Cat fleas (Ctenocephalides felis clade ‘Sydney’) are dominant fleas on dogs and cats in New South Wales, Australia: Presence of flea-borne Rickettsia felis, Bartonella spp. but absence of Coxiella burnetii DNA

The cat flea (Ctenocephalides felis) is the most common flea species parasitising both domestic cats and dogs globally. Fleas are known vectors of zoonotic pathogens such as vector-borne Rickettsia spp. and Bartonella spp. and could theoretically transmit Coxiella burnetii, the causative agent of Q fever. A total of 107 fleas were collected from 21 cats and 14 dogs in veterinary clinics, a feline rescue organisation and a grooming salon in New South Wales, Australia, to undergo PCR detection of Bartonella spp., Rickettsia spp. and C. burnetii DNA. Morphological identification confirmed that the cat flea (C. felis) is the most common flea in New South Wales, Australia, with only a single stick fast flea, Echidnophaga gallinacea recorded. The examined fleas (n = 35) at the cox1 locus revealed five closely related C. felis haplotypes (inter-haplotype distance < 0.5%). Multiplex TaqMan qPCR targeting the gltA (Rickettsia spp.) and ssrA (Bartonella spp.) genes was positive in 22.9% (95% CI: 11.8–39.3%) and 11.4% (95% CI: 3.9–26.6%) of samples, respectively. None of the DNA isolated from fleas was positive on TaqMan qPCRs targeting the C. burnetii IS1111, Com1 and htpAB genes. Co-infection of C. felis with Bartonella henselae and Bartonella clarridgeiae was demonstrated using gltA and ssrA Illumina next-generation amplicon sequencing. These findings reinforce the importance of flea control on domestic dogs and cats to effectively control the transmission of Rickettsia felis and Bartonella spp. The flea, however, is unlikely to be a vector of C. burnetii between companion animals and humans.


Introduction
The role of companion animal fleas in the epidemiology of Bartonella henselaethe causative agent of cat scratch diseaseis well documented (Klotz et al., 2011). However, the exact role the flea plays in the transmission of other vector-borne zoonotic pathogens remains undefined (Rickettsia felis) or unknown (Coxiella burnetii) (Schriefer et al., 1994;Cutler et al., 2010;Klotz et al., 2011;Abdad et al., 2011;Toman et al., 2012;Njeru et al., 2016;Ng-Nguyen et al., 2020). Recently, companion dogs and cats were implicated in the transmission of both latter pathogens in Australia, including a Q fever outbreak (caused by C. burnetii) in a small animal veterinary hospital and a cluster of five probable cases of human infection with R. felis (Kopecny et al., 2013;Williams et al., 2011).
A member of the family Rickettsiaceae, R. felis causes flea-borne spotted fever (FBSF) (Abdad et al., 2011). In humans, clinical manifestations include signs of pyrexia, headaches, maculopapular rash, myalgias and eschar (Schriefer et al., 1994;Abdad et al., 2011). Over 30 arthropod species are recognised as potential vectors, but the cat flea (Ctenocephalides felis) is considered the main reservoir and vector. Transmission is believed to occur when an infected flea bites or contaminates open wounds with its faecal matter (Legendre and Macaluso, 2017). In Australia, C. felis is the dominant flea species on domestic cats and dogs ( Slapeta et al., 2011), hosting R. felis across the east coast at a prevalence of 6.7-19.8% (Barrs et al., 2010;Teoh et al., 2018).
The role of the cat flea (C. felis) in the transmission of C. burnetii is unknown. Q fever in humans is asymptomatic in approximately 60% of cases (Toman et al., 2012). Clinical disease manifests as an acute flu-like syndrome with non-specific signs of chills, malaise, sweating or fatigue (Raoult et al., 2000). Individuals with compromised cardiovascular function or endothelial cell defects can develop persistent focal diseases such as endocarditis or vascular infection (Raoult et al., 2000;Kampschreur et al., 2014). In addition, a debilitating chronic fatigue syndrome is experienced by at least 20% of acute Q fever patients (Morroy et al., 2016). In adults of most other mammalian species, coxiellosis is subclinical; however, abortions, infertility, stillbirth and weak progeny have occasionally been associated with the disease, especially in domestic ruminants, the primary source of human infections (Sanford et al., 1994;Eibach et al., 2012). Infection occurs primarily through inhalation of aerosolised infected materials. Arthropod transmission is theoretically possible, although an uncommon route for C. burnetii (Porter et al., 2011;Duron et al., 2015). Little is known about the role fleas, such as C. felis, play in C. burnetii transmission (Loftis et al., 2006;Psaroulaki et al., 2014;Spitalsk a et al., 2015;Kamani et al., 2015). No association has been found between the presence of fleas and seropositivity to C. burnetti in dogs and cats (Ma et al., 2020).
The cat flea (C. felis) in Australia is distributed along the coastal regions with three genetically distinct clades (Crkvencic and Slapeta, 2019). The clade 'Sydney' of C. felis is the only species previously detected in New South Wales and in cooler regions of southern Australia, while the clade 'Cairns' and 'Darwin' display preponderance in tropical northern Australia and may also be found further south due to climate change (Crkvencic and Slapeta, 2019). The capacity of these genetic clades to carry zoonotic pathogens and those that exist globally within C. felis is not yet known (Lawrence et al., 2019).
The aim of this study was to confirm the presence of C. felis on dogs and cats and evaluate their role in carriage of DNA of selected zoonotic vector-borne diseases. In an effort to shed more light on the vector contribution of fleas towards Q fever and FBSF, we applied qPCR diagnostic assays to detect C. burnetii, R. felis and Bartonella spp. in fleas collected from dogs and cats visiting veterinary clinics, a rescue organisation and a grooming salon.

Flea collection
From November 2019 to May 2020 fleas were collected from around New South Wales, Australia (Table 1). Fleas were collected opportunistically from dogs and cats by veterinarians, veterinary nurses, groomers and shelter carers as part of their routine work. Fleas were placed in 70% ethanol and stored at room temperature before being transferred into the freezer (À20 C).
2.2. Morphological and molecular identification using cytochrome c oxidase subunit 1 All fleas were examined individually under a microscope (5-200Â, Olympus, Australia) and placed to a genus and species as previously described (Dunnet and Nardon, 1974;Lawrence et al., 2019). A minimum of one flea of each flea species (if more than one species was present or if the species was equivocal on morphological examination) from each animal was selected for molecular characterisation. Thirty-five fleas were incised on the dorsal caudal abdomen using a sterile scalpel blade before being placed in a microcentrifuge tube (Eppendorf, Macquarie Park, Australia) which was then incubated in a heat block set at 70 C for 15 min to evaporate the ethanol in which they were stored (Lawrence et al., 2015). Flea DNA was extracted using the Isolate II Genomic DNA kit (BioLine, Eveleigh, Australia) according to the manufacturer's protocol with DNA eluted in 100 μl of elution buffer and stored at À20 C. For each batch of DNA extractions, an extraction with no flea was included and the eluate served as a non-template control (negative extraction control, NEC).
Extracted flea DNA samples were subjected to conventional polymerase chain reaction (PCR) targeting cytochrome c oxidase subunit 1 (cox1) as previously described (Lawrence et al., 2015(Lawrence et al., , 2019. Polymerase chain reaction amplification was performed in a 30 μl reaction mixture containing 15 μl MyTaq Red Mix (BioLine), with 2 μl DNA and nuclease-free water. Assays were performed in a T100 cycler (Bio-Rad, Australia) with an initial denaturation at 95 C for one minute followed by 35 cycles of 95 C for 15 s, 55 C for 15 s, 72 C for 10 s, and a final elongation for 5 min at 72 C. All reactions were run with a NEC and sterile PCR water in place of DNA (nontarget control, NTC). Amplicons of cox1 were sequenced (Macrogen Ltd, Seoul, Korea) and DNA sequences were assembled using CLC Main Workbench 21 (CLC bio, Qiagen, Chadstone, Australia) and compared to reference cox1 haplotypes (h1-h90) sensu Lawrence et al. (2019) and associated with the three clades ('Sydney', 'Cairns', 'Darwin') sensu Crkvencic and Slapeta (2019).

Molecular detection and identification of vector-borne pathogens
including Rickettsia spp., Bartonella spp. and C. burnetii An aliquot of flea DNA underwent multiplex TaqMan qPCR targeting the C. burnetii multicopy insertion sequence gene IS1111 (146-bp amplicon), and two single copy genes: com1 (76-bp amplicon) the outer membrane protein-coding gene, and groEL (114-bp amplicon; heat-shock operon; htpAB) (de Bruin et al., 2011;Lockhart et al., 2011;Bond et al., 2016). Each reaction contained 1Â SensiFAST Probe No-ROX Kit (Bio-Line, Australia), primers and probe (Table 1), 2 μl DNA and nuclease-free water in a total volume of 10 μl. Duplicate assays were performed using a Bio-Rad-CFX Real-Time PCR Thermocycler (Bio-Rad Laboratories Pty Ltd, Gladesville, NSW, Australia) with an initial denaturation at 95 C for 3 min followed by 40 cycles of denaturation at 95 C for 10 s and annealing at 60 C for 40 s. Each qPCR run included a NTC (with nuclease-free water in place of DNA), and positive control C. burnetii DNA of known copy number (Amplirun® Vircell, Granada, Spain) at 1100, 550 and 25 copies of the C. burnetii genome per reaction corresponding to a quantification threshold (C t ) of~34 for IS1111 and a C t of~36 for both com1 and groEL. Samples were considered positive if amplification occurred for all three genes at or below the C t for each gene. A TaqMan probe-based real-time PCR for Rickettsia and Bartonella species targeting the citrate synthase (~75-bp amplicon; gltA) gene and a transfer-messenger RNA (~300-bp amplicon; ssrA) respectively, was used to screen DNA from individual fleas for vector-borne pathogens (Stenos et al., 2005;Diaz et al., 2012). The reactions were performed in duplicate using the CFX96 Touch TM Real-Time PCR Detection System (BioRad, Australia) and contained 1Â SensiFAST Probe No-ROX Kit (BioLine, Australia), primers and probes, 2 μl template DNA in a total volume of 20 μl. Data were analysed using the corresponding CFX Maestro 1.0 software (BioRad, Australia) as previously described ( Slapeta and Slapeta, 2016). Each run included positive and negative controls (NTC, NEC). Results were considered positive if duplicates yielded C t values < 36. Suspect positive results were determined when one or more repeats yielded C t values ! 36 and samples were considered negative if neither repeat crossed the threshold (C t > 40). Positive Bartonella results were sent to Macrogen for sequencing (Macrogen Ltd, Seoul, South Korea). Samples considered either positive or suspect positive for Rickettsia spp. (C t value < 38) were further identified using a pair of conventional nested PCRs targeting the outer membrane protein A (ompA) gene and gltA (Stenos et al., 2005;Slapeta and Slapeta, 2016). PCR products were sequenced at Macrogen Inc. (Seoul, Korea) and assembled using CLC Main Workbench 21 (CLC bio, Qiagen, Australia).
Illumina overhang adapters were added to two sets of amplicon primers to amplify a~301-bp and~360-bp region of the ssrA and gltA genes of Bartonella, respectively (Power et al., 2021). Next-generation sequencing (NGS) of first stage PCRs were performed in-house according to the 16S Metagenomic Sequencing Library Preparation instructions for the Illumina MiSeq System (Illumina, Australia). The ssrA and gltA conventional PCR reactions were performed with Q5 Hot Start High-Fidelity 2X Master Mix (New England Biolabs, Australia). The protocol for each Bartonella sample varied according to their starting DNA concentrations. The PCR products were visualized on a 2% agarose gel with the aid of GelRed to ensure the presence of PCR products of the required size. The ssrA and gltA PCR amplicons were submitted for NGS at the Ramaciotti Centre for Genomics, University of New South Wales, Australia, for sample preparation and paired-end 250 bp Illumina sequencing on MiSeq system (Illumina, Australia). The DADA2 pipeline (package version 1.19.2; Callahan et al., 2016) was used to infer the exact Bartonella amplicon sequence variants (ASVs) from flea samples as previously described in Power et al. (2021). Local reference Bartonella databases were imported into DADA2 to classify the ASVs using two different classification methods; the 'assignSpecies' and 'assign-Taxonomy' functions. ASVs that were not assigned to a gene and Bartonella species were manually removed. The ssrA and gltA reads were then tabulated separately and the proportions (%) of each ASV in each sample were calculated for both genes. To address the potential issue of misassigned reads, all ASVs with < 1% in a sample were manually removed and replaced with '0'. ASVs were visually inspected, and their species assignment confirmed in CLC Main Workbench 21 (CLC bio, Qiagen, Australia) and aligned to sequences from our local curated Bartonella reference databases (Power et al., 2021).

Results
A total of 107 fleas were collected opportunistically from 32 dogs and cats (HH-1 to HH-32) in New South Wales, Australia including a cat shelter, grooming salon, and veterinary clinics (Table 2). Most fleas (93.5%, 102/107) were morphologically identified as the cat flea (C. felis). Three Ctenocephalides sp. fleas had equivocal morphological characteristics due to being damaged or having an ambiguous second notch on the hind tibia between the apical and post-median setae. One flea (HH-21-2) from a dog from north-west New South Wales, Australia, was identified as the stick fast flea (Echidnophaga gallinacea). Overall, there were 33 males, 72 females, one C. felis whose sex was not possible to determine and one male E. gallinacea.
Amplification and DNA sequencing of Rickettsia-positive and suspect samples with conventional nested PCR targeting gltA and ompA genes revealed five fleas (HH-6-1, HH-13-1, HH-14-1, HH-15-1, HH-23-1) that sequenced as identical (100%) DNA with R. felis reference gltA (Parola et al., 2003) and ompA (Ogata et al., 2005). HH-18-1 amplified only using ompA assay and its DNA sequence was identical with the R. felis reference. An additional sample (HH-16-1) that amplified only gltA yielded a sequence that was 99% identical to the R. felis reference DNA with a variation of one single nucleotide at the gltA gene (Table 4). Sanger DNA sequencing of the ssrA product failed to unambiguously resolve the Bartonella ssrA DNA sequence. We used conventional PCR with Illumina tagged Bartonella-specific ssrA and gltA primers to amplify these regions from DNA of seven Bartonella spp. positive and suspect positive samples. Amplification was successful for all four positive samples but was unsuccessful for the three suspect positive samples. The positive amplicons were subject to Illumina DNA sequencing and on average yielded 30,811 paired-end good quality sequences for gltA and 18,102 paired-end good quality sequences for ssrA. The HH-18-1 sample showed a mixed pattern because for both gltA and ssrA, it had sequence reads matching B. henselae (strain Houston-1) and Bartonella clarridgeiae (strain 73). At ssrA, 41% of the reads belonged to B. henselae and 59% to B. clarridgeiae. At gltA, 25% of the reads belonged to B. henselae and 75% to B. clarridgeiae. The remaining three samples had only B. henselae sequences (Table 4; Fig. 1).

Discussion
Arthropods are known vectors of zoonotic pathogens worldwide, the most common of which is the cat flea, C. felis (Clark et al., 2018;Lawrence et al., 2019), a well-documented carrier of Rickettsia and Bartonella species (Barrs et al., 2010;Slapeta and Slapeta, 2016). In this study the presence of Rickettsia and Bartonella species DNA and that of C. burnetii, the causative agent of Q fever, in fleas of cats and dogs were investigated in Greater Sydney and rural communities in New South Wales. All but one flea were identified to be C. felis, consistent with previous studies in Australian companion animals ( Slapeta et al., 2011;Lawrence et al., 2014). The only other flea species identified in this study was the stickfast flea (E. gallinacea), which is rarely found on dogs and cats, since it traditionally favours residence on avian species ( Slapeta et al., 2011). This study confirmed that only the clade 'Sydney' of C. felis is present in New South Wales ( Slapeta et al., 2011;Lawrence et al., 2015;Chandra et al., 2017;Crkvencic and Slapeta, 2019). In neighbouring New Zealand, only the clade 'Sydney' of C. felis has been documented and a study by Chandra et al. (2017) found similar percentages of fleas positive for Bartonella and Rickettsia using multiplexed TaqMan qPCR on 38 C. felis DNA samples, with 5.3% (n ¼ 2) positive for Bartonella and 18.4% (n ¼ 7) positive for Rickettsia DNA. The identification of co-infection of B. henselae and B. clarridgeiae in a single flea was demonstrated using recently developed multi-locus Illumina next-generation amplicon sequencing, demonstrating the advantages that this technology and   approach enable over conventional PCR alone (Power et al., 2021).
Since the discovery of R. felis in the 1990s, the organism has frequently been detected in C. felis (Adams et al., 1990;Legendre and Macaluso, 2017). Detection of R. felis DNA in arthropods ranges from 1% to 100% across various countries (Reif and Macaluso, 2009). Our results of DNA detection of R. felis in fleas is similar to the 6.7-36.0% reported in other Australian studies (Schloderer et al., 2006;Barrs et al., 2010;Teoh et al., 2018) and 15.5% (Teoh et al., 2018) reported in coastal regions of New South Wales including the Central Coast, Northern Beaches, and Sydney. The cat flea (C. felis) likely plays an important role in the transmission of R. felis to humans, acting both as a vector and reservoir (Barrs et al., 2010;Hirunkanokpun et al., 2011;Legendre and Macaluso, 2017). The focus in Australia has been on fleas from coastal cities, presumably due to the importance of optimal temperatures and humidity for flea development in these geographical areas (Silverman and Rust, 1983;Metzger and Rust, 1997). Rickettsia felis is able to successfully maintain its population through both horizontal (Hirunkanokpun et al., 2011;Brown et al., 2015) and vertical transmission (Wedincamp and Foil, 2002) which might explain why densely populated areas in Australia have the highest incidence of human R. felis infection cases recorded (Teoh et al., 2016;Hii et al., 2017). Therefore, flea control in domestic animals is required to effectively reduce the transmission of R. felis to humans, and is especially important in densely populated areas.
Coxiella burnetii was not detected in any of the fleas collected. In cats and dogs, C. burnetii seroprevalence varies regionally, being the highest in cattery confined breeding cats (Shapiro et al., 2015) and dogs from rural communities (Shapiro et al., 2016(Shapiro et al., , 2017. By comparison, seropositivity in shelter or household Australian dogs and cats in urban areas are relatively low (Shapiro et al., 2015(Shapiro et al., , 2016; however, seropositivity as high as 55.9% has been found in dogs in rural communities near a previous Q fever outbreak (Ma et al., 2020). Given that the majority of fleas in the present study were obtained from either owned or stray animals in metropolitan areas, the absence of positive C. burnetii DNA was not unexpected. Although transmission of C. burnetii is most often airborne (Porter et al., 2011), C. burnetii DNA has been detected in a multitude of tick species that parasitise dogs and cats (Cooper et al., 2013;Duron et al., 2015) and ticks are theoretically capable of transmitting the bacterium through secretion of contaminated faeces (Korner et al., 2020) or tick bites . This is the first study to investigate for evidence of C. burnetii DNA in fleas collected from cats and dogs in Australia. In agreement with our findings, C. burnetii DNA was not detected in fleas from Slovakia ( Spitalsk a et al., 2015), Israel (Kamani et al., 2015) and Egypt (Loftis et al., 2006). In addition, Ma et al. (2020) found no association between the presence of fleas and seropositivity to C. burnetii in dogs or cats in Australia. The highest detection rate of C. burnetii DNA in fleas, however, has been reported in a study conducted in Cyprus which used a qPCR assay targeting the C. burnetii IS1111 insertion sequence to detect C. burnetii in DNA extracts from pooled flea sets. This study reported an overall C. burnetii DNA detection rate of 16.3% (25 of the 153 flea pools tested were positive) and a species specific pooled prevalence of 38.1% in dog fleas (Ctenocephalides canis) from foxes, 16.6% in C. felis fleas from both foxes and rats and 10.7% in Xenopsylla cheopis fleas from rats (Psaroulaki et al., 2014). The IS1111 is a commonly used target in qPCR due to its enhanced sensitivity of detection (Klee et al., 2006). Until quite recently C. burnetii was the only described species of its genus; however, it is now recognised that Coxiella-like endosymbiont bacteria, which are closely related to but genetically distinct from C. burnetii, are present in soft and hard ticks (Wilkinson et al., 2014). IS1111 has been demonstrated to be widespread in these Coxiella-like bacteria and is therefore not specific to C. burnetii (Duron, 2015). It is possible that Coxiella-like DNA may also exist in other arthropods such as fleas, and thus the sole use of IS1111 as a diagnostic indicator may be misleading due to the potential false positive amplification of DNA from Coxiella-like endosymbionts. Furthermore, the existence of the IS1111 insertion sequence in all strains of C burnetii has been a subject of recent debate (Marmion et al., 2005;Rolain and Raoult, 2005). Although the qPCR in the present study targeted the IS1111 insertion sequence, the inclusion of the single copy genes com1 and groEL in the qPCR assay increased the chances of detecting strain variants of C. burnetii that may not contain the IS1111 insertion sequence and additionally minimised the likelihood of obtaining false positive results due to amplification of Coxiella-like DNA.

Conclusions
The cat flea (C. felis) was confirmed as the most common flea on dogs and cats in New South Wales, Australia, and there was an absence of cox1 clades other than clade 'Sydney'. While DNA of the zoonotic pathogens R. felis and Bartonella spp. was demonstrated in fleas, C. burnetii DNA was not detected in this investigation, consistent with previous studies. A combination of molecular tools to characterise both the arthropods and the potential zoonotic pathogens enabled us to detect co-infection of Bartonella spp. and R. felis. Illumina next-generation amplicon sequencing was also applied to demonstrate co-infection of B. henselae and B. clarridgeiae.  The data were processed using DADA2 and perfect match to reference genomic sequence of Bartonella spp. is indicated by '*'. At gltA, alternative ASVs were detected for both Bartonella spp. sequences that were 1-2 nucleotide different to its reference sequence.

Funding
editing.

Data availability
The nucleotide sequence data generated in this study were deposited in GenBank (NCBI) under the accession numbers MZ381608-MZ381642 and MZ420158-MZ420169. Raw fastq sequence data was deposited at SRA NCBI BioProject: PRJNA737164. Sequence data, associated supplementary and additional data are available at LabArchives (https ://doi.org/10.25833/2c6t-2276).

Declaration of competing interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.