Antibodies reactive with Bartonella henselae and Ehrlichia canis in dogs from the communal lands of Zimbabwe

INTRODUCTION Ehrlichia canis is a Gram-negative bacterium that is an agent of canine monocytic ehrlichiosis. The organism is transmitted by Rhipicephalus sanguineus and, in the acute stage of infection, there is often fever, anorexia, lymphadenomegaly, splenomegaly and thrombocytopaenia. Most dogs survive the acute stage and enter the subclinical phase of the disease, which might last for years. During this phase, animals are apparently healthy although thrombocytopaenia is common. Dogs may spontaneously eliminate E. canis during the subclinical phase or go on to develop the chronic phase of the disease in which there is marked weight loss and signs resulting from pancytopaenia. While serosurveys have shown that high percentages of urban dogs in Zimbabwe have antibodies reactive with E. canis, there are no published data on infections in dogs in the rural communal lands. These are areas where traditional subsistence agro-pastoralism is practised and which contain 70 % of the national dog population. Bartonella henselae is a Gram-negative bacterium that is an emerging human and veterinary pathogen worldwide. The domestic cat is the natural host of B. henselae and high percentages of cats have asymptomatic bacteraemia which may persist for years. Seropositive animals have been found in Zimbabwe (24 %; 28/119) and South Africa (21 %; 11/52) and B. henselae has been isolated from the blood of domestic cats in Zimbabwe (10 %; 3/30) and South Africa (3 %; 1/31). In humans, B. henselae is an important emerging zoonotic agent that is often associated with contact with cats and their fleas. The organism has been implicated as an agent of an ever-increasing spectrum of diseases including cat-scratch disease, bacillary angiomatosis, endocarditis, bacteraemia, encephalopathy, neuroretinitis, osteomyelitis and peliosis hepatis. Immunosuppressed people are often at particular risk of infection with B. henselae and a recent study in South Africa found 10 % of outpatients attending HIV clinics in Johannesburg had Bartonella bacteraemia. Bartonella henselae is transmitted between cats by the cat flea, Ctenocephalides felis, which has a broad host range and is also commonly found on dogs. Recent studies have shown that B. henselae can infect dogs and cause clinical disease. Ctenocephalides felis is a common ectoparasite of dogs in southern Africa, particularly in dogs in the communal lands of Zimbabwe where ectoparasite control is seldom used. To determine if antibodies to B. henselae were present in dogs in Zimbabwe, a serosurvey on dogs from 5 widely separated communal lands was conducted. The dogs were also tested for antibodies to E. canis to provide data on the prevalence of canine ehrlichiosis in the rural areas of Zimbabwe. The results of our surveys are described in this report.


INTRODUCTION
Ehrlichia canis is a Gram-negative bacterium that is an agent of canine monocytic ehrlichiosis 29 .The organism is transmitted by Rhipicephalus sanguineus and, in the acute stage of infection, there is often fever, anorexia, lymphadenomegaly, splenomegaly and thrombocytopaenia.Most dogs survive the acute stage and enter the subclinical phase of the disease, which might last for years.During this phase, animals are apparently healthy although thrombocytopaenia is common.Dogs may spontaneously eliminate E. canis during the subclinical phase 6,21 or go on to develop the chronic phase of the disease in which there is marked weight loss and signs resulting from pancytopaenia.While serosurveys have shown that high percentages of urban dogs in Zimbabwe have antibodies reactive with E. canis [36][37][38] , there are no published data on infections in dogs in the rural communal lands.These are areas where traditional subsistence agro-pastoralism is practised and which contain 70 % of the national dog population 10 .
Bartonella henselae is a Gram-negative bacterium that is an emerging human and veterinary pathogen worldwide 23 .The domestic cat is the natural host of B. henselae and high percentages of cats have asymptomatic bacteraemia which may persist for years 32 .Seropositive animals have been found in Zimbabwe (24 %; 28/119) and South Africa (21 %; 11/52) 27 and B. henselae has been isolated from the blood of domestic cats in Zimbabwe (10 %; 3/30) 28 and South Africa (3 %; 1/31) 44 .
In humans, B. henselae is an important emerging zoonotic agent that is often associated with contact with cats and their fleas 23 .The organism has been implicated as an agent of an ever-increasing spectrum of diseases including cat-scratch disease, bacillary angiomatosis, endocarditis, bacteraemia, encephalopathy, neuroretinitis, osteomyelitis and peliosis hepatis 27 .Immunosuppressed people are often at particular risk of infection with B. henselae and a recent study in South Africa found 10 % of outpatients attend-ing HIV clinics in Johannesburg had Bartonella bacteraemia 19 .
Bartonella henselae is transmitted between cats by the cat flea, Ctenocephalides felis 14 , which has a broad host range and is also commonly found on dogs 49 .Recent studies have shown that B. henselae can infect dogs and cause clinical disease 20,31,39 .Ctenocephalides felis is a common ectoparasite of dogs in southern Africa 12 , particularly in dogs in the communal lands of Zimbabwe where ectoparasite control is seldom used.
To determine if antibodies to B. henselae were present in dogs in Zimbabwe, a serosurvey on dogs from 5 widely separated communal lands was conducted.The dogs were also tested for antibodies to E. canis to provide data on the prevalence of canine ehrlichiosis in the rural areas of Zimbabwe.The results of our surveys are described in this report.

Indirect fluorescent antibody assays
Bartonella henselae (Houston-I; ATCC 49882) was grown in Vero cells as described previously 27 .When 60-90 % of the cells were infected they were pelleted, washed and resuspended in phosphate buffered saline, applied (5µ aliquots) to the wells of 32 well Teflon slides and air-dried.E. canis (Oklahoma) was grown in DH82 continuous cell cultures, harvested and applied to the wells of Teflon-coated slides as described previously 38 .
Reactive antibodies were detected against B. henselae and E. canis using previously reported IFA procedures 27,37 and fluorescein isothiocyanate-labelled protein G conjugate (Biogenesis Inc, Sandown, NH, USA).Based on the results of previous studies, sera with IFA titres of ³1:80 were regarded as positive for previous exposure to Bartonella spp. 24,27or to Ehrlichia spp. 37,38.

Sera
Sera were obtained from 228 dogs in the 5 communal lands surveyed.Over half of the dogs sampled (142/228; 62 %) were 2 years of age or younger (Table 1).

Indirect fluorescent antibody assays
Dogs with antibodies against E. canis were found in all the communal lands sampled (Table 2) and seroprevalences varied from 15 % (7/48 in Matusadona) to 88 % (37/42 in Bikita).There were no significant differences between the overall seroprevalences in the different age groups of dogs studied.High antibody titres (arbitrarily defined as ³1:640) were only found in dogs 3 years of age or younger from Chinamora (5; 50 % of the positive dogs), Chiredzi (11; 69 % of positive dogs) and Bikita (23; 63 % of the positive dogs).
Dogs with antibodies against B. henselae were found in all communal lands apart from Chiredzi (Table 3).There was no obvious correlation between age and seropositivity.The highest seroprevalence was in dogs from Bikita (41 %) where the only high titres against B. henselae, arbitrarily defined as ³1:640, were found in 4 dogs which were 2 years of age or younger.

DISCUSSION
In 1990, Brooks 10 reported that dogs under a year of age constituted 33 % of the population in Manicaland Province in the east of Zimbabwe and that only 11 % of dogs were over 4 years of age.A similar age distribution was reported subsequently in 2000 13 and has now also been found in this study (Table 1).It appears that there has been little improvement in the very rapid turnover of dogs in the communal lands of Zimbabwe in the past decade.
This study is the 1st to show that rural dogs in southern Africa have high prevalences of antibodies to E. canis.The percentage of communal land dogs we found seropositive against E. canis (34 %) is similar to that described previously for urban dogs in Zimbabwe and South Africa (33-42 %) [36][37][38]43 . Altough local experience indicates infections with E. canis are frequent in southern Africa 29,53 , there is little direct supporting evidence.There are no African isolates of E. canis and only 3 reports identifying E. canis or a closely related species on the continent.In a recent report, only a few positive results were obtained when dogs suspected of having canine ehrlichiosis in South Africa were tested for E. canis with a highly sensitive and specific PCR assay 2 .In a study using PCR and sequencing, organisms closely related to E. canis have been identified in a sheep in South Africa 1 . Usingsimilar methods, organisms closely related to E. canis have been identified in cattle ticks from Mali and Niger, but DNA of Ehrlichia spp. was not found in 86 R. sanguineus from dogs in Mali and Sudan 42 .
Although antibodies to E. canis were detected, there is serological crossreactivity between members of the genus and we could not, then, determine the  Ehrlichia spp. that had infected the dogs we studied.There is extensive serological cross-reactivity between E. canis and E. chaffeensis 4 7 , the agent of human monocytic ehrlichiosis.The organism can also infect dogs and cause clinical and pathological signs indistinguishable from those caused by E. canis 7 .Although antibodies to E. chaffeensis have been found in both dogs 43 and people in southern Africa 11,45 , all available information indicates that E. chaffeensis only occurs in the USA and it is thus unlikely that infections with the organism influenced our results.Ehrlichia ruminantium (formerly Cowdria ruminantium) is the agent of heartwater, a disease of domestic ruminants that occurs widely in Africa and is transmitted by Amblyomma spp.Experimentally infected dogs do not show clinical signs or laboratory abnormalities but become bacteraemic for up to 3 weeks and seroconvert against E. ruminantium 26 .There is extensive antigenic cross-reactivity between E. canis and E. ruminantium with dogs experimentally infected with E. ruminantium becoming positive in IFA and immunoblots against E. canis 26,38 .Recently, PCR and sequencing studies identified the DNA of an E. ruminantium in dogs in South Africa 2 and Amblyomma spp.have been found on dogs in the region 12 .Although we did not determine the ticks on the dogs in our study, Amblyomma spp.occur widely in Zimbabwe 51 and it would appear likely that at least some of the seropositive dogs had been infected with an E. ruminantium or a closely related organism.
Other recognised Ehrlichia spp. that are closely related to E. canis and have serological cross-reactivity are E. muris which infects mice, and perhaps people, in Japan 25 and E. ewingii which is an agent of canine granulocytic ehrlichiosis that has only been reported in the USA 33 .Although it is very unlikely that these organisms influenced our results, there are other incompletely characterised Ehrlichia and Anaplasma spp. that have been described in southern Africa 1,2,51 and their role as pathogens in dogs is yet to be determined.With the apparent high exposure of dogs in southern Africa to Ehrlichia spp.and/or closely related organisms, further studies are indicated to determine the organisms involved and their role in the rapid turnover of dogs in the region.
Bartonella henselae is known to infect both cats 27,28,44 and people 19 in southern Africa and our study now provides evidence that infections also occur in dogs.We found a significant prevalence (14 %) of antibodies against B. henselae in dogs from widely separated communal lands in Zimbabwe.Similar seropre-valences have been found in dogs from the United Kingdom (6.5 %) 3 , Hawaii (3 %) 18 and Japan (8 %) 52 .These findings are consistent with the reported widespread distribution of B. henselae 27 and its vector, the cat flea 49 .
There are only limited data on the effects of B. henselae in dogs.Experimental infections cause no detectable clinical signs or result in short-lived bacteraemia 31 .Natural infections have been associated with peliosis hepatis 31 , a vasculoproliferative disorder characterised by cystic, blood-filled spaces in the liver, and pyogranulomatous hepatitis 20 .Bartonella henselae has also been found in 3 dogs suffering from various conditions with a wide variety of historical, clinical, haematological, and biochemical abnormalities 39 .
While there is no serological crossreactivity between Bartonella spp.and E. canis and spotted fever group rickettsiae 9 , serological cross-reactivity has been reported between B. henselae and other members of the genus.There is extensive cross-reactivity with B. quintana 22 which causes trench fever, bacillary angiomatosis and endocarditis in people and is transmitted by the human body louse (Pediculus humanus).Although B. quintana occurs in people in Africa 46 , there are no reports of infections occurring in dogs and it would appear unlikely, then, that the organism was responsible for the antibodies we detected.
Serological cross-reactivity has also been described between B. henselae and B. clarridgeiae 50 .The cat flea is the presumed vector of B. clarridgeiae 48 and the natural reservoir is the cat, in which the organism causes a chronic asymptomatic bacteraemia 34 .Bartonella clarridgeiae also infects dogs and has been associated with vegetative endocarditis 15 and Doberman hepatopathy 20 .Although there are no reports of B. clarridgeiae in Africa, the organism occurs very widely 30 in North and South America, Europe, South East Asia and New Zealand.It would appear likely, then, that B. clarridgeiae occurs in Africa and that infections with the organism occurred in at least some of the dogs that were studied.
Other Bartonella spp.are known pathogens of dogs but appear unlikely to have influenced our results.Bartonella vinsonii subspecies berkhoffii was the 1st Bartonella spp. to be identified as a pathogen in dogs 5 and has been reported in dogs with arrhythmias and endocarditis 8 , granulomatous lymphadenitis, granulomatous rhinitis 41 and anterior uveitis and choroiditis 40 .Although an IFA study using B. vinsonii subspecies berkhoffii as antigen found 65 % of dogs from the Sudan to be seropositive 17 , antibodies to B. vinsonii subspecies berkhoffii do not react with B. henselae 9 and would not thus have been detected in our study.Further experiments using specific serological and/or molecular tests for B. vinsonii subspecies berkhoffii are required to determine if the organism occurs in southern Africa.Bartonella elizabethae has been identified in a dog with a wide variety of clinical and laboratory abnormalities 39 but the organism has only been found in the Americas 4 and there appears to be only minor serological cross-reactivity between B. henselae and B. elizabethae 35 .Bartonella washoensis has been found in a dog with mitral valve endocarditis 16 and, although serological cross-reactivity with B. henselae has not been determined, it would appear unlikely that we detected antibodies against B. washoensis as the organism has only been described in the United States.
In summary, our study has shown that dogs in widely separated rural areas of Zimbabwe, and hence probably the region, are not uncommonly infected with E. canis and B. henselae or closely related species.Veterinarians should be aware that these organisms might cause disease in their canine patients and laboratories should offer appropriate diagnostic tests.Further studies are indicated to determine the Ehrlichia spp.and Bartonella spp. that occur in southern Africa and the role they play as pathogenic agents.