INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Borrelia burgdorferi, the etiologic agent of Lyme disease, can infect a variety of vertebrates, in which it causes disease or asymptomatic infections. The dog is the domestic animal at greatest risk, and it has been recommended as a sentinel animal for human Lyme disease (7, 19). Due to the lack of differential signs such as erythema migrans in infected dogs (1), laboratory methods are very important for canine Lyme disease diagnosis.

We had previously analyzed the antigenicity of the six invariable regions (IRs; i.e., IR₁ to IR₆) located within the variable domain of VlsE (13, 15), the B. burgdorferi lipoprotein that undergoes antigenic variation (23). We determined that while in humans and nonhuman primates only IR₆ is immunodominant, in mice not only IR₆ but also IR₂ and IR₄ are antigenic, with IR₆ and IR₂ being most frequently recognized (15). We developed a peptide-based enzyme-linked immunosorbent assay (ELISA) using as antigen a 26-mer synthetic peptide (C₆) based on the IR₆ sequence, and we determined that this assay is highly sensitive and specific for human Lyme disease serodiagnosis in the United States (14). We also determined that the C₆ ELISA could be used in Europe, insofar as it detected antibody in human patients that had culture-confirmed infections with either Borrelia garinii or Borrelia afzelii, the two genospecies most prevalent in Europe (18).

To identify a suitable probe for serology of canine B. burgdorferi infection, we first analyzed the antigenicity of each of the IRs of VlsE in dogs that had been experimentally infected with B. burgdorferi by tick inoculation. Like mice, dogs vigorously responded to both IR₂ and IR₆, with IR₆ stimulating a stronger and earlier antibody response than IR₂. Our further analysis ruled out the necessity of including IR₂ as a diagnostic antigen and demonstrated that IR₆ alone is enough as a probe for the serodiagnosis of canine Lyme disease.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Tick collection and dog inoculation. Thirty-three 6-week-old specific-pathogen-free beagles of both sexes were infected by tick inoculation as described previously (22). Ticks were field collected in Westchester County, New York. All dogs were infected with B. burgdorferi as evidenced by skin punch biopsy culture and PCR, which were conducted at 4 weeks after tick exposure (22). Serial blood samples were collected from all of the dogs at 2- to 4-week intervals for 17 weeks beginning at day 0 of the experiment. In some dogs, blood sampling continued until week 69 postinfection.

Negative control serum specimens and cutoff line. Seventy control serum specimens were collected from healthy dogs owned by students of a veterinary school in Louisiana. This panel of serum specimens was used to calibrate a cutoff line for serodiagnosis. The cutoff line was defined as the mean optical density (OD) value plus 5 standard deviations (SDs) of these 70 specimens. Lyme disease is not endemic in Louisiana, and the dogs did not have a history of travel to endemic areas.

Serum specimens from vaccinated dogs or dogs with infections other than B. burgdorferi. Fourteen blood samples were collected from dogs with leptospirosis (n = 5), Rocky Mountain spotted fever (RMSF; n = 2), or infection with Dirofilaria (n = 5), Babesia (n = 1), or Ehrlichia (n = 1) spp. An additional 15 serum specimens were collected from dogs vaccinated either with the outer surface protein A (OspA; n = 5), a whole fixed spirochete vaccine (n = 5), or the common vaccines received by puppies (distemper, adenovirus 2, parainfluenza, parvovirus, leptospirosis, and coronavirus [DA₂PPLCV]; n = 5). These serum samples contained antibodies to OspA or other spirochetal antigenic proteins, as appropriate, as determined by immunoblotting using B. burgdorferi whole-cell extracts as antigen.

Clinical serum specimens. A panel of 55 canine serum specimens was used to compare sensitivities as measured by kinetic ELISA (KELA), immunoblot analysis, and peptide-based ELISA. These samples were originally submitted for the serodiagnosis of Lyme disease and collected from dogs that were suspected of having Lyme disease. KELA and immunoblot assays were performed as previously described (22).

Synthetic peptide sequences, preparation, and biotinylation. Peptides were prepared using the fluorenylmethoxycarbonyl synthesis protocol (3) based on the sequences listed in Fig. 1. Synthetic peptides were covalently linked to biotin by the N-succinimidyl maleimide carboxylate method. The maleimide reagents were from Molecular Probes (Eugene, Oreg.), and the protocol suggested by the manufacturer was followed.

View larger version (19K): [in this window] [in a new window]   FIG. 1.   VlsE structure and sequences of synthetic peptides (15, 16). VlsE consists of two invariable domains at the amino and carboxyl termini and one variable domain at the center. The variable domain contains six variable regions, VRI through VRVI, and six invariable regions, IR1 through IR6. The six IR sequences were obtained from one cloned variable domain of VlsE expressed by the strain IP90 of B. garinii (13). The insert (framed) shows the IR6 sequences from IP90 of B. garinii (13) and strains 297 (11) and B31 (23) of B. burgdorferi sensu stricto. Bold letters indicate amino acids unique to each strain. The consensus sequences of the three overlapping peptides used in this study are depicted as C6N, C6M, and C6C.

Peptide-based ELISA. Peptide-based ELISA was conducted as described previously with minor modifications (13). Ninety-six-well ELISA plates were coated with 100 µl per well of 4-µg/ml streptavidin (Pierce Chemical Co., Rockford, Ill.) in coating buffer (0.1 M carbonate buffer [pH 9.2]) and incubated at 4°C overnight. The remaining steps were conducted in a rotatory shaker at room temperature. After two 3-min washes with 200 µl per well of phosphate-buffered saline (PBS)-Tween 20 (PBS/T, with PBS containing 0.1% Tween 20 [pH 7.4]) at 200 rpm, 200 µl of 5-µg/ml biotinylated peptide dissolved in blocking solution (PBS/T supplemented with 5% nonfat dry milk) was applied to each well. The plate was shaken at 180 rpm for 2 h. After three washes with PBS/T, 50 µl of dog serum diluted 1:200 with blocking solution was added to each well. The plate was incubated for 1 h at 180 rpm and then washed three times with PBS/T. Each well then received 100 µl of 0.35-µg/ml rabbit anti-dog immunoglobulin G conjugated to horseradish peroxidase (heavy and light chain specific; Sigma Chemical Co., St. Louis, Mo.), and dissolved in blocking solution. The plate was incubated for 1 h with shaking. After three washes with PBS/T, each for 3 to 4 min, the antigen-antibody reaction was probed using the TMB Microwell peroxidase substrate system (Kirkegaard & Perry Laboratories, Gaithersburg, Md.), and color was allowed to develop for 20 min. The enzyme reaction was stopped by addition of 100 µl of 1 M H₃PO₄. The OD was measured at 450 nm.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Canine antibody responses to IRs. To determine whether any IRs (Fig. 1) were antigenic in dogs, serum samples from eight animals that were infected with B. burgdorferi by tick inoculation were used. Blood samples were collected at 8 or 9 weeks postinoculation and analyzed for anti-IR antibody responses using peptide-based ELISAs. All of the eight dogs had a strong antibody response to IR₆ and a moderate to strong reactivity with IR₂ (Fig. 2). None of the eight dogs produced an antibody response to IR₁, IR₃, IR₄, or IR₅ except dog A96-2/2, which showed a very low level of antibody to IR₄ (data not shown).

View larger version (32K): [in this window] [in a new window]   FIG. 2.   Antigenicity of invariable regions of VlsE. Serum samples were collected from eight dogs at 0 (Pre-) and 8 or 9 weeks post (Post-)-inoculation. Animals were infected by tick inoculation. Antibody levels to the two IRs were assessed using C2 or C6 peptide-based ELISA. The baseline (OD = 0.096) was calculated from the mean OD value plus 3 SDs of prebleeds collected from the eight dogs when the two peptide-biotin conjugates were individually used as ELISA antigen.

IR₆ alone may be enough as a diagnostic probe. Our analysis of the antibody response to the IRs revealed that both IR₂ and IR₆ are immunodominant in dogs. However, the serum specimens we used were obtained at one time point during infection, and they may not represent the whole spectrum of antibody responses. The level of antibody to individual epitopes may fluctuate with time. To determine whether both the IR₂ and IR₆ sequences should be included in an ELISA in order to increase assay sensitivity, an analysis of the time course of the antibody responses to these two sequences was conducted. Serial bleeds collected from nine animals between 0 and 17 weeks after they were infected with B. burgdorferi by tick inoculation were used. Sera were analyzed for antibody reactivity with both IR₂ and IR₆, using peptide-based ELISAs. The response to IR₆ in all of the animals appeared earlier and was stronger than the response to IR₂, although specific antibodies to both sequences persisted throughout the study period (Fig. 3). These results indicated that IR₆ alone is sufficient for the serodiagnosis of canine Lyme disease.

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Antibody responses to IR2 and IR6 as a function of time postinfection. Serial blood samples were collected from nine dogs at 0, 3, 5, 7, 9, 13, and 17 weeks after tick exposure. Levels of antibody to the two IRs were assessed using C2 or C6 peptide-based ELISA. Mean OD values and standard errors are presented. The baseline (OD = 0.090) was calculated from the mean OD value plus 3 SDs of prebleeds collected from the nine dogs when the two peptide-biotin conjugates were individually used as ELISA antigen.

C₆ cannot be significantly shortened. C₆ is a 26-mer sequence. To determine whether its size could be reduced without affecting reactivity with canine antibodies, the C₆ peptide was epitope mapped. Three 14-mer overlapping peptides, C₆N, C₆M, and C₆C (Fig. 1), which reproduced the whole IR₆ sequence, were each used as ELISA antigens. Serum samples collected from eight dogs at 8 or 9 weeks after they were infected by tick inoculation were tested. Although all of the three overlapping peptides reacted with most of the bleeds, their OD values were much lower than with C₆ (Fig. 4), indicating that the C₆ length could not be shortened significantly.

View larger version (39K): [in this window] [in a new window]   FIG. 4.   Epitope mapping of IR6. Serum samples were collected from eight dogs at 8 or 9 weeks after tick inoculation. Levels of antibody to C6 and each of the overlapping 14-mers were assessed using peptide-based ELISA. The baseline (OD = 0.091) was calculated from the mean OD value plus 3 SDs of the eight prebleeds reacted individually with each of the overlapping peptides and C6.

Sensitivity of the C₆ ELISA for diagnosing canine Lyme disease. To assess diagnostic sensitivity of the C₆ ELISA, serial bleeds were collected from 33 dogs at 0 to 17 weeks and from four of these animals up to 69 weeks after the animals were infected with B. burgdorferi by tick bite. The bleeds were taken at 2- to 4-week intervals and assessed for antibody responses to IR₆ using C₆ alone as ELISA antigen. All of the infected animals started to show a strong antibody response at 4 or 5 weeks postinfection (Fig. 5). Although none of the animals (0 of 11) had a detectable antibody response at 2 weeks postinfection, 30% of the infected dogs (7 of 22) became positive as early as 3 weeks postinfection, indicating that this assay is capable of detecting an early infection. These responses lasted for the entire experimental period, up to 69 weeks. Thus, a chronic infection also was detectable with the C₆ ELISA. The serum specimens collected at 2 and 3 weeks postinfection were also analyzed by an ELISA in which a mixture of C₂ and C₆ was used as antigen. No additional positives were detected (data not shown). These data further confirmed that anti-IR₆ antibody alone can be used as an indicator of B. burgdorferi infection in dogs.

View larger version (17K): [in this window] [in a new window]   FIG. 5.   Sensitivity of the C6 ELISA for detection of early and late infection. Thirty-three dogs were infected by tick inoculation. Serial blood samples were collected from 11 dogs (group I) at 0, 2, 4, and 8 weeks and from the remaining animals (group II) at 0, 3, 5, 7, and 9 weeks postinoculation. From 13 dogs of group II, serum sampling continued at 17 weeks, and sampling from 4 dogs continued until week 69 postinoculation. Levels of antibody to C6 were assessed using the C6 ELISA. The cutoff line (OD = 0.376) was defined as the mean OD value plus 5 SDs of serum samples collected from 70 healthy dogs from an area where Lyme disease is not endemic.

Specificity of the C₆ ELISA. B. burgdorferi shares similar antigenic proteins, such as flagellins (2), heat shock proteins (20), and common antigens (8), with other pathogenic and nonpathogenic bacteria. Infections with these bacteria may cause a false positive in Lyme disease serology (21). Moreover, vaccination, especially with killed whole spirochetes, makes serology of canine Lyme disease even more complicated. To assess the specificity of the C₆ ELISA, serum specimens from 70 healthy dogs were analyzed for anti-IR₆ antibody. No significant antibody response was detected (Fig. 6). The mean ELISA OD of these sera plus 5 SDs was used as the cutoff value. None of the serum samples from 14 dogs with leptospirosis (n = 5), RMSF (n = 2), dirofilariasis (n = 5), babesiosis (n = 1), or ehrlichiosis (n = 1) contained a detectable anti-IR₆ antibody titer (Fig. 6). In addition, all 15 sera were negative from dogs that were vaccinated with either the OspA vaccine (n = 5), a whole-spirochete vaccine (n = 5), or the DA₂PPLCV puppy-vaccines (n = 5). These data indicated that the C₆ ELISA is not only very specific to B. burgdorferi but also uniquely specific for infection with B. burgdorferi, as opposed to vaccination with fixed B. burgdorferi or its antigens or with other vaccines commonly administrated to dogs.

View larger version (10K): [in this window] [in a new window]   FIG. 6.   Specificity of the C6 ELISA. Seventy negative control sera (healthy control) were collected from dogs in an area where Lyme disease is not endemic. Fourteen serum specimens (other infections) were collected from dogs with leptospirosis or RMSF or infected with Dirofilaria, Babesia, or Ehrlichia spp. Fifteen serum specimens (Vaccinated) were collected from dogs that had been vaccinated with the OspA vaccine (n = 5), a whole-spirochete vaccine (n = 5), or the DA2PPLCV vaccines (n = 5). Antibody levels were assessed using the C6 ELISA. The cutoff line (OD = 0.376) was defined as the mean OD value plus 5 SDs of the 70 healthy controls.

Comparison of conventional serodiagnostic methods with the C₆ ELISA. A panel of 55 clinical sera was used to compare the sensitivity of the C₆ ELISA with those of the KELA and immunoblot assay. This panel had been originally submitted for the serodiagnosis of canine Lyme disease. The C₆ ELISA detected 19 positive samples, while the KELA and immunoblot assays each produced 20 positive results (Table 1), thus yielding similar sensitivities. This serum panel also was analyzed by a peptide-based ELISA in which combined C₂ and C₆ were used as antigen. No enhanced sensitivity was achieved, confirming once again that C₂ is not required for the serodiagnosis of dog Lyme disease (data not shown).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

B. burgdorferi sensu lato is able to infect a variety of vertebrate hosts. Lameness is the key clinical sign of Lyme disease in dogs. Unfortunately, lameness is not a pathognomonic sign, and therefore reliable serology is important in canine Lyme disease diagnosis.

Blood samples collected from experimentally infected dogs were analyzed for antibody responses to all of the six IRs of VlsE. Unlike humans and monkeys but as with mice (15), all dogs responded to both IR₂ and IR₆, and one animal also responded weakly to IR₄ (Fig. 2). The evaluation of the levels of antibody to IR₂ and IR₆ in serial bleeds confirmed that at no time during the course of infection did the anti-IR₂ antibody level surpass that of the anti-IR₆ antibody (Fig. 3). Moreover, the anti-IR₆ response appeared earlier than the response to IR₂, and the two peptides, when combined in an ELISA, yielded no enhanced sensitivity when compared to C₆ used alone. Taken together, these results demonstrated that C₆ alone is sufficient as a diagnostic probe.

Our antigenic analysis of the IRs, using dog serum, echoes our previous finding that IR₂, IR₄, and IR₆ are antigenic in mice (15). These results also are in agreement with the conclusion drawn from our previous studies, namely, that these three IRs are exposed at the surface of the VlsE molecule (13, 17). Features of protein domains such as surface accessibility, hydrophilicity, flexibility, and proximity to a site recognized by helper T cells are all important in positively determining domain antigenicity (5). Unlike T-cell epitopes, all B-cell epitopes are presumably exposed at the antigen's surface (4).

The full length of the C₆ peptide is probably required for optimal antigenicity, as none of the 14-mers examined was as antigenic as the C₆ 26-mer (Fig. 4). In monkeys (n = 10), neither of the 14-mers examined was antigenic, except for C₆C, which yielded a low OD value with the serum of one animal (16). In humans (n = 10), only C₆M was antigenic and only in four patients, whereas in mice (n = 10), all of these 14-mers were antigenic or strongly antigenic in several animals (16). The C₆ epitopes recognized by dogs resemble those discerned by the antibody response of primate hosts, in that in most animals, C₆M was antigenic, albeit less than C₆ itself.

The antigenicity of the whole VlsE molecule has been investigated by Lawrenz and colleagues (12). These authors used a whole recombinant VlsE protein cloned from B. burgdorferi sensu stricto strain B31 as the ELISA antigen. Their study indicated that the recombinant VlsE ELISA could be a sensitive and specific method for the serodiagnosis of Lyme disease (12). Our previous antigenic analysis has revealed that IR₆ is the only immunodominant IR in humans (13, 14). Since VlsE is a variable antigen, its six variable regions (13, 23) (Fig. 1), regardless of their antigenicity, should be of no value for serodiagnosis. VlsE also contains two invariable domains at the amino and carboxyl termini, which collectively comprise one half of the entire VlsE molecule length (13, 23). It is unknown whether these two invariable domains are antigenic or have any diagnostic value. Recent data, however, indicate that these two invariable domains may not be conserved even among strains within the genospecies B. burgdorferi sensu stricto (9), although they remain unchanged during antigenic variation (23). Their role in serodiagnosis remains to be investigated. Our previous survey of a large number of human serum specimens collected from both U.S. and European Lyme disease patients demonstrated that IR₆ (C₆) alone can serve as a probe for the universal serodiagnosis of human Lyme disease (14, 17). The results presented herein strongly support the conclusion that the single probe IR₆ (C₆) is sufficient for canine Lyme disease serodiagnosis.

When a single peptide sequence is used as a diagnostic probe, its antigenic conservation and antigenicity must be considered. The IR₆ sequence remains unchanged during antigenic variation (23), and it is both structurally and antigenically conserved among pathogenic B. burgdorferi strains and genospecies (13). It is so immunodominant that all experimentally infected animals, including mice and monkeys in previous studies (13, 14) and dogs in this study, produce an early, persistent, and strong antibody response to this sequence. Its antigenicity was also underscored by surveys of a large number of human samples. When this sequence was used in ELISA, diagnostic sensitivities of >80% for early Lyme disease and nearly 100% for late Lyme disease were achieved (14, 18). Previous and current antigenic analyses indicate that mice and dogs respond to IR₆ as well as or better than monkeys and humans do (13-15, 18). Within 6 weeks, all experimentally infected monkeys produced a detectable antibody response to C₆ (14), although 5 weeks were required for all of the infected dogs to show a high titer of antibody to this peptide. Similarly, the sensitivities of the C₆ ELISA were 42, 72, 80, 83, and 90% with human samples collected at 1, 2, 3, 4, and 5 to 8 weeks, respectively, after the disease onset (14).

The C₆ ELISA was able to detect both early and chronic infections. Although we can offer no proof that the dogs remained infected, we have preliminary evidence that the C₆ antibody level diminishes significantly within a 12-week period following treatment of dogs with antibiotics (Liang et al., unpublished data). In contrast, all of the four (untreated) dogs tested herein maintained a strong anti-IR₆ response for at least 69 weeks postinfection (Fig. 5).

Serology of Lyme disease may be complicated by the low specificity of the conventional assays, in which whole-cell lysates of cultured spirochetes are used as antigen. Low specificity may be caused by cross-reactive antigens shared between B. burgdorferi and other pathogenic and nonpathogenic bacteria. Examples of such antigens are bacterial heat shock proteins (20), flagellin (2), and other common antigens (8). Unrelated bacterial infections may thus elicit antibodies that react with B. burgdorferi antigens, causing false positive results. In contrast to the existing diagnostic techniques, the C₆ ELISA is expected to be highly specific. None of the blood samples collected from 14 dogs with leptospirosis or RMSF or infected with Babesia, Ehrlichia, or Dirofilaria spp. contained detectable antibody to C₆. In fact, except for VlsE, which is expressed solely by B. burgdorferi, no other protein sequences homologous to IR₆ could be identified by BLAST searches in the National Center for Biotechnology Information database (13). More importantly, vaccination with the recombinant OspA or whole-spirochete vaccines did not induce an antibody response cross-reactive with C₆ (Fig. 6). In contrast, bacterin immunization resulted in multiple strong bands on B. burgdorferi immunoblots (6). These limited data suggest that the C₆ ELISA is not only specific but also usable in the current vaccination era.

It was possible to establish cutoffs in the C₆ ELISA that yielded 100% sensitivity and 100% specificity when the assay was evaluated using experimentally infected beagles (Fig. 5). The assay did not give false positive results when testing dogs that were known to be uninfected with B. burgdorferi or dogs infected with other organisms. As already mentioned above, it was of considerable interest that the assay did not react with sera from dogs that had been vaccinated with either the whole-cell or recombinant OspA commercial vaccine for Lyme disease or the common puppy shots (Fig. 6). This makes the assay more valuable in detecting infected animals despite their vaccinal status. However, when the assay was applied to 55 serum samples submitted to a diagnostic laboratory by veterinarians who requested Lyme serology, 5 samples were misclassified by the C₆ ELISA (Table 1) if immunoblot or KELA was used as the "gold standard" (10). This discrepancy may be explained by the fact that these 55 samples were not collected from well-defined clinical cases. It will be necessary to further assess these assays using a larger number of serum specimens from dogs with clinically well-defined Lyme disease. Unfortunately, the clinical diagnosis of Lyme disease is much more difficult in dogs than in humans, and such clinical samples are not easily available in large numbers. Experimental infection may be the best alternative to examine the potential of new serologic assays for canine Lyme disease, as we have done in this paper.