Validation of an ELISA method for the serological diagnosis of canine brucellosis
due to Brucella canis
Maria Zoraida Daltro de Oliveiraa, Vera Valec, Lara Keidd, Songeli Menezes Freirec, Roberto Meyerc,
Ricardo Wagner Portelac, Stella Maria Barrouin-Meloa,b,*
aLaboratório de Infectologia Veterinária, Escola de Medicina Veterinária, Universidade Federal da Bahia, Av. Ademar de Barros 500, 40170-000 Salvador, Brazil
bDepartamento de Patologia e Clínicas, Escola de Medicina Veterinária, Universidade Federal da Bahia, Av. Ademar de Barros 500, 40170-000 Salvador, Brazil
cLaboratório de Imunologia e Biologia Molecular, Instituto de Ciências e Saúde, e e Universidade Federal da Bahia, Vale do Canela S/N, Salvador, Bahia, Brazil
dDepartamento de Medicina Veterinária Preventiva e Saúde Animal, Faculdade de Medicina Veterinária e Zootecnia da Universidade de São Paulo, SP, Brazil
a r t i c l ei n f o
Received 23 October 2008
Accepted 10 July 2010
a b s t r a c t
In the present study, the validation of an enzyme-linked immunosorbent assay (ELISA) for serodiagnosis
of canine brucellosis is described. Two different antigenic extracts, obtained by heat or ultrasonic homog-
enization of microbial antigens from a wild isolate of Brucella canis bacteria, were compared by ELISA and
Western blot (WB). A total of 145 canine sera were used to define sensitivity, specificity and accuracy of
the ELISA as follows: (1) sera from 34 animals with natural B. canis infection, confirmed by blood culture
and PCR, as well as 51 sera samples from healthy dogs with negative results by the agar–gel immunodif-
fusion (AGID) test for canine brucellosis, were used as the control panel for B. canis infection; and (2) to
scrutinize the possibility of cross reactions with other common dog infections in the same geographical
area in Brazil, 60 sera samples from dogs harboring known infections by Leptospira sp., Ehrlichia canis,
canine distemper virus (CDV), Neospora caninum, Babesia canis and Leishmania chagasi (10 in each group)
were included in the study. The ELISA using heat soluble bacterial extract (HE-antigen) as antigen showed
the best values of sensitivity (91.18%), specificity (100%) and accuracy (96.47%). In the WB analyses, the
HE-antigen showed no cross-reactivity with sera from dogs with different infections, while the B. canis
sonicate had various protein bands identified by those sera. The performance of the ELISA standardized
with the heat soluble B. canis antigen indicates that this assay can be used as a reliable and practical
method to confirm infection by this microorganism, as well as a tool for seroepidemiological studies.
? 2010 Elsevier Ltd. All rights reserved.
Brucellosis comprises a spectrum of infectious diseases that af-
fects livestock, companion animals and humans as well as is
widely distributed in many countries (World Health Organization,
2008). The infection is transmitted to people by direct or indirect
contact with infected animals or their products, and most Brucella
species are pathogenic for humans (Delpino et al., 2004).
In dogs, the severity of brucellosis caused by Brucella canis
ranges from clinical normality to the occurrence of miscarriages,
poor reproductive performance in males and females or non-
specific signs such as lymphadenopathy and diskospondylitis
(Carmichael and Joubert, 1987; Kerwin et al., 1992). Infected dogs
can transmit the infection to other dogs or people even after the
bacteremia has ceased and without presenting clinical symptoms
of the disease (Carmichael and Shin, 1996; Lucero et al., 2005).
As clinical diagnosis in dogs must be confirmed by laboratory tests,
many serological methods have been developed. Such tests include
the tube agglutination test (TAT), the rapid slide agglutination test
(RSAT) or agar–gel immunodiffusion (AGID); however, all of these
assays present variable levels of false positivity (Johnson and
Walker, 1992; Mateu-de-Antonio et al., 1993; Baldi et al., 1994;
Carmichael and Shin, 1996). False negativity in laboratory tests
has also been reported (Keid et al., 2009). Therefore, bacterial cul-
tures from blood or other biological specimens are still necessary
to confirm a questionable serological diagnosis (Carmichael and
Shin, 1996; Baldi et al., 1997; Wanke, 2004).
Various ELISA-based approaches for serodiagnosis of the infec-
tion have been proposed, each one showing variable results
depending on the properties of the antigen used in the assay
(Johnson and Walker, 1992; Mateu-de-Antonio et al., 1993; Baldi
et al., 1994, 1997; Letesson et al., 1997). Recently, other methods
have been proposed to detect the infection in dogs, including an
immunochromatographic assay (Kim et al., 2007) and a polymer-
ase chain reaction method (Keid et al., 2007). Furthermore, new
0034-5288/$ - see front matter ? 2010 Elsevier Ltd. All rights reserved.
* Corresponding author at: Laboratório de Infectologia Veterinária, Escola de
Medicina Veterinária, Universidade Federal da Bahia, Av. Ademar de Barros 500,
40170-000 Salvador, Brazil. Tel.: +55 71 3283 6754; fax: +55 71 3283 6728.
E-mail address: firstname.lastname@example.org (S.M. Barrouin-Melo).
Research in Veterinary Science 90 (2011) 425–431
Contents lists available at ScienceDirect
Research in Veterinary Science
journal homepage: www.elsevier.com/locate/rvsc
methodological improvements in agglutination (Watarai et al.,
2007) and complement fixation tests (Adone et al., 2008), employ-
ing more efficient antigens, represent reliable tools for both veter-
inarian practitioners and public health purposes.
Antigens used in the immunodiagnosis of Brucella infections
consist of various somatic proteins and surface components, but
antigenic similarities occur between species within the same gen-
era, such as B. canis, B. ovis, B. suis or B. abortus (Myers et al., 1972;
Baldi et al., 1994, 1997; Goldbaum et al., 1999; Ebani et al., 2003;
Nielsen et al., 2004). Common epitopes have been described
between Brucella species and other bacteria, such as alpha-
proteobacteria from soil and plants, including Agrobacterium, Sino-
rhizobium and Ochrobactrum genera (Delpino et al., 2004). False
positivity due to cross-reactivity between Brucella and other path-
ogenic bacteria in the serological diagnosis includes Salmonella
(Nielsen et al., 2007), Yersinia enterocolitica and E. coli (Mateu-
de-Antonio et al., 1993; Bounaadja et al., 2009).
Since clinical signs of canine brucellosis are not pathognomonic,
the serotest for routine diagnosis must be able to differentiate the
animal’s antibody response to B. canis infection from the humoral
response elicited by other canine infectious agents that occur in
a same geographical area. This property is closely inherent to the
quality of the antigen used in the serotest. In Brazil, leptospirosis,
ehrlichiosis, babesiosis or leishmaniosis are examples of current
canine infectious diseases, among others such as canine distemper
and neosporosis. The AGID test is used as the official test for canine
brucellosis in Brazil.
We have recently developed an indirect ELISA using an antigen
obtained by heating a wild isolate of B. canis, which is easy to pre-
pare, has low associated costs and demonstrates good sensitivity
and specificity (Barrouin-Melo et al., 2007). In the present study,
we have compared this antigen with a bacterial sonicate, which
is also easy to obtain and does not require special or expensive
equipment, to enrich the antigen preparation with cytosolic pro-
teins specific for the genus Brucella (Carmichael and Joubert,
1987) and augment the specificity of the ELISA test.
2. Materials and methods
2.1. Dog sera selection
A total of 145 sera from dogs were employed in this study.
Among the samples, sera from 34 dogs examined and sampled at
the Faculty of Veterinary Medicine of University of São Paulo, Bra-
zil, with canine brucellosis confirmed by PCR and bacterial culture,
were used as positive controls. All of these positive control dogs
presented clinical signs of disease and were seropositive for canine
brucellosis by the agar–gel immunodiffusion test (AGID). The neg-
ative controls consisted of 51 samples obtained from healthy dogs
and bitches of different breeds and ages (from 6 months to 8 years
old), from controlled kennels with no history of brucellosis or other
infectious diseases, with negative test results in the AGID assay. All
of the 51 negative control sera samples were used for cut-off calcu-
lation by ROC (Receiver Operating Characteristics) curve, while 20
of these samples were used for cut-off calculation by the mathe-
matical formula as described by Frey et al. (1998), selected from
dogs less than 1 year old and therefore sampled before entering
Sera samples from 60 dogs with defined diagnosis, either by
clinical or laboratory methods, of different canine infectious dis-
eases, were used for comparative ELISA and WB tests using both
B. canis-derived antigens. These sera, in groups of ten samples,
were positive for Leishmania chagasi, Ehrlichia canis, Babesia canis,
Leptospira species, Neospora canis and canine distemper virus
(CDV) infection, and were tested and kindly provided by the Labo-
ratories of Zoonosis and Parasitology of UFBA (Salvador, Brazil) and
the Laboratory of Virology at the Federal University of Minas Gerais
(UFMG), Belo Horizonte, Brazil. All these samples presented nega-
tive results in the AGID assay for B. canis-specific antibodies. All
sera were stored at ?20 ?C before being tested by ELISA and Wes-
2.2. Bacteria cultures and antigen preparation
A strain isolate of B. canis was kindly provided by the Desidério
Finamor Research Institute (Rio Grande do Sul, Brazil) and kept in
Brucella broth (BBL Microbiological Systems, Cockeysville, USA) un-
til antigen preparation. The bacteria were isolated from the pla-
centa and fetuses of a bitch (positive by AGID test for B. canis,
which aborted between 42nd and 45th days of gestation), classi-
fied by microscopic and biochemical methods as B. canis (White
and Wilson, 1951; Vargas et al., 1996; Alton, 1998) and compared
with the RM6/66 B. canis reference strain (ATCC 23365 – donated
by the Laboratoire de Brucellose – Agence Française de Securité
Sanitaire des Aliments, Maison Alfort, France).
For antigen preparation, a heat soluble bacterial extract (HE-
antigen) was obtained following the method described by Myers
et al. (1972), with minor modifications. Briefly, the culture was
transferred to 5 mL of Brucella broth (BBL Microbiological Systems,
Cockeysville, USA), cultured for 48 h at 37 ?C and expanded in Roux
flasks containing 100 mL of Brucella agar (Difco Laboratories, De-
troit, USA) in the presence of oxygen for 48 h at 37 ?C, as described
by Carmichael and Bruner (1968). Bacterial cells were harvested
with 50 mL of sterile PBS (phosphate buffered saline; 150 mM
NaCl, 2.5 mM KCl, 1.5 mM KH2PO4, 9 mM Na2HPO4, pH 7.4) and
inactivated by heat (1 h, 56 ?C). The suspension was filtered
through sterile gauze and washed three times by centrifugation
(3500g, 10 min) in PBS. The pellet was then re-suspended in
10 mL of PBS and autoclaved at 120 ?C under 1.5 atmospheres for
For the sonicate antigen preparation (US-antigen), the PBS-
filtered suspension of inactivated bacteria was subjected to ultra-
sonic cell disruption by three 60 s cycles of ultra-sound at 40 Hz
(Branson Sonifier?450), in an ice bath, with 1 min intervals be-
tween cycles. After heat or ultrasonic treatment, the cells were
centrifuged at 12,000g for 20 min, at 4 ?C. The supernatants were
identified as HE-antigen (obtained by heat) and US-antigen (ob-
tained by sonication) and stored in 200 lL aliquots at ?20 ?C, until
their use in ELISA and SDS–PAGE/Western blotting techniques. The
protein concentration of both antigens was determined by the
Lowry method (Lowry et al., 1951).
2.3. Indirect ELISA technique
The indirect ELISA was standardized and performed as already
described (Mateu-de-Antonio et al., 1993; Carpenter, 1997), with
modifications for optimization as previously described (Barrouin-
Melo et al., 2007). The working dilution of the peroxidase-
conjugated anti-dog immunoglobulin, the optimum protein con-
centration of both antigen preparations and the serum sample
dilutions were determined by previous checkboard titrations, to
achieve suitable differentiation between positive and negative sera
samples. Sera dilutions were tested at 1:250; 1:500; 1:1000;
1:2000 and 1:5000. The same B. canis-positive and negative control
sera samples were used as standards throughout all assays to en-
sure reliability and reproducibility for the results. All incubations
were performed in a humidified chamber. Flat-bottomed polysty-
rene 96-well microtiter plates (Corning Inc., New York, USA) were
sensitized overnight at 4 ?C with 0.75 lg/well of HE-antigen or
0.5 lg/well of US-antigen, diluted in 0.05 M sodium carbonate–
bicarbonate buffer, pH 9.6. After being washed five times with
426de Oliveira M.Z.D. et al./Research in Veterinary Science 90 (2011) 425–431
PBS, pH 7.4, containing 0.05% (v/v) of Tween-20 (PBS-T), each well
of the antigen-coated plates was blocked with 200 lL of 5%
skimmed milk in PBS-T and incubated at 37 ?C for 1 h. After five
more washes, 100 lL of sera samples, diluted at 1:1000 in PBS-T
containing 1% (v/v) of skim milk, was added to each well in dupli-
cate. Plates were sealed and incubated at 37 ?C for 1 h. After five
washing cycles with PBS-T, 100 lL of anti-dog IgG conjugated with
peroxidase (Sigma, St Louis, USA), diluted at 1: 5000 in PBS-T, was
added to each well and the plates incubated at 37 ?C for 1 h. After
five washings as described above, the color reaction was developed
by adding 50 lL/well of a solution containing 1.0 mg/mL of o-
phenylenediamine dihydrochloride (OPD; Sigma, St Louis, USA) in
0.05 M citrate buffer (pH 4.0) with 0.04% (v/v) H2O2. Plates were
incubated in the dark for 10–15 min at room temperature. The
enzymatic reaction was stopped by the addition of 25 lL/well of
4 N H2SO4. Absorbance measurements were made at 492 nm, using
an automatic ELISA plate reader (550, Bio-Rad, Life Sciences, Her-
2.4. SDS–PAGE and Western blotting analyses
The heat soluble bacterial extract (HE-antigen) and the soni-
cate antigen preparation (US-antigen) were electrophoresed as
follows: 50 lg of each antigen was solubilized in sample buffer
and subjected to sodium dodecyl sulfate polyacrylamide gel elec-
trophoresis (SDS–PAGE) using a 12% polyacrylamide gel and ana-
lyzed with Coomassie blue (2.5% brilliant blue in 50% methanol,
10% acetic acid) staining. For Western blotting, the electrophore-
sed antigen was transferred to nitrocellulose membranes, in a hu-
mid procedure, employing 100 V for 1 h. Unbound sites on the
membranes were blocked with 5% skim milk in PBS-T and incu-
bated overnight at 4 ?C. The membranes were cut into 0.5 cm
wide strips after being washed five times with PBS-T. Each one
of 10 positive sera for canine brucellosis, 60 positive sera for dif-
ferent canine infections and 10 negative control sera, diluted
1:100 in PBS-T with 1% skim milk, were added to individual
strips, which were then incubated at 37 ?C for 1 h. Subsequently,
the strips were washed five times with PBS-T and blots developed
with horseradish peroxidase labeled anti-dog IgG (Sigma, St Louis,
USA), diluted 1:500 in PBS-T for 1 h at 37 ?C. After five washing
steps, the strips were immersed in a substrate solution made of
a 1:5 dilution of a 0.3% 4-Cl–naphtol in methanol and Tris-saline
buffer (0.05 M Tris–HCl, 0.2 M NaCl, pH 7.2), with 0.04% (v/v)
H2O2, and the reaction was developed in the dark at room tem-
perature for 15 min. A final washing step was performed once
with distilled water.
2.5. Statistical analysis
In the present study, sera from blood-culture- and AGID-
positive dogs were used as a reference for ELISA validation. The
AGID serotest was used as a reference because it is employed as
the official assay for canine brucellosis in Brazil.
The cut-off value was determined by a mathematical formula
for a statistically valid value, which defines the upper prediction
limit based on the upper tail of the t-distribution of 20 negative
control OD readings, at a confidence level of 99.5%. For this calcu-
lation, were utilized 20 sera from dogs with ages below 1 year old.
The cut-off was calculated as follows: cut-off = (X) + SD * t[1+(1/n)]
(Frey et al., 1998; Barrouin-Melo et al. 2007).
Specificity and sensitivity of the assay were calculated as
TP þ FN? 100E ¼
TN þ FP? 100
TP – Number of truly positive samples
TN – Number of truly negative samples
FP – Number of false positive samples
FN – Number of false negative samples.
The accuracy was determined by the number of discordant re-
sults divided by the number of total samples tested.
The sensitivity and specificity of the ELISA made with US-anti-
gen and HE-antigen were also calculated employing a cut-off value
obtained by the ROC (Receiver Operating Characteristics) curve
that presented the best specificity and sensitivity together. This
curve was obtained using EPIINFO Software, with values obtained
from all 34 positive control sera and 51 negative control sera ob-
tained from dogs showing no clinical signs of the disease. Sixty sera
samples from dogs with diseases other than brucellosis were in-
cluded as negative for Brucella infection in the negative control
group for the ROC curve calculation.
Confidence intervals for sensitivity, specificity and test accuracy
were determined by Wilson’s Score (Open EPI 1.1, Epi Info). Nor-
mality of the results distribution was tested using the Pearson’s
chi-square test. Comparisons of mean absorbance values for ELISA
readings of the groups of sera positive for different diseases, sera
positive for brucellosis and sera from healthy dogs were performed
by analysis of variance followed by Anova and Tukey’s Test (SPSS v.
The results of 51 sera samples from healthy dogs, all seronega-
tive by the AGID test for canine brucellosis, and 34 sera samples
from dogs positive for canine brucellosis by blood culture, PCR
and the AGID test were used as reference for the determination
of sensitivity and specificity of the ELISA. Table 1 shows these re-
sults, assayed with both HE-ELISA and US-ELISA. The cut-off absor-
bance value for the HE-ELISA was 0.235 and for the US-ELISA was
0.323, as determined by the Frey method.
As shown in Table 2, the HE-ELISA of dog sera presented a sen-
sitivity of 91.18%, identifying 31 of 34 sera from dogs positive for
canine brucellosis while the US-ELISA presented a sensitivity of
100% (34/34). The HE-ELISA results confirmed the AGID results in
all 51 sera, showing a specificity of 100%, while the US-ELISA
Distribution of positive and negative reactions for canine brucellosis by HE-ELISA and
US-ELISA using samples previously tested by the AGID reference serotest.
Negative control sera
Positive control sera for
*Dogs positive by blood culture and/or PCR for canine brucellosis.
Sensitivity, specificity and accuracy of HE-ELISA and US-ELISA serotests using 34
positive*and 51 negative sera samples tested by the AGID reference test.
HE-ELISA (%)US-ELISA (%)
*Dogs positive for canine brucellosis by blood culture or PCR.
Maria Zoraida Daltro de Oliveira et al./Research in Veterinary Science 90 (2011) 425–431
demonstrated a specificity of 84.31% (co-negativity in 43 out of 51
sera). The accuracy index was 96.47% for the HE-ELISA and 90.59%
for the US-ELISA.
To investigate the specificity of the ELISA with each of the anti-
gens, sera from dogs with infectious diseases common in our geo-
graphic area underwent comparative HE-ELISA and US-ELISA
Mean absorbance values for the HE-ELISA demonstrate a clear
discrimination between sera from dogs culture-positive for canine
brucellosis and sera from dogs negative for brucellosis or positive
for other diseases (Table 3). The US-ELISA showed higher mean
absorbance values and standard deviations for several sera sam-
ples among each group of diseases, despite the fact that the mean
OD value for the negative samples from healthy dogs remained low
(Table 3). Analysis of the mean absorbance values by the Mann
Whitney test (unpaired t test) for the HE-ELISA data, as presented
in Fig. 1, shows that the reactivity of sera from dogs positive for ca-
nine brucellosis was significantly higher than sera from other
groups (P < 0.001). There were no significant differences between
the mean absorbance values of sera from dogs with other diseases
in the HE-ELISA (Fig. 1).
The same sera from dogs with serological and clinical diagnoses
of different infections, healthy animals and animals positive for ca-
nine brucellosis were tested by US-ELISA. Their comparative reac-
tivities are shown in Fig. 2. In this plot, there is a difference
between positive and negative sera readings (P < 0.05). Sera from
dogs with other infections showed lower mean absorbance values
than the Brucella-positive group. Nevertheless, many samples from
the group with other infections had high OD readings, suggesting
cross-reactivity between antibodies in those sera and the Brucella
antigens present in the US preparation.
Sera from dogs with other infections were tested by Western
blot to determine the proteins responsible for the cross-reactivity,
which occurred mostly with the Brucella antigen obtained by ultra-
sound. Western blotting with ultrasound-extracted antigen re-
vealed that many proteins with molecular weights above 98 kDa
were recognized by sera from dogs with different infections
(Fig. 3). In the same antigen preparation, two bands of approxi-
mately 62 and 85 kDa, despite being recognized predominantly
by positive sera for canine brucellosis, were also recognized by ser-
um from a dog positive for babesiosis.
The Western blots of sera with higher OD values in the HE-ELI-
SA revealed that only one sample from a dog with leptospirosis re-
acted weakly with a band of approximately 20 kDa (Fig. 4). Bands
of approximately 18, 39 and 78 kDa were specifically recognized
only by sera from animals positive for canine brucellosis in both
antigen preparations (Figs. 3 and 4).
The calculations of the cut-off points, sensitivities and specific-
ities of the US-ELISA and the HE-ELISA were then performed by
ROC analysis, including the OD readings of the sera from the 60
dogs with other infectious diseases in the group of brucellosis-neg-
ative samples. With the cut-off point for OD readings at 0.278,
assuming the grey zone of 10% with the confidence interval of
0.250–0.306, as determined by the ROC curve, the sensitivity was
kept at 91% and the specificity at 95.7%, for the HE-ELISA
(Fig. 5a). Results for the US-ELISA showed a sensitivity of 77.3%
and a specificity of 75.5%, considering a cut-off point for OD read-
ing at 0.443 (Fig. 5b).
Distribution of optical density (OD) readings in HE-ELISA and US-ELISA assays of sera samples positive and negative for canine brucellosis and sera samples positive for other viral,
bacterial and protozoan infections.
Canine seranHE-ELISA US-ELISA
Mean OD value SDAbove cut-off n = 0.235Mean OD valueSDAbove cut-off n = 0.323
Positive for brucellosis*
Positive for leptospirosis
Positive for ehrlichiosis
Positive for babesiosis
Positive for leishmaniosis
Positive for neosporosis
Positive for distemper
*Positive for canine brucellosis by the AGID test.
brucell leptosp ehrlich babes leishneosp distemp negat
Sera from dogs with different infections
Absorbance 492 nm
Fig. 1. ELISA reactivities of sera from dogs with brucellosis, leptospirosis, ehrlich-
iosis, babesiosis, distemper, leishmaniosis and negative controls against B. canis-
antigens extracted by heat (HE-ELISA).
brulep ehrbablei neodisneg
Sera from dogs with different infections
Absorbance 492 nm
Fig. 2. Results US-ELISA of sera samples from dogs with brucellosis, leptospirosis,
ehrlichiosis, babesiosis, distemper, leishmaniosis, and negative controls.
428de Oliveira M.Z.D. et al./Research in Veterinary Science 90 (2011) 425–431
There is a need for sensitive, specific and automated serological
tests for the identification of B. canis-specific antibodies, as the
infection of dogs by this microorganism has become more wide-
spread and affected humans, particularly in large cities where the
disease has been reported in commercial breeding kennels (Vargas
et al. 1996; Poester et al., 2002).
In terms of sensitivity, ELISAs have been proven to be superior
to agglutination-based techniques for the serodiagnosis of canine
brucellosis (Wanke et al., 2002). Previous reports have described
efforts to obtain and apply suitable antigens in serological diagnos-
tic tests for canine brucellosis based on the ELISA technique with
variable results in terms of sensitivity, specificity and reproducibil-
ity (Johnson and Walker, 1992; Mateu-De-Antonio et al., 1993;
Wanke et al., 2002; Wanke, 2004). The different techniques used
to extract antigens from B. canis may interfere with their protein
composition or alter the primary structure of the epitopes, affect-
ing their function. In fact, some authors have shown that cytosolic
antigens can provide more sensitive and specific serotests than
outer membrane antigenic preparations from B. canis (Carmichael
and Joubert, 1987), while others have argued that there are no rel-
evant differences in ELISAs using either type of antigen (Wanke et
Recently, we have reported the standardization of an ELISA
using an easily obtained antigen extracted by heat from B. canis,
which presents good sensitivity and specificity (Barrouin-Melo
et al., 2007). In an attempt to improve the diagnostic accuracy of
Fig. 3. Western blot analysis of B. canis-antigens extracted by ultrasound (US-
antigen) after electrophoresis in a 12.5% polyacrylamide gel in the presence of
sodium dodecyl sulfate and transfer to nitrocellulose sheets. Numbers in the left
side indicate the molecular weights of various bands recognized by sera from dogs
with brucellosis (lanes 3 and 4) that are also recognized by sera from dogs with
leptospirosis (lanes 6–9), babesiosis (lanes 10–14), ehrlichiosis (lane 15), distemper
(lane 18) and leishmaniosis (lanes 19 and 20), but not by negative controls (lanes 1
Fig. 4. Western blot analysis of B. canis-antigens extracted by heat (HE-antigen)
after electrophoresis in a 12.5% polyacrylamide gel in the presence of sodium
dodecyl sulfate and transfer to nitrocellulose sheets. Numbers in the left side
indicate the molecular weights of bands recognized only by sera from dogs with
brucellosis (lanes 3 and 4) and not by sera from dogs with leptospirosis (lanes 6–9),
babesiosis (lanes 10–14), ehrlichiosis (lane 15), distemper (lane 18) and leishman-
iosis (lanes 19 and 20) or by negative controls (lanes 1 and 2).
Fig. 5. Diagram of the ROC curves for determination of the HE-ELISA (a) and US-
ELISA (b) cut-off points, sensitivities and specificities.
Maria Zoraida Daltro de Oliveira et al./Research in Veterinary Science 90 (2011) 425–431
the ELISA by enriching a B. canis antigen preparation with cytoplas-
mic antigens, we developed an antigen extracted from the bacteria
by ultrasound. In the present study, we compared the heat- and
performance in ELISA. Furthermore, the cross-reactivity due to rec-
ognition of B. canis proteins sera from dogs with other known
infections was assessed by Western blot.
The results presented herein, using 145 dog sera samples, show
that both ELISAs were highly sensitive for canine brucellosis; how-
ever, the antigen extracted by heat conferred better specificity than
the antigen extracted by ultrasound. Accordingly, the performance
of the US-ELISA, standardized with the antigen extracted by ultra-
sound, was considered limited, as 25 positive results from this test
were deemed false. This difference between ELISAs was displayed
by the ROC curve (Fig. 5).
As a diagnostic serological method, the ELISA has important
advantages over other serological tests commonly used for the
diagnosis of canine brucellosis, such as providing readily measur-
able results and being easy to perform and standardize. Although
the AGID test has reduced overall sensitivity and specificity (Zwir-
ner, 1996), this method, which uses B. ovis as a source of antigen
due to its similarity to B. canis, is currently the official test for sero-
logical diagnosis of canine brucellosis in Brazil, where the produc-
tion of test antigens is restricted to official state-level laboratories
(Poester et al., 2002). This is the reason why the AGID test was se-
lected for comparison with the indirect ELISAs evaluated and vali-
dated in the present study.
Using sera from dogs positive and negative by AGID as the ref-
erence serotest, the ELISAs with either antigen, extracted by heat
or ultrasound, showed a sensitivity of 91.1% and 100%, respec-
tively, which can be considered satisfactory. Nevertheless, the
higher specificity (100%) for the ELISA with antigen extracted by
heat versus the specificity (84.3%) for the ELISA with antigen ex-
tracted by ultrasound demonstrates the superior diagnostic accu-
racy provided by the heat-extracted antigen. The use of heat to
obtain antigen from B. canis may have favorably influenced the
specificity by disrupting epitopes that bind to antibodies directed
against other infectious agents. Considering that the diagnosis of
canine brucellosis commonly condemns the animal to be with-
drawn from reproduction, neutered or even sacrificed, false posi-
sensitivity. In fact, although ultrasound-extracted antigen con-
ferred 100% sensitivity to the ELISA, it also had several protein
bands recognized by sera from dogs with clinical and laboratorial
diagnosis of other infections by Western blot. Given that diseases
caused by other bacteria, such as Leptospira sp. and Ehrlichia canis;
protozoa, such as Leishmania chagasi, Babesia canis and Neospora
sp.; or even a virus, such as the canine distemper virus (CDV),
are common in our geographic area, the specificity of the diagnos-
tic method is critical. In the present study, the assessment of cross-
reactivity between antibodies specific for antigens from other
infectious agents common in dogs and those of B. canis revealed
that high-molecular-weight proteins, above 90 kDa, were more
commonly involved in such reactions than low-molecular-weight
proteins. On the other hand, the low-molecular-weight protein
fractions of Brucella seem to predominate in the reactions between
Brucella-specific antibodies and the HE-antigen, as could be seen in
the WB tests. In fact, outer membrane proteins are thought to con-
fer good sensitivity in serological tests for diagnosis of B. canis
infection (Cloeckaert et al., 1990; Wanke et al., 2002; Lopez et al.,
2005). The ELISA using heat-extracted antigen distinguished ca-
nine brucellosis, and this method could be used to differentiate
sera of dogs with canine brucellosis from sera of dogs with lepto-
spirosis, ehrlichiosis, babesiosis, leishmaniosis, neosporosis, dis-
temper and non-infected healthy animals. Despite the high
specificity of the HE-ELISA, as shown by the differences between
to evaluatetheir diagnostic
the mean absorbance values for canine brucellosis sera and sera
of all other diseases tested, as well as sera from non-infected
healthy animals, two sera from dogs with babesiosis and one lep-
tospirosis-positive sample presented discrete OD values. These re-
sults could be explained by the high incidence of infectious
diseases among stray dogs, which could have concomitant infec-
tions. When those sera from dogs with other known diseases were
included in the determination of a new cut-off for brucellosis-neg-
ative sera by ROC curve, the specificity dropped from 100% to
95.7%, which is still a very good index for a serotest. Given that
the AGID test has poor sensitivity (Zwirner, 1996), some of the neg-
ative sera with higher absorbance values in the HE-ELISA could
come from B. canis-infected animals that gave a false negative re-
sult by AGID.
In conclusion, the heat-extracted antigen presented better ELI-
SA results than the ultrasound-extracted antigen. Moreover, the
production of heat-extracted antigen does not require expensive
equipment or highly trained technicians, making it feasible for
application in ELISAs for field studies or population surveys. The
results presented here demonstrate that the antigen extracted by
heat from B. canis is highly suitable for use in the diagnosis of ca-
nine brucellosis by ELISA, creating a reliable and secure serological
We thank Lígia Paraguassú Batista from the Laboratory of
Microbiology; Universidade Católica de Salvador for technical
assistance; Zélia Inês Portela Lobato from the Laboratory of Veter-
inary Virology, UFMG; and Luis Fernando Pita Gondim from the
Laboratory of Veterinary Clinical Pathology, UFBA; for sera sam-
ples. This work was supported by the Fundação para o Amparo a
Pesquisa do Estado da Bahia (FAPESB) and the Conselho Nacional
de Desenvolvimento Científico e Tecnológico (CNPq).
Adone, R., Francia, M., Ciuchini, F., 2008. Brucella melitensis B115-based complement
fixation test to detect antibodies induced by Brucella rough strains. Journal of
Applied Microbiology 105, 567–574.
Alton, G.G., Jones, L.M., Angus, R.D., Verger, J.M., 1998. Techniques for the brucellosis
laboratory. INRA, Paris, p. 190.
Baldi, P.C., Wanke, M.M., Loza, M.E., Fossati, C.A., 1994. Brucella abortus cytoplasmic
proteins used as antigens in an ELISA potentially useful for the diagnosis of
canine brucellosis. Veterinary Microbiology 41, 127–134.
Baldi, P.C., Wanke, M.M., Loza, M.E., Monachesi, N., Fossati, C.A., 1997. Diagnosis of
canine brucellosis by detection of IgG antibodies against an 18 kDa cytoplasmic
protein of Brucella spp. Veterinary Microbiology 51, 273–281.
Barrouin-Melo, S.M., Poester, F.P., Ribeiro, M.B., Alcântara, A.C., Aguiar, P.H.P.,
Nascimento, R.M., Freire, S.M., 2007. Diagnosis of canine brucellosis by ELISA
using an antigen obtained from wild Brucella canis. Research in Veterinary
Science 83, 340–346.
Bounaadja, L., Albert, D., Chénais, B., Hénault, S., Zygmunt, M.S., Poliak, S., Garin-
Bastuji, B., 2009. Real-time PCR for identification of Brucella spp.: a comparative
study of IS711, bcsp31 and per target genes. Veterinary Microbiology 137, 156–
Carmichael, L.E., Bruner, D.W., 1968. Characterization of a newly recognized species
of Brucella responsible for infectious canine abortions. Cornell Veterinarian 58,
Carmichael, L.E., Joubert, J.C., 1987. A rapid slide agglutination test for the
serodiagnosis of Brucella canis infection that employs a variant (M?)
organism as antigen. Cornell Veterinarian 77, 3–12.
Carmichael, L.E., Shin, S.J., 1996. Canine brucellosis: a diagnostician’s dilemma.
Seminars in Veterinary Medical and Surgery (Small Animals) 1, 161–165.
Carpenter, A.B., 1997. Enzyme-linked immunoassays. In: ROSE, N.R. (Ed.), Manual of
Clinical Laboratory Immunology, 5th ed. ASM Press, Washington, DC, pp. 20–29.
Cloeckaert, A., De-Wergifosse, P., Dubray, G., Limet, J.N., 1990. Identification of
seven surface-exposed Brucella outer membrane proteins by use of monoclonal
antibodies: immunogold labeling for electron microscopy and enzyme-linked
immunosorbent assay. Infection and Immunity 58, 3980–3987.
Delpino, M.V., Fossati, C.A., Baldi, P.C., 2004. Occurrence and potential diagnostic
applications of serological cross-reactivities between Brucella and other alpha-
proteobacteria. Clinical and Diagnostic Laboratory Immunology 11, 868–873.
Ebani, V.V., Cerri, D., Fratini, F., Bey, R.F., Andreani, E., 2003. Serological diagnosis of
brucellosis caused by Brucella canis. New Microbiology 26, 65–73.
430de Oliveira M.Z.D. et al./Research in Veterinary Science 90 (2011) 425–431
Frey, A., Canzio, J.D., Zurakowski, D., 1998. A statistically defined endpoint titer Download full-text
determination method for immunoassays. Journal of Immunology Methods
Goldbaum, F.A., Velikovsky, C.A., Baldi, P.C., MoRt, S., Bacher, A., Fossati, C.A., 1999.
The 18 kDa cytoplasmic protein of Brucella spp., an antigen useful for diagnosis,
is a lumazine synthase. Journal of Medical Microbiology 48, 833–839.
Johnson, C.A., Walker, R.D., 1992. Clinical signs and diagnosis of Brucella canis
infection. Compendium on Continuing Education 14, 763–772.
Keid, L.B., Soares, R.M., Vasconcellos, A.S., Chiebao, D.P., Megid, J., Salgado, V.R.,
Richtzenhain, L.J., 2007. A polymerase chain reaction for the detection of
Brucella canis in semen of naturally infected dogs. Theriogenology 67, 1203–
Keid, L.B., Soares, R.M., Vasconcellos, S.A., Megid, J., Salgado, V.R., Richtzenhain, L.J.,
2009. Comparison of agar–gel immunodiffusion test, rapid slide agglutination
test, microbiological culture and PCR for the diagnosis of canine brucellosis.
Research in Veterinary Science 86, 22–26.
Kerwin, S.C., Lewis, D.D., Hriberkik, T., Partington, B., Hosgood, G., Eilts, B.E., 1992.
Diskoespondylitis associated with B. canis infection in dogs: 14 cases. Journal of
the American Veterinary Medical Association 201, 1253–1257.
Kim, J.W., Lee, Y.J., Han, M.Y., Bae, D.H., Jung, S.C., Oh, J.S., Ha, G.W., Cho, B.K., 2007.
Evaluation of immunochromatographic assay for serodiagnosis of Brucella canis.
Journal of Veterinary Medical Science 69, 1103–1107.
Letesson, J.J., Tibor, A., Eneide, V., Wansard, V., Weynants, P.D., Saman, V., 1997.
Humoral immune responses of Brucella-infected cattle, sheep, and goats to eight
immunosorbent assay. Clinical and Diagnostic Laboratory Immunology 4,
Lopez, G., Ayala, S.M., Escobar, G.I., Lucero, N.E., 2005. Use of Brucella canis antigen
for detection of ovine serum antibodies against B. ovis. Veterinary Microbiology
Lowry, O.H., Rosembrogh, N.J., Farr, A.L., Randall, R.J., 1951. Protein measurement
with the folin phenol reagent. Journal of Biology and Chemistry 193, 265–275.
Lucero, N.E., Escobar, G.I., Ayala, S.M., Jacob, N., 2005. Diagnosis of human
brucellosis caused by Brucella canis. Journal of Medical Microbiology 54, 457–
Mateu-De-Antonio, E.M., Martin, M., Soler, M., 1993. Use of indirect enzyme-linked
immunosorbent assay with hot saline solution extracts of a variant (M) strain of
inan indirect enzyme-linked
Brucella canis for diagnosis of brucellosis in dogs. American Journal of
Veterinary Research 54, 1043–1046.
Myers, D.M., Jones, L.M., Varela-Diaz, V.M., 1972. Studies of antigens for
complement fixation and gel diffusion tests in the diagnosis of infections
caused by Brucella ovis and other Brucellae. Journal of Applied Microbiology 23,
Nielsen, K., Smith, P., Conde, S., Draghi de Benitez, G., Gall, D., Halbert, G., Kenny, K.,
Massengill, C., Muenks, Q., Rojas, X., Perez, B., Samartino, L., Silva, P., Tollersrud,
T., Jolley, M., 2004. Rough lipopolysaccharide of Brucella abortus RB51 as a
common antigen for serological detection of B. ovis, B. canis, and B. abortus RB51
exposure using indirect enzyme immunoassay and fluorescence polarization
assay. Journal of Immunoassay & Immunochemistry 25, 171–182.
Nielsen, K., Smith, P., Yu, W.L., Halbert, G., 2007. Salmonella enterica serotype urbana
Immunochemistry 28, 289–296.
Poester, F.P., Gonçalves, V.S., Lage, A.P., 2002. Brucellosis in Brazil. Veterinary
Microbiology 90, 55–62.
Vargas, A.C., Lazzari, A., Dutra, V., Poester, F.P., 1996. Brucelose canina: Relato de
caso. Ciência Veterinária 26, 305–308.
Wanke, M.M., 2004. Canine brucellosis. Animal Reproduction Science 82–83, 195–
Wanke, M.M., Delpino, M.V., Baldi, P.C., 2002. Comparative performance of tests
using cytosolic or outer membrane antigens of Brucella for the serodiagnosis of
canine brucellosis. Veterinary Microbiology 88, 367–375.
Watarai, M., Kim, S., Yamamoto, J., Miyahara, K., Kazama, M., Matsuoka, S., Chimura,
S., Suzuki, H., 2007. A rapid agglutination assay for canine brucellosis using
antigen coated beads. Journal of Veterinary Medical Science 69, 477–480.
White, P.G., Wilson, J.B., 1951. Differentiation of smooth and nonsmooth colonies of
brucellae. Journal of Bacteriology 61, 239–240.
World Health Organization, 2008. WHO Recommended Standards And Strategies
For Surveillance, Prevention And Control Of Communicable Diseases. Available
Zwirner, N.W., 1996. ELISA. In: Margni, R.A. (Ed.), Imunología e Imunoquímica, 5th
ed. Editorial Médica Panamericana, Buenos Aires, pp. 798–820.
Maria Zoraida Daltro de Oliveira et al./Research in Veterinary Science 90 (2011) 425–431