Decoding Klebsiella pneumoniae in poultry chain: unveiling genetic landscape, antibiotic resistance, and biocide tolerance in non-clinical reservoirs

The rise of antibiotic resistance in the food chain is influenced by the use of antimicrobial agents, such as antibiotics, metals, and biocides, throughout the entire farm-to-fork continuum. Besides, non-clinical reservoirs potentially contribute to the transmission of critical pathogens such as multidrug-resistant (MDR) Klebsiella pneumoniae. However, limited knowledge exists about the population structure and genomic diversity of K. pneumoniae circulating in conventional poultry production. We conducted a comprehensive characterization of K. pneumoniae across the whole chicken production chain (7 farms; 14 flocks + environment + meat, 56 samples; 2019–2022), exploring factors beyond antibiotics, like copper and quaternary ammonium compounds (QACs). Clonal diversity and adaptive features of K. pneumoniae were characterized through cultural, molecular (FT-IR), and whole-genome-sequencing (WGS) approaches. All except one flock were positive for K. pneumoniae with a significant increase (p < 0.05) from early (n = 1/14) to pre-slaughter (n = 11/14) stages, most (n = 6/7) persisting in chicken meat batches. Colistin-resistant K. pneumoniae rates were low (4%-n = 1/24 positive samples), while most samples carried MDR strains (67%-n = 16/24) and copper-tolerant isolates (63%-n = 15/24, with sil and pco gene clusters; MICCuSO4 ≥ 16 mM), particularly at pre-slaughter. Benzalkonium chloride consistently exhibited activity against K. pneumoniae (MIC/MBC range = 4–64 mg/L) from representative strains independently of the presence or absence of genes linked to QACs tolerance. A polyclonal K. pneumoniae population, discriminated by FT-IR and WGS, included various lineages dispersed throughout the chicken’s lifecycle at the farm (ST29-KL124, ST11-KL106, ST15-KL19, ST1228-KL38), until the meat (ST1-KL19, ST11-KL111, ST6405-KL109, and ST6406-CG147-KL111), or over years (ST631-49 KL109, ST6651-KL107, ST6406-CG147-KL111). Notably, some lineages were identical to those from human clinical isolates. WGS also revealed F-type multireplicon plasmids carrying sil + pco (copper) co-located with qacEΔ1 ± qacF (QACs) and antibiotic resistance genes like those disseminated in humans. In conclusion, chicken farms and their derived meat are significant reservoirs for diverse K. pneumoniae clones enriched in antibiotic resistance and metal tolerance genes, some exhibiting genetic similarities with human clinical strains. Further research is imperative to unravel the factors influencing K. pneumoniae persistence and dissemination within poultry production, contributing to improved food safety risk management. This study underscores the significance of understanding the interplay between antimicrobial control strategies and non-clinical sources to effectively address the spread of antimicrobial resistance.


Introduction
Intensive poultry production is a crucial sector of the global food industry.It faces significant challenges in addressing the growing demand for poultry products, especially antibiotic-free chicken (Mottet and Tempio, 2017;Karcher and Mench, 2018).Ensuring biosecurity, which includes proper hygiene practices, vaccination programs, regular monitoring of flock health and effective farm management strategies, is essential to prevent disease transmission between flocks and farms.However, despite these efforts, intensive chicken production relies heavily on antibiotics and coccidiostats (Karcher and Mench, 2018;Gržinić et al., 2023).This reliance becomes even more concerning when considering that poultry presents potential risks to human health as it can be a reservoir of zoonotic pathogens causing infectious diseases and can contribute to the spread of antimicrobial resistance (AMR) within the food chain [Golden et al., 2021;European Food Safety Authority (EFSA) and European Centre for Disease Prevention and Control (ECDC), 2023].While antibiotics have traditionally been seen as the main drivers of AMR, recent restrictions on their use in food-producing animals, as well as alternative antimicrobial approaches, suggest the contribution of other compounds as selectors of antibiotic-resistant bacteria (Rebelo et al., 2023).
Metals and biocides are used in food-animal production for various purposes, including as feed additives, growth promoters, antiseptics, and disinfectants, to decrease the dependence on antibiotics [EMA Committee for Medicinal Products for Veterinary Use (CVMP) and EFSA Panel on Biological Hazards (BIOHAZ) et al., 2017;Rebelo et al., 2023].Copper-Cu is one of the metals commonly added to chicken feed.Its antimicrobial properties not only improve animal nutrition and productivity (e.g., by modulating gut microbiota) but also reduce disease risks, thereby boosting the overall flock health (Broom et al., 2021;El Sabry et al., 2021).Furthermore, biocides formulated with quaternary ammonium compounds (QACs) are used to disinfect surfaces, equipment, feeding systems, and water sources.Their versatility and broad-spectrum antimicrobial activity assist in controlling and preventing pathogens dissemination between flocks within poultry facilities (Chen et al., 2023).Genes associated with tolerance to metals and QACs often share the same genetic contexts with antibiotic-resistance genes (Slifierz et al., 2015;Li et al., 2022;Pereira et al., 2023).Thus, the use of copper-supplemented feed and QAC-based biocides could contribute to co-selection effects (Webber et al., 2015;Kampf, 2018;Rebelo et al., 2023).
Current intensive chicken production involves large-scale operations that span breeding and hatching to rearing, processing, and distribution.However, the impact of diverse antimicrobial strategies used throughout farm to fork on the spread of AMR remains underexplored [EMA Committee for Medicinal Products for Veterinary Use (CVMP) and EFSA Panel on Biological Hazards (BIOHAZ) et al., 2017;Rebelo et al., 2023].To effectively address AMR, a One Health approach -which emphasizes coordinated efforts across the domains of animals, humans, and the environment -is essential.This approach is not only crucial for elucidating the origins of less common foodborne pathogens such as K. pneumoniae but also for understanding emergent reservoirs and vectors of AMR genes outside the clinical setting (Wyres and Holt, 2018).Our prior study unveiled a significant occurrence of copper tolerance and multidrug resistance among K. pneumoniae strains found in chicken flocks (Mourão et al., 2023).We also identified K. pneumoniae lineages and plasmids carrying sil + pco copper tolerance and variable antibiotic resistance genes, resembling those identified in human clinical isolates worldwide (Mourão et al., 2023).These findings highlight the urgent need to further investigate the factors, drivers, and sources that contribute to the selection and persistence of multidrug resistant (MDR) K. pneumoniae within the poultry production chain.However, there has been limited research focusing on the early stages of chicken rearing (e.g., one-day-old chicks) and the in-house poultry environment, including cleaned poultry houses post-vacancy, where broilers are raised for 30-35 days before being sent to slaughter for meat production (Daehre et al., 2018;Zhai et al., 2020).
This study aims to provide a comprehensive analysis of the occurrence, diversity, and persistence of K. pneumoniae in the whole poultry production chain (from one-day-old chicks to chicken meat) between 2019 and 2022.Furthermore, we assessed the contribution of factors other than antibiotics (use of copper and QACs) as putative selective agents of AMR genes and bacteria that are clinically relevant.Identifying the transmission sources and pathways for MDR K. pneumoniae will pave the way for devising effective strategies to mitigate its dissemination, ensuring animal welfare, environmental sustainability, public health, and overall improved food safety.

Sampling design at the chicken farm and slaughterhouse processing plant
Our sampling included seven Portuguese intensive-based chicken farms with conventional indoor and floor-raised production systems in compliance with EU legislation, as indicated by the operator (ADAS, 2016).Six similar farms (arbitrarily designated as A, B, C, E, G, and H) previously studied (Mourão et al., 2023) and one (I) recently restructured with modern grow-out poultry house facilities were selected.In all farms, colistin was banned since January 2018, while copper was routinely used as an additive in inorganic formulation feed (far below the maximum dose of 25 mg/kg of Cu, according to EU Regulation 2018/1039).The vacancy period varied from 11 to >15 days, being each poultry house depopulated, cleaned, and disinfected using similar routine cleaning and disinfection procedures.The biocide active compounds used were benzalkonium chloride (BZC) or didecyl dimethyl ammonium chloride (DDAC) and hydrogen peroxide for water distribution systems.
In each of the seven farms, female, and male mixed one-day-old chicks (Ross 308 strain) were randomly distributed by two poultry houses at arrival (around 5,000-60,000 chickens per flock/house).A total of 14 flocks were followed and sampling (n = 56 samples) was carried out over three different periods during the production cycle (the length ranged from 26 to 43 days) between February and May 2022 (Figure 1).The initial sampling period included collecting samples from one-day-old chicken transport boxes (P0; n = 14 samples; each sample comprising 12 boxes).These were obtained using stick swabs, with each swab tube containing 10 mL of Buffered Peptone Water-BPW and used in 4 chicken boxes (a total of three swabs per sample).Additionally, the inside of cleaned poultry houses (P1; n = 14 samples) were sampled using the boot swab technique, which employs dry-foot swabs in a zigzag pattern, according to EU Regulation 200/2012.In the second period, the inside of grow-out houses containing the same flocks the day before slaughter (P2; n = 14 samples; after >25 days) was collected using the boot swab technique.In the third period, raw chicken meat samples (P3; n = 14 batches) from the same flocks were collected after slaughter and air chilling at the poultry production slaughterhouses, immediately before distribution for retail sale.Each sample (50 g) was processed as a pool of neck skin from 10 to 15 carcasses of the same batch (each batch corresponded to one flock from the same farm and poultry house slaughtered at the same time; according to EU Regulation 1086/2011).
All previous samples were collected in sterile plastic bags/ containers, transported at 4°C to the laboratory, and processed on the same day.Subsequent sample processing was performed using cultural/molecular approaches as described in the following sections.

Screening of Klebsiella pneumoniae
Klebsiella pneumoniae was recovered from Simmons citrate agar plates with 1% inositol (SCAi), the most suitable selective medium for Klebsiella recovery, directly from the suspended sample and after enrichment.A common initial step consisted of mixing swabs (three pooled swab tubes-P0 and two swab foot-P1 and P2) or weighing 25 g of pooled meat samples (P3) into 1/10 mL of BPW and BPW supplemented with 3.5 mg/L colistin.The direct culture method included spreading an aliquot of 100 μL of BPW and BPW + colistin after 1 h at room temperature (resuscitation step) on SCAi supplemented or not with colistin (3.5 mg/L).The enrichment approach involved the same procedure but after a previous incubation of BPW and BPW + colistin at 37°C for 16-18 h.All the SCAi plates were incubated at 37°C for 48 h.One to five colonies of each presumptive morphotype were selected for identification.Isolates were identified presumptively by CHROMagar™ Orientation and then by PCR for K. pneumoniae (Bialek-Davenet et al., 2014).

Klebsiella pneumoniae diversity between samples
We used Fourier Transform Infrared (FT-IR) spectroscopy with attenuated total reflectance (ATR) to infer isolates' relatedness between isolates identified in the same or different samples.After growth under standardized culture conditions (Mueller-Hinton agar; 37°C/18 h), a colony was directly deposited and air-dried on the ATR accessory of the FT-IR instrument.Spectra were acquired in a Spectrum Two instrument (Perkin-Elmer, United States) under standard conditions (4,000-600 cm −1 , 4 cm −1 resolution, and 16 scan co-additions).The region between 1,200 and 900 cm −1 was compared Sampling strategy at the chicken farm and slaughterhouse processing plant.Sample collection points are indicated by P0, P1, P2, and P3.Mourão et al. 10.3389/fmicb.2024.1365011Frontiers in Microbiology 04 frontiersin.org with each other, and with those from two different machine-learning classification models, used to predict K. pneumoniae capsular (KL) types: (i) a Random Forest classification model that enables identification of up to 33 KL-types from well-characterized international K. pneumoniae clones from the clinical setting (Novais Â. et al., 2023); and (ii) a Random Forest classification model to allow identification of up to 21 KL-types (11 in common with the previous model) based on a spectral database of poultry isolates identified in previous studies (Mourão et al., 2023).Isolate relatedness and prediction of KL-types were inferred as described previously (Novais Â. et al., 2023).Isolates predicted to have the same KL-type were considered putatively related.FT-IR-based assignments were confirmed by PCR of the wzi gene and further sequencing at Eurofins Genomics1 to infer K-type using BIGSdb2 (Brisse et al., 2013).

Antimicrobial susceptibility to copper
Copper-Cu susceptibility in representative isolates (different farms, flocks, and genomic backgrounds) was evaluated using the agar dilution method under anaerobic conditions (Mourão et al., 2016).Briefly, the MIC was determined using Mueller-Hinton 2 agar freshly prepared and supplemented with different copper sulfate (CuSO 4 ) concentrations (0.5 to 36 mM) and a final adjustment to pH = 7.2.The plates were then inoculated with 0.001 mL suspension (10 7 CFU/mL) of each isolate and incubated at 37°C under anaerobic conditions for 18 h-20 h.The MIC was identified as the lowest concentration where no visible growth was observed.Control strains included Escherichia coli ED8739 carrying the plasmid pRJ1004 with the sil and pco cluster (MIC CuSO4 = 16-20 mM) and Enterococcus lactis BM4105RF without acquired copper tolerance genes (MIC CuSO4 = 2-4 mM; Novais C. et al., 2023).All K. pneumoniae isolates were screened for the silA copper tolerance gene using a PCR, given its strong association with the presence of an intact sil operon and a Cu tolerance (CuT) phenotype (Mourão et al., 2016(Mourão et al., , 2023)).

Antimicrobial susceptibility to benzalkonium chloride
The MIC and minimum bactericidal concentrations (MBC) of benzalkonium chloride (BZC; CAS 68391-01-5, VWR) were determined for representative sequenced isolates (n = 45; different clones, sample types, farms, years and the presence or absence of QAC tolerance genes) using the Mueller-Hinton broth microdilution method (pH = 7.2; 37°C/20 h; Clinical and Laboratory Standards Institute, 2018).Briefly, a 96-well microtiter plate containing serial two-fold dilutions of BZC (concentration ranging from 0.125 to 128 mg/L) was used to assess the susceptibility of bacterial suspensions in log-phase growth (adjusted to reach a final inoculum of 5 × 10 5 CFU/ mL in each well) at 37°C for 20 h.Microdilution panels were freshly prepared before each assay.The first concentration of BZC without visible growth was considered the MIC (Clinical and Laboratory Standards Institute, 2018).
To determine the MBC BZC , 10 μL of each well without visible growth from the 96-well MIC plate were incubated onto brain heart infusion agar plates at 37°C for 24 h, as defined by the Clinical and Laboratory Standards Institute (1999).The MBC was defined as the lowest QAC concentration where the colony count was equal to or less than the CLSI-specified rejection value, based on the final inoculum count of each well (Clinical and Laboratory Standards Institute, 1999).Each experiment was repeated three times and the MIC/MBC values corresponded to the mean of these determinations.To ensure assay reproducibility, the Enterococcus faecalis ATCC 29212 (without any known QACs tolerance genes) was included as a control strain (MIC BZC and MBC BZC varied between 1 and 4 mg/L; Pereira et al., 2023).

Genomic analysis by whole-genome sequencing
We then aimed to elucidate the sources and persistence of specific clonal lineages and/or MDR plasmids along the chicken production.For that, we compared 68 genomes, including 48 sequenced de novo in the present study (n = 31 from 2022 farms and n = 17 from 2019 to 2020 farms) and 20 obtained in the previous study (Mourão et al., 2023), representing different farms, stages, timespans, and KL-types.Genomic DNA was extracted from the isolates using the Wizard Genomic DNA purification kit (Promega Corporation, Madison, WI) and the final concentration was measured with a Qubit 3.0 Fluorometer (Invitrogen, Thermo Fisher Scientific, United States).Subsequently, the isolates were sequenced using the Illumina NovaSeq 6,000 S4 PE150 XP (Illumina, San Diego, CA, United States) at Eurofins Genomics (see text footnote 1).FastQC v0.11.9 (Andrews, 10.3389/fmicb.2024.1365011Frontiers in Microbiology 05 frontiersin.org2010) and MultiQC v1.13.dev0 (Ewels et al., 2016) with default parameters were used for the quality control of the raw sequence data.
Plasmid replicon typing was performed on all WGS-selected isolates using ABRicate v1.0.1 (see text footnote 5) with PlasmidFinder database (from 18 July 2023) and pMLST v2.0 (Camacho et al., 2009;Carattoli et al., 2014) from the Centre for Genomic and Epidemiology.7 IncFII K plasmids were further characterized, as described by Villa et al. (2010). 8 We used the MOB-recon tool v3.1.0from the MOB-suite package to confirm the location of the metal tolerance genes and reconstruct putative plasmids based on draft assemblies (Robertson and Nash, 2018;Robertson et al., 2020).Metal tolerance genes were considered part of a specific plasmid when identified by MOB-recon or when located on the same contig as the replicon/incompatibility determinant.

Occurrence of Klebsiella pneumoniae by poultry samples
We detected K. pneumoniae in 43% (n = 24/56) of the samples collected throughout all stages of the poultry production chain.The highest occurrence was found in pre-slaughter fecal chicken samples (79%-11/14 flocks; in all but one farm), followed by derived chicken meat (50%-7/14 batches; all farms), and less frequently in one-day-old chicks (7%-1/14 flocks; one farm).Notably, K. pneumoniae was also detected in cleaned poultry houses (36%-5/14 flocks; four farms; Figure 2).Although differences were observed between farms, a significant increase in occurrence was found between one-day-old (1/14 flocks) and pre-slaughter (11/14 flocks) chickens (p < 0.05), and most (n = 6/7 flocks) of these maintained K. pneumoniae carriage in chicken meat at the slaughter stage.Only three out of 11 pre-slaughter flocks carrying K. pneumoniae were raised in cleaned poultry houses that had previously tested positive for these bacteria (Figure 2).

Copper tolerance of Klebsiella pneumoniae recovered from poultry chain
The copper tolerance gene silA was observed in 52% (n = 51/99) of K. pneumoniae isolates from all farms and most poultry samples with K. pneumoniae (63%-n = 15/24), with similar rates between sampling stages (p > 0.05; Figure 4).More than 80% (n = 43/51) of silApositive isolates were MDR (p < 0.05) and were also found to be more resistant to amoxicillin + clavulanic acid, ciprofloxacin, nalidixic acid, tetracycline, sulphonamides, gentamicin, and trimethoprim than the silA negative ones (p < 0.05; Figure 5).Copper susceptibility assays were performed in 53% (n = 52/99) of K. pneumoniae isolates carrying or not copper tolerance genes representative of different farms, flocks, KL-types, and antibiotic resistance profiles (Supplementary Table S1).Phenotypic results were congruent with the genotype since all isolates with MIC CuSO4 ≥ 16 mM carried the silA gene whereas those with MIC CuSO4 < 16 mM did not (Supplementary Table S1).

Benzalkonium chloride tolerance of Klebsiella pneumoniae recovered from poultry chain
The susceptibility to BZC was determined for 45 out of 99 K. pneumoniae isolates (all carrying genes previously associated with QAC tolerance), with diverse epidemiological and clonal backgrounds (Supplementary Table S2).The MIC BZC ranged between 4 and 64 mg/L, with MIC 50 = 16 mg/L and MIC 90 = 32 mg/L.The highest MIC BZC of 64 mg/L [non-wild-type using Epidemiological Cut-off-ECOFF of 32 mg/L proposed by Morrissey et al. (2014)] was observed in the ST280-KL23 lineage from a pre-slaughter chicken sample.Ten QACs' genotypes (19 to 24 genes each) were identified (Figure 6).A variable occurrence of  6).However, no differences were observed between MIC distribution and clones, sources, MDR phenotype, or the presence or absence of genes previously associated with QAC tolerance, including qac genes (Figure 6; Supplementary Table S2).The only exception was the ST6552-KL109 isolate lacking kmrA, coding for an efflux pump of the Major Facilitator Superfamily, and presenting MIC BZC of 4 mg/L.The MBC BZC for all tested isolates was the same as the MIC BZC .Klebsiella pneumoniae showing the highest MBC BZC of 32-64 mg/L comprised isolates from diverse farms, sources, and clones, including one of the most prevalent in the present study (ST11-KL106, farm H).

Discussion
This is a distinctive study that explores K. pneumoniae occurrence and diversity throughout the entire chicken farm-to-fork chain from a long-term perspective.It is also one of the few studies that offer evidence that poultry serves as a reservoir and source of K. pneumoniae strains with clinically relevant features, including genes coding for antibiotic resistance, metals, and/or biocide tolerance, while the identification of clones identical to those in clinical settings further supports K. pneumoniae as a foodborne pathogen.
Our data strongly emphasize that intensively farmed chicken production and their meat are relevant sources of K. pneumoniae and antibiotic-resistant isolates.This remains true even after EU veterinary antimicrobial sales decreased by 46.5% between 2011 and 2021 (European Medicines Agency, 2022).During the studied period (this study and Mourão et al., 2023), resistance remained high to antibiotics commonly used in poultry, such as tetracycline, sulphonamides, and/ or quinolones (Alliance to Save our Antibiotics, 2016; Gržinić et al., 2023).However, resistance to colistin, a critically important antibiotic for treating human infections caused by Gram-negative MDR strains, has significantly decreased from 61% in 2019-2020 samples to 4% in 2022.This reduction suggests that the efforts of the long-term ban on colistin use in food-animal production (>4 years) have yielded promising results not only by limiting mcr (Ribeiro et al., 2021) but also other colistin-resistance genotypes (Mourão et al., 2023).
In conventional chicken production farms, diverse antimicrobial treatments are commonly administered throughout  A neighbor-joining tree representing the phylogenetic relationships among the 68 Klebsiella pneumoniae genomes recovered between 2019 and 2022 was constructed from the Pathogenwatch pairwise-distance matrix [i.e., based on single nucleotide polymorphisms (SNPs) called in 1,972 core genes].Scale bar units represent substitutions per variant site.The SNPs among our isolates from the 15 main clusters (<10 alleles difference) are represented in each branch.The isolates from samples exhibiting gray shading correspond to the ones already published by Mourão et al. (2023).The associated metadata for all isolates was added using iTOL (https://itol.embl.de/).Each colored-filled shape represents the presence of relevant antibiotic resistance, virulence-associated, metal-tolerance genes, and plasmid replicons associated with well-defined incompatibility groups.Only known mutations conferring fluoroquinolone resistance are presented.Klebsiella intrinsic antibiotic resistance (bla SHV-1 , bla SHV-11 , bla SHV-26 , bla SHV-27 , bla SHV-28 , bla SHV-38 , bla SHV-142 , bla SHV-187 , fosA, oxqAB) and metal tolerance (arsBCR, cusABFCRS) genes were not represented.AGLY, aminoglycosides; BLA, β-lactams; CG, Clonal Group; FLQ, fluoroquinolones; KL, K-Locus; MLST, Multilocus Sequence Typing; PHE, phenicols; SUL, sulphonamides; TET, tetracycline; TMT, trimethoprim; Unk, unknown.Neighbor-joining trees representing the phylogenetic relationships among the 68 Klebsiella pneumoniae genomes recovered between 2019 and 2022 and those available in Pathogenwatch with less than 21 SNPs.(A) ST11-KL27, (B) ST11-KL111, (C) ST15-KL19, (D) ST15-KL146, (E) ST392-KL27, (F) ST1537-KL64, (G) ST280-KL23, and (H) ST1997-KL28.The genome selection was performed using the cgMLST single linkage clustering to include the ones with less than 10 allele differences (threshold = 10)*.Then these genomes were used to infer a neighbor-joining tree from the Pathogenwatch pairwise-distance matrix (i.e., based on single nucleotide polymorphisms-SNPs called in 1,972 core genes).Scale bar units represent substitutions per variant site.The number of substitutions between our isolates and the ones available in Pathogenwatch are represented in each branch.All isolates' associated metadata (country, collection date, and source of isolation) was added using iTOL (https://itol.embl.de/).*Klebsiella pneumoniae isolates from clonal lineages with <10 allele differences include: 2019-2020 collection-ST1-KL19, ST11-KL105, ST11-KL106, ST11-KL27, ST11-KL111, ST15-KL19, ST15-KL146, ST147-KL64, ST307-KL102, ST392-KL27, ST1537-KL64, and ST6250-KL155 (1); 2022 collection -ST11-KL106, ST11-KL111, ST15-KL19, ST147-KL64, ST280-KL23, ST307-KL102, ST525-KL45, ST1228-KL38, ST2055-KL14, ST1997-KL28, ST4110-KL3.The isolates exhibiting gray shading correspond to the ones already published by Mourão et al. (2023Mourão et al. ( ). 10.3389/fmicb.2024.1365011 .1365011Frontiers in Microbiology 12 frontiersin.orgthe fattening period, with differences by country and region (Joosten et al., 2019;Kasabova et al., 2021).In this study, all but one flock received antibiotics during the 26-43 days production cycle, which may explain the higher occurrence of MDR K. pneumoniae in pre-slaughter chickens compared to samples from previous stages.Furthermore, the contribution of poultry house management practices like inadequate disinfection or short vacancy periods between flocks, cannot be excluded (Daehre et al., 2018;Zhai et al., 2020;Franklin-Alming et al., 2021;Kaspersen et al., 2023).Even with standard cleaning and disinfection protocols across all the studied farms, a wide variety of K. pneumoniae strains, including MDR and clinically relevant lineages, can persist within the farm environment and/or until the meat becomes available to consumers.We found closely related MDR K. pneumoniae isolates in both cleaned poultry houses and pre-slaughter chickens (ST1228-KL38, ST29-KL24, ST280-KL23) from the same or different flocks or farms.The idea that strains re-introduction might come from parent flocks seems less likely since we observed minimal contamination (1 flock) during the reception of one-day-old chickens.This study also highlights water and feed as sources of MDR K. pneumoniae lineages (ST525-KL10, ST11-KL106), but at lower rates than chicken feces, as observed in Mourão et al. (2023).Noteworthy, some K. pneumoniae lineages (such as ST1-KL19, ST11-KL111, ST6405-KL109, and ST6406-CG147-KL111) persist throughout the chickens' lifecycle until their meat becomes available to consumers.Additionally, cross-contamination at slaughterhouses is also plausible given the identification of K. pneumoniae lineages in chicken meat samples that were not traced back to their originating farm.The higher contamination compared with the previous study (43%-50% vs. 17%) and the diversity of K. pneumoniae strains (Mourão et al., 2023), might be attributed to the cultural methodology employed (antibiotic-free selection using SCAi medium with and without BPW enrichment).This approach was specifically designed to enhance the investigation of the inherent K. pneumoniae populations, as previously reported (Rodrigues et al., 2022).FT-IR proved to be a reliable approach to assessing the clonal relationship between K. pneumoniae isolates from poultry origin, being able to correctly identify closely related isolates (representing 70% of the sample) and KL-types (78% predicted by the machine-learning models) and discard unrelated isolates or KL-types not represented by the models used, as happened for isolates from human origin (Novais Â. et al., 2023).The use of this technique represented a useful tool to quickly identify common isolates from the same or different samples and choose representative isolates to sequence by WGS, reducing the cost and time associated with typing.
The persistence of MDR K. pneumoniae lineages within the poultry chain (this study, Kaspersen et al., 2023;Mourão et al., 2023), suggests the presence of adaptive environmental factors beyond antibiotic resistance.We observed elevated rates of K. pneumoniae carrying sil and pco operons, along with a copper tolerance phenotype (>16 mM) in poultry samples.This suggests that the incorporation of copper-supplemented chicken feed might contribute to the selection of copper-tolerant and MDR strains within such production environments.However, earlier research, whether grounded in wet lab experiments or mathematical models, has indicated that the concentrations of copper necessary to foster the emergence of coppertolerant bacteria might be significantly below their corresponding MIC values (Gullberg et al., 2014;Arya et al., 2021).Furthermore, the coexistence of various pollutants appears to further lower the minimum selective concentration estimates for individual antimicrobials (Gullberg et al., 2014;Arya et al., 2021).The sil and pco operons are often located on plasmids that carry an array of other metal, qac, metabolic, and/or specific antibiotic resistance genes, fostering their persistence within the environment through co-selection events driven by antimicrobial usage, coupled with the frequent presence of mechanisms that prevent plasmid loss.
This study also reveals a high abundance of QACs tolerance genes in K. pneumoniae lineages, although they do not appear to correlate with phenotype, as observed in other studies regardless of testing methods (Abuzaid et al., 2012;Morrissey et al., 2014;Wu et al., 2015;Vijayakumar et al., 2018;Gual-de-Torrella et al., 2022).Nevertheless, our WGS approach identified a wide range of QAC tolerance genes, underscoring the pressing need for establishing reliable genotypicphenotypic correlations to elucidate QAC tolerance mechanisms (Hipólito et al., 2023).While several studies have reported that bacterial strains with an elevated MIC or MBC remained susceptible to the in-use BZC concentration (Vijayakumar et al., 2018;Maillard, 2022), it is essential to comprehend the environmental factors (e.g., temperature, pH) associated with the expression of these genes (Gualde-Torrella et al., 2022;Pereira et al., 2023) to clarify the environmental, clinical, or veterinary/industrial implications of bacteria with a reduced biocide susceptibility.Further studies are needed to investigate the impact of inappropriate biocide usage or low concentrations, which can act as stressors without killing bacterial pathogens, potentially promoting antimicrobial resistance, and facilitating the transfer of antimicrobial resistance genes (Maillard, 2022;Maillard and Pascoe, 2023).

Conclusion
Our study reveals chicken production as a significant reservoir hosting a diverse range of clinically relevant K. pneumoniae clones, including MDR, copper-tolerant and enriched in QAC tolerance genes.The identification of clones identical to those in clinical settings supports K. pneumoniae as a foodborne pathogen.Various sources of contamination (such as feed, water, poultry houses, and crosscontamination) contribute to the persistence of K. pneumoniae throughout the production chain, emphasizing that, despite a decrease 10.3389/fmicb.2024.1365011Frontiers in Microbiology 13 frontiersin.orgin its occurrence, certain clones still reach chicken meat even with implemented safety measures in place.
The co-occurrence of copper and/or QAC tolerance genes on highly prevalent MDR plasmids suggests that these have been circulating in various K. pneumoniae populations and phenotypic validation (at least in the case of copper) supports the possibility that these genes may play a role in the co-selection of these plasmids or strains under certain conditions within the food production chain or other environmental settings.
Further studies are needed to assess the implications of these K. pneumoniae lineages on food safety and the risk of transmitting antibiotic resistance to humans.Additional studies are imperative to elucidate the external factors (such as environmental conditions) that drive K. pneumoniae's adaptation toward antimicrobial resistance.Addressing these complexities can contribute to the development of effective strategies to safeguard animal welfare, enhance food safety, and mitigate public health risks associated with clinically-relevant K. pneumoniae lineages and antibiotic resistance.
susceptibility profile, and CuT based on the presence of the silA gene for further characterization by wzi sequencing and/or whole genome sequencing.

FIGURE 2
FIGURE 2 Occurrence of Klebsiella pneumoniae in poultry samples from the farm (P0, P1, and P2) and chicken meat (P3).*p < 0.05 (Fisher's exact test) when comparing P0 with P2.Farms were arbitrarily designated by capital letters as A, B, C, E, G, H and I.The presence of K. pneumoniae is indicated by a filled rectangle in the table.

FIGURE 3
FIGURE 3Sankey diagram representing, from left to right, the occurrence and diversity of Klebsiella pneumoniae by sample, farm, and KL-type.The width of each connection is proportional to the number of positive hits.The bold black line represents KL-types where 100% of the isolates were correctly identified by FTIR (KL3, KL14, KL19, KL21, KL23, KL38, KL102, KL109).The bold and dashed black line represents KL-types where at least 75% of the isolates were correctly identified by FTIR (KL64, KL111).The black dashed line represents KL-types where all the isolates exhibit highly related profiles recognized by FT-IR (KL10, KL30, KL35, KL45, KL124, KL61.KL13).The sensitivity and specificity of FTIR for KL-typing were 78% and 80%, respectively.Farms were arbitrarily designated by capital letters as A, B, C, E, G, H, and I.The Sankey diagram was generated using Tableau Desktop 2023.3 (https://www.tableau.com/).

FIGURE 6
FIGURE 6Heatmap representing the distribution of QAC tolerance genes across representative (n = 45) Klebsiella pneumoniae isolates.The presence of a specific QAC tolerance gene is indicated by a filled blue rectangle.