Candidatus Bartonella merieuxii, a Potential New Zoonotic Bartonella Species in Canids from Iraq

Bartonellae are emerging vector-borne pathogens infecting erythrocytes and endothelial cells of various domestic and wild mammals. Blood samples were collected from domestic and wild canids in Iraq under the United States Army zoonotic disease surveillance program. Serology was performed using an indirect immunofluorescent antibody test for B. henselae, B. clarridgeiae, B. vinsonii subsp. berkhoffii and B. bovis. Overall seroprevalence was 47.4% in dogs (n = 97), 40.4% in jackals (n = 57) and 12.8% in red foxes (n = 39). Bartonella species DNA was amplified from whole blood and representative strains were sequenced. DNA of a new Bartonella species similar to but distinct from B. bovis, was amplified from 37.1% of the dogs and 12.3% of the jackals. B. vinsonii subsp. berkhoffii was also amplified from one jackal and no Bartonella DNA was amplified from foxes. Adjusting for age, the odds of dogs being Bartonella PCR positive were 11.94 times higher than for wild canids (95% CI: 4.55–31.35), suggesting their role as reservoir for this new Bartonella species. This study reports on the prevalence of Bartonella species in domestic and wild canids of Iraq and provides the first detection of Bartonella in jackals. We propose Candidatus Bartonella merieuxii for this new Bartonella species. Most of the Bartonella species identified in sick dogs are also pathogenic for humans. Therefore, seroprevalence in Iraqi dog owners and bacteremia in Iraqi people with unexplained fever or culture negative endocarditis requires further investigation as well as in United States military personnel who were stationed in Iraq. Finally, it will also be essential to test any dog brought back from Iraq to the USA for presence of Bartonella bacteremia to prevent any accidental introduction of a new Bartonella species to the New World.


Introduction
Bartonella are fastidious, hemotropic, gram negative, rods that have an affinity for the erythrocytes, endothelial cells and macrophages of their hosts [1][2][3]. These bacteria cause vasoproliferative lesions by entering endothelial cells resulting in cellular proliferation and migration [4]. The organisms are highly adaptive which enables them to persist within the cells of their mammalian hosts for extended periods [2]. Culturing the bacteria is often frustrating [5,6], as colonies can take up to 45 days to grow on enriched blood media [2].
More than 25 species/subspecies of Bartonella and many Candidatus species have been identified in a wide range of domestic and wild animals [10; http://www.bacterio.cict.fr]. Of these, 14 are recognized as zoonotic or potentially zoonotic, including 10 isolated from dogs and cats.
Several Bartonella species, most of which are zoonotic, have been isolated or detected by PCR amplification and DNA sequencing from dogs, including B. vinsonii subsp. berkhoffii, B. henselae, B. clarridgeiae, B. rochalimae, B. quintana, B. koehlerae, B. elizabethae, B. vinsonii subsp. arupensis, B. taylorii, B. volans-like, B. bovis and B. bovislike [3,[17][18][19][20]. Recently, six Italian dogs and one Greek dog were infected with a novel, uncultured Bartonella sp. strain designated as strain HMD [GenBank accession number EF614393], which is phylogenetically similar, but distinct from B. bovis [19]. As the abnormalities observed in dogs infected with Bartonella spp. are very similar to the clinical and pathologic abnormalities seen in humans, the findings in this manuscript may prove to be of comparative medical importance [2,3]. The role of dogs as a primary reservoir for various Bartonella species is still unclear [2]. Nevertheless, dogs are excellent sentinels as part of a surveillance system for the detection of human infections [21]. In Iraq, limited information is available on zoonotic diseases in canids. A recent study investigated the presence of Q fever antibodies in military working dogs (MWD) deployed in Iraq and free-roaming dogs eliminated through a feral animal control program in locations throughout Iraq [22]. None of the MWD seroconverted versus 5.5% of the indigenous dogs. To our knowledge, presence of Bartonella infection has never been reported in Iraq. Therefore, the objective of the present study was to investigate the presence of Bartonella spp. by PCR amplification and to assess seroprevalence of B. henselae, B. clarridgeiae, B. vinsonii subsp. berkhoffii and B. bovis in domestic and wild canids of Iraq.

Materials and Methods
Subject enrollment, sample collection, and storage Whole blood samples (1.5-2 ml) were collected between February and December 2008 under the US Army feral animal control and zoonotic disease surveillance program from the saphenous or jugular veins in EDTA tubes from 193 canids, including 97 (50.3%) stray dogs (Canis familiaris), 57 (29.5%) jackals (Canis aureus) and 39 (20.2%) red foxes (Vulpes vulpes) located on several United States military bases throughout Iraq. Categorical age, sex, and location data was also collected with samples, when possible.

Ethics statement
All animals were chemically restrained prior to blood collection and humanely treated prior to euthanasia, in accordance with the rules of the ethic committee from the US Army feral animal control and zoonotic disease surveillance program and by the University of California, Davis (UCD) Animal Use and Care Committee (Protocol # 12668 originally approved on March 8, 2007) in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The blood samples were frozen and shipped to the UCD School of Veterinary Medicine. In a pilot study on 40 dogs, 10 jackals and 35 foxes ((McMillan-Cole, Master of Preventive Veterinary Medicine, Davis, CA; data not shown), either major bacterial contamination or lack of isolation of Bartonella was observed when blood culture were attempted. Therefore, the blood samples analyzed in the present study were only processed by DNA extraction.

Serological analysis
Presence of antibodies against B. henselae H1 (ATCC 49882) and U4 (UC Davis), B. clarridgeiae (ATCC51734), B. vinsonii subsp. berkhoffii genotype I (ATCC 51672) and B. bovis (strain 91-4T: CIP 106692T and CCUG 43828T) were tested on the serum or blood supernatant samples, using an indirect immune-fluorescent-antibody assay (IFA) as previously described [18,23,24]. Serum samples were diluted to 1:32 and 1:64 and 20 ml of each dilution was added to the test wells. Positive and negative controls were included on each slide. Three conjugate (Cappel fluoresceinconjugate goat anti-dog IgG fraction, MP Biomedicals, Aurora, OH, USA) dilutions were used for the four antigens being tested: 1:1400 (B. vinsonii subsp. berkhoffii and B. bovis), 1:2800 (B. henselae H1 and U4) and 1:3600 (B. clarridgeiae). For each well, 20 ml of the conjugates were added to the respective slides. The intensity of the fluorescence was graded subjectively by two independent masked readers, with sample fluorescence graded $2 at the 1:64 dilution reported as positive.

DNA extraction and PCR-RFPL
The DNA was extracted as previously described [25]. Polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) analysis of the gltA gene was then performed for all samples [25]. The primers used for the gltA gene were BhCS.781p 59-and BhCS.1137n 59, as previously described [26]. An approximately 400 bp fragment of the gltA gene was amplified and then verified by gel electrophoresis. The amplified product was then digested with restriction endonucleases TaqI

Sequencing and phylogenetic analysis
Two representative DNA products from the dog and jackal groups were further analyzed by PCR of the 16S-23S rRNA intergenic spacer region (ITS) [27] and of the rpoB gene [28] and submitted to sequencing. As the gltA gene sequence for the HMD strain had not been previously performed, DNA from strain HMD

Author Summary
Bartonellae are emerging vector-borne pathogens infecting erythrocytes and endothelial cells of various domestic and wild mammals. Blood samples were collected from domestic and wild canids in Iraq under the United States Army zoonotic disease surveillance program. Serology was performed using an indirect immunofluorescent antibody test for B. henselae, B. clarridgeiae, B. vinsonii subsp. berkhoffii and B. bovis. Overall seroprevalence was 47% in dogs (n = 97), 40% in jackals (n = 57) and 13% in red foxes (n = 39). Bartonella species DNA was amplified from whole blood and representative strains were sequenced. DNA of Candidatus Bartonella merieuxii, a new Bartonella species similar to but distinct from B. bovis, was amplified from 37% of the dogs and 12% of the jackals. B. vinsonii subsp. berkhoffii was also amplified from one jackal and no Bartonella DNA was amplified from foxes. Dogs were more likely to be Bartonella PCR positive than wild jackals and foxes, suggesting the role of dogs as reservoir for this new Bartonella species. As most Bartonella species isolated or detected in dogs are also infecting humans, it will be important to test Iraqi people, especially from Baghdad, and American veterans who served in Iraq for the presence of infection by Candidatus Bartonella merieuxii.
was submitted by one of the authors (EB), and after PCR testing with the appropriate primers, was submitted for sequencing (Davis Sequencing, Davis, CA, USA). Amplified products from the gltA gene, rpoB gene, and the ITS region were sequenced in both directions and consensus sequences were compared. The sequence alignments obtained in this study were also referenced against those of other known Bartonella species deposited in the GenBank/ EMBL/DDBJ database using BLAST [ Table 1]. The Clustal W program [29] was used to align each isolate and compare homologous gltA, rpoB and ITS spacer sequences to identify genetic variants. A phylogenetic tree was constructed from concatenated sequences of gltA, rpoB and ITS from representative Bartonella isolates, including ruminant-associated Bartonella species, by using neighbor-joining methods. Nucleotide substitution rates were calculated by Kimura's two-parameter distance model, as previously described [30], using MEGA version 5.0 software [31]. Bootstrap analysis was performed with 1,000 trials.

Statistical analysis
Bartonella prevalence data generated from diagnostic tests and binomial proportion confidence intervals (CI) were calculated using Wald and Fisher's exact methods where appropriate in OpenEpi, version 2.3.1 (Dean AG, Sullivan KM, Soe MM. OpenEpi: Open Source Epidemiologic Statistics for Public Health). All variables were initially screened using univariate logistic regression to check for statistically significant associations with serology and PCR outcomes. Due to quasi-complete separation of PCR data, species and location variables were reclassified in biologically meaningful groups prior to univariate analysis. Using forward stepwise selection methods, categorical variables were retained in separate multivariate logistic regression models for PCR and IFA. Variables that induced changes .10% in the b-coefficients of covariates after inclusion in the model were considered to be potential confounders, and evaluated along with biological reasoning and prior knowledge. Likelihood ratio tests comparing models with and without two-way interaction terms were used to determine presence of effect measure modification. All logistic regression models and resulting adjusted odds ratios with 95% CI were generated in SAS, version 9.3 (SAS Institute Inc., Cary, NC).

Dogs
Most of the dogs (79.38%; 77/97) were captured in Baghdad and surrounding suburbs (Inner Baghdad: 35, North Baghdad: 27, Outer Baghdad: 15). The other dogs were from locations classified as North Iraq (n = 15), West Iraq (n = 3) or South Iraq (n = 2). Forty-three (57.3%) of the 75 dogs for which sex was reported (missing from 22 dogs) were male. Age was not recorded for 27 animals, with 60 (85.7%) of the 70 dogs classified as adults. Thirtysix dogs (37.1%; 95% CI = 27.5%-46.73%) were PCR positive [ Table 2]. RFLP analysis of the gltA PCR product from the four enzyme digestions indicated similar banding patterns for all 36 dogs (data not shown). Banding pattern was similar to B. clarridgeiae using TaqI, HhaI and MseI. However, DdeI allowed differentiating these strains from B. clarridgeiae. A representative dog ''strain'' (F059) was chosen for further genetic analysis. Sequencing of the gltA gene fragment showed that the closest related Bartonella species were respectively B. bovis, B. schoenbuchensis, B. melophagi, B. chomelii and B. capreoli [ Table 3]. The partial gltA sequence of the HMD strain was deposited in GenBank under the number JX073031. Sequencing of the 16S-23S ITS spacer region indicated that DNA of ''strain'' F059 was 100% identical to the previously reported HMD strain [19] and most closely related to ruminant Bartonella species [ Table 3]. For the rpoB sequences, DNA of ''strain'' F059 was also 100% identical to strain HMD and with the closest related species being B. bovis and other ruminant Bartonella species [ Table 3] A phylogenetic tree of the concatenated sequences of the gltA and rpoB genes and the 16S-23S ITS indicted that all strains clustered together and aligned most closely to B. bovis ( Figure 1). However, partial sequences were less than 95% similar for all genes, suggesting that these canid strains likely belong to a new Bartonella species. DNA of other Bartonella species, including B. vinsonii subsp. berkhoffii was not PCR amplified from the blood of these dogs. The 36 PCR positive dogs were from either Baghdad (n = 30, 39%) or North Iraq (n = 6, 40%). Within Baghdad and its suburbs, 57.14% (20/35)

Jackals
Of the 57 jackals sampled, 55.6% (30/54) were collected from North Baghdad, 18.5% (10/54) from Outer Baghdad, 16.7% (9/ 54) from North Iraq, and 9.3% (5/54) from West Iraq (complete location data was missing for three jackals). The sex ratio of the  Table 2]. The other six jackals infected with a ''strain'' identical to strain HMD were also male, four being from North Baghdad and two from North Iraq. Five of these animals were reported as adults and age classification was missing from one sample. DNA of a representative jackal ''strain'' (F040) was chosen for genetic analysis. The gltA gene sequence aligned most closely with B. bovis and the other ruminant Bartonella species [ Table 4]. The 16S-23S ITS region sequence of DNA from jackal ''strain'' F040 was 100% identical to the HMD strain, aligning most closely with B. schoenbuchensis and the other ruminant Bartonella species. For the rpoB sequences, DNA amplification from jackal F040 was 100% identical to strain HMD and aligned most closely with B. bovis and the other ruminant Bartonella species [ Table 4].
Overall, 40.4% (23/57; 95% CI: 27.6%-53.1%) of the jackals were seropositive for at least one Bartonella antigen [ Table 2]. Of the seven PCR positive jackals, only one was seronegative, which was a Bartonella sp. strain HMD adult male from North Baghdad. All six other jackals were IFA positive for the four Bartonella antigens. Of the 17 seropositive jackals that were PCR negative, 70.6% (12/17) were seropositive for the four antigens tested.

Foxes
Blood samples were collected from 39 red foxes, comprised of 21 from West Iraq, 15 from Baghdad (four from Inner and 11 from Outer Baghdad) and three from North Iraq. There were 20 male and 19 female foxes, including 29 adults and 10 young individuals. None of the fox samples were Bartonella PCR positive [ Table 2]. Seroprevalence in foxes was 12.8% (5/39; 95% CI: 4.3%-27.4%).

Comparisons among canids
Species, sex, age and location data were collected for most of the canids and submitted to a univariate analysis [Tables 5 and 6). Species and age were then included in the multivariate models for both PCR and serology outcomes [Tables 7 and 8]. Adjusting for age, the odds of dogs being Bartonella PCR positive were 11.94 times higher than of wild canids (95% CI: 4.55-31.35). Adjusting for species, the odds of adult canids being PCR positive were 5.62 times higher than young canids (95% CI: 1.19-26.59). The ageadjusted odds of dogs being Bartonella seropositive were 8.71 times higher than for foxes (95% CI: 3.03-25.01). Similarly, ageadjusted odds of jackals being Bartonella seropositive were 5.64 times higher than of foxes (95% CI: 1.87-17.07).

Discussion
This study reports on the prevalence of Bartonella species in domestic and wild canids of Iraq and provides the first detection of Bartonella in jackals. Dogs and jackals were frequently infected with a new Bartonella species similar to strain HMD, initially described from dogs in southern Italy and Greece [19] and more recently in two dogs from Sri Lanka [32]. The PCR prevalence in Iraqi dogs (45.4%) was similar to the prevalence in the dogs from Basilicata, Italy (42.8%, 6/14).
Based on the criteria suggested by La Scola et al. [33], the low similarity percentages (,95.5%) between the Iraqi, Greek, Italian and Sri Lankan ''strains'' and B. bovis clearly suggest that the ''strains'' infecting dogs belong to a new Bartonella species. We

(2)
Vulpes vulpes (39)   propose Candidatus Bartonella merieuxii for this new species. B. vinsonii subsp. berkhoffii, a common Bartonella species isolated from dogs [3], was not detected in the Iraqi dogs and detected only from one jackal. Similarly, B. rochalimae was not detected from any of the canids, which is especially interesting as red foxes have been found to be infected by this Bartonella species in Israel and Europe [30]. Given the statistically significantly higher PCR prevalence of Candidatus B. merieuxii infection in dogs than in wild canids, domestic dogs may represent a major reservoir of this new species in Iraq. Because this new species is genetically related to B. bovis, we sought to investigate the seroprevalence of B. bovis in these canids in addition to the antigens usually used for canids in our laboratory. Most (86%) dogs and jackals that were PCR positive were also seropositive for B. bovis and for the three other antigens. The lowest seroprevalence observed was for B. clarridgeiae and B. vinsonii subsp. berkhoffii, suggesting that these Bartonella species may be found less frequently in Iraqi canids as compared to other parts of the world. The high seroprevalence in stray dogs and jackals correlates with previously reported studies on seroprevalence of Bartonella species in animals of the tropics, Middle East, and some African nations [1,2]. A recent study of stray dogs in the tropics [32] reported an overall seroprevalence of 8.3% (41/455), ranging from 0% in dogs from Vietnam to almost 11% in dogs from Bogota, Colombia. These data contrast with reports in domestic pet dogs in the United States and Europe, where overall seroprevalence is ,5% [1]. In Thailand, prevalence of Bartonella bacteremia in stray dogs was 31.3% (60/192) [17].
Cross-reactivity is known to occur between B. henselae and B. clarridgeiae, especially in human sera [2], and therefore it is unclear whether the seropositivity for more than one Bartonella species was due to cross-reactions or to co-infection or sequential infection to different Bartonella spp. over an extended time period. Given the relatively small sample sizes of some canid groups and locations, larger studies will be important to strengthen our findings.
To date, no vector has been proven as the source of infection for any Bartonella spp. in dogs. The current hypothesis on the mode of transmission of B. vinsonii subsp. berkhoffii involves ticks, because seropositive dogs were also frequently seropositive for other known tick-borne pathogens [3,34,35]. Similarly, although the body louse is considered the primary vector for B. quintana, B. quintana DNA has been amplified from both fleas and ticks, providing support for being possible Bartonella vectors [9,36]. Furthermore, Candidatus Bartonella merieuxii was detected in Rhipicephalus sanguineus ticks, as DNA was amplified from three tick pools including one pool from salivary glands from female ticks and one gut content pool collected two dogs PCR positive for Candidatus B. merieuxii [19]. Therefore, ticks may be possible candidate vectors for this new Bartonella species. Unfortunately, information on the presence of ectoparasites was not available and no quantitative measure of the   [2,3]. Domestic cats are generally asymptomatic carriers of B. henselae and B. clarridgeiae. Symptomatic bacteremic dogs are often seronegative and frequently manifest similar pathology and disease manifestations as reported in human patients [3,19,37]. Furthermore, most of the Bartonella species identified in sick dogs are also pathogenic for humans [2]. Bartonella quintana, for instance, causes a high percentage of endocarditis in people and has been associated with endocarditis in dogs [2,38,39]. Therefore, seroprevalence in Iraqi dog owners and bacteremia in Iraqi people with unexplained fever or culture negative endocarditis requires further investigation. This is especially true in Baghdad and the surrounding region, where most of the positive dogs were detected. Similarly, it will be important to investigate pre and post-deployment seroprevalence in United States military personnel who were stationed in Iraq. It would be critical to test for Bartonella bacteremia in veterans with unexplained fever, lymphadenopathy, or culture negative endocarditis. Finally, it will also be essential to test any dog brought back from Iraq to the USA for presence of Bartonella bacteremia to prevent any accidental introduction of a new Bartonella species to the New World.
In the context of zoontic bartonellosis, the dog appears to serve as both a sentinel and a reservoir for human infections. As a sentinel, infection in both dogs and humans (most often cases of endocarditis) has now been documented for eight Bartonella species or subspecies. As a reservoir, persistent infection in dogs has been documented in association with B. vinsonii subsp. berkhoffii, B. henselae and B. koehlerae [3]. As such, there are reports of dog bite transmission of B. henselae to people [40] and two recent publications that implicate needle stick transmission B. vinsonii subsp. berkhoffii or B. henselae from dogs to veterinarians [41,42]. In the context of diagnostic testing, several of our laboratories are attempting isolation of this novel Bartonella sp. to facilitate additional genetic characterization and the development and validation of Bartonella species specific serologic assays. In addition, attempts at isolation from sick soldiers deployed in Iraq would seem reasonable.
In summary, presence of Bartonella infection was detected in domestic dogs and wild canids in different areas of Iraq. Our findings identify a potential role for dogs as a major reservoir of a   new and potentially zoonotic Bartonella species.

Description of Candidatus Bartonella merieuxii sp. nov
Bartonella merieuxii (méri.eu.xii N. L. gen. N. merieuxii of Mérieux, in honor of Charles Mérieux, a French physician, founder of the Mérieux Foundation and former C.E.O. of the Mérieux Institute in Lyon, France. His broad interest in comparative medicine and in one health, within the Pasteur's tradition at its best (''No borders between the two medicines'' was his favorite saying), allowed him to build many bridges between physicians and veterinarians in the field of zoonoses. DNA of this species has been amplified and sequenced from dogs and wild canids (jackals) in the old world.