Seroprevalence Of IgG Antibodies Against Borrelia burgdorferi In Dogs In Belgrade Area, Serbia

Abstract In this study the seroprevalence of anti-Borrelia burgdorferi IgG antibodies in populations of naturally exposed dogs in the Belgrade area was determined. Serum samples of 215 dogs were tested with in-house made indirect immunofluorescent-antibody assay (IFA), enzyme-linked immunosorbent assay (ELISA), and Western blot assay (WB). Sera were from 34 pet dogs suspected of having Lyme borreliosis, 41 hunting dogs, 125 shelter dogs, and 15 clinically healthy pet dogs. Results of WB were used as a standard against which the performances of ELISA and IFA were evaluated. ELISA was significantly more sensitive than IFA (84.8 vs 67.9%), whereas the specificities of these tests were just slightly different (94.4% vs 92.6 %). Borrelia burgdorferi seroprevalence of 24.7% among tested dogs was recorded by WB test. The highest seroprevalence of 31.2% was found in the group of shelter dogs, while hunting and pet dogs showed lower seroprevalence of 19.5% and 17.6%, but the difference was statistically insignificant. These results are the first report on the seroprevalence of IgG antibodies to B. burgdorferi in dogs in the Belgrade area and could be of importance in the assessment of the risk of infection with B. burgdorferi in human population.


INTRODUCTION
Lyme borreliosis (LB) is a multisystemic zoonotic disease caused by Borrelia burgdorferi sensu lato (B. burgdorferi s.l.) complex and transmitted by ticks of the genus Ixodes in Europe where small mammals and birds are important reservoirs [1]. Borrelia burgdorferi is a complex classifi ed into more than 15 genospecies, including B. burgdorferi sensu stricto (B. burgdorferi s.s.), B. garinii, B. afzelii, B. spielmanii and B. bavariensis proved as a pathogenic, and B. valaisiana, B. lusitaniae, and B. bissetii thought to be potentially pathogenic [2]. LB has been detected in humans and in animals, as well. Canine LB was fi rst reported in USA, in 1984 [3]. Clinical manifestations of LB in dogs are nonspecifi c, they indicate disorders of the musculoskeletal, cardiovascular and nervous systems [4,5], and have been observed in 5-10% of dogs exposed to B. burgdorferi [6]. The non-specifi c clinical markers and an absence of signifi cant changes in hematological and biochemical parameters make it diffi cult to diagnose the disease. Due to the lack of pathognomonic clinical signs in LB, serodiagnostic techniques such as indirect fl uorescent antibody assay (IFA), enzyme linked immunosorbent assay (ELISA) and Western blot assay (WB) have become widely used in diagnostics and epidemiological surveys. Direct detection of the pathogen by culture or polymerase chain reaction (PCR) has a minor role in the routine diagnosis of LB. As far as pets are concerned, dogs have been identifi ed as competent reservoirs of B. burgdorferi [7]. Dogs outdoor behavior carries a high risk of contact with infected ticks, and they are considered to be good indicators of LB endemic areas. Therefore, seroprevalence of anti-Borrelia antibodies in dogs could be of importance for the identifi cation of LB endemic areas [8].
In the Belgrade area, human LB was diagnosed in 1987 [9] and according to reports of our relevant institution, Institute of Public Health of Serbia "Dr Milan Jovanović Batut", LB in this area is one of the most common zoonoses in humans [10]. The main vector of B. burgdorferi s.l. in Belgrade area is Ixodes ricinus. A high prevalence of borrelia in ticks collected in this area was shown. It ranges from 18.75 to 53.7%, depending on the investigated locality, year and method of detection [11][12][13]. Five different Borrelia genospecies were detected in I. ricinus ticks from the Belgrade region: B. burgdorferi s.s., B. garinii, B. afzelii, B. valaisiana, and B. lusitaniae [12,13]. In addition, three strains belonging to pathogenic B. burgdorferi s.s. and B. afzelii genospecies were isolated from small rodents captured in the Belgrade area [14].
Data on LB in the human population in Belgrade were available, which was not the case with data for dogs infected with borrelia in the same region. Therefore, the aim of the study was to investigate the incidence of anti-Borrelia burgdorferi antibodies in dog sera in Belgrade with in-house made IFA, ELISA and WB tests. Also, a comparison of sensitivity and specifi city of the in-house tests was conducted.

Animals
Dogs from the Belgrade area (municipalities: Rakovica, Čukarica, Zvezdara, Surčin, Voždovac, Palilula, and Savski Venac) were included in this study. Serum samples of 215 dogs of different breeds, age and of both sexes were taken in the period 2008-2010. The dogs were of four groups: hunting dogs (n=41), shelter dogs (n=125), pet dogs with clinical signs attributed to infection with B. burgdorferi (shifting limb lameness, swollen joints, fever, and lymphadenopathy) and with described tick exposure (n=34), and clinically healthy pet dogs with no tick exposure (n=15). Dogs in Serbia are not routinely vaccinated against LB so none of the dogs was vaccinated against LB.
Blood was taken from the cephalic vein. Serum was separated after spontaneous coagulation of blood at room temperature. The aliquots were stored at -20 0 C.

Serology
IFA, ELISA, and WB in-house tests made in our laboratories were used for detection of canine serum IgG antibodies against B. burgdorferi.

Control sera
Positive control sera were of subcutaneously immunized dogs (n=3), negative for anti-B. burgdorferi antibodies in WB. They were immunized with 0.5 ml of the whole cell sonicated cultures mixed with 0.5 ml of complete Freund's adjuvant. The antigen was administered three times at intervals of 7 days. Sera collected at day 21 after the fi rst immunization were used as a positive control.
Sera from dogs with no tick bite history and negative for B. burgdorferi infection (WB) were used as negative controls.

IFA
Cultured borreliae were precipitated by 30 min centrifugation at 10,000 x g, and washed 3 times in PBS (phosphate buffered saline; 10 mM sodium-phosphate, 0.8% NaCl, pH 7.2-7.4) with 5mM MgCl 2 . Serial dilutions of the antigen (1:5 -1:40 in PBS) were applied on 8 well multispot glass microscopic slides (Sigma). The slides were air dried and fi xed in cold acetone. The positive control serum (diluted 1:100 in PBS) was incubated with every dilution of the antigen. As secondary antibody polyclonal rabbit anti dog IgG antibodies conjugated with FITC (INEP-Zemun, Serbia) (diluted 1:16) were used. The antigen dilution giving 100 bright fl uorescent borreliae per fi eld (LEITZ-Orthoplan fl uorescence microscope, 400-fold magnifi cation) was declared as optimal. The slides with optimally diluted antigen were prepared and stored at -70 0 C until use.
For testing purposes, 1:100 diluted tested sera, and positive and negative control sera were placed on the antigen-coated multi spot glass slides and incubated for 30 min at 37 0 C, in a wet chamber. After being washed in PBS and distilled water, and dried, the secondary antibody was added. After 30 min, slides were washed, dried, mounted in glycerol/PBS and examined under the fl uorescence microscope. A positive reaction was one in which bright green fl uorescence was observed.

ELISA
When preparing the antigen for ELISA the cultured borreliae were pelleted by centrifugation, washed twice with PBS with 5mM MgCl 2 , and twice with PBS without Mg 2+ . The fi nal pellet was resuspended in a 1:1 solution of PBS and 1M NaCl, and sonicated on ice with a cell sonicator (MSE 150 watt, MSE Scientifi c Instruments, Sissex, England) [15,16]. The extract was centrifuged, and the protein concentration in the supernatant was determined by the BCA test (Micro BCA Protein Assay Kit, Pierce Technology, USA). Aliquots of the supernatants were kept at -20°C.
For the assay, the antigen (1μg/ml in 50 mM carbonate buffer, pH 9.6) was added in 96 fl at well polystyrene microtiter plates (Nunc, France). Plates were coated overnight at 4 0 C and then washed three times in PBS with 0.05% Tween 20 (PBS/Tween). Nonspecifi c binding was blocked with 5% non-fat milk (NFM, Carl Roth, GmbH, Karisruhe) diluted in the same buffer. After 1h incubation at room temperature and washing, a highly diluted (1:400 in PBS/Tween) [17] positive and negative control sera and test sera were added. After 2h the plates were washed and incubated for 1h with polyclonal goat anti-dog IgG peroxidase-conjugate (Sigma) secondary antibody (diluted 1:10,000 in PBS/Tween). The peroxidase reaction was visualized with 0.42 mM TMB (3,3',5,5'-tetramethylbenzidine) (Sigma). The reaction was stopped after 20 min with 2M H 2 SO 4 . The optical density was read at 450 nm (OD 450 ) in a multiplate reader (Multiscan Plus, Labsystem, Finland).
Based on the OD 450 values of 40 sera of apparently healthy dogs, negative for anti-B. burgdorferi IgG in WB, a cut off value of 0.440 was obtained. The value was the mean value of OD 450 increased by the value of three standard deviations. The coeffi cient of variation of up to 11% was obtained by testing 10 negative sera in three independent ELISA tests. The samples with OD 450 values within a range of cut off ± 11% were tested again.

WB
To obtain the antigen for WB cultured borreliae were pelleted by centrifugation, resupended in 10 mM Tris/HCl, 150 mM NaCl buffer, pH 8.0 containing 1 mM CaCl 2 and 1mM MgCl 2, and washed two times by centrifugation. The pellet was resuspended in the Tris/NaCl buffer without Ca 2+ and Mg 2+ , centrifuged and resuspended in 20 mM Tris/HCl, 300 mM NaCl buffer pH 8.0, containing 2 mM EDTA. The protein concentration was measured by the BCA test. Aliquots of the antigen were stored at -70 0 C.
The separated proteins were electrotransferred onto nitrocellulose (Hybond ECL, GE HealthCare LifeScience USA) by Multiphor II system (LKB, Uppsala, Sweden) in 25 mM Tris, 192 mM glycine, 20% methanol, 0.1% SDS transfer buffer. Nonspecifi c binding was blocked with 2.5% nonfat dried milk in 50 mM Tris/HCl, 100 mM NaCl, pH 7.5 buffer with 0.05% Tween 20 (TBS/Tween) for 1 h, at room temperature. The membrane was cut into 3-mm strips and incubated with test and control canine sera (diluted 1:100 in TBS/Tween), for 2h, washed with TBS/Tween and incubated for 1h with horseradish peroxidase-conjugated goat anti-dog IgG antibodies (Sigma) diluted 1:2,000 in TBS/Tween. After being washed in TBS/Tween and TBS without Tween, the antibodies reactivity was visualized with 1 mM 4-chloro-1-naphthol (Sigma)-0.03% H 2 O 2 . The molecular weight of labeled protein was determined using the ImageMaster, TotalLab v1.1 software (GE HealthCare LifeScience, USA). The interpretation of the WB results was done based on the criteria of Hauser et al. (18) where the reactivity with at least one of six protein bands (p83/100, p58, p56, OspC, p21 and p17) in WB with B. burgdorferi s.s. indicated a positive reaction.

Statistical analysis
Specifi city and sensitivity of IFA and ELISA was determined using WB as the reference test.
The specifi city (Sp) was estimated as the relative frequency of negative results in noninfected dogs and it was calculated using the following formula: The sensitivity (Se) was estimated as the relative frequency of positive results in infected dogs and it was calculated using the following formula: The chi-square test was used to analyze the difference between seroprevalence in groups of dogs. A difference with p-values of <0.05 was considered signifi cant. Statistical analysis was carried out by Microsoft Excel.

RESULTS
The presence of anti-Borrelia IgG antibodies in sera of 215 dogs was analyzed by IFA, ELISA and WB (Table 1). With WB, anti-Borrelia IgG antibodies were found in 53 (24.7%) of 215 sera. With ELISA, 54 (25.1%) and with IFA 48 (22.3%) anti-Borrelia IgG positive canine sera were found.
With WB used as the reference test, we determined the performance of IFA and ELISA and identifi ed true and false positive or negative canine sera ( Table 2). The sensitivity of IFA was 67.9% and the specifi city was 92.6%. The sensitivity of ELISA was 84.9% which was higher than the sensitivity of IFA. The specifi city of ELISA was 94.4%. Out of 215 analyzed canine sera, 93.5% (201/215) had the same reactivity (either positive or negative) in both tests.
The results of the chi-square test showed that statistically signifi cant differences in seropositivity were found only when the control group of clinically healthy pet dogs with no tick exposure was compared to the other groups. There were no statistically signifi cant differences in prevalence of anti-Borrelia IgG antibodies between groups when the control group was omitted.

DISCUSSION
The results of this study showed that anti-B. burgdorferi IgG antibodies persist in the population of dogs in the Belgrade area. High seroprevalence of 25% in naturally exposed dogs was confi rmed by different tests (IFA, ELISA and WB) widely used for serology diagnostics of LB. The reported tick infection rate in Belgrade is high [11][12][13], and reported high seroprevalence of anti-B. burgdorferi antibodies in dogs was somewhat expected.
In this study we devised and standardized immunodiagnostic assays based on the methodology applied for diagnosis of LB in humans [16]. It is known that the selection of B. burgdorferi s.l. strains for immunodiagnostic tests depends on their frequency in the investigated area. In the Belgrade region different genospecies of B. burgdorferi s.l have been detected, data regarding ticks infection rate with these genospecies are not consistent and protein profi les of every local strain have not been described. Therefore, a reference strain, the European isolate of B. burgdorferi s.s., IRS, was used as the antigen. It is chosen because European researches showed that tests based on the USA originating B31 strain exhibited low diagnostic sensitivity to detect LB in humans and dogs in Europe [19,20].
In order to determine the specifi city and sensitivity of in-house IFA and ELISA tests created for the detection of anti-B. burgdorferi antibodies it is required to either compare them with commercially available tests or to use highly specifi c and sensitive WB as a "gold standard" [21,22]. In this study, WB was used as the reference standard test. Despite the fact that sensitivity of any WB is not 100%, it is widely used in the evaluation of the results of IFA and ELISA [21,22], and its application precludes a possibility that false negative sera would remain undetected. Also, analyzing the reactivity of serum antibodies of dogs suspected of having LB and asymptomatic dogs from endemic areas with specifi c proteins of B. burgdorferi in WB is of importance in both clinical practice (were serum antibodies reactivity to specifi c proteins of B. burgdorferi s.l. could help in determining if observed clinical symptoms are associated with LB), and in epizootic studies (where it could be determined if positive serologic fi ndings obtained by ELISA and IFA are the result of natural infection, vaccination, or presence of cross-reactive antibodies) [17].
Due to the fact that beside B. burgdorferi s.s., abundant presence of B. afzelii was confi rmed in the Belgrade area [12][13][14] we additionally performed WB with a B. afzelii strain (human skin isolate obtained from University of Ljubljana, Slovenia) where positive reaction presented a reactivity with at least two of nine protein bands (p83/100, p58, p43, p39, p30, OspC, p21, p18/p17 and p14) (criteria of Hauser et al.) [18]. A statistical analysis revealed that there was a high compliance of the results of these two WB assays (article in press). Based on these results we considered that it was possible to use WB prepared with B. burgdorferi s.s. for estimating the prevalence of antibodies against B. burgdorferi s.l. in dogs in our region.
Analyzing the epizootic situation in neighboring countries and throughout Europe, it could be noticed that the canine seroprevalence rates varied depending on the investigated regions, applied immunodiagnostic tests, interpretations of results in terms of positive reactivity, and Borrelia strains used in the tests [21,23]. Data on the seroprevalence of dogs in Europe is mainly based on the results of IFA and ELISA, while WB was rarely used for estimating the epizootic situation [24,25].
With anti-B. burgdorferi IFA we detected 22.3% positive canine sera. Good specifi city but low sensitivity of IFA observed in this study was in accordance with the results of Lindenmayer et al. [22]. Similar sensitivity and specifi city of IFA was shown when reactivity of human serum against B. burgdorferi was estimated [16]. Regardless of the low sensitivity, IFA is frequently used in epizootic researches [26][27][28]. Yet, our results point out that with IFA test more than 30% of infected dogs remained undetected. This indicated a necessity of application of ELISA as a more sensitive test. With in-house ELISA we found the canine seroprevalence rate in the Belgrade area to be 25.1%. In Vojvodina Province (northern part of Serbia), with a commercial ELISA, similar canine seroprevalence of 25.5% was detected [24]. The congruence of canine seroprevalence between Belgrade area and Vojvodina is also present if we compare results of WB. We detected by WB 24.7% positive dog sera. Almost identical seroprevalence results (26.1%) were obtained for dogs in Vojvodina, when a commercial WB was applied [24]. It is interesting that Potkonjak et al. [29] reported canine seroprevalence rate of 8.1% in Vojvodina, obtained by using indirect ELISA, and rapid immunochromatographic and immunoblot tests for the confi rmation of reactive blood serums.
Numerous studies reported canine seroprevalence against Borreliae throughout Europe. Using ELISA, Turk et al. [30] examined random healthy dogs in the Zagreb area [Croatia] and found 5% of positive sera. Similar seroprevalence of 6.2% was reported for healthy dogs from Bulgaria [28]. In Romania, seropositivity detected by ELISA and IFA was in the range of 0-46%, depending on the region, with an overall prevalence of anti-Borrelia antibodies of 6.52% [31]. In different regions of Czech Republic canine seroprevalence detected by ELISA was between 0 and 28% with an average of 6.5% [20], while Kybicová et al., [32] detected IgG antibodies to B. burgdorferi in 10.3% of dogs. Out of 448 randomly selected dogs in Munich, Germany, 4.9% showed the presence of antibodies agains B. burgdorferi after screening with the SNAP 4Dx assay (IDEXX, Westbrook, Maine, USA) [33]. In the Netherlands, Goossens et al. [34] found a relatively high (17%), seroprevalence of B. burgdorferi among dogs of using whole-cell ELISA. On the other hand, lower seroprevalence rates in apparently healthy dogs were detected in Portugal (0.2%), Spain (0.4%), and Latvia (2.5%) [35][36][37]. Low seroprevalence of B. burgdorferi found among dogs in these countries could be explained by their position at the northern and southern borders of I. ricinus areal, where activity of ticks is spatially and temporally restricted, and consecutively exposure of dogs to Borreliae vectors is lower. In Voronezh Reserve, Russia, unexpectedly low seroprevalence of B. burgdorferi in dogs of 2.4%, in the area which has never been treated with acaricides, was reported [38].
Results of a survey carried out in the Netherlands showed that there was no difference in seroprevalence to B. burgdorferi between hunting (19.5%) and pet dogs (17.6%) [40]. However, Stefancíková et al. [23] showed that in Slovakia (area of Kosice) the seroprevalence rate in hunting dogs (40%) was signifi cantly higher than the one found in the service dog population (11.8%). Although seroprevalence rate of B. burgdorferi in shelter dogs was higher in comparison to hunting and suspected pet dogs in our study, the detected differences were not statistically signifi cant.
A close connection between the distribution of ticks of the genus Ixodes and prevalence of LB in both humans and dogs was shown, and it is believed that data on seroprevalence of anti-Borrelia antibodies in dogs can be useful in identifying potentially dangerous areas for both dog and human populations [8]. Results of this study are the fi rst report concerning seroprevalence of anti-Borrelia antibodies in dogs in the Belgrade area. The relatively high seroprevalence found in this study supports the previous thesis based on high borrelia prevalence in host-seeking ticks in this area, Belgrade being an endemic region of LB, and again confi rmed that dogs may be used as sentinels for human exposure to LB. In addition to previous studies [39], this study also contributes to the elucidation of the still incompletely known epizootic situation in the Belgrade area.