Increasing trend of antibiotic resistance in Shigella in Bangladesh: a plasmid-mediated transfer of mphA macrolide resistance gene

Shigellosis remains a common gastrointestinal disease mostly in children <5 years of age in developing countries. Azithromycin (AZM), a macrolide, is currently the first-line treatment for shigellosis in Bangladesh; ciprofloxacin (CIP) and ceftriaxone (CRO) are also used frequently. We aimed to evaluate the current epidemiology of antimicrobial resistance (AMR) and mechanism(s) of increasing macrolide resistance in Shigella in Bangladesh. A total of 2407 clinical isolates of Shigella from 2009 to 2016 were studied. Over the study period, Shigella sonnei was gradually increasing and become predominant (55%) over Shigella flexneri (36%) by 2016. We used CLSI-guided epidemiological cut-off value (ECV) for AZM in Shigella to set resistance breakpoints (zone-diameter ≤ 15 mm for S. flexneri and ≤ 11 mm for S. sonnei). Between 2009 and 2016, AZM resistance increased from 22% to approximately 60%, CIP resistance increased by 40%, and CRO resistance increased from zero to 15%. The mphA gene was the key macrolide resistance factor in Shigella; a 63MDa conjugative middle-range plasmid was harboring AZM and CRO resistance factors. Our findings show that, especially after 2014, there has been a rapid increase in resistance to the three most effective antibiotics. The rapid spread of macrolide (AZM) resistance genes among Shigella are driven by horizontal gene transfer rather than direct lineage.


Introduction
Shigella is the most common pathogen for gastrointestinal infection in developing countries and the leading cause of death among children < 5 years globally [1][2][3][4][5] . Shigella exneri is the predominant strain but Shigella sonnei is the uprising strain in low-and-middle-income countries (LMICs) including Bangladesh [5][6][7][8][9] . The sustained pressure of microbial infection and the tendency to quickly reduce the disease duration and severity has led to indiscriminate use of antimicrobials, therefore, triggering the raise of superbugs in developing countries 10,11 . World Health Organization (WHO) recommends cipro oxacin (CIP) as the rst-line therapy along with pivmecillinam, ceftriaxone, and azithromycin as alternative options. Due to the high CIP-resistance in Shigella in Bangladesh, the e cacy of CIP is currently in doubt. Recently, Shigella isolates have been reported to acquire resistant genes and plasmid with reduce susceptibility to uoroquinolones and third-generation cephalosporins 12,13 . Ceftriaxone resistance is low in Shigella but it is given parentally, therefore not encouraged for children 14,15 . Therefore, the macrolide AZM is widely used as the most preferred therapy for shigellosis in children 16 . There was no established clinical susceptibility breakpoints of AZM for Shigella before 2016 17 . Therefore, in Bangladesh, few studies reported AZM susceptibility for Shigella using different breakpoints which was inconsistent with current CLSI guideline 7,[18][19][20][21] . Several studies have reported the emergence of AZM resistance in Shigella spp. globally [22][23][24] and described the mechanism for AZM resistance 21 . To date, different molecular mechanisms involved in the development of resistance to AZM have been described. Shigella confers resistance to macrolides through variety of mechanisms include target site modi cation by methylases, enzymatic inactivation by esterases or phosphotransferases and through e ux pumps [25][26][27][28] . Several reports suggested that plasmid-mediated macrolide 2'phosphotransferase (mphA) mostly and esterase (ermB) for some instances inactivate macrolide through modifying its molecular structure 29,30 . Furthermore, conjugative R-plasmid mediated horizontal gene transfer (HGT) was demonstrated to be involved in the rapid transfer of genes responsible for resistance [31][32][33][34] . In 2015, middle range plasmid (50 MDa) mediated transfer of third generation cephalosporin resistance between Escherichia coli and S. sonnei was reported in Bangladesh 35 . Recently it has been described that a conjugative R-plasmid carrying azithromycin-resistance genes was involved in reduced susceptibility of S. exneri serotype 3a to AZM 33 .
Furthermore, given the limited treatment options for children with shigellosis, monitoring resistance rates and studying macrolide resistance mechanisms (AZMs) is not only a necessity but a task. Due to the rapid spread of the MDR phenomenon, advanced studies are always needed to assess and track real-time AMR burden in Shigella. In this study, we reported a trend towards AMR resistance in Shigella and the key mechanism of macrolide resistance in Shigella spp.

Susceptibility breakpoints for azithromycin and Shigella
Epidemiological cut-off values (ECVs) are not intended to determine clinical susceptibility cutoffs. Therefore, we performed a non-parametric Spearman rank test between the diameter of the inhibition zone of the azithromycin disk and the available MIC data for 32 S. exneri and 59 S. sonnei isolates. A signi cant correlation was observed between MIC and disc diffusion zone size for both S. exneri (rho, -0.907; P < 0.0001) and S. sonnei (rho, -0.862; P < 0.0001). We found no exception to determine the diameter of the disc diffusion zone of S. exneri and S. sonnei at the respective MIC values; zone diameter ≤ 15 mm in NWT S. exneri (MIC ≥ 16 µg/ml) and zone diameter ≤ 11 mm in NWT S. bulls (MIC ≥ 32 µg/ml) ( Figure 2). Antibiotic resistance pattern in Shigella spp.
The mphA gene conferring decreasing susceptibility to macrolide in Shigella spp.
We determined macrolide resistance genes among 37 AZM-resistant Shigella spp. which contains 14 S. exneri, 17 S. sonnei, 4 S. boydii and 2 S. dysenteriae. Out of the 37 AZM-resistant Shigella, 95% were positive for the mphA gene in the PCR test. The remaining 2 isolates did not show a band for any of the macrolide resistance genes studied. The AZM-resistant isolates of S. sonnei i with a zone diameter ≤11mm (MIC < 32 µg/ml) and AZM-resistant isolates of the other three species with a zone diameter ≤ 12 mm (MIC < 16 µg/ml) in disc diffusion method found positive for the mphA gene (Table 2).  Horizontal transfer of AMR Antimicrobial susceptibility testing con rmed that the transconjugants resistance to azithromycin, erythromycin, ampicillin and ceftriaxone, same as the donor stains (S. exneri K12582 and S. sonnei K12747) ( Table 3). The MIC of azithromycin was ≥ 256 µg/ml for all transconjugants. Plasmid analysis of transconjugants revealed that only 63 MDa plasmid was transferred ( Figure. 4A) from both donor Shigella isolates to E. coli K-12 recipient. The mphA gene was con rmed in the transconjugants and their plasmid DNAs by PCR (Figure. 4B).

Discussion
Antimicrobial resistance has been a long-persistent major public health issue, particularly in underdeveloped and developing nations where shigellosis is endemic. In this study, we report a rapid increase in the resistance to the rst-line antibiotics used to treat shigellosis, especially a 40% increase of AZM resistance among Shigella spp. in just two years (2014-2016) years. In fact, this is the rst report in Bangladesh showing a pattern of AZM-resistance in Shigella spp. following the publication of the CSLI de ned ECVs for AZM and Shigella. The altered temporal dominance of S. sonnei over S. exneri has been demonstrated and MRP-mediated HGT is considered to be the main mechanism of AMR spread. Several studies reported emergence of increasing S. sonnei worldwide including Bangladesh 6,8 . In 2001, 6% of S. sonnei was reported in Bangladesh 6,7,36 , which was increased to 54% in 2016. This acute temporal alteration of S. sonnei by 48% in just 15 years seems dramatic in geo-environmental timeframe, gives potential massages of weal and woes in parallel. Continuous improvement in the quality of global drinking water, rapid industrialization, improved nutritional status, better sanitation and less immune-crossreaction have been resulted in reduced less-adaptive Shigella spp. and increased more-adaptive S. sonnei load 6,8 . Simultaneously, antibiotic driven immense selection pressure and e cient dissemination channels can resonate the emergence of S. sonnei and signs chronic potential problems like spread of MDR S. sonnei 37,38 .
A decade ago, several drugs were considered to treat Shigellosis e.g cipro oxacin, tetracycline, chloramphenicol, ampicillin, trimethoprim-sulfamethoxazole, nalidixic acid etc 39 . Most of them have long since lost their effectiveness due to low intestinal absorption, cross-reactivity and mainly due to high resistance to Shigella.
After being the most preferred treatment option, CIP is seldom prescribed to treat Shigellosis in countries like Bangladesh currently because of its resistance mediated ine ciency 14,40 . In recent years, the prevalence of CIP resistance has been about 70% in patients of all ages, especially in Bangladesh 41,42 . In our study, we also found more than 70% CIP-resistance in 2016. Ceftriaxone is a potential alternative in shigellosis treatment but high cost and route of administration reduces its compatibility 14,15 . Moreover, a rapid increase of CRO-resistance was found in our study.
Empirically administered AZM offers an attractive option for its low frequent dosage system and high intracellular concentration in the colon of patients with shigellosis. But the absence of clinical or epidemiological cutoff values lead to unclear conclusions until 2016 [43][44][45] 44 . Therefore, this confusing situation regarding the AZM breakpoint is not over yet.
In the current study, we found sharp increase of AZM-resistance after 2014 and mphA gene was the key mechanism of resistance. Interestingly, in one of our recently published studies, we reported for the rst time of macrolide resistance pKSR100 when the plasmid carrying IS26-mphA-mrx-mphR(A)-IS6100 in Shigella isolated in Bangladesh the years after 2014 47 .
This pKSR100 plasmid is highly pathogenic and previously reported to be associated with intercontinental dissemination of macrolide resistance 33 . This is therefore a clear indication that the pKSR100-like plasmid induces an increase in AZM resistance in Shigella. By correlating the mechanism of AZM resistance, as CRO resistance increased signi cantly after 2014, the involvement of an emerging R plasmid carrying CRO-resistant factors in Shigella can be strongly inferred. In addition, the transfer of AZM, CRO and AMP resistance phenomena through one conjugative R-plasmid indicate the chance of rapid inter-species dissemination of resistance factors. In PFGE study, the AZM-resistant Shigella were not con ned in same pulsotype-cluster and they were not multiplied from same clone of AZM-resistant bacteria. These ndings indicate that horizontal transfer contributes more than direct lineage to spread AMR more rapidly.
Therefore, consistent antimicrobial resistance surveillance, resistance pro ling, and study of transmission dynamics of AMR resistance factors in MDR-Shigella are essential. In addition, effective antibiotics should be prescribed and advised carefully rather than switching to a new drug.

Study population
A total of 2407 Shigella strains were identi ed and isolated between 2009 and 2016 in the Clinical Microbiology Laboratory from the stool specimen of diarrheal patients admitted in icddr,b hospital unit, Dhaka, Bangladesh. Serotype of all the strains were con rmed in Laboratory of Gut-Brain Signaling, icddr,b, using standard microbiological and biochemical methods 48 . Among these strains, 770 isolates were subjected to antibiotic susceptibility test (AST) and further analysis. This study was reviewed and approved by institutional review board (IRB) of icddr,b, Dhaka, Bangladesh.

Serotyping of Shigella species
Isolated Shigella strains were sub-cultured on MacConkey agar (Difco, Becton Dickinson & Company, Sparks, Md.) plates, and incubate for 16 hours for optimum growth. Serotyping was performed by the slide agglutination test 49 . Two types of commercially available kits were used in this study; (i) antisera speci c for all type-and group-factor antigens (Denka Seiken, Tokyo, Japan) (ii) monoclonal antibody reagents (Reagensia AB, Stockholm, Sweden) speci c for all S. exneri type-and group-factor antigens. After serotyping, single colony of the strains was inoculated in Trypticase soy broth containing 0.3% yeast extract (TSBY), incubated for 16 hours and stored at -70°C with 15% glycerol afterwards.

Antibiotic susceptibility test (AST)
We determined the bacterial susceptibility to antimicrobial agents by the disc diffusion method according to the guidelines of CLSI 44 using Muller-Hinton agar and commercially available antimicrobial discs (Oxoid, Basingstoke, United Kingdom) 50 . We used E. coli (ATCC 25922) strain as negative control in AST. As per CLSI guideline, S. exneri with azithromycin disc diffusion zone diameter ≤15 mm and MIC ≥ 16 µg/ml respectively can be considered as NWT. In case of S. sonnei, only MIC (WT, ≤ 16 µg/ml and NWT, ≥ 32 µg/ml) was asserted in CLSI guideline 51 . In 2018, Thomas C. Darton et al. suggested disc diameter ≤ 11 mm as a cutoff value for S. sonnei. Thus, we aimed to con rm that disc diffusion zone diameter ≤ 11 mm for S. sonnei and ≤ 15 mm for S. exneri can be used to well discriminate AZM-resistant and sensitive isolates in our population. We followed Centers for Disease Control and Prevention (CDC) guided methodology for S. boydii and S. dysenteriae to de ne the susceptibility to AZM, where MIC ≥ 32 was de ned AZM resistant (non-wild type [NWT] isolates) 52 .  We used two multidrug resistant Shigella strains as donor strain and E. coli K-12 (NAL R , Lac + , F -) as the recipient, to conjugate described previously 57 . In our study, both of the donor strains had an MIC of ≥ 256 µg/ml to azithromycin and were positive for macrolide resistance factor mphA gene. Transconjugants were selected on MacConkey agar containing azithromycin (32 μg/ml: Sigma-Aldrich Corporation, St. Louis, Mo.) and nalidixic acid (32 μg/ml: Sigma-Aldrich Corporation, St. Louis, Mo.) that produce lactose-fermenting pink colonies of E. coli in contrast to non-lactose-fermenting pale colonies of Shigella isolate. As the recipient K-12 was lactose fermenting and susceptible to azithromycin, it can grow only if it receives the AZM resistance factor(s) from donor. The transconjugants were cultured onto MacConkey agar plates, and their identities were recon rmed as E. coli using with API 20E. The selected and con rmed transconjugants were then subjected to plasmid analysis and PCR con rmation for mphA gene.

Pulsed-eld gel electrophoresis (PFGE)
To observe the clonal relationship between azithromycin resistant and sensitive Shigella strains, a total of 11 Shigella strains (7 S. sonnei, 2 S. boydii, 2 S. exneri type 6) were compared using PFGE typing. Genomic DNA of Shigella strains was embedded in intact agarose gel and digested for 4 hour at 37°C with XbaI (New England Biolabs) restriction enzyme according to the PulseNet program 58,59 . CHEF-MAPPER system apparatus (Bio-Rad Laboratories) was used to separate the restriction fragments under suitable condition (switching time from 5 s to 35 s at 6 V cm−1 for 18 hours at 14°C). TIFF image of PFGE were analyzed using BioNumerics version 4.5 (Applied Maths, Kortrijk, Belgium) ngerprinting software. The dendrogram was generated by the UPGMA algorithm with the Dice-predicted similarity value of two PFGE patterns at 1.0% pattern optimization and 1.5% band position tolerance.

Declarations Ethical declaration
All experiments were performed in accordance with relevant guidelines and regulations, and all participants gave their written informed consent prior enrollment. The study was reviewed and approved by the Institutional Review Board (IRB) and the Ethical Committee of icddr,b, Dhaka, Bangladesh.