Prevalence and antimicrobial susceptibility of enterotoxigenic extra-intestinal Bacteroides fragilis among 13-year collection of isolates in Kuwait

Background Some strains of Bacteroides fragilis species are associated with diarrhea as a result of enterotoxin production (bft or fragilysin). Fragilysin is activated by C11 protease (fpn) and together with C10 protease (bfp) play a significant role in its invasiveness. The objectives of this study were to investigate the proportion of clinical isolates from extra-intestinal sources that are toxin producers and characterize the genes mediating toxin production. Clinical isolates submitted to our reference laboratory over the last 13 years were screened for toxin production using PCR technique. All stool isolates were excluded. The isolates were tested for their susceptibility to 8 antimicrobial agents by E test. Carbapenem resistance gene cfiA was detected by PCR. Results A total of 421 B. fragilis isolates were viable. Out of these, bft was detected in 210 (49.9%) isolates. Of the 210 bft-positive isolates, 171 (81.4%), 33 (15.7%) and 6 (2.8%) harbored bft-1, bft-2, and bft-3 genes, respectively. Twenty (9.5%) of the bft-positive strains originated from bloodstream infections. Twenty-five, 20 and 9 strains harbored bfp-1, bfp-2 and bfp-3 gene, respectively. Two, 3, 4 bfp isotypes were detected simultaneously in some of strains. The resistance rates against amoxicillin-clavulanic acid was 32%, clindamycin 62%, cefoxitin 26%, imipenem 11%, meropenem 17%, metronidazole 4%, piperacillin 61% and tigecycline 14%. A chromosomally located cfiA gene that encode metallo-β-lactamase was identified in only 34 isolates (16.2%). Conclusions The prevalence of enterotoxin-producing B. fragilis was high among the extra-intestinal isolates. Metronidazole was the most active agent against all isolates. There was no statistically significance difference between resistance rates among bft-positive and bft-negative isolates except for clindamycin.

contribution to the anaerobic microbiota of the gut is minimal based on studies on fecal samples.
An important virulence factor of B. fragilis is an enterotoxin. Enterotoxin-producing B. fragilis (ETBF) strain was first isolated from the stool of neonatal lambs with diarrhea in 1984 [3]. ETBF was later found to cause diarrhea in humans and has been defined as one of the 6 possible causes of diarrhea in children aged 1-5 years [4]. It has also been isolated infrequently from stool specimens of symptomatic adult patients [5]. Although some reports have put the prevalence of extra-intestinal ETBF at 6.2-38% [6,7], the prevalence of ETBF strains among B. fragilis isolates from diarrheagenic and extraintestinal clinical samples in Kuwait is unknown. The enterotoxin produced by B. fragilis is also known as fragilysin. It is a zinc-dependent, non-lethal, heat labile metalloprotease of about 20 kDa [8] which acts by cleaving the E-cadherin protein of zonula adherens and tight junctions in the intestinal epithelial cells leading to rearrangement of the actin of the cytoskeleton of the epithelial cells [9,10]. So far the gene mediating fragilysin (bft) has been well characterized into 4 isotypes, namely bft-1, bft-2, bft-3 and bft-4 [11]; the latter is mainly found in the Far East [7].
In addition, B. fragilis has another virulence factor, an endotoxin/lipopolysaccharide (LPS) with a demonstrable toxicity [12]. Once B. fragilis is exposed to antibiotics, it liberates endotoxin more than other Bacteroides species which may explain why this species is associated with clinical infections and higher mortality rate [13].
Some studies have demonstrated an association between increased prevalence of ETBF strains and inflammatory bowel diseases, such as Crohn's disease and ulcerative colitis [14], while a possible pathogenic role in the etiology of colorectal cancer [15] and bacteremia [16,17] has been suggested. Two cystein peptidase types with pathogenic role have recently been described: bfp1-4 genes encoding C10 peptidase and fpn gene encoding C11 peptidase (fragipain). Potential link between C10 peptidase and the pathogenesis of inflammatory bowel disease and sepsis has been documented and fragipain has been shown to activate B. fragilis enterotoxin [17,18].
The susceptibility of B. fragilis to metronidazole and carbapenem has been excellent with few anecdotal reports of resistance emerging in the literature over the last 3 decades [19][20][21]. They have remained the drugs extensively used for the treatment of infections caused by this opportunistic pathogen. The emerging reports of B. fragilis isolates resistant to these drugs are causing increasing concern to the infectious diseases and clinical microbiology experts worldwide. The trend of resistance to these drugs by B. fragilis in our country, and elsewhere in the Gulf countries, is not clearly defined at molecular level.
This study was designed to investigate the prevalence of ETBF and the bft genes among B. fragilis isolates collected over 13 years in the Anaerobic Reference Laboratory and to determine the presence of genes mediating carbapenemase (cfiA) production among carbapenemresistant B. fragilis isolates. It was also planned to investigate the prevalence of bfp1-4 and fpn genes in bft-positive and bft-negative strains.

Bacterial isolates
A total of 421 B. fragilis isolates were collected from the following different sources: wound infections (WIs), lower respiratory tract infections (LRTIs), bloodstream infections (BSIs), biopsy specimens, urine, and bile. As shown in Table 1, out of the 421 isolates, 210 (49.9%) harbored the bft gene. The majority of the bft-positive isolates were from WIs, followed by LRTIs, BSIs and biopsy specimens (BS).
Thirty-four (8.1%) and 19 (4.5%) of 421 isolates were resistant to meropenem and imipenem, respectively. Mechanism of carbapenem resistance revealed that all the carbapenem resistant strains were positive for the cfiA gene. This equates with 16.2% of the 210 bft-positive isolates. The sources of these resistant isolates were WIs [23], biopsy samples [4], BSI [3] and LRTIs [3]. A cfiA gene was detected in one B. fragilis strain that showed susceptibility to both imipenem and meropenem with MIC of 2 and 4 μg/ml, respectively. This isolate was cultured from a patient with wound infection. In addition, one B. fragilis isolate was resistant to both imipenem and meropenem with MIC of 8 and > 32 μg/ml, respectively, but cfiA-negative.

Discussion
B. fragilis can cause serious clinical infections thought to be related to the production of enterotoxin (bft), among other virulence factors. This toxin is activated by C11 (fpn) and C10 proteases (bfp) which help in the invasiveness of the organism. It has been shown that bft-positive B. fragilis are more invasive than bft-negative isolates in different types of infections and that blood culture isolates are more likely to carry bft enterotoxin gene than other isolates [24]. The prevalence of 49.9% for the enterotoxin producing-extra intestinal B. fragilis in our study is relatively high when compared with the figures of 14.4% reported in Poland [25], 18.6%, in Japan [16], 6.2-38% in USA [6,7], and 13-25% in Hungary [11,26]. Our data showed that the majority (81.4%) of the isolates contained bft-1 isotype compared with 15.7% of bft-2 and 2.9% bft-3 isotypes. It is pertinent to note that no bft-4 isotype strain was detected in this series. This order of prevalence of the isotypes is partially concordant with previous reports by Scotto d'Abusco et al., [27], Sarvari et al., [11] and Kierzkowska et al., [25]. In their study, Sarvari and colleagues [11] from Hungary reported the prevalence of 10% for the bft-1 isotype compared with 3% bft-2 but, unlike our study, they did not detect bft-3 isotype. This difference in the distribution of bft gene may be related to the severity of illness, prior antimicrobial therapy, type of the diet and thus gut flora and the method used to detect the enterotoxin. In addition, more than half (51.2%) of the isolates from bloodstream infections in Kuwait were enterotoxin producers which was much higher than previous reports from the USA and Japan (19)(20)(21)(22)(23)(24)(25)(26)(27)(28)   et al, 1996, respectively) [16,28]. In our study, simultaneous harboring of 2, 3 and 4 bfp isotypes occurred in 71, 33 and 6 isolates, respectively, which is higher than those reported by Sarvari et al., and among the most bft-positive strains, bfp-1 was the most prevalent isotype. This was discordant with the results of the study reported by Sarvari et al., in which the most common isotype was bfp-2 [11]. In addition, they did not find 4 isotypes in the same isolates [11]. Almost half of our isolates were bft-negative although they were pathogenic in a number of clinical scenarios. The explanation for this may be due to production of other virulence factors e.g. lipopolysaccharide (LPS) endotoxin especially after exposure to antibiotics [13] or both LPS and capsule that act as adhesion allowing the organism to become established at the site of infection and providing a nidus for abscess formation [1,24]. The majority (60%) of bft-positive B. fragilis isolates were from wound infections which was higher than that reported from Germany and USA (10%; 24) and from Hungary (51%; 11), but lower than that reported in Warsaw, Poland (67.5%; 28). This probably implies that bft-positive strains are more pathogenic in wound infections than the bft-negative strains.
Antimicrobial resistance is a growing problem all over the world including resistance phenomena by B. fragilis.
In this study, resistance to clindamycin (62%) was at an unacceptable level. This shows that B. fragilis in Kuwait are much more highly resistant to this agent than discordant reports elsewhere, such as 28.5% in Europe [29], 48.9% in Taiwan [30], 29.9% in USA [31] and 36.6% in China [32]. It is conceivable that the over use and abuse of this agent in almost all government and private hospitals as well as dental clinics in Kuwait is responsible for the alarming high resistance rate. Other relatively high unacceptable resistance level of 26% was recorded against cefoxitin. This is very disturbing finding as this agent is massively used for surgical prophylaxis by most of our surgeons in the country. Its empirical use must therefore be called to question. Metronidazole was the most active nonβ-lactam drug with 4% resistance rate. Despite being an active agent in our country, this resistance rate is much higher than those reported from other countries around the world [29][30][31][32]. Although the resistance rate to amoxicillin-clavulanic acid in our study is marginally higher than those reported around the globe, it is nonetheless at a very uncomfortable high level (32%). Another very interesting but disturbing finding, in our study, is the relatively high tigecycline resistance level of 14%. This is too high when compared with findings reported in the European study  (1.8%) by Nagy et al., [29], 0% in Taiwan study [30] and 5.4% in USA study [31]. B. fragilis resistance to carbapenem is often associated with production of a class B metallo-β-lactamase encoded by the chromosomal cfiA gene, in addition to outer membrane permeability barrier mechanism [33]. In our study, 11 and 17% of B. fragilis were resistant to imipenem and meropenem, respectively which are much higher than those reported in Europe (1.2% for imipenem) [29], USA (1.1% for imipenem and 2.5% meropenem) [31], Taiwan (8.5% imipenem and 9.9% meropenem) [30]. However, the B. fragilis resistance to imipenem and meropenem in our study were lower than reported in China (22.7 and 18.2%, respectively) [32]. One of our B. fragilis isolate was resistant to both imipenem and meropenem in the absence of cfiA gene which was similar to the report in studies by Soki et al., [34]. Our speculation is that this may be due to other mechanism of resistance, perhaps out membrane permeability problem. Detection of 16.2% cfiA gene is higher than 1.8% which was reported in Poland [25] but lower than that of 36.4% reported in China [32]. A cfiA gene was detected in one B. fragilis isolate that showed susceptibility to both imipenem and meropenem. This has been reported previously [35,36] and can be explained by the absence of insertion sequence upstream the gene leading to poor expression of cfiA gene.
Limitation of the study include retrospective collection of isolates and clinical data. Response to therapy could not be determined and insertion sequence elements were not done in the cfiA-positive isolates. Although to our knowledge the most common mechanism of carbapenem resistance in B. fragilis is the production of cfiA metallo-β-lactamase via activation of the cfiA gene by IS elements (high level resistance) or by activation of its putative own promoter other possible mechanisms, such as other carbapenemase genes and AmpC gene, were not investigated.

Conclusions
The prevalence of enterotoxin-producing B. fragilis strains among the clinical isolates of extra-intestinal origin was very high in our study. There was no statistically significance difference in the antibiotic resistance rates among bft-positive and bft-negative isolates except for clindamycin. In this study, metronidazole was the most active antimicrobial agent against enterotoxigenic B. fragilis isolates.

Study design
This was a multicenter prospective investigational study of stored B. fragilis isolates from 6 hospital microbiological laboratories (Mubarak, Amiri, Al Babtain, Ibn Sina, Adan and Maternity hospitals) in Kuwait. All clinical isolates obtained from proven cases of infections stored at − 80°C were resuscitated and viable strains investigated for enterotoxin production.

Bacterial strains
The bacterial strains were isolates collected during a 13years period, from 2006 through 2018, and stored in the Anaerobe Reference Laboratory, Faculty of Medicine, Kuwait University. The viable B. fragilis isolates were from proven cases of intra-abdominal infections, lower respiratory tract infections, bloodstream infections, wound infections and abscesses, managed in the 6 hospitals. All isolates were stored, in Brain Heart Infusion (BHI, Oxoid limited, Basingstoke, Hampshire, UK) broth containing 20% glycerol, at -80°C. During our investigation, isolates were subcultured on Brucella blood agar (Becton Dickinson, Heidelberg, Germany) incubated for 48 h at 37°C, in an Anoxomat Anaerobic WS800 system™ (MART Microbiology BV, Lichtenvoorde, Netherlands), in an atmospheric condition of 85% N 2 , 10% CO 2 , 5% H 2 . The identification was confirmed by a Matrix Assisted Laser Desorption/Ionization-Time of Flight Mass Spectrometry (MALDI-TOF MS; bioMérieux, L'Etoile, Marcy, France) analysis.

Antimicrobial susceptibility testing (AST)
The susceptibility of the isolates to 8 anti-anaerobic antibiotics was investigated by determining the minimum inhibitory concentrations (MICs) of the antibiotics using the E test method (bioMérieux) according to manufacturer's instructions. The antibiotics tested were the following: amoxicillin-clavulanic acid, clindamycin, cefoxitin, imipenem, meropenem, metronidazole, piperacillin and tigecycline. Susceptibility profiles of the isolates were determined according to the interpretative criteria recommended by the CLSI, 2018 [37]. B. fragilis ATCC 25285, and B. thetaiotaomicron ATCC 29741 were included as control in each run. Results for the isolates were accepted if the quality control strains results were within the established CLSI ranges (CLSI, 2018). MIC 50 , MIC 90 and percentage of resistance were calculated.

Molecular detection of fragilysin (bft) gene
Using previously published procedure, a PCR was performed for the detection of bft gene in all the B. fragilis isolates using bftF and bftR primers [18]. The genes, primer sequences, cycling conditions are shown in Table 6. The following positive controls were used: B. fragilis R19811 (bft-1), B. fragilis 1 ATCC 43858 (bft-2), and B. fragilis GAI 96462 (bft-3). Sequencing of the amplicons of the internal fragments of bft-1, bft-2 and bft-3 were performed using a GenAmp PCR system 9700 by cycling sequencing with BigDye® Terminator (AB Applied Biosystems, Carlsbad, California, USA).

Molecular detection of cfiA carbapenemase-producing gene
Production of carbapenemase by the isolates was detected in selected number of strains with very high MIC values for meropenem/imipenem using modified Hodge test. All bft-positive strains as well as imipenem and/or meropenem resistant B. fragilis (i.e. MIC ≥4 μg/ml) were screened for the presence of cfiA gene and confirmed by PCR, using published primers [38]. The genes, primer sequences, cycling conditions are given in Table 6. PCR was carried out in a volume of 25 μl. The PCR mix was obtained from Qiagen (Hilden, Germany) and the supernatant of boiled bacterial cells was used as a source of DNA template and the concentration of each primer was 25 pmol. PCR products was separated by agarose gel electrophoresis and stained with 1% ethidium bromide (Bio-Rad, Hercules, CA, USA) and visualized by UV light.

Statistical evaluation
The EpiCalc 2000, version 1.02 (Brixton Heath, Llanidloes, Powys, Wales, UK) was used to compare two proportionspercentages with 95% confidence interval and one sided P-value. WIs: Wound infections contributes to conception, acquisition and analysis, interpretation of data, and drafted the work. All authors have approved the submitted version and the modified version that involves the author's contribution to the study; AND have agreed both to be personally accountable for the author's own contributions and ensure that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved and the resolution documented in the literature. Availability of data and materials All data generated or analyzed during this study are publicly available and included in this published article.
Ethics approval and consent to participate non-applicable.
Consent for publication non-applicable.

Competing interests
The authors declare that they have no competing interests.