Rapid detection of extended spectrum β-lactamase producing Escherichia coli isolated from fresh pork meat and pig cecum samples using multiplex recombinase polymerase amplification and lateral flow strip analysis

The emergence and dissemination of extended-spectrum β-lactamase (ESBL)-producing Escherichia coli is a global health issue. Food-producing animals, including pigs, are significant reservoirs of antimicrobial resistance (AMR), which can be transmitted to humans. Thus, the rapid detection of ESBLs is required for efficient epidemiological control and treatment. In this study, multiplex recombinase polymerase amplification (RPA) combined with a single-stranded tag hybridization chromatographic printed-array strip (STH-PAS), as a lateral flow strip assay (LFA), was established for the rapid and simultaneous detection of multiple bla genes in a single reaction. Visible blue lines, indicating the presence of the blaCTX-M, blaSHV, and blaOXA genes, were observed within 10 min by the naked eye. The limit of detection of all three genes was 2.5 ng/25 μL, and no cross-reactivity with seven commensal aerobic bacteria was observed. A total of 93.9% (92/98) and 96% (48/50) of the E. coli isolates from pork meat and fecal samples, respectively, expressed an ESBL-producing phenotype. Nucleotide sequencing of the PCR amplicons showed that blaCTX-M was the most prevalent type (91.3–95.83%), of which the main form was blaCTX-M-55. The sensitivity and specificity of the RPA-LFA were 99.2% and 100%, respectively, and were in almost perfect agreement (κ = 0.949–1.000) with the results from PCR sequencing. Thus, the RPA-LFA is a promising tool for rapid and equipment-free ESBL detection and may facilitate clinical diagnosis in human and veterinary medicine, as well as AMR monitoring and surveillance.

Introduction collected, kept on ice, and transported to the laboratory for analysis within 24 h. Fifty pig cecum samples were obtained from a wholesale slaughterhouse in Suphanburi Province, Thailand, in February 2020. The history of antibiotic administration to the pigs was unknown. All samples were kept at 4˚C until processed.

Isolation of ESBL-producing E. coli
Each pork meat sample was gently chopped into small pieces, and then, 25 g of chopped pork was added to 225 mL of sterile peptone water and incubated at 37˚C overnight. One loop of the peptone culture was streaked onto MacConkey agar (Oxoid, UK) supplemented with 1 g/L cefotaxime (Oxoid, UK). Fecal material from cecum samples from each pig was collected using a sterile cotton swab and directly inoculated on MacConkey agar supplemented with 1 g/L cefotaxime. Three typical lactose fermented (LF) colonies were randomly selected from each sample, screened for E. coli on eosin methylene blue agar (Oxoid, UK), and identified by conventional biochemical testing, including MIO, TSI, LIA, citrate, and urea utilization [22]. If all the isolates from the same sample were designated as E. coli by the same biochemical results, then a single isolate was randomly selected.
All the selected E. coli isolates were subsequently confirmed by PCR amplification of UspA using E. coli-specific primers (Table 1). PCR included 0.05 μM of each primer, 200 μM dNTPs, 0.625 U Taq polymerase (NEB, Inc., USA), and 50 ng DNA template in a total volume of 25 μL. The PCR amplification conditions were as follows: 94˚C for 5 min, 35 cycles of 94˚C for 30 sec, 60˚C for 15 sec, 72˚C for 30 sec, and 72˚C for 5 min.

Characterization of ESBL genes by PCR-sequencing
Four common ESBL genes (bla CTX-M , bla OXA , bla SHV , and bla TEM ) were screened in all the bacterial isolates by PCR and confirmed by DNA sequencing. Each obtained sequence was subsequently compared to known genotype sequences in the NCBI database using the BLASTn program. In brief, whole-cell DNA from the bacterial isolates was prepared by the boiling method. PCR was subsequently performed using gene-specific primers (Table 1)

Multiplex RPA-LFA
A TwistDx basic kit (TwistDx, UK) was used for the amplification of the bla CTX-M , bla OXA , and bla SHV genes according to the manufacturer's instructions. The sequences of the specific tag-primers used in the RPA reactions are shown in Table 1 The purified RPA products (uniplex, duplex and triplex RPA) were detected using a singlestranded tag hybridization chromatographic printed-array strip (STH-PAS; TBA Co., Ltd., Japan). Briefly, 10 μL of 300 mM NaCl-modified dilution buffer and 1 μL of streptavidin-coated blue latex were added to 10 μL of the RPA product. The C-PAS4 membrane stick was dipped into the mixture for 5-10 min. The uniplex:duplex RPA products were evaluated at ratios (v/v) of 1:4, 2:4, 3:4 and 4:4. Each visible blue line indicated the presence of the respective ESBL target gene, which were from the bottom to top lines: bla CTX-M , bla SHV , and bla OXA (Fig 1).

Detection of the bla CTX-M , bla OXA , and bla SHV genes in ESBL-producing E. coli by the RPA-LFA
All the ESBL-positive E. coli isolates from the pork, and cecum samples were screened for the presence of bla CTX-M , bla OXA , and bla SHV by the RPA-LFA. The TwistDx basic kit and STH-PAS dipstick were employed using the conditions determined in the present study.

Data analysis
The results of the RPA-LFA and PCR detection were compared, and the sensitivity and specificity of the RPA-LFA were analyzed using Medcalc1 software (https://www.medcalc.org/ calc/diagnostic_test.php). The level of agreement between the two diagnostic methods was evaluated for significance using kappa (κ) statistics with QuickCals software (https://www. graphpad.com/quickcalcs/kappa1) [28].

Antimicrobial susceptibility patterns of E. coli isolated from retail pork meat and pig cecum samples
From the 100 pork samples individually collected from 100 retail markets, a single E. coli isolate was selected from 68 positive samples. Two and three isolates designated as E. coli with different biochemical results were selected from 9 and 4 samples, respectively. Thus, a total of 98 E. coli isolates were obtained from 81 pork samples. Of these screened E. coli isolates, 93.9% (92/98) were ESBL producers, and three isolates were resistant to carbapenem (Fig 2A).
From the fecal samples, a single E. coli isolate was selected from each of the 50 fecal samples collected from the slaughterhouse, and 96% (48/50) were ESBL producers ( Fig 2C). Almost all the ESBL-producing isolates (92.14%) were found to display MDR, and they were resistant to at least one agent in three or more antimicrobial categories. The ESBL-MDR isolates remarkably resisted trimethoprim-sulfamethoxazole, gentamycin, tetracycline, and chloramphenicol ( Fig 2B and 2D). Notably, a markedly higher percentage of ESBL-MDR isolates were found in the pig cecum samples (87.5-100%) than in the pork samples (45.6-88.0%).

PLOS ONE
Extended spectrum β-lactamase and RPA-LFA were determined as obvious blue lines. The LOD for all three genes was found to be 2.5 ng/ 25 μL (Fig 6).

Comparison of the RPA-LFA and PCR for the detection of the bla CTX-M , bla OXA , and bla SHV genes
To examine the efficiency of the RPA-LFA, a total of 140 ESBL-producing E. coli isolates (92 ESBL-positive pork isolates and 48 ESBL-positive fecal isolates) were tested by the RPA-LFA developed in the present study, and the results were compared to the results obtained by screening the same isolates by diagnostic PCR. Of 130 bla CTX-M PCR-positive samples, a discrepancy was revealed in one of the bla CTX-M , bla TEM samples by RPA-LFA (Table 3). According to agreement analysis, both methods showed completely matched results, with 129 positive and 10 negative bla CTX-M samples and two positive and 138 negative bla SHV and bla OXA samples, except one false-positive result (compared to the PCR) detected for bla CTX-M by the RPA-LFA (Tables 3 and 4). The otherwise concordant results of both methods provided Table 2

Discussion
Due to some of the limitations of phenotypic detection methods and the nonportability, expense, and technical skill required for the available advanced genotypic assays, the development of a more rapid and equipment-free system for the detection of ESBL genes is required. The aim of this study was to develop a rapid and simple diagnostic test for the most common ESBL genes encoding CTX-M, OXA, and SHV using an RPA-LFA method. The developed

PLOS ONE
Extended spectrum β-lactamase and RPA-LFA RPA could rapidly amplify bla CTX-M , bla OXA , and bla SHV at 37˚C in 30 min with easily interpreted visual results by the naked eye on lateral flow strips. Although, our RPA-LFA utilized two separate duplex (bla CTX-M , bla OXA ) and uniplex (bla SHV ) RPA reactions , the amplification was simple and rapid with no need for an expensive thermal cycler. Moreover, there is no need for gel preparation and staining, which shortens the detection time and decreases the amount of cumbersome processing. Multiplex isothermal amplifications are different among different strategies. Detection by multiplex loop-mediated isothermal amplification is difficult to perform because of the complicated structure of the primers and the recommended amplicon size of < 250 bp. Only a few multiplex helicase-dependent amplifications have been reported because the short amplicon size of < 150 bp is a limitation. Likewise, strand displacement amplification requires at least four primers/target and restriction enzyme digestion, which obstructs multiplexing [31]. The benefits of RPA are its low operating temperature (37-42˚C),

PLOS ONE
which is compatible with general heating instruments in the laboratory or even body temperature; its lack of requirement for tight temperature control and initial denaturation; and its rapid amplification time of within 5-25 min. The recombinase typically consumes all the available ATP within 25 min [32]. Although, our multiplex RPA primers were shorter (20-26 bp) than the recommended length of 30-35 bp, amplification of the bla CTX-M , bla OXA , and bla SHV genes was still accomplished and generated the product sizes from 296-593 bp. Longer primers with higher GC contents increase the probability of primer-dimer formation, while the optimal RPA amplicon size ranges from 100-200 bp. Longer products require a longer amplification time as well as an optimum ratio and concentration of primers [32]. Duplex and uniplex RPA products combined at a 2:4 (v/v) ratio provided specific detection of the three target genes on a lateral flow strip. Obvious blue lines for the specific RPA products corresponding to the bla CTX-M , bla SHV , and bla OXA genes appeared within 10 min. A false SHV result was demonstrated in the triplex RPA-LFA, which might have been due to the biotin primers self-pairing with the bla SHV -tag primers and subsequently hybridizing with the immobilized complementary oligonucleotides on the strip. The LOD was 2.5 ng/25 μL with no cross-reaction against seven common human pathogens. The LOD of the RPA-LFA depends on several factors, including the amplification efficiency, the multiplex level of each primer, self-pairing among primers, and the different (biased) amplification efficiencies of the different primers. Primer design and extensive optimization of multiplex nucleic acid amplification incorporated with LFA assays are critical steps to avoid primer dimers and nonspecific results [31]. The coupling of powerful amplification methods with simple detection platforms, especially lateral flow strips, increases the feasibility of their application in point of care or field detection [33]. Of the various LFA formats, the STH-PAS is a lateral flow dipstick format that allows the detection of multiple targets in a single reaction and easy visualization of the results in a short time by the naked eye. The optimum condition for STH-PAS detection depends on the stringency of the developing buffer, the ratio of products to developing buffer, and incubation time. For isothermal amplification, the reagent's instructions recommend using a modified dilution buffer and diluting the template five-to tenfold with TE buffer or water (TBA Co., Ltd.). The sensitivity and specificity of our RPA-LFA for the detection all three common ESBL genes in pork and pig cecum samples were 99.2-100% and in almost perfect agreement (κ = 0.949-1.00) with the PCR results. Multiplex PCR combined with STH-PAS has shown high sensitivity (93.3%) and specificity (99.1%) for the direct detection of carbapenemase genes from stool specimens within 2 h [34,35]. The transmission of the mcr-1 gene and carbapenem-resistant Enterobacteriaceae has been determined using a PCR-STH-PAS dipstick and pulse-field gel electrophoresis [35]. However, no multiplex RPA combined with STH-PAS has been reported before. A combination of other common drug resistance genes, including bla-TEM , plasmid-mediated AmpC beta lactamase genes, carbapenemase genes, and internal control lines, in one strip remains a challenge.
Overall, E. coli was found in 81% of the pork samples obtained from 100 retail markets in Bangkok, Thailand. This high level of contamination with E. coli is not only an indicator of poor hygiene but also indicates the distribution of antibiotic resistance in foods. Out of 98 E. coli isolates from pork that were tested, 93.9% were ESBL positive, and three isolates with carbapenem resistance were observed. Similarly, intestinal carriage of ESBL-positive E. coli in pigs was 96% in the tested slaughterhouse, which provides wholesale pork products throughout fresh markets in Bangkok and nearby provinces. High MDR profiles existed mostly in ESBLpositive E. coli, which implies a crisis in antibiotic usage. ESBL genes are located on plasmids, which are usually coexpressed with other plasmid-mediated drug resistance genes, such as quinolone, aminoglycosides, trimethoprim, and tetracyclines [36,37]. Similar to a previous report, a high incidence of ESBLs with resistance to other drug classes was observed among food animals and fresh meats in Thailand [11,38]. The presence of ESBL-MDR pathogens is a serious health problem due to the limited available drug regimens for treatment. Food producers, including farms and slaughterhouses, are important transmission sources of antibiotic resistance [39,40].
The frequency of CTX-M was high in both pork (91.3%) and pig cecum samples (95.8%), with the majority being CTX-M-55. Only four subtypes (CTX-M-55, CTX-M-14, CTX-M-15, and CTX-M-27) were detected in the pig fecal samples from the slaughterhouse, while 14 CTX-M subtypes were detected in raw pork samples from shops that came from various slaughterhouses. A high prevalence of CTX-M-55 in pork has been reported previously in many countries, including Cambodia, Hong Kong, and Vietnam [41][42][43]. A total of 212 ESBLproducing E. coli isolated from healthy subjects in the community and swine of Lamphun Province, Thailand, revealed that 95.8% were positive for CTX-M, with the most common subtype being CTX-M-55, followed by TEM (60.9%) and SHV (2.4%) [44]. The emergence of CTX-M-55 has also been observed in Salmonella isolates from raw meat and food animals in China and Cambodia [42,45]. However, the prevalence of ESBL subtypes varies in each geographical and epidemiological area. Antibiotic management is required in the whole production system, from farms to consumers. Moreover, the consumption of cooked meats is safer from food spoilage organisms and has a lower risk of antibiotic transmission than raw foods.

Conclusion
The RPA-LFA (using STH-PAS) is a rapid and reliable tool for the detection of ESBL-producing E. coli in pork and fecal samples. RPA amplification can be performed using simple heating equipment, and the results can be easily examined by the naked eye on a lateral flow strip. Thus, the assay is convenient for ESBL detection in food animals, patients, and environmental samples in low-resource settings.