Identification of Novel Zoonotic Activity of Bartonella spp., France

These bacteria may cause paucisymptomatic bacteremia and endocarditis in humans.

It has been assumed that each Bartonella species infected 1 or a few closely related mammalian reservoir hosts, in which infection caused long-lasting bacteremia.
Nonreservoir hosts were considered incidentally infected without bacteria being detected in blood. Recently, these assumptions has been contradicted by studies describing animal-associated Bartonella spp. indirectly associated with bacteremia and a spectrum of diverse symptoms in immune-competent persons who had contact with animals, arthropods, or both, which are natural routes of Bartonella transmission (5)(6)(7). In some cases, the source of infection remains unknown; ticks have been suggested as a possible source of animal-associated Bartonella infection in humans (6,(8)(9)(10).
Related to a patient's history of tick bites, it is common for physicians to suspect Lyme disease, some rickettsial diseases, or tickborne encephalitis. However, in many cases, the diagnosis is not confirmed by serologic or DNA-based tests. In recent years, alternate interpretations of Lyme disease serology have flourished, leading to considerable discord between formal institutions for infectious disease and patient advocacy associations. Thus, unexplained symptoms after tick bites have become an issue of increasing importance for patients and their physicians (11,12).
In this context, we screened for the presence of Bartonella in the blood of patients reporting tick bites and with unexplained and aspecific symptoms. Here we report the isolation and genomic sequencing of 6 Bartonella strains obtained by blood culture from 66 patients. Three strains were identified as B. henselae, and 3 other strains were identified as different animal-associated species (B. doshiae, B. tribocorum, and B. schoenbuchensis).

Patients
During January-June 2013, we conducted a study of a cohort of 66 French patients who had consulted their doctors for chronic symptoms appearing after a tick bite. The entire study protocol was approved by the ethics committee of the Institut Federatif de Recherche 48 under reference 13-022-1.
All patients associated symptom onset with tick bites that occurred during 2008-2012 (Table 1). At symptom onset, local doctors were consulted, and serologic tests for Lyme borreliosis were performed. All patient samples tested were seronegative for Lyme borreliosis bacteria; however, since that time, their symptoms had become chronic. The patients completed information forms giving informed consent for the use of their samples in the study. All of the patients lived in the countryside, where ticks were abundant and contact with wild animals was possible. The patients reported that they had not undergone antibacterial drug treatment for >3 months before the study.
We collected blood samples from each patient in ED-TA-containing sample tubes. For a control population, we used anticoagulated blood samples from 70 anonymous healthy blood donors from Paris (France). All samples (control and patients) were tested simultaneously.

Bartonella Isolation from Blood
To specifically isolate Bartonella spp., 100 mL of blood samples from patients or healthy donors were directly plated onto sheep blood agar plates and incubated at 35°C in a humidified atmosphere with 5% CO 2 for 45 days. The plates were assessed daily from days 7-45 before the culture was deemed negative (i.e., absence of colony in the absence of contamination) (1). Colony-forming units (CFU) were counted and bacteremia (UFD/mL of blood) evaluated.

Genome Sequencing, Assembly, and Analysis
We extracted genomic DNA from each isolated strain by using the EZ1 automated extraction system (QIAGEN, Hilden, Germany), following the manufacturer's recommendations. Bacterial genomic DNA was sequenced by using the Nextera XT DNA sample prep kit (Illumina Inc., San Diego, CA, USA) and a 2×250 paired-end protocol with the MiSeq pyrosequencer (Illumina), according to the manufacturer's instructions. We aligned each genome by using Mira version 3.2 software in the mapping mode (13). The resulting contigs were combined by using Opera version 1.2 (14) and GapFiller (15) software. Finally, the genomic assemblies were improved with manual refinement by using the CLC Genomics version 4.7.2 software package (CLC Bio, Aarhus, Denmark). Noncoding genes and miscellaneous features were predicted by using RNAmmer (16) and ARAGORN (17). Coding DNA sequences were predicted by using Prodigal (18), and functional annotation was achieved by using BLAST+ (19) and HMMER3 (20) against the UniProtKB database (21). Coding DNA sequences were also annotated by using the Clusters of Orthologous Groups databases (22) with blastp (http://blast. ncbi.nlm.nih.gov/Blast.cgi) default parameters.
Single-nucleotide polymorphisms (SNPs) among genomes were identified by using SNIT software (23). SNPs were searched in regions exhibiting >95% nt sequence identity and the SNIT software was used with default parameters except for the Tandem Repeat Finder filter for avoiding ambiguous SNPs in repeat regions. We also performed in silico DNA-DNA hybridization (DDH) between Bartonella strains by using GGDC software (24).

Taxonomic Classification
To determine the taxonomic classification of the 6 isolates, we used previously proposed criteria (25) in which the gltA and rpoB gene sequences from each strain were compared to those of validated published Bartonella species. These criteria classify Bartonella isolates within a particular species if they share >96% and 95.4% nucleotide sequence similarity for the gltA and rpoB genes, respectively (25). In our study, gltA and rpoB sequences were retrieved from the genomes.

Bartonella spp. Isolation
Bartonella spp. were isolated by prolonged culture from blood samples of 6 of the 66 patients who reported chronic symptoms following a tick bite. In contrast, samples from the 70 healthy blood donors remained negative after 45 days of incubation.
Bacteremia in the Bartonella infected patients increased from 50 to 850 CFU/mL. For 1 patient (case-patient 2), we had access to 2 blood samples that were taken at a 1-month interval. B. henselae was grown from the 2 samples, with similar bacteremia (50 and 60 CFU/mL, respectively), suggesting chronic bacteremia.
The case-patients who tested positive for Bartonella (Table 1) reported tick bites occurred 1-5 years before blood samples were collected. All of them live in the countryside, in contrast to the healthy blood donors, who were all from Paris, France. The main complaint of the case-patients was chronic fatigue, but they also reported other subjective or nonspecific symptoms (or both), such as headaches and myalgia. A qualifying characteristic of the 70 healthy blood donors was absence of chronic fatigue. Even though potential exposure to ticks is difficult to evaluate, because the anonymous blood donors all lived in Paris, we assumed they were not likely to have frequent tick exposure or wild animal contact.

Taxonomic Classification
Of the 6 Bartonella isolates from this study, 3 (MVT01, MVT02, and MVT03) were classified within the B. henselae species on the basis of both their phylogenetic position and gltA and rpoB sequence similarities ( Figure; Tables 1, 2). The isolates from samples from case-patient 2 at a 1-month interval shared 100% identity, based on gltA and rpoB gene comparison. Isolates MVT04, MVT05, and MVT07 were classified within the B. tribocorum, B. doshiae, and B. schoenbuchensis species, respectively. The assembly data and main genomic characteristics of each isolated strain are summarized in Table 3; in silico DDH values and SNP numbers are described in Table 4 (http://wwwnc.cdc.gov/EID/article/22/3/15-0269-T4.htm).
All studied strains displayed a similar genomic content when compared with reference genomes. The GGDC software we used proposes that DDH values >70% could classify isolates in the same species. Here, intraspecies values ranged from 80.3% to 100% (Table 4). Nevertheless, from 10 to 1,938 SNPs were identified among B. henselae isolates MVT01, MVT02, and MVT03 and from 693 to 2,093 SNPs when comparing these strains to B. henselae Houston-1 (Table 4), confirming that each strain was unique and did not result from cross-contamination or contamination from laboratory strains.
Of note, B. tribocorum isolate MVT04 and B. schoenbuchensis isolate MVT07 were the only 2 that exhibited plasmids. However, when compared with reference strain m07a, MVT04 and MVT07 carried a large plasmid and not the small plasmid homologous to the cryptic pBGR plasmid harbored by B. grahamii (26).

Discussion
In this study, animal-associated Bartonella isolates were individually cultured from the blood of patients who had been bitten by ticks and reported subjective symptoms, whereas no strains were isolated from healthy blood donors. This report describes the isolation of 3 different animal-associated Bartonella species from human samples, highlighting their potential novel zoonotic properties. Moreover, we found Figure. rpoB gene-based phylogenetic tree showing the relationships of 6 Bartonella isolates (underlined). Briefly, rpoB nucleotide sequences were aligned by using ClustalW software (http://www.clustal.org/clustal2/), and phylogenetic relationships were inferred by using the maximum-likelihood strategy and MEGA software (http://www.megasoftware.net). Bootstrap values above 70%, obtained from 500 analyses, are indicated at the nodes. Scale bar represents a 2% nucleotide sequence divergence. that zoonotic Bartonella spp. can be detected in the blood of afebrile patients, as has been shown for human-specific B. quintana and B. bacilliformis and as was recently reported for Candidatus Bartonella ancashi (27). Chronic bacteremia caused by infection by Bartonella spp. is welldescribed in many mammals, including humans (4,28). The Bartonella-mammalian host association is considered to be species-specific and attributable to co-evolution between host and pathogen (28). However, we show that animalassociated species can also chronically infect human blood, highlighting the possibility of host shift despite apparent host specificity (28,29).
Our findings also confirm studies identifying zoonotic Bartonella in the blood of patients with nonspecific complaints. Among them, B. henselae is well known worldwide  as a zoonotic agent infecting both cats and their fleas and has also been found in ticks (10). B. henselae has been detected in the blood of a patient without apparent symptoms 4 months after recovering from cat scratch disease. For this particular case, the sequence of manifestation of cat scratch disease, then bacteremia, followed by endocarditis was proposed because it has been known to occur for B. quintana bacteremia. One of the case-patients in this study owns a cat and may have been infected by this pet. The 3 other animal-associated species we detected should now be considered zoonotic Bartonella spp. B. doshiae and B. tribocorum are both rodent-associated species; in France and worldwide, these species have mainly been recovered from rats (Microtus agentiis for B. doshiae and Rattus rattus for B. tribocorum). B. schoenbuchensis is normally found in deer, elk, and cattle (37,38).
The zoonotic agents we isolated from patients from France have also been detected in animals in France. Similarly, in the United States and Thailand, Bartonella species known to be prevalent in animals have also been identified in humans: (B. henselae, B. vinsonii berkhoffii, and B. koehlerae in the United States (33,35) and B. tribocorum and B. rattimassiliensis in Thailand (39). Therefore, the zoonotic Bartonella species discovered in humans in this study generally appear to be related to the prevalence among animals.
The significance of these Bartonella spp. in the genesis of the clinical picture is difficult to determine. Bartonella spp. are present in ticks, and we have previously reported Bartonella infections following tick bites, such as SENLAT (scalp eschar and neck lymphadenopathy after tick bite [40]). However, the causal link between the conditions observed here, Bartonella and tick bite, cannot yet be concretely established, especially for persons with tick bites occurring up to 5 years previously, which introduces innumerable potential confounding exposures within the same period, including bites by other arthropods. For instance, 1 of the 3 patients with B. henselae bacteremia reported contact with cats; this contact was a more plausible source of infection than tick bites. Furthermore, it is crucial to determine whether Bartonella played a notable role in the observed pathologies, because treatment for chronic Bartonella bacteremia (as for B. quintana) is particularly arduous and may require 6 weeks of doxycycline treatment together with 3 weeks of gentamicin, as these are the only antimicrobial drugs known to be effective in eradication of Bartonella (1). Many Bartonella spp. can also cause endocarditis, including B. quintana and B. henselae; therefore, reports of rare cases of endocarditis attributed to zoonotic Bartonella such as B. kohlerae, B. alsatica, Candidatus B. mayotimonensis, B. vinsonii, or B. elizabethae may actually be the final manifestation of asymptomatic bacteremia, similar to that reported by our infected patients (28).
In summary, our major finding is the isolation of zoonotic Bartonella other than B. quintana in the blood of patients with poorly qualified syndromes. These results indicate that zoonotic Bartonella spp. infection may cause undifferentiated chronic illness in humans.