Serologic analyses of Peromyscus leucopus, a rodent reservoir for Borrelia burgdorferi, in northeastern United States

An enzyme-linked immunosorbent assay (ELISA) and indirect fluorescent-antibody test were used to detect antibodies to Borrelia burgdorferi, the causative agent of Lyme disease, in Peromyscus leucopus (white-footed mouse). Of the 661 mice captured in Connecticut, Rhode Island, and New York during 1980 and 1983 to 1987, 166 (25.1%) had antibodies to B. burgdorferi by ELISA. Comparative analyses of 210 serum specimens, collected in areas where Lyme disease is endemic, revealed a threefold difference in sensitivity between the ELISA (38.1% positive) and the indirect fluorescent-antibody method (12.4%). Although prevalence of seropositive P. leucopus was highest during June, elevated amounts of antibody (1:1,280 to 1:2,560) were detected in mice that harbored spirochetes during all seasons. Being reservoirs for B. burgdorferi, these rodents are suitable for monitoring spirochete infections at foci and should be included in field evaluations of control programs aimed at suppressing Lyme disease.

The association of I. dammini larvae and nymphs with P. leucopus is well documented (6,13,15,20,25). It is, therefore, appropriate to select this mammal for surveillance programs. Indirect fluorescent-antibody (IFA) methods have been used to detect antibody to B. burgdorferi (15,20), but among the mammals studied, the prevalence of seropositive white-footed mice has been relatively low. It is unclear whether these rodents have weak immunologic responses to B. burgdorferi or whether the results are due to low test sensitivity. Accordingly, this investigation was conducted to further evaluate an enzyme-linked immunosorbent assay (ELISA) developed earlier (24), to compare results with those obtained by IFA, and to determine whether ELISA is more suitable for monitoring immunologic responses to B. burgdorferi in P. leucopus populations.

MATERIALS AND METHODS
Sampling. White-footed mice were captured in Sherman box traps from woodlands in Connecticut, Rhode Island, and New York during 1980 and 1983 to 1987. Although rodents were collected during all seasons, the majority were ob-* Corresponding author. tained from May through August. Sites in Connecticut (East Haddam, Norwich, and Salem) are located in the southcentral or southeastern part of the state, a region where Lyme disease is prevalent in human populations (21,28). Mice were marked, released, and recaptured at East Haddam as previously described (20). In Rhode Island, P. leucopus was collected from inland areas, Block Island, Prudence Island, and Jamestown Island, whereas in New York, animals were captured in Westchester and Dutchess Counties. All study sites are within the geographical range of I. dammini. For controls, 60 additional mice were obtained from Newtown and Woodbridge, Conn., towns where Lyme disease and I. dammini are rare.
Serologic tests. Details on the use of IFA and ELISA methods and on quantitating total immunoglobulin to B. burgdorferi have been reported previously (20,24). To improve the sensitivity of the ELISA, the concentration of washed, whole-cell antigen used to coat the solid phase was increased from 7 Fig of protein per ml to 15 ,ug of protein per ml. In addition, the dilution of unconjugated rabbit anti-P. leucopus immunoglobulin was increased from 1:200 to 1:400, and the concentration of horseradish peroxidaselabeled goat anti-rabbit immunoglobulin G antibodies in phosphate buffer solutions was changed from 1:800 to 1:2,500. With these modifications, test sera were screened at dilutions -1:160; net absorbance values (i.e., net optical densities) greater than 0.39, 0.34, and 0.29 were graded as positive for the respective serum dilutions of 1:160, 1:320, and .1:640. These critical regions were determined by statistical analyses (3 standard deviations + mean) of net absorbance values for the respective serum dilutions of 60 normal specimens. Positive controls included serum samples from P. leucopus that had been found harboring B. burgdorferi. Procedures for blocking nonspecific binding sites and for the preparation of diluents, washes, and substrate remained unchanged. When new lots of enzyme-labeled antibodies or other reagents were purchased, tests were reevaluated with positive and negative serum controls and procedures were standardized accordingly. Uniformity of assay sensitivity was monitored daily by including the same positive and negative control sera on each plate. Controls were also included for antigens, conjugates, buffer solutions, and diluents.
To assess the specificity of the IFA method, 8 serum samples with antibodies to B. burgdorferi and 12 specimens that were nonreactive to this spirochete were screened against Borrelia hermsii (a relapsing fever spirochete), Treponema pallidum (the causative agent of syphilis in humans), and the following serovars of Leptospira interrogans: canicola, icterohaemorrhagiae, and pomona. The sources and preparation of these antigens have been reported previously (19).
Isolations. The blood, spleen tissues, or kidney tissues of white-footed mice were aseptically removed and processed for spirochete isolations as described previously (1,5,16). Briefly, 20 ,ul of blood or 1:10 dilutions of spleen and kidney tissues, triturated in 7 ml of Barbour-Stoenner-Kelly medium, were placed into duplicate tubes of this medium containing 0.1% agarose. Cultures were kept at 31°C for 3 to 6 weeks and examined for spirochetes by dark-field microscopy. Isolates were subcultured and identified as B. burgdorferi by fixing washed, whole cells to glass microscope slides and by applying murine monoclonal antibody (H5332) and appropriate conjugates in IFA tests (8,9). This monoclonal antibody was directed against outer surface protein A, a polypeptide with an approximate molecular mass of 31 kilodaltons. This protein is common to North American isolates of B. burgdorferi. Results verifying the presence of B. burgdorferi in rodents and arthropods from sites in Connecticut, Rhode Island, and New York have been published before (1,2,5,7,12,19).
Statistical analyses. To determine significant differences in percentages of positive sera, values were analyzed by a Student t test after arcsin transformation (27).

RESULTS
Comparative analyses of 210 P. leucopus sera revealed threefold greater sensitivity by modified ELISA (Table 1). In areas where Lyme disease is prevalent in human populations, the number of positive serum samples by ELISA (n = 80) greatly exceeded that determined by the IFA method (n = 26). Antibody titers were also correspondingly higher in the ELISA than in the IFA test. In analyses of an additional 24 serum samples, representing P. leucopus specimens from areas where I. dammini and Lyme disease are rare, results were negative in both assays. To assess reproducibility of the ELISA, 24 positive sera were retested. There were no differences in antibody titers for 10 samples, whereas twofold (n = 12) or fourfold (n = 2) variations were recorded for the remaining sera in the second trial.
Although antibodies to B. burgdorferi were present in P. leucopus during all months of the year in East Haddam, Conn. (Table 3), there were significantly more positive sera collected during June through August (22%) than during March through May (12%) (t = 2.1, P < 0.03). Antibody titers varied during all seasons, but titration endpoints peaked (1:2,560) during May, July, and August.  Multiple serum samples were obtained from 82 P. leucopus specimens in East Haddam during 1980, 1983, and 1984. Seroconversions (i.e., a negative-to-positive change in test results) were recorded for 6 mice from May through September (Table 4). Rising antibody titers were detected in 4 animals by ELISA, but IFA test results for these sera remained negative. The second or subsequent serum specimens from the remaining 76 mice were negative in either or both assays.
Infected white-footed mice had antibodies to spirochetes during all seasons. Of the 206 animals tested for B. burgdorferi, 72 (35%) harbored this bacterium in their blood, kidneys, or spleen ( Table 5). The majority of these (57.8% of 72) had detectable amounts of antibody to B. burgdorferi with titers ranging from 1:320 to 1:2,560 in ELISA. Although spirochetes were not isolated from the remaining 134 P. leucopus specimens, antibodies were detected in similar amounts in 10 mice.
Serum antibodies from 8 mice collected in Salem, an area where Lyme disease is highly endemic, cross-reacted in IFA tests against B. hermsii and T. pallidum. Similar tests with serovars of L. interrogans were negative. Antibody titers to B. hermsii differed twofold or less from homologous reactions to B. burgdorferi in all serum samples, whereas titration endpoints to T. pallidum were at least fourfold lower than those to the Lyme disease agent. All 12 serum specimens without antibodies to B. burgdorferi were likewise negative when tested against all other spirochetal antigens. DISCUSSION Antibodies to B. burgdorferi can be detected by either an IFA test or a modified ELISA, but the latter had greater sensitivity. The relatively low number of seropositive P. leucopus specimens reported earlier (20) was a conservative estimate of B. burgdorferi presence in endemic areas. Therefore, for surveillance purposes, ELISA is preferred because of greater sensitivity and ease of automation and standardization and because test results are more objective.
White-footed mice can harbor B. burgdorferi during all seasons. Anderson et al. (1) reported infection rates of .66% during the summer and c33% during the winter. In addition, the number of mice with antibodies to the Lyme disease agent varied seasonally and geographically. The highest titers of antibody and number of seroconversions occurred during the summer, a time when nymphs of I. dammini were active. Therefore, the warmer months appear to be most suitable for monitoring serologic responses to B. burgdorferi in these rodents.
Antibodies to B. burgdorferi were present in mice that had spirochete infections. Compared with those for other mammals, such as humans or dogs, antibody titers for these rodents were relatively low, regardless of the assay method used. Since some infected mice had no detectable antibody, immunologic response in P. leucopus may be delayed during initial and possibly late stages of infection. Even when immunoglobulins are produced, we suspect that they may be relatively ineffective in quickly neutralizing the spirochetes. Reservoir competence of P. leucopus has been confirmed (14,17), and although spirochetemia periods are unknown, these and possibly other vertebrate hosts (2, 3, 6) are  (22,26). Therefore, uninfected larval or nymphal ticks acquire B. burgdorferi by feeding on spirochetemic hosts and transmit the spirochetes when they refeed in later life stages. White-footed mice are abundant in woodlands, have relatively small home ranges, are easily caught, and often have numerous larvae and nymphs of Dermacentor variabilis and I. dammini attached (2,6,13,15,20,25). Antibodies produced against B. burgdorferi are now readily detected by ELISA, and methods are available for isolating this pathogen. Therefore, P. leucopus is ideal for studies designed to provide more precise information on the distribution and prevalence of B. burgdorferi infections in nature. Serologic analyses for the Lyme disease agent should be conducted along with isolation procedures. Furthermore, with growing interest in finding ways to decrease I. dammini populations and to reduce the risk of human infection, white-footed mice should be included in field evaluations of control efforts.
Tests on the specificity of our IFA method revealed cross-reactivity among the Borrelia and Treponema species. This has been demonstrated before (19,23) for antibodies in sera from humans and white-tailed deer (Odocoileus virginianus). However, since tick-borne relapsing fever is not known to occur in eastern United States and since T. pallidum does not infect P. leucopus, cross-reactivity with B. burgdorferi is probably insignificant. However, other unclassified spirochetes have been isolated from P. leucopus specimens and short-tailed shrews (Blarina brevicauda) in Connecticut and Minnesota (4). These organisms appear to be serologically distinct from species of Borrelia, Treponema, Leptospira, and Spirochaeta. Nonetheless, there might be other unknown spirochetes in nature that could stimulate immune responses in P. leucopus and lessen the specificity of these assays. Therefore, serologic evidence of B. burgdorferi infections should be supported by isolation and characterization of spirochetes in study areas.