Isolation of an antimicrobial‐resistant, biofilm‐forming, Klebsiella grimontii isolate from a reusable water bottle

Abstract A reusable water bottle was swabbed as part of the citizen science project “Swab and Send,” and a Klebsiella grimontii isolate was recovered on chromogenic agar and designated SS141. Whole‐genome sequencing of SS141 showed it has the potential to be a human pathogen as it contains the biosynthetic gene cluster for the potent cytotoxin, kleboxymycin, and genes for other virulence factors. The genome also contains the antibiotic‐resistant genes, bla OXY‐6‐4, and a variant of fosA, which is likely to explain the observed resistance to ampicillin, amoxicillin, and fosfomycin. We have also shown that SS141 forms biofilms on both polystyrene and polypropylene surfaces, providing a reasonable explanation for its ability to colonize a reusable water bottle. With the increasing use of reusable water bottles as an alternative to disposables and a strong forecast for growth in this industry over the next decade, this study highlights the need for cleanliness comparable to other reusable culinary items.

was an abundance of growth of presumptive Enterobacteriaceae.
Pure culture and sequencing showed it was likely to be an antibiotic-resistant, pathogenic strain of K. grimontii, which is able to form biofilms readily on abiotic surfaces.

| Isolation of SS141
A single sample was taken from the interior of the rim of a reusable water bottle using a Transwab ® Amies Charcoal Swab (MWE Medical Wire & Equipment) and kept at ambient temperature. The swab was used to directly inoculate CHROMagar™ Orientation (CHROMagar) chromogenic agar plate and incubated at 37°C for 18 hr. A single colony that was metallic green/blue in appearance was chosen and pure-cultured onto CHROMagar™ Orientation agar. The pure isolate was then kept in stocks at −80°C in 40% glycerol (Sigma, UK).
Acquired antimicrobial resistance genes located in the whole genome sequence or plasmids were searched for using ResFinder (version 3.2; Zankari et al., 2012) with an identity threshold of 60% and a minimum coverage of 60%, and metal resistance genes were

| Biofilm assays
The 96-well plate biofilm assay was performed as described previously (Hubbard, Jafari, Feasey, Rohn, & Roberts, 2019)  Gibco, Thermo Fisher Scientific), stained with 0.1% crystal violet (Sigma) followed by washing four times in PBS, and the stained biofilm dissolved in 30% acetic acid (Fisher Scientific). Finally, biofilm production was quantified at an optical density of 550 nm (OD 550 ) using a microplate reader.
Single tube biofilm assays were performed in 15-ml centrifuge tubes made from either polystyrene or polypropylene (Falcon ® ; Corning Life Sciences). Following incubation Luria Bertani broth, each culture was diluted 1/1,000 in 4 ml of M9 and incubated statically at 37°C or room temperature for 24 hr alongside a negative control of 4 ml M9 only. After incubation, each tube was washed four times with 6 ml of PBS and left to dry for 30-60 min and then stained with 5 ml 0.1% crystal violet solution for 15 min. The crystal violet stain was washed off with four washes of 6 ml PBS and left to dry for 60-120 min and photographed.

| Minimum inhibitory concentrations
Minimum inhibitory concentration (MIC) of AMP, AMX, AMC, CEF, CIP, OLA, CHL, and FOF to SS141 were carried out using cationadjusted Mueller Hinton Broth following the CLSI guidelines for antimicrobial susceptibility testing using the broth microdilution methods (CLSI, 2018).

| RE SULTS AND D ISCUSS I ON
The swab taken from a reusable water bottle was initially plated out on to the chromogenic CHROMagar™ Orientation agar routinely used for the identification of urinary tract pathogens (Samra, Heifetz, Talmor, Bain, & Bahar, 1998), which identified a significant amount of presumptive Enterobacteriaceae present on the swab. Due to the metallic blue/green pigmentation and morphology of the colonies on the chromogenic agar, it was determined that these colonies were probable Citrobacter, Enterobacter, or Klebsiella, which are often pathogenic. We therefore subcultured until it was a pure isolate, denoted as isolate SS141.
Following a hybrid assembly of the whole genome sequencing data, the isolate was identified as K. grimontii. It contains two low copy number plasmids: IncFII(K) (101 kb in size and 4.11× depth) and IncFIA(HI1) (134 kb in size and 1.98× depth) and a small extrachromosomal molecule, which was unable to be fully resolved and with extremely high copy number (between 1,348.61 and 1,357.92× depth). Antimicrobial resistance genes were not found to be present on any of these plasmids; however, putative heavy metal-resistant genes (silE, cusA, pcoR, and copA/B/C/D potentially conferring resistance to silver and copper) were found to be present on the IncFIA(HI1) plasmid and arsR, implicated in resistance to arsenic, was found on the chromosome of SS141. While K. grimontii can be asymptomatically carried by humans, it has also been associated with the cause of a number of infections including antibiotic-associated hemorrhagic colitis and bacteremia (Passet & Brisse, 2018).
Using PathogenFinder, we found SS141 is likely to be a human pathogen with a probability score of 0.842. Subsequently, we compared the SS141 genome to a clinical strain isolated from a patient's sputum, K. grimontii WCHKG020121 (Liu, Feng, et al., 2018) using ANI (Figure 1). We found that SS141 was very closely related to the clinical isolate K. grimontii WCHKG020121 (99.61%) and reference strain K. grimontii 06D021 (99.06%), further confirming that SS141 is K. grimontii. SS141 was also 99.12% identical to K. oxytoca JKo3 suggesting that K. oxytoca JKo3 is likely to have been misidentified and is actually K. grimontii (Figure 1). All these strains had a  Figure 1).
Four antimicrobial resistance genes were identified to be present on the SS141 chromosome using ResFinder, bla OXY-6-4 , which confers resistance to β-lactam antibiotics, including amoxicillin and ticarcillin (Fevre et al., 2005), and is an indicator of K. grimontii (Liu, Feng, et al., 2018;Passet & Brisse, 2018). Both oqxA and oqxB, which are involved in resistance to olaquindox (Hansen, Johannesen, Burmolle, Sorensen, & Sorensen, 2004), an antimicrobial used as a growth promoter in animals (Barber, Braude, Hosking, & Mitchell, 1979), and a variant of the fosfomycin-resistant gene fosA were all also identified and were all previously found to be present on the chromosome of several other K. grimontii isolates (Liu, Feng, et al., 2018). Resistance to the β-lactam antimicrobials AMP and AMX was confirmed (MIC of 16 and 32 µg/ml, respectively); however, SS141 was still sensitive to both AMC (2 µg/ml) and CEF (0.0625 µg/ml) according to EUCAST clinical breakpoints (Table 1). Resistance to fosfomycin was also confirmed with a MIC of 256 µg/ml, which is well over the clinical breakpoint of >32 µg/ml. Despite the identification of oqxAB genes in SS141 using ResFinder, the isolate was sensitive to OLA (16 µg/ ml) according to the previously determined breakpoint of >64 µg/ ml (Hansen et al., 2004;Liu, Wu, et al., 2018; Table 1). While oqxB2 was confirmed to be present in the genome following annotation using Prokka, we noticed that the oqxA gene identified by ResFinder was annotated by Prokka as bepF gene, which encodes for the protein BepF, part of the RND-type efflux pump BepFG, and has been shown to be involved in drug resistance to some degree in Brucella suis (Martin et al., 2009). Therefore, lack of confirmation of the presence of oqxA in the genome would explain the lack of resistance seen to olaquindox in the phenotypic assays. The oqxAB genes have also been reported to be present in the K. grimontii WCHKG020121 genome (Liu, Feng, et al., 2018), which align with 100% identity to  Figures A1 and A2 gastrointestinal tract (Wu, Lin, Hsieh, Yang, & Wang, 2011). We have confirmed that SS141 is a biofilm former, producing a visible and significant biofilm during growth in the minimal media M9 at 37°C (Ordinary one-way ANOVA, uncorrected Fisher's LSD pvalue = <.0001) and room temperature (Ordinary one-way ANOVA, uncorrected Fisher's LSD p-value = <.0001) compared with Escherichia coli 10129, a known biofilm producer (Figures 2 and 3a, ;Hubbard et al., 2019). As biofilm formation on a polystyrene 96-well plate and centrifuge tubes do not represent the same plastic commonly used in reuseable water bottles, we also tested the biofilm production potential of SS141 in centrifuge tubes and 96-well plates made from polypropylene, a significant component of reuseable water bottles.
We found that SS141 also produces significant and visible biofilms on  Tse et al., 2017). This cytotoxin has been previously associated with the cause of antibiotic-associated hemorrhagic colitis (Tse et al., 2017), further suggesting that SS141 is a pathogenic strain of K. grimontii.

| CON CLUS IONS
We describe the first instance of isolation of a K. grimontii isolate from a reusable water bottle. The isolate was found to form biofilms on polypropylene and polystyrene and carried multiple resistance genes to both metals and antibiotics. It is also likely to be pathogenic as the biosynthetic gene cluster for kleboxymycin is present within the genome. There is a strong forecast in the global market for reusable water bottles over the next decade (Accuray- Research-LLP, 2019), and this study highlights the aspect of cleanliness when it comes to repeated use of such household items. There is, as yet, no comprehensive scientific study of bacterial colonization of reusable water bottles, and perhaps this is needed to persuade manufacturers and users to promote and practice suitable washing regimens aimed at keeping bacterial load to a minimum.

ACK N OWLED G M ENTS
We would like to thank MWE, Medical Wire & Equipment, for their continued support of the Swab and Send citizen science project.

CO N FLI C T O F I NTE R E S T
None declared.