Mobile fosfomycin resistance genes in Enterobacteriaceae—An increasing threat

Abstract Antimicrobial resistance is one of the major threats to the health and welfare of both humans and animals. The shortage of new antimicrobial agents has led to the re‐evaluation of old antibiotics such as fosfomycin as a potential regimen for treating multidrug‐resistant bacteria especially extended‐spectrum‐beta‐lactamase‐ and carbapenemase‐producing Enterobacteriaceae. Fosfomycin is a broad‐spectrum bactericidal antibiotic that inhibits the initial step of the cell wall biosynthesis. Fosfomycin resistance can occur due to mutation in the drug uptake system or by the acquisition of fosfomycin‐modifying enzymes. In this review, we focus on mobile fosfomycin‐resistant genes encoding glutathione‐S‐transferase which are mainly responsible for fosfomycin resistance in Enterobacteriaceae, that is, fosA and its subtypes, fosC2, and the recently described fosL1‐L2. We summarized the proposed origins of the different resistance determinants and highlighted the different plasmid types which are attributed to the dissemination of fosfomycin‐modifying enzymes. Thereby, IncF and IncN plasmids play a predominant role. The detection of mobile fosfomycin‐resistant genes in Enterobacteriaceae has increased in recent years. Similar to the situation in (East) Asia, the most frequently detected fosfomycin‐resistant gene in Europe is fosA3. Mobile fosfomycin‐resistant genes have been detected in isolates of human, animal, food, and environmental origin which leads to a growing concern regarding the risk of spread of such bacteria, especially Escherichia coli and Salmonella, at the human–animal–environment interface.


2017
). Fosfomycin has a bactericidal effect by inhibiting the initial step in the biosynthesis of peptidoglycan. It acts as a phosphoenolpyruvate (PEP) analogue and binds the essential enzyme MurA (UDP-N-acetylglucosamine enolpyruvyl transferase) leading to bacterial cell lysis and death (Castañeda-García et al., 2013;Kahan et al., 1974).
The shortage of new antimicrobial agents has led to the re-evaluation of old antibiotics such as fosfomycin and colistin as potential treatment options against multidrug-resistant (MDR) bacteria especially extended-spectrum-beta-lactamase-and carbapenemase-producing Enterobacteriaceae, that is, the use of fosfomycin in other indications as uncomplicated UTI and potentially in combination with other antibiotics (Falagas et al., 2010;Samonis et al., 2012;Silver, 2017).
Three major fosfomycin resistance mechanisms have been described: (a) reduced permeability to fosfomycin, (b) modification of the antibiotic target MurA, and (c) antibiotic modification, whereas reduced drug uptake remains the most frequently encountered resistance mechanism among clinical isolates as well as in in vitro obtained mutants (Castañeda-García et al., 2013;Nilsson et al., 2003;Silver, 2017). In Escherichia coli, this chromosomally encoded resistance mechanism is based on mutations in two main nutrient transport systems which are responsible for the fosfomycin uptake, namely the glycerol-3-phosphate transporter (GlpT) and the glucose-6-phosphate transporter (UhpT) (Kahan et al., 1974).
In general, modification of the antibiotic target is one of the most common mechanisms to acquire antibiotic resistance in bacteria.
However, modifications of MurA seem to be rare in clinical isolates.
Nevertheless, it has been shown that overexpression of the murA gene contributes to the acquisition of clinical levels of fosfomycin resistance at low fitness cost (Couce et al., 2012). These chromosomal resistance mechanisms have been recently reviewed in detail (Castañeda-García et al., 2013;Silver, 2017).
Meanwhile, several fosfomycin-modifying enzymes have been characterized that lead to the inactivation of the drug.
Fosfomycin-producing microorganisms such as some strains of Streptomyces and Pseudomonas syringae exhibit antibiotic kinases (FomA or FomB) that modify and detoxify the antibiotic inside the cell by phosphorylation (Kobayashi et al., 2000). In clinical important pathogenic bacteria, fosfomycin resistance involves metalloenzymes that catalyze the opening of the epoxide (oxirane) ring by adding glutathione (FosA1-FosA10, FosL1-2, FosC2), bacillithiol (FosB), or water (FosX) (Rigsby et al., 2005). FosB (Mg 2+ -cofactor) type enzymes are found in low GC Gram-positive bacteria such as Bacillus subtilis, B. anthracis, Staphylococcus epidermidis, and S. aureus. FosX (Mn 2+ -cofactor) has been described in Gram-positive bacteria such as Listeria monocytogenes, Clostridium botulinum, and Brucella melitensis (Silver, 2017). In this review, we will focus on fosA and its subtypes, fosC2, and the recently described fosL1-L2 since these genes are mainly responsible for fosfomycin resistance in Enterobacteriaceae.
In Enterobacteriaceae, genes encoding fosfomycin-modifying enzymes are frequently found on plasmids, transposons, or within integrons. Regardless of the location (chromosomal or plasmid), insertion sequences are often found in proximity indicating mobilization of these genes. These facts enable rapid dissemination and indeed, fosfomycin-modifying enzymes are widespread in Asia where most of these genes initially have been described. Moreover, plasmids carrying fos-genes commonly harbor additional resistance genes that foster the chance of co-selection of fosfomycin resistance under the selective pressure by other antibiotics (Poirel et al., 2018).
The efficacy of fosfomycin against extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae is favorable although increasing fosfomycin resistance rates due to fosfomycin-modifying enzymes have been observed among ESBL-producing and carbapenemase-producing Enterobacteriaceae in the last decade, primarily in Asia but more and more globally (Baron et al., 2020;Benzerara et al., 2017;Chan et al., 2014;Fernandes et al., 2018;Hou et al., 2012;Wachino et al., 2010).
Co-selection and frequent use of fosfomycin are important factors in resistance development and dissemination. Fosfomycin is approved for use against uncomplicated UTI almost all over the world.
In some European countries, that is, Spain, France, Germany, Austria, and Greece, the intravenous formulation is administered in combination with other antibiotics for the treatment of nosocomial infections due to multidrug-resistant (MDR) Gram-positive and Gram-negative bacteria (Michalopoulos et al., 2011). In contrast, fosfomycin is not approved for veterinary use in most countries including China and Europe (Wang et al., 2017). However, fosfomycin is widely used in farms in Argentina, Brazil, and Central America, primarily for the treatment of infectious diseases in broiler chickens and pigs (Pérez et al., 2014).

| ORI G IN OF FOSA G ENE S
Ito and colleagues systematically surveyed the presence and distribution of the fosA genes in over 18,000 published genome sequences from 18 Gram-negative species (Ito et al., 2017). They propose the chromosomal location of fosA in the genomes of Providencia stuartii, Klebsiella pneumoniae, Serratia marcescens, Pseudomonas aeruginosa, Enterobacter aerogenes, Klebsiella oxytoca, Morganella morganii, Providencia rettgeri, and Enterobacter cloacae based on the high prevalence of fosA (>80%) in the respectively analyzed genomes. In contrast, fosA was rarely (<5%) detected in E. coli, Citrobacter freundii, and Acinetobacter baumannii which indicated a possible acquisition of the gene. Furthermore, they could show an extensive FosA sequence diversity both within and between species. Noteworthy, key residues in the active site of the enzyme, as well as residues involved in fosfomycin binding, are highly conserved across the different FosA proteins (Castañeda-García et al., 2013;Ito et al., 2017;Poirel et al., 2019).
Since the nomenclature of fosA has lacked consistency, it is proposed to distinguish between intrinsic versus acquired and chromosomal versus plasmid-mediated origin, respectively (Ito et al., 2017).
Thus, chromosomal or intrinsic fosA/FosA should be distinguished by adding the initials of the species whereas the numbering scheme should be reserved for acquired fosA/FosA (Ito et al., 2017).
The origins of plasmid-mediated FosA alleles are summarized in Table 1. The first plasmid-mediated fosA gene, fosA1 (also known as fosA Tn2921 ), was described in a clinical isolate in Spain. FosA1 was located within Tn2921 on plasmid pSU912 in S. marcescens (Navas et al., 1990). FosA1 is up to 100% identical with Fos EC , the fosfomycin-modifying enzyme found in the chromosome of E. cloacae which is the likely origin of FosA1 (Ito et al., 2017). Importantly, plasmid-mediated FosA1 is distinct from FosA SM which is generally present in the S. marcescens chromosome.
The origin of FosA3, the most widely disseminated acquired fosfomycin-modifying enzyme, was speculated to be K. pneumoniae after early reports (Lee et al., 2012;Wachino et al., 2010). The result of a more recent study showed that FosA3 is most closely related to FosA KG (>94% identity) from Kluyvera georgiana among known chromosomally encoded FosA (Ito et al., 2018). Furthermore, surrounding sequences were highly similar in K. georgiana YDC799 to corresponding regions of several plasmid-mediated fosA3 of E. coli (Ito et al., 2017). IS26 seems to be the key element in the mobilization of fosA3 since it commonly composes a composite transposon surrounding fosA3 by two IS26 elements oriented in the opposite direction. The same IS26-composite transposons frequently contain additionally a bla CTX-M gene (Yang et al., 2019). However, various truncations in the fosA3 surrounding regions by IS26 and other IS elements may give evidence for multiple independent mobilization events (Ito et al., 2018;Yang et al., 2019). The origin of FosA4, which shares 94% amino acid identity with FosA3, was similarly traced back to K. georgiana (Rodriguez et al., 2018).
The highly similar fosA5, fosA6, and fosA10 genes (less than 10 nucleic acid sequence differences) were proposed to originate from the K. pneumoniae chromosome (Guo et al., 2016;Huang et al., 2020;Ma et al., 2015). In all three genes, lysR genes truncated by IS10 were present upstream of the fosA gene. The downstream fragments of fosA5, fosA6, and fosA10 varied thus separate mobilization events of fosA KP were likely (Guo et al., 2016;Huang et al., 2020;Ma et al., 2015).
The plasmid-encoded fosA8, recently described in an E. coli isolate recovered from urine, showed the highest identity with a fosA gene identified in the chromosome of the enterobacterial species Lecleria adecarboxylata. This species is a rare source of infections in humans and is intrinsic resistant to fosfomycin probably due to high expression levels of the fosA LA gene .
FosA9 was first described in an E. coli isolate from a blood culture of a patient (ten Doesschate et al., 2019 (Ito et al., 2017).
The chromosomal progenitor of the recently reported plasmid-mediated fosfomycin resistance genes fosL1 and fosL2 and the origin of fosC2 is still unclear. FosC2 is a rarely described fosfomycin-modifying enzyme identified on a conjugative plasmid in E. coli in Japan (Wachino et al., 2010). This gene was not yet detected in Europe.

| PL A S MID T YPE S INVOLVED IN THE D ISS EMINATI ON OF FOS FOMYCIN -MODIF YING ENZ YME S AMONG ENTEROBAC TERIACE AE
The most common plasmid families in Enterobacteriaceae involved in the spread of antimicrobial resistance genes are IncA/C, IncL (previously designated IncL/M), IncF, IncI, IncH, and IncN (Carattoli, 2013;Rozwandowicz et al., 2018). In Europe, fosfomycin-modifying enzymes were present on plasmids belonging to the IncF, IncN, IncA/C, IncHI2, and IncX1 family, whereas IncF is the predominant plasmid incompatibility type. Some fosA-encoding plasmids were not typeable. An overview of fosA gene-carrying plasmids from enterobacterial isolates in Europe is shown in Table 2. FosA3 is the most frequently found fosfomycin-modifying enzyme worldwide (Yang et al., 2019). The dissemination of the fosA3 gene is closely associated with that of the ESBL gene bla CTX-M . In Europe, fosA3 was usually low copy number plasmids >100 kb in size (Carattoli, 2013).
IncFII plasmids (F2:A-:B-, F33:A-:B-and others) also play a key role in the dissemination of fosA3 among clinical isolates, companion, and food animals in Asian countries Hou et al., 2012;Yang et al., 2014).
Plasmids identified in Europe show highly similar molecular features even though a direct link could not be made in most cases. In France, a fosA5-carrying IncN plasmid was identified in a patient simultaneously with the original description of fosA5 on

FosA-encoding
an IncA/C background in China (Benzerara et al., 2017a;Ma et al., 2015). The fosA5-variant was previously designated as fosKP96 and has been described on 50 kb IncN plasmids in China in a urine and blood isolate, respectively .
Recently, a novel distinct fosA-variant, fosA8, was identified from a CTX-M-15-producing E. coli isolate recovered from urine in Switzerland. Mating-out experiments revealed the presence of fosA8 on a self-conjugative IncN plasmid that also carried a kanamycin resistance marker. In contrast to fosA3-plasmids, this plasmid did not carry the bla CTX-M-15 ESBL gene identified in this isolate .
In one isolate from sprouts, fosA3 was located on a conjugative Yes  self-transferable plasmids seem to play a major role in the spread of plasmid-mediated fosfomycin resistance determinants. Furthermore, resistance genes against aminoglycoside (in particular rmtB), florfenicol (floR), or sulfonamide (sul1, sul2) are often co-located with fosA genes.

| MOB ILE FOS FOMYCIN RE S IS TAN CE G ENE S IN ENTEROBAC TERIACE AE ISOL ATED AT THE H UMAN-ANIMAL-ENVIRONMENT INTERFACE IN EUROPE
The first plasmid-encoded fosfomycin-modifying gene, fosA, was reported in clinical isolates of Serratia marcescens in 1980 in Spain (Mendoza et al., 1980). Nine fosfomycin-resistant S. marcescens were studied in more detail and two types of fosfomycin resistance genes harboring conjugative plasmids were identified (Mendoza et al., 1980). In at least one case, fosA was associated with the insertion sequence IS2921 in the composite transposon Tn2921 (Navas et al., 1990). The results of a surveillance study performed fourteen years after the introduction of fosfomycin in Spain indicate that fosA is no longer confined to hospital environments since it could be detected in sewage samples in six out of seven provinces tested. Furthermore, fosA could be detected by hybridization in clinical isolates from three hospitals and the gene was carried by species of Enterobacteriaceae, Pseudomonas spp., and Acinetobacter spp. (Terán et al., 1988).
In the late nineties, an epidemiological multicentre survey was performed in Italy for which sixty out of 219 fosfomycin-resistant bacteria selected from more than 7400 urinary pathogens were screened for fosA and fosB. In only three enterobacterial isolates (1 Enterobacter sp. and 2 E. cloacae), fosA could be detected by PCR (Arca et al., 1997). However, the location of fosA genes is uncertain since hybridization using undigested plasmid DNA failed. With the current state of knowledge, a possible chromosomal location of fosA is likely (Ito et al., 2017).
After these early publications, reports on mobile fosfomycin resistance in Europe were sparse. The renewed interest in fosfomycin as a potential treatment against MDR Gram-negative bacilli and the first report of fosA3 2010 in Japan involving ESBL-producing E. coli reactivated the awareness of mobile fosfomycin resistance (Giske, 2015;Wachino et al., 2010). Fosfomycin-modifying enzymes, and especially FosA3, are meanwhile widespread in Asia among clinical isolates, pets, and livestock Ho, Chan, Lo, Law, Li, et al., 2013;Hou et al., 2012;Lee et al., 2012).
Characteristics of enterobacterial isolates encoding plasmid-mediated fosA in Europe are summarized in Table 3. The first report of fosA3 in Europe was in a Salmonella enterica serovar Corvallis strain isolated from a migratory wild bird in Germany. The fosfomycin resistance determinant was located on a large (187 kb) conjugative plasmid classified as IncA/C encoding bla NDM-1 , bla fosA3,sul1,sul2,strA,strB,aadA5,aphA6,tetA(A), mph(A), floR, dfrA7, and merA genes, which confer resistance to most β-lactam antibiotics including carbapenem, fosfomycin, co-trimoxazole, aminoglycoside, tetracyclines, and macrolides. The authors propose a possible origin of the plasmid in the Asiatic region according to complete plasmid sequence analysis and the migration route of Milvus migrans which includes Europe and northern Asia during the summer (Villa et al., 2015). However, retrospective studies revealed that fosA3 was already present in Europe at least since 2012 in clinical isolates in Portugal, France, and Switzerland, respectively (Benzerara et al., 2017;Mendes et al., 2016;Mueller et al., 2019). While the isolate from the patient in Portugal was associated with a urinary tract infection acquired after traveling to Asia (China, Philippines), there was no history of international travel in the patient population studied in France (Benzerara et al., 2017;Mendes et al., 2016). Of note, in the study of Mueller et al.  two fosfomycin-resistant E. coli isolates were identified which were tested negative for fosA1-A7 in PCR amplification although they exhibited an increased inhibition zone diameter on the agar plate in presence of FosA-inhibitor sodium phosphonoformate. In two follow-up studies, they identified three novel glutathione-S-transferase genes fosA8, fosL1, and fosL2, respectively (Kieffer et al., 2020;Poirel et al., 2019).
In more recent studies, fosA genes were reported from clinical isolates from Croatia, France, and Spain (Bedenić et al., 2018;Birgy et al., 2018;Loras et al., 2020) but only in the study from Spain were fosfomycin-resistant isolates explicitly investigated. Two isolates out of 39 randomly selected fosfomycin-resistant E. coli from urine samples carried a plasmid-borne fosA3 gene and both were ESBL-producer causing complicated UTI (Loras et al., 2020). The focus of the study from Croatia was the epidemiology, antibiotic resistance mechanisms, and routes of spread of OXA-48 carbapenemase among clinical isolates collected in 10 hospitals. Within this study, three fosA harboring enterobacterial isolates were identified (2 K. pneumoniae, 1 E. coli) whereof the later was an OXA-48 producing MDR UTI isolate (Bedenić et al., 2018). The report from France described the case of a 3-year-old boy originating from Bahrain who was hospitalized for pyelonephritis due to an ESBL-producing E. coli infection. Whole-genome sequencing has shown that the strain possesses a total number of seventeen antimicrobial resistance genes including the ESBL gene bla CTX-M-55 , the glutathione-S-transferase gene fosA, the extended-spectrum aminoglycoside resistance methylase gene armA, and the mobile colistin resistance gene mcr-1. The combination of these resistance genes is particularly worrying because colistin and fosfomycin remain one of the last-resort therapeutic options in the case of MDR Gram-negative bacteria (Birgy et al., 2018).
Besides the description of a fosA3-positive S. enterica serovar, Corvallis isolate from a migratory bird in Germany, mobile fosfomycin-modifying genes have been described in animals in   rRNA methylases such as rmtB, a gene-encoding pan-aminoglycoside resistance Yang et al., 2014). One bovine isolate and one isolate from a monkey encoded fosA3 and fosA4, respectively. FosA3 was co-located with rmtB and bla CTX-M-55 on a 48 kb IncF33:A1:B1 plasmid whereas fosA4 was detected and co-located with bla CTX-M-55 on a 145 kb IncF18:A-:B1 plasmid. Of note, the monkey isolate (fosA4) was of Asian origin as this animal was introduced to France through a French animal centre .
Also in France, ESBL-harboring plasmids from E. coli isolated from river water were studied recently. Of the 21 randomly selected isolates, one plasmid (IncF33:A-:B-) contained the resistance determinants bla CTX-M-55 , bla TEM-1b-like, and fosA and was highly similar to a plasmid from an E. coli of chicken origin in China (Baron et al., 2020).
In the context of a study on ESBL-encoding plasmids among E. coli isolates from fresh vegetables, three plasmids were found co-harboring bla CTX-M and fosA3. The conjugative plasmids were found in E. coli isolates originating from soy sprouts from the Netherlands in 2011 (Freitag et al., 2018). The presence of multidrug resistance plasmids in E. coli isolates in vegetables underlines the importance of greens as a possible reservoir for resistance genes.
Salmonella enterica serovar Infantis has emerged as the fourth most common serovar causing human salmonellosis in Europe, and it is one of the most frequently reported serovar in broilers and broiler meat (EFSA, 2016). This serovar has been associated with a resistance and virulence megaplasmids (270-320 kb) pESI (plasmid for emerging S. enterica serovar Infantis). One subgroup of this pESI-like plasmids includes a multidrug resistance plasmid encoding up to ten different resistance genes including bla CTX-M-65 and fosA3. Highly similar plasmids have been described in a human isolate from Italy, along with the poultry production chain in the United States and one human and two food isolates in Switzerland, respectively (Franco et al., 2015;Hindermann et al., 2017;Tate et al., 2017).
Tracking the transfer dynamics of resistance plasmids is an almost impossible task. However, the data point out that fosA3 was present in isolates of human, animal, food, and environmental origin.
At the human-environment interface, fosA3 was found to be located on the same plasmid backbone, that is, IncF. Surface waters are supposed to be putative reservoirs of multiresistant bacteria since they collect material from wastewater plants, water of urban or industrial effluents, and agricultural activities. Furthermore, wastewater has been described to reflect the current status of antibiotic-resistant bacteria in the population (Kwak et al., 2015). IncN plasmids carrying fosA3 seem to circulate among the human-food-animal-environment since such plasmids were described from all these origins.
Other plasmid incompatibility types than IncF and IncN appear sporadically which makes tracking even more difficult.
This review has several limitations. First, fosfomycin resistance is not routinely tested in most monitoring and surveillance studies and if so, the resistance mechanism is often not further analyzed.
Second, in many studies, fosfomycin resistance is not of main interest and is only detected by WGS and data analysis. These might cause underestimation of the current situation of the spread of mobile fosfomycin-modifying enzymes.

| CON CLUS IONS
The detection of mobile fosfomycin resistance genes in Enterobacteriaceae has increased in recent years. Certain fosA types (e.g., fosA3) are predominant and their occurrence is frequently linked to specific plasmids (e.g., IncFII family, IncN). Data from recent studies suggest that the therapeutic use of other antibiotic classes promotes the emergence and dissemination of fosfomycin-resistant Enterobacteriaceae, since additional resistance genes are often present, first, and foremost extended-spectrum beta-lactamases. Several studies highlighted in this review provide evidence for the transmission of mobile fosfomycin resistance genes in Enterobacteriaceae at the human-animal-environment interface.
The implementation of international and national mitigation strategies based on a One Health approach, including increased surveillance and monitoring of fosfomycin resistance in Enterobacteriaceae as well as the optimization of therapeutic strategies are necessary.

ACK N OWLED G M ENTS
We are grateful to the University of Zurich for financial support.

CO N FLI C T O F I NTE R E S T
None declared.

E TH I C S S TATEM ENT
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
Not applicable. Characterisation of plasmids harbouring extended-spectrum