Genome-Based Analysis of Klebsiella spp. Isolates from Animals and Food Products in Germany, 2013–2017

The increase in infections with multidrug-resistant and virulent Klebsiella pneumoniae (K. pneumoniae) strains poses a serious threat to public health. However, environmental reservoirs and routes of transmission for Klebsiella spp. that cause infections in humans and in livestock animals are not well understood. In this study, we aimed to analyze the distribution of antibiotic resistance genes and important virulence determinants (ybt, clb, iro, iuc, rmpA/A2) among 94 Klebsiella spp. isolates from different animal and food sources isolated between 2013 and 2017 in Germany. Antibiotic susceptibility testing was performed, and the genomes were sequenced by Illumina and Nanopore technology. Genetic relationships were assessed by conducting core genome multilocus sequence typing (cgMLST). Kleborate was used to predict resistance and virulence genes; Kaptive was used to derive the capsule types. The results revealed that 72 isolates (76.6%) belonged to the K. pneumoniae sensu lato complex. Within this complex, 44 known sequence types (STs), 18 new STs, and 38 capsule types were identified. Extended-spectrum beta-lactamase (ESBL) genes were detected in 16 isolates (17.0%) and colistin resistance in one (1.1%) K. pneumoniae isolate. Virulence genes were found in 22 K. pneumoniae isolates. Overall, nine (9.6%) and 18 (19.1%) isolates possessed the genes ybt and iuc, respectively. Notably, aerobactin (iuc lineage 3) was only detected in K. pneumoniae isolates from domestic pigs and wild boars. This study provides a snapshot of the genetic diversity of Klebsiella spp. in animals and food products in Germany. The siderophore aerobactin was found to be more prevalent in K. pneumoniae strains isolated from pigs than other sources. Further investigations are needed to evaluate if pigs constitute a reservoir for iuc lineage 3.


Introduction
Klebsiella ssp. are Gram-negative, rod shaped, facultative anaerobic bacteria that belong to the Enterobacteriaceae family. They are ubiquitous in soil, surface waters, plants, and intestines of animals and humans [1,2]. The Klebsiella pneumoniae (K. pneumoniae) sensu lato complex is clinically the most problematic and comprises the phylogroups K. pneumoniae (Kp1), K. quasipneumoniae subsp. quasipneumoniae (Kp2), K. quasipneumoniae subsp. similipneumoniae (Kp4), K. variicola subsp. variicola (Kp3), K. variicola subsp. tropica (Kp5), K. quasivariicola (Kp6), and K. africana (Kp7) and [3,4]. K. pneumoniae is the most prevalent species isolated from human infections [4,5]. Classically, K. pneumoniae is considered to be an opportunistic pathogen, causing infections in immunocompromised patients, including urinary tract infections, pneumonia, and bloodstream infections (BSIs) [6,7]. K. pneumoniae is the second most common cause of BSIs caused by Gram-negative bacteria [8]. Due to the acquisition of multiple antibiotic resistance genes, the treatment of K. pneumoniae infections has become challenging and multidrug-resistant Klebsiella spp. are considered a public health threat-especially for predisposed persons [9][10][11]. In the last decades, K. pneumoniae has also been reported as a cause of community-acquired infections, including liver abscesses, endophthalmitis, and meningitis in otherwise healthy individuals [7,12]. These infections are often caused by hypervirulent isolates representing specific K. pneumoniae lineages, e.g., sequence types ST23 and ST66 [13,14]. Hypervirulent strains possess a distinct portfolio of virulence factors, mainly several siderophore genes, and express specific capsule types (e.g., K1 or K2) [4,15]. Furthermore, hypervirulent K. pneumoniae often demonstrate a hypermucoviscous phenotype that can be determined by a 'string test' on agar plates [16]. Distinct hypervirulent strains that have also acquired carbapenemase genes can enter hospitals and spread among patients. This has mainly been described for China but can also occasionally be observed in Central Europe, including in Germany [17].
In animals, K. pneumoniae can cause infections including pneumonia, mastitis, and bacteremia [18]. While the molecular and epidemiological characteristics of clinically important, multidrug-resistant K. pneumoniae STs in humans are well studied, little is known of the K. pneumoniae population structure in animals [4]. Livestock colonized by K. pneumoniae is thought to be a potential reservoir for antimicrobial resistance determinants and virulence factors and could facilitate the spread to human isolates [19]. The objective of this study was to investigate the molecular-genetic characteristics of Klebsiella spp. isolates from animal and food sources in Germany via whole-genome sequencing (WGS)based analyses.

Klebsiella spp. Population Analysis
The study included 94 Klebsiella spp. isolates from animals and food sources of which most isolates (n = 58; 61.7%) were from pigs, pork, cattle, and milk. Furthermore, the collection contains isolates from vegetables, pets, livestock, and wild animals ( Figure 1). MALDI-TOF biotyping suggested the presence of 69 K. pneumoniae, 21 K. oxytoca, and 4 K. variicola, using the commercial MALDI-biotyper database comprising 8468 entries. Subsequent WGS-based analysis revealed that one proposed K. pneumoniae and ten proposed K. oxytoca isolates were mis-assigned and instead belonged to the species K. grimontii, and nine proposed K. oxytoca isolates to the species K. michiganensis. All K. grimontii genomes contained the beta-lactamase gene bla OXY 6 which is known to be specific to this Klebsiella species [20,21]. Within the K. pneumoniae sensu lato complex, 67 (71.3%) isolates were identified as K. pneumoniae, four (4.3%) as K. variicola subsp. variicola and one (1.1%) as K. quasipneumoniae subsp. similipneumoniae ( Figure 1).

Klebsiella spp. Antibiotic Resistance and Resistance Genes
The majority of the 94 study isolates (n = 73, 77.7%) were resistant to ampicillin. Resistance rates to third-generation cephalosporins, fluoroquinolones, and colistin were 5.3%, 3.2%, and 1.1%, respectively. Only one isolate showed resistance to all three classes (Supplementary Table S1). The genome-based prediction of resistance genes revealed the presence of various beta-lactamase genes (bla SHV, bla LAP , bla LEN , bla OKP-B , bla OXY , bla TEM ) in all but three Klebsiella spp. isolates (Table 1). Although only five isolates exhibited an ESBL phenotype (resistance to third-generation cephalosporins), WGS analyses identified 16 K. pneumoniae isolates that carried ESBL genes bla SHV-27 (n = 9), bla SHV-41 (n = 2), bla CTX-M-1 (n = 1), bla CTX-M-14 (n = 2), and bla CTX-M-15 (n = 2). However, subsequent sequence analysis of the bla SHV-27 and bla SHV-41 genes revealed a base-pair substitution (A to C) in all promoter sequences of these genes ( Figure 2). The combination of the amino acid substitutions in GyrA (S83I) and ParC (S80I) (n = 2) and the acquired resistance genes qnrB (n = 1) and qnrS (n = 1) were detected in three isolates with resistance to the fluoroquinolone ciprofloxacin. Sequence analysis of the single colistin-resistant isolate showed that the chromosomal gene mgrB was interrupted by an IS1-family transposase gene. Resistance to carbapenems was not observed in the study isolates.  Graph includes species (1st column), phylogroups (2nd column; [3,4]), isolate sources (3rd column) and presence of specific virulence genes (4th column).

Molecular Typing
MLST analysis of the 72 K. pneumoniae sensu lato complex isolates identified 62 STs, with ST107 (n = 4, 5.6%) as the most prevalent type (Table 2). Furthermore, 18 new STs were discovered. The genetic diversity was also reflected by capsule locus typing (Supplementary Table S1). The capsule loci KL30 (n = 7, 9.7%) was the most prevalentout of 38 capsule types (Figure 3). Furthermore, no association of specific ST or capsule type with isolation sources was identified. cgMLST revealed only three small clusters of genetically highly related isolates. Each cluster comprised only two isolates ( Figure S1), which did not show more than eight allele differences. Overall, cgMLST analysis revealed no prevalent genetic lineage (Figure 3).

Molecular Typing
MLST analysis of the 72 K. pneumoniae sensu lato complex isolates identified 62 STs, with ST107 (n = 4, 5.6%) as the most prevalent type (Table 2). Furthermore, 18 new STs were discovered. The genetic diversity was also reflected by capsule locus typing (Supplementary Table S1). The capsule loci KL30 (n = 7, 9.7%) was the most prevalentout of 38 capsule types ( Figure 3). Furthermore, no association of specific ST or capsule type with isolation sources was identified. cgMLST revealed only three small clusters of genetically highly related isolates. Each cluster comprised only two isolates ( Figure S1), which did not show more than eight allele differences. Overall, cgMLST analysis revealed no prevalent genetic lineage (Figure 3).

Presence of Virulence Factors
Genetic factors that are associated with hypervirulence were only detected in K. pneumoniae isolates. Overall, 22 K. pneumoniae genomes were positive for one or two siderophore systems ( Figure 1). Yersiniabactin (ybt) was present in nine (9.6%) isolates and 18 isolates (19.1%) contained aerobactin (iuc) ( Supplementary Table S1). Interestingly, aerobactin was only identified in domestic pig and wild boar isolates and belonged to the iuc lineage 3, whereas the presence of yersiniabactin was not associated with a specific isolation source. The addition of long-read sequencing for strain 30312,2 revealed a 171 kb IncFIB K plasmid carrying iuc3. Subsequent sequence analysis showed that the IncFIB K replicon was present in all iuc3 plasmids. Further virulence genes encoding for RmpA/A2, salmochelin, and colibactin were not detected in our study isolates.

Presence of Virulence Factors
Genetic factors that are associated with hypervirulence were only detected in K. pneumoniae isolates. Overall, 22 K. pneumoniae genomes were positive for one or two siderophore systems (Figure 1). Yersiniabactin (ybt) was present in nine (9.6%) isolates and 18 isolates (19.1%) contained aerobactin (iuc) ( Supplementary Table S1). Interestingly, aerobactin was only identified in domestic pig and wild boar isolates and belonged to the iuc lineage 3, whereas the presence of yersiniabactin was not associated with a specific isolation source. The addition of long-read sequencing for strain 30312,2 revealed a 171 kb IncFIBK plasmid carrying iuc3. Subsequent sequence analysis showed that the IncFIBK replicon was present in all iuc3 plasmids. Further virulence genes encoding for RmpA/A2, salmochelin, and colibactin were not detected in our study isolates.

Discussion
The analysis of 94 randomly collected Klebsiella spp. isolates from different animals and food sources showed that the species identification was not consistent between the two employed methods. Although MALDI-TOF spectrometry is commonly used in veterinary and medical microbiology laboratories, its discriminatory power depends on the underlying database, which requires continuous curation and updates, especially when closely related taxa such as Klebsiella spp. are analyzed. In this case, WGS-based analyses are better suited to species identification [4,5,22]. In accordance with previous studies, a

Discussion
The analysis of 94 randomly collected Klebsiella spp. isolates from different animals and food sources showed that the species identification was not consistent between the two employed methods. Although MALDI-TOF spectrometry is commonly used in veterinary and medical microbiology laboratories, its discriminatory power depends on the underlying database, which requires continuous curation and updates, especially when closely related taxa such as Klebsiella spp. are analyzed. In this case, WGS-based analyses are better suited to species identification [4,5,22]. In accordance with previous studies, a substantial proportion of the study isolates (n = 67, 71.3%) were classified as K. pneumoniae sensu stricto, suggesting a wide prevalence also in a veterinary context. Such isolates appeared as food colonizers or as infectious agents among livestock animals [23][24][25].
Klebsiella spp. are rarely detected in animals and food samples [4]. Nevertheless, there are reports of sequence types ST11, ST15, ST25, and ST23 being isolated from livestock and companion animals [26][27][28][29][30][31]. These globally distributed STs are associated with the majority of nosocomial and community-acquired K. pneumoniae infections in humans [4]. In this study, we detected one K. pneumoniae ST11 and one ST147 isolate in pig and cattle, respectively. Since these STs are proposed to be human-associated, it is conceivable that they were transmitted from humans to animals [32]. Further human-associated or hypervirulent STs were not detected. However, our study collection contains a high genetic diversity, with 62 STs among 67 K. pneumoniae sensu stricto isolates. From this set, no predominant host-associated sequence type or reservoir could be determined. ST107 (four isolates, three with capsule type KL10) was the most prevalent ST among the 94 study isolates. K. pneumoniae of this ST have been occasionally isolated from human patients [33][34][35].
Antibiotic use in veterinary medicine is thought to increase the risk of antibioticresistant bacteria that can be transmitted to humans [36]. High prevalence rates of Escherichia coli with ESBL-mediated resistance to third-generation cephalosporins and MCR-1mediated colistin resistance have been reported from livestock and food in recent years [37][38][39]. As expected, the resistance to ampicillin was frequent among the study collection due to the presence of intrinsic beta-lactamase genes in Klebsiella spp. (bla SHV , bla OXY , bla OKP , bla LEN ). ESBL genes were detected in 16 of 94 isolates, but only five isolates with bla CTX-M genes showed resistance to third-generation cephalosporins (ST11, ST13, ST107, ST1017, ST1836). The remaining eleven isolates harbored bla SHV-27 and bla SHV-41 genes, but also a base-pair substitution in their promoter sequence, that has been described to be associated with a weak bla gene expression and non-ESBL phenotype [40]. Resistance to colistin was detected in only one (ST11, pig) of the 94 Klebsiella spp. isolates, but the underlying mechanism was not MCR-1-mediated, as described for E. coli from livestock [39]. Instead, an insertion in the mgrB gene was detected, which results in a truncated and most likely non-functional protein [41]. Changes in the mgrB gene have frequently been described as a cause of colistin resistance in K. pneumoniae [42].
Virulence genes were present in only 22 K. pneumoniae isolates. Interestingly, the iuc gene encoding the siderophore aerobactin was associated with K. pneumoniae isolates of pig and wild boar origin ( Figure 1, Table S1). Although these isolates differed in ST and capsule type, all harbored the iuc3 gene. The iuc lineage 3 is associated with self-transmissible IncFIB K and IncFII K virulence plasmids [43]. Since all 22 iuc3 isolates harbored plasmids of the IncFIB K type, we believe that domestic pigs could be a reservoir for K. pneumoniae carrying iuc3 on plasmids. However, this hypothesis requires further experimental proof using larger study collections.

Bacterial Isolates
A total of 90 Klebsiella spp. isolates were collected by the Chemical and Veterinary Analysis Agency Stuttgart between 2013 and 2017 as part of their routine diagnostic activities, one isolate was added from the Bundesinstitut für Risikobewertung (BfR) and three additional isolates from 2004 and 2006, respectively, were included as reference isolates. Further information about the isolates can be found on MALDI-UP [44].

Species Identification and Antibiotic Susceptibility Testing (AST)
Klebsiella species identification was conducted using a Bruker Microflex LT MALDI TOF MS after performing an extraction protocol in combination with the Biotyper database "MBT 8,468 MSP library" as recommended by the manufacturer. The species was assigned if the first two hits matched and showed a score >2.0. AST was performed using the broth microdilution assay according to CLSI guidelines. An interpretation of the minimal inhibitory concentrations was carried out using the European Committee on Antimicrobial Susceptibility Testing (EUCLAST) clinical breakpoints v10 derived for the human clinical context (http://www.eucast.org/clinical_breakpoints, accessed on 25 March 2021) (Supplementary Table S1).

Conclusions
This study demonstrates the genetic diversity of Klebsiella spp. isolates from animals and food products. Despite this diversity, we identified several well-known epidemic clones associated with nosocomial infections and multidrug resistance in humans, such as K. pneumoniae ST11 and ST147. The investigation of antibiotic resistance genes showed that the presence of the ESBL genes bla SHV-27 and bla SHV-41 did not result in resistance to third generation cephalosporins due to a mutation in their promoter sequences. The virulence factor aerobactin was the most prevalent siderophore locus in our collection, and our data suggest that pig isolates may act as a reservoir for the siderophore iuc lineage 3 in K. pneumoniae.