Occurrence, Typing, and Resistance Genes of ESBL/AmpC-Producing Enterobacterales in Fresh Vegetables Purchased in Central Israel

Beta-lactam resistance can lead to increased mortality, higher healthcare expenses, and limited therapeutic options. The primary mechanism of beta-lactam resistance is the production of extended-spectrum beta-lactamases (ESBL) and AmpC beta-lactamases. The spread of beta-lactamase-producing Enterobacterales via the food chain may create a resistance reservoir. The aims of this study were to determine the prevalence of ESBL/AmpC-producing Enterobacterales in vegetables, to examine the association between EBSL/AmpC-producing bacteria and types of vegetables, packaging, and markets, and to investigate the genetic features of ESBL-producing isolates. The antibiotic susceptibilities were determined using VITEK. Phenotypic ESBL/AmpC production was confirmed using disk diffusion. ESBL-producing isolates were subjected to Fourier-transform infrared (FT-IR) spectroscopy and to whole genome sequencing using Oxford Nanopore sequencing technology. Of the 301 vegetable samples, 20 (6.6%) were positive for ESBL producers (16 Klebsiella pneumoniae and 4 Escherichia coli), and 63 (20.9%) were positive for AmpC producers (56 Enterobacter cloacae complex, 4 Enterobacter aerogenes/cancerogenus, and 3 Pantoea spp., Aeromonas hydrophila, and Citrobacter braakii). The blaCTX-M and blaSHV genes were most common among ESBL-producing isolates. The beta-lactamase genes of the ESBL producers were mainly carried on plasmids. Multilocus sequence typing and FT-IR typing revealed high diversity among the ESBL producers. AmpC producers were significantly more common in leafy greens and ESBL producers were significantly less common in climbing vegetables. The presence of ESBL/AmpC-producing Enterobacterales in raw vegetables may contribute to the dissemination of resistance genes in the community.


Introduction
Bacteria carrying gene coding for extended-spectrum beta-lactamases (ESBL) and AmpC beta-lactamases exhibit hydrolytic activity in response to numerous beta-lactam antibiotics, resulting in ineffective antibiotic treatment.Beta-lactam antibiotics are the mainstay of treatment against many infections caused by Gram-negative bacteria [1].Therefore, the spread of ESBL/AmpC-producing Gram-negative bacteria poses a growing concern worldwide and is of clinical and epidemiological importance [2].The case fatality rate for ESBL-producing Enterobacterales infections ranges between 12% and 41% [3], and the in-hospital costs of a single ESBL infection reach up to US$30,093 [4].Similarly, higher Antibiotics 2023, 12, 1528 2 of 13 mortality and increased hospital costs were observed for patients infected with AmpCproducing bacteria compared with susceptible isolates of the same organisms [5].
The spread of antimicrobial resistance occurs through the dissemination of clones or via horizontal gene transfer.The most prevalent and clinically relevant ESBL enzymes are TEM (Temoneira, named after the patient infected with the first isolate expressing TEM-1), SHV (sulfhydryl reagent variable), and CTX-M (cefotaxime-hydrolyzing betalactamase isolated in Munich), with CTX-M enzymes emerging as the most common [6].While TEM-and SHV-type ESBLs are highly related, differing by only a few amino acid substitutions, CTX-M-type ESBLs are more genetically diverse [6].Most ESBL-encoding genes are located on plasmids.The resistance-conferring plasmids of incompatibility (Inc) groups IncF, IncI1, and IncC are associated with ESBL-producing Enterobacterales.IncF plasmids are the predominant group, and can be further categorized into IncFII, IncFIA, and IncFIB [7,8].AmpC beta-lactamases are clinically important cephalosporinase class C enzymes, which are commonly chromosomally encoded in certain species, but may also be plasmid-mediated [9].
ESBL-producing Enterobacterales were originally considered nosocomial pathogens and the main reservoir of carriers was among patients hospitalized in acute care and post-acute care facilities.In the last two decades, infections with these resistant organisms have been increasingly reported in outpatient settings [10].The appearance of resistant Enterobacterales in the community could be the result of spread from healthcare institutions into the community, or could stem from an environmental source (e.g., domestic animals or contaminated food) [10].Raw vegetables may constitute a source of AmpC/ESBL-producing Enterobacterales.Possible contamination pathways include contaminated biological fertilizer or water, poor hygiene practices of the workers involved in food production, ineffective food safety management, or unsanitary vegetable packaging or display [11,12].
In this study, we aimed to determine the prevalence of ESBL/AmpC-producing Enterobacterales in fresh vegetables in Israel and to examine the association between EBSL/AmpCproducing bacteria and types of vegetables, packaging, and markets.In addition, we aimed to analyze the genetic features of ESBL-producing isolates in order to investigate their resistance genes and plasmid carriage status.
Whole genome sequencing (WGS) revealed that all ESBL-producing isolates harbored at least one ESBL variant (blaTEM, blaSHV, and blaCTX-M) in different combinations.The blaCTX-M, blaSHV, and blaTEM genes were detected in 95%, 75%, and 50% of the ESBL isolates, respectively.The predominant allele of blaCTX-M was blaCTX-M-15.Six (30%) ESBL-positive K. pneumoniae isolates exhibited co-occurrence of all three families (blaTEM+ blaSHV+ blaCTX-M).Two ESBL isolates carried blaCTX-M without an additional bla gene, whereas co-occurrence of blaTEM+ blaSHV, blaCTX-M+ blaTEM, and CTX-M+ blaSHV was found in one, three, and eight ESBL-producing isolates, respectively (Table S2).All ESBL-producing isolates carried genes conferring resistance to aminoglycosides and fluoroquinolones, excluding one ESBL-producing E. coli isolate which did not carry any aminoglycoside-resistant genes.Three K. pneumoniae ESBL-producing isolates harbored the aac(6 )-Ib-cr gene, which confers resistance to both aminoglycosides and fluoroquinolones.All ESBL-producing K. pneumoniae isolates carried fosA, a resistance determinant to fosfomycin, whereas fosA was not identified in any E. coli ESBL-producing isolates.Genes conferring resistance to disinfectants and quaternary ammonium compounds were identified in 50% of the ESBL-producing isolates (Table S2).
Of the 14 genes conferring resistance to beta-lactams, 10 genes were associated with phenotypic resistance to all the beta-lactams tested, excluding ceftazidime, and to amikacin.The blaSHV-11 gene was only associated with phenotypic resistance to ceftazidime.The two blaCTX-M-types were associated with phenotypic resistance to all the beta-lactams tested and to amikacin.Of the nine genes linked to aminoglycoside resistance, seven were associated with phenotypic resistance to amikacin, of which two genes were also associated with phenotypic gentamycin resistance.An association was found between resistance to trimethoprim/sulfamethoxazole and the presence of sul2 (Figure 3).The distribution of ESBL genes carried on each plasmid type among all ESBL-producing isolates is presented in Figure 4.All isolates, excluding three, carried plasmids containing ESBL genes; the gene most frequently carried by plasmids was blaCTX-M, followed by blaTEM.The IncFIB plasmid replicon group was identified in 75% of the ESBL-producing isolates.Two plasmid types, IncFIB(K) and IncFIB(K)(pCAV1099-114), were most common in the ESBL sample.Plasmid type IncFIB(K) appeared in nine isolates (45%), which seven isolates carried the combination of blaCTX-M+blaTEM-1B genes on the plasmid and two carried only blaCTX-M-15.Plasmid type IncFIB(K)(pCAV1099-114) was identified in six isolates (30%) and harbored blaCTX-M.An overview of the plasmid types most frequently found, IncFIB(K) and IncFIB(K)(pCAV1099-114), is displayed in Figures 5A and  5B, respectively.IncFIB(K) was associated with a variety of resistance gene cassette arrays among the nine isolates carrying IncFIB(K) (Figure 5C), whereas IncFIB(K)(pCAV1099-114) was more homogeneous among the six isolates carrying the plasmid (Figure 5D).was identified in six isolates (30%) and harbored blaCTX-M.An overview of the plasmid types most frequently found, IncFIB(K) and IncFIB(K)(pCAV1099-114), is displayed in Figure 5A and Figure 5B, respectively.IncFIB(K) was associated with a variety of resistance gene cassette arrays among the nine isolates carrying IncFIB(K) (Figure 5C), whereas IncFIB(K)(pCAV1099-114) was more homogeneous among the six isolates carrying the plasmid (Figure 5D).was identified in six isolates (30%) and harbored blaCTX-M.An overview of the plasmid types most frequently found, IncFIB(K) and IncFIB(K)(pCAV1099-114), is displayed in Figure 5A and Figure 5B, respectively.IncFIB(K) was associated with a variety of resistance gene cassette arrays among the nine isolates carrying IncFIB(K) (Figure 5C), whereas IncFIB(K)(pCAV1099-114) was more homogeneous among the six isolates carrying the plasmid (Figure 5D).The molecular epidemiology of the ESBL isolates was very heterogeneous.Of the 16 ESBL-producing K. pneumoniae isolates, multilocus sequence typing (MLST) identified 12 distinct sequence types (ST).One K. pneumoniae isolate belonged to an emerging epidemic clone ST ST15, two belonged to ST17, and two isolates belonged to ST25.The remaining K. pneumonia isolates each belonged to a different ST (ST2074, ST37, ST1887, ST29, ST405, ST985, ST1799, ST551, ST3057, and two undetermined STs).Fourier-transform infrared (FT-IR) spectroscopy typing of K. pneumoniae did not reveal any clusters, except for one cluster comprising two ESBL-producing K. pneumoniae isolates with a similar genetic and resistance profile but different STs (ST17 and ST25), both isolated from packaged sprouts collected on different days from different market types (Figure 6A).Likewise, MLST revealed high diversity among the four ESBL-producing E. coli isolates (ST535, ST999, ST21, and one undetermined ST).The FT-IR typing of E. coli did not yield any clusters, confirming the diversity among the E. coli strains (Figure 6B).The molecular epidemiology of the ESBL isolates was very heterogeneous.Of the 16 ESBL-producing K. pneumoniae isolates, multilocus sequence typing (MLST) identified 12 distinct sequence types (ST).One K. pneumoniae isolate belonged to an emerging epidemic clone ST named ST15, two belonged to ST17, and two isolates belonged to ST25.The remaining K. pneumonia isolates each belonged to a different ST (ST2074, ST37, ST1887, ST29, ST405, ST985, ST1799, ST551, ST3057, and two undetermined STs).Fourier-transform infrared (FT-IR) spectroscopy typing of K. pneumoniae did not reveal any clusters, except for one cluster comprising two ESBL-producing K. pneumoniae isolates with a similar genetic and resistance profile but different STs (ST17 and ST25), both isolated from packaged sprouts collected on different days from different market types (Figure 6A).Likewise, MLST revealed high diversity among the four ESBL-producing E. coli isolates (ST535, ST999, ST21, and one undetermined ST).The FT-IR typing of E. coli did not yield any clusters, confirming the diversity among the E. coli strains (Figure 6B).

Discussion
The vegetable industry in Israel includes a wide variety of crops, with a production volume of 1.2 million metric tons in 2021 [22].While studies of ESBL/AmpC-producing bacteria present in vegetables have been conducted in Europe, Africa, Asia and America [23], this is the first such study conducted in Israel.
Currently, nearly 90% of Israel's treated wastewater is reused for irrigation purposes [24].Furthermore, Israel has a high rate of 3GCR (>20%) in community settings and ESBL carriers in the community are common [13].Nevertheless, the prevalence of ESBL in vegetables in Israel obtained from our study is low.Other countries have also demonstrated relatively high 3GCR rates in community settings but a relatively low prevalence of ESBL/AmpC in vegetables.For example, countries with high 3GCR, such as South Korea, Romania, Brazil, and Nepal [25], exhibited low (<10%) [20,26] or medium (10-20%) [14,27] ESBL/AmpC prevalence in vegetables.Similarly, countries with a medium 3GCR rate, such as the United States and United Kingdom [25], have low ESBL/AmpC prevalence in vegetables [15,28].In contrast, a correlation between 3GCR prevalence in human isolates and prevalence of ESBL/AmpC in vegetable samples was found in countries such as the Netherlands (low 3GCR [25] and low ESBL/AmpC prevalence [16,17,29]), India (high 3GCR [25] and high ESBL prevalence (>20%) [18]), and Spain (medium 3GCR [25] and medium ESBL prevalence [19]).
The ESBL prevalence in our study was 6.6%, whereas other studies reported a wide range of prevalence between 0.06 and 25.4% [14][15][16][17][18][19].The ESBL-positive E. coli strains comprised 1.3% of our sample, while previous studies found an ESBL E. coli prevalence of 0-13.8% [26][27][28].The percentage of AmpC-producing isolates in our study was 20.9%, and in contrast, previous studies reported a range of 0.47-14.3%[15,17,20,21].We demonstrated that the prevalence of ESBL/AmpC-producing isolates was associated with the types of vegetables, packaging, and markets; therefore, differences in results between these various studies are most likely explained by these factors.This supports the proposition by van Hoek et al. that the wide variety in prevalence of ESBL-producing isolates reported in different studies was due to differences in vegetable types [17].Moreover, sample size, testing method, hygiene awareness in each country, and local beta-lactamase prevalence may impact the results as well.
K. pneumoniae was the predominant ESBL-producing bacterium in our study, followed by E. coli.This observation may suggest that ESBL-producing Klebsiella spp.are more capable of spreading and persisting in the environment than ESBL-producing E coli, as proposed previously [30].Indeed, the prevalence of ESBL-producing Klebsiella spp. in community-acquired infections has been shown to be steadily increasing [31].
ESBL was scarcely found in climbing vegetable plants, likely because they grow above ground and therefore are less exposed to soil and contaminated fertilizer or wastewater, thus reducing the possibility of contamination [12].This finding supports previous work that revealed negligible microbial risk in cucumbers and tomatoes [32].In our study, the majority of AmpC isolates came from leafy green vegetables, in agreement with the World Health Organization's statement that leafy vegetables were the greatest concern in terms of microbiological hazards in vegetables [33].
Most of the ESBL-producing bacteria in our study exhibited resistance to fluoroquinolones.The proportion of ESBL-producing isolates' resistance to fluoroquinolones has increased over time [34].In studies conducted 20 years ago, 45% of ESBL-producing isolates were resistant to fluoroquinolones [35], compared to 70% in our study.Worldwide, the majority of ESBL-producing isolates are susceptible to carbapenems; thus, carbapenems are the mainstay of treatment against infections caused by ESBL-producing pathogens [36].All ESBL-producing isolates in our study were susceptible to carbapenems.They were also all susceptible to amikacin, as previously reported among hospitalized and community patients [37].
In our study, all but one E. coli ESBL-positive isolate produced CTX-M.Previous studies have also found that CTX-M was the most predominant type of ESBL among bacteria found in vegetables [18,23,29].CTX-M-15 was the prevalent allele, constituting 80% of our sample, consistent with previous studies [36,38].
The dissemination of ESBL genes can be attributed to highly transmissible plasmids.In our ESBL sample, 85% of the isolates originating from fresh vegetables carried a plasmid coding for beta-lactam resistance, indicating a threat of resistance dissemination by horizontal gene transmission via plasmids.Among our ESBL-producing isolates, the IncFIB plasmid was the most predominant replicon group, in agreement with previous reports on vegetable samples [14,39].Moreover, the IncFIB plasmid group is prevalent in aquatic environments [40] and in clinical settings [41], indicating the vast transmission and dissemination of the IncFIB plasmid.The high prevalence of the IncF plasmid may be attributed to the presence of addiction systems, i.e., systems which increase plasmid stability by ensuring that only cells containing the plasmid survive, whereas those that have lost the plasmid are eliminated.Therefore, the IncF plasmid serves as an efficient genetic platform to spread ESBL resistance [8].
While in clinical settings it is more common for isolates to be clonally related [36], in the community, isolates are varied, as shown by the wide genetic diversity among the ESBL-producing isolates in our study.The broad assortment of resistance genes found in all the ESBL-positive isolates, specifically resistance to clinically relevant antimicrobials, is a concerning finding.One ESBL-producing K. pneumoniae belonged to a high-risk epidemic clone associated with nosocomial infections in humans (ST15) [42], suggesting dispersal of ESBL genes via raw vegetables.
Beta-lactam resistance poses a challenge for therapeutic treatment, and results in prolonged hospital stays and higher mortality rates [43].Food screening for ESBL/AmpCproducing bacteria is essential to protect health.Consuming raw contaminated vegetables may result in the transfer of resistance genes located on mobile elements, such as ESBL genes, to opportunistic intestinal pathogens [44][45][46].Furthermore, extensive exposure to AmpC-producing bacteria may result in colonization [1].The US Food and Drug Administration has issued a guidance document for the produce industry that outlines principles and practices to minimize microbial food safety hazards during fresh fruit and vegetable production, but the guidance is a mere recommendation and not a regulation that can be enforced [47].
Our study has several limitations.First, while vegetables from a variety of markets were investigated, we did not include vegetables collected directly from farms or from produce factories.Future studies should consider testing those sites in order to provide information on critical points in the farm-to-consumer continuum.Second, we tested for the presence of ESBL/AmpC in fresh vegetables, but did not quantify the bacterial load (CFU) found in the vegetable samples.Third, it was beyond the scope of this study to examine the extent of the public health risks from consumption of ESBL/AmpC via fresh vegetables.

Study Design and Sample
In July and August 2021, 301 vegetable samples were purchased from 70 unique markets in the central region of Israel, all within a 60 km radius.The markets included supermarket chains, i.e., large grocery stores which offer a wide variety of products (n = 17); local minimarkets, i.e., small grocery stores which sell a limited selection of basic food products (n = 20); and outdoor markets, i.e., open-air markets with individual stands (n = 33 different stands from three outdoor markets).From each market, we collected different types of vegetables, based on availability.The vegetable types included climbing vegetables, i.e., vegetables grown vertically by supports, such as trellises (tomatoes and cucumbers); leafy greens, i.e., plant leaves eaten as food (parsley, cilantro, and lettuce); and vegetables grown in environment-controlled beds, i.e., vegetables grown on a prepared bed with controlled temperature, soil mixture type, and light conditions (sprouts and mushrooms).Of the sample, 132 vegetables were packaged in sealed plastic bags and 169 were unpackaged.

Specimen Processing
After purchase, the samples were transported to the laboratory in coolers.All samples were immediately stored in plastic bags at 4 • C and processed within three hours of their acquisition.A total of 25-40 g of each unwashed vegetable sample from a single market was sliced and inoculated in 30 mL Brain Heart Infusion broth (Hylabs, Rehovot, Israel).Following overnight incubation under aerobic conditions at 35 ± 2 • C, the broth was subcultured on a selective screening agar plate, CHROMagar ESBL (Hylabs, Rehovot, Israel).Identification of the species level and antibiotic susceptibilities were determined with VITEK ® MS or VITEK ® 2, using the GN card for identification and the AST-N395 card for susceptibilities (bioMe'rieux SA, Marcy l'Etoile, France).The antibiotics tested were ampicillin, cefazolin, cefuroxime, ceftriaxone, ceftazidime, cefoxitin, ertapenem, meropenem, amikacin, gentamicin, ciprofloxacin, and trimethoprim/sulfamethoxazole.

FT-IR Typing
FT-IR is a typing method which generates a spectrum based on the absorption of infrared light of the bacterial composition.These spectra reflect the specific signatures of specimens and can be grouped into a cluster according to spectra similarity [49].FT-IR was performed as described previously [49].Briefly, isolates were grown at 35 ± 2 • C on blood agar plates (HyLabs, Rehovot, Israel) and samples were prepared in quadruplicate according to the IR Biotyper (Bruker, Leipzig, Germany) manufacturer's instructions.The spectra and cluster analysis were generated by the OPUS software (Bruker, version 7.5).The cut-off ranges recommended by the manufacturer for E. coli and K. pneumoniae were 0.15-0.18and 0.2-0.25,respectively.The ESBL-producing isolates were subjected to FT-IR typing with the addition of two unrelated known ESBL-producing E. coli and K. pneumoniae from our collection, serving as controls.

WGS and Bioinformatics Analysis
The ESBL-positive isolates were subjected to WGS.DNA extraction was performed on the MICROLAB Nimbus workstation (Hamilton, Reno, NV, USA) with a STARMag 96 Universal Kit (Seegene, Seoul, Republic of Korea).Samples were sequenced with Oxford Nanopore at SNPsaurus, Colorado, Oregon.Resistance genes were detected using ResFinder (https://cge.cbs.dtu.dk/services/ResFinder/---version25.4.23,accessed on 7 August 2023).Pangenome analysis of all genomes was performed with version 3.12.0 of Roary.The resulting core genome alignment was then used to generate a phylogenetic maximum likelihood tree using version 8.2.12 of RAxML with the GTRGAMMA model.Genes were classified as plasmid-related if they were located on the same contig as the plasmid rep gene, whereas genes that were not located on the plasmid contig were classified as neither plasmid nor chromosomal.A polymerase chain reaction (PCR) of blaCTX-M, blaTEM, and blaSHV was performed for two ESBL-negative isolates serving as negative controls (an AmpC-positive isolate and an ESBL/AmpC-negative Acinetobacter spp.isolate), using specific primers [50][51][52] (Table 2).

Statistical Analysis
Odds ratios and their 95% confidence intervals were calculated to compare the prevalence of AmpC-or ESBL-producing bacteria between vegetable types, packaging, and market types.The association between phenotype and genotype was assessed using a chi-square test.All the analyses were performed in Stata version 14.2 (StataCorp, College Station, TX, USA).

Conclusions
In our study, we detected ESBL/AmpC-producing bacteria in fresh vegetables.Further analysis revealed that most of the ESBL genes were carried on plasmids.The presence of ESBL/AmpC resistance in bacteria from vegetables highlights the need for food screening and monitoring of antimicrobial resistance in fresh produce, including plasmid identification, to increase food safety and limit the dispersal of resistance genes.The plasmid-borne ESBL and AmpC in bacterial populations in fresh vegetables constitute a reservoir of antibiotic resistance genes and may contribute to global resistance dissemination in the community.Further studies are needed to evaluate the impact of the ESBL/AmpC-producing bacteria found in raw vegetables on human health.

Figure 1 .
Figure 1.Prevalence of ESBL-and AmpC-producing Enterobacterales and distribution according to bacteria.

Figure 4 .
Figure 4. Distribution of ESBL genes per plasmid types among the ESBL-producing isolates.

Figure 5 .
Figure 5. Plasmid types and rooted distance trees of ESBL-positive isolates.(A,B) Backbone structure of a representative plasmid sequence of (A) the IncFIB(K) plasmid type and (B) the

Figure 4 .
Figure 4. Distribution of ESBL genes per plasmid types among the ESBL-producing isolates.

Figure 4 .
Figure 4. Distribution of ESBL genes per plasmid types among the ESBL-producing isolates.

Figure 5 .Figure 5 .
Figure 5. Plasmid types and rooted distance trees of ESBL-positive isolates.(A,B) Backbone structure of a representative plasmid sequence of (A) the IncFIB(K) plasmid type and (B) theFigure 5. Plasmid types and rooted distance trees of ESBL-positive isolates.(A,B) Backbone structure of a representative plasmid sequence of (A) the IncFIB(K) plasmid type and (B) the IncFIB(K)(pCAV1099-114) plasmid type.On plasmid type IncFIB(K), the CTX-M gene refers to either CTX-M-14 or CTX-M-15, depending on the isolate.Blue, CDS; green, rep; red, CARD (Comprehensive Antibiotic Resistance Database); black, GC content.(C,D) Rooted distance tree of ESBL-positive isolates carrying the (C) IncFIB(K) plasmid type and (D) IncFIB(K)(pCAV1099-114) plasmid type.

Figure 6 .
Figure 6.Genotypic and phenotypic characteristics of ESBL-producing isolates.(A) K. pneumoniae isolates and (B) E. coli isolates.The dendrogram on the right displays the relative distance of the isolates according to Fourier-transform infrared (FT-IR) typing.

Figure 6 .
Figure 6.Genotypic and phenotypic characteristics of ESBL-producing isolates.(A) K. pneumoniae isolates and (B) E. coli isolates.The dendrogram on the right displays the relative distance of the isolates according to Fourier-transform infrared (FT-IR) typing.

Table 2 .
List of primers used in this study.