INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Borreliosis, a multisystemic infectious disease of humans and some animal species is caused by spirochetes of the Borrelia burgdorferi sensu lato group. The three pathogenic genospecies known to occur in Europe are B. burgdorferi sensu stricto, Borrelia garinii, and Borrelia afzelii (4, 46, 49). A fourth genospecies, Borrelia valaisiana (former group VS 116), is widely distributed in Europe but its pathogenicity is not yet clear (35). In humans, Lyme borreliosis (LB) can be recognized by an expanding, sometimes migrating erythematous lesion (EM). Simultaneously with the EM, immunoglobulin M (IgM) and often IgG antibodies against specific antigens of B. burgdorferi sensu lato develop (1, 35, 43). In early LB, a response against the 41-kDa flagellin and the 21- to 23-kDa outer surface protein, OspC, is mounted, and later in the course of disease responses against an expanding number of proteins can be measured by immunoblotting (2, 8, 11, 18, 19, 51). Antibodies against the 31- to 34-kDa OspA, the major protein expressed when the spirochete inhabits the tick midgut, only develop in late LB with chronic often antibiotic-resistant arthritis (2, 24, 25). OspA is downregulated and OspC is expressed when spirochetes migrate from the midgut to the salivary gland of the tick and are subsequently transmitted to the host (10, 15, 37). In humans and dogs vaccinated with a recombinant OspA, bactericidal antibodies which are protective against infection develop (30, 34, 45). Paradoxically, in symptomatic humans and hamsters this naturally occurring bactericidal activity apparently does not resolve the disease (5, 13). In the hamster model the three genospecies are able to cause infection separately and at the same time elicit non-cross-reactive protective bactericidal activity (27).

Borreliosis can also occur in dogs, for which clinical symptoms were defined as malaise (caused by fever and showing as inappetence) and lameness (23). Apart from antibodies against the 41-kDa flagellin protein, which can be cross-reactive, antibodies against 39-, 30-, 28-, 26-, 25-, and 19-kDa proteins are frequently seen in Borrelia-exposed dogs (16, 23). In a wooded area in The Netherlands where Lyme borreliosis is endemic all household dogs developed antibodies against Borrelia, whereas a control group of dogs living in an area where the disease is not endemic did not show such antibodies (21). Therefore, it was concluded that all these Dutch dogs had Borrelia infections. By PCR we found that in dogs clinically suspected of having borreliosis the frequency of infection by Borrelia, often by more than one species at the same time, was much higher than in dogs that remained asymptomatic (22). Diseased dogs with clinical symptoms such as malaise (in most cases accompanied by fever) and lameness had a very high titer during the symptomatic period, which persisted in chronically diseased dogs or diminished when dogs recovered (21).

Serum can exert antibody-independent borreliacidal activity through complement. In human sera, the intensity of this bactericidal activity differs between strains from the three pathogenic European species (47). Canine bactericidal activity through complement has not yet been tested and may exert differential protection against the different genospecies.

The goal of this study is to characterize the specific immune responses of Dutch dogs against infection by one or several species of the B. burgdorferi sensu lato group. We determined antibody-independent and antibody-dependent bactericidal activities in sera of symptomatic and asymptomatic dogs and investigated the expansion of the antibody response in the course of symptomatic infection. Sera were tested for bactericidal activity and specific antibodies against B. burgdorferi sensu stricto, B. garinii, and B. afzelii by in vitro bactericidal assays and by Western blotting.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Borrelia isolates. Three Borrelia strains representing the three major pathogenic genospecies were examined in this study. The specific isolates studied were B31 (B. burgdorferi sensu stricto), A87S (B. garinii), and pKo (B. afzelii). Strains B31 and pKo were both high-passage reference strains, but A87S was passaged less than 15 times. The isolates were stored at 70°C in 50% glycerol peptone and cultured in modified Barbour-Stoenner-Kelly (BSK) medium at 33°C. These isolates were used for the immobilization assays as well as for the preparation of antigen for immunoblots.

Dogs studied and serum samples. Dogs living in a wooded environment in the south of The Netherlands are all heavily infested by naturally occurring ticks during consecutive tick seasons, especially in May and June (20). These dogs were monitored in a local veterinary clinic for at least 5 years, and some of them developed symptoms compatible with canine LB as described by Jacobson et al. (23). Dogs were not vaccinated against borreliosis. The diagnosis of symptomatic borreliosis was made when dogs had a period of symptoms in which malaise (listlessness or inappetance) was followed by a period of lameness. Dogs without this combination of symptoms were referred to as asymptomatic. Results of the concurrent serological monitoring by whole-cell enzyme-linked immunosorbent assay (ELISA) were not used as an entry criterion. In the present study, 15 sera from dogs symptomatic for borreliosis were further analyzed by Western blotting and immobilization assays and compared with sera from 15 asymptomatic age-matched controls. For the symptomatic dogs we recognized three patterns in the course of the disease: dogs that recovered from disease, dogs with intermittent recurring disease, and dogs with progressive disease (Table 1). The age at which first symptoms were observed for symptomatic dogs is indicated, as is the age of occurrence of a high peak titer in asymptomatic control dogs. Ten dogs showed symptoms before their 8th year of life. In 13 of 15 symptomatic dogs malaise was accompanied by fever (>39.0°C). Only one of the asymptomatic dogs (dog 39) developed fever, which was explained by another infectious disease. Several organ systems were involved in most of the symptomatic dogs. In 13 of the symptomatic dogs one or several treatments with antibiotics were given (amoxicillin at 10 mg/kg of body weight twice daily, orally for 14 days) in at least one of the disease episodes. Treatment was usually given when very high fever was noticed and may have influenced the course of the disease, especially in three younger dogs that were treated during first symptoms and that completely recovered. However, in the other dogs, recovery from disease episodes occurred with and without treatment, and under both conditions episodes recurred. To exclude other causes of fever of undetermined origin, malaise, and lameness, the clinical workup when appropriate included radiology, laboratory work in search of immune-mediated diseases (determination of antinuclear antibody and rheumatoid factor), and histology on biopsies. Moreover, from all symptomatic dogs complete hematological (complete and differential red and white blood cell and platelet counts) and serum biochemistry profiles (blood urea nitrogen, creatinine, glucose, electrolytes, liver enzymes, bilirubin, protein electrophoresis) were obtained at several time points in the course of disease (during the 5 years of monitoring). For most symptomatic dogs the test results suggested acute bacterial infection (neutrophilia with left shift) or prolonged antigenic stimulation (mild lymphocytosis and eosinophilia and polyclonal gammopathy) as the cause of disease. Asymptomatic dogs were not treated with antibiotics unless another infectious disease was diagnosed (Table 1).

Antigen preparation. Spirochetes were grown in 50 ml of BSK medium at 33°C until the stationary phase was reached and the concentration of the spirochetes was approximately 5 × 10⁷/ml. Cells were harvested by centrifugation at 5,000 × g for 20 min and washed three times with 50 mM Tris-HCl (pH 7.4). Protein concentrations were determined as described by Lowry et al., and the preparations were stored at 20°C (28). The amount of protein used for the gel electrophoresis was the same for each gel.

Whole-cell ELISA. Whole-cell antigen was prepared from B. burgdorferi sensu stricto strain B31 by sonication as described previously (21). After the incubation with serum, horseradish peroxidase-conjugated anti-dog IgG (1/3,000 dilution; Organon Teknika, Turnhout, Belgium) was used in combination with the chromogenic agent (ABTS [2,2'-azinobis(3-ethylbenzthiazolinesulfonic acid]; Sigma Chemical Co., St. Louis, Mo.). Optical density at 405 nm was measured using a Titertek Multiscan ELISA reader. Antibodies were determined by end point titration; each serum was tested in a dilution range from 1/20 to 1/2,560. Higher dilution titers were extrapolated from the optical density values. Cutoff values were calculated on the basis of the results of the pre-tick-bite sera of 12 young dogs. Samples were considered positive if they had an end point titer of 1/320 (21). Sera from 52 dogs in an area where the disease is not endemic (New Zealand) tested negative, as did sera from 16 SPF dogs vaccinated and challenged with Leptospira icterohemorrhagiae(21). Sera from all dogs in this study from the area of endemicity in the south of The Netherlands showed seroconversion to titers of 1/320 and higher in the first or second tick season. Dogs remaining asymptomatic were seen to reach persistent titers of 1/320 and 1/640; only occasionally was a higher titer observed (21). However, the symptomatic dogs showed a steep rise in titers to 1/2,560 or higher before or concurrent with the development of symptoms. In most of the symptomatic dogs these high titers persisted for 1 year or longer (Table 1).

SDS-PAGE and Western blots. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed on whole-cell lysates of B. burgdorferi sensu lato by a modification of the methods described by van Dam et al. (46). Briefly, the antigen was diluted 1:1 with 2% SDS sample buffer and was boiled for 5 min. This suspension was electrophoresed on 13% polyacrylamide gels (15 by 10 cm). The gels were run at 50 mA for 3 to 4 h. Adjacent to the cell lysate, prestained low-molecular-mass markers (Bio-Rad, Münich, Germany) were applied in an extra lane. The separated proteins were blotted overnight onto nitrocellulose at 50 mA in a carbonate buffer (10 mM NaHCO₃ and 3 mM Na₂CO₃ containing 20% methanol). Blots were blocked with nonfat dried milk in an incubation buffer (10 mM Tris-HCl, 500 mM NaCl, 0.5% Tween 20; pH 7.5) for 1 h at 22°C and cut into 4-mm strips. Antigen strips were incubated with 1:100 dilutions of test serum for 120 min, washed three times for 5 min each time (with 10 mM Tris-HCl, 500 mM NaCl, 0.5% Tween 20; pH 7.5), and incubated with horseradish peroxidase-conjugated goat anti-dog IgG antibodies (Nordic, Breda, The Netherlands) diluted 1:1,000 in phosphate-buffered saline (PBS). After three washes, two times as described above and one time with PBS, the antibody reactivity was visualized by incubation with 4-chloro-1-naphthol-H₂O₂ for 15 min. When immunoblots were performed with monoclonal antibodies (MAb), 1 ml of a 1:50 dilution of the culture supernatant was incubated instead of the test serum and 1 ml of a 1:7,500 dilution of alkaline phosphatase-conjugated goat anti-mouse IgG antibodies was used as a second antibody (Promega, Leiden, The Netherlands). The reactive protein bands were visualized with nitroblue tetrazolium and 5-bromo-4-chloro-3-indolylphosphate (Promega). Protein bands found by immunoblotting were scored at their molecular masses and intensities for further evaluation. In this study very vague bands were not included.

MAb. Six MAb were used to locate the major proteins on the Western blots. H9724 recognizes a 41-kDa flagellar protein (p41) in all three species tested (19). LA 26 is directed to the 31-kDa outer surface protein A (OspA) of B. burgdorferi sensu stricto and B. afzelii, whereas LA 31 is directed to OspA of B. burgdorferi sensu stricto and B. garinii (46). OspB was located with MAb 84C in all three strains tested (40). Finally L22 1F8 and L22 C11 were used to locate a 21- to 23-kDa protein (OspC); L22 1F8 reacts with OspC of all three Borrelia species (48, 50), whereas L22 C11 is directed to OspC of B. garinii and B. afzelii (48). For the outer surface proteins the apparent molecular masses differed for the three strains. OspB differed in apparent molecular mass between 33 kDa for strain A87S and 34 kDa for strains B31 and pKo. OspA had an apparent molecular mass of 31 kDa in strains B31 and A87S. Strain pKo expressed a 31-kDa band in the immunoblot, which showed no reactivity to a B. afzelii-specific anti-OspA MAb. OspC MAb to strains A87S and pKo reacted with protein bands with an apparent molecular mass of 21 kDa. Strain B31 expressed a 21-kDa protein that did not show reactivity with the specific MAb.

Bactericidal assays. Borrelia isolates were thawed and grown to a density of approximately 10⁷ spirochetes per ml of BSK, as judged by dark-field microscopy. An aliquot of this suspension was added to an aliquot of heat-inactivated test serum and an aliquot of serum of SPF dogs, referred to as normal canine serum (NCS), as a complement source to give a final volume of 100 µl. To assess the bactericidal activity of NCS (i.e., antibody-independent killing of spirochetes), an aliquot of heat-inactivated NCS and the active NCS was added to the spirochetes, as described previously for human normal serum (47). The concentrations of NCS used in the final assays differed for the three genospecies because of their different sensitivities to the bactericidal activity of the complement source. To assess the bactericidal activity of the serum of dogs attending the clinic (i.e., antibody-dependent killing of spirochetes), 15% NCS was added for strain A87S and 25% SPF serum was added for B31 and pKo. To avoid the presence of particles that could diminish the visibility of the spirochetes, all sera were centrifuged for 5 min at 14,000 × g and 4°C before use. Experiments were performed in duplicate in a 96-well microtiter plate. The plate was sealed and incubated at 33°C. After 0, 1, 3, and 5 h of incubation an aliquot of 5 µl was drawn from each well to assess the mobility and the extent of bleb formation of the spirochetes by dark-field microscopy. Immobilized, blebbed spirochetes were considered nonviable (47). In negative-control experiments heat-inactivated SPF serum was added to the suspension containing spirochetes with or without test serum. Borreliacidal activity of the test serum was corrected according to the formula corrected immobilization (CIM) = percentage of immotile spirochetes in test serum and NCS  percentage of immotile spirochetes in NCS only/(100%  percentage of immotile spirochetes in NCS only). A CIM of 20% or more was considered significant immobilizing activity of the test serum.

Detection of bacterial DNA by PCR. From eight symptomatic dogs and four asymptomatic dogs tissue biopsies could be tested for the presence of Borrelia DNA. Positive tissue biopsies included skin, synovial tissue, heart, liver, bladder wall, and bone marrow tissue, and cerebrospinal fluid. All specimens were included in a study described elsewhere (22). After homogenization of tissues and DNA extraction, part of the 5S-to-23S rRNA spacer region was amplified by PCR. Amplification products were hybridized with specific probes for B. burgdorferi sensu stricto, B. garinii, B. afzelii, and B. valaisiana. Details of all procedures have been described earlier (22).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Antibodies found on immunoblots. The 15 sera of symptomatic dogs and 15 sera of asymptomatic dogs were compared by Western blotting with three strains providing the antigens. A total of 40 antigens of B. burgdorferi sensu stricto, 41 antigens of B. garinii, and 39 antigens of B. afzelii reacted with antibodies from at least one of the sera. More bands were detected on immunoblots with sera of symptomatic dogs that were under 8 years of age when the first symptoms occurred than with sera of older symptomatic dogs or with sera of asymptomatic dogs (Table 2). The sera of the younger dogs were mostly sampled during early, and for some dogs temporary, stages of the disease; the sera of the older dogs were sampled during later stages of disease, and these dogs had progressive disease (Table 1).

View larger version (54K): [in this window] [in a new window]   FIG. 1.   (A) Immunoblots with dog sera using different Borrelia strains as sources of antigens. Lane 1, negative-control dog; lane 2, asymptomatic dog 35; lane 3, symptomatic dog 33; lane 4, symptomatic dog 48; lane 5, positive-control dog. The molecular masses of the proteins were assessed by running a prestained low-molecular-mass marker adjacent to the cell lysate. All three strains were tested for the presence of OspA, OspB, OspC, and flagellin with MAb. (B) Immunoblots with dog sera using strains A87S and B31 as sources of antigens. Lane 1, negative-control dog; lane 2, asymptomatic dog 57; lanes 3 and 4, preclinical sera from dog 33; lane 5, acute-phase serum from dog 33; lane 6, chronic-phase serum from dog 33; lanes 7 and 10, preclinical serum from dog 28; lanes 8 and 11, acute-phase serum from dog 28; lanes 9 and 12, convalescent serum from dog 28. Identification of the proteins was as for panel A.

Bactericidal activity of complement. Immobilizing activity against three spirochetal strains was determined with NCS. NCS in a concentration of 25% in the test medium had little immobilizing activity against strain B31 of B. burgdorferi sensu stricto (mean ± SD, 4% ± 2%) and against B. afzelii strain pKo (6% ± 4%). Strain A87S was more sensitive to NCS, and incubation of strain A87S with 25% NCS resulted in high levels of immobilization (ranging from 31 to 95% in seven experiments with an average of 61% ± 24%). Thus B. burgdorferi strain B31 and B. afzelii strain pKo were resistant to the bactericidal activity of dog complement. In contrast, B. garinii strain A87S was susceptible to the bactericidal activity of NCS. With human sera and sera from other mammals also, differences in the complement susceptibilities of the genospecies have been described (26, 47). Complement-mediated killing in natural hosts could have ecological implications for the Borrelia species as it might determine the reservoir competence (26). In this respect, it is interesting to note that the dog is a competent reservoir for B. burgdorferi sensu stricto (31).

Antibody-mediated bactericidal activity. Sera from all 15 symptomatic and 11 of the 15 asymptomatic dogs were tested in an immobilization assay with representative strains of the three different Borrelia species (Fig. 2 and Table 3). The immobilization by antibodies was corrected for the immobilization by NCS. A CIM of >20% was considered significant. Two sera from dogs 33 and 48, which are both Bernese mountain dogs, immobilized nearly all spirochetes of strain pKo (CIM of 80 to 100%) in the assay and had reactivity against many bands of this same strain on immunoblots (Table 3). The serum of four dogs (dogs 14, 40, 58, and 72) immobilized strain A87S. Two dogs (dogs 72 and 79) had serum that immobilized B. burgdorferi sensu stricto strain B31. The serum of dog 72 had a CIM against A87S (60 to 80%), but not against B31, during high peak titers and symptoms. When serum was tested 2 years later during one of the intermittent episodes with symptoms, the CIM against A87S was less (20 to 40%) while a CIM against strain B31 had developed (60 to 80%). In total, sera from 7 of the 15 symptomatic dogs immobilized one or two strains and none of the 11 asymptomatic dogs had serum that immobilized one of the three strains (P = 0.0080 by chi-square test).

View larger version (25K): [in this window] [in a new window]   FIG. 2.   Antibody-dependent immobilizing activities of sera from symptomatic and asymptomatic dogs and antibody-independent immobilizing activities against three Borrelia strains after 5 h of incubation at 33°C. Immobilizing activities of sera from symptomatic and asymptomatic dogs were corrected for the immobilizing activity of complement, according to the formula CIM = percentage of immotile spirochetes in test serum and NCS  percentage of immotile spirochetes in NCS only/(100%  percentage of immotile spirochetes in NCS only). Open bars, CIM < 20% (i.e., negative); hatched bars, CIM = 20 to 40% or 40 to 60% (i.e., positive) or 60 to 80% or 80 to 100% (i.e., strongly positive). Multiple sera from dogs in different stages of disease were tested. All dogs, except one, with a positive test result reacted against one of the strains in single or in consecutive sera. Dog 72 had immobilizing activity against strain A87S in one serum sample (72a) and against strain B31 and strain A87S in a consecutive serum sample (72b).

View larger version (63K): [in this window] [in a new window]   FIG. 3.   Whole-cell ELISA, as reciprocal titer, and antibody-dependent immobilizing activity, as CIM (in percent; presented as in Fig. 2), of consecutive sera from three dogs sampled before, during, and after disease. Symptomatic periods are represented by M (malaise) and L (lameness), and the arrows indicate when the serum used in immunoblots was obtained (see Fig. 1). (A) Dog 28. No immobilizing activity against any of the three strains was seen. This dog was one of the three dogs in this study that completely recovered from disease. (B) Dog 14. Bactericidal activity was only directed against B. garinii strain A87S. (C) Dog 33. Bactericidal activity was only directed against B. afzelii strain pKo.

Detection of Borrelia DNA in tissue biopsies. From 12 dogs tissue specimens were available. Seven out of eight symptomatic dogs and three out of four asymptomatic dogs were tested positive for Borrelia by PCR (Table 3). From all seven symptomatic dogs B. burgdorferi sensu stricto was amplified. Three of these dogs also contained B. garinii DNA, and two of these dogs also contained B. afzelii DNA. In one dog DNA from four Borrelia species was found. B. burgdorferi sensu stricto DNA was not amplified from any of the asymptomatic dogs. B. garinii, B. afzelii, and B. valaisiana were each detected once among asymptomatic dogs.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Sera from pet dogs which were symptomatic or asymptomatic for borreliosis and which were monitored clinically and serologically during a 5-year period were sequentially collected. All dogs were exposed to Ixodes ricinus ticks and developed an antibody response in whole-cell ELISA employing B. burgdorferi sensu stricto antigens in contrast to a control group of dogs from an area where the disease is not endemic (21). In the present study, sera from 15 dogs symptomatic for borreliosis and sera from a comparable control group of 15 asymptomatic dogs were further analyzed by Western blot and immobilization assays.

All sera contained multiple antibodies against Borrelia as detected by Western blotting, whereas sera from only one out of four of the negative-control dogs showed a sole band against the 41-kDa flagellin protein. The reactivity in the sera from asymptomatic dogs was usually limited to the 41-, 39-, and, to a lesser extent, the 66- to 60-kDa regions. Therefore we regard antibodies against these proteins in dogs as markers of exposure to B. burgdorferi sensu lato. Sera from symptomatic dogs had a broader spectrum of reactivity, especially sera from dogs with occurrence of the first symptoms before the 8th year of life (Table 3). In preclinical sera from symptomatic dogs antibodies against p41 and p39 were already present long before the onset of disease. This is in accordance with our hypothesis that these antibodies are a consequence of exposure to spirochetes but that they are not necessarily related to clinical disease. Antibodies against the 41-kDa flagellin protein may be due to cross-reactivity of antibodies directed at the flagella of other bacteria (29, 39). However, the 39-kDa protein is Borrelia genus specific, and no cross-reactivity of antibodies against this protein has been demonstrated (29, 38, 41). The 66-kDa protein is in the range of the cross-reacting heat shock proteins, and three strains reacted with the serum of the positive-control dog, exclusively infected with B. burgdorferi sensu stricto. Therefore, this protein is probably not a recently described outer membrane protein which is species specific within the B. burgdorferi sensu lato group (3).

In sera from symptomatic dogs that were older than 8 years of age when the first symptoms occurred, the immune reactivity on Western blots was diminished. These dogs fulfilled the clinical entry criteria for borreliosis, and in three out of four tested by PCR, B. burgdorferi sensu lato DNA was detected in organ tissues (22). Moreover, they exhibited a strong rise in whole-cell ELISA titer before or during the onset of clinical manifestations, and therefore another disease as the cause of symptoms is not likely, although it is not excluded. It is known that the immune response can change during old age (36). Alternatively, frequent reinfections together with persistent infection may cause antigenic changes in the spirochete, resulting in a shift to in vivo-expressed antigens which we could not measure. Antigenic shifts may be part of the immune evasion strategies of the spirochete as determined in persistently infected laboratory mice (9).

In the acute-phase sera from young dogs (under 8 years on occurrence of the first symptoms), reactivity with seven proteins of B. afzelii, five proteins of B. garinii, and two proteins of B. burgdorferi sensu stricto was associated with disease. Investigators studying European human sera found more reactions with the B. afzelii and B. garinii reference strains than with the B. burgdorferi sensu stricto strains (18, 19, 33). We found strong cross-reactivity of specific antigens on B. afzelii immunoblots with serum from the control dog infected with B. burgdorferi sensu stricto. Although cross-reactivity with homologous antigens of B. afzelii must be accounted for in the evaluation of the immunoblots (33), more probably reinfections and mixed infections with various Borrelia genospecies account for the discrepancy between the seroreactivities of the dogs and the PCR results showing most frequently B. burgdorferi sensu stricto in these dogs (Table 3) (22).

Bactericidal antibodies were found in six acute-phase sera from 10 symptomatic dogs that were younger than 8 years on occurrence of the first symptoms but in none of those from the 11 asymptomatic dogs tested. In one dog the target of these bactericidal antibodies changed over time from B. garinii to B. burgdorferi sensu stricto. In the other five dogs consecutive sera were bactericidal only against one bacterial genospecies, B. garinii or B. afzelii. It is remarkable that B. burgdorferi sensu stricto DNA was amplified from tissue biopsies from symptomatic dogs and not from asymptomatic dogs. This may indicate that B. burgdorferi sensu stricto strains are more virulent in dogs and that a protective immune response is more difficult to elicit. So far, only B. burgdorferi sensu stricto has been shown to be virulent in dogs in an animal model (44). Alternatively, the species that elicits bactericidal activity may be the cause of disease, as has been shown in the hamster model (27).

Three young dogs (dogs 28, 47, and 63; Table 1 and Table 3) without bactericidal activity had OspA reactivity on immunoblots using strains B31 and A87S with acute-phase and convalescent sera (Fig. 1B; dog 28). Interestingly, these three dogs recovered from disease and had low convalescent whole-cell ELISA titers (Fig. 3A; dog 28). In the mouse model the presence of anti-OspA antibodies in late disease was associated with accelerated resolution of disease (12, 42). For these three dogs the disease was suspected upon occurrence of the first symptoms and treatment was immediately initiated (Table 1), which could have facilitated recovery and may have influenced the antibody spectrum. Four of the six young dogs with bactericidal activity had bactericidal antibodies against B. garinii in their serum. Three of these had a preferential reactivity against B. garinii-specific OspC, which is in line with studies that report that this protein is capable of inducing the production of highly specific borreliacidal antibodies shortly after natural infection (6). Two of these dogs were investigated for the presence of spirochetal DNA in their tissues, and both were found to be infected with B. burgdorferi sensu stricto and B. garinii. This may be in line with the hypothesis that a heterogeneous population of spirochetes is delivered to the host, which would result in changes of the immune response over time enabling the infection to persist (14). Two other young dogs, which were both Bernese mountain dogs, developed a marked and persistent bactericidal antibody response against strain pKo (B. afzelii) as well as a preferential response against this strain, including a response against OspC, as detected with immunoblots. Subsequently to the development of borreliacidal antibodies both Bernese mountain dogs developed a lifetime progressive disease, apparently not prevented by these antibodies.

Downregulation of antigens recognized by bactericidal antibodies and coinfection with a different strain not recognized by bactericidal antibodies could both be involved in the persistence of infection. Alternatively, persistent and recurring symptoms could be caused by autoreactive antibodies. Probably, there are different mechanisms for disease, which may have different outcomes depending on the host immune system idiosyncrasies.

In conclusion, although both naturally exposed and infected dogs have moderately titered to high-titered antibodies as measured by whole-cell ELISA, symptomatic dogs produce a much wider spectrum of antibodies, including immobilizing antibodies. Western blots especially may be helpful in confirming the diagnosis of canine borreliosis.