extracted from the column with a buffered solution (pH 9.0) containing 10 mM Tris-HCl and 0.5 mM EDTA. All steps were performed with the ready solutions contained in the kit. Measurement of the DNA concentration of Y. pseudotuberculosis was carried out on a NanoDrop^TM^ Lite Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) in [ng/µL]. The obtained DNA was used as a positive control. The number of DNA copies per µL was calculated based on the following equation for calculation of the mass of 1 molecule of DNA: m [g]= n [bp] × 1.096 × 10^−21^, wherein “m” is the mass of one molecule of DNA and “n” is the number of bp for the strain of Y. pseudotuberculosis IP32918 used (4.7 Mb).

2.6. DNA Isolation from Raw Goat Milk Contaminated with Yersinia pseudotuberculosis

DNA isolation from Y. pseudotuberculosis-contaminated raw goat’s milk samples was performed with the FavorPrep Milk Bacterial DNA Extraction Kit (FAVORGEN BIOTECH CORP., Taiwan). The kit was chosen because the protocol includes steps for eliminating lipids and proteins, which are known PCR inhibitors. The milk samples were centrifuged at 18,000× g (all subsequent centrifugations were performed at the same speed) to pellet the bacteria, bacterial fragments, etc.; then, the supernatant was removed and the pellets were lysed at 37 °C with a lysis buffer containing lysozyme and at 60 °C after adding of proteinase K to each sample following the protocol of the manufacturer. The DNA was precipitated with 96% ethanol and purified through silica gel columns using the washing solutions in the kit. DNA was extracted with 50 µL of the supplied elution TE buffer (pH = 9.0) containing 10 mM Tris-HCl and 0.5 mM EDTA. The DNA quantity was measured by a NanoDrop^TM^ Lite Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) in [ng/µL].

2.7. Loop-Mediated Isothermal Amplification (LAMP) Assay

For the aim of this study, first, the sensitivity of one LAMP kit using two different dyes—a dsDNA-sensitive one, which changes its color from transparent in negative samples to light blue in positive samples and one metal ion (Mg^2+^)-sensitive dye, which changes color from violet in negative samples to blue in positive samples (EURx Blue-LAMP Kit 2X #E1410-01, Gdańsk, Poland)—was evaluated. All reaction mixtures were prepared following the instructions of the manufacturers. The primer set used, which targets the inv gene, was selected from the scientific literature [36,42] and includes 3 pairs of primers as follows: two outer (F3—5′-CTC GTC GCG TGA TTT CTC C-3′, B3—5′-GAT CTA CCC CGA CAG TGA GT-3′), two inner (FIP—5′-CCA GTT GTG GGA GTG CAG GTA ACT ATA AAG AGC GCC CAG CC-3′, BIP—5′-CAC CGG TGA GCG TGT TGC TTT GTG TAA TTG ATC CCG GCA GT-3′) and two loop primers (LF—5′-CAT TCG CGC GCA AAT CC-3′, LB—5′-GCA ACG CAA CCC TTA TGC-3′). The manufacturer’s instructions of all kits were followed by preparing the master mixes. Three stock solutions of primers were prepared as follows: F3/B3 mix—5 µM of each primer, FIP/BIP mix—40 µM of each primer, and LF/FB mix—20 µM of each primer. All reactions were performed at 65 °C for 35 min. The reaction volume was 25 µL, containing 2 µL DNA input from the respective dilution. The calculation of the DNA copies per mL was performed, having in mind that 1 mL of milk was used for DNA extraction; the DNA was eluted with 50 µL TE buffer, 2 µL of which was used for the LAMP reaction.

2.8. Quantitative Detection by Digital Droplet PCR (ddPCR)

Digital droplet PCR was applied for absolute quantification of Y. pseudotuberculosis in the dilutions prepared with DNA isolated from pure IP32918 bacterial culture, as well as in the dilutions of the artificially contaminated milk samples after DNA isolation. The analysis was performed using the Bio-Rad system (Bio-Rad Laboratories, Inc., Hercules, CA, USA) for ddPCR, which includes a QX200^TM^ Droplet Generator, PX1™ PCR Plate Sealer, C1000 Touch™ Thermal Cycler Chassis with reaction module C1000™/C1000 Touch™ and a QX200^TM^ Droplet Reader. The primers and probe were selected from published data [42] and target the ail gene of Y. pseudotuberculosis: Yps 1 (forward)—5′-CGT CTG TTA ATG TGT ATG CCG AAG-3′, Yps 2 (reverse)—5′-GAA CCT ATC ACT CCC CAG TCA TTA TT-3′, and Yps-probe—5′-6-FAM-CGT GTC AAG GAC GAT GGG TAC AAG TTG G-BHQ1-3′ (purchased for synthesis at Microsynth, AG, Balgach, Switzerland). In order to adapt the TaqMan protocol, which was previously created for real-time PCR, to the requirements of the ddPCR used in the current study, we purchased a probe with fluorophore 6-FAM (6-Carboxyfluorescein) and quencher BHQ1 (black hole quencher 1) because they are recommended by the ddPCR producer Bio-Rad and are aligned with the laser scanners of the QX200^TM^ Droplet Reader. The master mix was prepared with “ddPCR™ Supermix for Probes” (#1863026, Bio-Rad Laboratories, Inc., Hercules, CA, USA): 2× Supermix for Probes; 500 nM of each primer, 200 nM probe, 2 µL DNA template, and reaction volume 20 µL. The reaction protocol followed the instructions of the manufacturer: 10 min at 95 °C (enzyme activation), 30 sec at 94 °C (denaturation), 1 min at 60 °C (annealing/extension), 45 cycles of amplification, 10 min at 98 °C (enzyme deactivation), and 30 min at 4 °C (hold). The data analysis was performed with QuantaSoft Software (v. 1.7.4, Bio-Rad Laboratories, Inc., Hercules, CA, USA). The ddPCR reaction was performed in parallel with DNA isolated from pure Y. pseudotuberculosis culture and with the DNA from the artificially contaminated milk samples. The calculation of the DNA copies per mL was performed having in mind that 1 mL of milk was used for DNA extraction; the DNA was eluted with 50 µL TE buffer, 2 µL of which was used for the LAMP reaction.

3. Results

3.1. Bacteriological Findings in Raw Goat Milk

The cultivation of the milk sample on BHI agar at 37 °C for 36 h (after 2 weeks of enrichment at 4 °C) resulted in the growth of five morphologically different colonies, as follows: (1) large shiny S-colony (peach color) with smooth edges and slightly convex pointed center; (2) light yellow, convex S-colony; (3) white, medium large, flat S-colony; (4) creamy non-glossy R-colony with raised center and flat translucent halo; and (5) light beige, medium large, convex S-colony. Each colony was stained by Gram, whereby colonies 1, 2, 3 and 5 were determined as Gram positive, and colony 4 as Gram negative. The colonies were identified biochemically with the BD Phoenix™ M50 system, and the results are presented in Table 1. As is visible from the results, the milk used in the current study was not contaminated with Yersinia spp.

3.2. Artificial Contamination of Milk with Yersinia pseudotuberculosis and Determination of Colony-Forming Units

After the artificial contamination of the milk sample with eight logarithmic dilutions of Y. pseudotuberculosis, starting from a stock solution in PBS with density 0.5 McFarland, the bacterial count in PBS dilutions of 10^−4^, 10^−5^ and 10^−6^ was determined on BHI agar (Figure 2). The concentration of Y. pseudotuberculosis estimated in the three dilutions was as follows: dilution 10^−4^–5.6 × 10^3^/mL; dilution 10^−5^–5.6 × 10^2^/mL; and dilution 10^−6^–5.6 × 10^1^/mL.

3.3. Comparison of the Sensitivity of Two LAMP Kits with ddPCR by Using DNA Isolated from Pure Yersinia pseudotuberculosis Culture

For this aim, serial dilutions of pure Y. pseudotuberculosis DNA were prepared after measuring the DNA concentration on a NanoDrop^TM^ Lite Spectrophotometer. The initial DNA concentration was set on 2 × 10^6^ DNA copies/µL and diluted six times to obtain in total seven 10-fold serial dilutions. Given that 2 µL of each dilution was added to a PCR tube, the highest final concentration of the DNA template in tube 3 (Figure 3) was 4 × 10^4^ per reaction. The LAMP reaction is very sensitive towards contamination and therefore two negative controls were used—one from the PCR box where the LAMP master mix was prepared and one from the PCR bench where the DNA template was added to the tubes. First, a ddPCR was performed with 2 µL of each dilution in order to quantify the exact number of the DNA copies per dilution and to be able to evaluate the sensitivity of the LAMP kit regarding each of the dyes used (Figure 3, Table 2).

As is visible from Figure 3 and Table 2, the concentrations of the pure Y. pseudotuberculosis DNA dilutions determined by ddPCR almost fully correspond to the preliminary calculations based on the spectrophotometric measurement. These dilutions were used to carry out LAMP reactions with the chosen commercial kit (Figure 4).

The results from the electrophoresis reveal that the EURx kit exhibits high sensitivity. Lines 8 and 9 represent the highest dilutions where DNA input is very low are slightly visible on the gel electrophoresis, wherein the intensity is more pronounced by the kit with the blue dye. The color contrast in the same samples is not distinguishable from the negative controls.

The sensitivity of LAMP assay (visible colorimetric and gel electrophoresis detection limit) when using pure DNA was compared to that of ddPCR in Table 3. Both BLUE LAMP kits were characterized with a detection limit of 1.43 × 10^2^ DNA copies per reaction regarding visible colorimetric detection. The detection limit of the DNA gel electrophoresis approach was 1.17 × 10^1^ DNA copies per reaction for the blue dye and 1.43 × 10^2^ DNA copies per reaction for the violet dye variant. The differences in the sensitivity of the colorimetric LAMP detection if the result is read visually by the change in the color of the corresponding dye and the ddPCR is Δlog = 1.0 × 10^2^ less, with the ddPCR protocol being more sensitive.

3.4. Quantification of Yersinia pseudotuberculosis in Artificially Contaminated Raw Goat Milk by Using ddPCR

Digital droplet PCR was performed in order to quantify the number of DNA copies in the artificially contaminated milk samples (Figure 5). The data are given in Table 4. The concentration in the dilutions decreased uniformly in logarithmic order with a 10-fold difference between each sample. Dilutions from 10^−2^ to 10^−8^ are quantified as long as the concentration in the first dilution is high and only positive droplets were generated.

The concentration of the inoculated bacteria in CFU/mL determined on the BHI agar plates was compared to that estimated by ddPCR. As the results presented in Table 5 show, the number of CFU/mL determined with ddPCR was significantly more than that counted on agar plates. The difference between both approaches for quantification (Δlog) of the Y. pseudotuberculosis CFUs (colony-forming units) increases from less than one (8 × 10^0^) up to more than two logarithms (6.9 × 10^2^) as the dilution rate decreases, which could be due to the presence of viable but non-culturable bacteria.

3.5. Comparison of LAMP Sensitivity with That of ddPCR by Using DNA Isolated from Artificially Contaminated Raw Goat Milk

LAMP reactions were performed with DNA isolated from the dilutions of artificially contaminated milk. The results from the colorimetric visualization and the concomitant gel electrophoresis are presented in Figure 6. The detection limits are given in Table 6. The BLUE LAMP KIT with violet dye turned out to be less sensitive regarding the color change in this experiment than the blue dye, for which color changes can be distinguished up to a concentration of 1.48 × 10^2^ DNA copies per reaction. The change in the other dye from violet to blue can be distinguished visibly up to a concentration of 1.25 × 10^3^ DNA copies per reaction. The result from gel electrophoresis reveals higher sensitivity than the visible colorimetric detection, as expected. The kit variation with the metal ion-sensitive violet dye detects 0.8 × 10^0^ DNA copies per reaction, whereas the variation with the dsDNA-binding blue dye detects one logarithm less DNA copies.

4. Discussion

The need for fast and efficient methods for rapid point-of-care detection of specific nucleic acids by isothermal amplification reactions has led to the development of a wide variety of LAMP protocols and products for UV/Vis detection of the investigated samples. The main advantage of the LAMP method is that high yields of the product are obtained in a short time under constant temperature in only three main steps of DNA amplification (sample preparation, amplification of genetic material and detection). Moreover, the Bst polymerase enzyme is characterized by stability and high tolerance to amplification inhibitors. The reaction conditions were described in detail by Notomi at al. in the year 2000 and require the presence of two types of oligonucleotide primers (outer—F3 and B3 and inner—FIP and BIP) that recognize six different DNA regions. A third pair of primers can be added to the reaction mixture—the so-called loop primers LF (loop forward) and LB (loop backward)—the presence of which reduce run time [41,45] and increase specificity. These primers anneal to the loop and generate more amplicons as well as inorganic pyrophosphate as a byproduct of the reaction, which is used in some approaches for colorimetric visualization of positive samples. The final products of amplification represent a mixture of looped DNA fragments and different lengths of the straight double-stranded part, which on agarose gel electrophoresis form a typical DNA ladder [31]. Reaction products can also be visualized by turbidimetry [34], colorimetric dyes [46] and fluorescent dyes [35,37], thus shortening the electrophoresis step and the time to obtain the result. The LAMP method is also characterized by some shortcomings, such as the difficult design of primers due to the specific requirements for the oligonucleotides and the target DNA sequence [47]. The high sensitivity of the method requires that the operating conditions are equivalent to those for quantitative PCR. The LAMP approach is less versatile than PCR, and multiplex reaction approaches are still less developed than classical PCR, although extensive work is also underway in this direction. The reaction products are not of a specific size for the amplified amplicon, but are a mixture of DNA fragments of different sizes and are visualized on gel electrophoresis as a DNA ladder. Despite the listed shortcomings, which are expected to be overcome in the future with the improvement of the method, the studies published so far show that the LAMP method has the potential for rapid identification of infections at an early stage or the proof of traces of pathogens in clinical samples, food products and other species research purposes. A number of research groups are working on the development and validation of LAMP protocols for the identification of Yersinia pathogens. However, there is still no method validated and introduced into routine diagnostic practice for the detection of pathogens with LAMP. A number of studies are currently underway to improve the sensitivity of the imaging approaches used, as well as their repeatability [34,37,48].

Considering the importance of the LAMP reaction as a potential on-field or point-of-care diagnostic method due to its rapidity, sensitivity, specificity, cost-effectiveness and ease of implementation, in the present study, we compared for the first time the sensitivity of one commercial LAMP kit in two variants—one with containing a dsDNA-binding dye and the other containing a metal ion (Mg^2+^)-sensitive dye—for visible detection of the sought pathogen Y. pseudotuberculosis with that of ddPCR in order to ensure precise determination of the detection limit. Color visualization of the positive LAMP samples was compared with UV detection of the DNA product in the same samples on agarose gel. The study was performed with samples artificially contaminated with a selected Y. paseudotuberculosis strain because the surveillance of this food pathogen in milk is low according to the available scientific data discussed in detail in the next paragraphs and it is difficult to collect enough data with statistical significance in a short period of time.

In the current study, as to our knowledge, for the first time the primers/probe set of Lambertz et al. [8] was used in a ddPCR protocol for the identification of Y. pseudotuberculosis in artificially contaminated milk samples. The amplification protocol was identical to the original one, with the exception of the primer’s concentration, which in the current study was diminished from 900 nM to 500 nM (following the Bio-Rad instruction for TaqMan-based ddPCR), and the choice of fluorophore (6-FAM) and quencher (BHQ1) was tailored to the reader’s laser channels. As expected, the ddPCR protocol achieved very high sensitivity—2.0 × 10^1^ DNA copies/mL—which is comparable to the results for artificially contaminated carrots in the study of Lambertz et al. [8]. Several studies focused on PCR inhibitors; additionally, the manufacturer of the ddPCR machine used in the current study (Bio-Rad, USA) states that ddPCR is less affected by PCR inhibitors than classical PCR or real-time PCR [49,50,51,52,53], which makes the protocol a reliable approach for DNA enumeration in milk samples and comparative evaluation of the sensitivity of other DNA amplification-based methods such as LAMP. The difference between the number of viable CFUs counted on agar plates prepared with selected artificially contaminated milk dilutions and the number of DNA copies detected with the ddPCR protocol could be due to the presence of viable but non-culturable bacteria, as well as the presence of DNA from non-viable cells or the DNA extraction efficiency itself. Najdenski et al. proposed a new protocol for the detection and enumeration of Y. enterocolitica in milk samples and proved that there is a difference between CFU/mL counted on agar plates and the number of DNA copies detected via the real-time TaqMan-based PCR technique, especially in samples with low DNA input [54], which corresponds to our results. The presented variation in the ddPCR protocol based on the protocol of Lambertz et al. cited above could be used for the identification and enumeration of Y. pseudotuberculosis in milk samples after the selection of an appropriate fluorophore and quencher for the QX200^TM^ Droplet Reader and optimization of the primers/probe concentration.

Analysis of the data obtained from the LAMP protocol shows a very high sensitivity of both dyes, which turn blue when the reaction product accumulates in a positive sample. The result from the DNA electrophoresis confirms the high sensitivity of the LAMP reaction. The result obtained for the colorimetric sensitivity of both dyes for the detection of Y. pseudotuberculosis DNA isolated from pure bacterial culture revealed equal potential regarding the color change (1.43 × 10^2^ detected DNA copies/reaction). However, the gel electrophoresis approach showed better sensitivity to the variation in the dsDNA-binding dye (1.17 × 10^1^ DNA copies per reaction vs 1.43 × 10^2^ for the violet dye). The colorimetric detection limit of the kit based on color change in the violet metal ions (Mg^2+^)-sensitive dye in artificially contaminated milk samples was 1.25 × 10^3^ DNA copies/reaction, corresponding to 3.1 × 10^4^ DNA copies/mL. When gel electrophoresis of the same reactions was performed, 0.8 × 10^0^ DNA copies/reaction were detected (corresponding to 2.0 × 10^1^ DNA copies/mL) (Table 6). The dsDNA-binding dye, which changes color from transparent to blue in positive reactions, showed a colorimetric detection limit of 1.48 × 10^2^ DNA copies/reaction (corresponding to 3.4 × 10^3^ DNA copies/mL) and 1.36 × 10^1^ DNA copies/reaction (corresponding to 3.4 × 10^2^ DNA copies/mL) on gel electrophoresis (Table 6). Interestingly, the results obtained with DNA isolated from artificially contaminated samples revealed a better colorimetric detection limit for the dsDNA-binding dye, whereas on gel electrophoresis, the signal was stronger and detected DNA copies in samples with lower DNBA input for the variant with the metal-sensitive dye. However, the colorimetric LAMP detection of DNA isolated from milk samples reveals lower sensitivity. We can assume that the differences in the detection limit between DNA isolated from pure bacterial culture and that extracted from Y. pseudotuberculosis-contaminated milk samples representing the highest dilutions could be due to the following main reasons: (1) the low DNA input in the starting DNA sample, which does not guarantee taking the same amount of DNA copies each time; (2) the presence of PCR inhibitors such as calcium ions, lipids and proteins, which are contained naturally in milk and dairy products, and (3) the DNA extraction method, which should ensure a high level of purification from PCR inhibitors. Currently, there are only few reports on the inhibition of the LAMP reaction by known PCR inhibitors such as calcium chloride, bile salts, urea, humic acid, EDTA, etc. One of the most comprehensive investigations was published by Nwe et al. in 2024 [55]. The authors investigated the effects of different inhibitors on the LAMP reaction, including calcium ions and some proteins. Their findings indicate that although the onset of the amplification could be delayed, this does not significantly affect the final accumulation of LAMP amplicons in end-point reactions, except in cases of very high concentrations of PCR inhibitors exceeding those sufficient for PCR inhibition. The authors used real-time LAMP with Eva Green dye, which is a dsDNA-binding dye, and the metal ion-sensitive indicator hydroxy naphthol blue (HNB), which binds Mg^2+^ ions (the violet dye used in the current study belongs to the same group of dyes) for comparison. Nwe et al. managed to prove that the end-point detection method with colorimetric dyes is less affected by PCR inhibitors than fluorescent real-time LAMP and that in general LAMP is more robust than classical PCR and tolerates higher concentrations of known PCR inhibitors, except calcium ions at concentrations above 0.8 mM. It is known that EDTA could also be a PCR or LAMP inhibitor [56]. Inhibition depends on the concentration of EDTA in the final PCR reaction [57]. According to Huggett et al., the PCR reaction is completely inhibited by 4 mM of EDTA. The TE buffers used in our study contain 0.5 mM EDTA, which, according to the experimental data published by Huggett et al., does not inhibit the PCR reaction. According to Sakatoku et al. [58], 10 mM of EDTA was necessary to inhibit the LAMP reaction, which excludes the EDTA concentration in the TE buffer used for the elution of our samples as a possible inhibitor of the LAMP reaction. The chemistry of dyes could also be a reason for the differences in the result regarding the DNA isolated from milk samples, having in mind the possible presence of inhibitors. However, the data about the interaction between different types of dyes and LAMP inhibitors are very scanty. Both dyes used in our study rely on different principles for reaction product detection. The blue dye detects the LAMP amplicon by binding dsDNA, whereas the violet dye changes color depending on the formation of insoluble complexes between Mg2+ ions and the pyrophosphate released upon dNTP incorporation into the amplicons. It is known that inhibitors such as calcium chloride (contained in milk) and IgG can delay the reaction’s onset [55] but do not inhibit the formation of the reaction product. Calcium ions can stabilize secondary and tertiary DNA structures, and this could be a reason why the dsDNA-binding dye has a one logarithm lower detection limit on gel electrophoresis than the metal-sensitive dye in samples with very low DNA input, but further investigations are needed to prove this hypothesis. DNA isolation methods should also be carefully selected before starting such investigations. In previous studies, the efficiency of different kits for the isolation of DNA from milk was compared [54,59,60]. Looking for a protocol that provides extraction and purification of DNA samples and based on our previous laboratory experience in comparison to different kits for DNA extraction from milk, we chose the FavorPrep Milk Bacterial DNA Extraction Kit developed especially for milk samples. The quality of the isolated DNA was confirmed with a ratio of 260/280 between 1.7 and 1.9 by measurement of the DNA concentration with a NanoDrop^TM^ Lite Spectrophotometer. The high sensitivity of ddPCR is additional proof that the kit achieves isolation of a low number of DNA copies. The amount of extracted DNA is comparable to similar results obtained in previous studies with other DNA isolation kits [54]. The differences between the detection limit of both variants of the kit in the frame of the same set of samples (either DNA from pure bacterial culture or from milk samples) could most probably be due to the low DNA concentration of the highest dilutions. However, in order to confirm with certainty which is the exact reason for the difference in Y. pseudotuberculosis detection in samples originating from different matrices (in our study, DNA from pure bacterial culture vs DNA from milk samples), further in-depth chemical studies should be carried out to identify if there are residual calcium ions, humic acid, fats and proteins in the extracted DNA, which could be a subject of a follow-up study.

Especially important to achieve wide applicability of the LAMP method applied in our study for detection of Y. pseudotuberculosis is the use of kits which allow colorimetric detection in the visible spectrum that do not require the presence of expensive equipment and can be applied at the site of the epidemiological outbreak (point-of-care diagnosis). To date, the dyes used include various pH-sensitive [61] or metal-sensitive [36] indicators, chelating agents [34], DNA intercalating dyes [36], etc. The dyes used for visual evaluation of positive and negative LAMP in this study belong to the following two groups: the dye which changes the color from transparent to blue is a dsDNA-binding dye, whereas the other die, which changes from violet to blue belongs to the group of the metal-sensitive dyes which bind to Mg^2+^ ions. A number of studies are currently underway to improve the sensitivity and reproducibility of the imaging approaches used. The large number of amplicons, as well as the accumulation of magnesium pyrophosphate as a reaction byproduct, allow for rapid monitoring of positive and negative samples in real time or at the end of the reaction using fluorescent or colorimetric readout dyes changing color through dsDNA binding or Mg^2+^ binding [34,37,48]. From the experiments carried out, it can be concluded that visual detection of Y. pseudotuberculosis with the blue dsDNA-binding LAMP dye (10^3^ DNA copies/mL), although less sensitive than gel electrophoresis detection, is comparable to that of the classical microbiological approaches based on the cultivation of the bacteria on culture medium and therefore has the potential to be implemented for point-of-care diagnostics after the necessary validations of the protocol. When isolating Y. pseudotuberculosis DNA from milk samples, the sensitivity of the dyes at low DNA concentrations depends most probably on the potential presence of residues of PCR inhibitors and the DNA extraction method. In the current study, the dsDNA-binding dye was less sensitive than the Mg^2+^-sensitive dye regarding gel electrophoresis detection, which might be due to the principle of color change. However, the blue dsDNA-binding dye is characterized by a better contrast between positive and negative samples. The difference between the detection limit of both kits is Δlog = 10^1^ DNA copies/reaction (Table 6) in the frame of the same set of samples and as discussed above is most probably due to the low DNA input. Remarkably, the detection limit of the LAMP kit in both variants when gel electrophoresis is applied for the final read of the results is comparable to that of the ddPCR protocol. Finally, the results obtained in the current study demonstrated that the blue dsDNA-binding dye should be preferable for visual point-of-care or on-field detection of Y. pseudotuberculosis because of the better contrast in the color change between positive and negative samples, which allows visual detection of one logarithm lower DNA copies/mL than the metal-sensitive violet dye.

5. Conclusions

In conclusion, the current study evaluates for the first time the sensitivity of one commercial LAMP kit with two different dyes (one dsDNA-binding and one metal ion (Mg^2+^)-sensitive) for visual detection of Y. pseudotubercusis in artificially contaminated goat milk samples. The sensitivity of the dyes was compared both among the dyes themselves and to that of the ddPCR method. The data obtained show that the sensitivity of colorimetric visual detection with the blue dsDNA-binding dye is comparable to that of classical microbiological methods, whereas the gel electrophoresis approach reveals a detection limit close to that of the ddPCR protocol and allows highly sensitive qualitative pathogen detection. Our results show that the dsDNA-binding dye could be recommended for point-of-care and on-field detection of Y. pseudotuberculosis because the color change offers a visual distinguishment between negative and positive samples by microbial contamination of 10^3^ DNA copies/mL. The data presented in the current study will contribute to the future validation of an appropriate LAMP protocol for the detection and identification of Y. pseudotuberculosis. The conducted comparison also contains important information for LAMP implementation as potential rapid method not only for point-of-care and on-field testing but also for routine laboratory diagnostics after the necessary validation procedures, with more virulence genes specific for identification of Y. pseudotuberculosis and comparison between different DNA amplification-based methods for the detection of this food pathogen with primer sets targeting the same genes. Although the data defined here provide a reliable basis for interpreting the LAMP results, more in-depth chemical analyses should be performed in future studies in order to clarify the reasons for the differences between the detection limit of the dyes by DNA inputs lower than 10^2^ DNA copies per reaction.