Lack of correlation between Bartonella DNA detection within fleas, serological results, and results of blood culture in a Bartonella-infected stray cat population


Corresponding author and reprint requests: B. La Scola, Unité des Rickettsies, CNRS UPRESA 6020, Faculté de Médecine, 27 Boulevard Jean Moulin, 13385 Marseille Cedex 05, France
Tel: +33 91 38 55 17
Fax: +33 91 83 03 90


Objective  To correlate the presence of different Bartonella species in the blood of a stray cat population trapped on a French military base with specific antibodies and species detected in cat fleas.

Methods  The prevalence of Bartonella bacteremia was investigated in 61 cats by plating frozen whole blood on blood agar plates. Identification of isolates and detection of Bartonella DNA from cat flea batches from ten cats was achieved by PCR amplification and sequencing. Antibody detection was performed by microimmunofluorescence.

Results  We obtained 38 isolates of Bartonella from blood. Sixteen were identified as B. clarridgeiae, 15 as B. henselae genotype/serotype Houston 1 (type I), and seven as B. henselae genotype/serotype Marseille (type II). B. henselae was detected in five fleas, and B. clarridgeiae in one flea. Sixty-one per cent of the cats had detectable antibodies against at least one species or serotype. Sixteen cats had antibodies against only one antigen. For each species, the distribution of bacteremia among the cats could not be correlated with either the distribution of infected fleas or the distribution of specific antibodies.

Conclusions  The lack of correlation between Bartonella DNA detection within fleas, serological results, and results of blood culture is probably due to a lack of natural heterologous protection between species or serotypes. Cats suffer bacteremia with three Bartonella species and should therefore be considered the reservoirs of at least three human pathogens.


Human infections due to Bartonella species are widely considered to be emerging diseases, although they also include long-recognized diseases such as Carrion's disease due to B. bacilliformis, trench fever due to B. quintana, and cat-scratch disease (CSD) due to B. henselae and B. clarridgeiae[1–3]. Newer clinical manifestations, such as bacillary angiomatosis and peliosis hepatitis caused by both B. henselae and B. quintana, chronic lymphadenopathy due to B. quintana, and endocarditis due to B. henselae, B. quintana, and, in one case, B. elizabethae, have been recently identified [1,4,5]. The natural host of B. henselae and B. clarridgeiae is the cat. The vector of B. henselae is the cat flea, whereas the vector of B. clarridgeiae is presently unknown, although this species has also been detected in cat fleas [6]. Both species can be isolated from the blood of apparently healthy cats. The prevalence of Bartonella spp. isolated from the blood of cats ranges from 4% to 70%, according to the geographic location of the survey and cat population [7–9].

In 1996, Drancourt et al. reported a new serotype of B. henselae named Marseille [10], which also genotypically differed from previously characterized strains. The authors found that two clinical isolates of B. henselae, from a patient with endocarditis and a patient with CSD, possessed slight differences in 16S rRNA gene sequence when compared to that of the Houston-1 type strain. Later, by determining the 16S rRNA gene sequence of the 18 B. henselae strains isolated in our laboratory (unpublished data), we found that they all belonged to one or the other of these two different genotypes, one being B. henselae Houston [11], and the other B. henselae Marseille [12]. Investigators in The Netherlands [13] also demonstrated two 16S rRNA gene variants in B. henselae DNA amplified from samples from CSD patients. This was shown by analysis of the 16S−23S rRNA gene spacer PCR fragments and 16S rRNA gene PCR products digested with AluI. The presence of two genotypes was later confirmed in France [14] and in Germany [15,16], based on sequencing of the 16S rRNA encoding gene, the genotype Houston being named type I and the genotype Marseille type II. In a study reporting the prevalence of B. henselae in bacteremic cats, the authors also reported the presence of two different genotypes circulating in the cat population [7,14]. Two bacteremic cats were co-infected with the two different genotypes [14].

Our study was performed in the context of a control campaign for stray cats in a French military base. Our work consisted of the determination by blood culture and serology of the prevalence of infection with Bartonella spp. PCR was also used to detect Bartonella spp. in cat fleas collected from the animals tested.

Materials and methods

Cat sampling

The study was performed between June 1997 and June 1999 in the context of a stray cat eradication campaign around the canteens of a French military base. Sixty-one cats were trapped using oral Dexeutanol (Instituto DEX, Seville, Spain) mixed with fresh meat, according to the manufacturer's instructions. Sedated cats were then anesthetized with Imalgène (Rhône Mérieux, Lyon, France), and approximately 3 mL of blood was collected aseptically with a syringe by intracardiac puncture. A 2-mL aliquot of each sample was inoculated into a serum-separating tube for serologic analysis, and the remainder was placed in a lithium–heparin tube for culture and frozen at −80 °C. Euthanasia of cats was performed by intracardiac inoculation of Dolethal (Vétoquinol SA, Lure, France), according to the manufacturer's directions.

Isolation procedure

Frozen blood samples were thawed to room temperature before plating. One milliliter of blood was inoculated onto Columbia 5% sheep blood agar plates (BioMerieux, Marcy l'Etoile, France). Plates were placed in polythene bags and incubated at 37 °C in 5% CO2 (Genbag CO2 system, BioMerieux). The plates were briefly held at a 45° angle to allow the blood to flow across the agar. Plates were initially set agar side down but were inverted after 24 h. Plates were examined weekly for evidence of growth for up to 3 months. When Bartonella-like colonies were observed, confirmation of their identity was achieved using the PCR-based methods outlined below.

Identification of isolates

Presumptive identification of isolates was done by determination of oxidase and catalase reactions and by microscopic examination after Gram and Gimenez staining. DNA extracts were prepared from suspect colonies for use as templates in PCR amplification using the QIAamp Blood Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Molecular identification was based on 16S rRNA gene amplification and sequencing, as previously described [17,18]. Sequences were compared with DNA sequence databases using the program BLAST 2.0 (The National Center for Biotechnology and Information).

Detection of Bartonella spp. in cat fleas

Ten batches of three to five cat fleas were obtained from ten cats. Fleas in each batch were separately crushed in sterile water. DNA extracts were prepared from crushed fleas using the QIAamp Tissue Kit (Qiagen), according to the manufacturer's instructions. Detection of Bartonella spp. was performed using amplification and sequencing of the 16S rRNA gene and intergenic spacer region (ITS), as previously described [19]. Sequences were compared with DNA sequence databases using the program BLAST 2.0. As a control of PCR amplification, we used the 18Saidg−18Sbi primer pair, which allows amplification of an 18S rRNA gene fragment of arthropods. Consensus forward primer 18Saidg (5′-TCTGGTTGATCCTGCCAGTA-3′) was determined after alignment of 18S rRNA sequences of Drosophila melanogaster (GenBank accession number M21017) and Aedes aegypti (GenBank accession number M95126). The consensus reverse primer 18Sbi primer was that described by DeSalle et al. [20].

Serologic procedure

B. henselae Houston (strain Houston-1, ATCC 49882), B. henselae Marseille (strain URLLY 8, CIP 104756) and B. clarridgeiae (strain Houston-2, ATCC 51734) were propagated in 150-cm2 culture flasks containing ECV 304 human endothelial cell monolayers for use as antigens for an immunofluorescence (IFA) assay. Antigens were applied by a pen nib to each well of 30-well microscope slides (Dynatech Laboratories Ltd, Billingshurst, UK), air-dried, and fixed in acetone for 10 min. Each antigen was applied at different sites in each well to enable sera to be tested against each antigen simultaneously. Sera were diluted 1 : 50 and 1 : 100 in phosphate-buffered saline (PBS) with 3% non-fat dry milk and applied to the wells on the slides; slides were incubated in a moist chamber for 30 min at 37 °C, and then subjected to three 10-min washes in PBS. After air-drying, bound antibody was detected using a fluorescein isothiocyanate-conjugated goat anticat IgG (Jackson Immunoresearch Laboratories, West Grove, PA, USA) diluted 1 : 300 in PBS. Incubation, washing and drying were performed as described above. The slides were mounted in buffered glycerol (Fluoprep, BioMerieux) and observed with a Zeiss epifluorescent microscope at ×400 magnification. After this screening, serial two-fold dilutions from 1 : 50 to 1 : 800 were made of positive sera, and the antibody titers were determined as above.

Statistical analysis

Pearson's chi-square test was used to compare data. A difference was considered significant when P < 0.05.


Culture and identification

Bartonella spp. were isolated from 38 of 61 blood samples (62%). Comparison of 16S rRNA gene sequences of these isolates identified three different sequences. The first sequence (16 isolates; 42%) was identical to that of B. clarridgeiae, the second sequence (15 isolates; 40%) was identical to that of B. henselae Houston-1, and the third sequence (seven isolates; 18%) was identical to that of B. henselae Marseille.

Detection of Bartonella spp. from cat fleas

ITS amplification and sequencing led to the detection and identification of Bartonella DNA from six of ten (60%) flea batches—B. henselae was detected in five of these batches and B. clarridgeiae in one (Table 1); 16S rRNA gene fragments could not be amplified from these batches, so differentiation of B. henselae Houston from B. henselae Marseille DNA was not possible.

Table 1.  Results of B. henselae detection in fleas and blood in the group of 10 cats from which fleas were collected
Flea detectionBacteremiaCats (n = 10)
B. henselae Houston2
B. henselaeB. clarridgeiae1
B. henselae1
B. clarridgeiae1
B. henselaeB. henselae Houston3

Serologic analysis

An IFA titer of ≥ 1 : 50 was considered as positive. In total, 37 (61%) sera were positive for at least one antigen, with titers ranging from 1 : 50 to 1 : 400. Differences in titers between the three antigens were easily determined, even when they differed by only one serum dilution, because all antigens were read simultaneously, having been spotted into the same well on the microscope slide. Most cat sera were reactive to the three antigens tested (Table 2). Nevertheless, five sera were reactive to B. clarridgeiae only, and 11 to B. henselae only. With the exception of one serum, for which antibodies to B. henselae Marseille only were detected, all sera reactive to B. henselae were reactive to both serogroups of the species. Among the 32 sera with antibodies against B. henselae, 27 had the same antibody titer to both serogroups, one had an antibody titer to Houston serotype higher than to Marseille serotype, and four had an antibody titer to Marseille serotype higher than to Houston serotype.

Table 2.  Patterns of reactivity according to the antigen tested and results of blood cultures for the 61 tested cats
Blood culturesAntibody detectionCat (n = 61)
  1. BHH, B. henselae Houston antigen; BHM, B. henselae Marseille antigen; BC, B. clarridgeiae antigen.

B. henselae Houston+++6
B. henselae Marseille3
B. clarridgeiae+++9

Correlation between bacteremia, detection of Bartonella in fleas and antibody titers

Three of the four negative flea batches were obtained from bacteremic cats. Two of the six positive flea batches, including that from which B. clarridgeiae DNA was amplified, were obtained from blood-culture-negative cats. Four of the six positive batches were obtained from bacteremic cats. For three of these cats, the same species, B. henselae, was isolated in blood and detected in fleas. For one cat, B. clarridgeiae was isolated from blood and B. henselae was detected in infesting fleas. As three of five bacteremic cats were infested with fleas infected with the same species as that isolated in blood culture, and three of five blood-culture-negative cats were also infested with Bartonella-carrying fleas, we concluded that there was no apparent correlation between the bartonella bacteremia status of the cat and the presence of Bartonella spp. within infesting fleas.

Among the 37 cats seropositive against Bartonella spp., 12 had negative blood cultures and 25 had positive blood cultures. Twenty-one of 38 cats possessed antibodies only against the Bartonella spp. isolated from their blood, whereas 12 of 23 non-bacteremic cats possessed antibodies which reacted with at least one species antigen (P = 0.8). Of the 22 cats which had yielded B. henselae, ten possessed B. henselae-specific antibodies, whereas of the 16 cats which were infected with B. clarridgeiae, 11 had B. clarridgeiae-specific antibodies (P = 0.15). Thus, the results of blood culture were not apparently correlated with the results of serology.


In this study, more than half (62%) of the cats were bacteremic due to Bartonella spp. This infection rate is similar to those found in previous studies of stray cats (Table 3), which have reported prevalences of 22–59% [7–9,13]. Factors which appear to influence the prevalence of bacteremia include cat age, presence/absence of flea infestation, and the geographic region of the survey [6,8]. The Bartonella spp. isolated in this study belong to three different genotypes, namely B. clarridgeiae and the two variants of B. henselae. No simultaneous infection with different genotypes was observed, although this occurrence has previously been described [6,14,21]. In our study, DNA was amplified from sweeps of at least 50 colonies with no subsequent ambiguity in the 16S rDNA sequences, but this approach does not preclude the possibility of co-infection if colonies of one genotype significantly outnumbered those of another.

Table 3.  Results of blood culture, serology and DNA detection of Bartonella from cats in different studies, with most remarkable data reported in each study
ProcedureSpeciesCats (n)Isolation (%)Antibodies (%)FleasMost remarkable dataRef.
  1. LC, lysis–centrifugation; BCB, inoculation in blood culture bottle; FHB, frozen heparinized blood; BP, blood plating.

LC–BPB. henselae195989.5NDBacteremia up to 12 months[34]
252848 Increase in blood culture and serology
positivity in cats from CSD patients
LC-BPB. henselae
B. koehlerae
6141NDNDDetection of a new genotype
later named B. koehlerae[35]
Isolation of B. henselae from fleas
LC-BPB. henselae I, II13322507/27 (26%)Detection of B. clarridgeiae in fleas[21]
B. clarridgeiae    Detection of two B. henselae genotypes 
LC-BP, BCBB. henselae I, II9453NDNDDetection of two B. henselae genotypes[7]
B. clarridgeiae    Improvement of B. clarridgeiae by the
use of blood culture bottles
LC-BPB. henselae20539.581NDBacteremia associated
with young cats and antibodies, especially high titers
FHB-BPB. henselae I, II6162616/10 (60%)Detection of two B. henselae genotypesThis study
B. clarridgeiae    Detection of B. clarridgeiae in fleas
Lack of correlation between
bacteremia, serology and PCR on fleas

This work confirms the report of Brenner et al. [22], who demonstrated that prior freezing of whole blood improved recovery rates of B. henselae. Before implementing this procedure, the same cat population had failed to yield any Bartonella isolates (unpublished data). This procedure also facilitated the isolation of B. clarridgeiae, and we obtained 16 isolates using this approach, whereas Heller et al. [7] were unable to isolate B. clarridgeiae by direct blood plating. The efficacy of this procedure is thought to be due to the lysis of erythrocytes, within which B. henselae persists [23]. That the isolation of B. clarridgeiae was also achieved by this procedure may also be due to the release of this species from an intraerythrocytic location.

Although kittens and young cats have been found to be more frequently bacteremic than older cats [7,9,24], a lack of perinatal transmission of B. henselae in experimentally infected cats has been demonstrated [25]. Transmission of B. henselae between cats has been demonstrated to occur through cat flea contact [25,26], and as transmission by flea bite was not observed in an experimental model, it is considered that B. henselae is excreted in flea feces and enters its host via broken skin associated with cuts or scratches [27]. Transmission of B. clarridgeiae has not been previously investigated, but the detection in this study and the study of Bergmans et al. [21] of this species by PCR amplification within cat fleas demonstrate that this vector may well be involved.

Interestingly, the cat infested with B. clarridgeiae-positive fleas was bacteremic due to B. henselae, demonstrating that bacteremic cats can harbor fleas infected with another Bartonella species. This observation may be explained by the fact that after cat fleas have fed on bacteremic cats, they excrete viable Bartonella bacteria for up to 9 days [27,28] on uninfected or previously infected cats.

Detectable antibodies to Bartonella spp. were found in 61% of the cats, a prevalence similar to those reported in other studies (3.7–81%) [8,29–31]. The antibody response to B. henselae has been demonstrated to occur 2 weeks after infectious challenge [32]. As cats can remain bacteremic for up to 32 weeks [25], the presence of both bacteremia and elevated antibody titers is a common feature. Most cats had elevated antibodies to B. henselae and B. clarridgeiae, although, as 16 cats had antibodies to only one species, this pattern of dual infection probably reflects infections by both species rather than serologic cross-reactivity. Whether these infections are simultaneous or successive is, however, unclear. These results are in accordance with the study of Yamamoto et al., who demonstrated lack of heterologous protection between B. clarridgeiae and each of the two serotypes of B. henselae[33]. This finding, together with the fact that the two serotypes of B. henselae are human pathogens, as is probably B. clarridgeiae, suggests that cats could constitute a reservoir of three different pathogens with no cross-protection. It is not known whether human patients with a disease caused by one type of B. henselae can subsequently experience an infection with the other, but it is likely, as in the cat, that there is no cross-protection. Any vaccine strategy needs to take this problem into account.

In our study, the results of blood culture did not correlate with Bartonella DNA detection within fleas, as this detection occurs equally in bacteremic and non-bacteremic cats. Furthermore, the blood-infecting bacteria were often different from those detected within fleas. The presence of specific antibodies is also not correlated with bacteremia, as seroprevalence does not differ in bacteremic and non-bacteremic cats and is not related to the species isolated in blood culture. Thus, results of serology and Bartonella DNA detection in fleas are not predictive of the presence or the identity of infecting bacteria. The lack of correlation between serological results, results of blood cultures and Bartonella DNA detection within fleas also demonstrates that possible strategies to prevent the spread of infection among cats are to eliminate flea infestation or to develop multivalent vaccines that ensure protection against B. clarridgeiae and both genotypes/serotypes of B. henselae.