One Health Perspective on Infections in Nephrology: A Parallel Survey on Carbapenem- and Multidrug-resistant Klebsiella Pneumoniae and Pseudomonas Aeruginosa in High-risk Renal Patients and Farm Animals in Italy

Silvia Bonardi (  silvia.bonardi@unipr.it ) University of Parma: Universita degli Studi di Parma https://orcid.org/0000-0003-3079-9290 Enrico Fiaccadori University of Parma: Universita degli Studi di Parma Antonio Parisi Istituto Zoopro lattico Sperimentale della Sardegna G Pegre Gerardo Manfreda Università degli Studi di Bologna: Universita di Bologna Sandro Cavirani University of Parma: Universita degli Studi di Parma Clotilde Silvia Cabassi University of Parma: Universita degli Studi di Parma Michele Luca D’Errico University of Parma: Universita degli Studi di Parma Giovanni Pupillo University of Parma: Universita degli Studi di Parma Alice Sabatino University of Parma: Universita degli Studi di Parma Costanza Spadini University of Parma: Universita degli Studi di Parma Cecilia Crippa Università degli Studi di Bologna: Universita di Bologna Gisella Pizzin University of Parma: Universita degli Studi di Parma Ferdinando Ruocco Italian Veterinary Services Frédérique Pasquali Università degli Studi di Bologna: Universita di Bologna

Written informed consent was obtained from the patients when possible or from their closest relatives when available. The Ethics Committee of the Area Vasta Emilia Nord (AVEN) approved the study (protocol n. 24497).

2 Pig sampling
From October 2016 to December 2017, 300 nishing pigs were randomly selected during 23 visits in a slaughterhouse of Parma province. The animals belonged to 300 batches and were reared in 95 farms. The caecal content of pigs was aseptically collected immediately after evisceration.

Pig sample testing
A 10 g aliquot of faecal material was suspended in 90 ml of Buffered Peptone Water (BPW; Biolife Italiana, Italy) and incubated at 37 ± 1°C for 18 -22 h. Thereafter, cultures plating and colonies selection were performed as described in 2.2.1.

Kirby-Bauer Disk Diffusion Susceptibility test
Human and porcine Gram-negative isolates were screened by the Kirby-Bauer Disk Diffusion Susceptibility test following EUCAST (2018) [17] and considering both clinical and screening breakpoints.

Whole genome sequencing
Based on MIC, selected K. pneumoniae and P. aeruginosa isolates were tested by WGS. Whole-genomic DNA was extracted from K. pneumoniae and P. aeruginosa isolates using the MagAttract HMW DNA Kit (Qiagen, Hilden, Germany). The puri ed DNA concentration and the quality parameter ratio 260/280 were measured by BioSpectrometer uorescence (Eppendorf). Whole genomes were sequenced on Illumina MiSeq platform (Nextera library, paired-end reads). The INNUca v3.2 pipeline was used to quality check and de novo assemble reads into contigs. The pipeline includes SPAdes 3.11 as de novo assembler and provides Multi Locus Sequence Typing (MLST) pro le.
SNP calling was performed on reads using snippy 4.0 pipeline with default setting (https://github.com/tseemann/snippy). An alignment of core SNPs was produced to infer a high-resolution phylogeny. A maximum likelihood (ML) tree was constructed using the iQTree program and was viewed using iTOL v4. Analysis of resistome of genomes was performed using ABRicate [19]. With this tool a BLAST search of genes included in the Res nder database was performed on de novo assemblies [20,21].
De novo assemblies were annotated by RAST version 2.0 [22]. Gbk les were used as input les for visualisation of the genetic environment of carbapenemase encoding genes by SnapGene [23]. Assembled sequences are available at NCBI [24] under BioProject PRJNA587603.

Pseudomonas aeruginosa of human and porcine origin
Two CR P. aeruginosa strains were isolated from patients. Apparent prevalence was 1.1% (2/187) (95% CI: 0.3%-3.8%) ( Table 1). The NEF23 strain belonged to ST938 and its resistome consisted of ve genes, including bla OXA−396 and bla PAO, while the NEF156 strain belonged to ST395 and its resistome consisted of six genes, including bla OXA−488 and bla PAO (Table 4). In pigs, four CR P. aeruginosa strains were detected. Speci cally, one ST938 isolate (CRE102) was identical to the NEF23 from a 83-old male patient hospitalized one month after pig slaughter for an episode of AKI due to sepsis, on previous chronic kidney disease. Moreover, based on SNP calling, the two genomes differed by only 41 SNPs suggesting the close genetic relationship among the porcine and the human isolates (Fig. 3, Table S3). No direct epidemiological link was found.
The remaining P. aeruginosa isolates of porcine origin belonged to ST274, ST782 and ST885 and all of them harboured the bla PAO gene. In addition, bla OXA−50 (ST782 and ST885) and bla OXA−486 (ST274) were identi ed together with other AMR genes (Tables 1 and 4).

Single Nucleotide Polymorphisms (SNPs) calling
Results on de novo assembly statistics are reported in table S1. SNP calling con rmed results of MLST typing with a higher level of discriminatory power (Figs. 2 and 3). Regarding K. pneumoniae, within the same ST types pairwise SNP differences were 41 (ST37), 51 (ST277), 55 (ST258), ranging from 29 and 89 (ST554) and 1071 (ST307). Isolate NEF123 (ST4525) showed 292 and 296 SNPs differences to NEF112 and NEF48 both belonging to ST37, demonstrating the close genetic relationship of the two ST-types (Fig. 2, table S2). Not surprisingly, between genomes of different ST-types the genetic distance increased with pairwise SNPs differences ranging from 905 to 32805.

Localisation of carbapenemase-encoding genes
Within the meropenem-resistant K. pneumoniae isolates, the carbapenemase-encoding genes bla KPC−2 , bla KPC−3 or bla VIM−1 were detected. To evaluate their mobilisation potential, their genetic environments were investigated. bla KPC−2 and bla KPC−3 were included in a complete Tn4401 (Figs. 4 and 5). Tn4401 was previously described as a transposon belonging to the Tn3 family ad involved in bla KPC mobilisation [25].
bla VIM−1 was located close to other AMR determinant genes (Fig. 6). Although the contig was too short to con rm the localisation of the gene in a class 1 integron, the sequence showed 99% identity in comparison to K. pneumoniae genomes (CP032178.1, MF344563.1) in which the sequence was included in a class 1 integron within a transposon (data not shown). Moreover, this gene was already detected within class 1 integrons as part of a Tn402 transposon [26].
Regarding P. aeruginosa, bla OXA and bla PAO showed genetic environments which did not show any speci c sequence of mobilizable genetic elements (MGEs) con rming previous data (Figs. 7 and 8) [27].

Klebsiella pneumoniae in humans
In our study, K. pneumoniae ST258 bla KPC-2, which is one of the most common in Italian hospital settings [7], was identi ed in two patients only (22.2%) tested on the same day. Instead, K. pneumoniae ST554 bla KPC-3 has been the only clone isolated from the two renal wards for 10 months (July 2017-May 2018), suggesting a long-lasting outbreak involving six patients. MDR to amynoglicosides, β-lactams, macrolides, phenicols, sulphonamides, uoroquinolones and fosfomycin was associated to carbapenem resistance.
To our knowledge, this is the rst report of K. pneumoniae ST37 producing VIM-1 in Italy. The two isolates were detected from the same patient (72 years old male with AKI) tested twice during a three-month period (April-July 2017. The isolates carried the bla VIM−1 gene within a class 1 integron, along with additional ESBL-encoding genes (bla CTX−M−15 , bla SHV−12, bla OXA−1 ) and several AMR determinants. High levels of AMR were identi ed also in the new sequence type ST4525 bla VIM−1 (NEF123), including ESBL genes (bla CTX−M−15 , bla SHV−12 , bla OXA−1) and genes for resistance to aminoglycosides, uoroquinolones, macrolides, fosfomycin, chloramphenicol, sulphonamides, and trimethoprim.

K. pneumoniae in pigs
K. pneumoniae has been associated with sporadic cases of disease in pigs, including pneumonia, mastitis and septicaemia [28]. In our study, CR K. pneumoniae was never isolated from swine, not only because carbapenems are banned in veterinary treatments, but also for the high biosecurity measures commonly applied in pig farms, which exclude access of pigs to the natural environment or anthropic sources of AMR bacteria [29].

Pseudomonas aeruginosa in humans
Although P. aeruginosa β-lactams resistance in usually mediated by the chromosomally encoded AmpC type β-lactamase induced by contacts with sub-MIC levels of β-lactams, the microorganism can acquire resistance through chromosomal mutations as well as acquisition of antimicrobial resistance genes horizontally transferred [30].
The two CR ST938 and ST395 strains did not harbour carbapenemase-encoding genes, but bla OXA and bla PAO genes, con rming that decreased susceptibility to meropenem could be conferred by oxacillinases [31].

Pseudomonas aeruginosa in pigs
In most of cases, P. aeruginosa has not a primary role in pig infections [32]. Interestingly, two porcine strains harboured β-lactamases genes, such as bla OXA−50 which is an intrinsic oxacillinase able to confer decreased susceptibility to ampicillin, ticarcillin and meropenem [31].

Shared P. aeruginosa ST938 between humans and pigs and implications for safety
This is the rst report of an ESBL-MDR P. aeruginosa strain isolated both from humans and pigs in Italy, whose resistance to carbapenems is due to βlactamases genes other than carbapenemases. P. aeruginosa ST938 carrying bla OXA−396 and bla PAO was isolated from a pig slaughtered in February 2017 and from an 83-old male patient hospitalized one month later for an episode of AKI due to sepsis, on previous chronic kidney disease. Even if no epidemiological links between these ndings were found, the occurrence of the same CR strain in humans and pigs re ects the One Health concept of bacterial sharing between humans and food animals.
Since other isolates of MDR P. aeruginosa harbouring ESBL genes were found in pigs, the potential risk of pig meat contamination and subsequent human exposure to MDR opportunistic pathogens could not be excluded. Especially raw pork and products thereof (e.g. sausages, salami) could represent a risk for humans.

Conclusions
In recent years, the concept of One Health has been reformulated underlying the role of geographically close ecosystems in the occurrence of traits which have an impact on human, animal, plant, and environmental health, such as AMR [33].
In our study based on the One Heath perspective of bacterial exchange between human and animal compartments, K. pneumoniae strains harbouring bla KPC-2 , bla KPC-3 and bla VIM-1 were isolated from patients only. This nding is consistent with the European ban of carbapenem use in food-producing animals.
Nevertheless, despite treatments bans, environmental overlapping could cause microbial and MGEs exchange between humans and animals, causing wide distribution of AMR bacteria in different settings. On the contrary, P. aeruginosa was found both in human and porcine entities, but the isolates did not carry carbapenemase genes and their carbapenem-resistance was associated to bla OXA-50 family and bla PAO genes. In addition, MDR microorganisms were mostly detected in K. pneumoniae strains from hospitalized patients, probably because of the selection pressure following antimicrobial treatments. The surprising similarity in AMR of P. aeruginosa strains detected from pigs and patients could be due to improper use in pig husbandry of the same antimicrobial agents used in the hospital setting. However, the circulation of bacterial strains between the animal and human compartments cannot be excluded.

Declarations
Funding: the study was funded by Cariparma Foundation Grants No 2016.0108 and No 2018.0176 Con icts of interest: The authors declare no con icts of interest.