First Report of IMI-2-Producing Enterobacter bugandensis and CTX-M-55-Producing Escherichia coli isolated from Healthy Volunteers in Tunisia

The aim of this study was to characterize the prevalence of fecal carriage of extended-spectrum beta-lactamases and carbapenemase-producing Gram-negative bacteria among healthy humans in Tunisia. Fifty-one rectal swabs of healthy volunteers were plated on MacConkey agar plates supplemented with cefotaxime or imipenem. The occurrences of resistance genes, integrons, and phylogroup typing were investigated using PCR and sequencing. The genetic relatedness of isolates was determined by pulsed-field-gel-electrophoresis (PFGE) and multilocus-sequence-typing (MLST). Whole-genome-sequencing (WGS) was performed for the carbapenem-resistant isolate. Sixteen ESBL-producing Escherichia coli isolates and one carbapenem-resistant Enterobacter bugandensis were detected out of the fifty-one fecal samples. The ESBL-producing E. coli strains contained genes encoding CTX-M-15 (n = 9), CTX-M-1 (n = 3), CTX-M-27 (n = 3), and CTX-M-55 (n = 1). Three CTX-M-1-producers were of lineages ST131, ST7366, and ST1158; two CTX-M-15-producers belonged to lineage ST925 and ST5100; one CTX-M-27-producer belonged to ST2887, and one CTX-M-15-producer belonged to ST744. Six isolates contained class 1 integrons with the following four gene cassette arrangements: dfrA5 (two isolates), dfrA12-orf-aadA2 (two isolates), dfrA17-aadA5 (one isolate), and aadA1 (one isolate). E. bugandensis belonged to ST1095, produced IMI-2 carbapenemase, and contained qnrE1 and fosA genes. A genome-sequence analysis of the E. bugandensis strain revealed new mutations in the blaACT and qnr genes. Our results reveal an alarming rate of ESBL-E. coli in healthy humans in Tunisia and the first description of IMI-2 in E. bugandensis.


Introduction
Antimicrobial resistance is observed in both pathogenic and normal commensal bacteria. The bacterial microbiota contains various antimicrobial resistance genes (ARGs), even in individuals with no history of exposure to antibiotics [1]. Antimicrobial-resistant bacteria in the gut microbiota of humans and animals may pose a serious threat as they can spread to other hosts or transfer genetic resistance elements to other members of the microbiota, including pathogens [2,3].
Enterobacteriaceae are a large family of Gram-negative bacteria that includes bacteria of the normal gut microbiota but also a number of pathogens for humans. They are a common cause of urinary-tract infections and hospital-acquired pneumonia. Beta-lactam resistance in these pathogens is most commonly due to the expression of beta-lactamases, including extended-spectrum beta-lactamases (ESBLs) and carbapenemases. ESBLs mediate resistance to all penicillins, expanded-spectrum cephalosporins (ESCs) (e.g., ceftazidime, cefotaxime, and ceftriaxone), and aztreonam, but not to cephamycins (cefoxitin and cefotetan) and carbapenems. Carbapenemases confer resistance to penicillins and carbapenems, which are the drug of choice for the treatment of severe infections caused by ESBL-producing Enterobacteriacae (ESBL-E), and to ESCs used in the treatment of some of those infections (KPC, NDM, OXA, VIM, etc.). To date, many different types of ESBLs and carbapenemases have emerged worldwide (http://www.bldb.eu) accessed on 1 November 2022 [4].
Carbapenem-and ESC-resistant Enterobacteriaceae are rising on the African continent, reaching alarming levels in some countries. In a recent review on carbapenemase-producing Enterobacteriaceae (CPE) in Africa, their prevalence in hospital settings was reported as ranging from 2.3% to 67.7% in North Africa and from 9% to 60% in sub-Saharan Africa [5]. Moreover, the Mediterranean and North African countries have recently been designated as CPE-endemic areas, where ESBL-E and carbapenem-resistant Enterobacteriaceae (CRE) have been frequently reported from clinical, non-clinical, and environmental sources [5,6]. However, there is still limited information regarding the prevalence and characteristics of CPEs and ESBL-E among healthy humans in African countries. Data on the prevalence of ESBL-E in the community remain scarce in Tunisia. While the rate of fecal carriage of ESBL-E has mainly been investigated during nosocomial outbreaks [7], only one study was conducted in 2011 in a community setting. This study reported a low fecal carriage rate (7.3%) of ESBL-E among healthy individuals [8]. In the current report, we aimed to assess the prevalence of ESBL-Es and CREs among healthy humans in Tunisia in order to determine the beta-lactam resistance genes and to characterize their clonal relatedness.

The Fecal Carriage Rate of Detected ESBL and Carbapenemase-Producing Enterobacteriaceae and Beta-Lactamases
Sixteen (31.4%) out of fifty-one fecal samples yielded ESBL-E, while CPE was identified in only one sample. All ESBL-E were E. coli isolates. The prevalence of different ESBL types among the E. coli isolates is shown in Table 1. ESBL-E. coli all possessed a CTX-M-type ESBL: bla CTXM-15 (n = 9), bla CTX-M-1 (n = 3), bla CTX-M-27 (n = 3) and bla CTX-M-55 genes (n = 1). Three of the CTX-M-positive isolates were also positive for bla TEM-1 .
A variety of resistance genes located outside integrons were observed among our strains: tet(A) or tet(B) (in 12 tetracycline-resistant strains), and sul2, sul3 (in the 16 sulphonamideresistant strains).

Molecular Typing of Isolates
Phylogenetic analysis of E. coli isolates revealed that most ESBL isolates (9/16; 56.3%) were classified within phylogenetic group A, six (37.5%) were members of group D, and one (6.2%) was a member of group B2. The PFGE analysis showed that all of the isolates were genetically unrelated (supplementary Figure S1). MLSTs were determined for seven selected ESBL-E. coli depending on the type of CTX-M produced. ST131, ST7366, and ST1158 were detected in three bla CTX-M-1 -containing isolates; ST925, and ST5100 were found in two bla CTX-M-15 -containing isolates; ST2887 in one bla CTX-M-27 -producing isolate; and ST744 in one bla CTX-M-55 -producing isolate ( Table 2).

Genotypic and Phenotypic Characterization of the Carbapenem-Resistant Isolate
Genotypic and phenotypic identification showed that the carbapenem-resistant isolate, the 9i strain, belonged to the species Enterobacter bugandensis. PCR and sequencing revealed the presence of bla IMI-2 . This strain demonstrated multidrug resistance to all tested antibiotics except colistin. A whole-genome-based phylogenetic tree created using TYGS illustrated that strain 9i is close to the E. bugandensis EB-247 (NZ_AP022508.1) species (supplementary Figure S2). The ANI values of strain 9i with all genomic sequences of E. bugandensis strains deposited in GenBank (since 2022) range from 85.65% for E. bugandensis (GCF 002890755.1) to 95.23% for E. bugandensis strain 153_ECLO (NZ_JVSD00000000.1) (supplementary Table  S5). These results, indicating an ANI value above the 95% cutoff for the same species, demonstrate that the 9i strain is related to the E. bugandensis species.
E. bugandensis 9i (NZ_JAMYWA000000000.1) consists of 5.12 Mbp, distributed over 103 contigs and organized in 67 scaffolds with a fold coverage of 80×, having an N50 contig size of 221,597 kb and an average G + C content of 56%. Mining the 9i genome sequence revealed the qnrE1 and fosA genes encoding resistance to fluoroquinolone and fosfomycin, respectively. Interestingly, a novel variant of the ACT AmpC beta-lactamase encoding gene (Locus tag_M6B24_02185) was identified. In fact, the AmpC gene encoded a protein (MCP1112056.1) that showed 99% similarity with the beta-lactamase genes coding for ACT-77 (WP_060573544.1). The AmpC sequence identified in our study includes SNPs encoding four amino-acid substitutions at positions 35 (H/N), 115 (E/D), P 203 (R/K), and 267 (P/S) of the amino-acid chain (supplementary Table S1). On the other hand, we detected a new variant of the qnrE gene with a 99.38% similarity in identity to QnrE3 (CP014280.1) compared to the Qnr proteins available in the databases; and three aminoacid alterations were observed in the current variant (G7S), (A94D) and (A99V). The amino-acid alterations (G7S) and (A99V) were detected in previously published Qnr family proteins under accession numbers WP_218307411.1 and WP_039992288.1, respectively (supplementary Table S2), while the alteration at position 94 is new.

Discussion
This study was conducted to determine the prevalence of ESBLs and CPEs among Enterobacteriaceae isolates recovered from the feces of healthy individuals in Tunisia. Our findings showed that 31.4% of volunteers were ESBL-E carriers. In addition, one healthy volunteer was colonized by an IMI-2-producing E. bugandensis; to the best of our knowledge, this is the first detection of an asymptomatic CPE carrier, not only in Tunisia but also in North Africa.
The prevalence of fecal ESBL-E found in our study is almost four times higher than it was when we reported on it nine years earlier [8]. Comparing these two studies, we found that the increased prevalence of fecal ESBL-E is associated with a change in CTX-M-type distribution; indeed, CTX-M-1 was the only enzyme responsible for the ESBL phenotype in the population studied in 2011, while today more CTX-M alleles, including CTX-M-27, 55 and 15, are present in the digestive tract of the Tunisian population analyzed.
The dissemination of ESBLs among the community confirms that they are not confined solely to the clinical environment. The variability and proportion of ESBLs detected in this study seem to be in agreement with the current epidemiological data on the spread of ESBL E. coli in Tunisia. Indeed, our results showed that the most frequent ESBL among human E. coli samples is CTX-M-15, which was found in nine samples. This enzyme was also detected in ESBL-producing isolates obtained from the hospital environment, wastewater, food, and animals [9][10][11]. On the other hand, CTX-M-1 constitutes the most common ESBL of animal origin, and it is also frequent in the natural environment and in food [2,[12][13][14].
CTX-M-27-producing E. coli isolates are less common and have been recently reported in municipal wastewater [11]. Interestingly, the bla CTX-M-55 gene found in one ESBL-E. coli has never been reported in Tunisia in human isolates, and only in a few E. coli isolates of animal origin [14].
The high fecal colonization of ESBL and Carbapenemase-producing Enterobacteriaceae was attributed to the effects of different risk factors. Although their origins and characteristics were not investigated in our study, possible reasons for the acquisition of resistance could be the extensive utilization of self-administered antibiotics in the Tunisian community and the consumption of animal products harboring resistant bacteria [2].
A study carried out by Kothari et al., wherein gut colonization of exclusively breastfed healthy neonates was investigated, showed widespread resistance to ampicillin (87%) and cephalosporins among Enterobacteriaceae. The authors hypothesized that the acquisition of resistance genes was achieved through breastfeeding and contact with pets, as well as through horizontal transfer in the gut microbiota [28]. Likewise, in a Swedish study, tetracycline resistance was detected in 12% of commensal E. coli strains from infants who had not received tetracycline [29]. Furthermore, two studies from Lebanon addressed the risk factors for ESBL carriage in nursing-home residents in Beirut and Tripoli. In Beirut, constipation and antibiotic intake were independent risk factors for ESBL carriage, whereas in Tripoli, only antibiotic administration was found [30,31].
PFGE analyses and MLST typing revealed high genetic diversity among ESBL-E. coli isolates. These isolates harbored different bla CTX-M alleles and belonged to different nonepidemic E. coli sequence types, suggesting that commensal E. coli acts as a reservoir of bla CTX-M genes. Through horizontal gene transfer, these bla CTX-M genes may spread in the community amongst different bacterial species and strains. Their dissemination is often associated with highly mobilizable gene platforms, including plasmids and transposons; bla CTX-M gene transfer is largely promoted by plasmids, which are often self-conjugative and carry additional resistance determinants [2]. Transposable elements carried by the plasmid can further mobilize the genes by copying them onto another plasmid or chromosome and so increase their potential for transmission to additional bacterial recipients through conjugation [2,8].
It is of note that the pandemic lineage of extraintestinal pathogenic E. coli ST131, known to cause urinary and bloodstream infections, was also observed. Our data are consistent with those obtained in other recent studies in the Netherlands and California, where the ESBL-E. coli ST131 strain was detected in the intestine of healthy subjects [32,33].
Interestingly, reduced susceptibility to imipenem was noted in only one strain isolated from a female volunteer. PCR and sequencing results confirmed by NGS analysis data showed that it was an IMI-2-producing E. bugandensis. This is a newly described species that was initially found during a neonatal outbreak in Tanzania [34]. Until now, only three studies have reported E. bugandensis from human and non-human origin, including the International Space Station, cattle vermicompost, and clinical samples [35][36][37]. Despite the fact that IMI carbapenemases are mainly found in Enterobacter spp., no study has reported the presence of such enzymes among E. bugandensis, with the exception of a recent study where the already published E. cloacae genome MBRL 1077, originally reported as an IMI-1producing E. cloacae strain isolated in the USA, was reclassified as E. bugandensis [36,37]. Here, we report the first recognition of IMI-2-producing E. bugandensis isolated from healthy humans.

Isolation and Identification
Fifty-one rectal swabs were collected during the period from May 2018 to September 2018 from healthy human volunteers (with an age range of 3 months to 55 years) living in different urban areas in Tunisia. We excluded subjects who had undergone any antibiotic therapy or hospitalization in the 3 months prior to fecal sampling.
The samples were seeded on two MacConkey agar plates supplemented with cefotaxime (2 µg/mL) and imipenem (1 µg/mL), respectively, and were incubated at 37 • C for 24 h. For samples showing growth on the selective medium, at least three colonies were selected for subsequent characterization. Resistant isolates were identified using the VITEK 2 Compact system (bioMérieux, Marcy-l'Étoile, France) and by species-specific polymerase chain reaction (PCR) (amplification of the uidA gene) [8]. Only one ESBL or carbapenemase-positive isolate per sample was studied further.

Genetic Typing of Isolates
The clonal relationship among ESBL-producing E. coli isolates was determined with pulsed field gel electrophoresis (PFGE) using the XbaI enzyme, as previously described [39]. Seven selected ESBL-producing E. coli isolates were also characterized by multilocus sequence typing (MLST) in order to determine the corresponding sequence type (ST) and clonal complex (CC). Nucleotide sequences of the housekeeping genes were submitted to the Escherichia coli MLST Database: (http://mlst.warwick.ac.uk/mlst/dbs/Ecoli) [40]. The isolates were assigned to the phylogenetic groups A, B1, B2, or D using a PCR strategy with specific primers for chuA, yjaA, and TspE4.C2 determinants [41].

Whole-Genome Sequencing (WGS), Assembly, and Annotation of E. bugandensis 9i
Genomic DNA was extracted using the QIAamp DNA Mini Kit according to the manufacturer's recommendations. Whole-genome sequencing (WGS) was carried out on a MiSeq machine (Illumina Inc., San Diego, CA, USA). Quality assessment of the reads was performed using the FASTQC quality control tool version 0.10.0 [46]. Quality filtering was performed with the Trimmomatic program [47]. The reads assembling was performed by first using the Velvet de novo assembler for short read set assemblies [48]. Next, the results were reassembled with SPAdes 3.11.1 using multiple values for k-mer size (here, 33, 55, and 91) in order to enhance the quality of reads assemblies [49]. DNA contamination and confidence estimation for single-cell genome data were checked using ACDC software 2021-2022 [50]. Genome annotation was performed with NCBI Pro-karyotic Genome Annotation Pipeline (PGAP) [51]. This Whole-Genome Shotgun project has been deposited at DDBJ/ENA/GenBank under the accession no JAMYWA000000000. The version described in this paper is version JAMYWA010000000.
Whole-genome-based taxonomic analysis of E. bugandensis 9i was then performed using the Type Strain Genome Server (TYGS) (at https://tygs.dsmz.De) [52]. The phylogenomic tree was constructed using FastME from the Genome BLAST Distance Phylogeny (GBDP) [53]. Then, pair-wise-genome comparisons were conducted using GBDP and intergenomic distances which were obtained with algorithm 'trimming' and distance formula d5 [52]. Finally, trees were rooted at the midpoint and branch supports were inferred from 100 pseudo-bootstrap replicates [54].
The comparative genomic analysis was based on the average nucleotide identity using both best hits (one-way ANI) and reciprocal best hits (two-way ANI) between the genome sequence and all genome assemblies of Enterobacter bugandensis species (n = 172) available in GenBank (as of December 2022) using the ANI calculator with the program JspeciesWS [55]. Typically, the ANI values between genomes of the same species are above 95% [56].

Conclusions
Several ESBL-E. coli and imipenem-resistant E. bugandensis were found among healthy populations with a high fecal carriage rate. To the best of our knowledge, this is the first study detecting the fecal carriage of IMI-2-producing E. bugandensis in healthy populations in Africa and worldwide. However, due to the small sample size in our study, further studies are required to provide more compelling evidence in the near future.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics12010116/s1. Table S1: Analysis of mutations in ACTi gene after comparison with ACT-77 (WP_060573544.1); Table S2: Analysis of mutations in qnr gene in E. bugandensis 9i genome sequence; Table S3: Primers of the target β-lactamase genes; Table S4: Primers of the target antimicrobial resistance genes; Table S5: ANI values of E. bugandensis 9i isolate with all genome sequences of E. bugandensis deposited in genbank (as of December 2022) Figure S1: Pulsed-field gel electrophoresis (PFGE) profiles of the ESBL-producing E. coli isolates recovered from the faecal samples of healthy humans; Figure S2: Phylogenomic tree based on genome sequences in the TYGS tree inferred with FastME 2.1.6.1 from Genome BLAST Distance Phylogeny approach (GBDP). The branch lengths are scaled in terms of GBDP distance formula d5. The numbers above branches are GBDP pseudobootstrap support values > 50% from 100 replications. The tree was rooted at the midpoint.
Author Contributions: R.B.S., N.K., H.-I.O. and K.B.S. contributed to the design and implementation of the research. Drafting of the manuscript was carried out by R.B.S., N.K. and C.T. Acquisition, analysis, and interpretation of data were performed by R.B.S., A.A., I.C. and C.T. The NGS of IMI-2-producing E. bugandensis and interpretation was performed by A.N., A.L. and A.W. All authors discussed the results and commented on the manuscript, which was limited to those who contributed substantially to the work reported. All authors have read and agreed to the published version of the manuscript. Informed Consent Statement: All participating adults provided a priori written informed consent before taking part in the study; the parents/guardians of participating children provided written consent prior to the sampling. The data were analyzed anonymously. Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.