Detection of Arcobacter Species in Human Stool Samples by Culture and Real - time PCR

Objective : The clinical significance of Arcobacter species has not been established due to a lack of suitable detection methods. Material and Methods : A total of 1,650 stool samples submitted to the Clinical Laboratory of Heidelberg University Hospital were inoculated onto agar plates selective for Campylobacter species isolation and incubated at 37℃. Results : Four (0.24%) of the samples were positive for Arcobacter butzleri isolates. Genus-specific primers for real-time PCR were designed to identify Arcobacter species. Of the 1,650 stool samples tested, twelve (0.73%), including the four culture-positive samples, were positive for Arcobacter species by real-time PCR. The sensitivity of real-time PCR was 10 4 CFU g -1 stool, 50 CFU reaction -1 using a stool sample. Conclusions : Although the sensitivity of real-time PCR was relatively low compared with other PCR methods, the present method detected a broad range of Arcobacter species. The combination of the stool culture using agar selective for Campylobacter species and real-time PCR for Arcobacter species may be clinically useful for the diagnosis and epidemiology of Arcobacter species infections.


Introduction
The genus Arcobacter belongs to the class Epsilonproteobacteria, which includes two families, family Helicobacteraceae with Helicobacter pylori, which is famous as a cause of gastric cancer, and family Campylobacteraceae. The family Campylobacteraceae includes two genera, the genus Campylobacter and the genus Arcobacter, and the genus Arcobacter is currently containing at least 28 species 1) . Arcobacter species are present in various animals, including livestock and environments in some regions and countries. Although infection caused by Arcobacter species is rare, three species, Arcobacter butzleri, Arcobacter cryaerophilus and Arcobacter skirrowii, have been associated with enteritis and occasional bacteremia in human 2) .
The morphological and biochemical properties of Arcobacter species are similar to those of Campylobacter species, as are the clinical features of patients infected with these pathogens are similar to those of Campylobacter species 3) . However, the bacteriological properties of Arcobacter species differ from those of Campylobacter species. For example, Arcobacter species can grow in the presence of oxygen (aerotolerance) and at lower temperatures (15 to 37℃) than Campylobacter species (35-42℃) 4

) 5) .
Symptoms of A. butzleri infection include abdominal pain, diarrhea, nausea, vomiting and fever 6) . Moreover, A. butzleri infection is more frequently associated with persistent and watery diarrhea, and less frequently associated with bloody diarrhea, than Campylobacter jejuni infection 6) . Arcobacter species infection has become increasingly important as an emerging gastrointestinal infection 3) . Arcobacter species infection in patients with other underlying diseases, including diabetes mellitus 7) , liver cirrhosis 8) , cancer and chronic renal failure 9) , may become severe enough to require hospitalization and medication. An outbreak of A. butzleri infection was reported to affect 10 children in a nursery and elementary school in Italy 10) . Moreover, bacteremia caused by A. butzleri in a neonate was regarded as vertically transmitted 11) .
The prevalence of Arcobacter species in stool of patients with enteritis have been reported to range from 0.1% to 1.5% using culture methods, and from 0.4 to 57.0% using PCR methods 3) 4) 5) 12) . However, the rates of detection of Arcobacter species in clinical laboratories are likely underestimated because there are no standardized protocols for the detection of these species 3) and because it is difficult to identify Arcobacter species using routine methods for bacterial identification 13) . This study describes a method to detect clinical isolates of Arcobacter species in stool samples, involving culture of stool samples and genus specific realtime PCR.

Patient stool samples for culture and real-time PCR
A total of 10,133 stool samples were submitted to the microbiology laboratory of Heidelberg University Hospital in Germany from December 2015 to March 2017. These samples were maintained at room temperature and tested within 24 hours by culturing and isolation of bacterial pathogens. Following culture inoculation, the remaining samples were stored at -20℃. Of all the stored samples, 1,650 samples were tested for real-time PCR. All samples tested for real-time PCR were obtained from patients in the internal medicine department of the University Hospital or the affiliated local primary care hospitals. Multiple samples obtained from a single patient were tested individually.

Identification of Arcobacter species from stool
culture Stool samples were cultured by routine methods to isolate species of the genera Aeromonas, Campylobacter, Salmonella, Shigella, and Yersinia. Briefly, stool samples were streaked on Columbia agar containing 5% sheep blood, on xylose-lysine-deoxycholate (XLD) agar (Becton Dickinson, Franklin Lakes, NJ, USA), on Campylobacter selective agar (Campylosel agar ® ; bioMerieux) and on cefsulodinirgasan-novobiocin (CIN) agar (Becton Dickinson) using PREVI ® Isola automated streaker (bioMerieux). Columbia and XLD agar plates were incubated under aerobic conditions at 37℃ for 24 hours; Campylobacter selective agar plates were incubated under microaerobic conditions at 37℃ for 48 hours; and CIN agar plates were incubated under aerobic conditions at 30℃ for 24 hours. Stool samples were also transferred to Selenite broth (Becton Dickinson) for enrichment; after incubation under aerobic conditions at 37℃ for 24 hours, the broth was streaked on XLD agar plates and incubated under aerobic conditions at 37℃ for 24 hours. Colonies on the culture plates were identified by MALDI Biotyper ® (Bruker Daltonik, Billerica, MA, USA) which is based on proteomic fingerprinting using matrix assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) technique for colony identification, because mass spectrometry is more reliable in identifying Arcobacter species 14) . We identified colonies with the score 2.0 or greater as Arcobacter species.

Identification of Arcobacter species by realtime PCR 1) Primer and probe design
The 23S rRNA sequences of 16 Arcobacter species were obtained from National Center for Biotechnology Information (NCBI) Taxonomy Database in January 2016 1) . The sequences were aligned with BioEdit v7.0.5 15) and the primers and a probe were designed using Primer3 16) to amplify regions of A. butzleri 23S rRNA. Specificity of the primers was checked using the primer-blast tools from NCBI. The primers were complementary to nucleotides 1593-1612 (forward primer) and 1893-1910 (reverse primer) and the probe to nucleotides 1849-1875, with the reverse primer and probe binding to the reverse strand (Table-1). The size of PCR product was 317 base pairs. The specificity of the primers for Arcobacter species 23S rRNA was shown in Figure-1, by alignment with the sequences of representative strains of Helicobacter pylori, Helicobacter coli, Campylobacter jejuni and Campylobacter coli. Alignment was performed using MEGA7 17) .

2) Preparation of pooled stool samples
Mixtures of stool samples were prepared from 10 individual samples. Briefly, a pea-sized amount or 250 μl of each sample was transferred to a sterile tube containing 5 ml nuclease-free water. Each of the 165 pooled stool samples was homogenized and filtrated with a 5 μm syringe filter. Filtrates were collected and centrifuged at 3,000 rpm for 10 minutes. The supernatants were discarded and DNA was extracted from the re-suspended pellets.
3) DNA extraction DNA was extracted from the pooled samples using QIAamp Fast DNA Stool Mini Kit ® (Qiagen, Venlo, the Netherlands), according to the manufacturerʼs instructions. If the pooled sample was positive for Arcobacter species by real-time PCR, DNA was extracted separately from each of the 10 individual stool samples that comprised the mixture.

4) Real-time PCR
Real-time PCR was performed using a Smartcycler ® (Cepheid, Sunnyvale, CA, USA). Each 25 μl reaction mixture consisted of 12.5 μl of 1 × Takyon Master Mix ® (Eurogentec, Seraing, Belgium), 6.63 μl of nuclease free water, 0.25 μl of 500 nM forward primer, 0.25 μl of 500 nM reverse primer, 0.38 μl of 150 nM probe and 5 μl of template DNA. Five μl of nuclease free water was used as a negative control and 5 μl of 1.0 × 10 -2 ng/ml A. butzleri DSM_8739 (type strain) DNA solution was used as a positive control. The amplification protocol consisted of an initial denaturation at 95℃ for 3 minutes, followed by 40 cycles of denaturation at 95℃ for 45 seconds, annealing at 61℃ for 30 seconds and extension at 72℃ for 30 seconds. PCR for DNA extracted from the mixture of 10 samples and following PCR using DNA extracted separately from each of the 10 individual stool samples used the same PCR protocol.   The sequences of 23S rRNA of the representative strains of the class Epsilonproteobacteria, of which whole-genome sequences were available in NCBI genome site, were collected and aligned with the primers to detect Arcobacter species in this study. The primers developed in this study were specific for Arcobacter species but could not detect Helicobacter species and Campylobacter species.

5) Sequencing of the PCR products
The PCR product was sent out to GATC Biotech (Konstanz, Germany) and sequenced with the same primer sets for PCR, by Sanger sequencing method. Each sequence was submitted to leBIBI-QBPP r_procaryota_LSU-rDNA-23S_stromgemt database 18) to confirm that the sequence of the PCR product was that of an Arcobacter species and to identify each species of Arcobacter. 6) Stool culture from frozen PCR-positive stool samples Stool samples positive for Arcobacter species by real-time PCR and stored at -20℃ were retrieved and cultured. Briefly, 1 gram of each stool sample was transferred to 9 ml Arcobacter enrichment broth, containing 24 g/l Arcobacter broth (Thermo Scientific, Waltham, MA, USA) and selective supplement (100 mg l -1 5-fluorouracil, 10mg l -1 amphotericin B, 16 mg l -1 cefoperazone, 32 mg l -1 novobiocine and 64 mg l -1 trimethoprim) 19) . After incubation under microaerobic conditions at 37℃ for 48 hours, the broth was streaked onto Campylobacter selective agar (Campylosel agar ® ; bioMerieux) and incubated under microaerobic conditions at 37℃ for 120 hours. Colonies were identified by MALDI Biotyper ® (Bruker Daltonik). 7) Sensitivity and specificity of real-time PCR assay for Arcobacter species (Figure-2) Sensitivity of real-time PCR assay was assessed using a stool sample negative for A. butzleri by real-time PCR (background matrix), spiked with A. butzleri DSM_8739. Sensitivity was also assessed using a PBS spiked with A. butzleri DSM_8739. Briefly, 200 mg aliquots of stool (background matrix) or 200 μl aliquiots of PBS were seeded with 10-fold serial dilutions of A. butzleri DSM_8739 solution at concentrations ranging from 1.0 × 10 6 CFU ml -1 to 10 1 CFU ml -1 (or g -1 of stool). The negative control consisted of unseeded stool (background matrix) or PBS. DNA was extracted and real-time PCR was performed as above to determine sensitivity of the real-time PCR assay.
Specificity of the primers was assessed using enteropathogenic bacteria including A. butzleri. Briefly, DNA of Aeromonas hydrophila, Arcobacter butzleri_DSM8739, Campylobacter coli, Campylobacter jejuni, Escherichia coli O-157, Salmonella typhimurium and Shigella sonnei was extracted respectively and PCR was performed using Assessment of real-time PCR assay sensitivity with or without stool. Two hundred milligram aliquots of stool (background matrix) or 200 μl aliquiots of PBS were seeded with 10-fold serial dilutions of Arcobcter butzleri DSM_8739 solution at concentrations ranging from 1.0 × 10 6 CFU ml -1 to 10 1 CFU ml -1 (or g -1 of stool). DNA was extracted and real-time PCR was performed to detect Arcobcter butzleri to know sensitivity of the assay with or without stool. Thermal Cycler Dice ® (Takara Bio Inc., Shiga, Japan) according to manufacturerʼs instructions. To prove that the DNA of enteropathogenic bacteria was contained in each reaction tube, PCR was performed using primers designed to amplify 16S rRNA (Table-2).

Identification of Arcobacter species using bacterial culture (Figure-3)
Of the 1,650 stool samples tested, thirteen were positive when inoculated on agar plates selective for Campylobacter species and incubated at 37℃ under microaerobic conditions. Of these thirteen isolates, seven were identified as Camylobacter jejuni, one as C. coli, one as Campylobacter species, and four (0.24%) as Arcobacter butzleri when analyzed by MALDI-TOF. No other bacterial pathogens causing enteritis were isolated from these four stool samples containing Arcobacter butzleri isolates, and no other Arcobacter species was detected in any of the isolates tested by MALDI-TOF.

Identification of Arcobacter species using realtime PCR (Table-3, Figure-2)
Of the 1,650 stool samples, fourteen (0.85%) showed positive reaction by real-time PCR. Sequencing of these fourteen real-time PCR products confirmed that twelve were of Arcobacter species. Of the twelve positive samples, eleven were identified as A. butzleri and one was A. skirrowii. Thus, twelve (0.73%) of the 1, 650 stool samples tested were positive for Arcobacter species by realtime PCR, with this method having an accuracy of 85.7% (twelve per fourteen real-time PCR-positive samples) in detecting Arcobacter species 3) . Other two real-time PCR products could not be identified by sequencing and searching in leBIBI-QBPP database.
All four culture-positive samples were included in the twelve real-time PCR-positive samples for Arcobacter species (Figure-3). This result indicates that real-time PCR is more sensitive than culture methods in detecting Arcobacter species in stool samples.

Stool culture from frozen PCR-positive stool samples
The twelve real-time PCR-positive samples were thawed and culured again. However, Arcobacter species did not grow up from these samples.

Sensitivity and specificity of real-time PCR method for Arcobacter species
The sensitivity of real-time PCR for Arcobacter species was 685 copies of PCR products per reaction when assessed using the PCR products of A. butzleri DSM_8739, 10 2 CFU ml -1 (2 CFU/reaction) when using DSM_8739 solution in PBS, 10 4 CFUg -1 stool (50 CFU reaction -1 ) when using a stool sample spiked with DSM_8739 (Table-4). Six clinical isolates of enteropathogenic bacteria, including one isolates of C. coli and one of C. jejuni, were negative for Arcobacter species by real-time PCR (Table-5).

Discussion
Culture identification of Arcobacter species is an essential procedure to know the clinical importance of this pathogen. This study demonstrated that a selective agar for Campylobacter species was also useful in detecting Arcobacter species in stool samples by appropriate culture condition. This selective agar contains vancomycin (10 mg l -1 ), cefoperazone (32 mg l -1 ) and amphotericin B (3 mg l -1 ) to eliminate intestinal microbiota 20) . Arcobacter as well as Campylobacter species are intrinsically resistant to these antibiotics 21) . Both Arcobacter and Campylobacter species can grow at 37℃ with microaerobic condition 4) 5) . However, it is difficult to distinguish colonies of these bacteria on blood agar plates, as both are small, white-to-gray in color, have similar microscopic features (Gramnegative curved rods) and biochemical properties except for their different aerotolerance properties. Both MALDI-TOF and PCR/DNA sequencing are fast and reliable methods for distinguish these genera as well as the identification of Arcobacter species 14) .
The prevalence rates of enteritis due to Arcobacter species in human are unclear in many countries including Japan, although they have been reported to range from 0.1% to 1.5% 3) . In this study, 0.2% of samples from patients with diarrhea in Germany, as same as previous report. Standardized protocols for detecting Arcobacter species are necessary to determine its prevalence and causative agents among patients with enteritis.
PCR-based methods are useful for detecting Arcobacter species in stool samples and have been found to be more sensitive than culture methods, as shown in previous study and this study 3) . However, using PCR methods, the prevalence of Arcobacter species in diarrheal stool samples have been reported to range from 0.4-57.0% 4) 5) 12) . The different prevalence rates of Arcobacter species among studies may result from differences in pretest probabilities, culture conditions, including medium and temperature or from differences in the sensitivities of PCR-based methods.
The isolation of Arcobacter species from stool samples using culture methods requires inoculation of these samples as soon as possible onto Campylobacter selective agar plates. Freezing of fecal samples or storage prior to isolation of bacteria was shown to reduce recovery of Arcobacter species as same as this study 22) . Moreover, A. butzleri isolates could not be recovered from PCR-positive stool Yamauchi 23) . The sensitivity of real-time PCR in the present study was almost same as previous study. A realtime PCR method showed detection limits of 4.2 CFU per reaction in broth and the detection limits of real-time PCR in the present study was 2 CFU per reaction in PBS (10 2 CFU ml -1 PBS) 24) . Another multiplex PCR method showed a sensitivity for detecting A. butzleri and A. cryaerophilus of 10 3 CFU g -1 chicken skin/meat sample 25) , and the sensitivity of real-time PCR in the present study was 10 4 CFU g -1 stool (50 CFU per reaction). The different sensitivity among study may result from differences in background material of A. butzleri DSM_8739. DNA extraction from stool as performed in this study is more difficult than it from broth, PBS and skin/meat, because of solid components and inhibitory compound known to be present in stool.
Our real-time PCR method has also advantages in detecting a wide range of Arcobacter species and may therefore be better clinically than more specific methods. Furthermore, our method has sufficient specificity. It detects Arcobacter species but does not detect other enteropathogenic bacteria including C. jejuni and C. coli, though unspecific amplification was observed in two clinical stool samples. For the reason listed above, our real-time PCR method demonstrated the value of screening test of stool samples for Arcobacter species.
The major limitation of this study was that initial screening by real-time PCR was performed for the pooled samples, containing 10 stool samples in one reaction. Sensitivity and positivity rate might be higher if each sample was individually analyzed. Another limitation of this study was that although the proportion of Arcobacter species is extremely rare, the number of study sample is insufficient for evaluation.
In conclusion, we were able to detect Arcobacter species using a routine stool culture method for isolating Campylobacter species, with assistance of MALDI-TOF. Our real-time PCR method showed higher positivity rate than the culture method. Our genus-specific real-time PCR method has possibility to detect a broad range of Arcobacter species and could be used to assess the epidemiology of infectious diseases caused by Arcobacter species. Further investigation of the usefulness of PCR-guided culture using fresh stool samples is necessary.