Evaluation of high-resolution melt curve analysis for rapid differentiation of Campylobacter hepaticus from other species in birds

Spotty liver disease (SLD) is a bacterial disease of chicken, causing mortalities and reduction in egg production, hence, contributing to economic loss in the poultry industry. The causative agent of SLD has only recently been identified as a novel Campylobacter species, Campylobacter hepaticus. Specific primers were designed from the hsp60 gene of Campylobacter hepaticus and PCR followed by high-resolution melt curve analysis was optimised to detect and differentiate three species of Campylobacter (Campylobacter coli, Campylobacter jejuni and Campylobacter hepaticus). The three Campylobacter species produced a distinct curve profile and was differentiated using HRM curve analysis. The potential of the PCR-HRM curve analysis was shown in the genotyping of 37 Campylobacter isolates from clinical specimens from poultry farms. PCR-HRM curve analysis of DNA extracts from bile samples or cultures from bile samples, were identified as Campylobacter hepaticus and confirmed by DNA sequencing. The DNA sequence analysis of selected samples from each of the three HRM distinctive curves patterns showed that each DNA sequence was associated with a unique melt profile. The potential of the PCR-HRM curve analysis in genotyping of Campylobacter species was also evaluated using faecal specimens from 100 wild birds. The results presented in this study indicate that PCR followed by HRM curve analysis provides a rapid and robust technique for genotyping of Campylobacter species using either bacterial cultures or clinical specimens.


Introduction
Spotty liver disease (SLD) is a disease of poultry that causes mortality and reduced egg production. It mainly affects free-range laying hens. On post mortem examination there are multiple 1-2mm pale focal lesions in the liver [1]. Spotty liver disease was first reported from the Eastern states of Australia in the 1980s [2] and subsequently has been reported from Europe, The field isolates were obtained from bile and faecal specimens collected at post mortem examination from the gall bladder and large intestine of chickens, respectively. The bile and faecal specimens were collected into 5 mL sterile tubes which were stored in zip lock bags, on ice in a thermally insulated container until received at the laboratory where they were stored at -20˚C prior to DNA extraction. All isolates are listed in Table 1 with source details of Campylobacter species.

DNA extraction
Total genomic DNA was extracted from bile specimens and bacterial cultures using Wizard 1 SV Genomic DNA purification system (Promega, Australia). The Quick-DNA Faecal/soil Microbe Miniprep Kit (Zymo Research, USA) was used to extract DNA from faecal samples, according to the manufacturer's instructions. Faecal swab samples of the same species of wild birds were pooled in groups of five before DNA extraction. All extracted DNA was quantified using the Nanodrop 2000 (Thermo Fisher Scientific, Australia). The concentration of each DNA sample was adjusted to 5 ng/μl for subsequent PCR or stored at -20˚C until use.

PCR amplification
The hsp60 gene was used to design the PCR primers (hsp60-F 5'-AAGAAATATTACAGC AGGAG-3' and hsp60-R 5'-GCATACCTTCAACAACATT-3'). The hsp60 gene has been successfully used in PCR for detection of enteric bacteria and different species of Campylobacter [23][24][25]. The PCR amplification was performed in 25 μl reaction volume on a Rotor-Gene™ 6000 thermal cycler (Qiagen, Melbourne, Australia). PCR amplification was performed in 25 μl reaction volume and the reaction mixture contained 3 μl of extracted genomic DNA, 25 μM of each primer, 1.5 mM MgCl 2 , 1250 μM of each dNTP, 5 μM SYTO 9 green fluorescent nucleic acid stain (Invitrogen, Australia), 5 x GoTaq Green Flexi Reaction Buffer and 1 U of Go Taq DNA Polymerase (Promega, Australia). The optimal PCR conditions were initial denaturation at 96˚C for 3 min, 35 cycles of 96˚C for 30 s, 55˚C for 30 s and 72˚C for 30 s, and a final extension of 72˚C for 5 min. In each set of PCR, HV10 C. hepaticus genomic DNA and distilled water were included as positive and negative controls, respectively.

High-resolution melt curve analysis
On completion of PCR, HRM curve analysis was performed. The PCR products were subjected to 0.5˚C/s ramping between 50˚C and 90˚C. All samples were tested in triplicate and their melting profiles were analysed using Rotor-Gene 1.7.27 software and the HRM algorithm provided. Normalisation regions of 76-77˚C and 83.5-84.5˚C were applied for analysis. The HV10 isolate (C. hepaticus) was set as a 'genotype' (GenBank accession CP031611), and the average HRM genotype confidence percentage (GCP) (the value attributed to each isolate compared with the genotype, with a value of 100% indicating an exact match) for the replicates was predicted by the software. The GCPs for C. hepaticus known isolates were averaged and the standard deviation (SD) calculated and used to establish the GCP range for the C. hepaticus cut-off point. The cut-off point was applied in HRM analysis to evaluate the differentiation power of the assay to discriminate the Campylobacter spp.

Sequencing and nucleotide sequence analysis of PCR amplicons
Selected PCR amplicons (B1, C1, F2, PR1, PRC1, UN1, ATCC29428, ATCC33559 and HV10) were purified using the Wizard 1 SV Gel and PCR Clean-Up System (Promega, Australia) according to the manufacturer's instructions. Purified amplicons were sequenced in both directions with the same primers as used for PCR by Australian Genome Research Facility Ltd (AGRF Ltd., Brisbane, Australia). The sequences were analysed using Clustal W [26] and DNASTAR (Meg Align) software. GenBank accession numbers were assigned to the nucleotide sequences of the Campylobacter isolates and reference strains (Table 1).

Detection limit of the assay
The detection limit of the assay on DNA extracted from the bile sample was determined using dilutions of quantified C. hepaticus DNA. DNA extracted from C. hepaticus was serially diluted 10-fold from 1 ng to 10 −8 ng. Each DNA dilution was tested in PCR. To evaluate the specificity of the assay, DNA was extracted from seven bacterial strains of genetically similar genera (Klebsiella, Pseudomonas, Enterobacter, Staphylococcus, streptococci, E. coli and Pasteurella) and tested in PCR-HRM.

Results
Gel electrophoresis demonstrated that the PCR generated only amplicons of expected size of 269 bp and non-specific DNA amplification was not observed.

Differentiation of C. hepaticus, C. coli and C. jejuni strains using conventional and normalised HRM curve analysis
The three Campylobacter species, (ATCC33559 C. coli, ATCC29428 C. jejuni and HV10 C. hepaticus) each produced one peak in conventional melt curve and distinctive normalised curves (Fig 1a and 1b). The mean melting points and SD of C. coli, C. jejuni and C. hepaticus were 80.50 ±0.00˚C, 81.65±0.00˚C and 80.25±0.14˚C, respectively.

Non-subjective differentiation of Campylobacter species using GCP values and a mathematical calculation
The HRM curve analysis produced a distinct profile for C. hepaticus positive isolates. The positive control C. hepaticus (HV10) produced one peak and field samples (samples 1-40) also produced one peak at 80.15-80.40˚C. The C. jejuni samples (ATCC29428, NCTC11351 and BAL172236) produced a single peak at 81.75-82.0˚C, and the C. coli samples (ATCC33559 and C699) at 80.65-80.75˚C (Table 1). The negative control samples did not produce a peak. The conventional and normalised curves produced by C. coli samples was close to the curves range produced by the C. hepaticus isolates (Fig 2a and 2b).
Using GCPs of all C. hepaticus isolates, a cut-off value was generated as a mathematical model to assess the relationship of the isolates without requiring visual interpretation by the operator. The mean GCP of all C. hepaticus specimens was 92.4 and the mean SD was 4.1. The value of 5SD was subtracted from the average GCP to determine a cut-off point. A cut off point value of 71.9 was determined for C. hepaticus genotypes. Thus, the GCP range of the C. hepaticus isolates was determined to be 71.9-100 and was used for detection of all Campylobacter isolates. When C. hepaticus reference strain (VH10) was used as the reference genotype with a cut off value of 71.9, all C. hepaticus isolates produced a GCP (72-93) higher than 71.9 and genotyped as C. hepaticus. All C. jejuni and C. coli isolates also generated GCPs between 0.0-0.6 and 40.3-63.1, respectively, which were less than cut-off point (71.9) and therefore were automatically genotyped as 'variation'. When the C. hepaticus was used as reference genotype, the highest GCP values of C. coli and C. jejuni were 63.1 and 0.6, respectively. Therefore, the GCP gap between C. hepaticus and C. coli was about 9 and between C. hepaticus and C. jejuni was 71. To show the differentiation power of PCR-HRM in discriminating between C. hepaticus and C. coli and C. jejuni, the mean GCP values of each Campylobacter isolate was plotted in a dot plot which shows the GCP gap when C. hepaticus (HV10) is used as the reference genotype (Fig 3). The PCR-HRM had a higher discrimination power in differentiating between C. hepaticus and C. jejuni isolates.

Detection of minor variations in hsp60 nucleotide sequence by the newly developed PCR-HRM curve analysis technique
To confirm that the differentiation of Campylobacter species by HRM curve analysis was related to variations in the nucleotide sequences of the amplicons, nucleotide sequences of amplicons from each distinct curve profile were determined and compared. The PCR amplicon for C hepaticus, C. jejuni and C. coli reference strains was 269 bp in length. One of C. hepaticus isolates (F8-1) had two substitutions (T to G and G to A) at positions 259 and 260, respectively. C. coli and C. jejuni strains each had 25 and 27 nucleotide substitutions when compared with C. hepaticus HV10 (Fig 4).

Sensitivity and specificity of the assay
The concentration of DNA extracted from C. hepaticus positive control was measured using NanoDrop 2000 (ThermoFisher Scientific, Australia) and serial 10-fold dilutions were prepared. Each dilution was tested in PCR and 10 μl of each amplicon was run on 1.5% of agarose gel electrophoresis. The DNA sample was detectable up to 10 −3 ng of C. hepaticus DNA (S1 Fig). None of seven bacterial strains (Klebsiella, Pseudomonas, Enterobacter, Staphylococcus, streptococci, E. coli and Pasteurella) produced a curve and remained negative using the PCR-HRM. The specificity of the assay was 100% as confirmed with DNA sequencing of Campylobacter species used in this study.

Evaluation of PCR-HRM assay for differentiation of Campylobacter spp. in wild waterfowl
Out of 20 pooled samples (each pooled sample consisted of five faecal specimens of a particular wild duck species), two samples from Pacific black duck (PBD-13 and PBD-14) were found positive in PCR-HRM. Both samples produced similar melt curves (81.5±0.06) which were higher than melting points of C. coli (80.5±0.00), C. jejuni (81.2±0.02) and C. hepaticus (79.9 ±0.04) and therefore, generated distinct normalised curves (Fig 5). The amplicons of 272 bp were detected in the two groups of samples (both from Pacific black ducks) with identical nucleotide sequences (Fig 6). Sequence comparisons using BLAST showed 87.1% sequence similarity with Campylobacter canadensis (C. canadensis) and 85.8% similarity with C. coli. The phylogenetic analysis showed that Campylobacter species detected in wild waterfowls were genetically closer to C. canadensis when compared with C. coli, C. jejuni and C. hepaticus in their hsp60 gene (Fig 6).

Discussion
Spotty liver disease, caused by the bacteria C. hepaticus, is a disease of significance to the poultry industry, and may continue to increase in prevalence as the layer industry moves away from cages to free range and barn husbandry systems. Diagnosis of SLD can be difficult in the field as the majority of birds in a flock will appear to be in good health, with sick birds showing only mild depression or found dead [12]. To limit the spread of disease and to treat affected birds, the identification of the disease-causing pathogen needs to be rapid.
This study describes the development and assessment of a PCR with HRM curve analysis for the detection of C. hepaticus. Molecular tests are a more rapid method for detection of bacterial pathogens than traditional culture methods particularly for slow growing bacteria such as C. hepaticus. A PCR has been developed and optimised to detect C. hepaticus [16], and also a multiplex PCR has been developed which could simultaneously and specifically identify the presence of C. jejuni, C. coli and C. hepaticus [27]. While PCR and multiplex PCR are suitable for diagnosis of C. hepaticus, running samples on agarose gel electrophoresis is required in both methods to reveal the results which takes about one hour to conclude. While pre-poured agarose gels or buffer-free electrophoresis can save time, still loading samples and running the gel and obtain photographic record of the results is a time-consuming exercise. The use of HRM curve analysis is a relatively new technology which provides further automation by eliminating the need for gel electrophoresis and provides faster analytics to identify the genotype of samples in less than 15 minutes when compared to gel electrophoresis and saves time to pour, load and run the gel and therefore reduces time and labour-cost. The use of PCR-HRM curve analysis as a rapid genotyping method compared with PCR and gel electrophoresis has been well documented before [21,28]. A multiplex-PCR and HRM curve analysis for differentiation of C. jejuni and C. coli suitable for use with poultry faecal or carcass swab specimens has also been reported [14].
This study assessed the use of bile specimens collected from dead or culled birds at post mortem examination for diagnostic purposes. Campylobacter spp. are recognised as bile resistant [29] and the level of C. hepaticus detected in the bile of hens has been shown to be considerably higher than in the liver [16], therefore bile was selected as the preferred sample for use in this study, however, a few faecal samples were also included in this study.
Targeting hsp60 gene in PCR for detection and differentiation of enteric bacterial pathogens including Campylobacter species has been reported to be a highly specific and reproducible approach [25,30]. The specificity of this test was assessed by the inclusion of reference isolates of the two common Campylobacter spp, C. coli and C. jejuni, as well as C. hepaticus negative and positive field samples. This PCR with HRM curve analysis was able to correctly detect C. hepaticus, and differentiate C. hepaticus from C. coli and C. jejuni. The sensitivity of the test was determined using serial 10-fold dilutions of the C. hepaticus reference sample with the sensitivity determined to be 10 −3ng of C. hepaticus DNA. The sensitivity of PCR and culture in detection of Campylobacter species has been reported to be very similar [31].
The positive samples in PCR-HRM which were collected from wild Pacific black duck were genetically 87.1% similar to C. canadensis which has been reported in captive whooping cranes (Grus Americana) in Canada [32]. While partial hsp60 gene sequences of wild bird samples (PBD-13 and PBD- 14) show similarity with C. canadensis, full genome sequence of these samples is required for confirmation of the species. Among different Campylobacter species, C. coli, C. jejuni, C. lari and C. hepaticus are commonly associated with birds [8,[32][33][34][35]]. While C. coli, C. lari and C. jejuni have also been reported in waterfowl, C. jejuni has been found more prevalent in faeces of waterfowl [36]. There is limited information on the biology of C. canadensis and its role in human health.
Banowary, Dang [14] concluded that PCR with HRM curve analysis is a powerful and valuable tool to study the nucleotide diversity of amplicons between tested specimens. The test enables detection of multiple species in the same test and is suitable for automation which makes this method a powerful diagnostic screening tool. There is also a reduction in the risk of cross-contamination as the PCR with HRM uses closed tubes, reducing the risk of DNA contamination of controls and samples [18].
This PCR-HRM curve analysis is suitable for the testing of birds suspected to have died due to SLD. This test would also be suitable for epidemiological studies at the processing plant, where a large volume of samples could be collected and tested to assess the prevalence of SLD on farms and regions. There is still considerable research required to understand the spread and pathogenicity of C. hepaticus and the test developed in this study could be useful to study disease prevalence. The PCR-HRM curve analysis showed both specificity and sensitivity for detection of C. hepaticus. This assay represents a rapid and sensitive diagnostic tool for assisting in the diagnosis of SLD in poultry.  F8-1, PR1, PRC8, PRC1, F2, C1 and B1) and wild birds (PBD-13 and PBD-14). Sequences were compared with those of Campylobacter reference strains. The tree was constructed using the Neighbour-joining method (DNASTAR software).