Evaluation of the available animal models for Bartonella infections

Abstract

The diseases caused by the Bartonella genus of bacteria are clinically diverse, and can be challenging to cure. The study of bartonellosis has been hampered by the lack of a suitable animal model. Preclinical studies for novel therapeutics and a competent host for vector transmission studies are needed to fill critical knowledge gaps. The studies included here are a representation of in vivo Bartonella research and the corresponding challenges. This review examines the current state of available animal models by assessing the success of various model species and strains in Bartonella infection. With a focus on the strengths and weaknesses of current animal models, the importance of these models for improvement of human health and veterinary care is emphasized.

1. Introduction

Bartonellosis is a class of infections by the bacteria in the Bartonella genus. Infections with these bacteria are perceived as rare, making it difficult to study the effect of these infections on the human host. Bartonella is a genus of vector-borne, gram-negative bacteria that can be transmitted by sandflies, lice, fleas, and possibly ticks [1]. In addition, direct transmission from cats is well documented [2]. The bacteria are present in the saliva and/or feces of the vector. After the host has been bitten by the vector, the bacteria are either injected into the host by way of the saliva or introduced to the lower layers of dermal tissue by scratching infected feces into the bite site. Once the bacteria enter the bloodstream, they infect erythrocytes and travel throughout the host body. The primary niche for this genus of bacteria has not yet been identified and very well may be species or host specific. Infection with differing species can have a large impact on the pathology, diagnosis, treatment, and prognosis of a patient.

The family of diseases characterized as bartonellosis is diverse and not well-understood. As with Lyme disease, which is also vector-borne, many of the associated symptoms are non-specific and the severity of clinical presentation varies widely. This complicates the diagnosis of bartonellosis and results in only the severely ill, who are unable to clear the infection by themselves or with minimal medical intervention, being formally diagnosed. An extremely important consideration is the effectiveness of antimicrobial treatments for Bartonella infections. The multiple species of Bartonella exhibit variability in antibiotic susceptibility, and treatment is often determined by clinical manifestation rather than the infecting species [3]. Indeed, there is currently no single treatment that is effective for all Bartonella-associated diseases [4]. A reliable preclinical animal model for the testing of existing and novel therapeutics will propel this field forward. The purpose of this review is to evaluate the most commonly used animal models in Bartonella disease research, the work that has been done to develop these models, and what is lacking in the Bartonella field.

2. Human bartonellosis

A variety of disease manifestations are commonly seen in the clinical setting. A brief description of some of these manifestations is included in this section. When considering the diagnosis of bartonellosis, it is important to consider that complicated pathologies can correlate with any of multiple Bartonella species. Table 1 compares the most frequently identified clinical findings with which Bartonella species they have been associated.

Table 1.

Human disease manifestations and associated Bartonella spp. disease.

Disease	Bartonella spp.	ref
Cat scratch disease	B. henselae, B. quintana, B. clarridgeiae	[[5], [6], [7], [8]]
Carrion's disease Verruga peruana	B. bacilliformis	[9]
Trench Fever	B. quintana	[10]
Bacillary angiomatosis	B. henselae, B. quintana	[11]
Bacteremia	B. henselae, B. quintana, B. clarridgeiae, B. elizabethae, B. bacilliformis	[[12], [13], [14], [15]]
Fever of Unknown Origin	B. henselae, B. quintana, B. rochalimae B. tamiae, B. rattimassiliensis, B. elizabethae, B. tribocorum	[[16], [17], [18], [19]]
Blood-culture negative endocarditis	B. quintana, B. henselae, B. alsatica, B. koehlerae, B. vinsonii subsp. arupensis, B. vinsonii subsp. berkhoffi, B. elizabethae, B. mayotimonensis	[15,[20], [21], [22], [23], [24], [25]]
Ocular	B. henselae, B. grahamii, B. elizabethae	[[26], [27], [28]]
Neurological	B. henselae, B. quintana, B. vinsonii subsp. berkhoffi, B. koehlerae, rare rodent species of Bartonella	[[29], [30], [31], [32]]

3. Carrion's disease

Carrion's disease is a severe form of bartonellosis caused by Bartonella bacilliformis. This pathogen and its associated diseases are localized to the Andes Mountain ranges in South America. Carrion's disease can be classified into two separate disease states: the acute hemolytic phase also known as Oroya fever, with a mortality rate of 44–88%, and a cutaneous manifestation known as verruga peruana. However, with proper diagnosis and treatment, the mortality rate drops significantly [33].The clinical presentation of Carrion's disease is established and quite different from the presentation of other forms of bartonellosis. For these reasons, B. bacilliformis is not discussed in this review. For more information in Carrion's disease or B. bacilliformis please refer to Garcia-Quintanilla's “Carrion's disease: more than a neglected disease” review [9].

4. Cat scratch disease

Globally, most Bartonella infections are associated with Bartonella henselae. This species of Bartonella is transmitted to cats by cat fleas (Ctenocephalides felis) and is the causative agent of cat scratch disease (CSD) in humans. Transmission to humans occurs by bites or scratches of infected cats [34]. Reported symptoms include low grade fever, swollen lymph nodes, and a papule at the scratch site. In a healthy, immunocompetent patient, these symptoms typically subside on their own, without treatment or with minimal care [35]. However, in immunocompromised patients, these signs can progress into more severe manifestations such as endocarditis, encephalitis, bacillary angiomatosis, or neuroretinitis.

5. Trench fever

Another commonly recognized form of bartonellosis is trench fever, historically caused by Bartonella quintana-infected human body lice (Pediculus humanus corporis). Today, trench fever is seen among the homeless populations in multiple metropolitan areas [36]. In addition to the high fever associated with this disease, symptoms include severe headaches, eye pain, and intense muscle pain in the legs. The most distinguishing feature of this illness is the high fever which lasts for many days before subsiding and returning approximately a week after. This cyclic pattern can be repeated many times, leading to an extended disease course.

6. Endocarditis

Patients that present with infectious endocarditis often have a history of valve replacements, heart transplants, or otherwise compromised cardiac tissue [37]. Bartonellosis endocarditis is often difficult to diagnose and relies on blood and valve biopsy cultures. However, due to the fastidious nature of Bartonella spp., the samples are often returned as “blood culture negative endocarditis” or BCNE. Patient outcomes are heavily dependent on timely treatment. The formation of a bacterial biofilm is a likely cause of Bartonella-induced endocarditis, typically associated with B. quintana and B. henselae, but also others (Table 1). It is imperative that Bartonella infections are diagnosed prior to biofilm formation and subsequent endocarditis so that the infection can be properly treated and eradicated.

7. Bacillary angiomatosis and bacillary peliosis

Bacillary angiomatosis and peliosis describe abnormal growth of blood vessels near the skin or internal organs, respectively, which can continue to grow uncontrollably like tumors [38,39]. Bacillary peliosis is often found in blood-containing organs such as the liver and spleen [40]. This form of angiomatosis is typically found on the kidneys. When Bartonella infection (Table 1) is the etiological cause of the bacillary angiomatosis or peliosis, the bacteria can often be identified in the biopsies of the vascular tumors [[41], [42], [43]].

8. Importance of animal models

The various forms of bartonellosis are identified primarily in immunocompromised individuals. The subtle progression of the disease is not well understood, particularly when the disease becomes self-limiting or becomes latent in the vertebrate host. Evidence from many other vector-borne pathogens indicate that early diagnosis and treatment leads to better outcomes [44]. Relevant, accurate, and applicable animal models are necessary for understanding the early stages of this disease; only then will we have the tools necessary for drug development and prevention strategies. The inability to monitor the disease states of naturally occurring Bartonella infections highlights the need for the development of a proper animal model.

A complicating factor in monitoring disease states of vector-borne pathogens is the concept of reservoirs and hosts. Typically, a reservoir can remain infected with minimal to no clinical or pathological findings. The pathogen uses the reservoir host to multiply and spread to potential vectors [45]. This differs from a susceptible host, which shows signs of infection according to disease pathology. Some of the most common reservoirs for Bartonella species include cats, wild mice, dogs, rats, and macaques [46,47]. Bartonella has been identified in several other animals, but it is unknown if these animals serve as susceptible hosts or reservoirs. These include foxes, horses, bats, and deer.

A common example of this reservoir/host/vector/pathogen relationship is Borrelia burgdorferi. It is well documented that the white-footed mouse or other reservoir host is necessary for pathogen acquisition to the ticks [48]. These mice can be chronically infected without exhibiting the signs commonly associated with long-term Lyme disease such as facial palsy, arthritis, endocarditis, or inflammation of the brain and spinal cord [49,50]. In the case of B. henselae, cats do occasionally develop signs such as endocarditis, neuroretinitis, bacillary angiomatosis, and/or pyrexia, complicating the division of host and reservoir [51].

The development of an animal model is complicated by the host specificity of each pathogen and the ability for many of the host species to act as reservoirs. While the bare minimum requirement for an animal model would be successful infection, disease models would include pathologies commonly seen in human and animal patients. Infection models have been developed using rodents such as mice and rats. These models become bacteremic for acute and/or chronic infections but much like natural reservoirs, do not show clinical signs or clinical signs have not been assessed. The studies presented here aimed to identify disease models by examining infection rate, clinical signs, and/or histological changes. This review does not focus on the use of animal models for studying Bartonella biology in natural hosts [52,53].

It is unknown if animals tested in field studies have any disease pathology. The recurrent nature of Bartonella complicates field sampling as infected animals may be in a dormant stage of the disease [54]. Blood samples taken from captured animals or ectoparasites removed from the animals and can be tested [54], but these tests are only able to identify potential bacteremia in the animal, which may not be sustained [54]. Even when animals do show bacteremia, without extensive internal organ tissue biopsies, there is no way to determine if the animal is exhibiting any symptoms or microscopic pathology. There is little to no survey data regarding the health of the animal, and tissue tropism is unknown because biopsies are not taken. These factors make it difficult to determine the exact reservoir hosts for these pathogens.

There has been some success in the development of an infected model defined by bacteremia (Table 2, Table 3). However, the correlation of using mice, a reservoir host, as a model for a rodent-associated Bartonella presents a challenge. Although rodent-associated Bartonella species have been identified in human cases, many of these species are also associated with domestic animals such as dogs and cats [51,55].

Table 2.

Description of various Bartonella spp. in mouse strains and the reported results.

Mouse Strain	Bartonella spp	Inoculation route (dose if available	Methodology	Results (# bacteremic mice, length of bacteremia, PCR positives w/ or w/out tissue identification)	Clinical Findings
Swiss Webster	B. elizabethae	SC (101–106 CFU) [56]	Blood culture	1/6 at 106 [56] 1/6 at 105 [56]	Clinical signs not assessed
	B. tamiae	SC (106–107) [57]	PCR Pathology	3/12 [57]	Multiple findings in the skin, liver, kidney, heart, and spleen [57]
	B. birtlesii	IV (3 × 105CFU) [58]	Cell culture	9 weeks [58]	Clinical signs not assessed
BALB/c	B. elizabethae	SC (101–106 CFU) [56]	Blood culture	1/6 at 106 [56]	Clinical signs not assessed
	B. birtlesii	IV (3 × 105–107) [58] SC (3.4–107) [58] ID (9 × 105) [58] PO (4 × 103-3 × 108) [58] IV (5 × 107 CFU) [59]	Blood culture [58] Blood and biopsy culture [59]	5 weeks [58] Bacteremia - 40 days [59] Culture positive blood, spleen, and liver [59]	Clinical signs not assessed
	B. henselae	Blood transfusion [60] IP 104 CFU [60,61]	Nested PCR [60] Nested PCR [61]	IP 2/4 Liver & Spleen [60] Transfusion 2/4 Spleen [60] 2/4 at 4 days [61] 4/4 at 21 days [61]	Positive Skin biopsy at day 4 (confocal) [61]
C57Bl/6	B. elizabethae	SC (101–106 CFU) [56]	Blood culture	2/6 at 106 [56]	Clinical signs not assessed
	B. birtlesii	IV (3 × 105 CFU) [58] SC (5 × 103–1.4 × 107 CFU) [58] SC (3.4-800 CFU) [58]	Blood culture	IV 4 weeks [58] SC +30 days [58]	Clinical signs not assessed
OF1	B. birtlesii	IV (5 × 108 CFU) [62] Tick-feeding [62]	PCR of blood and/or blood/biopsy culture	3/3 gDNA in cultures of blood and livers [62]	Clinical signs not assessed
CD1	B. tribocorum	SC (10, 100, 1000 CFU) [63]	Blood culture	21/36 [63]	Clinical signs not assessed
Immuno-compromised	B. birtlesii	SC (6 × 105 CFU) [58] 0.12–0.4 OD600⁎ [64]	Cell culture [58] Blood culture [64] Pathology [64]	All mice up to 40 days [58] 2 months [64]	“Obvious and striking liver pathology” [64]
	B. henselae	IP (103 CFU) [65] IP (2 × 108 CFU) [66]	Nested PCR [65] PCR [66] Pathology [66]	4/4 All livers and one spleen [65] All infected livers [66]	B. henselae can infect sickle-disease RBCs [65] More infected hepatic cells in immune-compromised mice [66] Changes in cytokine and immune cell concentrations [66]
	B. taylorii	0.12–0.4 OD600⁎ [64]	Blood culture [64] qPCR [64] Pathology [64]	2 months by culture [64] +120 days by qPCR [64]	“Obvious & striking liver pathology” [64] Enlarged Spleens, granulomatous nephritis, splenomegaly [64]
	B. grahamii	0.12–0.4 OD600⁎ [64]	Blood culture [64]	2 months [64]	Clinical signs not assessed

ID – intradermal, IV – intravenous, SC – subcutaneous, IP - intraperitoneal injection routes.

* 1 OD₆₀₀ = 8 × 10⁹ genome equivalents(1).

Table 3.

Experimental infections of various animal models for Bartonellosis.

Animal Model	Bartonella species	Inoculation route (dose if available)	Successfully infected animals (culture/PCR)	Clinical Findings
Mice	Strain-dependent, see Table 2 for more details
Cats
	B. henselae [67]	ID (8 × 102 CFU, 1 × 106 CFU) IV (5 × 101 CFU, 2 × 105 CFU)	ID (8/8), IV (2/16)	Bacteremia, no clinical abnormalities
	B. henselae, multiple strains [68]	ID (6.6 × 106 CFU/ml to 9.6 × 107 CFU/ml)	12/12	Bacteremia, fever in 6/6 cats infected with feline strain
	B. henselae [69]	Donor blood (IM and IV) and urine (IM)	Donor blood (12/18) urine (0/6)	Fever (6/18), lymphadenopathy (8/18), neurological issues (2/18)
	B. henselae, B. clarridgeiae [70]	Donor blood (IV)	18/18 (multiple inoculations if negative)	Multiple histopathological findings in all animals, single cat with CNS abnormalities
	B. henselae [71]	Infected fleas ID (5 × 106 CFU) close contact	Fleas (5/5) ID (5/5) close contact (0/2)	Bacteremia, no clinical abnormalities
	B. vinsonii, B. rochalimae, B. bovis, B. quintana [72]	ID (2.84 × 106 CFU/ml to 3.6 × 109 CFU/ml) IV (2 × 104 CFU/ml)	3/33 total	Only 3 bacteremic cats, authors report multiple animals seroconverted. No clinical signs.
Dogs
	B. koehlerae, B. henselae, B. vinsonii [73]	SC (5 × 104 TCID50B. henselae) (3 × 104 TCID50B. vinsonii)	2/2 (B. henselae and B. koehlerae)	Only 1 tissue in each animal was positive post necropsy, no clinical signs of infection
	B. henselae, B. v. berkhoffi, B. rochalimae [74]	ID (2.4 × 106–1.1 × 109 CFU/ml B. henselae) (7.6 × 107–2.4 × 109 CFU/ml B. v. berkhoffi) (9.2 × 106 CFU/ml B. rochalimae)	0/6 bacteremia, but 6/6 seroconverted (B. henselae), 3/4 (B. v. berkhoffi), 2/2 (B. rochalimae)	Swelling (B. v. berkhoffi, B. rochalimae) or necrotic lesions (B. henselae) at inoculation site
Non-Human Primates
	B. quintana [75]	IV (lice intestine broth, unknown dose)	7/7 Rhesus macaques	3/7 animals had fever
	B. quintana [76]	IV (unknown dose)	4/4 Rhesus macaques	Bacteremia, no clinical abnormalities
	B. quintana [77]	ID (6.6 × 107 CFU/ml)	Single Rhesus macaque	Bacteremia, no clinical abnormalities
	B. bacilliformis [78]	ID (9.5 × 107 CFU/ml) IV (1.1 × 106 CFU/ml)	6/6 Owl monkeys (Aotus nancymaae)	Parasitemia via Giemsa staining of blood, lesions at ID sites of infection

ID – intradermal, IV – intravenous, SC – subcutaneous injection routes.

The studies presented here represent a thorough survey of the published experimental data regarding Bartonella infections. Table 2 summarizes the collection of mouse data by comparing strain of mouse used, species of Bartonella utilized, and the associated findings. This table also helps to highlight the variety of methods used to determine infection. While many of the studies presented here claim a successful animal model, the benchmarks for success vary widely. It is also important to note that customary publishing practices of presenting only positive data skew the literature. A compilation of studies that did not produce successful animal models could lead to more insights regarding Bartonella infection.

9. Description of animal models

Animals used for modeling range from small rodents to non-human primates, with each species having the ability to mimic some aspect of human disease. In the case of Bartonella, a well-established animal model has not been developed for human-specific strains. Many factors play a role in the difficulty in developing this model. Bartonella species are very host-specific when it comes to infection. Even when one species has developed the ability to infect more than one host species, the range is limited and often evolutionarily related. Fig. 1 gives an overview of animals that have commonly been used and the associated bacterial species. Here we describe experimental Bartonella infections in many mouse strains and other potential hosts and assess their suitability as an animal model.

Fig. 1.

For each host shown, the associated Bartonella species used for experimental infection is linked. Some have been tested in multiple hosts.

10. Mice

10.1. Swiss Webster

Swiss Webster (SW) is an outbred strain of mice commonly used in toxicology, cancer research, and general research. This strain of mice often becomes bacteremic with rodent-associated Bartonella spp. In one study, SW mice infected with B. birtlesii were bacteremic, according to blood cultures, for up to 9 weeks [58]. A study of SW mice infected with B. elizabethae had less success in that only 2 of the 36 mice inoculated became bacteremic [56]. This is contrary to SW mice that were inoculated with three isolates of B. tamiae, in which all 12 mice developed a thickened and tough skin at the inoculation site that resolved by week 4 post-infection. The mice also developed subcutaneous masses on the lateral thorax and inguinal lymph node enlargement, some of which persisted until the end of the study at 6 weeks. PCR analysis of the masses showed B. tamiae DNA was present in mice inoculated with each isolate [57]. Histopathology of biopsies taken during necropsy show extensive immunological changes. Granulomas were noted on the liver, lymph nodes, and kidneys. Pyrogranulomas were found in the lymph nodes and spleen. Additional histological findings included necrotizing dermatitis and myocarditis with granulomas in the right and left atria [57].

10.2. OF1

The OF1 mouse is an outbred strain commonly used in toxicology, teratology, pharmacology, and physiology. They are known for their rapid growth and high breeding success. OF1 mice were used in one study to examine the transmission potential of B. birtlesii by Ixodes ricinus ticks. A single mouse was infected by intravenous injection and was positive when tested for B. birtlesii for at least 2.5 weeks with semi-nested PCR. The mouse continued to be used as a bacteremic source of blood for the ticks. Of note, I. ricinus ticks were able to acquire the bacteria from the mice and transmit it back into naïve mice [62]. The combination of these data suggests that B. birtlesii can infect OF1 mice, yet higher numbers are needed to be certain.

10.3. BALB/c

This mouse strain is considered a multipurpose model and is used for hybridoma development, monoclonal antibody production, and infectious disease research, including multiple Bartonella infection studies. A comparison of injection methods showed that BALB/c mice were more bacteremic when subcutaneously injected with B. birtlesii compared to intravenous injection, and bacteremia persisted for 5 weeks post-injection [58]. Intradermal injection resulted in a quicker emergence of bacteremia, although the infection was cleared quicker and bacteremia was lower than in mice infected by the subcutaneous route. Another study of BALB/c mice, this time infected with B. henselae by intraperitoneal injection, showed the presence of Bartonella DNA in the spleen, liver, and skin as soon as 4 days post-injection. Blood samples taken from the mice at days 4 and 21 showed Bartonella DNA was present in all of the 21 day samples [61]. This study highlights the need for extended experiments to fully understand the complexities of Bartonella infections. In this strain of mice, B. elizabethae was able to infect 1 out of 6 mice within one week. This infection lasted 7 weeks as measured by blood culture [56]. Another study looked at infections of B. henselae by intraperitoneal injection and by blood transfusion. The study showed that 2 of 4 animals had Bartonella DNA in their spleens and one had Bartonella DNA in the liver. Blood from infected mice was transfused into naïve mice, of which 2 out of 4 mice had Bartonella DNA detected in their spleen. Interestingly, despite infectious blood being transfused, the blood collected from all recipient animals was negative by PCR [60]. Another study result showed that infection with B. henselae resulted in bacterial colonization of the liver three days after infection. Bartonella DNA was detectable from the blood and liver at 3- and 7-days post-infection, and at 3-, 7-, and 14-days post-infection in the spleen. IgM was detectable in the plasma of infected mice after 3 and 7 days but undetectable by 14-days post-infection [79]. Infection with B. birtlesii resulted in mice becoming bacteremic on average 8 days after infection and peaking on 14-days post-infection. These mice were no longer bacteremic at 10 weeks, and none of the mice relapsed in the following 4 weeks [80]. Another BALB/c mouse study aimed to evaluate the different routes of inoculation of B. henselae regarding the generation of immune responses. When mice were inoculated by intraperitoneal (IP) injection, the liver and mesenteric lymph nodes were positive as early as 6-h post-injection and the spleen, liver, and kidney were positive up to 7 days, according to PCR. Bartonella DNA was not amplified in whole blood samples of these mice. None of the inoculation routes at standard doses were able to produce bacterial colonies with culture. However, at high doses, IP-injected mice did have bacteria present in the tissues within 24 h, despite all blood cultures remaining negative [81].

10.4. C57BL/6

C57BL/6 mice are one of the most used strains in biomedical research. This breed of mouse is particularly useful due to its permissive background for transgenic/knockout model development and genetic homogeneity. In Bartonella research, C57BL/6 mice have been infected with B. elizabethae and B. birtlesii. When inoculated with B. elizabethae, 2 out of 6 mice became bacteremic at 2- and 3-weeks post-infection and remained bacteremic for 7–8 weeks. No illness was observed and no samples were collected for histopathology [56]. Conversely, B. birtlesii inoculation led to high levels of bacteremia after just 1 week but also cleared as quickly as 5 weeks post-injection. It is important to note that the total number of infected mice was not reported in this study and samples were not taken for pathology [58]. This strain of mouse has been used to develop mutant strains that are deficient in IL-10, CD4, and CD8. In one study, IL-10-deficient mice did not become infected with B. birtlesii. When compared to controls, CD8-deficient mice had no change in bacteremia while bacteremia in CD4-deficient mice indicated an increase in length and quantity of bacterial burden. When both CD4 and CD8 were knocked out, the bacteremia was intermediate to the two single knock outs, indicating that the infection is partially controlled by helper T cell responses [58]. This comprehensive analysis of mouse models compared mouse strain, route, and dose of B. birtlesii for the duration of bacteremia. The study found that the mice became bacteremic early with intradermal inoculation, but longer-lasting bacteremia was induced with subcutaneous (SC) or intravenous routes. Another study showed that SC injection of 10⁸ CFUs of B. henselae and B. grahamii were required to induce transient local inflammation [82]. Lymphadenopathy was observed in mice infected with either strain but was more pronounced with B. henselae. Lymph nodes from 5 out of 6 mice infected with B. henselae contained culturable bacteria at day 1, but by day 20 had become sterile. However, in 5 out of 15 of the mice tested, DNA could still be found at day 21 in the lymph nodes and spleen. Researchers stated that mice infected with B. grahamii showed similar results. Despite the lymphadenopathy and cultivatable bacteria in the lymph nodes and spleens, no live bacteria or DNA were found in the liver, heart, or blood up to 21 days [82]. Another study showed that the liver and spleen are capable of clearing culturable bacteria within 6 days and can only be found in a few brain and blood samples before 6 h [83]. Despite the quick clearance of the bacteria, B. henselae DNA was found in all livers tested at 3 months post-infection. Histopathology of the livers at 3 days, 2 weeks, 4 weeks, and 3 months showed the progression of pathology. At 3 days post-infection, the tissues had few small aggregates of lymphocytes and monocytes. After 2 weeks post-infection, the granulomatous lesions became more obvious and consisted of lymphocytes, monocytes, and epithelioid cells. At week 4, the inflammatory lesions were greatly reduced and by month 3, the tissues were devoid of inflammatory lesions [83]. These studies differ from Arvand et al. [84] who found cultivatable bacteria up to 3 days post-infection from liver and up to 6 days post-infection from the spleens. Bartonella DNA could be amplified from the livers of infected mice up to 12 weeks post-infection. The infected livers did have mononuclear cell infiltrates at 2 weeks post-infection and these lesions expanded in size and number up to 12 weeks post-infection. The mice in this study seroconverted at 2 weeks post-infection which continued until 12 weeks post-infection. At the end of the study at 20 weeks, serology and pathology had returned to normal levels [84].

10.5. CD1

CD1 mice are an outbred strain which makes them genetically diverse, like Swiss-webster mice. Despite their popular use, only one study was identified in which CD1 mice were inoculated with two strains of B. tribocorum isolated from two different species of mouse. Both strains had success infecting these mice with 67% (12/18) and 61% (11/18) efficiency. Of the 23 mice, some of them had relapsing bacteremia and 8/30 (26.7%) still had bacteremia at the end of the 27-week study [63].

10.6. SCID

SCID (severe combined immunodeficiency disease) mice have a genetic mutation that affects B and T lymphocyte maturation. The beige mutation results in defective natural killer cells. Because of these mutations, SCID and SCID/beige mice are ideal models for studying the immune system, pathogenicity of microbes, and tumor growth. In the cases of Bartonella, exposure to the pathogen does not always lead to pathology in the host, so the lack of an intact immune system in these species allows the Bartonella to invade host cells more readily. Since atypical bartonellosis presents in immunocompromised patients more often, the use of immunocompromised mice may better mimic the disease process in humans. An experimental infection of SCID and SCID/beige mice with B. taylorii, B. birtlesii, B. doshiae, and B. grahamii showed that the bacteria can persist in the bloodstream for up to 2 months [64]. SCID mice had low levels of B. taylorii, but high levels of bacteremia were found in SCID/beige mice infected with B. taylorii, B. grahamii, and B. birtlesii. The difference in bacteremia between SCID and SCID/beige indicates that natural killer cells are partially responsible for clearance of the bacteria. SCID/beige mice infected with B. taylorii showed extensive levels of pathology in the liver, kidneys, and spleen. Other studies utilized SCID mice infected with B. henselae and noted pathologies in the liver of infected animals with no notable differences in the spleens or aorta of these animals [85].

10.7. AHNAK knockout

At least one study has used AHNAK knockout mice to study the effects of B. henselae infection. This mutant mouse strain no longer produces the Neuroblast Differentiation-Associated Protein, a large structural scaffold protein. This protein has multiple roles in the cell but primarily functions as a component of calcium signaling during CD4+ T cell activation. Although the researchers were able to identify B. henselae DNA in both the wild-type and mutant mice, closer histopathological analysis showed that only a few cells were infected in the liver of wild-type mice, but significantly more cells were infected in the knockout mice. After day 18, granulomatous mononuclear cells were identified in the liver of both the wild-type and knockout mice, with significantly more in the knockout group. Like SCID mice, the lack of T cell activation results in a more severe form of the disease [66].

11. Cats

Cats are an outbred animal model often used to research genetic and neurological conditions, as well as a naturally-occurring AIDS model from Feline Immunodeficiency Virus (FIV) [86]. Like mice, cats are a natural reservoir and host for some Bartonella species, predominantly B. henselae. Survey data has also determined that B. clarridgeiae and B. koehlerae can use felines as a reservoir [51,67,[87], [88], [89]]. Cats infected with Bartonella can have asymptomatic intraerythrocytic bacteremia that can last for months or years [90]. The percentage of bacteremic cats in households of patients with Cat Scratch Disease (CSD) is significantly higher than in domestic environments where CSD is not present [91].

Even though they lack clinical signs, infection rates in cats are high and relapsing bacteremia can be identified using common laboratory techniques such as PCR, blood culture, or immunoblot. Experimental infection of cats with B. henselae indicated that intradermal injection led to higher bacteria counts when compared to intravenous injection [67]. In a comparison of bacteremia duration between three Bartonella species including B. koehlerae, B. henselae, and B. clarridgeiae, B. henselae remained in the blood for 37–77 days, while B. koehlerae was evident by bacteremia for 70–78 days with no sign of relapsing bacteremia in either. However, when felines were infected with B. clarridgeiae, they were bacteremic for 263–363 days, including relapses. IgG and IgM antibodies were detected as early as 15 days post infection and persisted past 200 days (the full length of the study) [68].

Clinical disease in cats has been documented by Kordick, et al. [69] after intramuscular or intravenous injection of donor blood infected with B. henselae. After infection, some cats became febrile, but only 2 out of 14 animals exhibited clinical signs. This appears consistent with what is observed in the clinical setting [69]. A later study utilized blood from B. henselae or B. clarridgeiae-infected donor cats for intravenously inoculation. These cats exhibited acute anemia and fever, as well as intermittent bacteremia, but infection was not accompanied by clinical signs. One cat did experience focal motor seizures, nystagmus, and intermittent rigidity in the first 4 months. Out of the 18 cats in the experiment, 9 tested positive by either PCR or serology. At necropsy, many of the cats displayed histopathology of the lymph nodes, spleen, kidneys, heart, and eyes. Bartonella DNA was identified in the brain, lymph nodes, lung, left ventricle, liver, and kidney [70].

Horizontal and vertical transmission of B. henselae between cats has rarely been studied but has not been successful. In a preliminary study focusing on both naturally and experimentally infected cats, Abbot et al. described the lack of direct cat-cat transmission [67]. One study aimed to determine the ability of B. henselae to infect newborn cats by infecting the pregnant queens in mid- to late-pregnancy. All the queens became bacteremic within 2 weeks and were bacteremic at the time of delivery. Of the five pregnancies, three resulted in live births. Kittens born to these three queens were not bacteremic by qPCR or culture. Interestingly, kittens from each litter exhibited markedly different antibody responses including transient colostral antibodies, extended seropositivity, and no maternal IgG antibodies [68].

Additional Bartonella spp. were tested to determine their ability to infect cats [68]. B. vinsonii, B. rochalimae, B. bovis, and B. quintana were intradermally injected into cats aged 6–18 months (1/15 cat was inoculated by IV, 1/15 cat was age 3 yrs). Bacteremia was measured in only 3 out of 4 B. rochalimae-infected animals, and one of these animals displayed evidence of recurrent bacteremia. Although many of the cats never exhibited bacteremia, many of them seroconverted. Time elapsed between injection and seroconversion differed for each species. B. vinsonii-infected cats had measurable antibodies 3–4 weeks after injection, while B. rochalimae-infected cats became serologically positive by 14 days post injection. The cats inoculated with the feline strain of B. bovis mounted a rapid response with 5 of 6 cats seroconverting in the first 7–10 days while the bovine B. bovis strain seroconversion took 2–3 weeks. Interestingly, cats infected with either of two strains of B. quintana showed opposite effects. Cats infected with the human strain of B. quintana seroconverted in 2–3 weeks and remained positive through the end of the experiment. Conversely, cats injected with the feline strain of B. quintana did not show any signs of a serological response [72].

12. Dogs

Clinical signs of Bartonella-infected dogs include endocarditis, myocarditis, vasculitis, and granulomatous disease [92,93]. In one study, two golden retrievers that had naturally been infected with B. koehlerae were experimentally infected with B. vinsonii berkhoffi or B. henselae. B. koehlerae was identified by PCR and culture prior to experimental infection, but no Bartonella DNA was identified after. Both dogs became seropositive to their respective bacteria between weeks 2–4, and cultures from one animal had B. henselae in the bone marrow and B. koehlerae in the lungs. The dog infected with B. vinsonii berkhoffii did generate an antibody response but none of the tissues collected for analysis grew Bartonella in culture. There were, however, significant pathological findings. Gross lesions were seen in the to the lungs and pleura, adrenal cortex, salivary gland and trigeminal ganglion [73]. This study reveals that infection of dogs with these strains of Bartonella can lead to multiorgan pathology.

B. henselae and B. vinsonii infections were repeated in another study using beagles, which also included B. rochalimae. In this study, B. henselae caused severe and necrotic lesions at the site of inoculation which persisted for 10–20 days. All the B. henselae-infected dogs did seroconvert within 2 weeks of inoculation, although none of the dogs became bacteremic. For the animals infected with the other Bartonella species, there were no clinical or physical abnormalities except mild inflammation at the injection site. Dogs inoculated with B. vinsonii berkhoffi had a persistent bacteremia that lasted for 60- and 70-days post-inoculation. Antibody titers were low but measurable in these dogs for up to 90 days. In the case of B. rochalimae, the dogs became bacteremic at days 19 and 24 post-inoculation although the bacteremia only lasted 2–2.5 weeks [74].

13. Non-human primates (NHP)

Several screening studies in animal-care facilities have found that non-human primates (NHPs) harbor Bartonella quintana without any clinical signs, indicating they are a natural reservoir for the bacteria. Few experimental studies have been conducted to examine Bartonella infection in non-human primates. The first documented study, performed in 1953 by Mooser and Weyer, used the contents of B. quintana-infected lice (Pediculus spp) intestines to intravenously infect 7 rhesus macaques [75]. Infection was confirmed when uninfected lice were allowed to feed on the animals and subsequent microscopic examination of louse hemolymph produced Bartonella-positive slides. The macaques were bacteremic for approximately 100 days, after which uninfected lice did not become positive. The animals did not show clinical signs of infection throughout the study, although the authors did note a transient rise in body temperature in 3 out of 7 animals. A later study found that B. quintana isolated from rhesus macaques during surveillance was able to be cultured and intravenously injected into 4 non-infected animals [76]. All animals displayed intermittent, transient bacteremia throughout the next 100 days after infection, and none of the animals displayed clinical signs of disease. In both studies, the exact inoculation dose was unknown. A macaque model of infection for B. quintana has been described for bacteremia, with published data from one representative animal [77]. This animal was infected with 6.6 × 10⁷ CFU of B. quintana via the intradermal route and subsequently displayed intermittent levels of bacteremia over 100 days, but no clinical signs of disease.

In addition to the studies using Bartonella quintana, one publication describes the susceptibility of owl monkeys to Bartonella bacilliformis infection [78]. Animals were infected with 1.1 × 10⁶ CFU intravenously (n = 3) or 9.5 × 10⁷ CFU intradermally (n = 3). In the intradermal-inoculated group, one animal presented with fever and skin lesions. Two out of 3 animals in each group presented with bacteremia as measured through Giemsa staining. However, PCR and blood culture tests were negative in both groups. This study demonstrated that a species of new world non-human primate could become infected and present mild disease with a different Bartonella strain.

A compilation of the studies conducted in cats, dogs and primates is presented in Table 3. An important limitation to the studies in rodents, dogs and cats is the detection of Bartonella DNA exclusively as evidence for infection. Ideally, cultures would be used to validate a productive infection, because DNA could be detected from environmental contamination. A recent study indicated that blood supplementation could enrich the culture and increase the chances of live bacteria acquisition [94].

14. Conclusion

The data presented in this review stands as an indicator of the convoluted and complex nature of animal models for bartonellosis research. An ideal animal model for bartonellosis would accurately represent human clinical presentation and have measurable changes in pathology and bacteremia. Without a normalized set of clinical signs or symptoms, identifying a suitable pathological model is difficult. Where dogs may exhibit pathology, cats typically only exhibit bacteremia. Only certain combinations of Bartonella species and mouse strains lead to persistence and rarely, pathology. This is likely due to the utilization of multiple reservoir hosts by the ubiquitous Bartonella genus of bacteria.

None of the models presented here meet all criteria for an accurate, measurable, and pathological model. Selection criteria should be heavily dependent on experimental needs. For example, although Swiss Webster strain mice display pathological signs in many tissues, the low measurable infection rate limits it's use in antibiotic treatment studies. However, this same strain of mouse may be useful when studying disease progress or severity. Using non-traditional models such as cats and dogs may be of use in some Bartonella studies. Cats display extended bacteremia states as seen in some human cases of bartonellosis, indicating they could be a good model for chronic infections with serological findings (Table 3).

In lieu of a pathological model, infectious models serve as the primary data source for therapeutic drug testing. Without primary research examining the changing pathologies from multiple strains and mutations of Bartonella spp., treatments cannot be broadly targeted. As co-infections between Bartonella and other vector-borne pathogens, such as B. burgdorferi, emerge as a factor in disease progression, the need for a versatile pathological model becomes more apparent. Despite the difficult circumstances presented here, it is the aim of these authors and many others in the field of Bartonella to identify the best model for the research purpose.

Funding statement

This research was supported by the Steven & Alexandra Cohen Foundation.

Declaration of Competing Interest

The authors have no conflicts of interest to declare.

Data availability

No data was used for the research described in the article.