End-point rapid detection of total and pathogenic Vibrio parahaemolyticus (tdh⁺ and/or trh1⁺ and/or trh2⁺) in raw seafood using a colorimetric loop-mediated isothermal amplification-xylenol orange technique

Bacterial strains, culture conditions, and genomic DNA template preparation

Vibrio parahaemolyticus reference stains, DMST 15285 (tdh⁺/trh⁻) (obtained from the Department of Medical Science, Ministry of Public Health, Nonthaburi Province, Thailand), TISTR 1596 or ATCC 17802 (tdh⁻/trh2⁺) (obtained from the Thailand Institute of Scientific and Technological Research), and Vp10/5 (tdh⁺/trh1⁺) (obtained from an oyster, Chatuchak District, Bangkok, Thailand in 2018) were utilized as the positive controls to optimize and validate LAMP-XO assay. The reference stains were cultured in 5 mL tryptic soy broth (TSB) (Merch KGaA, Darmstadt, Germany) supplemented with 2% NaCl (BDH Poole, USA) in a shaking incubator at 250 rpm for 16 h at 37 °C. A volume of three mL of the culture broth was transferred into a 1.5-mL microcentrifuge tube and centrifuged at 10,000 rpm for 2 min to remove the supernatant. The pellet was resuspended in Tris-EDTA (TE) buffer (Invitrogen, Grand Island, NY, USA), and subjected to genomic DNA (gDNA) extraction using GenUP™ kit (Biotechrabbit, Berlin, Germany) according to the manufacturer’s protocol. The gDNA concentration was measured using a nanodrop (DS-11 FX+, spectrophotometer/fluorometer; DeNovix, Wilmington, DE, USA). The gDNA solution was aliquoted and kept at −80 °C.

Sample collection, V. parahaemolyticus detection, storage, recovery, and gDNA template preparation

A total of 102 raw seafood samples were used in this study. Of the 102 samples, 83 were purchased from fresh markets and supermarkets in Bangkok, Thailand of which 30 and 53 were purchased in 2018 and 2021 to 2022, respectively. Of the 102 samples, 16 samples were V. parahaemolyticus isolates obtained from Pacific white shrimp collected from different shrimp farms in eastern Thailand in 2013 and three samples were V. parahaemolyticus isolates obtained from Pacific white shrimp collected from North Vietnam in 2016. Eighty three samples purchased from markets comprised of six crabs (five blue swimming crabs and one red swimming crab), 15 fish (five groupers, five giant sea perches, two mackerels, one ornate threadfin bream, one bluefin tuna, and one salmon), 31 mollusc shellfish (15 oysters, six green mussels, five blood cockles, two short-necked clams, two spiral babylon snails, and one scallop), 21 shrimp (one giant tiger prawn and 20 Pacific white shrimp) and 10 squids (six splendid squids, three octopuses, and one giant squid tentacle). For V. parahaemolyticus detection of the purchased seafood, the samples were separately put into a sterile plastic bag and taken to a laboratory in a cooler bag containing ice and were handled within 2 h after sample purchasing. Briefly, 2.5 g of each sample was cut aseptically and immersed into 22.5 mL of TSB (Merch KGaA, Germany) supplemented with 2% NaCl (BDH Poole, Rahway, NJ, USA) for bacterial enrichment. The sample was gently mixed by hand and incubated at room temperature (RT) for 30 min before being removed from culture broth which was further incubated at 37 °C for 16 h. A total of three mL of culture broth was subjected to gDNA isolation. The extracted gDNA was used as a template for LAMP-XO, standard LAMP, qPCR, and conventional PCR assays. For V. parahaemolyticus isolation, one loop full of the culture broth was streaked onto thiosulfate-citrate-bile salts-sucrose agar (TCBSA) (BD, Sparks, USA) on which V. parahaemolyticus produced opaque and blue-green color with 2–3 mm in diameter colonies. The presumptive V. parahaemolyticus colonies were picked up separately and transferred to CHROMagar™ Vibrio (CHROMagar, Paris, France) on which V. parahaemolyticus produced mauve color colonies. For long-term storage, a single mauve colony of V. parahaemolyticus on CHROMagat™ Vibrio was cultured in TSB (Merch KGaA, Germany) supplemented with 2% (w/v) NaCl (BDH Poole, Rahway, NJ, USA) at 37 °C for overnight. A total of 400 µL of 50% glycerol (Sigma-Aldrich, St.Louis, MO, USA) and 600 µL of an overnight culture were mixed aseptically in a sterile 1.5-mL microcentrifuge tube and kept at −80 °C. For bacterial recovery, ice crystals on top of a glycerol stock were aseptically scraped and transferred into TSB supplemented with 2% NaCl and incubated in a shaker incubator at 37 °C with 250 rpm shaking for overnight. The presumptive culture was subjected to species-specific rpoD-qPCR assay for confirmation of species.

Optimization and validation of LAMP-xylenol orange assay

Optimization of LAMP-xylenol orange (LAMP-XO) assay was carried out using 4 sets of previously described primers targeting four different genes: Set 1 for rpoD gene for total V. parahaemolyticus (pathogenic and non-pathogenic strains) detection (Nemoto et al., 2011; Lamalee et al., 2023) and sets 2–4 for tdh, trh1, and trh2 genes, respectively, for V. parahaemolyticus pathogenic strain detection (Table 1) (Nemoto et al., 2009; Yamazaki et al., 2010). The optimization for each gene was done separately. For rpoD detection, in addition to the main primers utilized, we designed four additional primers (loop forward and loop backward 2; F1c2 and B1c2, and forward inner and backward inner 2; FIP2 and BIP2) (Table 1) to further improve the reaction kinetics (Jaroenram et al., 2022). The primers were examined for possible cross dimerization by basic local alignment search tool (BLAST) (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Briefly, the protocol (Table S1) was done in a 25-µL reaction mixture containing each target-specific primer set (forward inner primer (FIP), backward inner primer (BIP), forward outer primer (F3), backward outer primer B3), forward loop primer (LF), and backward loop primer (LB)) at different amounts (Table 1), dNTP mix (New England Biolabs, Ipswich, MA, USA), betaine (Sigma-Aldrich, St.Louis, MO, USA) MgSO₄ (New England Biolabs, Ipswich, MA, USA), Bst 2.0 WarmStart DNA polymerase (New England Biolabs, Ipswich, MA, USA), 2.5 µL of 1× low-buffer solution with pH 8.5 (100 mM (NH₄)₂SO₄, 500 mM KCl, 20 mM MgSO₄, and 1% Tween-20), and 1 µL of a gDNA template. The final volume was adjusted to 25 µL using UltraPure™ distilled water (DW) (Invitrogen, Grand Island, Germany). The negative control containing only DW (no gDNA templates) was included in each run. LAMP reaction was done in a Loopamp Realtime Turbidimeter LA-320C (Eiken Chemical Co Ltd, Tokyo, Japan) at a given condition (temperature and time) followed by DNA polymerase inactivation at 80 °C for 5 min. Following this initial protocol, the optimal incubation temperature was first determined, in that the LAMP reactions were carried out at various temperatures (60, 63, and 65 °C) for 75 min. The obtained optimal temperature was then subjected to optimizing six respective parameters: dNTP mix (1.2–1.8 mM), betaine (0.2–0.8 M), MgSO₄ (4–10 mM), Bst 2.0 WarmStart DNA polymerase (6–12 U), reaction time (30, 45, 60, and 75 min), and XO dye, (0.03–0.12 mM) (Table S2). The last parameter was performed in a heat block, and the result was inspected visually. Color of LAMP-XO was changed from purple to yellow in a positive test while color was still purple in a negative test. To confirm LAMP-XO results, LAMP amplicons were analyzed by 3% agarose gel electrophoresis (AGE) (Vivantis, Malaysia), stained with ethidium bromide (Invitrogen, Waltham, MA, USA) and visualized under UV illumination. Each parameter used 1 µL gDNA of 10⁶ copies/µL/reaction as a template except for the incubation temperature and time that used 1 µL aliquot of 10-fold serially diluted DNA (10⁴, 10³, 10², 10 copies/µL) instead. The DNA copy has been calculated using the formular “amount of DNA (ng) × 6.022 × 10²³/ length of a DNA template (bp) × 1  × 10⁹ × 650” (https://www.technologynetworks.com/tn/tools/copynumbercalculator). For each primer set/target gene, any given temperature, time, and components’ concentration that maximize DNA amplification based on signal intensities by the turbidimeter and the degree of color change from purple (negative) to yellow (postitive) was selected to establish the standard LAMP-XO protocol.

Standard LAMP assay

The standard LAMP assay was non-colorimetric, adopted from the standard LAMP protocol suggested by New England Company Ltd. (https://international.neb.com/protocols/2014/06/17/loop-mediated-isothermal-amplification-lamp). It was used as a control assay to test the efficiency of LAMP-XO. Its reaction components were almost similar to those of the optimized LAMP-XO, only excluding XO, and the low-buffer that was substituted by 1× ThermoPol-supplied reaction buffer (20 mM Tris-HCl, 10 mM (NH₄)₂SO₄, 10 mM KCl, 2 mM MgSO₄ and 0.1% Triton X-100, pH 8.8). The LAMP reaction was performed in a Loopamp Realtime Turbidimeter LA-320C (Eiken Chemical Co Ltd, Tokyo, Japan), at 65 °C (75 min for rpoD, tdh, trh1, and trh2). The results were reported in a real-time amplification plot format. For result confirmation, LAMP products were inspected using 3% AGE.

Conventional PCR assay

Conventional PCR assay was performed using species-specific toxR primers for V. parahaemolyticus detection and tdh, trh1, and trh2 primers for V. parahaemolyticus pathogenic strain detection (Table S3) (Tada et al., 1992; Kim et al., 1999; Messelhäusser et al., 2010). PCR reaction was carried out in a 25-µL reaction mixture containing 12.5 µL GoTaq^® Green Master Mix solution (Promega, Madison, USA), 0.4 µM each forward primer and backward primer for toxR, tdh, trh1, and trh2 primers and 1 µL of a gDNA template. A final volume of a 25-µL reaction mixture was adjusted with UltraPure™ DW (Invitrogen, Grand Island, Germany). The PCR amplification was done in T100 Thermal Cycler No. 186-1096 (Bio-Rad, Hercules, CA). PCR products were analyzed using 1.5% AGE (Vivantis, Malaysia), stained with ethidium bromide (Sigma-Aldrich, St.Louis, MO, USA) and observed under a UV light. A positive control and a negative control were included in each run.

Quantitative PCR assay

A quantitative PCR (q-PCR) assay was done using species-specific rpoD primers for V. parahaemolyticus detection and tdh, trh1, and trh2 primers for V. parahaemolyticus pathogenic strain detection (Table S4) (Nemoto et al., 2009; Messelhäusser et al., 2010; Yamazaki et al., 2010; Nemoto et al., 2011). A 20-µL reaction mixture consisted of 10 µL of 2× KAPA SYBR FAST qPCR Master Mix Universal (Sigma-Aldrich, St.Louis, MO, USA), 10 mM each forward primer and backward primer, and 1 µL aliquot of 20 ng/µL of a gDNA template. A final volume was adjusted using UltraPure™ DW (Invitrogen, Grand Island, Germany). The qPCR amplification was carried out in a Rotor-Gene Q R10116106 (Qiagen Hilden, Germany). A positive control, a negative control, and a standard curve were included in each run.

Comparative detection limit of LAMP-XO, standard LAMP, conventional PCR, and qPCR assays by gDNA

The concentrations of gDNA from the reference strains, DMST 15285, TISTR 1596 or ATCC 17802, and Vp10/5, were measured using a nanodrop (DS-11 FX+, spectrophotometer/fluorometer; DeNovix, Wilmington, DE, USA) and subjected to serial dilution in a 10-fold manner from 10⁴–10⁰ copies/µL. The DNA solution was used as a template for LAMP-XO, standard LAMP, qPCR, and conventional PCR assays.

Comparative detection limit of LAMP-XO, standard LAMP, conventional PCR, and qPCR assays by spiked seafood samples

Three Pacific white shrimp from a supermarket were sterilized by autoclaving. Lack of V. parahaemolyticus contamination was later confirmed by a culture method using TCBSA (Difco Laboratories, Sparks, MD, USA). Briefly, 2.5 g of each Pacific white shrimp was spiked with one mL aliquot of a 10-fold serial dilution of overnight culture of V. parahaemolyticus reference strains to generate inoculating levels of 10⁴–10⁰ colony-forming unit (CFU)/2.5 g sample. The inoculating levels were calculated based on the assumption of OD₆₀₀ = 1 = 8  × 10⁸ CFU/mL. A negative control was a non-spiked sample. Each sample was adjusted to 22.5 mL using TSB supplemented with 2% NaCl, mixed gently by hands and stored at RT for 30 min. All samples were further incubated at 37 °C for 4 h prior to gDNA extraction. The obtained gDNA was subjected to LAMP-XO, qPCR, and conventional PCR assays.

Detection of total V. parahaemolyticus(pathogenic and non-pathogenic strains) and pathogenicV. parahaemolyticus (tdh⁺and/ortrh1⁺and/ortrh2⁺) in raw seafood samples by LAMP-XO andstatistical analysis

For LAMP-XO validation using seafood samples, LAMP-XO reaction for each target gene was carried out using optimal reagent concentrations and conditions as shown in Table 2. Final results of LAMP-XO validation were cross-compared with conventional PCR and qPCR. Specimens were classified as true positive, true negative, false positive, or false negative for each test under evaluation compared to qPCR as a gold standard. The clinical sensitivity, specificity, positive predictive values (PPV), negative predictive values (NPV), diagnostic accuracy, percent overall agreement (POA), and 95% CIs of LAMP-XO, conventional PCR, and qPCR assays were analyzed by SAS software (SAS Institute Inc., Cary, NC, USA). The clinical sensitivity was calculated as (number of true positives)/(number of true positives + number of false negatives) × 100, and the clinical specificity was calculated as (number of true negatives)/(number of true negatives + number of false positives) × 100. The PPV was calculated as (number of true positives)/(number of true positives + number of false positives) × 100, and the NPV was calculated as (number of true negatives)/(number of true negatives + number of false negatives) × 100. The accuracy was calculated as (number of true positives + number of true negatives)/(total number of patients) × 100. The degree of agreement between two diagnostic tests was measured by the concordance response rate (percentage of responses with both positive or both negative results). The POA indicates the percentage of relationship between varied results of comparative and reference methods. Note that the clinical (statistical) sensitivity refers to the ability of the validated assay to correctly identify real positive samples (V. Parahaemolyticus contaminated samples), while the clinical (statistical) specificity refers to the ability of the validated assay to correctly identify real negative samples (V. Parahaemolyticus-free samples). The significant difference between two detection methods, i.e., LAMP-XO and qPCR and LAMP-XO and conventional PCR for each gene was analyzed with a McNemar chi-square test and a P value < 0.05 was considered significant.

Optimization of LAMP-XO assay conditions

LAMP reactions for each target gene (rpoD, tdh, trh1, and trh2) (Table 1) were performed in a low-buffer solution (pH 8.5) based on the method previously reported to leverage the effect of a pH drop for colorimetric product (Jaroenram, Cecere & Pompa, 2019). Since LAMP reaction standard temperatures range from 60–65 °C, the incubation temperature at which a template could be best amplified was firstly optimized, followed by LAMP reagents, reaction time, and XO concentrations (Table S2).

In the LAMP reaction cocktail, the optimal XO concentration would qualitatively generate the most substantial colorimetric change from purple (negative) to yellow (positive) upon a decrease in pH (< 6.7) in the presence of the LAMP reaction by-products. The XO concentration producing the most distinctive color difference between the positive and negative test results was selected (Table S5). Using a high XO concentration is a concern since it interferes in a weak positive reaction. Overall, different target genes have different components’ concentrations, except the incubation temperature (65 °C) and amplification time (75 min) which is in common as reported in Table 2. The performance of optimal LAMP-XO was compared with conventional PCR and qPCR assays (Table S6).

Comparative detection limit of LAMP-XO, standard LAMP, conventional PCR, and qPCR assays using a 10-fold serial dilution of gDNA

To test the analytical sensitivity of the assays, LAMP-XO, standard LAMP, PCR, and qPCR assays were conducted for each target gene/primer set with the same set of a gDNA template ranging from 10⁴–10⁰ copies/reaction (Table 3, Fig. 1). Primers targeting rpoD were used for total V. parahaemolyticus (pathogenic and non-pathogenic strains) detection in LAMP-XO, standard LAMP, and qPCR assays while primers targeting toxR were used for total V. parahaemolyticus (pathogenic and non-pathogenic strains) detection in PCR. Primers targeting tdh, trh1, and trh2 were used for pathogenic V. parahaemolyticus (tdh⁺ and/or trh1⁺ and/or trh2⁺) detection. For LAMP-XO assay, rpoD and tdh primers indicated positive amplification at 10⁴–10² copies, while trh1 and trh2 primers did at 10⁴–10³ copies. Thus, the DLs of the naked eye detection of LAMP-XO for rpoD and tdh genes were 10²copies/reaction and for trh1 and trh2 genes were 10³ copies/reaction (Fig. 1A, top). These colorimetric results were in agreement with those confirmed by AGE (Fig. 1A, bottom). Standard LAMP assay determined DLs of 10² copies/reaction for rpoD and tdh and 10³ copies/reaction for trh1 and trh2 for both turbidity measurement (Fig. 1B, top) and AGE (Fig. 1B, bottom). PCR produced DLs at 10³ copies for toxR and 10⁴ copies for tdh, trh1, and trh2 (Fig. 1C), while qPCR assay produced DLs at 10¹ copies/reaction for tdh and 10² copies/reaction for rpoD, trh1, and trh2 (Fig. 1D).

The LAMP-XO assay exhibited comparable DLs to standard LAMP for all 4 sets of primers. Compared to conventional PCR, its DLs were 10 times more sensitive for toxR, trh1, and trh2 and 100 folds greater for tdh. LAMP-XO also had equivalent DL to qPCR for rpoD but 10 times less sensitive for tdh, trh1, and trh2. This finding demonstrated that the LAMP-XO was as sensitive as standard LAMP and qPCR, but much more sensitive than PCR.

Comparative detection limit of LAMP-XO, standard LAMP, conventional PCR, and qPCR, assays using spiked shrimp samples

Prior to testing the analytical sensitivity of the assays, V. parahaemolyticus-negative shrimp samples were separately spiked with the reference strains at the concentrations of 10⁴–10⁰ CFU/2.5 g spiked shrimp. Amplification of V. parahaemolyticus gDNA was not observed by qPCR in a negative control (a non-spiked sample). The DL of LAMP-XO for rpoD, tdh, and trh2 was 10⁰ CFU/2.5 g spiked shrimp, and for trh1 primers was 10³ CFU/2.5 g spiked shrimp (Table 4, Fig. 2A, top). All of the colorimetric results were in accordance with the AGE results except for trh1 detection where AGE could detect down to 10² and 10¹ CFU/2.5 g spiked shrimp (Fig. 2A, bottom). However, no distinct positive (yellow) test results were observed at these dilutions. The detection limits (DLs) of standard LAMP (Fig. 2B) and qPCR assays (Fig. 2D) for all 4 sets of primers were 10⁰ CFU/2.5 g spiked shrimp whereas those of PCR were 10⁰ and 10¹ CFU/2.5 g spiked shrimp for toxR and tdh, respectively and 10³ CFU/2.5 g spiked shrimp for trh1 and trh2 (Fig. 2C).

LAMP-XO assay revealed comparable DLs to standard LAMP and qPCR for rpoD, tdh, and trh2 and 1,000 times less sensitive than that of qPCR for trh1. Compared to conventional PCR, limits of LAMP-XO detection were 10 times more sensitive for tdh and 1,000 times more sensitive for trh2. These findings were in accordance with those using a 10-fold serial dilution gDNA which demonstrated overall LAMP-XO was more sensitive than PCR whereas LAMP-XO sensitivity was comparable to those of standard LAMP and qPCR except for trh1.

LAMP-XO assay for the detection of total V. parahaemolyticus (pathogenic and non-pathogenic strains) and pathogenic V. parahaemolyticus (tdh⁺ and/or trh1⁺ and/or trh2⁺) in raw seafood samples

LAMP-XO assay efficacy was validated using 102 raw seafood samples purchased from fresh markets and supermarkets and collected from shrimp farms in Thailand and North Vietnam. The colorimetric results were compared with the results from PCR and qPCR. Using qPCR as a gold standard, LAMP-XO and PCR assays identified 76/102 and 74/102 samples positive for V. parahaemolyticus contamination, respectively (Table 5). Thus, the clinical sensitivity, specificity, PPV, NPV, diagnostic accuracy, and percent overall agreement (POA) were 100% for all categories for LAMP-XO, but were 97.4%, 100%, 100%, 96.4%, 98.4%, and 98.0%, respectively, for PCR. No statistically significant difference was observed for V. parahaemolyticus detection between qPCR and LAMP-XO and between qPCR and PCR (P > 0.05).

For pathogenic strain detection, 76 samples positive for V. parahaemolyticus by a gold standard qPCR were tested for tdh, trh1, and trh2 genes. LAMP-XO for tdh detection had sensitivity, specificity, PPV, and NPV of 90.5%, 100%, 100%, and 88%, respectively. The diagnostic accuracy and POA between LAMP and qPCR results were 94.4% and 97.4%, respectively (Table 6). PCR for tdh diagnosis showed 38.1% sensitivity, 98.2 specificity, 96.8% PPV, and 52.4% NPV. The diagnostic accuracy and POA between PCR and qPCR results were 62.7% and 81.6%, respectively (Table 6). No statistically significant difference was observed for tdh detection between qPCR and LAMP-XO (P > 0.05) while there were statistically significant differences between qPCR and PCR and between LAMP-XO and PCR (P < 0.05).

LAMP-XO for trh1 detection demonstrated 75% sensitivity, 100% specificity, 100% PPV, and 73.5% NPV. The diagnostic accuracy, and POA between LAMP-XO and qPCR results were 85.3% and 96.1%, respectively (Table 7). PCR for trh1 detection exhibited sensitivity, specificity, PPV, and NPV of 58.3%, 100%, 100%, and 62.5%, respectively. The diagnostic accuracy and POA between PCR and qPCR results were 75.4% and 93.4%, respectively (Table 7). No statistically significant difference was observed for trh1 detection between qPCR and LAMP-XO and between qPCR and PCR (P > 0.05).

LAMP-XO for trh2 detection exhibited 100% for sensitivity, specificity, PPV, NPV, diagnostic accuracy, and POA which were similar to those of qPCR. For trh2 detection, PCR revealed 55.6% sensitivity, 100% specificity, 100% PPV, and 61% NPV. The diagnostic accuracy and POA between PCR and qPCR results were 73.8% and 89.5%, respectively (Table 8). The statistically significant difference was observed for trh2 detection between qPCR and PCR and between LAMP-XO and PCR (P < 0.05).

Overall, LAMP-XO yielded results comparable to those of qPCR for rpoD, tdh, trh1, and trh2 detection since no statistically significant difference was observed between the 2 methods. Compared to PCR, LAMP-XO significantly demonstrated greater performance for tdh and trh2. For trh1, LAMP-XO was prone to have higher performance to PCR. However, no statistically significant difference was observed for trh1 detection between the 2 methods.

Distribution of pathogenic genes in V. parahaemolyticus-positive samples

Based on the results of a gold standard qPCR and the presence or absence of the tdh or trh1 or trh2 toxin genes, the 76 V. parahaemolyticus-positive samples could be classified into 8 groups: tdh⁺/trh1⁻/trh2⁻, tdh⁺/trh1⁺/trh2⁻, tdh⁺/trh1⁻/trh2⁺, tdh⁺/trh1⁺/trh2⁺, tdh⁻/trh1⁺/trh2⁻, tdh⁻/trh1⁺/trh2⁺, tdh⁻/trh1⁻/trh2⁺, and tdh⁻/trh1⁻/trh2⁻ (Table 9). Most of V. parahaemolyticus isolates were tdh⁻/trh1⁻/trh2⁻ (64.5%, 49/76) and the rest were pathogenic strains with tdh⁺ and/or trh1⁺ and/or trh2⁺. Among pathogenic strains (27 isolates), tdh⁺/trh1⁻/trh2⁺ strains were predominant (11 isolates) followed by tdh⁺/trh1⁺/trh2⁻ (6 isolates), tdh⁺/trh1⁺/trh2⁺ (3 isolates), tdh⁻/trh1⁻/trh2⁺(3 isolates), tdh⁻/trh1⁺/trh2⁻ (2 isolates), tdh⁺/trh1⁻/trh2⁻ (1 isolate), and tdh⁻/trh1⁺/trh2⁺ (1 isolate).