Article

The prevalence of Bartonella, hemoplasma, and Rickettsia felis in domestic cats and in cat in Ontario Ali Kamrani, Valeria R. Parreira, Janice Greenwood, John F. Prescott

Abstract The prevalence of persistent bacteremic Bartonella spp. and hemoplasma infections was determined in healthy pet cats in Ontario. Blood samples from healthy cats sent to a diagnostic laboratory for routine health assessment over the course of 1 y were tested for Bartonella spp. using both polymerase chain reaction (PCR) and blood culture, and for the presence of hemoplasma by PCR. The overall prevalence of Bartonella spp. by PCR and by culture combined was 4.3% (28/646) [3.7% (24/646) Bartonella henselae, 0.6% (4/646) Bartonella clarridgeiae]. The novel B. henselae PCR developed for this study demonstrated nearly twice the sensitivity of bacterial isolation. The overall prevalence of hemoplasma was 4% (30/742) [3.3% (25/742) Candidatus haemominutum, 0.7% (5/742) Mycoplasma haemofelis]. There was no significant difference between the prevalence of by season or by age (# 2 y, . 2 y). Candidatus Mycoplasma turicensis was identified, for the first time in Canada, in 1 cat. The prevalence of Bartonella (58%) and hemoplasma (47% M. haemofelis, 13% M. haemominutum) in blood from a small sampling (n = 45) of stray cats was considerably higher than that found in healthy pet cats. The prevalence of Rickettsia felis in cat fleas was also assessed. A pool of fleas from each of 50 -infested cats was analyzed for the presence of R. felis by PCR. Rickettsia felis was confirmed, for the first time in Canada, in 9 of the 50 samples. Therefore, the prevalence of Bartonella and hemoplasma infection in healthy pet cats is relatively low. Further, the control of cat fleas is important because of the public health significance of Bartonella and R. felis infection.

Résumé La prévalence d’une bactériémie persistante à Bartonella spp. et d’une infection à hémoplasma a été déterminée dans une population de chats en santé provenant de l’Ontario. Des échantillons sanguins de chat en santé envoyés à un laboratoire de diagnostic pour un bilan de santé de routine durant une période d’une année ont été éprouvés pour Bartonella spp. par réaction d’amplification en chaîne par la polymérase (PCR) et par hémoculture, et pour la présence d’hémoplasma par PCR. La prévalence globale de Bartonella spp. par PCR et culture combinés était de 4,3 % (28/646) [3,7 % (24/646 Bartonella henselae, 0,6 % (4/646) B. clarridgeiae]. L’épreuve PCR développée pour la présente étude avait une sensibilité presque le double de l’isolement bactérien. La prévalence globale d’hémoplasma était de 4 % (30/742) [3,3 % (25/742) Candidatus Mycoplasma haemominutum, 0,7 % (5/742) M. haemofelis]. Il n’y avait pas de différence significative entre les prévalences d’infection selon la saison ou le groupe d’âge (# 2 ans, . 2 ans). Candidatus Mycoplasma turicensis a été identifié chez un chat pour la première fois au Canada. La prévalence de Bartonella (58 %) et d’hémoplasma (47 % M. haemofelis, 13 % M. haemominutum) dans le sang d’un échantillonnage restreint (n = 45) de chats errants était considérablement plus élevée que chez des chats en santé. La prévalence de Rickettsia felis chez des puces de chats a également été évaluée. Un pool de puces provenant de chacun de 50 chats infestés par des puces a été analysé par PCR pour la présence de R. felis. La présence de R. felis a été confirmée pour la première fois au Canada dans 9 des 50 échantillons. Ainsi, la prévalence d’infection par Bartonella et hémoplasma chez des chats en santé est relativement faible. De plus, la maîtrise des puces de chat est importante compte tenu de l’impact sérieux en santé publique des infections par Bartonella et R. felis. (Traduit par Docteur Serge Messier)

Introduction B. henselae, B. clarridgeiae, and B. kohlerae (2–4), all of which may cause various clinical manifestations in humans (4,5). Bartonella henselae is Healthy cats may be subclinical carriers of blood-borne bacterial the major causative agent of cat scratch (CSD) in immuno- infections that are potentially serious zoonotic pathogens and which competent people and of bacillary angiomatosis and hepatic peliosis may cause severe in other cats. The genus Bartonella is in patients with a defective , especially those infected comprised of fastidious hemotropic that have been increas- with human immunodeficiency virus (HIV) (5,6). As a result of the ingly identified in a wide range of domestic and wild animals (1). development of sensitive molecular techniques for the diagnosis of Domestic cats are the primary reservoir of 3 species of Bartonella, Bartonella organisms in human patient tissue samples, and because

Department of Pathobiology, and Centre for Public Health and Zoonoses, University of Guelph, Guelph, Ontario N1G 2W1 (Kamrani, Parreira, Prescott); Vita-Tech Laboratories, 1345 Denison Street, Markham, Ontario L3R 5V2 (Greenwood). Address all correspondence to Dr. J.F. Prescott; telephone: (519) 824-4120, ext 54453; fax: (519) 824-5930; e-mail: [email protected] Received September 6, 2007. Accepted January 8, 2008.

2008;72:411–419 The Canadian Journal of Veterinary Research 411 of the increasing numbers of patients with defective immune sys- kit (QIAamp DNA Blood Mini Kit; Qiagen, Mississauga, Ontario). tems through HIV infection or the use of immunosuppressive drugs, Purified DNA was eluted in 150 mL elution buffer and stored at Bartonella spp. have emerged as agents of zoonotic infections (7). This 220°C. Pooled flea samples were prepared for DNA extraction by study is the first to use both blood culture and polymerase chain first rinsing in sterile distilled water for 5 min at room temperature. reaction (PCR) to investigate the prevalence of Bartonella bacteremia Each sample was macerated completely in a 1.5 mL microcentrifuge in both pet and stray domestic cats in Ontario. tube using a sterile pipette tip. The DNA was extracted, according to Feline hemobartonellosis is a bacteremic infection in cats that may the manufacturer’s instructions, using the QIAamp DNA Blood Mini be associated with severe hemolytic . The etiologic agents of Kit (Qiagen), eluted in 100 mL elution buffer and stored at 220°C. To this infection, Mycoplasma haemofelis and Mycoplasma haemominutum, monitor the extraction process for any foreign DNA contamination, may cause severe , and subtle anemia or subclinical a negative “DNA extraction” control, consisting of 200 mL sterile infection, respectively, in cats (8). Candidatus Mycoplasma turicensis phosphate buffered saline (pH 7.2, PBS), was included in each batch is a recently recognized hemotropic mycoplasma, first identified of samples processed. in 2005 in Switzerland in a cat with hemolytic anemia (9). This Various primer combinations, each targeting the 16S–23S rRNA study is the first to investigate the prevalence of the hemoplasmas, intergenic region of the genus Bartonella, were assessed for their including Candidatus Mycoplasma turicensis, in pet and stray cats amplification efficiency of plasmid DNA containing the target in Ontario. gene of B. henselae (ATCC 4988). Oligonucleotide sequences of each Rickettsia felis is a recently recognized agent of flea-borne spotted primer are listed in Table I. The following primer combinations fever in humans (10–12). The cat flea is the major vector of this organ- were examined: primer pair JF 311/JR 473, as described by Jensen ism. In endemic areas of the United States, peri-urban are et al (15); primer pair MF 321/MR 983, as described by Maggi the major reservoir host of R. felis. The reservoir potential of cats has and Breitschwerdt (16); and, a novel combination of the primers yet to be determined, although experimental infection of cats with MF 321/JR 473. The PCR amplifications were performed in a 25 mL R. felis has been demonstrated (13,14). One objective of this study reaction mix containing a ready-to-use PCR buffer that included a was to determine if R. felis is present in cat fleas in Canada. proof-reading hot start DNA polymerase (Platinum PCR SuperMix High Fidelity; Invitrogen, Burlington, Ontario), 0.3 pmol/mL of each primer, and 2 mL DNA template (0.02 ng, plasmid DNA containing Materials and methods B. henselae target gene). Use of proof-reading DNA polymerase on Bartonella DNA when samples were extracted from blood, removed Blood samples some nonspecific amplification observed with Taq polymerase (data Blood from healthy domestic cats in southern Ontario was col- not shown). Amplification conditions used in this assay consisted of lected in tubes containing ehtylenediamine tetra-acetic acid (EDTA) denaturation at 94°C for 2 min, 45 cycles of 30 s of denaturation at anticoagulant, and submitted to a laboratory (Vita-Tech Laboratories, 94°C, 30 s of annealing at 58°C, 30 s of extension at 68°C, followed Markham, Ontario) as part of a routine “wellness” program for by an additional cycle of 7 min extension at 68°C. The predicted size cats. Care was taken to exclude samples from animals that were of the amplicons resulting from this PCR reaction were calculated either suspect or clinically ill. A convenience sampling method was to be 152 bp, 136 bp, 233 bp, 242 bp, and 187 bp for B. henselae, used, whereby approximately 45 cats . 2 years old, and 14–45 cats, B. clarridgeiae, B. elizabethae, B. vinsonii subsp. berkhoffii, and B. bovis, # 2 years old, were collected each month. A total of 742 samples respectively. The PCR amplification products were visualized by were collected over a period of 12 mo (April 2006 to March 2007). ethidium bromide fluorescence after electrophoresis in a 3% agarose Blood samples were also collected from 45 stray cats, found within gel. The primer pair MF 321/JR 473 was used in subsequent PCR Wellington County, Ontario. Blood was stored at 270°C prior to amplification of DNA isolated from feline blood and pooled flea testing. samples. A single-step PCR assay first described by Jensen et al (17) was Cat fleas used for the detection of the hemoplasmas. Oligonucleotide primer Fleas were collected in the summer of 2006 from cats submit- sequences are listed in Table I. Since Candidatus M. turicensis has ted for care to 3 local Humane Societies located in Guelph (n = 4), the same amplicon size as M. haemofelis using this amplification Kitchener-Waterloo (n = 7), and Cambridge (n = 3). In addition, protocol, Candidatus M. turicensis-specific primers (Table I) were fleas were collected from cats within the Wellington County area used in a second PCR protocol on all M. haemofelis PCR-positive (n = 36), and submitted to the Ontario Veterinary College for cases identified. This second PCR amplifies a 1317 bp region of the teaching and research purposes. One flea sample consisted of 16S rRNA gene of Candidatus M. turicensis. For both PCR reactions 1–5 fleas pooled from each flea-infested cat. Fleas were iden- for hemoplasma detection, 2 mL of DNA template was added to tified as Ctenocephalides felis based on current taxonomic keys make a total 25 mL reaction mixture containing 0.8 pmol/mL of each (http://www.icb.usp.br/~marcelcp/Ctenocephalidesfelis.htm) and primer and a ready-to-use PCR mixture containing a proof-reading stored in 70% ethanol prior to testing. hot-start DNA polymerase (Platinum PCR SuperMix High Fidelity, Invitrogen). A touch-down PCR program was used: 95°C for 5 min, DNA extraction and PCR amplification 35 cycles of 95°C for 30 s, 60°C to 50°C for 30 s/cycle (the annealing The DNA for PCR amplification was extracted from a 200 mL temperature is decreased by 1°C every cycle until 50°C, after which aliquot of each EDTA-treated blood sample using a commercial each cycle was for 30 s), 68°C for 2 min, and finally, 68°C for 10 min.

412 The Canadian Journal of Veterinary Research 2000;64:0–00 Table I. Sequence of oligonucleotide primers used for PCR assays Primer (target gene) Nucleotide sequence 59-39 Organism detected Reference JF 311 CCTTCGTTTCTCTTTCTTC Bartonella spp. (15) (16S–23S rRNA) JR 473 AACCAACTGAGCTACAAGCC Bartonella spp. (15) (16S–23S rRNA) MF 321 AGATGATGATCCCAAGCCTTCTGG Bartonella spp. (16) (16S–23S rRNA) MR 983 TGTTCTAACAACAATGATGATG Bartonella spp. (16) (16S–23S rRNA) HF (16S rRNA) ACGAAAGTCTGATGGAGCAATA Feline hemoplasma organisms (17) HR (16S rRNA) ACGCCCAATAAATCCGRATAAT Feline hemoplasma organisms (17) MTF GAACTGTCCAAAAGGCAGTTAGC M. turicensis (9) (16S rRNA) MTR AGAAGTTTCATTCTTGACACAATTGAA M. turicensis (9) (16S rRNA) CSF (gltA) GGGGGCCTGCTCACGGCGG Rickettsia spp. (19) CSR (gltA) ATTGCAAAAAGTACAGTGAACA Rickettsia spp. (19) OMF (rOmpB) CCGCAGGGT TTGTAACTGC Rickettsia spp. (19) OMR (rOmpB) CCTTTTAGATTACCGCCTAA Rickettsia spp. (19)

The PCR amplification products were visualized by ethidium bro- Services Division, University of Guelph, Guelph, Ontario) using mide fluorescence following electrophoresis in a 2% agarose gel. an ABI Prism 3100 Genetic Analyzer, and the Big Dye Terminator The DNA extracts from pooled fleas and blood from stray cats Reaction Kit (Version 3) (Applied Biosystems) with either uni- were amplified targeting 2 different genes of R. felis in 2 separate versal or specific primers. Sequencing was performed directly on reactions. The first pair of primers amplified a 381 bp sequence of amplicons resulting from PCR amplification using primers specific the citrate synthase (gltA) gene of the genus Rickettsia (Table I). The for B. clarridgeiae, R. felis, or Candidatus M. turicensis. The PCR second pair of primers amplified a 840 bp sequence of an outer mem- products resulting from DNA amplification specific for B. henselae brane protein B gene (rOmpB) of R. felis (Table I). For both reaction were cloned into pGEM-T Easy vector (Promega; Fisher Scientific, mixes, 5 mL of DNA template was added to make a total 25 mL vol- Nepean, Ontario) and sequenced. The obtained sequences were ume of PCR reaction mix containing 13 PCR buffer (New England compared to known sequences held in the GenBank database using Biolabs, Pickering, Ontario), 0.2 mM dNTPs (Roche Diagnostics the nucleotide-nucleotide BLAST function. GmbH, Mannheim, Germany) 2.5 mM Mg21, 1 pmol/mL of each primer and 1 U of Taq polymerase (New England Biolabs). The Blood culture for Bartonella spp. PCR was carried out with an initial denaturation at 94°C for 3 min Initially, 2 culture methods were used for isolation of Bartonella. followed by 40 cycles of denaturation at 94°C for 30 s, annealing at The first was direct culture of 200 mL of each blood sample on Brain 50°C for 30 s, and elongation at 72°C for 30 s. The amplification was Heart Infusion Agar (BHIA) (Difco, Detroit, Michigan, USA) with completed with an additional elongation at 72°C for 7 min. The PCR 5% chocolatized sheep blood (“chocolate agar”). Agar plates were products were electrophoresed through 2% agarose gel and visual- incubated at 35°C in a 5% CO2 water-saturated atmosphere for ized, after ethidium bromide labelling, under UV light. 6 wk. The DNA was extracted from suspect colonies (small, slow All PCR assays were carried out in a Model T Gradient ther- growing) and sequences specific to Bartonella spp. were targeted mocycler (Biometra, Göttingen, Germany). To avoid PCR ampli- as described previously. The other method used was a novel liquid fication product carryover, DNA extraction, PCR reaction set-up, medium, Bartonella-Alphaproteobacteria growth medium (BAPGM) and gel electrophoresis were each carried out in a different room. (18). A 200 mL aliquot of blood was inoculated into a tube containing Appropriate negative reagent controls and positive DNA plasmid BAPGM. Each tube was cotton plugged and incubated for 12 d at controls, containing the target gene of each relevant organism, were 35°C in a 5% CO2, water-saturated, atmosphere. After incubation, included in each PCR assay. 200 mL of the blood-inoculated medium was used for DNA extrac- tion and amplification. In addition, 200 mL of the blood-inoculated DNA sequencing ­liquid medium that had been incubated for 12 d was plated on BHIA To confirm the identity of PCR-amplified products, selected ampli- with 5% chocolatized sheep blood. The plates were then incubated cons were sequenced (Guelph Molecular Supercentre, Laboratory for another 12 d at 35°C, as described. The DNA was isolated from

2000;64:0–00 The Canadian Journal of Veterinary Research 413 for 12 d at 35°C. The DNA was extracted from the BAPGM and tested using the Bartonella spp. specific PCR. Aliquots of the inoculated BAPGM media were subcultured onto chocolate agar for a further 12 d; this study was done in triplicate. Statistical analysis The association of categorical variables of age categories and seasons of the year with each infection were tested by a chi-squared or Fisher’s exact test. The level of significance was considered at P , 0.05. All of the statistical analyses were carried out using sta- tistical software (MINITAB, version 15.1; Minitab, State College, Pennsylvania, USA). Results

PCR assay for Bartonella detection Various primer combinations, each targeting the 16S–23S rRNA intergenic region of the genus Bartonella, were assessed for their efficiency of amplification of plasmid DNA containing the target gene of B. henselae. Based on band intensity, the novel combination of the forward primer MF 321 and the reverse primer JR 473 (Table I) (9,15–17,19) gave better amplification than the original combination of MF 321 and MR 983 or JF 302 and JR 473 suggested by Maggi and Breitschwerdt (16) and by Jensen et al (15), respectively (Figure 1), and was therefore used throughout the study. The PCR was opti- mized for DNA template concentration and annealing temperature (data not shown). Results of PCR on blood from healthy cats A total of 646 and 742 blood samples, from July 2006 to March 2007, were tested by PCR for the presence of Bartonella and hemoplasma, respectively. The overall results of specific PCR on blood samples Figure 1. Comparison of the efficiency of the combination of primers are shown in Table II, while the monthly results for PCR are given in JF-JR and MF-JR. Lane A — amplification of a 170 bp PCR product of Table III. The prevalence of infection with Bartonella and hemoplasma 16S–23S rRNA intergenic spacer region of B. henselae by primers JF-JR. Lane B — amplification of a 152 bp PCR product of 16S–23S rRNA inter- in different seasons is illustrated in Figure 2. There was no associa- genic space of B. henselae by primers MF-JR. The same concentration of tion between the presence of Bartonella spp. and of hemoplasma in the cloned plasmid has been used as a DNA template for both reactions. healthy cats (Fisher exact probability test: P = 0.42). L: 50 bp. Agarose gel 3%. No association was found between age category, # 2 and . 2 years old, for either infection (P = 0.064 for Fisher’s exact test for Bartonella suspect colonies (small, slow growing) and tested in the PCR assay infection and P-value = 0.457 for Fisher’s exact test for hemoplasma for confirmation as Bartonella spp. infection). No association was found between each infection and the season of sampling (P = 0.36 for chi-squared test for Bartonella infec- Comparison of the sensitivity of blood culture and tion and P = 0.33 for chi-squared test for hemoplasma infection). PCR for Bartonella detection After applying Candidatus M. turicensis-specific PCR to all The sensitivity of PCR and of blood culture on solid or in liquid M. haemofelis PCR-positive samples, 1 cat from the domestic pet media for detection of Bartonella spp. in blood was compared in the cat population yielded a PCR product consistent with the expected following way. Colonies of an isolate of B. henselae made from one Candidatus M. turicensis amplicon of approximately 1300 bp. The of the cats in the study were suspended in 700 mL sterile PBS. After sequenced PCR product was 99% identical to, and had the high- homogenization of the suspension, 10-fold serial dilutions were est homology with, the 16S rRNA gene sequence of Candidatus made in PBS until 1029; 200 mL of each dilution was used for DNA M. turicensis in GenBank (Accession number DQ464422.1). extraction and PCR amplification using primers specific for Bartonella spp. In addition, 200 mL of each dilution was plated on chocolate Comparison of the sensitivity of culture and PCR agar. The number of colony-forming units for each dilution was Comparison of the sensitivity of detection of Bartonella, which determined following 12 d incubation at 35°C. Furthermore, 200 mL had been serially diluted in saline, showed that PCR was up to of each dilution was inoculated directly into BAPGM and incubated 10 times more sensitive than culture on chocolate agar, and that PCR

414 The Canadian Journal of Veterinary Research 2000;64:0–00 Table II. Results of specific PCR on DNA extracted from the blood of pet and stray domestic cats, as well as fleas isolated from stray cats Sample source Organism amplified Pet cat Stray cat Fleas B. henselae 3.2% (21/646) 57.7% (26/45) 90% (45/50) B. clarridgeiae 0.6% (4/646) 0 12% (6/50) M. haemofelis 0.6% (5/742) 46.6% (21/45) 10% (5/50) M. haemominutum 3.3% (25/742) 13% (6/45) 56% (28/50) R. felis N/A 0 18% (9/50)

Table III. Results of specific PCR and direct culture of blood in chocolate agar for the detection of Bartonella spp. in pet cats in Ontario (April 2006 to March 2007) Method, age Number of positive samples identified per month category Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Total positive Culture 0/16 0/11 0/20 0/41 1/37 2/45 1/44 0/42 1/27 0/45 2/45 0/25 7/398 (1.8%) B. henselae # 2 years old PCR N/A N/A N/A 2/43 2/37 5/45 N/A 2/43 1/28 0/45 2/45 1/25 15/311 (4.8%) B. henselae # 2 years old PCR N/A N/A N/A 0/43 0/37 0/45 N/A 1/43 0/28 0/45 0/45 1/25 2/311 (0.6%) B. clarridgieae # 2 years old Culture 1/45 0/45 2/30 0/41 0/45 0/45 1/40 3/45 0/42 0/43 0/45 0/32 7/498 (1.4%) B. henselae . 2 years old PCR N/A N/A N/A 0/45 1/45 1/38 N/A 0/45 0/42 2/43 0/45 2/32 6/335 (1.8%) B. henselae . 2 years old PCR N/A N/A N/A 0/45 1/45 1/38 N/A 0/45 0/42 0/43 0/45 0/32 2/335 (0.6%) B. clarridgieae . 2 years old detected between 1 and 10 colony-forming units. The broth culture culture for B. henselae (Table III). Bartonella clarridgeiae was not iso- medium BAPGM, which was subcultured on chocolate agar, was lated from any of the samples tested. found to be between 100 and 1000 times less sensitive than direct culture on chocolate agar. Culture in BAPGM broth, followed by Results of PCR on blood from stray cats DNA extraction and PCR analysis was found to be about 100 times The results of PCR for Bartonella spp., hemoplasma, and R. felis less sensitive than direct culture on chocolate agar. Because of the shown in Table II indicate the high prevalence of B. henselae (58%) toxicity of the BAGPM medium identified in the controlled sensitiv- and the absence of B. clarridgeiae in these cats. The high prevalence of ity experiments, its use was discontinued after the first 200 blood M. haemofelis (47%) and its greater prevalence than M. haemoninutum samples were cultured. (13%) is also shown (Table II). Results of blood culture on chocolate agar Results of PCR on fleas Initially 938 blood samples were cultured for Bartonella detection Results of specific PCR on pooled flea samples are shown in from April 2006 to March 2007. Of these samples, 42 blood samples Table II. Using primers targeting the gltA gene of R. felis, 5 of were discarded because of contamination. There was no significant 50 samples were identified as positive and confirmed to be posi- association between Bartonella isolation and season (chi-squared test tive by sequencing of the resulting amplicons In contrast, 9 of 1.15, P = 0.76), nor between Bartonella isolation and age (Fisher’s 50 samples were identified as being positive using primers target- exact probability test, P = 0.78). With the exception of 3 samples, all ing the rOmpB of R. felis. Sequencing of the amplicons showed 99% blood samples positive by culture were also PCR positive. Direct homology to the rOmpB sequence of R. felis in GenBank (Accession PCR on blood was found to be 1.7 times more sensitive than blood Number CP000053.1). Three samples that yielded PCR products

2000;64:0–00 The Canadian Journal of Veterinary Research 415 5 and culture (25). In some other studies, blood culture has been 4 shown to be more sensitive than PCR (25). Culture of blood would

3 intuitively be an optimal approach to diagnosis, because of the rela- tively large volumes of blood that can be assessed compared with 2 PCR. We observed that PCR was, surprisingly, more sensitive than Prevalence (%) Prevalence 1 blood culture. The novel liquid culture medium (BAGPM) described 0 by Maggi et al (18) was, however, far less sensitive than culture of Summer Fall Winter Seasons blood in chocolate agar or direct PCR. Recently described changes in BAPGM may have improved the sensitivity of this medium (25). B. haenselae B. clarridgeiae M. haemofelis M. haemominutum We isolated no B. clarridgeiae which underlines its difficulty of isola- tion (26). Figure 2. The prevalence of blood-borne infection in domestic cats in Ontario from summer 2006 to winter 2007 detected by PCR. Summer was The prevalence of bacteremia in healthy cats obtained in the pres- defined as July to September, Fall as October to December, and Winter as ent study was 3.8% and 1.6%, based on the PCR and culture meth- January to March. ods, respectively, for a combined prevalence of 4.3%. It appears that Bartonella bacteremia is uncommon in healthy pet cats in Ontario, of the size expected for B. henselae were cloned into pGEM-T Easy which may reflect the type of cats sampled as well as the geographic vector, and sequenced. Resulting sequences shared 99% homology location. Since these were animals in which blood profiles were being to a 16S–23S rRNA ITS sequence of B. henselae Houston-1 strain in analyzed for evidence of “wellness,” it is likely that these cats were GenBank (Accession Number BX897699.1). To confirm the identity well cared for and would be unlikely to have fleas. It was, however, of PCR products identified as B. clarridgeiae, 2 samples were found to impossible to obtain such details. In the United States, Guptill et al be 98% identical to the 16S–23S rRNA ITS sequence of B. clarridgeiae (26) found 6%, 12%, 28%, and 33% prevalence rates for Bartonella Houston-2 strain in GenBank (Accession Number AF312497). bacteremia in pet cats in Chicago, Washington DC, California, and Florida, respectively. Since Ontario is most similar climatically to Chicago, our findings are consistent with those of Guptill et al (26). Discussion As in the present study, Guptill et al (26) observed no association This is the first systematic study of the prevalence of Bartonella between Bartonella bacteremia and season of sampling. The preva- spp. and hemoplasma infection in healthy cats in Ontario, and lence of Bartonella bacteremia is significantly higher in regions with the first to report the presence of R. felis in cat fleas in Canada. higher average temperatures and humidity, favoring flea infestation, The study also documents the uncommon presence of Candidatus than in regions with temperate climatic conditions (26). Flea infesta- M. turicensis. There was a marked contrast between the relatively tion, age , 13 mo, and adoption from a shelter or as a stray cat all low prevalence of these bacteremic infections in healthy pet cats appear to be significant risk factors for Bartonella bacteremia (27,28). versus their high prevalence in stray cats, and a strikingly higher In the current study, although the prevalence of Bartonella bacteremia frequency of M. haemofelis in stray cats than in pet cats. In contrast was higher in the age group of # 2 years old, the difference was to the blood from stray cats, the dominant hemoplasma species in not significant. Lappin et al (23) conducted a PCR-based study of fleas was M. haemominutum, as it was in pet cats, but the prevalence Bartonella species in 92 flea-infested cats in Alabama, Maryland, and of B. henselae in fleas was similarly very high. Texas, and found B. henselae in 32 (34.7%), B. clarridgeiae in 9 (20.6%), The PCR assay suggested by Jensen et al (15) for diagnosis of and both species in 7 (7.6%) cats, suggesting an association between bartonellosis has been widely used in various epidemiological fleas and bacteremia. Many studies have determined the prevalence studies (20–23). The primers used may nonspecifically amplify the of Bartonella bacteremia in different countries. In the Netherlands, ubiquitous organisms of Mesorhizobium spp., so that Maggi and 22% of 113 cats were positive by blood culture (27). In France, 16.5% Breitschwerdt suggested that other primers be used (16). However of domestic cats in Paris were found to be bacteremic (28). In the these primers amplify a region of B. henselae in which there are vari- United , 34 (9.4%) out of 360 cats sampled were bacteremic able repeats, resulting in amplicons of unpredictable sizes (24). The (29). It is difficult to make direct comparisons with other studies use of a novel combination of these primers (MF321/JR473) was since variables such as climate, flea infestation, age, and source all chosen for study in this investigation. These primers amplify the appear to influence the prevalence of bartonellosis. most common veterinary-relevant Bartonella spp. but do not target The prevalence of Bartonella bacteremia in stray cats in Ontario the variable repeat region of B. henselae, thus yielding PCR products was remarkably high compared with pet cats. This finding is gener- of predictable and consistent size. In addition, PCR using this novel ally supported by the results of other studies (26) that suggest that primer pair may be more sensitive for the detection of Bartonella spp. being a stray cat is a risk factor for bartonellosis (27,28,30). Although in blood than other PCR assays reported. a higher prevalence of Bartonella bacteremia has been reported from For clinical samples, PCR detection of Bartonella spp. was almost tropical countries (31,32), the high prevalence of Bartonella infection twice as sensitive as culture of blood samples on chocolate agar, in the current study is striking, considering the differences in climatic which was less than the sensitivity assessed using serially diluted conditions in Ontario compared to tropical countries or the southern cultures. Since 3 culture-positive samples were missed by PCR, the United States. The prevalence of Bartonella infection in stray cats new PCR approach used here is not perfect, so that the best method probably reflects higher exposure to cat fleas, which are considered for diagnosis of Bartonella bacteremia may be a combination of PCR the major route of Bartonella transmission between cats (26,28), and

416 The Canadian Journal of Veterinary Research 2000;64:0–00 is supported by the very high prevalence of Bartonella spp. in cat this organism usually causes subclinical or mild “flu-like” symp- fleas in Ontario (Table II). toms, serious or even fatal infections especially in elderly and young Numerous studies have determined the prevalence of hemoplasma people have been reported (11,12). The presence of this organism in infection in healthy and diseased cats in different countries. In a cat fleas in Ontario underlines the need for flea control in domestic PCR-based study, Jensen et al (17) found an overall prevalence of cats and for physicians to consider this infection in people presented 19.5% hemoplasma in a population of healthy and anemic cats in with febrile illness and flea bites. the United States. Although similar numbers of anemic and nonane- Bartonella DNA was amplified from 49 of the 50 flea pools, includ- mic cats were infected with M. haemominutum, significantly higher ing 43 infected with B. henselae, 4 with B. clarridgeiae, and 2 that were numbers of anemic cats were infected with M. haemofelis. In a PCR- dually infected. This is higher than observed elsewhere (23,37,39). based study of healthy and diseased cats from Saskatchewan and The relatively high prevalence of infection of cat fleas in the current Alberta (33), most cats with regenerative anemia were infected with study might relate to differences in sensitivity of the PCR assays M. haemofelis rather than M. haemominutum. In the United Kingdom, used in these studies. The infection rates for hemoplasma in pooled Tasker et al (34) found the overall prevalence of hemoplasma infec- fleas (33 of 50, 10% and 56% for M. haemofelis and M. haemominutum) tion in healthy and sick cats to be 18.5% but, in contrast with the are higher than the rates obtained in similar studies by Shaw et al results of Jensen et al (17), most sick-anemic cats were infected with (37) (49%, all M. haemominutum) and Lappin et al (23) (M. haemofelis M. haemominutum rather than M. haemofelis. In a study of 92 healthy 3.3%, and M. haemominutum 23.9%), but the higher prevalence of cats in the United States (23), 7.6% and 21.7% were infected with M. haemominutum is similar. Since fleas are hematophagous, the M. haemofelis and M. haemominutum, respectively. The prevalence detection of pathogen DNA in fleas may reflect the existence of of hemoplasma infection in healthy cats in the current study was these infections in their hosts as well, but the prevalence of infection low, consistent with our earlier speculation that these cats would in the source cats was not determined. However the conclusion is be unlikely to be exposed to risk factors for hemoplasma infec- that cats infested with C. felis must be exposed to a combination of tion, and M. haemoninutum was more common in healthy cats than flea-associated bacterial organisms, and that these may be present M. haemofelis, a finding consistent with that of most previous studies. at high prevalence in fleas. By contrast, the prevalence of hemoplasma infection in the stray cats was considerably higher. This may reflect the likelihood that Acknowledgments being a stray implies a greater probability of flea-infestation and of fighting, but no details were available to support this specula- The authors thank the Ontario Veterinary College’s Pet Trust tion. Besides the high prevalence, a second striking feature in stray for financial support. We are also grateful to Drs. A. Abrams-Ogg, cats was the relatively high prevalence of M. haemofelis compared H. Cai, and C. Ribble for discussions during the course of this work, to M. haemominutum. One explanation might be the ability of and to Drs. K.R. Macaluso and B. Willi for provision of control DNA M. haemofelis to produce prolonged bacteremia than M. haemominu- from R. felis and M. turicensis. Dr. H. Cai kindly provided control tum and Candidatus M. turicensis (35). DNA from M. haemofelis and M. haemominutum. Dr. D. Bienzle pro- Candidatus Mycoplasma turicensis was recently discovered in a vided blood from the stray cats. The authors also thank the Humane Swiss cat with severe hemolytic anemia (9). In the current study, Societies of Cambridge, Kitchener, and Guelph for assistance with one presumptively M. haemofelis PCR-positive samples was found to obtaining cat fleas. be Candidatus M. turicensis, the first time this has been reported in North America. In a study of 713 healthy and ill cats of Switzerland, References Willi et al (35) later reported a 1.1% infection rate for Candidatus M. turicensis, all from ill cats. An infection rate for Candidatus 1. Breitschwerdt EB, Kordick DL. Bartonella infection in animals: M. turicensis of 2.3%, 10% and 26% was observed in a sample of Carriership, reservoir potential, pathogenicity, and zoonotic healthy and ill cats in the United Kingdom, Australia, and ill cats potential for human infection. Clin Microbiol Rev 2000;13: of South Africa (36). The current study suggests that the infection is 428–438. uncommon in cats in Ontario. 2. Koehler JE, Glaser CA, Tappero JW. Rochalimaea henselae infection. Rickettsia felis was identified here in cat fleas for the first time in A new with the domestic cat as reservoir. J Am Med Canada. Others have found somewhat similar or higher infection Assoc 1994;27:531–535. rates elsewhere (10,37,38). The PCR based on the rOmpB gene was 3. Kordick DL, Hilyard CEJ, Hadfield TL, et al. Bartonella clarridge- twice as sensitive as that based on the gltA gene. Despite the high iae, a newly recognized zoonotic pathogen causing inoculation prevalence of Bartonella and hemoplasma infection, no stray cat papules, fever, and lymphadenopathy (cat scratch fever). J Clin blood was PCR positive for R. felis. Natural infection of cats with Microbiol 1997;35:1813–1818. this organism has not been reported. Besides opossums, which are 4. Avidor B, Graidy M, Efrat G, et al. Bartonella koehlerae, a new the reservoir of this organism in North America, it is believed that cat-associated agent of culture-negative human endocarditis. fleas are the main reservoir for R. felis, since vertical transmission of J Clin Microbiol 2004;42:3462–3468. this organism has been reported (14). Experimental infection studies 5. Anderson BE, Neuman MA. Bartonella spp. as emerging human suggest that cats may naturally be infected by R. felis, but that the pathogens. Clin Microbiol Rev 1997;10:203–219. period of bacteremia is short (13). The current study supports the 6. Arvand M, Wendt C, Regnath T, Ullrich R, Hahn H. Characteriza­ conclusion that R. felis bacteremia in cats is rare. In people, although tion of Bartonella henselae isolated from bacillary angiomatosis

2000;64:0–00 The Canadian Journal of Veterinary Research 417 lesions in a human immunodeficiency virus-infected patient in species, Ehrlichia species, and Anaplasma phagocytophilum Germany. Clin Infect Dis 1998;26:1296–1299. DNA in the blood of cats with anemia. J Feline Med Surg 2007; 7. Boulouis HJ, Chang CC, Henn H, Kasten RW, Chomel BB. Factors 9:1–9. associated with the rapid emergence of zoonotic Bartonella infec- 23. Lappin MR, Griffin B, Brunt J, et al. Prevalence of Bartonella tions. Vet Res 2005;36:383–410. species, haemoplasma species, Ehrlichia species, Anaplasma 8. Sykes JE. Feline hemotropic mycoplasmosis (feline hemobar- phagocytophilum, and Neorickettsia risticii DNA in the blood of tonellosis). Veterinary Clinics of North America, Small Animal cats and their fleas in the United States. J Feline Med Surg 2006; Practice 2003;33:773–789. 8:85–90. 9. Willi B, Boretti FS, Cattori V, et al. Identification, molecular 24. Dillon B, Iredell J, Breitschwerdt EB, Maggi RG. Potential limita- characterization, and experimental transmission of a new hemo- tions of the 16S-23S rRNA intergenic region for molecular detec- plasma isolate from a cat with hemolytic anemia in Switzerland. tion of Bartonella species. J Clin Microbiol 2005;43:4921–4922. J Clin Microbiol. 2005;43:2581–2585. 25. Duncan AW, Maggi RG, Breitschwerdt EB. A combined approach 10. Marquez FJ, Muniain MA, Rodriguez-Liebana JJ, Del Toro MD, for the enhanced detection and isolation of Bartonella species Bernabeu-Wittel M, Pachon AJ. Incidence and distribution in blood samples: Pre-enrichment liquid culture followed pattern of Rickettsia felis in peridomestic fleas from Andalusia, by PCR and subculture onto agar plates. J Microbiol Methods southeast Spain. Ann N Y Acad Sci 2006;1078:344–346. 2007;69:273–281. 11. Oliveira RP, Galvão MAM, Mafra CL, et al. Rickettsia felis 26. Guptill L, Wu CC, HogenEsch H, et al. Prevalence, risk fac- in Ctenocephalides spp. fleas, Brazil. Emerg Infect Dis 2002;8: tors, and genetic diversity of Bartonella henselae infections in 317–319. pet cats in four regions of the United States. J Clin Microbiol 12. Richter J, Fournier PE, Petridou J, Häussinger D, Raoult D. 2004;42:652–659. Rickettsia felis infection acquired in Europe and documented by 27. Bergmans AM, de Jong CM, van Amerongen G, Schot CS, polymerase chain reaction. Emerg Infect Dis. 2002;8:207–208. Schouls LM. Prevalence of Bartonella species in domestic cats in 13. Wedincamp J, Jr, Foil LD. Infection and seroconversion of cats The Netherlands. J Clin Microbiol 1997;35:2256–2261. exposed to cat fleas (Ctenocephalides felis bouché) infected with 28. Gurfield AN, Boulouis HJ, Chomel BB, et al. Epidemiology of Rickettsia felis. J Vector Ecol 2000;25:123–126. Bartonella infection in domestic cats in France. Vet Microbiol 14. Boostrom A, Beier MS, Macaluso JA, et al. Geographic association 2001;80:185–198. of Rickettsia felis-infected opossums with human murine typhus, 29. Birtles RJ, Laycock G, Kenny MJ, Shaw SE, Day MJ. Prevalence Texas. Emerg Infect Dis 2002;8:549–554. of Bartonella species causing bacteraemia in domesticated and 15. Jensen WA, Fall MZ, Rooney J, Kordick DL, Breitschwerdt companion animals in the United Kingdom. Vet Rec 2002; EB. Rapid identification and differentiation of Bartonella spe- 151:225–229. cies using a single-step PCR assay. J Clin Microbiol 2000;38: 30. Fabbi M, De Giuli L, Tranquillo M, Bragoni R, Casiraghi M, 1717–1722. Genchi C. Prevalence of Bartonella henselae in Italian stray cats: 16. Maggi RG, Breitschwerdt EB. Potential limitations of the 16S-23S Evaluation of to assess the risk of transmission of rRNA intergenic region for molecular detection of Bartonella Bartonella to humans. J Clin Microbiol 2004;42:264–268. species. J Clin Microbiol 2005;43:1171–1176. 31. Chomel BB, Carlos ET, Kasten RW, et al. Bartonella henselae 17. Jensen WA, Lappin MR, Kamkar S, Reagan WJ. Use of a poly- and Bartonella clarridgeiae infection in domestic cats from the merase chain reaction assay to detect and differentiate two Philippines. Am J Trop Med Hyg 1999;60:593–597. strains of Haemobartonella felis in naturally infected cats. Am J Vet 32. Marston EL, Finkel B, Regnery RL, et al. Prevalence of Bartonella Res 2001;62:604–608. henselae and Bartonella clarridgeiae in an urban Indonesian cat 18. Maggi RG, Duncan AW, Breitschwerdt EB. Novel chemically population. Clin Diagn Lab Immunol 1999;6:41–44. modified liquid medium that will support the growth of seven 33. Kewish KE, Appeleyard GD, Myers L, Kidney BA, Jackson ML. Bartonella species. J Clin Microbiol 2005;43:2651–2655. Mycoplasma haemofelis and Mycoplasma haemominutum detection 19. Rolain JM, Franc M, Davoust B, Raoult D. Molecular detection by polymerase chain reaction in cats from Saskatchewan and of Bartonella quintana, B. koehlerae, B. henselae, B. clarridgeiae, Alberta. Can Vet J 2004;45:749–752. Rickettsia felis, and Wolbachia pipientis in cat fleas, France. Emerg 34. Tasker S, Binns SH, Day MJ, et al. Use of a PCR assay to assess Infect Dis 2003;9:338–342. the prevalence and risk factors for Mycoplasma haemofelis and 20. Pons I, Sanfeliu I, Quesada M, et al. Prevalence of Bartonella ‘Candidatus Mycoplasma haemominutum’ in cats in the United henselae in cats in Catalonia, Spain. Am J Trop Med Hyg 2005;72: Kingdom. Vet Rec 2003;152:193–198. 453–457. 35. Willi B, Boretti FS, Baumgartner C, et al. Prevalence, risk fac- 21. Hackett TB, Jensen WA, Lehman TL, et al. Prevalence of DNA tor analysis, and follow-up of infections caused by three feline of Mycoplasma haemofelis, ‘Candidatus Mycoplasma haemominu- hemoplasma species in cats in Switzerland. J Clin Microbiol 2006; tum’, Anaplasma phagocytophilum, and species of Bartonella, 44:961–969. Neorickettsia, and Ehrlichia in cats used as blood donors in the 36. Willi B, Tasker S, Boretti FS, et al. Phylogenetic analysis of United States. J Am Vet Med Assoc 2006;229:700–705. “Candidatus Mycoplasma turicensis” isolates from pet cats in the 22. Ishak AM, Radecki S, Lappin MR. Prevalence of Mycoplasma United Kingdom, Australia, and South Africa, with analysis of haemofelis, ‘Candidatus Mycoplasma haemominutum,’ Bartonella risk factors for infection. J Clin Microbiol 2006;44:4430–4435.

418 The Canadian Journal of Veterinary Research 2000;64:0–00 37. Shaw SE, Kenny MJ, Tasker S, Birtles RJ. Pathogen carriage by 39. Kelly PJ, Meads N, Theobald A, Fournier PE, Raoult D. Rickettsia the cat flea Ctenocephalides felis (Bouche) in the United Kingdom. felis, Bartonella henselae, and B. clarridgeiae, New Zealand. Emerg Vet Microbiol 2004;102:183–188. Infect Dis 2004;10:967–968. 38. Schloderer D, Owen H, Clark P, Stenos J, Fenwick SG. Rickettsia felis in fleas, Western Australia. Emerg Infect Dis 2006;12:841–843.

2000;64:0–00 The Canadian Journal of Veterinary Research 419