CRISPR 2 PCR and high resolution melting profiling for identification and characterization of clinically-relevant Salmonella enterica subsp. enterica

Isolation and identification of Salmonella isolates in stool of diarrheal patients

Stool samples were collected from patients (n = 26) with diarrhea during May 2016 to July 2017 at Phayao Ram Hospital, Phayao province, Thailand. Diarrhea is defined as patients showing related symptoms of fecal incontinence diagnosed by a physician and participants signed the consent forms after brief explanation of the research project. Samples were incubated in buffered peptone water (Oxoid, Hampshire, UK) at 37 °C overnight, plated on SS and XLD agar (Oxoid, Hampshire, UK), and suspected Salmonella colonies indicated by using a triple sugar iron (TSI) slant (Oxoid, Hampshire, UK) assay and lysine iron agar (LIA) (Biomedia, Nontanuri, Thailand). Conventional biochemical tests (glucose fermentation, lactose oxidation, gas and H₂S production, urease assay, methyl red staining, and indole, motility and Voges-Proskauer tests) were performed according to ISO 6579:2002.

The study protocols were approved by the Ethical Committee of Phayao University (no. 57 02 04 0020).

Determination of β-lactam antibiotic resistance profile and ESBL phenotype

Susceptibility to β-lactam antibiotics was performed using a disk diffusion method according to Clinical and Laboratory Standards Institute (CLSI) guidelines (CLSI, 2016) using ampicillin (AMP, l0 µg), cefotaxime (CTX, 30 µg), ciprofloxacin (CIP, 5 µg), ertapenem (ETP, 10 µg), and gentamycin (GEN, 10 µg) discs (Oxoid, Hampshire, UK), with Escherichia coli ATCC 25922 as control. ESBL phenotype was evaluated using a double-dish method (CLSI, 2016) employing CTX alone and in combination with clavulanic acid (10 µg) (Oxoid, Hampshire, UK), with in-house known ESBL-producing and -negative Escherichia coli strains as controls. A rapid ESBL was slightly modified from the ESBL NDP (Nordmann, Dortet & Poirel, 2012), which is based on observation of color change due to acid production of ESBL-producing bacteria. Salmonella isolate was inoculated in one ml of NB broth (Oxoid, Hampshire, UK) and incubated at 37 °C overnight, each experiment conducted in duplicate. Then culture was centrifuged at 13,000 g for 2 min, pellet washed twice with 700 µL of 1X 10 mM Tris HCl pH 8.0 containing 1 mM EDTA (TE) buffer and mixed with 100 µL of lysis buffer [(B-PERII Bacterial Protein Extraction Reagent)] (Pierce/Thermo Scientific, Villebon-sur-Yvette, France), followed by 100 µL of revelation solution (0.05% (w/v) phenol red) with and without 6 g/ml cefotaxime and incubation at 37 °C for 10 min; yellow color in the former indicating positive result.

Determination of Salmonella serotypes by HRM-PCR

DNA was extracted from Salmonella isolates as previously described (McNerney et al., 2017). In brief, one ml aliquot of an overnight culture was sedimented as described above, washed twice with 400 µL of TE buffer, resuspended in the same volume of TE buffer, incubated at 80 °C for 20 min, and cooled to ambient temperature. Then, a 50 µL aliquot of lysozyme solution (10 mg/mL) was added and the solution incubated at 37 °C for one hour with occasionally shaking, followed by addition of 75 µL of 10% SDS/proteinase K (10 mg/mL) solution, vigorous vertexing and incubation at 65 °C for 10 min. Following addition of 100 µL of 5 M NaCl and 100 µL of prewarmed (65 °C) 10% N-cetyl-N,N,N,-trimethyl ammonium bromide (CTAB)/ 5 M NaCl solution, the mixture was further incubated at 65 °C for 10 min, followed by addition of 750 µL of chloroform : isoamyl alcohol (24 :1) and centrifugation at 11,000 g at 4 °C for 5 min. DNA in the upper aqueous was precipitated with ethanol, resuspended in 50 µL of double-distilled water and stored at −20 °C until used.

Multiplex HRM-PCR was performed using a combination of primers to amplify fljB (170 bp), gyrB (171 bp) and ycfQ (241 bp) (Table 1). HRM-PCR mixture (10 µL) contained 1 µL of DNA, 0.1 pmol of gyrB, 0.075 pmol of fljB and 0.075 pmol of ycfQ primer pairs and 2 µL of HOT FIREPol EvaGreen: no ROX Mix (Solis Biodye, Tartu, Estonia). Thermocycling was performed in a BIO-RAD CFX96™ Real-Time System (Bio-Rad, Hercules, CA, USA) as follows: 95 °C for 12 min, followed by 40 cycles of 95 °C for 10 s, 60 °C for 10 s and 72 °C for 20 s. Samples were then heated at 95 °C for 1 min, cooled to 40 °C for 1 min and then heated from 70 to 95 °C at 0.2 /s, with 25 fluorescence data acquisitions/°C. HRM profiles were generated using a Precision Melt Analysis software V 1.2 (BIO-RAD, Hercules, CA, USA) with sensitivity setting at 0.30, temperature shift at threshold 5, pre-melt normalization range from 80.87 to 81.51 °C, and post-melt normalization range from 87.17 to 87.92 °C. Following normalizing and temperature shifting, difference plots were generated relative to HRM profile of S. Bareilly (as baseline).

Hierarchical clustering of HRM curves using dynamic time warping (DTW) algorithm

A dendrogram of normalized HRM curves was constructed using a DTW algorithm to determine distance measurements (Lu et al., 2017); the entire dendrogram construction was performed in Python programming language (Van Rossum, 1995). In short, a smooth spline approximation was determined from each normalized HRM curve using cubic splines of splrep function in scipy module (Jones, Oliphant & Peterson, 2001) followed by a rate curve calculated from negative first derivative of the resulting spline employing a splev function. The curve was then z-normalized using a zscore function to calculate DTW distances, in which l$\stackrel{ˆ}{}$2-norm is the distance function (only between 80 and 94 °C). Hierarchical clustering based on DTW distances was performed based on a neighbor-joining method to generate a dendrogram.

DNA profiling by ERIC PCR

ERIC-PCR mixture contained 0.2 µL of extracted Salmonella DNA, primer set (Table 1), 2 µL of HOT FIREPol Blend Master Mix Plus 10 mM MgCl₂ (Solis Biodye), and double-distilled water to make 10 µL. Thermocycling was performed in Veriti Thermal Cycler, Applied Biosystems (Thermo Fisher Scientific, Waltham, MA, USA) as follows: 95 °C for 15 min; followed by 30 cycles of 95 °C for 1 min, 54 °C for 2 min and 72 °C for 4 min; and a final step at 72 °C for 7 min. Amplicons were separated by 4% agarose gel-electrophoresis, stained with RedSafe dye (INiRON, Kirkland, WA, USA) and recorded with a Molecular Imager Gel DOC™ XR+ (Bio-Rad, Berkeley, CA, USA) equipped with an Image Lab™ software as JPEG images at 300 dpi resolution.

Analysis of ERIC PCR amplicon profile and phylogenetic tree

Amplicon patterns generated by ERIC-PCR were analyzed and employing a curve-based algorithm (Pearson correlation) together with an open-source GelJ software (Heras et al., 2015) to create a similarity scale, and analysis of constructed phylogenetic tree clusters was performed by an unweighted pair-group using arithmetic averages algorithm (UPGMA).

CRISPR 2 uniplex and HRM-PCR assays of Salmonella isolates

Salmonella isolates (n = 55) included, in addition, previously characterized Salmonella spp. [S. 4,[5],12:i:- (n = 15), S. Typhimurium (n = 6) and other Salmonella serotypes (n = 8)] (Poonchareon et al., 2019a; Poonchareon et al., 2019c). CRISPR 2 uniplex PCR mixture (10 µL) contained 1 µL of DNA, primer set (IDT, Singapore) (Table 1) and 2 µL of HOT FIREPol Blend Master Mix Plus 10 mM MgCl₂ (Solis Biodye). Thermocycling was carried out as described above but using the following conditions: 95 °C for 12 min; 35 cycles of 94 °C for 60 s, 59 °C for 60 s and 72 °C for 90 s; and a final step at 72 °C for 7 min. Amplicons were analyzed and recorded as described above using 2% agarose gel-electrophoresis. Amplicons of S. 4,[5],12:i:- isolates from this study and those previously characterized (nos. 1, 23, 25, 35, 56, 76, 142, 152, 157, and 249) (Poonchareon et al., 2019a; Poonchareon et al., 2019c) were purified from PCR clean up & gel extraction kit; GeneDirex (Bio-Helix, Keelung City, Taiwan) and sequenced by First BASE Lab (Seri kembangan, Selangor, Malaysia). DNA sequences were aligned and compared with a BioEdit software package (https://bioedit.software.informer.com).

CRISPR 2 HRM-PCR mixture (10 µL) contained 1 µL of DNA, 2 pmol of B1 and 2 pmol of B2 (IDT) (Table 1), and 2 µL of HOT FIREPol EvaGreen: no ROX Mix (Solis Biodye) and thermocycling was performed as described above but under the following conditions: 95 °C for 12 min, followed by 35 cycles of 95 °C for 60 s, 59 °C for 60 s and 72 °C for 90 s. Samples were then heated at 95 °C for 1 min, cooled to 40  °C for 1 min and HRM profiles generated and analyzed as described above but using HRM profile of S. Agona as baseline. Clustering of S. 4,[5],12:i:- isolates was achieved with a pre-melt normalization range of 87.7–88.2 °C and a post-melt normalization range of 91.5–91.7 °C, with S. 4,[5],12:i:- isolate 76 as a reference isolate.

Characterization of CRISPR 2 amplicon sequences

CRISPR 2 DNA sequences in fasta format were uploaded to https://crisprcas.i2bc.paris-saclay.fr/CrisprCasFinder/Index with default setting. Information of CRISPR 2 region (molecular signatures of CRISPR 2 locus), e.g., DR consensus sequence, DR length and number of spacers, were provided directly in the form of DNA sequences or numeral data.

Identification of bla alleles

Reaction mixture (10 µL) for multiplex PCR of bla alleles of interest contained 1 µL of DNA, bla primer sets (IDT) (Table 1) and 2 µL of HOT FIREPol Blend Master Mix Plus 10 mM MgCl₂ (Solis Biodye) and thermocycling was conducted as described above but using the following conditions: 95 °C for 12 min; 30 cycles of 95 °C for 40 s, 60 °C for 40 s and 72 °C for 60 s; and a final step at 72 °C for 7 min. Amplicons were analyzed (using 1.5% agarose gel) and recorded as described above.

Salmonella serotypes of clinical isolates from stools of diarrheal patients, Phayao Ram Hospital, Phayao province

Multiplex HRM-PCR of Salmonella fljB, gyrB and ycfQ amplicons from 26 patients, stool sample isolates generated 11 distinct HRM profiles (HRM_1-11) with HRM_1 predominant (n = 10, indicative of S. 4,[5],12:i:-), followed by HRM_6 (n = 4, indicative of S. Enteritidis) and HRM_8 (n = 3, indicative of S. Stanley) (Fig. 1). HRM_5 profile (n = 2) was similar to those previously identified for Salmonella serotypes S. Agona, S. Corvallis, S. Derby, and S. Kedougou (Poonchareon et al., 2019b). The other seven HRM profiles were not characterized and considered as indicative of rare or unknown Salmonella serotypes.

ERIC PCR was applied to characterize genetic relatedness of S. Enteritidis (HRM_6), HRM_2, _3, _4, _5, _7, _9, _10, and _11 Salmonella isolates. Phylogenetic tree constructed from ERIC PCR amplicon sizes revealed these Salmonella isolates clustered into three distinct clades (Fig. 2). Salmonella isolates 378 and 454 (HRM_5) were in clade 1 and assigned S. Weltevreden. All S. Enteritidis isolates (HRM_6) were grouped in clade 2, and Salmonella isolates with serotypes unidentified from multiplex HRM-PCR profiles were grouped together in clade 3 and still remained untyped except for Salmonella isolate 412 (HRM_4) that was assigned S. 4,[5],12:i:-.

Phylogenic tree based on multiplex HRM-PCR of clinical Salmonella isolates and association with other characteristics

A DTW algorithm was applied to transform HRM-PCR profiles into numerical data to allow construction of a phylogenetic tree (Lu et al., 2017), resulting in grouping of the 26 clinical isolates into six clades (Fig. 3). S. 4,[5],12:i:-, S. Enteritidis and S. Stanley was clustered in clade 2, 4 and 6, respectively while HRM types of unidentified serotypes were mainly distributed in clade 3 and 5 except HRM_5 (S. Weltevreden) and isolate 412 (HRM_4, S. 4,[5],12:i:-) was in clade 1 and 3 respectively. It is worth noting that S. 4,[5],12:i:- isolates were clustered in clades 2 (n = 8) and 3 (n = 1), but three were unassigned. Ampicillin resistance was predominantly associated with blaTEM, and two ESBL-producing Salmonella isolates (based on both double-dish and the rapid ESBL assay) harbored blaCTX group 1. One ESBL-producing S. 4,[5],12:i:-, isolate 412 (assigned by ERIC-PCR, Fig. 2) demonstrated a multidrug-resistant phenotype, including CIP resistance.

CRISPR 2 uniplex and HRM-PCR analyses

In order to simplify Salmonella typing based on multiplex PCR of fljB, gyrB and ycfQ, uniplex PCR of CRISPR 2 locus was developed. The 26 clinical Salmonella isolates each generated a single amplicon, size of which was indicative of serotype when compared with known Salmonella serotypes, namely, 830 bp of S. Enteritidis and 1,700 bp of S. 4,[5],12:i:- (Fig. 4). Salmonella isolate 412, shown by ERIC-PCR to be S. 4,[5],12:i:-, generated an amplicon of 1,700 bp. However, samples 378 and 454 (S. Weltevreden) and 312, 413 and 495 (S. Stanley) failed to demonstrate amplicons of consistent sizes. Amplicon sizes of the remaining samples did not correspond to the three Salmonella serotype standards. Thus, CRISPR 2 uniplex PCR, although simpler to perform, had limited ability in providing definitive identification of Salmonella serotypes based on amplicon sizes.

HRM temperature-shifted difference profiles (relative to S. 4,[5],12:i:- isolate 76) of CRISPR 2 uniplex amplicons of the 11 S. 4,[5],12:i:- isolates from this study and other previously identified samples produced three different profiles (clusters 1-3) (Fig. 5). Cluster 1 contained five known human and four from minced pork samples, cluster 2 four isolates from this study and three known human isolates and cluster 3 seven isolates from this study (including Salmonella isolate 345 and 392 that were not grouped in any of the six clades inferred from multiplex HRM PCR profiling (Fig. 3) and Salmonella isolate 412) and one known human isolate (Table 2). The three ESBL-producing Salmonella clinical isolates in cluster 1 carried a novel variant DR consensus sequence (Fabre et al., 2012). Interestingly, ESBL-producing Salmonella isolate 412 (cluster 3), isolated in 2017, was not grouped with the other three ESBL-producing S. 4,[5],12:i:- isolates collected earlier, suggesting the possibility of a different evolution trajectory of isolate 412.

Seasonal prevalence of Salmonella serotypes collected from stools of patients, Phayao Ram Hospital

Over a period of one calendar year, infection at Phayao Ram Hospital of Salmonella serotypes identified using multiplex HRM-PCR profiling peaked during the summer period (April to August) (Fig. 6), with, as expected, S. 4,[5],12:i:- being predominant, but all serotypes were represented. Although minor peaks were discernable, they are not of significance owing to the limited number of Salmonella isolates examined.