Background

The genus Bartonella (family Bartonellaceae; order Rhizobiales) comprises facultative intracellular alphaproteobacteria. An increasing number of Bartonella species are recognized as zoonotic pathogens [1]. In humans, bartonellosis can have a variety of mild and unspecific clinical signs and symptoms, but can also be life threatening [2, 3]. Bartonellae can be transmitted to humans indirectly from blood-sucking arthropod vectors through the scratches of an infected reservoir host (e.g. cats) or via contact with infectious faeces of arthropod vectors. Fleas play a major role in the transmission of Bartonella, especially the cat flea (Ctenocephalides felis), one of the most common flea species in central Europe, which is host opportunistic, and thus a common source of infection of Bartonella, and especially Bartonella henselae [4]. Direct transmission through contact with reservoir hosts can not be excluded as a possible transmission path, although it is considered highly unlikely.

A wide variety of mammals are suspected of being reservoir hosts of Bartonella spp. [2]. Among mammals, small mammals, including bat and rodent species, are the group that harbours by far the highest diversity of Bartonella spp. [5]. Moreover, high prevalences of Bartonella spp. have been detected in rodents, the most speciose group of mammals [6]. As many rodent species live in proximity to humans in many parts of the world, human-rodent interactions are not unlikely. Nonetheless, very low numbers of clinical cases of bartonellosis in humans have been reported in the context of rodent-associated Bartonella. Thus, the purpose of this review is to determine explanatory factors for this unexpected finding. Therefore, we decided to explore (i) potential risk factors for humans; (ii) clinical cases described in recent years connected with rodent-associated bartonellae; and, additionally (iii) the growing trend in the scientific literature of newly described Bartonella taxa, including the reporting of a high number of Candidatus species. This review is focussed on rodent, lagomorph and other small mammal species (with the exception of bats) as they may be sympatric and share the same ectoparasites and Bartonella species.

Current knowledge on rodent reservoirs of Bartonella spp.

Global distribution of Bartonella spp. in small mammals

Rodent-associated bartonellae are distributed worldwide and have been the subject of research on almost every continent. For a systematic review of papers describing the detection of Bartonella in small mammal species, the following search engines were used: Google Scholar, PubMed, and Google. The following terms were searched for, solely or in combination: ‘Bartonella’; ‘rodent’; ‘small mammal’; and, in addition [name of any country in the world]; or [name of any continent in the world]. Furthermore, a separate search was conducted with the search term [name of any small mammal genus] in combination with one of the following terms: [Bartonella]; [Bartonellae]; [Bartonellosis]. Only studies published in the English language and in peer-reviewed journals were taken into consideration. In total, 132 studies were included in the analysis, representing research on Bartonella in a total of 231 small mammal species and subspecies (excluding bats). Research on Bartonella in small mammals has been conducted in 67 of the 195 countries (34.4%) of the world. Most of these studies were conducted in North America [in both Canada and the USA (100%)], followed by Europe [25 out of 44 countries (54.6%)] and Asia [19 out of 48 countries (39.6%)]. The continents/regions for which the lowest numbers of studies were reported are as follows: Africa [15 out of 54 countries (27.8%)], Oceania [two out of 14 countries (14.3%)], and Latin America and the Carribean [four out of 33 countries (12.1%)] (Additional file 1: Table S1). Thus, small mammals in a large number of countries have not yet been investigated for the presence of Bartonella spp. Most of these countries are located in Latin America, the Middle East, and Central Africa, and the lack of published data from them might be partly due to their economic and/or political situation. However, it is considered important that studies are especially carried out in countries in Central Africa, as they are among those with the lowest health coverage [7]. People from these areas make up a large proportion of those who most frequently need treatment for neglected tropical diseases [7], and infections with rodent-borne Bartonella spp. can also be expected to occur more frequently in these areas. It is also worth noting that studies were more frequently conducted in some countries than in others (e.g. there were 17 from the USA but only one from Argentina). The studies undertaken in the USA were conducted thoroughly and, in total, reported 25 small mammal species positive for Bartonella (Fig. 1).

Fig. 1
figure 1

The number of Bartonella-positive small mammal species per country (listed in Additional file 1: Table S1)

The most frequently studied genus was Rattus, and in particular the two cosmopolitan species Rattus norvegicus and Rattus rattus. Members of the genera Apodemus, Bandicota, Microtus, Mus, and Myodes were very often associated with Bartonella spp. The most studied rodent species was R. norvegicus (43 studies from four of the seven continents), followed by R. rattus (41 studies), Mus musculus (25 studies) and Clethrionomys glareolus (24 studies). The five Bartonella species most frequently detected in small mammal hosts were Bartonella grahamii (found in 53 small mammal species, and in 31 countries), Bartonella elizabethae (found in 43 small mammal species, and in 34 countries), Bartonella tribocorum (found in 30 small mammal species, and in 27 countries), Bartonella taylorii (found in 27 small mammal species, and in 21 countries), and Bartonella queenslandensis (found in 29 small mammal species, and in 13 countries). Small mammals positive for Bartonella were found in 65 (97.0%) of 67 investigated countries (they were not found in Hungary and Pakistan). The five most frequently listed Bartonella species in Additional file 1: Table S1 may, however, be a distortion as, for example, B. elizabethae is one of the first rodent-associated species to have been described (in 1993) whereas other rodent-associated species such as B. kosoyi were not described until much later (in 2018). Furthermore, it should be noted that various methods were used in the studies, and that not all the published sequences had a homology of 100%.

Clinical cases of bartonellosis in humans in the context of rodent-associated Bartonella spp.

Clinical symptoms/clinical cases, diagnostics, and pathogenicity

The most frequently described Bartonella species pathogenic for humans include the human-specific species Bartonella bacilliformis (transmitted via sand flies), which causes Carrion’s disease in South America; the zoonotic, cat-transmitted species B. henselae, which is responsible for cat scratch disease; and the human-specific species Bartonella quintana (transmitted via body lice), the causative agent of trench fever [8]. Much less is known about human infections with other Bartonella spp. In general, endocarditis, lymphadenopathy and neuroretinitis are common symptoms of severe cases of bartonellosis [9]. A detailed PubMed analysis (performed on 2 October 2021) with the search terms ‘Bartonella [species]’ where the species was one of the 33 Bartonella species given in Table 1 [e.g. (‘B. alsatica’) and (‘infection’)] revealed only 14 publications citing evidence for human infections (see Table 2).

Table 1 Published Bartonella species and subspecies, the status of their published name according to the International Code of Nomenclature of Prokaryotes (ICNP), year first described, and first-mentioned host reservoir(s) together with their category and taxonomic order
Table 2 Human infections by small mammal-associated Bartonella spp.

When high diagnostic standards were applied (including direct pathogen detection via culture or PCR), only eight Bartonella species or subspecies (Bartonella vinsonii subsp. arupensis infections, B. elizabethae infections, Bartonella alsatica infections, B. tribocorum or B. vinsonii subsp. vinsonii infections, Bartonella doshiae, B. grahamii, Bartonella rattimassiliensis) were reported (in total, 24 confirmed patient cases; see Table 2).

The analysis of the frequencies of clinical entities showed that 12 patients suffered from acute febrile illness (most likely associated with bacteremia/blood stream infections) (50%), three patients from endocarditis or prosthetic valvular graft infections (12.5%), three patients from lymphadenopathy (12.5%), two from nonspecific symptoms (8.3%), and one each from bacillary angiomatosis, hepatic lesions or neuroretinitis (4.2%). Two of these patients were immunocompromised (human immunodeficiency virus infection, leukemia); no clear association with an underlying comorbidity was reported for the remaining patients. The reported antibiotic therapy regime varied but often included the administration of a macrolide combined with doxycycline for some weeks, which often resulted in clinical improvement.

From a clinical point of view, ‘acute febrile illness’ and ‘endocarditis’ can be classified as ‘bacteremia/blood stream infections’, which 15 of 24 reported patients (62.5%) suffered from. Although these cases were anecdotal, it can be suggested that human infections by rodent-associated Bartonella spp. are rare. To our knowledge, there are several possible reasons for this low number of case reports: (i) physicians are very likely unaware of Bartonella infections (especially when rodent-associated), and thus do not include them in their differential diagnosis; (ii) laboratories may not able to detect these pathogens due to their fastidious nature, and because their diagnostic portfolio does not include PCR tests for the detection of Bartonella spp. or they do not carry out long, sterile microaerophilic incubations for the cultivation of samples from patients. Moreover, (iii) bartonellosis might only cause mild and unspecific symptoms; and (iv) rodent-associated Bartonella infections of humans may simply be rare medical entities. A possible molecular explanation for the latter is the host restriction of Bartonella species mediated by their respective Trw type IV secretion systems (T4SSs) [10]. The Trw T4SS (originally described as a plasmid conjugation system) is crucial for adhesion to erythrocytes and subsequent erythrocyte invasion, and Bartonella with mutations in the trwE gene (signature-tagged mutagenesis) are unable to establish long-lasting bacteremia in certain rodent infection models. It has been demonstrated that the Trw systems of certain Bartonella spp. are responsible for species-specific host-restricted adhesion to erythrocytes. For instance, the Trw T4SS of B. tribocorum mediates a significant bacterial infection in Wistar rats, but infection human erythrocytes is 23 times less efficient. It seems likely that infections of humans by rodent-adapted Bartonella spp. rarely occur because the rodent-pathogen Trw system and human erythrocyte host receptors simply do not match.

Groups at risk of Bartonella infection in the context of rodent-associated Bartonella spp.

Many Bartonella species are pathogenic for humans. However, B. henselae, B. bacilliformis and B. quintana cause most cases of Bartonella disease in humans [8, 11]. Veterinarians, veterinary nurses and people that work with and care for animals seem to be at increased risk of infection as they are particularly exposed to reservoir hosts and vectors of Bartonella spp. [12,13,14]. Oteo et al. [15] found that 11.2–56% of tested veterinary professionals in Spain showed seroreactivity for B. henselae, B. quintana, and/or B. vinsonii berkhoffii. Bartonella spp. were even isolated from 7.9% of the positive individuals, although all of them were asymptomatic [15]. Bartonella henselae is also reported to have possibly contributed to the death of two veterinarians [16]. Cat and dog owners also appear to be at increased risk of infection. Transmission of B. henselae is associated with scratches received from both cats and dogs [17]. Owners of a cat ≤ 12 months old have an increased risk of infection with B. henselae compared to those with a cat > 12 months old [18]. Forest workers and orienteers seem to be the other groups at risk [19, 20]. Furthermore, a higher risk of infection has also been described for homeless people, alcoholics, and drug addicts who administer substances intravenously [21, 22]. Though an intravenous transmission route seems unlikely, one study did show that drug addicts who administered substances intravenously were more at risk of contracting Bartonella spp. [20]. Infestation with ectoparasites such as lice and fleas due to poor hygiene may also lead to bartonellosis, especially in homeless people [23].

The risk factors for rodent-associated Bartonella infections in humans are similar to those mentioned above for Bartonella transmitted via other animals (Table 2). Most of the patients listed in Table 2 were either young, old, pregnant or immunosuppressed. We assume that inclusion in one of these groups is a risk factor for developing clinical symptoms after infection with rodent-associated bartonellae because these groups are associated with an impaired or not yet fully developed immune system. Thus, rodent-associated bartonellosis seems to be opportunistic and might be more likely to develop when a person has a pre-existing medical condition. Furthermore, the studies showed that being homeless [24], abusing drugs [20], or being in contact with animals, e.g. through hunting or animal breeding, may increase the risk of rodent-associated Bartonella infection.

Reservoir role and clinical cases of pet animals infected with rodent-associated Bartonella spp.

Cats are known hosts of B. henselae (and Bartonella clarridgeiae and Bartonella kohlerae) and dogs of Bartonella rochalimae. Thus far, there have only been occasional reports of clinical symptoms in cats and dogs related to Bartonella spp. infection, and even fewer related to rodent-associated bartonellae. Whether bartonellae are primary or opportunistic pathogens for cats and dogs is not entirely clear. Clinical manifestations of bartonellosis are rarely seen in domestic cats, and to the best of our knowledge, there have been no case reports of rodent-associated bartonellosis in them. However, there is one report of a cat which was thought to have transmitted rodent-associated B. grahamii to a human via a scratch [25]. Unlike cats, dogs may develop severe clinical symptoms of bartonellosis that are similar to those displayed by humans [26]. Thus far, rodent-associated B. elizabethae, B. grahamii, B. taylori and a Bartonella volans-like strain have been detected in dogs [27,28,29]. However, only B. elizabethae infections could be linked directly to a canine clinical case. An 8-year-old dog suffering from unspecific symptoms including lethargy, appetite and weight loss was diagnosed with B. elizabethae infection in the blood stream. The dog died immediately before the diagnosis was confirmed, and no other pathogen was detected in the blood [28]. Furthermore, there is one record of a dog with a previously unspecified clinical record which was found to be positive for a strain of B. volans after its death [29]. Bartonella grahamii, B. elizabethae, B. taylorii were found to have a moderate prevalence (9.4%) in stray dogs without a clinical record in Thailand, highlighting the potential reservoir competence of dogs for rodent-associated bartonellae [27].

Current insights into Bartonella taxonomy with a focus on recently discovered small mammal-associated Bartonella spp.

Bartonellae can be divided into eubartonellae and other ancient clades according to their genetic features [5]. Eubartonellae can further be subdivided into four lineages, one of which is the most diverse with regard to potential host species as well as species and subspecies of Bartonella. There are presently 84 known species and subspecies of Bartonella, of which 38 were initially found in specimens belonging to the order Rodentia, followed by 10 species each in specimens of the orders Carnivora and Artiodactyla, and seven in humans. The number of newly described species has been increasing in the past decade [5]. Forty-five new Bartonella species have been proposed and/or published since 2011 (Table 1). Most of these newly described Bartonella species (n = 30) were reported in wild small mammal or specifically rodent species such as Acomys russatus (golden spiny mouse), Mastomys erythroleucus (Guinea multimammate mouse), and Pachyuromys duprasi (fat-tailed gerbil) (Table 1). Out of the 51 proposed or published rodent- or other small mammal-associated species, 19 have Candidatus status.

There is an increasing number of newly discovered, not yet fully characterized, Bartonella species with Candidatus status. Labelling a potentially new species Candidatus is a new concept that began in the 1990s [30] and allows researchers to propose prokaryotic taxa that are well characterized but as yet uncultured. In contrast to official species names, there is no official process for reviewing proposed Candidatus species and their names before they are published (A. Oren, personal communication). Authors are welcome to submit the proposed names to the Judicial Commission on Prokaryote Nomenclature of the International Committee on Systematics of Prokaryotes (ICSP), but they are not obliged to do so. The committee reviews Candidatus species through an extensive literature review, as many Candidatus names have not been qualitatively validated or do not follow the rules of the International Code of Nomenclature of Prokaryotes [31], which explains why many Candidatus names in use for newly described Bartonella species are malformed. However, suggested corrections for these malformed names are regularly published [32]. Nevertheless, due to the large number of new Candidatus taxa being proposed, particularly those with rodents and other small mammals as their origin and likely reservoir hosts, the lists of corrected names are not exhaustive and constantly evolving [31, 32].

Thus, the above should be borne in mind with respect to newly described Candidatus species and their published names. A mandatory submission process for the validation of species names prior to the publication of newly proposed Candidatus species would help to avoid the time-consuming renaming process carried out by the ICSP, and would further help to avoid the circulation in the literature of malformed Candidatus names. For example, the ICSP proposes that ‘Bartonella bandicootii’ should be re-named ‘Bartonella paramelis’ (Table 1). ‘Bandicoot’ is the English name of the animal from which this species of Bartonella was isolated, but the proposed specific epithet ‘bandicootii’ (rather than ‘bandicooti’) is both malformed and in violation of recommendation 6 (3) of the ICNP. The genus name of the host animal, which is Parameles, should be used as the basis for the specific epithet of the Bartonella species rather than the English common name, bandicoot, which is why the current name of this species should be Bartonella paramelis [33].

Furthermore, an increasing number of Bartonella species without Candidatus status have not been validly published according to the ICNP. In total, there are 44 Bartonella species names currently in use that have not been validly published. Of these species, 40 have been described since 2007, and 17 are from rodents or other small mammals (Table 1). As the number of newly described Bartonella species is increasing, in particular with respect to those isolated from rodents and other small mammals, it would be helpful if their names were proposed for approval under the ICNP before they were published, and that these should only be published when the relevant dataset allows these Bartonella species to be clearly differentiated from known species and subspecies of the genus. At present, it is not always easy to define a species or subspecies, but with whole genome sequencing becoming more affordable, it is highly likely that this issue will be resolved in the future through the sufficient description of new Candidatus species through the use of this genetic technique [34].

Further reasons for low numbers of cases of bartonellosis in humans despite high prevalences of Bartonella spp. in rodents and other small mammals

At first glance, it appears to be a phenomenon that Bartonella spp. from small mammals seem to be the least pathogenic for humans and their companion animals, and especially so in the case of rodents, as these comprise the group of small mammals with the highest prevalences and diversities of recorded Bartonella species, are commonly found in urban areas, and are known to harbour and spread disease [35]. Recent work demonstrated that one third of Norway rats in the Belgian region of Flanders harboured B. tribocorum, yet no human cases of infection with this parasite have been reported there [36]. Although it does not seem plausible that there is no transmission of Bartonella spp. from rodents to humans, in particular when taking into consideration the prevalences of Bartonella spp. in rodents and the frequent proximity of the latter to humans, there are several factors that might account for this.

Rodent-associated Bartonella spp. are probably arthropod-borne pathogens, but transmission through rodent scratches or bites (as is often reported for B. henselae infections) cannot be completely ruled out [1, 37]. Several studies have reported possible vertical transmission in naturally infected rodents [37,38,39]. However, a study on infected Clethrionomys glareolus could not experimentally prove transplacental or transovarial transmission of B. taylorii and B. grahamii to the offspring. There is one report of possible vertical transmission of B. birtlesii in BALB/c mice [40]. Most rodent-associated flea species are host specific and do not infest humans, which further limits potential zoonotic transmission of Bartonella spp. However, a few flea species are known to be more host opportunistic [41]. Yersinia pestis, the causative agent of the plague, is known to be transmitted from rats to humans mainly through Xenopsylla cheopis, the oriental rat flea [42], which can cause severe outbreaks of the disease. Xenopsylla cheopis is also known to harbour DNA of zoonotic Bartonella spp. such as B. elizabethae, and it was experimentally shown that X. cheopis can excrete this species over time [43, 44]. Nonetheless, there are, to the best of our knowledge, no reported cases of bartonellosis in humans previously infested with this flea species.

Another possible means of infection is direct contact with a reservoir host. Cats may transmit B. henselae to humans through their scratches [45]. There is even one report of a cat transmitting B. grahamii, which is a rodent-associated pathogen, to its owner [25]. The explanation for this was that the cat came into contact with an infected rodent when it caught it with its paws, which then became infectious. As rodents are not predators of humans and tend to avoid them, direct contact between the two is highly unlikely, with the exception of small mammals kept as pets [46]. However, indirect transmission via a cat is a possible, though unlikely, transmission path [25].

Notwithstanding the cases described above, rodent-associated bartonellae are the principal cause of bartonellosis in humans. However, how and to what extent rodents and their ectoparasites are involved in the zoonotic transmission cycle of Bartonella is not fully understood. Studying the transmission of Bartonella spp. from rodents to humans would help in assessing the potential risk of the former for the latter.

Conclusions

The zoonotic transmission cycle of rodent-associated bartonellae is not fully understood. It is especially unknown whether Bartonella infections in humans arise from direct contact with small mammals or rather indirectly via infestation with a rodent-associated ectoparasite vector. The total number of confirmed human cases of bartonellosis worldwide caused by rodent-associated bartonellae is much lower than expected when taking into account the abundance of Bartonella spp. in rodents, and possible reasons for this have been discussed in this review. Many small mammal species are considered reservoir hosts for the increasing number of newly described Bartonella spp., although some of the latter have yet to be experimentally confirmed. Even though most Bartonella spp. show high adaptation to their hosts, this is not the case for all rodent-associated species. That is why further experimental research is needed to increase our understanding of host–pathogen interactions between rodents and Bartonella species.