Novel strains of Klebsiella africana and Klebsiella pneumoniae in Australian Fruit Bats (Pteropus poliocephalus)

Over the past decade human associated multidrug resistant (MDR) and hypervirulent Klebsiella pneumoniae lineages have been increasingly detected in wildlife. This study investigated the occurrence of K. pneumoniae species complex (KpSC) in grey-headed flying foxes (GHFF), an Australian fruit bat. Thirty-nine KpSC isolates were cultured from 275 GHFF faecal samples (14.2%), comprising K. pneumoniae (sensu stricto) (n=30), Klebsiella africana (n=8) and Klebsiella variicola subsp. variicola (n=1). The majority (79.5%) of isolates belonged to novel sequence types (ST), including two novel K. africana STs. This is the first report of K. africana outside of Africa and in a non-human host. A minority (15.4%) of GHFF KpSC isolates shared STs with human clinical K. pneumoniae strains, of which, none belonged to MDR clonal lineages that cause frequent nosocomial outbreaks, and no isolates were characterised as hypervirulent. The occurrence of KpSC isolates carrying acquired antimicrobial resistance genes in GHFF was low (1.1%), with three K. pneumoniae isolates harbouring both fluoroquinolone and trimethoprim resistance genes. This study indicates that GHFF are not reservoirs for MDR and hypervirulent KpSC strains, but they do carry novel K. africana lineages. The health risks associated with KpSC carriage by GHFF are deemed low for the public and GHFF.

The majority (~85%) of KpSC isolates associated with clinical disease belong to K. pneumoniae (sensu stricto) [23], however, K. variicola subsp.variicola and K. quasipneumoniae are classified as emerging human pathogens [27,28].In contrast, the recently described K. quasivariicola has .CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.It is The copyright holder for this preprint this version posted July 12, 2021.; https://doi.org/10.1101/2021.07.12.451971 doi: bioRxiv preprint been reported in only a few clinical cases, and K. africana in a single human clinical case [24].
The hypervirulent phenotype is associated with the presence of specific pathogenicity and virulence factors, including capsule types K1 and K2, serotypes O1 and O2, capsule upregulator genes rmpA and rmpA2, and siderophores yersiniabactin, aerobactin, salmochelin and colibactin [30].Whilst MDR and hypervirulent K. pneumoniae are usually distinct strains, there are recent reports of convergence, with isolates exhibiting both MDR and hypervirulence traits [23].
Klebsiella is intrinsically resistant to ampicillin and amoxicillin due to the presence of core chromosomal beta-lactamase (bla) producing genes [6], designated blaSHV in K. pneumoniae, blaLEN in K. variicola and K. quasivariicola, and blaOKP in other KpSC [6,24,31,32].Acquired AMR in K. pneumoniae is typically associated with acquisition of resistance genes via horizontal transfer of mobile genetic elements, such as plasmids and transposons [33] and integrons [34].
One type of integron, the clinical class 1 integron, has been significant for the emergence and ongoing dissemination of AMR in Gram-negative bacteria including K. pneumoniae [35].Class 1 integrons capture and express diverse antibiotic resistance genes (ARGs) via the integrase gene (intl1) and a promoter (Pc), and are typically characterised by a 3'-conserved segment (qacE∆1-sul1) [34].K. pneumoniae is proposed to play an important role in the acquisition of AMR genes from environmental bacteria and horizontal transmission via mobile genetic elements to other Enterobacteriaceae [35,36].Notably, the emergence and dissemination of plasmid-borne blaSHV .CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.It is The copyright holder for this preprint this version posted July 12, 2021.; https://doi.org/10.1101/2021.07.12.451971 doi: bioRxiv preprint genes associated with ESBL activity [37,38] and carbapenemase genes (blaNDM-1, blaKPC, blaOXA-48) [36], are now widely found in diverse Enterobacteriaceae.
The One Health approach to AMR highlights the need for ecological studies of K. pneumoniae in diverse animal species to identify reservoirs of clinically important lineages [23].Surveillance of wildlife for pathogenic and antimicrobial resistant bacteria is particularly important for species inhabiting urban environments, such as bats, due to the zoonotic risks posed to humans in close proximity.
Class 1 integrons harbouring ARGs to narrow-spectrum penicillins, trimethoprim and aminoglycosides were detected in faecal DNA from grey-headed flying fox (GHFF; Pteropus poliocephalus) a large fruit bat endemic to Australia [22], however, the bacterial hosts of these integrons have not been resolved.This study aimed to examine K. pneumoniae ecology in GHFF and to define the diversity, pathogenicity and virulence factor occurrence, and AMR gene carriage in isolates from GHFF.

Sample collection
GHFF have established large colonies, comprising several thousand to >50,000 individuals, in both urban and rural environments [40,41].The distribution of GHFF spans eastern Australia, from Adelaide (South Australia) to Ingham (Queensland).Faecal samples (n=275) were collected between 2017 and 2018 from free-living wild GHFF (n=255) inhabiting urban environments and captive GHFF (n=20) that were undergoing rehabilitation.GHFF faecal samples were collected from four wild colonies; Camellia Gardens (CG), (n=50), Centennial Park (CP) (n=52) and Blackalls Park (BP) (n=101) in New South Wales (NSW), and Adelaide Botanic Park (ABP) (n=52) in South Australia (SA).Sampled captive animals were from the ABP colony that were recovering in a rehabilitation facility after heat stress (Fauna Rescue of South Australia), Mylor, SA (n=20).Faecal samples were collected either opportunistically from plastic drop sheets placed under roosting flying foxes or directly from individual GHFF via a rectal swab for the ABP colony.The FecalSwab TM (COPAN, Brescia, Italy) system was used for both sampling methods.
Samples were stored at 4˚C and cultured within 72 h of collection.

Ethics
All sample collections were conducted under approvals from animal ethics committees at Macquarie University (No. 2017/013) and The University of Adelaide (No. S-2015-028), NSW Government Scientific Licence (No. SL101898) and SA Department of Environment and Water Wildlife Scientific Permit (M-23671-1,2 and 3).
or Serratia sp. were selected for further analyses.To distinguish Klebsiella sp. from Serratia sp., isolates were evaluated for gelatin metabolism by inoculation into 5 mL of nutrient gelatin (NG) containing 13 g/L nutrient broth (Oxoid, Hamphire, United Kingdom) and 120 g/L commercial bovine gelatine (Dr.Oetker, Bielefeld, Germany).The NG samples were incubated for 48-96 h and tested after 48 h, 72 h and 96 h for a positive gelatin metabolism reaction by transferring to 4˚C for 30 min.Isolates that were negative for gelatin metabolism (solid at 4˚C) after 96 h were deemed presumptive Klebsiella sp. or Enterobacter sp. with those positive for gelatin metabolism (liquid at 4˚C) eliminated from further analyses.Bacterial cell lysate (95˚C 15 min; >10000 g 10 min) was used as the DNA template in KpSC MLST and class 1 integron PCRs.

Preliminary identification of KpSC
Preliminary identification of KpSC isolates was performed by MLST PCRs targeting three of seven house-keeping genes (gapA, mdh and phoE) using primer sets listed in Table S1 [25].DNA from K. pneumoniae strain DAR_Y9835_6192 was used as a positive control in all PCRs.PCRs were performed using GoTaq® Colorless Mastermix (Promega, Madison, USA), 0.4 µM of each primer and 2 µL DNA.The PCR conditions were 94°C for 2 min, then 35 cycles of 94°C for 20 s and a primer specific annealing temperature for 30 s (Table S1) and extension at 72˚C for 30 s, with a final extension at 72°C for 5 min [43].Isolates returning a positive result for all three housekeeping genes when analysed by agarose gel electrophoresis were presumed to be K. pneumoniae and underwent whole genome sequencing and class 1 integron screening.

Whole genome sequencing and genetic characterisation
DNA was extracted using the GenFind v2 kit (Beckman Coulter) and libraries prepared using the NexteraFlex kit (Illumina Inc.) as per the manufacturers' guidelines.Libraries were sequenced on the Illumina NextSeq500 platform (150bp paired-end reads, n=52 presumptive KpSC).Following preliminary analysis, a single K. africana isolate (FF1003) was selected for additional long read sequencing using the Oxford Nanopore MinION R9 device (Oxford Nanopore Technologies) as described previously [44].Illumina reads (of 150 bp) were quality trimmed using Cutadapt v1.16 [45] and de novo assembled using SPAdes v3.1.12[46] optimised with Unicycler v0.4.7 [47].The FF1003 genome was fully resolved using Unicycler's hybrid assembly approach.Illumina reads were also mapped to the K. pneumoniae SGH10 reference chromosome (GenBank accession: CP025080) and single nucleotide variants called using the RedDog v1b 10.3 pipeline (https://github.com/katholt/RedDog). Genomes were excluded from further analysis if any of the following quality control criteria were met; i) total assembly size <5Mbp or >6Mbp; ii) mean read depth <20x (calculated from mapping stats); iii) mapping coverage <85%; iv) ratio of heterozygous to homozygous variant calls >0.4 (cut-off determined empirically); v) species not part of the KpSC (see below).

Phenotypic antibiotic resistance testing
Isolates harbouring acquired AMR genes (n=3) (as identified by WGS) and representative isolates of all other KpSC STs (n=12) underwent phenotypic antibiotic susceptibility testing for ESBL activity and for resistance to quinolones/fluoroquinolones, trimethoprim +/-sulfamethoxazole and aminoglycosides (Table S4).Antibiotic susceptibility testing was performed according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) disk diffusion method [54] and isolates were evaluated as susceptible or resistant using EUCAST breakpoint criteria (v 9.0 available at http://www.eucast.org/clinical_breakpoints/)(Table S4).EUCAST breakpoints were unavailable for nalidixic acid and susceptibility was instead determined using the Clinical and Laboratory Standards Institute (CSLI) breakpoint criteria (CLSI M100 ED29:2019 available at https://clsi.org/standards/products/free-resources/access-our-free-resources/)(Table S4).

Occurrence of KpSC in GHFF
Of 50 presumptive K. pneumoniae isolates identified by MLST, WGS confirmed 39 genomes passed quality control and belonged to the KpSC, with the majority belonging to K. pneumoniae (n=30) and the remaining to K. africana (n=8) and K. variicola subsp.variicola (n=1) (Fig. 1).
The KpSC isolates were detected in three of four wild GHFF colonies, with frequencies ranging from 0.0% to 36.0%(overall mean 9.0%) (Fig. 1).For the captive GHFF, K. pneumoniae was detected in 80% of faecal samples from the single wildlife rehabilitation facility in this study (Fig. 1).
Eighteen isolates from the CG colony represented eight STs; K. pneumoniae n=6 STs and K. africana n=2 STs (Fig. 2).Four KpSC isolates were cultured from the BP colony, with two being distinct K. pneumoniae STs, one K. africana and one K. variicola, and one K. pneumoniae isolate was cultured from the ABP colony (Fig. 2).Although the captive Mylor GHFF showed a high occurrence of K. pneumoniae, 15 of 16 isolates were ST5035 and showed little genetic diversity (Fig. 2).Simpson diversity in the wild GHFF was 0.88 overall, which ranged from 0.84 in the CG colony to 1.0 in the BP colony, however, the captive GHFF group showed almost complete uniformity (Simpson diversity = 0.13).
STs were typically clustered by location, with the two notable exceptions being K. africana ST4938 which was detected in GHFF from the CG (n=6) and BP (n=1) colonies, and K. pneumoniae ST4919 in the ABP (n=1) and CG (n=1) colonies (Fig. 2).

Acquired antimicrobial resistance genes
Only one of 13 GHFF KpSC STs (ST1017) was associated with acquired ARGs, which equated to an overall occurrence of 1.1% across all sampled GHFF (wild 1.2% and captive 0.0%).ST1017 was cultured from three faecal samples from wild GHFF, with all three isolates carrying two AMR genes, dfrA14 and qnrS1, which confer resistance to trimethoprim and fluoroquinolones respectively (Fig. 2).Phenotypic resistance to trimethoprim and ciprofloxacin, and intermediate resistance to nalidixic acid, was confirmed for all three ST1017 isolates by disk diffusion antibiotic susceptibility testing (Table S5).
Class 1 integron screening of GHFF KpSC isolates detected IntI1 in all three ST1017 isolates however, amplification of the gene cassette array was not successful.BLASTn searches for these genes in the relevant genome assembly graphs indicated that qnrS1, intI1 and dfrA14 were coharboured on a single replicon, which formed a closed circularised path through the graph which was subsequently shown to carry an IncN plasmid replicon marker, suggesting this structure represented an antimicrobial resistance plasmid.DfrA14 was associated with IntI1, however, the class 1 integron was truncated and the 3'-conserved segment replaced by mobC and an IS6100 transposase.A BLASTn search found the ST1017 IntI1-drfA14-mobC-IS6100 integron sequence was an identical match to >100 K. pneumoniae associated plasmids in GenBank, including pathogenic strains (accessions CP052219, CP052525 and MN218814).The qnrS1 gene was associated with an IS26 transposase, with both inserted into a tra operon.
No KpSC STs (0/13) harboured acquired AMR genes conferring ESBL or carbapenemase activity, with all tested isolates susceptible to amoxicillin plus clavulanic acid, cephalexin, cefotaxime and imipenem in disk diffusion antibiotic susceptibility testing (Table S5).

Phylogenetic comparison of human and GHFF K. africana isolates
The GHFF K. africana isolates were compared to the only two previously reported K. africana isolates; a MDR clinical isolate (ERR315152, ST2831) causing a disseminated infection in Kenya and a human faecal isolate (ERR2835900, ST3291) from Senegal [24] (Fig. 3).The human isolates were separated by 20,512 chromosomal single nucleotide variants (SNVs) and both were distantly related to the GHFF isolates (≥16,249 pairwise chromosomal SNVs).However, seven of the eight GHFF isolates were clustered together with little genetic diversity evident between them (≤53 SNVs) (Fig. 3).Interestingly, these included isolates from two different sites (Camelia Gardens, n=6 and Blackhall's Park, n=1).The eight GHFF isolates differed from the rest by ~23,300 SNVs.
All K. africana isolates carried intrinsic blaOKP-C-1 genes which exhibited variations in their nucleotide sequences, however, the translated amino acid sequences were identical for all isolates.

Discussion
The detection of two novel K. africana STs in GHFF effectively doubled the number of known lineages globally and increased the collective K. africana pool from two isolates to 10 isolates.
This was a highly unexpected finding given that K. africana has previously only been isolated from humans in Africa [24].The K. africana ST4938 bat isolates detected at two colonies geographically separated by 100km, exhibited low genetic diversity (0-53 SNVs).GHFF typically forage within 50 km from the colony location, however they can also travel longer distances between colonies over several days [15,16].These findings suggest a common source for the ST4938 by GHFF in the CG and BP colonies and possible horizontal microbial transmission between individuals roosting in close proximity [56].The K. africana isolates were only detected in GHFF from a small geographical region and over a short temporal sampling period of four weeks, so it remains unclear if K. africana are part of the GHFF endemic microbiome.Diverse KpSC have been isolated from numerous environmental niches including plants, soil, insects and the aquatic environment [36].Given the arboreal nature of wild GHFF, environmental reservoirs of K. africana such as fresh water or food resources, including native fruits and flowers, orchard fruits and insects, present possible sources [57,58].Wider sampling and high-resolution analysis of microbiome composition would assist in elucidating the frequency and ecology of K. africana in GHFF microbiomes.
In wild GHFF, the occurrence of KpSC isolates harbouring acquired ARGs was 1.2%, which was lower than the previously reported occurrence of class 1 integrons in wild GHFF (5.3%) [22].
Resistance to one critically important antimicrobial category, the fluoroquinolones, was detected in the GHFF K. pneumoniae isolates, whereas ESBL and carbapenemase activity was absent [3,4].These findings are comparable with an earlier study of AMR in Enterobacteriaceae from wild Australian mammals, which also found an absence of ESBL and carbapenemase producing K.
pneumoniae and a low occurrence of resistance to other clinically important antimicrobials (range 0.0% to 10.1%) [59] and were not considered MDR [61].Two GHFF strains shared STs with carbapenemase producing strains reported to cause localised outbreaks ST105 [62] and ST661 [63], however neither GHFF isolate harboured any resistance genes.These findings suggest that although GHFF and humans share K. pneumoniae STs, the isolates detected in GHFF did not harbour MDR or virulence factors that are frequently associated with isolates causing MDR human nosocomial and community acquired infections.
Phylogenetic analysis suggests K. africana and K. quasivariicola are the closest relatives to the clinically important K. pneumoniae (sensu stricto) taxa [24].The GHFF K. africana isolates exhibited characteristics associated with K. pneumoniae clinical infections, including the O2 serotype [55] and yersiniabactin [23].Additionally, the human clinical K. africana isolate ST2831 carried diverse AMR genes, including the ESBL blaCTX-M-15 gene.These findings indicate K.
africana strains are capable of acquiring virulence factors and AMR genes, and have the potential to be opportunistic pathogens of GHFF and humans [55].
The considerable variation of KpSC occurrence in GHFF from different locations may relate to differences in exposure to environmental sources of KpSC, such as water and food resources [36], and frequency of transmission between individuals, which is influenced by colony structure and density [56].In addition to the apparent local clonal expansion of ST5035 in the captive Mylor GHFF, the clustering of three STs in the CG colony also suggests local clonal expansion, which may have contributed to the elevated occurrence of KpSC at CG compared to other wild colonies.
Despite the detection of K. pneumoniae strain ST5035 in 75% of captive GHFF, it was notably absent in the parent colony (ABP), suggesting acquisition from a common source whilst in captivity, in combination with a high frequency of transmission between individuals.The captive GHFF were group-housed in an outdoor aviary with exposure to multiple potential sources of K. pneumoniae ST5035, including animals (poultry, dogs, captive kangaroos and other wildlife) and food (cultivated fruits) [36].The stark contrast in the occurrence of K. pneumoniae between wild and captive GHFF, demonstrates widespread transmission of a potential bacterial pathogen in group-housed captive wildlife.Widespread acquisition (36.4% occurrence) of a class 1 integron harbouring the resistance gene aadA2 has also been demonstrated in group-housed captive GHFF [22].This potential for widespread dissemination of bacterial pathogens and AMR genes in grouphoused captive GHFF should be considered as a potential health risk for both human carers and GHFF in rehabilitation programs.
Studies examining bacterial pathogens and AMR in wildlife typically focus on the potential for wildlife to act as bacterial reservoirs [64,65] and largely ignore the potential impacts of anthropogenic pathogens colonising wildlife (reverse zoonosis) [66].Each year, several thousand sick and injured GHFF require veterinary care [67], which may include antimicrobial therapy.
The detection of fluoroquinolone resistant K. pneumoniae isolates capable of causing opportunistic infections in humans, and possibly in GHFF, is concerning, especially as enrofloxacin (a veterinary fluoroquinolone) is frequently prescribed to GHFF in care.
Antimicrobial administration to GHFF carrying resistant isolates may result in a poor treatment response and prognosis for recovery, but also may increase dissemination to other GHFF in care, to human carers and into the environment post release.Additionally, increasing numbers of GHFF require veterinary care each year due to the cumulative impacts of wild fires, heat stress events, habitat loss and food shortages [67,68].

Fig. 1 .
Fig. 1.The occurrence and distribution of K. pneumoniae species complex isolates in GHFF from
[10]cently, diverse strains of MDR K. pneumoniae, Escherichia coli and ot her Enterobacteriaceae harbouring plasmid-borne carbapenem resistance genes were detected in 40% of Australian silver gulls (Chroicocephalus novaehollandiae) at one site, demonstrating widespread transmission of ARGs in Australian wildlife[10].In captive GHFF, no KpSC isolates harbouring acquired ARGs were found, which is in contrast to previous reports of a high occurrence of class 1 integrons in captive GHFF (41.2%)[22].This absence of ARGs in KpSC isolates from captive GHFF group-housed in the same rehabilitation facility, may be explained by the widespread carriage (75% occurrence) of a single KpSC strain, K. pneumoniae ST5035, which did not harbour class 1 integrons or AMR genes.Of the four K. pneumoniae STs shared by humans and wild GHFF, three have been reported as