Phenotypic and Genetic Characterization of Klebsiella pneumoniae Isolates from Wild Animals in Central Italy

Simple Summary Klebsiella pneumoniae is an opportunistic bacterium known as a nosocomial pathogen, but it is able to colonize a multitude of ecological niches, both free-living and host-associated. For this reason, the presence of antimicrobial resistance genes in wildlife is an indicator that resistant bacteria of human- or livestock-origin are widespread in the environment. Wild animals can act as efficient reservoirs and epidemiological links between humans, livestock, and natural environments. Despite the renewal of interest in epidemiology, there is a critical lack of an unbiased ecological approaches to defining reservoirs and sources of infections. The aims of the present study were to investigate the occurrence of K. pneumoniae in wildlife, to characterize antimicrobial features through phenotypic and whole genome sequencing approaches, and to identify similar available sequences that could suggest a transmission event. Abstract Despite Klebsiella pneumoniae being widely recognized as a nosocomial pathogen, there is a critical lack in defining its reservoirs and sources of infections. Most studies on risk factors have focused on multidrug-resistant (MDR) isolates and clinically-oriented questions. Over a two-year period, we sampled 131 wild animals including mammal and bird species from three regions of Central Italy. All typical colonies isolated from the analytical portions were confirmed by real-time PCR and identified by MALDI-TOF mass spectrometry (MALDI-TOF MS). All confirmed K. pneumoniae isolates were tested for antimicrobial susceptibility to 29 antimicrobials and subjected to whole genome sequencing. Typical colonies were detected in 17 samples (13%), which were identified as K. pneumoniae (n = 16) and as K. quasipneumoniae (n = 1) by MALDI-TOF MS. The antimicrobial susceptibility profile showed that all the isolates were resistant to β-lactams (ceftobiprole, cloxacillin, cefazolin) and tetracycline; resistance to ertapenem and trimethoprim was observed and nine out of 16 K. pneumoniae isolates (56.2%) were classified as MDR. Genomic characterization allowed the detection of fluoroquinolone resistance-associated efflux pumps, fosfomycin and β-lactamase resistance genes, and virulence genes in the overall dataset. The cluster analysis of two isolates detected from wild boar with available clinical genomes showed the closest similarity. This study highlights the link between humans, domestic animals, and wildlife, showing that the current knowledge on this ecological context is lacking and that the potential health risks are underestimated.


Introduction
The Central Italy region is characterized by a considerable extent of rural areas and natural parks, which host many wildlife mammal and birds' species, and at the same time by the presence of urbanized and industrial areas. The overlapping of the human activities

The sample Collection
During the period 2020-2021, 119 wild animals including 10 mammal species (wild boar (Sus scrofa), red deer (Cervus elaphus), roe deer (Capreolus capreolus), fallow deer (Dama dama), European badger (Meles meles), red fox (Vulpes vulpes), wolf (Canis lupus italicus), porcupine (Erethizon dorsatum), otter (Lutra lutra), and hedgehog (Erinaceus europaeus)) and 5 bird species (magpie (Pica pica), jay (Garrulus glandarius), goose (Anser anser), sparrow hawk (Accipiter nisus) and starling (Sturnus vulgaris)) were collected in Central Italy, mainly in Abruzzo, Molise, and Lazio Regions. Most of them were collected as part of the regional plans of epidemiological surveillance and monitoring of diseases in wild fauna. Bird species were collected in the frame of the national plan of surveillance of West Nile disease and Usutu. Wild animals of this study died due to anthropic activities (gunshot wounds or impacts with vehicles) and the carcasses were subjected to necropsy for feces and tissue sample collection. A total of 131 diagnostic samples including faces (n = 98), intestine (n = 13), and brain (n = 20) were collected and analyzed at Istituto Zooprofilattico Sperimentale dell'Abruzzo e del Molise "G. Caporale".

Bacterial Isolation and Identification
Analytical portions of 25 g of each sample were enriched in 225 mL of Buffered peptone water (BPW) (Biolife Italiana, Monza, Italy) for 24 ± 1 h at 37 ± 1 • C. For analytical portions lower than 25 g, a volume of BPW to maintain a weight/volume ratio of 1:10 was added to the sample. Thereafter, the enrichment was streaked onto Simmons Citrate agar + Inositol (SCAI) and onto Klebsiella ChromoSelect Selective agar base + selective supplement (Sigma-Aldrich, St. Louis, MO, USA), followed by incubation at 44 ± 1 • C for 48 ± 1 h.
Both from SCAI and Klebsiella ChromoSelect Selective agar, typical yellow mucoid colonies and purple colonies, respectively, were picked up and subcultured onto nutrient agar (Microbiol & C., Cagliari, Italy) at 37 ± 1 • C for 24 ± 1 h, in order to confirm them by real-time PCR.

Real-Time PCR
All typical colonies were subjected to DNA extraction using DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany). Then a real-time PCR was performed in order to detect K. pneumoniae and related species (Kp1 to Kp7 phylogroups, making up the Kp complex) by specific amplification of the zikr region according to Barbier et al. [17] with minor modifications. All real-time PCR assays were performed on a 7500 Fast Real-Time PCR System (Thermo Fisher Scientific, Lenexa, KS, USA) with the following temperature program: 95 • C for 20 s and 40 cycles at 95 • C for 3 s and 60 • C for 30 s. Melt curves were generated with temperature increments of 0.3 • C per cycle from 60 to 95 • C. DNA was amplified in a 20 µL PCR mix containing 10 µL of Kapa syber fast rox qPCR MasterMix (Roche, Basel, Switzerland), 2 µL of each primer (final concentration 300 nM), 2.5 µL of template DNA, and 3.5 µL PCR grade water (Thermo Fisher Scientific).

MALDI-TOF Mass Spectrometry
All real time PCR positive colonies were subjected to matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis.
The acquisition and analysis of mass spectra were carried out by Microflex LT/SH™ spectrometer (Bruker Daltonik GmbH, Bremen, Germany), which was operated in linear positive mode covering the molecular weight range of 2 to 20 kDa. Each spot was hit with 240 shots in several points with a pulsed nitrogen laser ( intestine (n = 13), and brain (n = 20) were collected and analyzed at Istituto Zooprofilattico Sperimentale dell'Abruzzo e del Molise "G. Caporale".

Bacterial Isolation and Identification
Analytical portions of 25 g of each sample were enriched in 225 mL of Buffered peptone water (BPW) (Biolife Italiana, Monza, Italy) for 24 h ± 1 h at 37 °C ± 1 °C. For analytical portions lower than 25 g, a volume of BPW to maintain a weight/volume ratio of 1:10 was added to the sample. Thereafter, the enrichment was streaked onto Simmons Citrate agar + Inositol (SCAI) and onto Klebsiella ChromoSelect Selective agar base + selective supplement (Sigma-Aldrich, Missouri, USA), followed by incubation at 44 °C ± 1 °C for 48 h ± 1 h.
Both from SCAI and Klebsiella ChromoSelect Selective agar, typical yellow mucoid colonies and purple colonies, respectively, were picked up and subcultured onto nutrient agar (Microbiol & C., Cagliari, Italy) at 37 °C ± 1 °C for 24 h ± 1 h, in order to confirm them by real-time PCR.

Real-Time PCR
All typical colonies were subjected to DNA extraction using DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany). Then a real-time PCR was performed in order to detect K. pneumoniae and related species (Kp1 to Kp7 phylogroups, making up the Kp complex) by specific amplification of the zikr region according to Barbier et al. [17] with minor modifications. All real-time PCR assays were performed on a 7500 Fast Real-Time PCR System (ThermoFisher Scientific) with the following temperature program: 95 °C for 20 s and 40 cycles at 95 °C for 3 s and 60 °C for 30 s. Melt curves were generated with temperature increments of 0.3 °C per cycle from 60 to 95 °C. DNA was amplified in a 20 µL PCR mix containing 10 µL of Kapa syber fast rox qPCR MasterMix (Roche, Basel, Switzerland), 2 µL of each primer (final concentration 300 nM), 2.5 µL of template DNA, and 3.5 µL PCR grade water (ThermoFisher Scientific).

MALDI-TOF Mass Spectrometry
All real time PCR positive colonies were subjected to matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis.
In brief, bacterial pure and fresh colonies were applied in a 96-well steel target plate (Bruker Daltonik GmbH, Bremen, Germany) using a toothpick overlaid with 1 µL of αcyano-4-hydroxycynnamic acid-matrix solution (HCCA, Bruker Daltonik GmbH, Bremen, Germany), and air dried at room temperature (18-27 °C). For each isolate, three replicates were performed.
The acquisition and analysis of mass spectra were carried out by Microflex LT/SH™ spectrometer (Bruker Daltonik GmbH, Bremen, Germany), which was operated in linear positive mode covering the molecular weight range of 2 to 20 kDa. Each spot was hit with 240 shots in several points with a pulsed nitrogen laser (ʎ = 337 nm) at a frequency of 60 Hz. Prior to analysis, the instrument was calibrated using an extract of the Escherichia coli DH5α strain, with two additional proteins: RNase A of 13,683.2 Da and Myoglobin of 16,952.3 Da (BTS, Bruker Daltonik GmbH, Bremen, Germany). The mass spectra were generated automatically by the MBT Compass Explorer 4.1.70 software, which compares the sample mass spectrum to the reference mass spectra in the database and calculates a logarithmic score value (0-3.0) reflecting the similarity between the sample and the reference spectrum. If the log (score) value is ≥ 2.0, the bacterial identification is reliable at the species level; if it is between 1.7 and 1.99, it is reliable at the genus level, while if the log (score) is <1.7, bacterial identification is not reliable by this technique. λ = 337 nm) at a frequency of 60 Hz. Prior to analysis, the instrument was calibrated using an extract of the Escherichia coli DH5α strain, with two additional proteins: RNase A of 13,683.2 Da and Myoglobin of 16,952.3 Da (BTS, Bruker Daltonik GmbH, Bremen, Germany). The mass spectra were generated automatically by the MBT Compass Explorer 4.1.70 software, which compares the sample mass spectrum to the reference mass spectra in the database and calculates a logarithmic score value (0-3.0) reflecting the similarity between the sample and the reference spectrum. If the log (score) value is ≥2.0, the bacterial identification is reliable at the species level; if it is between 1.7 and 1.99, it is reliable at the genus level, while if the log (score) is <1.7, bacterial identification is not reliable by this technique.

Antimicrobial Susceptibility Testing
All confirmed K. pneumoniae isolates were tested for antimicrobial susceptibility to 29 antimicrobials using the agar disk diffusion and the microdilution methods. The antimicrobials tested are representative of the major classes of antimicrobial drugs β-lactams, quinolones, and aminoglycosides important to both veterinary and human medicine. For antimicrobial susceptibility testing by microdilution method, K. pneumoniae isolates stored in cryobank tubes (Pro-Lab Diagnostics, Richmond Hill, Canada), were streaked on nutritive agar plates (Microbiol, Cagliari, Italy) and incubated at 37 • C for 24 h. At the end of the incubation, the purity of the culture was visually checked for morphology and 3-5 colonies were picked and suspended in a tube containing 5 mL of Sensititre demineralized water (Thermo Fisher Scientific, Lenexa, KS, USA) to reach a final turbidity equal to 0.5 McFarland standard (Biomeriux Italia, Florence, Italy). Immediately after, 10 µL of each inoculum was further added to a tube containing Sensititre Cation Adjusted Mueller-Hinton Broth (SCAMHB) (Thermo Fisher Scientific, Lenexa, KS, USA).
For antimicrobial susceptibility testing by disk diffusion method, K. pneumoniae isolates stored in cryobank tubes (Pro-Lab Diagnostics, Richmond Hill, Canada), were streaked on nutritive agar plates (Microbiol, Cagliari, Italy) and incubated at 37 • C for 24 h. At the end of the incubation, the purity of the culture was visually checked for morphology and 3-5 colonies were picked and suspended in a tube containing 5 mL of meat broth (IZSAM, Italy) to reach a final turbidity equal to 0.5 McFarland standard (Biomeriux Italia, Florence, Italy). Immediately after, each inoculum was spread on Mueller-Hinton agar (Liofilchem, Teramo, Italy) using a sterile swab. The antibiotics (Oxoid Ltd., Basingstoke, UK) tested by disk diffusion method were as follows: cefpodoxime-clavulanic acid (CD 11 µg), ceftobiprole (BPR 5 µg), kanamycin (KAN 30 µg), nalidixic acid (NAL), netilmicin (NET 10 µg), trimethoprim (TMP 5 µg), and cloxacillin (CLO 30 µg). The interpretation of disk diffusion results was carried out in accordance with EUCAST guidelines [18].
Multidrug-resistant (MDR) isolates were defined as those isolates resistant to ≥3 antimicrobial classes in addition to ampicillin, to which all K. pneumoniae isolates are known to be resistant [16].

Next-Generation Sequencing (NGS) and Data Analysis
All the isolates identified by MALDI-TOF MS as K. pneumoniae were also subjected to DNA extraction according to Portmann et al. [20]. Cultures were grown overnight in nutrient agar (Microbiol & C., Cagliari, Italy) at 37 ± 1 • C and DNA extraction was performed on single colonies using the DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions.
Then, 1 ng of DNA of each sample was used for library preparation using Nextera XT DNA chemistry (Illumina, San Diego, CA, USA) according to the manufacturer's protocols. Whole genome sequencing was performed using the NextSeq 500 platform (Illumina, San Diego, CA, USA) with the NextSeq 500/550 mid output reagent cartridge v2 (300 cycles, standard 150-bp paired-end reads) For the WGS data analysis, an in-house pipeline [21] was used, which included steps for trimming using Trimmomatic v.0.36 [22] and a quality control check of the reads with FastQC v.0.11.5 [23]. Genome de novo assembly of the reads was performed using SPAdes v.3.11.1 [24]. The genome assembly quality check was performed with QUAST [25].

Typing, AMR, and Virulence Gene Profiles
The species detection and capsule (K) and O antigen (LPS) serotype prediction for each draft genome were performed using Kleborate [26] and Kaptive [27], respectively. Then, the assemblies were uploaded onto the BIGSdb platform [28] to calculate the sequence type (ST) using the K. pneumoniae MLST scheme [29].
All the assemblies obtained in this study were analyzed using Kleborate to detect the presence of acquired virulence loci associated with invasive infections such as siderophores yersiniabactin (ybt), aerobactin (iuc) and salmochelin (iro), the genotoxin colibactin (clb), and the hypermucoidy locus rmpADC.
PlasmidFinder 2.1 [36] was used to verify the presence of plasmid replicons carrying AMR gene resistance.
Finally, the visualization of the phenotypic antimicrobial susceptibility profiles was assessed by heatmap, using the Morpheus tool [37].

Clustering Analysis
cgMLST analysis was performed to verify the relatedness among the 16 K. pneumoniae draft assemblies according to the Pasteur Institute's scheme of 629 loci [38] hosted in BIGSdb database. The resulting neighbor joining (NJ) tree was visualized using Interactive Tree of Life (iTOL) [39].
In order to identify similar available sequences that could be indicative of a transmission event or outbreak using an allele-based method, the K. pneumoniae genomes that showed particular AMR and virulence patterns were uploaded onto the Pathogenwatch platform.

Bacterial Detection and Identification Results
From a total of 131 samples tested, K. pneumoniae typical colonies were detected by plate culture method in 17 samples (13%). All 17 presumptive colonies were confirmed by real-time PCR and identified as K. pneumoniae by MALDI-TOF MS. The isolates were isolated from wild animals distributed mainly in the Abruzzo, Molise, and Lazio regions ( Figure 1). The metadata related to the 17 isolates are summarized in Table 1.

Whole Genome Sequencing and Typing Results
The average reads quality score after trimming returned 34. The species detection results from Kleborate, countering those obtained from MALDI-TOF, showed that 16 genomes belonged to K. pneumoniae and one to K. quasipneumoniae (2021.TE.10), isolated from a European badger. This result was not surprising since K. quasipneumoniae spectra were not available in the MBT Compass Library.
Among the sixteen K. pneumoniae genomes, Kleborate identified 11 KL and 7 O antigen serotypes. Meanwhile the single K. quasipneumoniae isolate was predicted as a KL35 and O3/O3a serotype.
All the results on genomes' typing are reported in Table 2.

Antimicrobial Susceptibility Testing
The antimicrobial susceptibility profiles of isolates are reported in Figure 2. All 17 isolates (100%) were resistant at least to one antimicrobial. Regarding β-lactams, all K. pneumoniae isolates were resistant to ampicillin, cloxacillin, and cefazolin; on the other hand, all K. pneumoniae isolates were susceptible to cefpodoxime, ampicillin/sulbactam, ceftazidime, ceftriaxone, and cefuroxime. For aztreonam, ticarcillin-clavulanic acid, piperacillintazobactam, and cefepime, all isolates showed an intermediate susceptibility profile. Meanwhile, 3 of the 16 isolates (18.8%) were resistant to cefoxitin and 13 of the 16 isolates (81.3%) were resistant to ceftobiprole. About carbapenems, all isolates were susceptible to meropenem, but 7 of the 16 isolates (43.8%) were resistant to ertapenem. All the 16 K. pneumoniae isolates were susceptible to cefpodoxime-clavulanic acid, except one (2021.TE.3) that had intermediate susceptibility.
All the 16 K. pneumoniae isolates were susceptible to aminoglycosides (amikacin, gentamicin, netilmicin, and kanamycin), tigecycline, and quinolones (nalidixic acid). Importantly, all isolates were resistant to tetracycline and intermediate for tobramycin and ciprofloxacin. The interpretation of antimicrobial susceptibility profile was not possible for cephalotin due to the lack of both EUCAST and CLSI breakpoints.
No plasmids were found in 7 K. pneumoniae isolates isolated from fallow deer, roe deer, deer, and one European badger.

cgMLST Analysis
Cluster analysis based on allele calling was performed on 16 out 17 isolates confirmed as K. pneumoniae.

Discussions
Our results indicate that overall, the occurrence of K. pneumoniae in wild animals in Central Italy is low (12.2%). The phenotypic and genomic characterization of AMR in 16 K. pneumoniae and one K. quasipneumoniae isolates from wild animals allow us to deepen knowledge on this pathogen in this ecological niche.
The antimicrobial susceptibility testing results highlighted significant levels of phenotypic resistance against β-lactams (ampicillin, cloxacillin and cefazolin), ertapenem, tetracyclines, and trimethoprim. Furthermore, we observed the presence of nine MDR K. pneumoniae isolates from a wolf, a deer, a magpie, a European badger, two fallow deer, and three wild boars. Moreover, investigating all the resistance genes detected in the dataset, bla SHV genes were observed in all the 16 K. pneumoniae isolates and bla OKP-A-2 in only the K. quasipneumoniae isolate, as previously described by Wyres et al. [40]. Only one isolate harbored other resistance genes such as sul2, tetA, strA, and strB, which are responsible for resistance to sulfonamides, tetracycline, and aminoglycosides, respectively, although full consistency with the phenotypic profile was not observed.
The widespread use of tetracycline in the veterinary sector and agriculture for its broad spectrum of activity has increased the bacterial community's tolerance to this compound [41], even if only one isolate detected in a wild boar harbored the tetA gene.
On carbapenems, more than half of the isolates were resistant to ertapenem, even if none of the carbapenemase genes were found. The presence of resistant isolates in our study could be linked to the loss or deficiency of outer membrane porins (OMPs) that play an important role in conferring carbapenem resistance in K. pneumoniae [40]. Carbapenems are last resort drugs used in clinical settings, and although the prevalence of carbapenemresistant Enterobacteriaceae (CRE) is low in animals, it is increasing and has been reported in companion animals, livestock, and even in wildlife and environmental samples [2]. The unusually high resistance rate to this compound may be explained by the increasing contact between wild animals and human-related environments, such as urban, agricultural, or livestock wastewaters, which are known to be some of the major reservoir of antibiotics and resistant bacteria [5].
Many available studies have been carried out on other Enterobacteriaceae, in particular Escherichia coli, and have shown high antibiotic resistance rates from wild animals in other Italian regions [8][9][10][11]. In Italy, few data are available on K. pneumoniae occurrence in wild animals and its characterization of AMR, and the data concerns wild birds only [42,43]. The available studies in Europe concern wild boars from Algeria [44], Chinese hares [45], hedgehogs in Spain [46], and European mouflons in Austria and Germany [47].
On mobile genetic elements, both the incompatibility group (Inc) plasmids and the colicin (Col)-type plasmid were detected in most of the isolates. Additionally, the replicon plasmid pK2044 was detected in the hypervirulent ST23 isolate from a wild boar, representing a possible concern. Indeed, the plasmid pK2044 was first described in 2009 by Wu et al. [48] and was found in an invasive K. pneumoniae isolate that caused liver abscesses and meningitis in humans.
Recently, the emergence from clinical specimens of a hypervirulent ST23 clone carrying carbapenemase genes has subverted the "classical" division among hypervirulent and AMR clones [49]. Moreover, the convergence of antimicrobial resistance and virulence in hypervirulent isolates represents a public health concern due to the onset of difficult-to-treat infections in previously healthy adults [49].
As previously mentioned, the hypervirulent ST23 strain isolated from wild boar was shown to be similar to clinical isolates carrying carbapenemase genes, as revealed by the cgMLST analysis. In this study, the similarity among K. pneumoniae isolates isolated from a wild boar and clinical cases was also demonstrated for the MDR ST35 clone. These results are concordant with other authors [5,50,51], which show that identical or near identical strains belonging to the same clonal complex could circulate in wildlife, humans, and domestic animals.
In this study, isolates were mainly detected from the intestinal contents, confirming the carrier function of animals. Although the necropsies did not reveal any lesions, four isolates were detected in the brain and we cannot exclude the pathogenic potential of K. pneumoniae. Furthermore, some of these wild animals, such as wild boar, are also food-producing animals (e.g., Italy), where, according to local tradition, it is customary to eat seasoned raw meat products (e.g., salami, sausages). These traditional foods represent a possible way of transmission to humans, as has been demonstrated for other pathogens such as Brucella spp., and Mycobacterium spp. [52].
The zoonotic potential of K. pneumoniae in food-producing animals was widely reviewed by Davis and Price [53] and recently by Rodrigues et al. [54]. They described the contamination of milk and a variety of retail meats including poultry, beef, and pork, as well as freshwater fish and prawns, by K. pneumoniae. Furthermore, Davis et al. [55] and Rodrigues et al. [54] described the phylogenetic relatedness between urinary tract infection isolates with retail meat isolates sampled in the same geographic area.
In this work, various species of wild animals characterized by distinct ecological contexts have also been considered. According with Ramey & Ahlstrom [56], these species, as wild boars, can be defined as peridomestic animals, because they reliant upon humandominated activities or managed by human habitats. The wildlife can be exposed to anthropogenic antibiotic resistance contamination through agricultural activities, livestock production practices, waste disposal, and wastewater management.
The overlapping of wildlife habits with livestock and human ones highlight that wild animals can represent a potential reservoir of AMR bacteria and can reveal the impact of human activities in the environment [1,52].
For this, in-depth knowledge about the ecological context of wildlife can provide insight into the environmental pathways through which antimicrobial determinants can be lost or acquired [57].

Conclusions
Despite the renewal of interest in K. pneumoniae epidemiology, there is a critical lack of an unbiased ecological approaches to defining the reservoirs and sources of infections. Most studies on risk factors of K. pneumoniae colonization or infection have focused on clinically-oriented questions, such as detection of ESBL-producing K. pneumoniae. The current knowledge of the ecological context of K. pneumoniae is still lacking, and the potential health risks are underestimated.
This study highlighted the link between the interface of humans, animals, and environment, showing once more the importance of establishing collaboration strategies between public health actors to control and prevent the spread of AMR bacteria.
For these reasons, continuous data collections in molecular epidemiology and resistance development in wildlife, environment, and foods must be implemented.