Western immunoblot analysis for distinguishing vaccination and infection status with Borrelia burgdorferi (Lyme disease) in dogs

Transcription

1 J Vet Diagn Invest 11: (1999) Western immunoblot analysis for distinguishing vaccination and infection status with Borrelia burgdorferi (Lyme disease) in dogs David T. Gauthier, Linda S. Mansfield Abstract. Serodiagnosis of Lyme borreliosis in dogs is complicated by the use of commercially available Lyme disease vaccines that may cross-react with certain diagnostic assays. Western immunoblotting may be used to distinguish between dogs naturally exposed and those vaccinated against Borrelia burgdorferi. Because current vaccines are not 100% efficacious and dogs may be vaccinated after natural exposure, certain dogs may show serum antibody responses against both natural and vaccine exposure (dual status). In this study, samples from 17 nonexposed, 17 B. burgdorferi-bacterin vaccinated, 13 naturally exposed, and 8 dual-status dogs were tested by western immunoblot to determine if dual-status dogs could be reliably differentiated from naturally infected or vaccinated dogs. Reaction to outer surface protein A antigen of B. burgdorferi (31 kd) was a consistent marker for vaccination, appearing in all samples from vaccinate and dual-status dogs and in no samples from single-status naturally exposed dogs. Antibodies to 4 bands, at 80, 39, 29, and 28 kd, were present in all naturally infected and dual-status dogs. No samples from vaccinated or nonexposed dogs were reactive to all 4 of these bands simultaneously. Thus, vaccine and natural exposure produce differing antibody responses, whereas dual-status dogs produced the full antibody response of both types of exposure. Naturally acquired Borrelia burgdorferi infection in dogs may cause episodic fever, lethargy, anorexia, lymphadenopathy, and polyarthritis. 1,19 Less commonly, effects on cardiac, renal, neurologic, and ocular systems may occur. 12,17,26 Any combination and severity of these signs may be seen, making clinical diagnosis of Lyme disease difficult. Serodiagnosis of canine Lyme disease is constrained by low specificity, lack of agreement between current assays for B. burgdorferi antibody, and inability to distinguish between natural infection and vaccination status. The indirect fluorescent antibody assay (IFA) and enzyme-linked immunosorbent assay (ELISA) tests detect anti-borrelia serum Ig 13,20 but do not distinguish between naturally induced and vaccinal antibody titers. Borrelia IFA and ELISA tests may be confounded by cross-reactivity among antibodies to other Borrelia species, leptospiral serovars, and oral treponemes. 25,27,28 The inability to distinguish between naturally induced and vaccinal antibody titers can be a problem in serodiagnosis of canine Lyme disease because both killed-bacterin a,b and recombinant B. burgdorferi outer surface protein A (OspA c ) Lyme disease vaccines are currently licensed for use in dogs. Immunogenicity of B. burgdorferi antigens depends on the mode of introduction. 1,3,15,29 Western immunoblotting detects antibody response to individual B. burgdorferi proteins From the Animal Health Diagnostic Laboratory, A12 Veterinary Medical Center, College of Veterinary Medicine, Michigan State University, East Lansing, MI Received for publication April 15, and thus may be used to distinguish between natural exposure and vaccination. Serodiagnosis of Lyme disease in dogs is further complicated by the presence of dogs of dual status, i.e., that have been both vaccinated and naturally exposed to B. burgdorferi. Vaccination with Lyme bacterin has been demonstrated to be efficacious, although not in all cases. 18 Mouse protection assays also suggest less than 100% efficacy of vaccination. 2 Dogs may also be vaccinated after natural infection has occurred. Thus, concurrent vaccine and natural antibody responses may be the result of natural infection before or after vaccination. The ability of western immunoblots to distinguish between dogs that are bacterin-vaccinated, naturally infected, or both has been indicated 16 but not confirmed by studies using dual-status animals. This study was focused on determining criteria for distinguishing between naturally infected and vaccinated dogs by identification of status-specific reactive protein bands on western immunoblots. Immunoblots of serum samples from dual-exposure dogs were then examined using these criteria to determine if they could be reliably differentiated. The results confirm that western blots can be utilized for this purpose. Materials and methods Experimental groups. Four experimental groups of dogs were used for this study. Group 1 consisted of 13 dogs that were antibody positive due to natural infection with B. burgdorferi. Age and sex were evenly distributed across the group. These dogs came from Lyme disease-endemic areas of the country, including 259

2 260 Gauthier, Mansfield Wisconsin 6 and New England 7,22. All dogs were positive for total IgG antibody to B. burgdorferi by IFA (titers of 1: to 1:20,480). A titer of 1: had been previously determined as the minimum for a positive designation. Case histories were examined for the clinical signs of Lyme disease to corroborate IFA results. Criteria for inclusion as positive controls were 1) history of travel to a Lyme disease-endemic area, 2) no history of vaccination against Lyme disease, 3) no previous degenerative joint disease, 4) positive response if treated with tetracycline or amoxicillin, and 5) at least 2 of 5 clinical signs known to be associated with Lyme disease (anorexia, lethargy, lameness, lymphadenopathy, or fever). 1,16,19 Any of the following diagnoses excluded dogs from group 1: positive reaction to Rocky Mountain spotted fever (Rickettsia rickettsii), antinuclear antibodies, lupus erythematosus, rheumatoid factor, or positive Coomb s test. In addition, complete blood count, serum biochemistry, joint radiography, or urinalysis results suggestive of another differential diagnosis excluded dogs from the positive control group. Group 1 also included serum from 1 dog determined to be positive for natural infection via western blot at another laboratory. d The 8 group 2 dogs had concurrent natural infection and vaccination antibody responses. Four of these dogs were obtained in the same manner as those in group 1. Selection criteria for these dogs were the same as for those in group 1 except that these dogs also had been vaccinated for Lyme disease. Group 2 also included serum d from 4 dogs evaluated as positive for natural infection and vaccination via the same western blot analyses as used for the single dog in Group 1. The western blot used by this laboratory was validated by evaluation of serial weekly blood samples from dogs experimentally exposed to ticks and comparison to dogs of known vaccination history. Group 3 consisted of 17 dogs vaccinated in the laboratory with one of 2 Lyme disease vaccines. a,b These dogs were all from nonendemic areas and had no history of travel to endemic regions. Age and sex were evenly distributed. Negative antibody response to B. burgdorferi was confirmed by IFA at the beginning of the experiment. Nine dogs received Solvay vaccine, and 8 dogs received Ft. Dodge vaccine. Both vaccines were administered intramuscularly in 1-ml doses twice over a 2-wk interval according to the manufacturers instructions. Serum was collected for IFA and western immunoblot analyses at 2 3 wk and 6 8 mo after vaccination. Two blood samples taken 2 wk after vaccination were subsequently damaged and could not be utilized for the western immunoblot. Therefore, the sample size for postvaccination 1 is 15, and the sample size for postvaccination 2 is 17. Group 4 dogs were the 17 Group 3 dogs before vaccinations were administered. These dogs had no vaccination and no natural exposure to B. burgdorferi based on case history. All dogs in this group tested for antibodies to B. burgdorferi by IFA. IFA. Serum samples were assayed for total IgG antibody to B. burgdorferi with an IFA test using the following procedure. Twelve-well glass slides were plated with B. burgdorferi (AT , B31 type strain, 3 4 passage) spirochetes to evenly cover each well, dried, and fixed for 5 min in acetone. The wells were exposed to serial dilutions of canine sera beginning at 1: and incubated at 37 C for 30 min. A positive control was included. e The wells were washed with phosphate-buffered saline (PBS; 0.01 M Na 2 HPO 4, 0.15 M NaCl, ph 7.3) and treated with 1: 25 goat anti-dog IgG fluorescein isothiocyanate conjugated antibody f in 1:1,000 Evans blue in PBS. After incubation for 30 min at 37 C, the slides were washed with PBS and observed using a fluorescence microscope. 8 Titer was established as the reciprocal of the last dilution with positive fluorescence. Western immunoblot. Serum samples were tested for antibody response to specific B. burgdorferi antigens using a western immunoblot assay. Borrelia burgdorferi (AT , B31 type strain, 3 4 passage) was cultured to high concentration in BSK-H medium h with 6% rabbit serum h at 37 C. Spirochetes were centrifuged 3 times at 10,000 g for 30 min each time, washing each time with Tris buffer (0.005 M Tris, M ethylenediamine tetraacetic acid [EDTA], 0.1 M NaCl, ph 7.4). The pellet was resuspended in distilled water to 0.5 mg/ml as previously described. 5 An equal volume of lysing buffer was added (0.1 M Tris, 10% glycerol, 2% sodium dodecyl sulfate [SDS], 5% -mercaptoethanol, 0.1% bromphenol blue, ph 8.0), and the solution was denatured at 95 C for 5 min. After denaturation, spirochete solutions were stored at 20 C until used. Immediately before electrophoresis, spirochete solutions were thawed and denatured again at 95 C for 5 min. After cooling, 100 g of the lysed and denatured spirochete solution was electrophoresed at 175 V on a 15% acrylamide gel until the dye front reached the bottom of the gel. Gels of 10%, 12%, and 10 20% (continuous) acrylamide concentration were also utilized, although all reported results are derived from 15% gels, which gave the best separation of relevant bands. A 74-mm-long loading well was used. For reference purposes, biotinylated broad range SDS-polyacrylamide gel electrophoresis (PAGE) size standard markers i were run on the same gel. The separated proteins were transferred to PVDF membrane. i The transfer was carried out in buffer (0.4 M Tris, 3 M glycine, 20% methanol) in distilled water at 100 V ( 180 ma) for 2 hr followed by 150 V for 30 min. The

3 Western blot for B. burgdorferi status 261 membranes were separated from the gels, incubated overnight in blocking buffer (0.05 M Tris, 0.15 M NaCl, M EDTA, 0.05% Tween-20, 2% bovine serum albumin), and air dried. Membranes were either used immediately after drying or stored at 20 C until used. Membranes were placed into a slot-blot apparatus, i dividing the blot into lanes. Separate lanes were incubated with 1:100 serum samples in blocking buffer at room temperature overnight. The immunoblots were then washed 3 times for 5 min each in wash buffer (0.05 M Tris, 0.15 M NaCl, M EDTA, 0.05% Tween-20, ph 7.4) and incubated for 3 hr with horseradish peroxidase (HRP)-labeled goat anti-dog IgG antibody (H L) k at a concentration of 1:500. Immunoblots were washed with wash buffer as described above and developed with an AEC (3-amino-9-ethylcarbazole) kit h for 2 3 minutes. Molecular weights of protein bands were determined using biotinylated SDS-PAGE size standard markers run adjacent to B. burgdorferi lysed antigen. ExtrAvidin h -HRP with AEC developer was used to visualize standards. A linear semilogarithmic regression was calculated from markers between and including 97.4 kd and 21.5 kd. Band sizes were extrapolated from this trendline. Cross-reactivity. The ability of antibodies from dogs exposed to Leptospira spp., Rocky Mountain spotted fever (Rickettsia rickettsii), Ehrlichia canis, E. risticii, Babesia canis, and oral Treponema spp. to cross-react on the immunoblots was evaluated. These samples form group 5. Seven groups of samples were tested for cross-reactivity: 1) 2 samples positive for Leptospira interrogans serovars canicola, grippotyphosa, hardjo, autumnalis, and pomona (titers of 1: 400 and 1:800; microtiter agglutination test), 2) 5 samples antibody positive by IFA for E. risticii (1:320 to 1:), l 3) 5 samples antibody positive by IFA for E. canis (1:320 to 1:81,920), l 4) 1 sample antibody positive by IFA for Rocky Mountain spotted fever, 5) 4 samples antibody positive by IFA for B. canis, 6) 5 samples from dogs exhibiting advanced periodontal disease, and 7) 2 samples from dogs antibody positive by IFA for E. canis ( 1:240), E. risticii ( 1:80), and B. canis ( 1:80). l In all cases, procedures were identical to those described above for western immunoblotting. Results IFA. IFA results of control group (prevaccine) and vaccine group samples are shown in Table 1. All samples were positive to at least a titer of 1: by 2 3 weeks postvaccination. However, by 6 8 months, 4 Solvay-vaccinated dogs and one Ft. Dodgevaccinated dog were ( 1:) by IFA. The Table 1. Reciprocal IFA titers of group 3 (vaccinated) dogs before vaccination (Prevacc), 2 3 weeks after vaccination (Postvacc 1), and 6 8 months after vaccination (Postvacc 2). Negative results indicate an IFA titer below 1:. Titers for dogs receiving either Solvay or Ft. Dodge vaccines are indicated separately. Fort Dodge vaccine (n 8) Prevaccination Solvay vaccine (n 9) Prevaccination mean time of sampling after vaccination was not significantly different between groups of dogs receiving Ft. Dodge and Solvay vaccines (P 0.1; Student s t- test). Naturally exposed nonvaccinated dogs exhibited IFA titers from 1:20,480 to 1:. The sample with the lowest positive titer in this group (1:) had previously tested positive by ELISA at another laboratory. Dual-status dogs had IFA titers of 1:320 to 1: 20,480. Western immunoblot. Immunoblots of all dogs showed status-specific patterns with particular bands above and including 27 kd. Bands below this did not exhibit status-specific patterns and were excluded from the analysis. Additionally, some bands above this mass were determined to have nonspecific and inconsistent reactivity and were likewise excluded. Gels with other acrylamide concentrations were used to more closely examine bands in higher ( 60 kd) and lower ( 30 kd) size ranges, but 15% was determined to be optimal for separation of bands of interest in this study. Total reactivity of canine sera from the 4 infection groups to various B. burgdorferi proteins of interest is shown in Table 2. All visible bands that could be positively identified as such were recorded, regardless of intensity. Some status-specific differences in banding intensity were observed. Representative blots from all infection groups are shown in Fig. 1. Group 1: natural infection. Reaction to 80-, 62-, 54-, 39-, 29-, and 28-kD bands was seen in blots of all naturally infected dogs studied. The latter 2 bands appeared as a distinct doublet directly under the level of the 31-kD protein standard (carbonic anhydrase) (Fig. 1). The 62- and 39-kD bands appeared as distinct bands on the leadings edges of diffuse bands at kd and kd. Almost all (12 of 13; 92.3%) dogs in this group reacted to a band at 73 kd. Reactivity to

4 262 Gauthier, Mansfield Table 2. Frequency of protein bands reactive by western blot for dogs of each infection status (group). Entries are given as number of serum samples reactive (% samples reactive) to each protein band. Prevaccination (Prevacc) are nonexposed dogs. Cross-reaction (Crossrx) includes dogs positive for non-borrelia organisms tested for cross-reactivity to whole cell B. burgdorferi lysate. 1 (Vacc 1) are the prevaccination dogs 2 3 weeks after the last vaccination. 2 (Vacc 2) are the same dogs 6 8 months after vaccination. Dogs vaccinated with either Solvay or Ft. Dodge vaccines are pooled in prevaccination and both postvaccination groups. NI naturally infected dogs with no vaccination status. Dual dogs with concurrent natural infection and vaccination. Infection status (group)* Band size (kda) NI (I) n 13 Dual (II) n 8 Vacc. 1 (IIIa) n 15 Vacc. 2 (IIIb) n 17 Pre-vacc. (IV) n 17 Cross-rx (V) n (92.3) 10 (76.9) 2 (15.4) 10 (76.9) 7 (53.8) 3 (23.1) 2 (15.4) 6 (75.0) 4 (50.0) 4 (50.0) 7 (87.5) 2 (25.0) 7 (46.7) 6 (40.0) 6 (40.0) 2 (13.3) 2 (13.3) 8 (53.3) 1 (6.7) 10 (66.7) 9 (60.0) 15 (100.0) 15 (100.0) 2 (13.3) 3 (20.0) 8 (53.3) 5 (29.4) 3 (17.6) 9 (52.9) 17 (100.0) 17 (100.0) 3 (17.6) 3 (17.6) 4 (23.5) 7 (41.2) 4 (23.5) 14 (66.7) 2 (9.5) 14 (66.7) 1 (4.8) 1 (4.8) 3 (14.3) 17 (81.0) 1 (4.8) 13 (61.9) 2 (9.5) 2 (9.5) * No. of serum samples reacting to band by immunoblot (%). Figure 1. Representative western immunoblots of dogs in groups 1 4. Band sizes (kd) are given to the right of the image. Positions of molecular mass markers are given to the left of the image. I Dogs with concurrent vaccination and natural infection; II naturally infected dogs without vaccination; III vaccinated dogs at 2 weeks postvaccination; IV dogs with no exposure to B. burgdorferi or Lyme disease vaccine. any protein at 31 kd was not observed in naturally infected dogs. The OspA antigen of B. burgdorferi appears at this location in vaccinated dogs. 3,9,15 Limited (3 of 13; 23%) reactivity to an antigen of B. burgdorferi at 34 kd was observed. The position of this band and the reactivity of an identically positioned band in dual-status and vaccinated dogs indicate that it is OspB of B. burgdorferi. Group 2: dual status. In dual-status dogs, reactivity to all bands except 31 (OspA) and 34 (OspB) kd was similar to that in the naturally infected dogs. All dogs had antibodies to 80-, 73-, 62-, 54-, 50-, 39-, 29-, and 28-kD proteins. All dual-exposure dogs showed reactivity to OspA, and most (7 of 8; 87.5%) had antibodies against OspB. Group 3: vaccinated. All 2-weeks postvaccination dogs reacted to both OspA and OspB antigens. Limited reactivity was seen in 2-weeks postvaccination dogs to the 80-kD (7 of 15; 46.7%), 68-kD (6 of 15; 40.0%), 54-kD (8 of 15; 53.3%), 39-kD (9 of 15; 60.0%), 29-kD (2 of 15; 13.3%), and 28-kD (3 of 15; 20.0%) bands. No serum from vaccinated dogs was reactive to both 29- and 28-kD bands. None of the bands were increased in intensity at 6 months, and many were either attenuated or absent. Antibody response to OspA and OspB was still detectable in all samples, but OspB had in some instances decreased to the limits of detection for this test. Reaction to the 80- kd band was observably weaker in vaccinated dogs than in naturally infected or dual-status dogs (Fig. 1). This difference was much more apparent 6 months af-

5 Western blot for B. burgdorferi status 263 ter vaccination, at which time this band had disappeared altogether in some samples. Group 4: nonexposed. In nonexposed dogs, reactivity to the kD band was most common (7 of 17; 41.2%). Reactivity to other bands, when present, was weak. No nonexposed dogs were reactive to the band at 39 kd. Cross-reactivity. Among samples studied for crossreactivity of antibodies to non-borrelia organisms on B. burgdorferi immunoblots, none displayed combinations of bands that mimicked vaccinated, naturally exposed, or dual-status samples. Because samples containing antibodies to the same non-borrelia organism did not show consistent patterns of band reactivity, they were not given a separate status (e.g., E. risticii positive). Therefore, all potentially cross-reactive samples were combined into 1 group (Table 2). Frequency of reactivity to most bands in the cross-reactive group was greater than that in the nonexposed group. The most commonly reactive bands in the cross-reactive group were 73 kd (14 of 21; 66.7%), kd (14 of 21; 66.7%), kd (17 of 21; 81.0%), and 29 kd (13; 61.9%). Discussion In this study, all dogs vaccinated against B. burgdorferi had antibodies that detected OspA antigen at 31 kd and OspB antigen at 34 kd, with 1 exception for the latter in the dual-status group. A strong reaction to these antigens in vaccinated dogs has been observed previously. 3 In this study, intensity of reactivity to the OspA band remained high in vaccinated animals regardless of time postvaccination up to 6 months. Neither unexposed nor naturally exposed dogs were reactive to the OspA antigen, with 1 very weak exception in a nonexposed dog. OspA has greatly reduced expression during tick feeding and transmission of B. burgdorferi, offering 1 explanation why naturally infected dogs do not mount a reaction to this major surface antigen. 9 In this study, the intensity of reactivity to OspB appeared to drop off more rapidly than that to OspA and was extremely weak in some samples 6 months after vaccination. OspB was absent in 1 of 8 dual-status dogs and weakly reacted in 3 of 13 naturally infected dogs. OspA was a more reliable marker for vaccination status because of the waning of antibody response to OspB over time and the potential for OspB reactivity in naturally infected dogs. Using OspA as a marker for prior vaccination, it was possible to distinguish dual-status and vaccinated dogs from naturally infected and dogs in this study. Criteria were then established for separating the naturally infected from the dogs. Six bands, at 80, 62, 54, 39, 29, and 28 kd; were present in all samples from naturally infected dogs. The band at 80 Table 3. Criteria developed by this study for differentiating infection status. Status Negative Vaccinate Naturally Infected Dual Status Bands required none 31 kda 80, 39, 29, 28 kda 80, 39, 31, 29, 28 kda kd was well separated from all other bands and reacted intensely in all samples from naturally infected dogs. Two samples from the nonexposed group did react with this band, but at very low intensity. The band at 39 kd was found in all naturally infected dogs, in no dogs from the nonexposed group, and in 1 dog from the cross-reactive group. Care must be exercised to distinguish this band from the kD diffuse band, which was commonly seen in dogs from all groups (Fig. 1). Although both 29- and 28-kD bands were found in the nonexposed and cross-reactive groups, only 1 sample contained antibodies to both. This sample was not reactive to 39- or 80-kD bands. The 62- and 54-kD bands were eliminated as possible markers because of potential difficulty in consistently distinguishing them from closely comigrating bands that were reactive in nonexposed dogs. Based on these observations, we established the presence of 80-, 39-, 29-, and 28-kD bands as the criterion for natural infection (Table 3). All naturally infected and no nonexposed or cross-reactive dogs fit this criterion. Banding patterns were much more similar between vaccinated dogs and dual-status dogs than between naturally infected and dogs. In general, reactivity of serum from vaccinated dogs to 41-kD bands was lower in number and intensity than that for dual-status dogs. When the criteria for natural infection of 80-, 39-, 29-, and 28-kD bands were applied to dual-status and vaccinate samples, all dual-status and no vaccinate samples tested positive for natural infection. Reaction to both 29- and 28-kD bands was very important in the differentiation of dual status vs. vaccinate samples; 3 vaccinate samples were reactive to 80-, 39-, and either 29- or 28-kD bands. All 4 bands were not persistent in vaccinated animals, decreasing in number recognized between the first and second postvaccination samples. Additionally, reaction to 80- kd bands in vaccinated dogs was visibly weaker than that in dual-status or naturally infected dogs. The 39-kD band cited in this study runs on the leading edge of the 41 kd protein flagellin. 4 Flagellin is an early appearing antigen in Lyme disease serodiagnosis 1,4,8,10,33,34 and is known to cross-react with a wide variety of other spirochetes and nonspirochetal bacteria. 25,28 Its occurrence in all experimental groups of this study is therefore not surprising and supports the idea

6 264 Gauthier, Mansfield that whole flagellar antigen is not a useful antigen in Lyme disease serodiagnosis. In other studies, the potential usefulness of recombinant internal fragments of B. burgdorferi flagellin in immunoblots 31,32 and ELISA 23,24 has been demonstrated. The identity of the 39-kD band was not determined here, but 2 possibilities are suggested. The 39-kD band could be a protein with roughly the same mobility on SDS-PAGE as flagellin. Alternatively, the 39-kD mass suggests that this protein may be P39, whose importance as a marker of early Lyme disease in humans 8,21 and in dogs 3 has been cited. Reaction at 39 kd using monoclonal antibodies is specific to naturally infected dogs, suggesting that differentiation of natural infection and vaccination might be accomplished using this protein. 3 Antibody production to P39 has been documented in mice infected with viable B. burgdorferi but not in mice infected with inactivated or noninfectious B. burgdorferi. 29 Because the 39-kD band was found in vaccinated dogs in this study, it is unlikely that this protein is P39. Repeating this study with recombinant P39 3 may be instructive. The identities of the 80-, 29-, and 28-kD bands were not determined here, but bands of similar sizes have been identified. A band of 83 kd has been described as occurring in all samples of naturally and artificially infected dogs. 15 Reaction was weaker to this band in artificially exposed dogs, as seen in the present study. This band may be identical to a 93-kD band detected in other studies 10,15,30,33,34 that has been cited as important in detecting late-stage disease in humans. 34 Comparison of the 80-kD band of this study to bands of similar masses in other studies may be difficult because of the degree of distortion that often occurs near the top of miniblots. Bands running just under the 31- kd OspA band have been cited in other studies 3,15,16 and were specific to natural infection. 3 The separation of this band into 2 bands, 29 and 28 kd, in this study may be attributed to differences in gel composition and SDS-PAGE conditions between studies. Because no bands found in this study reacted only with samples from naturally exposed dogs and not with vaccinated or dogs, it was necessary to create a criterion for positive samples based on the presence of several bands at once. Because all bands used in the positive criterion have the potential to individually appear on immunoblots of nonexposed or vaccinate samples, a small number of non-naturally exposed samples may be reactive against all of the positive bands simultaneously. In this case, an erroneous result would be produced. This result was not seen in the present research but may appear with larger sample sizes. To assess the degree of cross-reactivity of sera from dogs positive for non-borrelia organisms against whole cell B. burgdorferi lysate on immunoblots, tests were performed on 21 samples positive for 1 or a combination of the following agents: E. canis, E. risticii, B. canis, L. interrogans (multiple serovars), R. rickettsii, and oral Treponema spp. Dogs from the crossreactive group exhibited much the same pattern on immunoblot as did the prevaccinated nonexposed dogs. Certain bands (73, 65 64, 42 41, and 29 kd) were found in a higher frequency in the cross-reactive dogs than in nonexposed dogs. This result was expected, because many of the cross-reactive dogs had advanced periodontal disease or high antibody titers against non- Borrelia organisms. The majority of dogs in the nonexposed group were in good health and had not been in areas with high tick populations. Differences in western blot protocol may produce immunoblot banding patterns that differ between laboratories. Methods of antigen preparation, including sonication and/or denaturation of whole cells, purification, and passage number, may all affect the presence and reactivity of various antigens. Strain of B. burgdorferi used is also an important factor, especially in the lower mass ( 45 kd) proteins. 14 Additionally, reagent formulation, antibody dilutions, blotting conditions, and other considerations may cause discrepancies in recognized protein bands. Strategies have been devised to overcome the problems of interlaboratory variation and to increase the specificity and sensitivity of human Lyme disease serodiagnosis, including recombinant antigen immunoblots 31,32 and ELISA. 11,24 Interpretation of these results, however, remains dependent on the specific method used, and there are no universally accepted criteria for dogs. The protein bands cited in this study as the criterion for determination of natural infection (80, 39, 29, 28 kda) have not been tested against monoclonal antibodies, so caution should be exercised when equating them to bands of similar masses obtained with different blots. Further characterization of these bands and larger sample sizes are needed before these results can be extrapolated to populations and reliable sensitivity/specificity estimates can be made. This study has, however, confirmed that criteria can be established for distinguishing 4 B. burgdorferi infection statuses in dogs: nonexposed, vaccinated, naturally infected, and concurrently naturally infected and vaccinated. Acknowledgements We thank Dr. Ibulaimu Kakoma and Dr. Richard H. Jacobson for providing serum samples and Alice Murphy, Mary Rossano, Elisa Mazzaferro, and Leslie Marsh for technical assistance. Sources and manufacturers a. Solvay, Mendota Heights, MN. b. Fort Dodge Laboratories, Fort Dodge, IA.

7 Western blot for B. burgdorferi status 265 c. Rhone Merieux, Athens, GA. d. Dr. Richard H. Jacobson, Diagnostic Laboratory, College of Veterinary Medicine, Cornell University, Ithaca, NY. e. VMRD, Pullman, WA. f. ICN Biomedicals, Costa Mesa, CA. g. Carl Zeiss, Thornwood, NY. h. Sigma, St. Louis, MO. i. Bio-Rad, Hercules, CA. j. Schleicher & Shuell, Keene, NH. k. Kirkegaard & Perry Laboratories, Gaithersburg, MD. l. Dr. Ibulaimu Kakoma, Laboratories of Diagnostic Veterinary Medicine, University of Illinois, Urbana, IL. References 1. Appel MJG, Allan S, Jacobson RH, et al.: 1993, Experimental Lyme disease in dogs produces arthritis and persistent infection. J Infect Dis 167: Barthold SW, Bockenstedt LK: 1993, Passive immunizing activity of sera from mice infected with Borrelia burgdorferi. Infect Immun 61: Barthold SW, Levy SL, Fikrig E, et al.: 1995, Serologic responses of dogs naturally exposed to or vaccinated against Borrelia burgdorferi infection. J Am Vet Med Assoc 207: Berland R, Fikrig E, Rahn D, et al.: 1991, Molecular characterization of the humoral response to the 41-kilodalton flagellar antigen of Borrelia burgdorferi, the Lyme disease agent. Infect Immun 59: Bradford MM: 1976, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein dye binding. Anal Biochem 131: Burgess EC: 1986, Natural exposure of Wisconsin dogs to the Lyme disease spirochete. Lab Anim Sci 36: Ciesielski CA, Markowitz LE, Horsley R, et al.: 1989, Lyme disease surveillance in the United States, Rev Infect Dis 11:A Craft JE, Fischer DK, Shimamoto GT, Steere AC: 1986, Antigens of Borrelia burgdorferi recognized during Lyme disease: apearance of a new immunoglobulin M response and expansion of the immunoglobulin G response late in the illness. J Clin Invest 78: desilva AM, Telford SR, Brunet LR, et al.: 1996, Borrelia burgdorferi OspA is an arthropod-specific transmission-blocking Lyme disease vaccine. J Exp Med 183: Dressler F, Whalen JA, Reinhardt BN, Steere AC: 1993, Western immunoblotting in the serodiagnosis of Lyme disease. J Infect Dis 167: Gerber MA, Shapiro ED, Bell GL, et al.: Recombinant outer surface protein C ELISA for the diagnosis of early Lyme disease. J Infect Dis 171: Grauer GF, Burgess EC, Cooley AJ, Hagee JH: 1988, Renal lesions associated with Borrelia burgdorferi in a dog. J Am Vet Med Assoc 93: Greene RT, Levine JF, Breitschwerdt EB, et al.: 1988, Clinical and serologic evaluations of induced Borrelia burgdorferi infection in dogs. Am J Vet Res 49: Greene RT, Walker RL, Burgess EC, Levine JF: Heterogeneity in immunoblot patterns obtained by using four strains of Borrelia burgdorferi and sera from naturally exposed dogs. J Clin Microbiol 26: Greene RT, Walker RL, Nicholson W, et al.: Immunoblot analysis of immunoglobulin G response to the Lyme disease agent (Borrelia burgdorferi) in experimentally and naturally exposed dogs. J Clin Microbiol 26: Jacobson RH, Chang YF, Shin SJ: 1996, Lyme disease: laboratory diagnosis of infected and vaccinated symptomatic dogs. Semin Vet Med Surg Small Anim 11: Levy SA, Duray PH: 1988, Complete heart block in a dog seropositive for Borrelia burgdorferi: similarity to human Lyme carditis. J Vet Intern Med 2: Levy SA, Lissman BA, Ficke CM: 1993, Performance of a Borrelia burgdorferi bacterin in borreliosis-endemic areas. J Am Vet Med Assoc 202: Levy SA, Magnarelli LA: 1992, Relationship between development of antibodies to Borrelia burgdorferi in dogs and the subsequent development of limb/joint borreliosis. J Am Vet Med Assoc 200: Lindenmayer J, Weber M, Bryant J, et al.: 1990, Comparison of indirect immunofluorescent-antibody assay, enzyme-linked immunosorbent assay, and western immunoblot for the diagnosis of Lyme disease in dogs. J Clin Microbiol 28: Ma B, Christen B, Leung D, Vigo-Pelfrey C: 1992, Serodiagnosis of Lyme borreliosis by western immunoblot: reactivity of various significant antibodies against Borrelia burgdorferi. J Clin Microbiol 30: Magnarelli LA, Anderson JF, Kaufman AF, et al.: 1985, Borreliosis in dogs from southern Connecticut. J Am Vet Med Assoc 186: Magnarelli LA, Fikrig E, Padula SJ, et al.: 1996, Use of recombinant antigens of Borrelia burgdorferi in serologic tests for diagnosis of Lyme borreliosis. J Clin Microbiol 34: Magnarelli LA, Flavell RA, Padula SJ, et al.: 1997, Serologic diagnosis of canine and equine borreliosis: use of recombinant antigens in enzyme-linked immunosorbent assays. J Clin Microbiol 35: Magnarelli LA, Miller JN, Anderson JF, Riviere GR: 1990, Cross-reactivity of nonspecific treponemal antibody in serologic tests for Lyme disease. J Clin Microbiol 28: Munger RJ: 1990, Uveitis as a manifestation of Borrelia burgdorferi infection in dogs. J Am Vet Med Assoc 197: Schillhorn van Veen TW, Murphy AJ, Colmery B.: 1993, False positive Borrelia burgdorferi antibody titers associated with periodontal disease in dogs. Vet Rec 132: Shin SJ, Chang YF, Jacobson RH, et al.: 1993, Cross-reactivity between B. burgdorferi affects specificity of serotests for detection of antibodies to the Lyme disease agent in dogs. Vet Microbiol 36: Simpson WJ, Burgdorfer W, Schrumpf ME, et al.: 1991, Antibody to a 39-kilodalton Borrelia burgdorferi antigen (P39) as a marker for infection in experimentally and naturally infected inoculated animals. J Clin Microbiol 29: Volkman DJ, Walker RL, Gorevic PD, et al.: Characterization of an immunoreactive 93-kDa core protein of Borrelia burgdorferi with a human IgG monoclonal antibody. J Immunol 146: Wilske B, Fingerle V, Herzer P, et al.: 1993, Recombinant immunoblot in the serodiagnosis of Lyme borreliosis. Med Microbiol Immunol 182: Wilske B, Fingerle V, Preac-Mursic V, et al.: 1994, Immunoblot using recombinant antigens derived from different genospecies of Borrelia burgdorferi sensu lato. Med Microbiol Immunol 183: Zöller L, Burkard S, Schafer H: 1991, Validity of western immunoblot band patterns in the serodiagnosis of Lyme borreliosis. J Clin Microbiol 29: Zöller L, Cremer J, Faulde M: 1993, Western immunoblot as a tool in the diagnosis of Lyme borreliosis. Electrophoresis 14: