Antibiotic Sensitivity Screening of Klebsiella spp. and Raoultella spp. Isolated from Marine Bivalve Molluscs Reveal Presence of CTX-M-Producing K. pneumoniae

Klebsiella spp. are a major cause of both nosocomial and community acquired infections, with K. pneumoniae being responsible for most human infections. Although Klebsiella spp. are present in a variety of environments, their distribution in the sea and the associated antibiotic resistance is largely unknown. In order to examine prevalence of K. pneumoniae and related species in the marine environment, we sampled 476 batches of marine bivalve molluscs collected along the Norwegian coast. From these samples, K. pneumoniae (n = 78), K. oxytoca (n = 41), K. variicola (n = 33), K. aerogenes (n = 1), Raoultella ornithinolytica (n = 38) and R. planticola (n = 13) were isolated. The number of positive samples increased with higher levels of faecal contamination. We found low prevalence of acquired resistance in all isolates, with seven K. pneumoniae isolates showing resistance to more than one antibiotic class. The complete genome sequence of cefotaxime-resistant K. pneumoniae sensu stricto isolate 2016-1400 was obtained using Oxford Nanopore and Illumina MiSeq based sequencing. The 2016-1400 genome had two contigs, one chromosome of 5,088,943 bp and one plasmid of 191,744 bp and belonged to ST1035. The β-lactamase genes blaCTX-M-3 and blaTEM-1, as well as the heavy metal resistance genes pco, ars and sil were carried on a plasmid highly similar to one found in K. pneumoniae strain C17KP0055 from South-Korea recovered from a blood stream infection. The present study demonstrates that K. pneumoniae are prevalent in the coastal marine environment and that bivalve molluscs may act as a potential reservoir of extended spectrum β-lactamase (ESBL)-producing K. pneumoniae that may be transmitted through the food chain.


Introduction
The genus Klebsiella contains several species known to cause nosocomial infections [1,2] and some that cause community acquired infections [3,4]. Klebsiella spp. are widely distributed outside the

Sampling
A total of 204 batch samples of bivalve molluscs were collected along the Norwegian coast from September 2019 to March 2020. An additional 272 samples collected in 2016 were included in the study. The samples comprised 384 blue mussels (Mytilus edulis), 48 oysters (Crassostrea gigas), 24 scallops (Pecten maximus), five horse mussels (Modiolus modiolus), three ocean quahogs (Arctica islandica), two cockles (Cerastoderma edule), two carpet shells (Politapes rhomboides) and one sand gaper (Mya arenaria). Even though not a bivalve mollusc, seven sea urchins (Strongylocentrotus droebachiensis) were also included. In total, 476 samples covering 77 different production areas and five non-rearing locations along the Norwegian coast was included in the study. Samples from production areas were collected through the surveillance programme on bivalves conducted by the Norwegian Food Safety Authority (NFSA). A detailed overview of samples and sampling locations is provided in Supplementary Table S1.

Sample Preparation
Each batch sample comprised 10-20 individual bivalves. Live and closed bivalves were cleaned under cold tap water before they were opened using a sterile knife. Approximately 80-100 g soft tissue and intra-valvular fluid was weighed into sterile plastic bags (VWR, Radnor, PA, USA) and homogenised for 2.5 min using a stomacher (Seward, UK).

Isolation and Identification of Presumptive Klebsiella spp.
Aliquots of 25 g were transferred to new sterile plastic bags and diluted 1:10 in Buffered Peptone Water (BPW) (VWR, USA), homogenised for 30 s and incubated aerobically at 37 • C for 18-24 h. After incubation, 10 µL of the enrichment cultures were streaked on Simmons Citrate Agar (Bio-Rad, Hercules, CA, USA) supplemented with 1% Myo-Inositol (Sigma-Aldrich, St. Louis, MO, USA) (SCAI), a highly selective media for the isolation of Klebsiella spp. and Raoultella spp. [29] and incubated aerobically at 37 • C for 48 h. Samples collected in 2016 had been enriched in BPW by the same protocol and stored at −80 • C in 20% glycerol. Before the samples were analysed, they were thawed in room temperature and approx. 1.5 mL transferred to 10 mL BPW and incubated at 37 • C over night. Yellow colonies representing presumptive Klebsiella spp. were sub-cultured to obtain pure cultures. The obtained isolates were cultured overnight on Plate Count Agar (PCA) (Oxoid, UK) at 37 • C. Colonies were transferred directly to disposable 96 spot targets (Bruker, Germany) and covered with 1 µL HCCA matrix (Bruker, Germany). The spots were air dried and the isolates were identified using Matrix Assisted Laser Desorption Time of Flight Mass Spectrometry (MALDI-TOF MS) (Bruker, Germany).

PCR Amplification of bla SHV
To confirm the absence of bla SHV K. pneumoniae isolate 2016-1400 was subjected to PCR amplification of the bla SHV gene. DNA was extracted using the DNeasy blood and tissue kit (Qiagen, Hilden, Germany). Each 20 µL reaction contained 4 µL 5X Phusion HF buffer, 0.4 µL 10 mM dNTP mix, 0.2 µL of each 50 µM SHV specific primer (SHVF: 5 -ATGCGTTATATTCGCCTGTG-3 , SHVR: 5 -TGCTTTGTTATTCGGGCCAA-3 ) [42], 0.2 µL Phusion DNA polymerase, 1 µL template DNA and 14 µL nuclease free water. PCR amplification was performed using the GeneAmp PCR system 9700 Thermal Cycler (Applied Biosystems, Waltham, MA, USA) and the following conditions: initial denaturation at 98 • C for 30 s, 30 cycles of 98 • C for 5 s, 62 • C for 5 s and 72 • C for 10 s, with a final extension at 72 • C for 3 min. K. pneumoniae CCUG 10,785 was included as a positive control and a bla SHV -negative E. coli as a negative control. The PCR products were resolved on a 1% agarose gel stained with Gel Red Nucleic Acid Stain (Biotium, Fremont, CA, USA) and visualised on a Bio-Rad ChemiDoc system (Bio-Rad, USA).

Filter Conjugation
Conjugation was carried out according to the method described by Jutkina et al., 2016 [43]. Briefly, kanamycin (KAN) and rifampicin (RIF) resistant gfp marked E. coli recipient was grown in Mueller Hinton broth (MHB) (Oxoid, UK) with 50 µg/mL KAN (Sigma-Aldrich, St. Louis, MO, USA) at 30 • C with shaking overnight. The CTX and AMP resistant donor was grown over night in MHB supplemented with 2 µg/mL CTX (Sigma-Aldrich, St. Louis, MO, USA) under the same incubation conditions. The donor and recipient were centrifuged at 2755× g for 15 min. and washed twice in phosphate buffered saline (PBS) (Sigma-Aldrich, St. Louis, MO, USA) before final resuspension in PBS. The conjugation mixtures were prepared by mixing equal aliquots of donor and recipients (1:1 ratio). The conjugation mixture was pipetted on to 0.45 µm filters (Merck Millipore, Burlington, MA, USA) and placed on Mueller Hinton (MH) (Oxoid, UK) agar plates and incubated at 37 • C overnight. After incubation, the filter was removed and placed in a falcon tube with 10 mL PBS and sterile glass beads and the cells were removed from the filter by vortexing at maximum speed for 90 s. Serial dilutions up to 10 −6 was prepared in PBS and 100 µL spread in duplicates on CHROMagar orientation (CHROMagar, Paris, France) plates supplemented with 50 µg/mL KAN, 50 g/mL RIF and 100 µg/mL AMP (Sigma-Aldrich, St. Louis, MO, USA) and 50 µg/mL KAN, 50 µg/mL RIF and 2 µg/mL CTX. The plates were incubated at 37 • C for 24-30 h.

Distribution of Klebsiella spp. and Raoultella spp. in Marine Bivalves
From the 476 samples, presumptive Klebsiella spp. were detected in 41% (n = 194) of the samples, with some samples positive for several morphotypes. A total of 204 isolates were obtained and identified as members of the genera Klebsiella and Raoultella using MALDI-TOF MS. In total, 78 isolates were identified as K. pneumoniae, 41 as K. oxytoca, 33 K. variicola, one K. aerogenes, 38 as R. ornithinolytica and 13 isolates were identified as R. planticola (Table 1). The frequency of samples positive for Klebsiella spp. and/or Raoultella spp. increased with higher levels of faecal contamination as expressed by the number of E. coli detected, with a total of 24% (n = 51), 48% (n = 96) and 81% (n = 56) of the samples positive from < LOQ (n = 213) areas, class A (n = 194) areas and class B (n = 69) areas, respectively. The most frequently isolated species from class A and B areas was K. pneumoniae and K. oxytoca from locations where E. coli MPN/100 g was < LOQ (Figure 1). Detailed overview of isolates is provided in Supplementary Table S3.

Antimicrobial Susceptibility Patterns of Klebsiella spp. and Raoultella spp.
Among the K. pneumoniae isolates, resistance to more than one agent was seen in only eight isolates. Three MDR isolates were detected (Supplementary Table S4), while one isolate displayed phenotypic resistance to cefuroxime, cefotaxime and ampicillin, as well as intermediate susceptibility to aztreonam.

Genome Sequencing
The sequenced genome of K. pneumoniae sensu stricto isolate 2016-1400 was de novo assembled into two contigs, one 5,088,943 bp chromosome and one 191,744 bp plasmid. The isolate belonged to ST1035 and had the wzi allele 116, corresponding to capsule locus (KL) type 57. Further analysis of the sequenced genome revealed that the isolate carried bla CTX-M-3 , bla TEM-1 , oqxA, oqxB, fosA and erm(D) conferring resistance to erythromycin but lacked the bla SHV gene. This was confirmed by PCR analysis. A comparison of 2016-1400 with SHV-1-harbouring K. pneumoniae ST1035 genomes (ENA run accession number ERR4859177 and ERR3416161) showed that there had likely been a deletion of a 10.1 Kbp region, which included bla SHV-1 , due to the insertion of an IS5 family transposase between a hypothetical protein and diguanylate phosphodiesterase in our isolate.
Several heavy metal resistance genes, including the plasmid borne copper resistance system(pco) gene cluster, the arsenic resistance genes (ars) and the sil operon genes conferring resistance to silver were identified. Virulence genes involved in iron acquisition (ent and fep), adherence (ecp), magnesium uptake (mgt) and immune evasion (ompA) were also detected in 2016-1400 (Table 3).   bla CTX-M-3 and bla TEM-1 , as well as the heavy metal resistance genes were carried on a IncFIB(K)/IncFII plasmid highly similar to the190 582 bp non-conjugstiveCP052387.1 IncFIB(K)/IncFII plasmid (100% sequence coverage and 99.96% nucleotide identity) from a clinical K. pneumoniae strain (Figure 2). The plasmid was not transferable by filter conjugation and carried few conjugal transfer genes. These plasmids carried all of the same genes, except an IS630-like element ISSpu2 family transposase, which was carried by 2016-1400 only. Compared to CP052387.1, this insertion occurred between two base pairs in a pseudogene, an incomplete hypothetical protein. The 5,083,236 bp chromosome of the same published genome (CP052386.1) was also similar to 2016-1400, with 99.62% sequence coverage and 99.43% nucleotide identity.

Discussion
Here we present a comprehensive study on the prevalence of K. pneumoniae and related species in marine bivalves collected from both areas used for commercial production of bivalve molluscs for human consumption along the Norwegian coast as well as non-rearing locations in Western and Southern Norway. We further show low prevalence of acquired antibiotic resistance in these isolates. However, one K. pneumoniae isolate carried a clinically important and potentially mobile ESBL gene.
K. pneumoniae sensu stricto isolate 2016-1400 recovered from M. edulis at a production area in Middle-Norway displayed phenotypic resistance to cefotaxime and carried a plasmid encoding the bla CTX-M-3 gene, first described in clinical E. coli and Citrobacter freundii isolates in Poland in 1996 [44]. The CTX-M genes originated from Kluyvera spp. and were spread through mobilisation from their chromosomal position [14,45] and since they have disseminated worldwide [14]. Cephalosporins are among the most commonly used classes of antibiotics worldwide [46] and ESBL-producing K. pneumoniae represents a threat to the public health as it limits the available therapeutic options [47]. The plasmid also carried a bla TEM-1 gene, a common plasmid-borne resistance gene among clinical Gram-negative bacteria, primarily conferring resistance to penicillins and first-generation cephalosporins [48]. No gene encoding the chromosomal bla SHV was identified in K. pneumoniae 2016-1400. Interestingly, no chromosomal bla SHV was detected in the chromosome (CP052386.1) of K. pneumoniae strain C17KP0055 from South-Korea, which is highly similar to the strain in our study. This is rare but absence of SHV has previously been reported in some K. pneumoniae strains [49,50].
A recent study on antibiotic resistance in E. coli from marine bivalves collected in Norway identified bla CTX-M genes (bla CTX-M-15 and bla CTX-M-14 ) in only 1% of the isolates [22]. This is in line with our findings, indicating a low prevalence of CTX-M-producing Enterobacterales in the Norwegian marine environment. Although the occurrence of ESBL-producing Enterobacterales in Norway is low compared to many other countries, the prevalence of clinical ESBL-producing Klebsiella spp. isolates is increasing [15]. Hence, the presence of ESBL-producing Enterobacterales in reared bivalves intended for human consumption is concerning.
Furthermore, the K. pneumoniae 2016-1400 plasmid carried genes conferring resistance to silver, copper and arsenic. The plasmid was highly similar to the plasmid of a K. pneumoniae strain (C17KP0055) isolated from the blood of a South-Korean patient, encoding the indistinguishable ARGs and heavy metal resistance genes. Co-localisation of heavy metal resistance genes and ARGs on the same mobile genetic element (MGE) increases the chance of co-selection by heavy metals [51]. High copper and arsenic levels have been detected in some marine environments in Norway [52] and this may contribute to the co-selection of this plasmid/strain. K. pneumoniae ST1035 has been associated with human infections [53][54][55][56]. Furthermore, capsule type (K) 57 has been identified in clinical isolates and associated with pyogenic liver abscess (PLA) [57,58]. However, isolate 2016-1400 did not harbour any of the genes associated with hypervirulent K. pneumoniae (hvKp) or the capsule types (K1 and K2) which are mainly associated with hvKP. Analysis of K. pneumoniae strain C17KP0055 showed that this strain belonged to the same ST and did not harbour any of the hvKP associated virulence genes.
One MDR K. pneumoniae isolate (2019-1764) was recovered from a location close to the city centre of Bergen. The location is likely to be heavily influenced by anthropogenic activities, indicating that this isolate is of human or animal origin. This could pose a threat to the public health as the sampling location is in close proximity to bathing areas [59].
The overall results from our study are in accordance with previous studies on AMR in bacteria from the Norwegian marine environment, where low prevalence of acquired resistance was found in E. coli and Vibrio spp. [22,60,61]. In the present study, acquired resistance was only found in few K. pneumoniae isolates and in none of the other isolated species. These results reflect the restrictive use of antibiotics in Norway, both in clinics, food production and companion animals. In Norwegian clinical K. pneumoniae isolates, resistance to amoxicillin-clavulanic acid, trimethoprim, trimethoprim-sulfamethoxazole, cefuroxime and mecillinam are the most commonly observed resistance phenotypes [15]. Resistance to these agents, except mecillinam, was also found among the marine K. pneumoniae isolates. No carbapenemase-producing isolates were recovered. This is in contrast to the increase of CR-KP in some European countries [13] as well as several reports on the occurrence of CR-KP in the environment [59,[62][63][64]. Although all of the isolated species are considered to be intrinsically resistant to aminopenicillins due to the presence of chromosomal class A β-lactamases [5,65], we found some ampicillin-susceptible isolates among all species, except K. aerogenes and R. planticola. This has previously been described in Klebsiella spp. isolates [66][67][68][69] and is in K. pneumoniae likely caused by differential expression [70] or lack of the bla SHV gene.
K. pneumoniae was the most common species isolated from our samples. This is in accordance with a study by Podschun et al. [28] on Klebsiella spp. in surface waters from fresh, brackish and seawater in temperate regions. The presence of K. oxytoca seems to be less dependent on faecal contamination than K. pneumoniae. As K. oxytoca is known to be present in terrestrial environments [71], there are several transmission routes for this species to the marine environment. The relative distribution of the isolated genera in our study is in contradiction to the current description of the genus Raoultella as an environmental genus known to be associated with aquatic environments [6,65,72]. Even from samples with E. coli MPN/100 g below the LOQ, Klebsiella was the more frequently isolated compared to Raoultella.
This study provides enhanced understanding about the prevalence and AMR of K. pneumoniae found in the Norwegian marine environment. Although low prevalence of acquired resistance was detected, our study demonstrates presence of ESBL-producing K. pneumoniae in an area used for production of marine bivalves for human consumption. This indicates that bivalve molluscs may act as a potential reservoir of ESBL-producing K. pneumoniae for transmission to the community through the food chain. Our study also highlights the importance of the marine environment in dissemination of opportunistic human pathogens and ARGs.