Edwardsiella piscicida DC1 was isolated from diseased turbot (Scophthalmus maximus) reared in a commercial fish farm located in Qingdao, Shandong province, which also showed high pathogenicity to Lateolabrax maculatus. Another eight clinically-common pathogens were evaluated in this study, including Pseudomonas fluorescens, Leclercia adecarboxylata, Vibrio rotiferianus, Vibrio scophthalmi, Vibrio alginolyticus, Vibrio hyugaensis, Haemophilus piscium and Vibrio azureus, which were all isolated and stored in our lab. All strains were verified by 16S rDNA gene sequencing in Sangon Biotech Co., Ltd. (Shanghai, China). All strains were cultured in a Luria-Bertani broth (LB) medium at 28°C.

For clinical specimen preparation, 30 clinical samples were prepared. Briefly, clinically healthy L. maculatus (average weight 14.7 g) were purchased from a commercial fish farm located in Yantai, Shandong province. Before experiment, fish were randomly sampled for the examination of E. piscicida presence by plate counts. For experimental infection, E. piscicida DC1 were cultured in LB medium to an OD₆₀₀ = 0.8, L. maculatus were challenged via intraperitoneal (i.p.) injection with 100 μl of E. piscicida DC1 (10⁷ cfu/ml) and maintained at 22°C. At 0, 12, 24, 36, and 48 h post-challenge (hpc), the liver, spleen, and kidney were obtained aseptically from the fish (three fish/time point). For E. piscicida detection, the tissues obtained above were homogenated in PBS, spread onto the LB agar plates, and incubation at 28°C for 48 h, furthermore, the observed bacterial colonies were verified by 16S rDNA sequencing (Zhang et al., 2014). The samples were further confirmed by quantitative real-time PCR as reported previously (Reichley et al., 2015). A total of 24 positive and 6 negative fish samples were prepared for further RT-RAA analysis. For environmental samples preparation, nine sea water and nine sediment samples were collected from the intertidal zone of Yantai, China. Before experiment, these samples were randomly sampled for the examination of E. piscicida presence by plate counts. Different concentrations of E. piscicida cultured above were added, and verified as above. A total of 12 positive (6 sea water and 6 sediment) and 6 negative samples (3 sea water and 3 sediment) were prepared for further RT-RAA analysis.

For bacterial DNA isolation, 1 ml sample of pure cultured bacterial suspension was obtained. The DNA extraction was performed using the TIANamp Bacteria DNA Kit (Tiangen Biotech Co, Ltd, China) following to the manufacturer’s instructions. For clinical samples’ DNA extraction, 50 mg samples were homogenized in 200 μL PBS and the total DNA was extracted using CTAB method (Guerra et al., 2020). The DNA were eluted in 100 μL of nuclease-free water and stored at a temperature of -20°C until utilized.

The EvpP gene was selected as a target gene for detecting E. piscicida. The EvpP gene sequence was obtained from E. piscicida DC1, and submitted into the GenBank (accession no. OR670479.1). Using primers EvpP-F1/R1 ( Table 1 ), we amplified a 591 bp fragment of the EvpP gene and inserted it into the pMD-19T vector (TAKARA, Dalian, China). The recombinant plasmid was confirmed by sequencing in Sangon Biotech Co., Ltd. and named pEvpP. The recombinant plasmid’s concentration was quantified using Nano-500 micro-spectrophotometer (Allsheng, Hangzhou, China) and the DNA copy number was determined using the formula: DNA copy number (copies/μL) = [6.02 × 10²³ × plasmid concentration (ng/μL) × 10⁻⁹]/[DNA length × 660]. Then, the standard recombinant plasmids diluted in a 10-fold manner ranging from 2.1×10⁸ to 2.1×10⁰ copies/μL and stored at −20°C until further use.

Other nine EvpP genes of different E. piscicida strains were obtained from the National Center for Biotechnology Information (NCBI) database. The DNAMAN software was utilized to conduct multiple sequence alignment in order to identify the conserved regions of the EvpP gene. The primers and probe for RT-RAA were designed within the highly-conserved regions using an online tool “Primer & probe design for RPA/RAA” (https://ezassay.com/primer), adhering to the guidelines for designing RT-RAA primers and probe. To ensure specificity, a Primer-BLAST search was performed against the NCBI database to verify the sequences the primer and probe. Furthermore, the possibility of primer dimers and hairpins was evaluated by utilizing the online OligoEvaluator software (http://www.oligoevaluator.com). Finally, the primer and probe were synthesized and purified using high-performance liquid chromatography in Sangon Biotech Co., Ltd.

First, an ideal probe (p308–358) was designed ( Table 1 , Figure 1 ). Then, six candidate primer pairs (F264–293/R405–434, F265–294/R404–433, F266–295/R403–432, F267–296/R402–431, F267–297/R401–431, F268–297/R401–430) were designed ( Supplementary Table S1 ; Figure 1 ). The RT-RAA assay was performed with the above primers and the probe, and the primer pair with the highest end fluorescence intensity value (RFU value) and better “S” amplification curve was selected for further analysis. All reactions were performed three times.

A commercial RAA fluorescence method kit (Zhongce, Hangzhou, China) was used to perform the RT-RAA assay in a reaction volume of 20 μL according to the manufacturer’ s instructions. Each primer and probe described above was diluted to a final concentration of 10 μM. The reaction mixture contained 2.0 μL of DNA template that was extracted, 10 μL of reaction buffer, 5.56 μL of DNase-free water, 0.4 μL of each primer, 0.24 μL of probe, 1.0 μL of magnesium acetate and reactive dry powder. Following the addition of all remaining components to the reactive dry powder, magnesium acetate is subsequently introduced to the tube lids. Subsequently, the tube lids are securely closed and the reaction tubes are subjected to gentle agitation by inverting them 5-6 times, followed by centrifugation at a low speed for a duration of 10 seconds. Afterward, the resulting tubes are transferred to the Bio-rad CFX96 Touch Real-Time PCR Detection System (Bio-rad, Hercules, CA, USA) and maintained at a temperature of 39°C for 20 minutes. Each experimental run includes both a positive control (pEvpP plasmid) and a negative control.

The qPCR assay was performed in 20 μL reaction volume using Pro Taq HS Premix Probe real-time PCR Kit III (Accurate Biotechnology Co., Ltd, Hunan, China) according to the manufacturer’ s instructions. The primers (F3/R3) and probe (Probe-qPCR) for aPCR assay were listed in Table 1 . In the 20 μL reaction mixture, there was 2.0 μL of extracted DNA template, 10 μL of 2 × Pro Taq HS Probe Premix III, 4.4 μL of DNase-free water, 1.0 μL of each primer F3/R3 (10 μM), 1.6 μL of Probe-qPCR (10 μM). Next, the reactions were examined on the Bio-rad CFX96 Touch Real-Time PCR Detection System (Bio-rad, Hercules, CA, USA). The qPCR procedure involved an initial denaturation step at a temperature of 95°C for a duration of 30 s, followed by 40 cycles of amplification (with a denaturation step at 95°C for 5 s and an annealing step at 60°C for 30 s). Every trial includes both a positive control (pEvpP plasmid) and a negative control.

To assess the specificity of the RAA assay, eight additional bacterial pathogens were included: P. fluorescens, L. adecarboxylata, V. rotiferianus, V. scophthalmi, V. alginolyticus, V. hyugaensis, H. piscium and V. azureus. The genomic DNA of these bacteria were extracted as described above and then utilized as the template. The specificity test was performed under the aforementioned conditions and system, and repeated more than three times with each template and negative controls.

To determine the sensitivity of the RAA assay, plasmid pEvpP was diluted in 10-fold serial dilutions ranging from 2.1×10⁵ to 2.1×10⁰ copies/μL, with six replicates for each dilution. For comparison, the same templates were tested in parallel using the qPCR assay. The sensitivity assays were carried out utilizing the aforementioned conditions and systems, and were repeated more than three times with each template and negative controls.

In order to assess applicability of the RT-RAA assay for E. piscicida detection in clinical and environmental samples, bacterial DNA were extracted from a total of 48 samples. Among these samples, 30 were clinical samples (24 positive and 6 negative), 9 were sea water samples (6 positive and 3 negative), and 9 were sediment samples (6 positive and 3 negative). Subsequently, the RAA assay was conducted on all 48 samples. For comparative purposes, the same samples were also subjected to the qPCR assay for E. piscicida in parallel.

Probit analysis for the detection limit of the RT-RAA assay and qPCR assay was performed at a 95% probability level. The kappa and p values of the RT-RAA and qPCR assays were calculated. All statistical analysis was carried out with SPSS 21.0 (IBM, Armonk, NY).

Ten EvpP gene sequences of E. piscicida were obtained from the GenBank database. Based on the multiple sequence alignment analysis, conserved nucleotide residues were identified and visually emphasized with a black background. In order to design primers and probe, 10 EvpP sequences were aligned with F2, R2, and Probe-RAA primers. The sequence alignment revealed high conserved regions for primers and probe design ( Figure 1A ). After primer screening, the optimal primer pair F268–297/R401–430 was screened out according to the amplification curve and renamed F2/R2 ( Figure 1C , Table 1 ).

The RAA assay showed a positive result for E. piscicida and negative results for P. fluorescens, L. adecarboxylata, V. rotiferianus, V. scophthalmi, V. alginolyticus, V. hyugaensis, H. piscium and V. azureus ( Figure 2 ). The results showed that the established RT-RAA assay exhibited a considerable degree of specificity towards its intended targets, with no cases of cross-reactivity observed with other prevalent marine pathogenic bacteria.

The pEvpP plasmid, which was serially diluted from 2.1×10⁵ to 2.1×10⁰ copies/μL, was subjected to testing using the RT-RAA assay. The results indicated that consistent signals were observed within the range of 2.1×10⁰ and 2.1×10⁵ copies/μL for each reaction. As a result, the detection limit of the RAA assay was determined to be 42 copies (0.14 fg) per reaction with a 95% probability ( Table 2 , Figure 3 ) (Probit analysis, p=0.015), which aligns with the detection limit of 42 copies per reaction for the qPCR method ( Table 2 , Figure 3 ) (Probit analysis, p=0.02).

In this RT-RAA assay, a total of 48 samples, consisting of 24 positive clinical samples, 6 positive sea water samples, 6 positive sediment samples, and 12 negative samples (6 clinical samples, 3 sea water samples, 3 sediment samples), were utilized, and the results were comparable with those obtained by qPCR. Using qPCR, 36 out of the 48 samples were identified as positive. The RT-RAA assay successfully detected and distinguished all 36 positive samples with complete accordance, sensitivity, and specificity ( Table 3 , Figure 4 ). Notably, no significant disparities were observed between the detection outcomes of the RT-RAA assay and qPCR. The RT-RAA assay yielded a kappa value of 1.0 (p < 0.001).