Identi cation of enterotoxigenic Bacteroides fragilis in patients with diarrhea: a study targeting 16S rRNA, gyrB and nanH genes

Rosa Amiri Tehran University of Medical Sciences Zahra Norouzbabaei Tehran University of Medical Sciences Naeemeh Kalali Tehran University of Medical Sciences Sedighe Ghourchian Tehran University of Medical Sciences Mehdi Yaseri Tehran University of Medical Sciences Alireza Abdollahi Tehran University of Medical Sciences Masoumeh Douraghi (  mdouraghi@tums.ac.ir ) Tehran University of Medical Sciences https://orcid.org/0000-0001-5861-3182


Background
The members of Bacteroides fragilis group are opportunistic pathogens that cause severe infections such as intra-abdominal, pelvic, and brain abscesses, peritonitis and sepsis [1]. Among Bacteroides fragilis group, B. fragilis is the most important and virulent species [2]. B. fragilis isolates are classi ed into the non-enterotoxigenic B. fragilis (NTBF) and the enterotoxigenic B. fragilis (ETBF) strains colonizing the human intestinal tract [3]. NTBF strains can cause extra-intestinal infections such as sepsis, abscesses, and necrotizing skin and soft tissue infections [3]. ETBF strains are found to be associated with acute diarrhea, gastrointestinal tract infections, in ammatory bowel disease (IBD) and recently with colorectal cancer [4,5]. Of virulence determinants, B. fragilis toxin (BFT) or fragilysin is a metalloprotease that encoded by a 6-kb pathogenicity island [6]. bft gene has three subtypes which have similar function, but the potency of these subtypes seems to be different as follows: BFT-2 > BFT-1 > BFT-3 [7]. The strains harboring bft-1 gene are often isolated from adults with diarrhea while the strains carrying bft-2 are predominantly recovered from children suffered from antibiotic-associated diarrhea [7].
The members of B. fragilis group are almost indistinguishable phenotypically [8]. The most clinically relevant species is B. fragilis and the resistance to antibiotics is not uncommon among the isolates [9]. Of great concern, B. fragilis have the ability to express and transfer antimicrobial resistance genes in human gut microbiota [10]. The selective media such as Bacteroides Bile Esculin agar (BBE) is commonly used for isolation of members of the B. fragilis group; however, the overgrowth of enterococci and lactobacilli in such media could be an issue. This issue is partly related to the increasing resistance of enterococci to gentamicin which is frequently used as a supplement in BBE agar [11]. In addition, most phenotypic methods are time-consuming, laborious, couldn not clearly distinguish closely related species and require long incubation time [12]. Also, the automated identi cation methods, such as matrix-assisted laser desorption ionization time-of-ight mass spectrometry (MALDI-TOF MS) are not affordable in laboratories with limited resources [8]. Therefore, the correct identi cation and differentiation of this species using the candidate genetic markers is of prime importance.
Among the different biochemical and molecular methods which were developed for identi cation of anaerobic bacteria, 16S rDNA sequencing is considered as a reference method for de nitive identi cation [13][14][15]. This method is robust, accurate and is useful for identi cation of new and non-culturable microorganisms. Furthermore this method can be used for taxonomic purpose [13] and distinguishing the related species from one another such as B. fragilis group species which are phenotypically similar [15]. In spite of bene ts of 16S rDNA sequencing, it was not used routinely in laboratory for identi cation of bacteria because of the high cost of sequencing [13,15], particularly in the resource limited laboratories. Here, we aimed to identify ETBF and the bft subtypes among the isolates recovered from patients with diarrhea. Also we investigated whether PCR assays of 16S rRNA gene-targeted Bacteroides fragilis group, DNA gyrase subunit B (gyrB) and neuraminidase (nanH) genes are useful determinants for identi cation of B. fragilis compared with 16S rRNA gene sequencing as the reference method.

Setting
This study was conducted on 530 patients with diarrhea who were admitted to the pediatrics hospital (n=91) or one of two referral hospitals (n=439) in Tehran, Iran, between June 2016 and April 2017. The demographic characteristics (age and gender) of patients were recorded from the medical les. The feces specimens were cultured on BBE agar and incubated at 37 ˚C for 24-48 hours in an anaerobic atmosphere using the Anoxomat jar system (MART Microbiology B.V., the Netherlands). Gray colonies with dark background and positive for esculin, were supposed to be a member of B. fragilis group. These Gramnegative coccobacilli were cultured on Brucella Blood Agar supplemented with 5 mg/l hemin and 10 mg/l vitamin K1 (Sigma, USA). The isolates were stored at -20 ˚C until further analysis.

PCR assays of 16S rRNA gene-targeted Bacteroides fragilis group (BFG), gyrB and nanH
To identify the isolates of BFG, the group-speci c primers (Bfra-F/g-Bfra-R) targeting 16S rRNA gene was used [17]. For identi cation of the most important species of BFG, B. fragilis, genespeci c primers were used to amplify gyrB (gB904F/gB1272R) [18] and nanH (GAI 11/GAI 12) genes [19]. B. fragilis ATCC 23745 was used as a reference strain. A representative PCR product for each gene was sequenced and the results were compared with those deposited in GenBank using BLAST (http://www.ncbi.nim.nih.gov/blast/cgi).

Antimicrobial susceptibility tests (AST)
The minimum inhibitory concentrations (MICs) of metronidazole, cefoxitin, moxi oxacin, piperacillinetazobactam and tetracycline against ETBF isolates using MIC Test Strips (Lio lchem, Italy) were determined and interpreted as recommended by Clinical and Laboratory Standards Institute [23]. The fresh colonies of ETBF with turbidity of 1.0 McFarland standard was inoculated into Brucella blood agar supplemented with 5 mg/l hemin and 10 mg/l vitamin K1. Plates were incubated at 37 °C anaerobically for 24 h and then the MICs were read and recorded.
Nucleotide sequence accession number The nucleotide sequences of 16S rRNA gene fragment for the all isolates tested were deposited in GenBank as indicated in Table S1. The nucleotide sequences of the representative amplicon of gyrB and 16S rRNA for BFG were deposited in GenBank under the accession number MG252858 and MG388289, respectively.
Of these 100 isolates, 77 produced amplicons using PCR primers targeting gyrB or nanH genes. As shown in the Venn diagram ( Figure 2); a subset of 32 isolates generated amplicons using the primer sets for gyrB as well as nanH. A subset of eight isolates was exclusively positive for gyrB and was missed by the primer targeting nanH. A subset of 37 isolates was only nanH -positive. Overall, 40 isolates yielded an amplicon using primers targeting gyrB gene while 69 were positive for nanH. Compared to the reference method, the speci city and accuracy of the PCR targeting gyrB gene was higher than of nanH (Table 1).  26.6% of the B. fragilis isolates were ETBF (one from an adult and three from children) and the remaining isolates were classi ed as NTBF. All ETBF isolates were positive for gyrB as well as nanH. The ETBF isolates harbored bft-1 and bft-2 subtypes: three were bft-1 positive (one from an adult and two from children) and only one was bft-2 positive from a child). All of the isolates which were toxin positive with PCR showed cytopathic effect in cell culture assay. The properties of the 100 isolates in relation to the presence of nanH and gyrB genes are shown in Table S1. The antibiotic susceptibility pro le of ETBF and their antibiotic resistance genes were shown in Table 2.

Discussion
There is a dearth of data on distribution of ETBFs in patients with diarrhea in Iran-the second largest country in the Middle East. Accordingly, we conducted the current study to assess the distribution of ETBF in children and adults who suffered from diarrhea. In the present study, we found that more than onefourth of B. fragilis isolates harbored bft gene and altered HT-29/C1 morphology. The sequence of bft gene was identical to those deposited at the GenBank. Of four ETBF isolates, three were recovered from children and one from an adult. In the current study, three ETBF isolates harbored bft-1 and only one isolate was positive for bft-2. All but one of ETBF isolates tested were susceptible to cefoxitin, metronidazole, moxi oxacin and tetracycline. Previous studies pointed to the association between ETBF and diarrhea, particularly in children. In addition, bft-1 is identi ed as the most prevalent subtypes of bft in patients with diarrhea [27][28][29]. Due to the small number of ETBF tested in this study, we could not reach a clear conclusion about the distribution of bft subtypes in children and adults as demonstrated previously by others [30,31]. Another nding of this study is that the majority of B. fragilis isolates obtained from patients with diarrhea did not harbor the bft gene. A possible explanation for this nding is that the virulence factors other than BFT might be implicated in pathogenesis of B. fragilis or these isolates might be found as a part of microbiota in these patients. In other words, the diarrhea causes might not be associated with the B. fragilis and its etiology remains unknown. Narimani et al. (2016) demonstrated that none of the B. fragilis isolates harbored bft gene as their isolates were obtained from healthy individuals in Tehran, Iran [32]. In a study recently conducted in the north-west of Iran, four ETBF isolates were recovered from diarrheal fecal samples of the 100 outpatient and hospitalized children under the age of 5 years [33]. A study by Zamani et al. (2017) conducted in Tehran and aimed to provide evidence of association between B. fragilis and ulcerative colitis (UC) and non-in ammatory bowel disease (nIBD). They detected bft genes in 51.4% and 1.6% of UC and nIBD samples, respectively: all bft positive isolates harbored the bft-1 [34]. Another study was conducted by Rashidan et al. (2018) in Tehran and they isolated B. fragilis group strains from biopsy specimens of IBD (38%) and nIBD (25%) cases.
ETBF was isolated from 6/19 IBD (31.5%) and 2/5 nIBD (40%) cases [35]. Several studies showed that the distribution of ETBF was uneven in various regions around the world. The usefulness of conserved genes for molecular identi cation has been highlighted in previous studies [18,19,36], proposing the genetic identi cation as the gold standard for identi cation of anaerobic bacteria [37,38]. 16S rRNA gene sequencing is the reliable method to be used for de nitive species-level identi cation of anaerobic bacteria. However, this method is not at reasonable turnaround time and prices to be used routinely by microbiology laboratories for identi cation of various species. Hence, we sought the performance of the PCR assays which were previously applied for identi cation of BFG or the most virulent species, B. fragilis. Of them, the PCR assay of 16S rRNA gene-targeted BFG was 32.7% in agreement with 16S rRNA gene sequencing. In other words, this PCR-based method was able to identify only 26 of isolates which belong to BFG according to 16S rRNA gene sequencing and 74 isolates which are not a member of BFG yielded an ampilcon by this PCR. We found that this PCR is not good enough to detect members of BFG as the sequencing data obtained in the current study was in slight agreement to PCR assay of 16S rRNA gene-targeted BFG.
Of various candidate genes for identifying B. fragilis, we selected nanH and gyrB as they were found to be potentially reliable gene for the identi cation and discrimination of B. fragilis from other species belonging to B. fragilis group. nanH gene encodes the neuraminidase that is an enzyme produced by all B. fragilis strains and enhances their pathogenicity [19]. We found that the all the isolates which were identi ed as B. fragilis, yielded an amplicon using primers targeting nanH genes. However, there were 54 isolates of non-B. fragilis either other species of Bacteroides or other genera produced amplicons by primers targeting nanH. This PCR was also in slight agreement to 16S rRNA gene sequencing and had low speci city to identify B. fragilis. Jotwani et al. (1995) and also Kuwahara et al. (1996) [19]. The other candidate gene is gyrB which encodes the β-subunit of DNA gyrase and it also seems to be a speci c gene for B. fragilis and able to differentiate this species from other Bacteroides [18]. We noted that the PCR assay targeting gyrB was 65% in agreement to 16S rRNA gene sequencing which was higher than that of nanH. The speci city of this PCR was also higher than that of nanH. Lee et al. (2010) used the speci c primers which could discriminate B. fragilis from Bacteroides ovatus and Prevotella melaninogenica [18]. The major discrepancy between genes in identi cation of B. fragilis might be partly related to high level of sequence conservation of nanH [39]. Of genes examined here, gyrB is more speci c than nanH for identifying B. fragilis.

Conclusion
We showed that less than 1% of patients with diarrhea harbored ETBF. Here, we noted the slight agreement between PCR-based methods for identi cation of B. fragilis and 16S rRNA gene sequencing as the reference method.

Declarations
Ethics approval and consent to participate The study protocol was in accordance with ethic guidelines of Tehran University of Medical Sciences (TUMS) and the study was approved by the review board at TUMS, Tehran, Iran.

Consent for publication
Not applicable.

Availability of data and material
Data sharing is not applicable to this article.

Con ict of Interest
The authors declare that they have no con ict of interest. Figure 1 Diagram of the culture method and the PCR assays of 16S rRNA gene-targeted BFG, gyrB or nanH for identi cation of B. fragilis in fecal specimens from patients with diarrhea.

Figure 2
Venn diagram illustrating the frequency of unique (N= 8 for gyrB and N= 37 for nanH) and shared (N= 32) positive results for gyrB and nanH among the isolates tested.