In vitro Synergistic Activities of Fosfomycin in Combination with Other Antimicrobial Agents Against Carbapenem-Resistant Escherichia coli Harboring bla NDM-1 on the IncN2 Plasmid and a Study of the Genomic Characteristics of These Pathogens.

Purpose: The spread of New Delhi metallo-β-lactamase (NDM) encoded by the bla NDM gene has been a global health crisis for many years. Most of bla NDM-harboring bacteria commonly carry various antimicrobial resistance (AMR) genes on their chromosomes or plasmids, leading to limited treatment options. Thus, we aimed to evaluate the synergistic effects of fosfomycin in combination with other antimicrobial agents against bla NDM-harboring carbapenem-resistant Escherichia coli (CREC) and to characterize the whole-genome and plasmid sequences of these pathogens.
Methods: Thirty-eight CREC isolates were collected from patients in the Medicine Ward, Songklanagarind Hospital, Thailand. The activity of fosfomycin in combination with other antimicrobial agents against CREC isolates harboring bla NDM on the plasmid was evaluated using the checkerboard method. In this method, the serial dilutions of two antibiotics were mixed with the cultured CREC, the mixtures were incubated, and FICI was calculated to interpret the synergistic activity of the combination. The whole-genome and particular plasmids of these pathogens were sequenced using next-generation sequencing. Sequence analysis, especially on antimicrobial resistance (AMR) genes, mobile-genetic elements (MGEs), and virulence genes was performed using many bioinformatics tools.
Results: Of the E. coli 38 isolates, only 3 isolates contained the bla NDM-1 gene, which is located on the IncN2 plasmid. The combinations of fosfomycin with aminoglycosides, colistin, tigecycline, sitafloxacin, and ciprofloxacin were synergies against bla NDM-1-harboring CREC isolates. Genomic analysis revealed that these isolates harbored many β-lactam resistance genes and other AMR genes that may confer resistance to aminoglycoside, fluoroquinolone, rifampicin, trimethoprim, sulfonamide, tetracycline, and macrolide. Also, various MGEs, especially the bla NDM-1-bearing IncN2 plasmid, were present in these isolates.
Conclusion: Our study demonstrated some synergistic effects of antimicrobial combination against CREC isolates harboring bla NDM-1 on the IncN2 plasmid. Also, our data on the whole-genome and plasmid sequences might be beneficial in the control of the spread of bla NDM-1-harboring CREC isolates. The linkages between bla NDM-1-carrying plasmid, patient information, and time of collection will be elucidated to track the horizontal gene transfer in the future.

Introduction

Carbapenem-resistant Enterobacteriaceae (CRE) has been a serious public health concern for many decades. Carbapenems were an important, life-saving mainstay treatment for many years, but due to widespread over- and inappropriate use, in recent years widespread resistance to this antibiotic has developed.1 Importantly, as CRE can hydrolyze almost all β-lactam antibiotics, the treatments are limited, resulting in a high mortality rate.

CREC resists carbapenem through various mechanisms, especially the production of carbapenemase enzymes.2 The carbapenemase-encoding genes are usually located on mobile genetic elements (MGEs) that can transfer these particular genes to other pathogenic bacteria. Among them, the blaNDM gene is one of the most prevalent and strong carbapenemase genes with high hydrolysis activity.3 The outbreak of blaNDM gene has been reported in many parts of the world such as Australia (1.6%), America (11%), Africa (11%), Europe (17%), and Asia (58%) and the evidence therefore confirm the global spread of the blaNDM gene.4 Moreover, the infection caused by this pathogen seriously threaten the patient with a high mortality rate.5 The blaNDM gene can be detected on many types of plasmids and transposons (Tn), especially Tn125.6 A previous study reported that Acinetobacter spp. acted as the intermediate host to disseminate the blaNDM-1 into Enterobacteriaceae.7 The blaNDM-1 gene is mostly found within the bracket of two copies of ISAba125 in the form of Tn125, which provides a −35 region promoter for the expression of blaNDM-1. However, the genetic content of Tn125 can be truncated due to the insertion of various insertion sequence (IS) elements.8 Outbreaks of blaNDM have been reported on almost all continents. Twenty replicon types of blaNDM-bearing plasmids, IncHI2, IncHI3, IncN, IncN2, IncC, IncB/O/K/Z, IncFIA, IncFIB, IncFIC, IncFIII, IncHI1, IncL/M, IncP, IncR, IncT, IncX1, IncX3, IncX4, IncY, and ColE10, have been identified.3 The co-existence of other antimicrobial resistance (AMR) genes located on these plasmids has led to the current crisis and challenging clinical situation.

In the treatment of CREC infections, the widespread use of carbapenems as monotherapy has resulted in increasing resistance to these antibiotics and other β-lactams. A combination of antibiotics therefore could be an optional therapy to reduce further resistance development and/or enhance the antimicrobial activity. A previous study reported that the combination of a carbapenem (e.g., ertapenem) with fosfomycin resulted in a synergistic effect of increased antimicrobial activity. This theory could lead to effective alternative treatments for infections caused by carbapenem-resistant Enterobacteriaceae (CRE).9

Although some antibiotics such as fosfomycin, tigecycline, and colistin are considered as a treatment option for the CRE infections, fosfomycin is mostly considered for severe patients due to its broad-spectrum bactericidal activity and less toxicity than other antibiotics of choice. However, many studies suggested that fosfomycin should be combined with other antibiotics to avoid the fosfomycin resistance that might be rapidly developed during the therapy.10,11 Additionally, there are few studies reporting the fosfomycin-based combination against blaNDM-1-harboring CREC. Thus, the objective of this study was to assess the synergistic activities of fosfomycin in combination with other antimicrobial agents against blaNDM-1-harboring CREC, and to genetically characterize the entire genome of these pathogens.

Materials and Methods

Bacterial Isolation and Identification

The bacterial isolates were obtained from colonized patients in the Medicine Ward, Songklanagarind Hospital, between March and April 2017. One µg/mL of meropenem containing MacConkey agar plates were used to screen and select carbapenem-resistant Gram-negative bacteria.12 The strain of Escherichia coli was identified by Matrix-Assisted Laser Desorption/Ionization Time-of-Flight (MALDI-TOF).

Antimicrobial Susceptibility Testing (AST) and Detection of the blaNDM Gene

Thirty-eight E. coli isolates were confirmed to be carbapenem-resistant isolates by evaluation of the minimum inhibitory concentrations (MICs) of imipenem and meropenem through the broth microdilution method. E. coli ATCC 25922 was used as the quality control.13 The blaNDM gene was detected among these CREC isolates by a PCR method using the NDM-F primer (GGTTTGGCGATCTGGTTTTC) and the NDM-R primer (CGGAATGGCTCATCACGATC).10,14 The PCR products were then checked using agarose gel electrophoresis and the expected size was 621 bp. In addition, the MIC values of other 9 antibiotics (fosfomycin, ciprofloxacin, doripenem, levofloxacin, tigecycline, sitafloxacin, colistin, gentamicin, and amikacin) and 1 β-lactamase inhibitor (sulbactam) (Sigma-Aldrich) were evaluated against blaNDM-harboring isolates by the broth microdilution method, according to the Clinical and Laboratory Standards Institute (CLSI) guidelines, 2016 or US Food and Drug Administration (FDA) breakpoints for tigecycline or European Committee on Antimicrobial Susceptibility Testing (EUCAST) for colistin.13,15,16

Evaluation of Synergistic Effects

All blaNDM-harboring CREC isolates were assessed for the synergistic activity of fosfomycin in combination with imipenem, meropenem, doripenem, gentamicin, amikacin, ciprofloxacin, levofloxacin, sitafloxacin, colistin, tigecycline, and sulbactam (Sigma-Aldrich, USA) using the checkerboard method. As we used fosfomycin, the Clinical and Laboratory Standards Institute (CLSI) recommended adding 25 mg/mL of glucose-6-phosphate (G-6-P) in the 96-well plates containing Mueller-Hinton broth (MHB). The Fractional Inhibitory Concentration Index (FICI) was used to interpret the synergism (FICI ≤ 0.5), indifference (0.5 < FICI ≤ 4.0), or antagonism (FICI > 4.0).

DNA Extraction and Whole-Genome Sequencing (WGS)

The genomic DNA (gDNA) of all blaNDM-harboring CREC isolates was extracted using a TIANamp Bacterial DNA Kit (Tiangen, Beijing, China), following the manufacturer’s instructions. Then, the quality of the gDNA was checked in terms of concentration and purity with a Thermo Scientific™ NanoDrop 2000 and gel electrophoresis, respectively. The qualified DNA samples were then sent to Beijing Genomics Institute (BGI), Beijing, China, for conducting short-read sequencing (WGS) with the use of a BGISEQ-500 sequencer (Beijing Genomics Institute, Beijing, China).

Plasmid Extraction and Sequencing

Plasmid DNA was extracted by the alkaline lysis method, as previously described.17 Briefly, the bacteria were treated with a base solution (0.2 N NaOH-1% SDS) to denature the double-strand DNA. An acid solution (4M potassium acetate-2M acetic acid) was immediately added to neutralize the mixture, which was then centrifuged at high speed to separate the plasmid DNA from the chromosomal DNA. The purity of the plasmid DNA was visualized by agarose gel electrophoresis. The qualified plasmids were then sequenced using NovaSeq 6000 sequencer (Beijing Headquarters Novogene Co., Ltd. Beijing, China).

Data Analysis of WGS and Plasmids

All sequence reads were de novo assembled using Unicycler v0.4.7.18 The assembled sequences were then annotated using Prokka v1.12.19 For analysis of these sequence reads, the contigs of all CREC isolates were uploaded to the Center for Genomic Epidemiology (CGE) (), a free online bioinformatics service allowing scientists to analyze sequencing data in relation to infectious diseases. Here, we used ResFinder 4.1 ()20 and PlasmidFinder 2.1 (),21 with a set of 95% identity and 80% minimum length for the detection of acquired AMR genes and plasmids, respectively. Meanwhile, VirulenceFinder 2.0 ()22 with a set of 90% identity and 60% minimum length and MLST 2.0 (Multi-Locus Sequence Typing) ()23 were used to identify virulence genes and sequence types (STs), respectively. In addition to the CGE website, we also used other web-based bioinformatics tools. For other mobile-genetic elements (MGEs), insertion sequences (ISs) and integrons were also investigated using blastn against the IS database and integron_finder, respectively. The bacteriophage genomes were explored using phaster (),24 while CRISPR-Cas regions were detected using CRISPRCasFinder (). The bacteriocin-encoding genes were also investigated using BAGEL4 ().25

Pan-Genome and Phylogenetic Analysis

The pan-genome profiles of blaNDM-harboring CREC isolates were compared to published genomes from Africa (n = 11), Austria (n = 39), Germany (n = 120), and Thailand (n = 12), which are available from the National Center for Biotechnology Information (NCBI).26–29 The pan-genome analysis was performed using Roary v3.13.0,30 with minimum blastx identity and core definition threshold at 95% and 99%, respectively. A phylogenetic tree was generated from the pan-genome alignment, based on accessory genes, using a maximum likelihood method (1,000 bootstraps). The interactive visualization of the phylogenetic tree was performed using the Geneious R10.2631 and Phandango website ().32 Additionally, a pan-genome frequency plot, a pie chart of the pan-genome, and a presence and absence matrix against a phylogenetic tree were also created using roary_plots script ().

Data Availability

The assembled genomes of the blaNDM-1-harboring CREC isolates from this study have been submitted to the NCBI GenBank under BioProject number PRJNA780210 with BioSample numbers SAMN23132882, SAMN23132883, and SAMN23132884.

Results

Bacterial Isolates and Clinical Data

A total of 38 E. coli isolates were collected from the patients and were classified the carbapenem resistance. Due to a screening method can be used to collect the isolates that provide non-susceptible to the carbapenem (intermediate resistance and resistance), only 6 (15.8%) out of 38 isolates were then identified as CREC by phenotypic method and a half of the CREC isolates (n = 3) were positive for the blaNDM gene. These 3 blaNDM-harboring isolates were obtained from the rectums of 2 female and 1 male patients, who had cardiac arrest or pulmonary edema as common underlying disease. Importantly, all 3 patients have previously received carbapenem and other antibiotics, which might be the reason for carbapenem-resistance isolates.

Antimicrobial Susceptibilities in CREC Isolates

The MIC values of the 11 tested antibiotics against the 3 isolates of blaNDM-1-carrying CREC are shown in Table 1. The 3 isolates were resistant to imipenem, meropenem, and doripenem with high MIC values ranging from 16 to 64 µg/mL. These isolates were resistant to all tested fluoroquinolones, ciprofloxacin (32 µg/mL), and levofloxacin (64 and 128 µg/mL), while sitafloxacin resistance (8 µg/mL) was found in 2 isolates. The low level of colistin resistance was also observed in all isolates. In addition, resistance to gentamicin and fosfomycin was detected in 2 (16 and 32 µg/mL) and 1 (256 µg/mL) isolates, respectively. However, all isolates were still susceptible to tigecycline and amikacin.

Table 1

Minimum Inhibitory Concentration (MIC) Values of 11 Antibiotics and 1 β-Lactamase Inhibitor Against blaNDM-1-Harboring CREC Isolates

Isolate Code	MIC Value (µg/mL)
	FOS	IPM	MEM	DOR	CIP	LVX	TIG	STFX	CT	GN	AMK	SUL
CREC003	0.25 (S)	32 (R)	64 (R)	16 (R)	128 (R)	32 (R)	0.5 (S)	8 (R)	4 (R)	32 (R)	2 (S)	128
CREC004	0.5 (S)	16 (R)	32 (R)	32 (R)	64 (R)	32 (R)	0.5 (S)	4 (I)	4 (R)	16 (R)	2 (S)	128
CREC038	256 (R)	32 (R)	64 (R)	32 (R)	128 (R)	32 (R)	1 (S)	8 (R)	4 (R)	1 (S)	2 (S)	128

Abbreviations: FOS, fosfomycin; IPM, imipenem; MEM, meropenem; CIP, ciprofloxacin; DOR, doripenem; LVX, levofloxacin; SUL, sulbactam; TIG, tigecycline; STFX, sitafloxacin; CT, colistin; GN, gentamicin; AMP, amikacin; R, resistance; I, intermediate resistance; S, susceptible.

Synergistic Effects Against CREC Isolates

The synergy study showed that synergistic effects of fosfomycin plus gentamicin, amikacin, ciprofloxacin, sitafloxacin, colistin, or tigecycline were observed in 1 isolate, while there was no difference in the combination of fosfomycin with imipenem, meropenem, doripenem, and levofloxacin in any isolates (Table 2). The antagonistic effect of fosfomycin in combination with sulbactam was seen in 1 isolate (Table 2).

Table 2

The Synergistic Effects of Fosfomycin in Combination with Other Antimicrobial Agents Against blaNDM-1-Harboring CREC Isolates

Combination	No. of Isolates
	Synergy	Indifference	Antagonistic
Fosfomycin + Imipenem	0	3	0
Fosfomycin + Meropenem	0	3	0
Fosfomycin + Doripenem	0	3	0
Fosfomycin + Gentamicin	1	2	0
Fosfomycin + Amikacin	1	2	0
Fosfomycin + Ciprofloxacin	1	2	0
Fosfomycin + Levofloxacin	0	3	0
Fosfomycin + Sitafloxacin	1	2	0
Fosfomycin + Sulbactam	0	2	1
Fosfomycin + Colistin	1	2	0
Fosfomycin + Tigecycline	1	2	0

Sequence Types (STs) and Antimicrobial Resistance (AMR) Genes in CREC Isolates

In the MLST analysis, CREC003 and CREC004 were identified in ST448, while CREC048 was identified in ST131, as shown in . The study found that the blaNDM-1 gene was present in the 3 isolates noted previously. Other β-lactam resistance genes, blaTEM-1C, blaTEM-57, blaCTX-M-14, blaCTX-M-15, and blaCMY-2, were also found in these CREC isolates. Additionally, the CREC isolates carried other AMR genes, conferring resistance to aminoglycosides (aac(3)-IId, aph(3”)-Ib, aph(6)-Id, aadA5, aadA16, and aac(6’)-Ib-cr), fluoroquinolone (qnrB6), rifampicin (ARR-3), trimethoprim (dfrA17 and dfrA27), sulfonamide (sul1 and sul2), tetracycline (tet(A)), and macrolide (mph(A)). The distribution of AMR genes is illustrated in Figure 1 and .

Figure 1

Distribution of antimicrobial resistance (AMR) genes in blaNDM-1-harboring CREC isolates.

Mobile Genetic Elements (MGEs) in CREC Isolates

Various mobile genetic elements (MGEs) were identified in the blaNDM-1-harboring CREC isolates, as shown in Figure 2 and –. For plasmids, we found IncFIA, IncFII(K), and IncN2 plasmids in all isolates. The IncA/C2, IncFIB (AP001918), IncFII (pAMA1167-NDM-5), IncQ1, and IncX4 plasmids were identified in only the ST448 isolates, while the IncFII (pRSB107) plasmid was only observed in the ST131 isolate.

Figure 2

Distribution of mobile genetic elements (MGEs) in blaNDM-1-harboring CREC isolates.

Many insertion sequences (ISs), namely IS102, IS1222, IS2, IS200F, IS4, IS5, IS903, ISEc31, ISKpn26, ISLad1, ISSen4, and ISYen3, were also detected in all CREC isolates. ISEc17, ISEc27, ISEc44, ISEch5, and ISKpn43 were only found in the ST448 isolates, while ISEc13, ISEc24, ISEc45, ISKpn37, ISPlge3, and ISSen1 were only identified in the ST131 isolate. For integrons, a class 1 integron with the carriage of aac(6)-Ib gene responsible for aminoglycoside resistance was detected in the ST131 isolate.

Virulence-Associated Genes in CREC Isolates

Besides the identification of AMR genes, we also looked for virulence-associated genes and the results are shown in Figure 3 and . We found the fyuA, gad, irp2, terC, and traT genes in all 3 blaNDM-1-harboring CREC isolates, while the afaA, afaC, afaD, lpfA, and nfaE genes were only detected in the ST448 isolates, and the chuA, yfcV, iss, iha, ompT, iucC, iutA, sitA, kpsE, kpsMII_K5, papA_F43, sat, and usp genes were only found in the ST131 isolate.

Figure 3

Distribution of virulence-associated genes in blaNDM-1-harboring CREC isolates.

CRISPR-Cas Region in CREC Isolates

In the analysis of the CRISPR-Cas system, 5 to 7 CRISPR positions were found in 3 isolates of blaNDM-1-harboring CREC. The CRISPR positions contained 1 to 18 spacers. Additionally, 2 positions of cas locus was also identified in all ST448 isolates, which were further classified as CRISPR-Cas type-IF and type-IE. The details of the CRISPR-Cas region are provided in Tables 3 and .

Table 3

CRISPR-Cas System in blaNDM-1-Harboring CREC Isolates

Isolate Code	Source of Isolation	Sequence Type (ST)	CRISPR Region			Cas Region
			CRISPR Position	No. of Direct Repeat (DR)	No. of Spacer	Cas Position	cas Locus	Cas Type
CREC003	Rectum	448	Position 1	2	1	Position 1	cas6_csy3_csy2_csy1_cas3-cas2_cas1	Cas type-IF
			Position 2	19	18	Position 2	cas2_cas1_cas6_cas5_cas7_cse2_cse1_cas3	Cas type-IE
			Position 3	14	13
			Position 4	2	1
			Position 5	12	11
			Position 6	10	9
			Position 7	2	1
CREC004	Rectum	448	Position 1	14	13	Position 1	cas1_cas3-cas2_csy1_csy2_csy3_cas6	Cas type-IF
			Position 2	19	18	Position 2	cas2_cas1_cas6_cas5_cas7_cse2_cse1_cas3	Cas type-IE
			Position 3	2	1
			Position 4	12	11
			Position 5	10	9
			Position 6	2	1
			Position 7	2	1
CREC038	Rectum	131	Position 1	2	1	-	-	-
			Position 2	2	1
			Position 3	5	4
			Position 4	7	6
			Position 5	8	7

Bacteriophage Genome in CREC Isolates

We found bacteriophage genomes in these isolates, as shown in Figure 4 and . PHAGE_Entero_mEp460 genome was found in all CREC isolates. Two bacteriophage genomes, PHAGE_Entero_cdtI and PHAGE_Klebsi_4LV2017, were only detected in the ST448 isolates, while 4 bacteriophage genomes, PHAGE_Escher_TL_2011b, PHAGE_Entero_BP_4795, PHAGE_Entero_P88, and PHAGE_Pectob_ZF40, were only found in the ST131 isolate.

Figure 4

Presence of bacteriophage genomes in blaNDM-1-harboring CREC isolates.

Bacteriocin Class in CREC Isolates

The results of bacteriocins are shown in Figure 5 and . We found that 3 genes, encoding bottromycin, sactipeptides, and colicin, were detected in the ST131 isolate, while only bottromycin was identified in the ST448 isolates.

Figure 5

Presence of bacteriocin-encoding genes in blaNDM-1-harboring CREC isolates.

blaNDM-1-Bearing IncN Plasmid in CREC Isolates

All 3 E. coli isolates carried the blaNDM-1 gene on a plasmid (pSK20-NDM1, pSK21-NDM1, and pSK22-NDM1) belonging to the IncN2 incompatibility group. The structure of the blaNDM-1-bearing IncN2 plasmid is shown in Figure 6. The pSK20-NDM1, pSK21-NDM1, and pSK22-NDM1 plasmids were closely similar, all having a length of 41,190 bp. All of the IncN2 plasmids had the same backbone, consisting of genes encoding for plasmid stability (stbA, stbB, and stbC), replication (repA), and conjugal transfer (tra). Only one antibiotic resistance gene was found, blaNDM-1. The genetic environment of blaNDM-1 shares a single conserved region, which was a Tn3-like structure, ISAba125, ISEc33, blaNDM-1, blaMBL, ΔtnpF, ISSen4, and Tn5403. The blaNDM−1 gene was bracketed by two insertion sequences (IS), ISEc33 and ISSen4, which belong to the IS630 and IS3 families, respectively. The upstream ISAba125 was interrupted by the insertion of ISEc33 but not the part of the promoter involved in the expression of blaNDM-1. The sequence analysis revealed that the accessory module was inserted into the IncN2 plasmid between the fipA and nuc genes.

Figure 6

Structure of IncN2 plasmid in blaNDM-1-harboring CREC isolates.

Pan-Genome and Phylogenetic Tree of CREC Isolates

The results of the pan-genome analysis are shown in Figure 7, while the phylogenetic tree is visualized in . From 26,093 genes recorded in our study, 23,331 (89.41%) and 2,762 (10.59%) genes were identified as accessory and core genes, respectively. In comparison to previously published genomes, the CREC003 and CREC004 isolates (ST448) could be grouped in the same clade, while the ST131 isolate (CREC038) was located in a different clade (Figures 7 and S1). The ST448 isolates contained the ymfD, ymfE, and nfaA genes encoding the e14 prophage with SAM-dependent methyltransferase, the e14 prophage with inner membrane protein, and the non-fimbrial adhesin 1, respectively. The flhC_1, mntH_2, alba, moaA_2, and csoR genes encoding the flagellar transcriptional regulator FlhC, the divalent metal cation transporter MntH, the antilisterial bacteriocin subtilosin biosynthesis protein AlbA, the GTP 3’,8-cyclase, and the copper-sensing transcriptional repressor CsoR were present as important features in ST131 isolate.

Figure 7

Genomic diversity among blaNDM-1-harboring CREC isolates compared to previously published genomes.

Discussion

In 2019, the Centers for Disease Control and Prevention (CDC) considered the infection caused by CRE as an urgent threat, due to the high level of estimated cases in hospitalized patients, mortality rates, and attributable health-care costs. CRE, particularly CREC, developed resistance mechanisms for many antibiotic classes such as carbapenems, aminoglycosides, and fluoroquinolones.33,34 Here, we report on the synergistic effects of fosfomycin in combination with other antimicrobial agents against blaNDM-1-harboring CREC isolates as well as the genomic characteristics of these pathogens. According to the MIC results, we found that all 3 blaNDM-1-harboring CREC isolates were resistant to more than 3 antibiotic classes, indicating multidrug resistance in these isolates. Although they were susceptible to tigecycline and amikacin, the use of these antibiotics as monotherapy still needs to be of concern since the target bacteria can develop resistance mechanisms during therapy.35 We then evaluated the synergistic activity of antibiotic combination for combating these pathogens. In the synergism study, we found that the use of fosfomycin combined with aminoglycosides, colistin, tigecycline, and sitafloxacin provided synergistic effects against the ST131 isolate of the blaNDM-1-harboring CREC. Similar to previous study, Cebrero-Cangueiro et al (2021) found that fosfomycin plus gentamicin showed a synergistic effect against carbapenemase-producing Enterobacteriaceae (CPE).36 The synergistic effects of fosfomycin plus colistin and tigecycline have been also reported in approximately 20% of the carbapenem-resistant K. pneumoniae (CRKP).10 The combination of fosfomycin with ciprofloxacin showed the synergistic effect against only 1 isolate of the ST448 in our study. The benefit of combination therapy is to suppress the emerging resistance mechanisms that probably occur during the treatment.37

In WGS analysis, 2 and 1 isolates of blaNDM-1-harboring CREC were identified in ST448 and ST131, respectively. These STs have been previously identified as an internationally spread clone.38,39 The E. coli ST131 clone has been reported as the major clone causing human infections and is predominantly found in Southeast Asia.40 The previous study has been demonstrated that the majority of this strain carried the blaKPC gene which commonly produces Klebsiella pneumoniae carbapenemase (KPC) and this gene is easily transferred to another bacteria through the MGEs. Thus, it raises the concern about the increase of carbapenem-resistant strains.41 In AMR detection, we found that the CREC isolates carried many AMR genes conferring resistance to carbapenems, third-generation cephalosporins, other β-lactams, aminoglycosides, fluoroquinolones, rifampicin, trimethoprim, sulfisoxazole, tetracycline, and macrolides. These isolates have also been classified as multidrug-resistant (MDR) isolates, which was in concordance with our MIC results. In Germany, Welker et al (2020) reported that E. coli ST131 carried both carbapenem and other β-lactam resistance genes.42 In South Korea, Shin et al (2021) also demonstrated that 23 AMR genes, especially the blaNDM gene, were present in E. coli isolates.43 Here, we also assessed the correlation among the antimicrobial susceptibility, synergistic activity of fosfomycin-based combination, and WGS results. The combinations of fosfomycin with aminoglycosides were synergy in the ST131 isolate having low MIC values of gentamicin and amikacin, which carried a lower number of aminoglycoside resistance genes than the ST448 isolates having a high MIC value of gentamicin (Tables 1 and 2, Figure 1).

In our study, MGEs were also investigated to assess the dissemination mechanisms of the AMR genes. A previous study reported the presence of 16 replicon types of blaNDM-1-bearing plasmids in Enterobacteriaceae, namely ColE, IncA/C, IncN2, IncFIA, IncFIB, IncFII, IncHI1, IncHI3, IncHIB, IncL/M, IncP, IncR, IncT, IncX1, IncX3, and IncY.3 In this study, IncFIA, IncF(II)K, and IncN2 plasmids were detected in all CREC isolates, while the blaNDM-1 gene was identified in the IncN2 plasmid. The blaNDM-1-bearing IncN2 plasmid shared >99% nucleotide identity and 100% query coverage with pJN24NDM1 (GenBank accession no. MK368725), pC2972-5-NDM (GenBank accession no. CP039806), pTR3 (GenBank accession no. JQ349086), and pNDM-ECS01 (GenBank accession no. KJ413946), according to the BLAST search. The pJN24NDM1 is an IncN2 plasmid first identified in an E. coli ST405 isolate from the abdominal fluid of a patient in China,44 while the pC2972-5-NDM and pTR3 plasmids have been reported in K. pneumoniae isolates from China45 and Singapore,46 respectively. These highly similar plasmids detected from different geographic areas and bacterial species can be assumed to spread by lateral dissemination.47 The first blaNDM-1-bearing IncN2 was identified in NDM-1-producing E. coli, which was isolated from a patient hospitalized in Bangladesh.48 In 2014, blaNDM-1-bearing IncN2 (pNDM-ECS01) was also detected in E. coli ST131 from Central Thailand,49 similar to our study. This could imply that the E. coli ST131 strain might be widely disseminated throughout Thailand. The IncN2 plasmid has the potential to horizontally transfer the blaNDM gene to other pathogenic bacteria, potential worldwide dissemination of the blaNDM gene. In addition, a previous study demonstrated that the blaNDM-1 gene located in transposon Tn125 was responsible for the blaNDM-1 spread among Acinetobacter spp.7 The structure of the Tn125 prototype includes ISAba125, blaNDM-1, bleMBL, ΔtrpF, dsbC, cutA, ΔgroES, groEL, ISCR27, and ISAba125. In a comparative analysis with the Tn125 prototype, the Tn125-like element in plasmids from this study lacked dsbC, cutA, ΔgroES, groEL, ISCR27, and downstream ISAba125.50 The sequences revealed that the accessory module was located between the fipA and nuc genes. We hypothesize that these two regions (between fipA and nuc genes) probably act as the main integration sites for the insertion of transposable elements within IncN plasmids. Same site insertion for the accessory module has also been reported in other IncN plasmids, pKT58A (GenBank accession no. JX065631) and pRSB206 (GenBank accession no. JN102344).51,52 Thus, this site might act as a significant integration site in IncN2 plasmids. Currently, IncN plasmids have been found all over the world with the highest prevalences in China and the United States.44

Furthermore, many ISs were present in these isolates, which are components of transposons providing a cut-and-paste mechanism.53 As shown in Figure 6, ISEc33 and ISSen4 bracketed the blaNDM-1 on IncN2 plasmid. This structure was the same as the previously reported p271A plasmid, which was isolated from E. coli ST101.54 Notably, we found a class 1 integron in the ST131 isolate, which also provides the ability to drive AMR genes to other bacteria.

To check the pathogenicity of these CREC isolates, the gene encoding siderophore receptor (fyuA), glutamate decarboxylase (gad), high molecular weight protein 2 non-ribosomal peptide synthetase (irp2), tellurium ion resistance protein (terC), and outer membrane protein complement resistance (traT) were identified in all CREC isolates. Most virulence genes can also be horizontally transferred to other bacteria through MGEs, similar to AMR genes.55

In the study of bacterial adaptive immunity system against foreign genetic elements, we found CRISPR-Cas systems in all CREC isolates. These CRISPR-Cas systems contained many spacers, indicating that these isolates may have been previously infected with bacteriophages or invaded by foreign DNA fragments. Bacteriophage genomes were also detected in all CREC isolates. The partial genome of Escherichia phage was identified in the ST131 isolate, while 4 partial genomes of Enterobacteria phage were present in the ST131 and/or ST448 isolates. Notably, we found the genomes of the Klebsiella phage in the ST448 isolates and Pectobacterium phage in the ST131 isolate. This may indicate that these CREC isolates had been priorly infected by bacteriophages with the lysogenic life cycle.56 Also, we hypothesize that these partial phage genomes might carry the AMR genes, as described in a previous study.57 To prove this phenomenon, long-read WGS is needed to elucidated further.

In the investigation of gene encoding bacteriocins, bottromycin was found in all isolates, while sactipeptides and colicin were only detected in the ST131 isolate. Bottromycin has been previously detected in E. coli isolated from the intestine of humans.58 It commonly provides a natural antimicrobial activity, particularly the modified bottromycin A2 which provides activity against methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE).59

In our study of the genomic diversity of our blaNDM-1-harboring CREC isolates compared to other published genomes, we found that the specific features of the ST448 and ST131 isolates were different from various STs. In ST488 isolates, the prophage-encoding proteins may be associated with bacterial adaptation to the bacteriophages,60,61 while non-fimbrial adhesin 1 is probably associated with cell attachment to abiotic surfaces and/or host cells, and the biofilm formation.62 For the ST131 isolate, the FlhC, MntH, AlbA, and CsoR proteins are probably related to the production of flagellar, the transportation of divalent metal cation, the production of antilisterial bacteriocin subtilosin, and the expression of a copper-sensing repressor, which might be beneficial for bacterial evolution and pathogenicity.63–66

Conclusion

The combinations of fosfomycin with aminoglycosides, colistin, tigecycline, and sitafloxacin might be considered as a treatment of infections caused by ST131 carrying the blaNDM-1 gene. Our study of the whole-genome and plasmid sequences of the blaNDM-1-harboring CREC isolates provides significant information, especially the presence of AMR genes, MGEs, virulence genes, genes involving bacterial defense mechanisms, and the structure of blaNDM-1-bearing IncN2 plasmid. These particular isolates contained specific features compared to some published genomes, which are probably associated with bacterial adaptation. The findings can be expected to be useful in the development of appropriate treatments and the control of these infections in the future.