Polyclonal Spread of Fosfomycin Resistance among Carbapenemase-Producing Members of the Enterobacterales in the Czech Republic

ABSTRACT Fosfomycin (FOS) has been recently reintroduced into clinical practice, but its effectiveness against multidrug-resistant (MDR) Enterobacterales is reduced due to the emergence of FOS resistance. The copresence of carbapenemases and FOS resistance could drastically limit antibiotic treatment. The aims of this study were (i) to investigate fosfomycin susceptibility profiles among carbapenem-resistant Enterobacterales (CRE) in the Czech Republic, (ii) to characterize the genetic environment of fosA genes among the collection, and (iii) to evaluate the presence of amino acid mutations in proteins involved in FOS resistance mechanisms. During the period from December 2018 to February 2022, 293 CRE isolates were collected from different hospitals in the Czech Republic. FOS MICs were assessed by the agar dilution method (ADM), FosA and FosC2 production was detected by the sodium phosphonoformate (PPF) test, and the presence of fosA-like genes was confirmed by PCR. Whole-genome sequencing was conducted with an Illumina NovaSeq 6000 system on selected strains, and the effect of point mutations in the FOS pathway was predicted using PROVEAN. Of these strains, 29% showed low susceptibility to fosfomycin (MIC, ≥16 μg/mL) by ADM. An NDM-producing Escherichia coli sequence type 648 (ST648) strain harbored a fosA10 gene on an IncK plasmid, while a VIM-producing Citrobacter freundii ST673 strain harbored a new fosA7 variant, designated fosA7.9. Analysis of mutations in the FOS pathway revealed several deleterious mutations occurring in GlpT, UhpT, UhpC, CyaA, and GlpR. Results regarding single substitutions in amino acid sequences highlighted a relationship between ST and specific mutations and an enhanced predisposition for certain STs to develop resistance. This study highlights the occurrence of several FOS resistance mechanisms in different clones spreading in the Czech Republic. IMPORTANCE Antimicrobial resistance (AMR) currently represents a concern for human health, and the reintroduction of antibiotics such as fosfomycin into clinical practice can provide further option in treatment of multidrug-resistant (MDR) bacterial infections. However, there is a global increase of fosfomycin-resistant bacteria, reducing its effectiveness. Considering this increase, it is crucial to monitor the spread of fosfomycin resistance in MDR bacteria in clinical settings and to investigate the resistance mechanism at the molecular level. Our study reports a large variety of fosfomycin resistance mechanisms among carbapenemase-producing Enterobacterales (CRE) in the Czech Republic. Our study summarizes the main achievements of our research on the use of molecular technologies, such as next-generation sequencing (NGS), to describe the heterogeneous mechanisms that reduce fosfomycin effectiveness in CRE. The results suggest that a program for widespread monitoring of fosfomycin resistance and epidemiology fosfomycin-resistant organisms can aide timely implementation of countermeasures to maintain the effectiveness of fosfomycin.

IMPORTANCE Antimicrobial resistance (AMR) currently represents a concern for human health, and the reintroduction of antibiotics such as fosfomycin into clinical practice can provide further option in treatment of multidrug-resistant (MDR) bacterial infections. However, there is a global increase of fosfomycin-resistant bacteria, reducing its effectiveness. Considering this increase, it is crucial to monitor the spread of fosfomycin resistance in MDR bacteria in clinical settings and to investigate the resistance mechanism at the molecular level. Our study reports a large variety of fosfomycin resistance mechanisms among carbapenemase-producing Enterobacterales (CRE) in the Czech Republic. Our study summarizes the main achievements of our research on the use of molecular technologies, such as next-generation sequencing (NGS), to describe the heterogeneous mechanisms that reduce fosfomycin effectiveness in CRE. The results suggest that a program for widespread monitoring of fosfomycin resistance and epidemiology fosfomycinresistant organisms can aide timely implementation of countermeasures to maintain the effectiveness of fosfomycin. among these strains has been conducted recently. Moreover, as reported in previous studies, the Czech Republic has an increasing number of cases of disease caused by carbapenemase-producing E. coli and C. freundii, and the occurrence of fosfomycin resistance mechanisms in such isolates represents a concerning public health issue (35)(36)(37).
The aim of our study was to characterize the epidemiology of FOS resistance in the Czech Republic clinical setting in E. coli and Citrobacter carbapenemase producers. Moreover, an additional aim was to characterize the resistance mechanisms of FOS resistance through the detection of fosA-like genes and through the detection of specific point mutations in the associated transporters and regulators involved in uptake.

RESULTS
All 293 Enterobacterales strains were carbapenemase producers: 132/293 produced NDM-type enzymes (  Step 3) GlpR acts as a repressor for GlpT expression. In the cytoplasm, G3P attaches to GlpR, blocking its binding to glpT promoter. (Step 4) PtsI transfers a P group from PEP (2-phosphoenolpyruvate) to PtsH. The P group is then transferred to CRR by PtsH. CRR-P and PtsI activate CyaA (4). CyaA is an adenylate cyclase that converts the ATP to cAMP. cAMP binds to the CRP, and the cAMP-CRP complex promotes expression of both GlpT and UhpT (5). (Step 5) cAMP-CRP complex promotes GlpT expression, binding to glpT promoter. The activity of the cAMP-cAMP receptor protein (CRP) complex is enhanced by FNR (6). (Step 6) UhpT promotes the entry of G6P and FOS into the cell. The presence of G6P enhances the expression levels of UhpT. (Step 7) The UhpABC system promotes the expression of UhpT. UhpC binds extracellular G6P, and through UhpB, a phosphate group is transferred to UhpA (UhpA-P). (Step 8) UhpA-P is the activate form of UhpA and, together with cAMP-CRP complex, starts UhpT transcription, binding to the uhpT promoter. (Step 9) PtsG promotes the entry of glucose into the cell (7). (Step 10) GlpQ is a periplasmic glycerophosphodiester phosphodiesterase that converts periplasmic glycerophosphodiesters to G3P (8). PCR investigations of the two strains detected fosA10 in an NDM-producing strain of E. coli (ECO49406) and a fosA7 gene in a VIM-producing C. freundii strain (CFR50714). fosA10 was successfully transferred by conjugation, while attempts to transfer fosA7 through conjugation and transformation failed.
Based on short-read data and conjugation experiments, fosA10 was located on the IncK plasmid. fosA10 was in a genomic cassette of 3,835 bp consisting of excA-DEAD box-fosA10-lysR-IS10R. The BLAST results showed that the cassette shared 100% query and identity with the fosA10 cassette of the IncB/O/K/Z plasmid p542093_1 (accession no. CP091410.1) and 65% query and 100% identity with the fosA10 cassette reported in an IncFII plasmid (pHNPK9-Fos; accession no. MT074415.1) (16) collected from veterinary E. coli isolates in China. pHNPK9-fosA10 cassette (4,328 bp) differed only by (i) containing two copies of IS10 flanking fosA10 and the DEAD box and (ii) lacking excA (Fig. 2).
The FosA10-producing strain ECO49406 exhibited a wide range of single amino acid substitutions, categorized as neutral, in GlpT, UhpB, CyaA, PtsI, and GlpR. ECO49406 carried  the deleterious mutation G359E (P score, 23.077) in CyaA (Fig. 7). No alteration in the target MurA, the transporter UhpT, or the regulators PtsH, UhpA, UhpC, FNR (fumarate and nitrate reduction regulatory protein), and CRR were detected in any of the E. coli strains studied. These results imply a link between ST and mutations in the FOS pathway.
Regarding C. freundii isolates, three of nine belonged to ST98, two to ST95, two to ST65, one to ST673, and one to ST19. Analysis of the FOS pathway highlighted a similar link between ST and certain amino acid substitutions (Fig. 8). Two FOS r ST65 strains (CFR47299 and CFR47462) shared the same neutral substitutions in GlpT, UhpB, CyaA, UhpC, and GlpQ. Additionally, both isolates missed the first 4 amino acids (aa) (MLSI) and had the deleterious substitutions K6L, P7N, and A8Q (P scores, 24.8, 28.2, and 23.4, respectively) in GlpT. Two FOS s ST95 strains (CFR56415 and CFR51929) accumulated identical neutral substitutions in UhpB, CyaA, UhpC, and PtsG but shared the deleterious mutation V766A in CyaA (P score, 23.13). The same deleterious change in CyaA occurred in strains CFR47298 and CFR50714, belonging to ST98 and ST673, respectively. In contrast, strain CFR47298 had deleterious deletions in GlpT (Y406del; P score, 211.6) and in UhpC (F112L; P score, 25.28). Strain CFR67526, the only ST19 strain, shared the same reported neutral alteration in CyaA. Interestingly, the glpT sequence had an insertion of 25 nucleotides (nt) at position 788, leading to a 1,384-nt instead of a 1,359-nt gene. The frameshift mutation affects the amino acids from position 251 to 263aa, which could possibly affect the activity of the transporter (Fig. 8). No alterations in MurA and in the regulators UhpA, PtsH, PtsI, GlpR, CRP, and CRR were detected in any of the C. freundii strains studied. These results together show that the occurrence of mutations at different levels can decrease FOS susceptibility.

DISCUSSION
Fosfomycin has regained importance in clinical practice and has offered an alternative first-line option against MDR bacterial infections. The pathway of fosfomycin inside bacterial cells depends mainly on GlpT and UhpT activity (2). Acquiring mutations in GlpT and UhpT can impair their transport activity, decreasing FOS uptake into the bacterial cell and hence FOS effectiveness (41)(42)(43). However, modifications in these proteins have a high fitness cost, leading to the predominance of FOS-susceptible strains (44).
In the current study, all sequenced E. coli strains had amino acid substitutions in GlpT. The substitutions E448K, Q444E, and E443Q, categorized as neutral by PROVEAN, were reported previously by Takahata et al. and Sorlozano-Puerto et al. and are recognized as not impacting GlpT functionality (3,9). L297F has been categorized as neutral (P score, 22.375) and was reported previously by Sorlozano-Puerto and colleagues (9); however, the impact on GlpT function has not been investigated yet. Two FOS r E. coli ST131 strains (ECO52246 and ECO52259) contained W28del, classified as deleterious (P score, 212.042). We speculate that there is a possible impact of W28del on GlpT activity, leading to a FOS r profile in E. coli ST131 strains. E. coli ST131 is a hypervirulent and pandemic clone (45) associated with the global spread of ESBLs such as CTX-M-15, KPC-like, and NDM-like enzymes (46)(47)(48). The acquisition of additional antimicrobial resistance traits in such successful clones can impact the clinical outcome of infections by such isolates and reduce antibiotic availability.
UhpB is a component of the UhpABC system and is a membrane-associated protein kinase that autophosphorylates and subsequently transfers its phosphate group to  Here, we describe the first genetic analysis of mutations detected in FOS s and FOS r C. freundii isolates and their effect on FOS susceptibility. In our study, several amino Interestingly, all C. freundii isolates except C. freundii ST65 strains had the deleterious mutation V766A in CyaA (P score, 23.126). Regarding the UhpB regulator, all C. freundii harbored the neutral substitutions L342R and D430N and the mutations A75V (P score, 21.66) and S461N (P score, 20.7) (except for C. freundii ST65 strains). Regarding the FOS r C. freundii ST98 strain (CFR 47298), we report the deletion Y406del (P score, 211.6) in GlpT and the deleterious single substitution F112L (P score, 25.28) in UhpC, which could be implicated in FOS resistance. C. freundii ST65 and ST98 are emerging clones involved in the spread of ESBLs. Samuelsen et al. described two cases of CMY-481OXA-10-coproducing C. freundii ST65 from clinical samples in Denmark, while Schweizer et al. described an outbreak event in Germany, caused by KPC-2-producing C. freundii ST98 isolates (49,50). The occurrence of deleterious substitutions in proteins implicated in FOS uptake could decrease the effectiveness of FOS and its use against infections by clinically relevant clones such as carbapenemase-producing C. freundii ST98 isolates. E. coli ST648 is an international high-risk and pathogenic clone, in both clinical and veterinary settings (51). Recently, it has been recognized as a pandemic clone, able to carry carbapenemases such as KPC-2 (52), NDM-1 (53)(54)(55), and OXA-48 (53,55). In 2016, Yang et al. reported an E. coli ST648 strain coproducing NDM-5, CTX-M-55, MCR-1, and FosA3 from a duck in China (56). Here, we report the first case of E. coli ST648 coproducing NDM-5 and FosA10 isolated from humans in the Czech Republic. The genomic environment of fosA10 showed perfect identity with the fosA10-carrying IncI1 plasmid obtained from clinical E. coli ST227 strains in the United Kingdom and close similarity with the fosA10-carrying IncFII plasmid from veterinary E. coli ST38 collected in China (57). The global ST648 epidemiology and our findings focus attention on the ease of acquisition of MDR genes in this clone, which could drastically reduce the number of still-active antibiotics. Moreover, the findings highlight (i) the transition of fosA10 from veterinary to clinical settings, (ii) the ability of the fosA10 cassette to fit in both the IncFII and IncB/O/K/Z environments, and (iii) the ability of fosA10-carrying IncK plasmid to switch from minor to successful clones, such as ST648.
Moreover, we report the first case of a C. freundii ST673 isolate producing VIM and carrying a new FosA7 variant, named FosA7.9. The FosA7.9 gene was inserted in a wellconserved cassette, surrounded by two copies of the HNH endonuclease gene. The HNH endonuclease is a group of homing endonucleases that can act as selfish genetic elements, like transposons, breaking DNA double strands and allowing the acquisition of functional attributes to the host cell, such as antimicrobial resistance (AMR) genes (58,59). Moreover, the genomic environment of the fosA7 cassette, flanked by HNH genes, was shared with two clinical C. freundii ST396 isolates collected in China (38). These results and the lack of any insertion elements surrounding fosA7.9 suggest a possible role for HNH endonucleases in slowly spreading new AMR traits in low-risk and silent hosts, such as C. freundii (35).
Conclusions. These results show the emergence of FOS r among CRE from clinical settings in the Czech Republic. The 3-year study revealed a decrease in FOS susceptibility among carbapenemases-producing E. coli and Citrobacter strains. In our investigation, the decrease of FOS s could be largely attributable to impairment in GlpT and CyaA activity, significantly reducing the permeability to FOS. To our knowledge, we report the first isolation of FosA10-producing E. coli ST648 and the emergence of a
Identification of strains was confirmed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) with MALDI Biotyper software (Bruker Daltonics, Bremen, Germany). The production of carbapenemases (metallo-b-lactamase, OXA-48, and KPC) was assessed with the double-disc synergy test with EDTA, the temocillin disc test, and the phenylboronic acid test (37). FOS MICs were evaluated using ADM and interpreted according to EUCAST clinical breakpoints (v 12.0) and guidelines. Production of FosA-like and FosC2 enzymes was detected by the PPF test (60).
WGS and analysis. A total of 15 strains (6 E. coli and 9 C. freundii strains) were selected as representative for genomic content and FOS pathway mutation analysis. The selection was based on FOS MICs: the selection included 11 strains (four E. coli and seven C. freundii strains) with FOS MICs of $64 mg/mL (resistant [R]) and four strains (two E. coli and two C. freundii strains) with FOS MICs of #16 mg/mL (S). The four FOS-susceptible strains were included to compare differences in FOS resistance (FOS r ) and FOS susceptibility (FOS s ) profiles. In detail, the selection for FOS s E. coli included FOS MICs corresponding to the FOS epidemiological cutoff (ECOFF) value and one dilution beyond the FOS ECOFF value according to EUCAST (https://mic.eucast.org/search/?search%5Bmethod%5D=mic&search%5Bantibiotic%5D=100& search%5Bspecies%5D=-1&search%5Bdisk_content%5D=-1&search%5Blimit%5D=50). For FOS s C. freundii isolates, FOS ECOFF values are not available; thus, low-level susceptibility profiles for FOS were selected (FOS MIC = 8 mg/mL and 16 mg/mL).
Genomic DNA was extracted using a NucleoSpin microbial DNA kit (Macherey-Nagel, Germany). WGS was performed on seven selected strains with the NovaSeq 6000 system with a 2 Â 250 paired-end run following Nextera XT library preparation (Illumina Inc., San Diego, CA, USA). The remaining eight strains belonged to two different projects (Table 1), and WGS was previously performed with both the Illumina MiSeq platform (Illumina Inc., San Diego, CA, USA) and the Sequel I platform (Pacific Biosciences, Menlo Park, CA, USA) (36). Reads were assembled using SPAdes software (62). Assembled sequences were annotated using the RAST (Rapid Annotation using Subsystems Technology) server (63). The resistome, plasmid replicons, mobile elements, multilocus sequence types (MLST), and plasmid MLST (pMLST) were determined by uploading the assembled sequences to

Molecular Insight into FOS Resistance Mechanisms
Microbiology Spectrum chromosomal environments was created by using EasyFig (69) and the graphic editor Procreate (Savage Interactive, Tasmania, Australia). Phylogenetic analysis. Phylogenetic relationships between the selected sequenced isolates and global genomes were investigated. Phylogenetic trees were obtained using core genome, recombination, and SNPs by using parsnp v1.2, available in the harvest suite (70), and using a corresponding reference genome. Graphic illustration of the trees was build with the Interactive Tree Of Life (iTOL) (https:// itol.embl.de/) (71). For the construction of the SNP-based phylogenies, 160 Escherichia coli genomes and 112 Citrobacter freundii genomes were downloaded from the NCBI assembly database, including complete and draft genomes. E. coli ECO49406 and C. freundii RHBSTW-00135 were use as respective references. The evolutionary analysis of FosA-like proteins in Enterobacterales was conducted by MEGA 11 (72), using the maximum-likelihood method and the Jones-Taylor-Thornton (JTT) matrix-based model (73).
Protein mutations. The effects of amino acid alterations on the biological function were predicted using the online PROVEAN (Protein Variation Effect Analyzer) platform (http://provean.jcvi.org/index .php) (76). PROVEAN predicts protein sequence variations, including single or multiple amino acid substitutions, insertions, or deletions. The platform produces a delta alignment score based on the reference and variant versions of a protein query sequence with respect to sequence homologs collected from the NCBI protein database through BLAST. For each substitution, the tool provides a score (P score) categorized in three classes: (i) if the P score is equal to or below the cutoff of 22.5, the protein alteration is categorized as deleterious (potential loss of protein structure or function); (ii) if the P score is above the threshold, the alteration is marked as neutral (no alteration in the structure or function of the protein) (9). Amino acid variations in MurA, GlpT, UhpT, UhpA, UhpB, UhpC, CyaA, PtsI, PtsH (phosphocarrier protein HPr), GlpR, CRP, CRR (enzyme IIA [Glc]), PtsG (phosphotransferase system [PTS] glucose-specific EIICB component), FNR (fumarate and nitrate reduction regulatory protein), and GlpQ (glycerophosphodiester phosphodiesterase) were investigated (accession numbers are reported in Table S1).
Data availability. The nucleotide sequence of fosA7.9 has been uploaded to GenBank under the accession number ON245013. GenBank accession numbers of the sequenced strains are presented in Table 1.