Co-occurrence of mcr-3 and fosA3 in IncP plasmid in ST131 Escherichia coli : A novel case

Introduction: Plasmid-mediated colistin resistance genes, especially mcr-3 combined with the fosfomycin resistance gene fosA3 , are a grave health concern. Our study was designed to determine the epidemiological characteristics of the combination of mcr-3 and fosA3 in Anhui province, China. Methodology: A total of 127 multi-drug-resistant (MDR) E. coli strains were assessed for antibiotic resistance/sensitivity to detect mcr-3 and fosA3 using polymerase chain reaction (PCR) and sequencing. The genes of interest were conjugated using EC600, and replicon and sequence types (STs) were identified by PCR-based replicon typing (PBRT) and multilocus sequence typing (MLST). Cluster similarity and genomic relatedness among the positive isolates were confirmed by Xbal PFGE. Results: The processed E. coli isolates were highly resistant to the tested antibiotics; the prevalence of mcr-3 was 0.78% in the transferable IncP-type plasmid in ST131, whereas fosA3 prevalence was 38.58% among different transferable plasmids, including IncFIIK, IncFII and IncA/C, and in various STs including ST69, ST1193, ST12, ST46, ST57, ST1196, ST38, ST95, ST131, ST7584 and ST10184. Both were successfully transferred to EC600. The Xbal PFGE cluster exposed similarities among the STs. Conclusions: Our results show that to control the spread of colistin and fosfomycin resistance genes in human pathogens, the ban on colistin must be continued in animal feeding farms not only in China but around the world; additionally, awareness platforms on the use of colistin must be implemented and strict policies in poultry and pig farms must be maintained. Furthermore, fosfomycin misuse by patients and overuse by physicians must be strictly managed to stop the spread of fosfomycin resistance.


Introduction
Antimicrobial resistance (AMR) has become a major public health problem globally. The massive and inappropriate use of antimicrobial agents in agriculture and medicine is partially or entirely responsible for the increased spread of multi-drug resistance (MDR). Poisoning by MDR pathogens causes more than 70,000 human deaths in the United States (US) each year [1,2]. A group led by Professor Lord Jim O'Neil estimates that by 2050, MDR will cause 10 million deaths worldwide. While the accuracy of this alarming prediction is uncertain, it is recognised that AMR is a massive burden on multiple levels (economic, community, experimental and public health) [3], underlining the importance of a coordinated international response to prevent and control the global spread of AMR [4].
Colistin is a series of non-ribosomally synthesised cationic antimicrobial cyclic peptides (CAMPs) [5] that is widely used in agricultural and medical treatments [6,7]. It was previously thought that the main targets of colistin were negatively charged lipids, specifically, a fraction of lipopolysaccharides (LPS) in the outer leaflets of the outer membranes of bacteria [8]. Despite the risk of nephrotoxicity and neurotoxicity, colistin is still used for the treatment of severe infections caused by MDR pathogens (in particular, carbapenemaseproducing Enterobacteriaceae) [9][10][11][12].
Intrinsic polymyxin resistance is limited to the previously resistant population. The therapeutic utility of colistin as the antibiotic of last resort against carbapenem-resistant superbugs may be influenced by recent developments and the discovery of plasmidmediated colistin resistance determinants (mcr-1; mcr-10) [17,18].
Currently, ten slightly different variants of the mcr gene (mcr-1-mcr-10) have been discovered in bacteria isolated from animals, food, farms, humans and the environment. As a result, the problem of mcr transmission is worsening by the day. The first mcr-1 gene-carrying plasmid was identified in E. coli, Aeromonas and Proteus of animal and human origin in America, Europe and Asia. mcr-3 shows 47% and 45% nucleotide sequence identity with mcr-2 and mcr-1, respectively, and mcr-4 has been identified in Salmonella from humans and pigs in Italy, Spain and Belgium. mcr-5 has been reported in S. enterica in Germany; mcr-6 has 87.9% identity with mcr-2 and has been reported mainly in Europe; mcr-7 and mcr-8 have been reported in China, and both have been detected in K. pneumoniae; mcr-9 has been detected in a human patient in the US; and mcr-10 has recently been found in various Enterobacteriaceae in several countries [19].
Based on the epidemiological and geographical distribution of mcr, mcr-3 appears to be the second most prevalent variant after mcr-1. Phylogenetic analysis shows that mcr-3 and mcr-1 are evolutionarily distinct. Currently, mcr-3 has been found on three continents, i.e., Asia, North America and Europe. Bacterial infections in animals, including pigs, cows and goats, are treated with colistin in Europe, while in many Asian countries, such as China and Japan, colistin is used as a growth promoter, especially in pigs and poultry. Such indiscriminate use of antibiotics has led to the emergence of new colistin resistance variants like mcr-3 [20].
Fosfomycin, a broad-spectrum antibiotic, entered clinical practice in 1996 and is extremely important in the treatment of humans. Fosfomycin inhibits bacterial cell wall synthesis and, as a reactive antibiotic against gram-negative and gram-positive bacteria, is commonly used to treat lower urinary tract infections. It functions well synergistically with cephalosporin, aminoglycosides and daptomycin.
Recently, fosfomycin was proposed for the treatment of MDR bacterial infections. Three types of enzymes, fosA, fosB and fosX, which are primarily responsible for clinical resistance, are encoded both chromosomally and in plasmids. More than ten fosA genes have been discovered, of which fosA3 is the most common variant. It is mainly horizontally distributed and predominantly found in Asia [21,22].
In our study, a total of 127 isolates of colistin-and fosfomycin-resistant strains were collected from the tertiary A hospital in Hefei, China. We performed various techniques for the genomic analysis of these antibiotic resistance genes, including antibiotic susceptibility testing, multilocus sequence typing (MLST) and polymerase chain reaction (PCR)-based replicon typing (PBRT), conjugation, and Xbal pulsedfield gel electrophoresis (Xbal-PFGE).

Study design and sample collection and identification
To identify the spread of colistin resistance variants (mcr-1-mcr-10), especially in combination with fosfomycin resistance genes (fosA1-fosA10) in E. coli, we conducted a study at a major teaching hospital in Anhui, China. Bacterial isolates were collected between April 2018 and May 2019. A total of 127 samples, including urine (n = 39), sputum (n = 38), wound (n = 27) and blood (n = 23), were received from The First Affiliated Hospital of the University of Science and Technology of China (USTC). All the isolates were grown on MacConkey agar at 37 °C overnight, and the following day, single colonies were selected and grown on Luria-Bertani (LB) broth overnight for 12-16 hours. To precisely identify the bacteria, the 16s rDNA gene was extracted from the LB broth and sequenced. The sequencing statistics were then analysed using EzBioCloud (www. ezbiocloud.net) and BLAST (www.ncbi.nlm.nih.gov/blast).

Antibiotic resistance and detection of antibiotic resistance genes
The identity of all 127 E. coli isolates was confirmed by 16s rDNA sequencing. The boiling method was followed for the extraction of DNA from the E. coli strains [23]. The detected genes included the colistin resistance genes mcr-1-mcr-10 and the fosfomycin resistance genes fosA1-fosA10. The tested antibiotics included cefotaxime, ceftriaxone, cefepime, meropenem, aztreonam, fosfomycin, amikacin, tigecycline and colistin. For the interpretation of the results and determination of breakpoint values, the breakpoint values of the European Committee on Antimicrobial Susceptibility Testing (EUCAST; www.eucast.org) and the recommendations of the Clinical Laboratory Standards Institute (CLSI-2019) were used, following a previously described protocol [24]. The PCR products were sequenced by General Biosystems Co., Ltd. (Hefei, China).

Restriction enzyme analysis with pulsed-field gel electrophoresis (REA-PFGE)
The Xbal PFGE experimental procedure was carried out according to the PFGE protocol to identify genomic similarity [26]. Bionumerics V8.0 (Applied-Maths, Sint-Martens-Latem, Belgium) was used for the Xbal PFGE gel analysis, the dendrogram was produced based on the unweighted pair-group method with arithmetic mean (UPGMA), and the Dice similarity coefficient was used for the cluster investigation with 1.5% position tolerance.

Replicon typing
The plasmids of the fosfomycin and colistin resistance determinants and their incompatibility groups were identified using the PBRT Kit 2.0 (Diatheva, Italy). This PCR-based replicon typing kit was used to identify more than 29 different incompatible plasmid groups [18].

Transferability of resistance determinants
A conjugation experiment was performed to examine the transferability of the resistance genes of interest. E. coli strains bearing two fosA3 genes and one mcr-3 gene were randomly selected as donors, while EC-600 (Rif R -Nal R ) was used as the recipient bacterial isolate. A previously described protocol was followed for the conjugation [18]. Finally, the results were confirmed by antibiotic susceptibility testing and plasmid characterisation using the PBRT Kit.

Statistical analysis
Bionumerics V8.0 (Applied-Maths, Sint-Martens-Latem, Belgium) was used for the chart construction and Xbal PFGE gel analysis, the dendrogram was produced based on UPGMA, and for the cluster investigation, the Dice similarity coefficient was calculated with 1.5% position tolerance.

Ethical approval
The study was approved (approval number 2020KY-191) by the ethical committee of The First Affiliated Hospital of the USTC.

Antibiotic resistance profile and screening for fosfomycin and colistin resistance determinants
The E. coli strains in this study were collected from different units of the hospital, including the intensive care unit, gerontology, paediatrics, oncology and urinary surgery. A total of 127 E. coli isolates were confirmed by 16S rDNA sequencing, and all the strains were highly resistant to the tested antibiotics, with 65% being resistant to amikacin, a high proportion of 88% being resistant to aztreonam, and 60% and 72% being sensitive to meropenem and tigecycline, respectively. The resistance profile is shown in Figure 1. The genes encoding resistance to fosfomycin were assessed to identify fosA genes, and 49 out of the 127 E. coli strains displayed resistance to fosfomycin with a minimum inhibitory concentration of ≥ 256 µg/mL. The prevalence of fosfomycin resistance noted was 38.58%. The strains showing resistance to colistin with a minimum inhibitory concentration of ≥ 4 µg/mL were processed further to identify mcr genes. Of the 127 E. coli isolates, only one strain was resistant to colistin (mcr-3 detected in 0.78%). No other fosA genes and mcr genes were detected in this study. The primers used for mcr-3 and fosA-3 detection are described in Table 2.

Xbal PFGE
The Xbal PFGE technique was used for the molecular typing of the E. coli isolates. Bionumerics V8.0 was used for the cluster analysis of the Xbal PFGE gel; after the isolates were successfully digested using the Xbal restriction endonuclease, the cluster was exposed, as shown in Figure 3, representing the dissimilarity among the STs of Xbal PFGE.  Figure 2. MLST result of E. coli isolates (n=127), the sequence types identified in our report are presented, and the minimum spanning tree constructed using the genomic sequences of these identified ST's and MLST alleles, while the chart was constructed using the Bionumerics software volume 8.0. ST1196 representing the highest occurrence, while ST10184 was representing the lowest occurrence. The nodes represent the STs, the diameters of the nodes represent the number of isolates, and the length of the branches represents the number of distinct alleles among the seven MLST alleles. The corresponding sequence types are labeled on the nodes.

Replicon types
Four different plasmid replicons were detected among the 127 E. coli isolates. E. coli carried the fosA3 gene on IncFIIk, IncFII and IncA/C and mcr-3 combined with fosA3 in an IncP-type plasmid. The results are detailed in Table 3.

Conjugation
The conjugation experiment was performed to detect the transferability of the plasmids carrying fosA3 and mcr-3 resistance determinants. The plasmids of interest were successfully trans-conjugated to EC600 (Rif R -Nal R ), and the resulting trans-conjugants were checked using PCR-based replicon typing. The results are described in Table 3. To further confirm the accuracy of the transferability, fosA3-and mcr-3specific plasmid PCR was performed and the determinants were observed in the trans-conjugants.

Discussion
mcr-3 and its genomic variants have been detected globally since the gene was first discovered in Shandong province in China. In this study, we identified mcr-3 in combination with the fosA3 gene in E. coli samples from The First Affiliated Hospital of the USTC. It suggested that the occurrence of the colistin resistance gene mcr-3 in Anhui province is under control. This may be a symptom of a wider decline in mcr prevalence, which may be partly due to restrictions on the use of colistin in Chinese livestock farms and improved husbandry practices. However, further research is needed. Previously published data indicate a high prevalence of animal mcr-3 (> 9.5%) in other provinces of China. However, the prevalence of mcr-3 in our study was relatively low [27]. The low prevalence of mcr-3 in our study (Table 3) may be due to sampling times or parameters of colistin resistance different from those in other studies. mcr-3 was detected both alone and in combination with the fosfomycin resistance gene fosA3 in our research. PCRbased replicon typing (Table 3) confirmed that the mcr-3 gene in our study was present in a transferable IncPtype plasmid. Isolated mcr-3 was processed to detect IS1294 inversion sequences; however, none were detected in our study. Several studies have suggested that mcr-3 is structurally different from other colistin resistance determinants; this may be why other genes, such as mcr, did not show combinations in our research. Recently, several published articles reported the spread of mcr-3 in environmental samples and hospital wastewater [28]. Similar to mcr-1, mcr-3 has been reported several times as an extended-spectrum betalactamase (ESBL) and metallo-β-lactamase (MBL), especially for blactxm-15 and bla ctxm-55 in Asian countries, while bla ndm-1 , bla ndm-5 and bla kpc-2 have been reported worldwide [29]. To our knowledge, the co-occurrence of mcr-3 and fosA3 was first noted in our clinical isolation report. A Spanish study [30] reported on the epidemiological characterisation of fosA3, showing a prevalence of 16.30% across seven different hospitals in Madrid; the STs responsible for fosA3 (ST69 and ST4038) among the 55 samples examined were also reported. As we tested more samples (n = 127), fosA3 prevalence was approximately 38.58% (Figure 1) among the different STs in our study, including ST69, which supports the accuracy of our work. fosA3 has also been previously identified [31] among different transferable plasmids with sizes between 40 and 60 kb. Since the discovery of mcr-1, other mcr-like genes have been reported globally, mainly pig-derived mcr-3 and chicken-derived mcr-7. mcr-3 and mcr-7 are reported to have very similar structures and probably originate from Aeromonas species in aquatic environments, supporting the detection of mcr-3 in environmental and hospital-derived fluids. mcr-3 and mcr-1 have been detected globally, mainly in Spain and New Zealand, and have different origins including food, humans, the environment and animals [32][33][34]. An article published in 2018 reported on the spread of mcr-3 in China; 0.75% of the tested samples were mcr-3-positive; overall, eight positive human and animal samples were identified among the 13 different provinces of China [35]. The prevalence of mcr-3 (0.78%) was slightly in our study, possibly due to study and sample load variation. The study also reported the prevalence of mcr-3 in the IncP-type transferable plasmid (Table 3), thus strengthening our study as we also report the duplicate transferable plasmid accounts for the spread of mcr-3 in Anhui province. On the other hand, the co-occurrence of fosA3 and mcr-3 have not been published before.
The genomic characterisation of mcr-3 revealed its location in many plasmids, usually with sizes over 200 kb, and was reported in a study published by the American Society of Microbiology that detected it in a 261-kb IncHI2-type plasmid [36]. In our investigation, the ST analysis identified ST131 ( Figure 2) as responsible for mcr-3 presence in E. coli; other studies have suggested different STs for mcr-3 dissemination. Furthermore, a swine sample analysis from Vietnam revealed that the spread of mcr-3 occurred via ST69 and ST1081. The differences between the two regions may have caused the observed variation in STs [37].

Conclusions
In conclusion, to curb the spread of colistin resistance genes among human pathogens, it is necessary to continue the ban on the use of colistin in livestock farms, not only in China but worldwide, maintain vigilance platforms on colistin use, and implement strict policies on poultry and pig farms. mcr-3 in combination with other resistance determinants or fosA3 poses an extreme risk; therefore, the excessive and inappropriate use of antibacterial agents should be monitored and addressed to combat such problems in humans in the future, and alternative therapies need to be explored for the management of infections, especially in animal farms and the hospital setting.