Diagnosis of canine brucellosis by ELISA using an antigen obtained from wild Brucella canis

ARTICLE IN PRESS

wwwelseviercomlocatervsc

Research in Veterinary Science xxx (2007) xxxndashxxx

Diagnosis of canine brucellosis by ELISA using an antigenobtained from wild Brucella canis

Stella Maria Barrouin-Melo a Fernando Padilla Poester b Marcos Borges Ribeiro cAdriano Costa de Alcantara c Paulo Henrique Palis Aguiar a Ivana Lucia Nascimento c

Robert Eduard Schaer c Roberto Meyer Nascimento c Songelı Menezes Freire c

a Departamento de Patologia e Clınicas Escola de Medicina Veterinaria Universidade Federal da Bahia Av Adhemar de Barros

500 Salvador BA CEP 40170-110 Brazilb Instituto de Pesquisas Veterinarias Desiderio Finamor (FEPAGRO) Estrada do Conde 6000 Eldorado do Sul RS CEP 92990-000 Brazil

c Laboratorio de Imunologia e Biologia Molecular Instituto de Ciencias da Saude Universidade Federal da Bahia

Vale do Canela SN Salvador BA CEP 40110-100 Brazil

Accepted 19 February 2007

Abstract

An indirect ELISA test was developed for the diagnosis of Brucella canis infection in dogs A bacterial whole cell extract was used as asolid phase antigen using B canis isolated from an infected animal Sera from culture-positive and healthy negative animals were used asinternal reference controls The cut-off point was determined by a mathematical formula for a statistically valid value which defined theupper prediction limit based on the upper tail of the t-distribution of 21 negative control sera readings for the confidence level of 995The sensitivity and specificity of the ELISA test were 95 and 91 respectively The ELISA test showed a significant concordance index(K = 084) with the agar gel immunodiffusion test The reliability of the ELISA for the detection of infected animals was established by adouble blind study testing 280 sera provided by serum banks from different diagnostic and research institutions and analyzed by ROCCurve 2007 Elsevier Ltd All rights reserved

Keywords Brucella canis ELISA Serodiagnosis Canine brucellosis

1 Introduction

Canine brucellosis is a zoonotic bacterial infectionwhose clinical diagnosis is difficult to perform (Berthelotand Garin-Bastuji 1993 Mateu-de-Antonio et al 1993Carmichael and Shin 1996) Human infections have been

0034-5288$ - see front matter 2007 Elsevier Ltd All rights reserved

doi101016jrvsc200702006

Abbreviations ELISA enzyme-linked immunosorbent assay AGID agar

gel immunodiffusion TAT tube agglutination test B canis Brucella canisB ovis Brucella ovis OD optical density

Corresponding author Tel +55 71 3263 6705 fax +55 71 3263 6718E-mail address barrouinufbabr (SM Barrouin-Melo)

Please cite this article in press as Barrouin-Melo SM et al DiagnoSci (2007) doi101016jrvsc200702006

reported being laboratory technicians and dog ownersthe most exposed to Brucella canis (Godoy et al 1977Lucero et al 2005) The disease has a worldwide distribu-tion and represents a great cause of economic losses inbreeding kennels (Carmichael 1966 Wanke 2004)Despite serological tests are routinely used the blood cul-ture is considered the only definitive test for the diagnosisof canine infection (Carmichael and Shin 1996 Baldiet al 1997 Wanke 2004) Clinical canine brucellosis iscommonly confirmed by different serological tests whichinclude (i) tube agglutination test (TAT) (ii) rapid plateagglutination test (RPAT) both with or without additionof 2-mercaptoethanol and (iii) agar gel immunodiffusion

sis of canine brucellosis by ELISA using an antigen Res Vet

2 SM Barrouin-Melo et al Research in Veterinary Science xxx (2007) xxxndashxxx

ARTICLE IN PRESS

test (AGID) using antigens prepared either with B canis

or Brucella ovis (Serikawa et al 1989 Baldi et al 19941997 Carmichael and Shin 1996 Lucero et al 2002)Cross-reactions are common occurring with bacteria likePseudomonas aeruginosa mucoid Staphylococcus sppand Bordetella bronchiseptica among others producingfalse positive results (Johnson and Walker 1992 Mateu-de-Antonio et al 1993 Baldi et al 1994 Carmichaeland Shin 1996)

Tests like the counterimmunoelectrophoresis (Myersand Varela-Diaz 1980) the indirect immunofluorescencethe complement fixation test (Schlemper and Vaz 1990Johnson and Walker 1992 Carmichael and Shin 1996)and the enzyme-linked immunosorbent assay (ELISA)(Berthelot and Garin-Bastuji 1993 Baldi et al 1994) haveshown to be effective for the diagnosis of B melittensis B

ovis and B abortus

Several authors have recommended the use of ELISAfor the detection of anti-B canis antibodies in dogs butvariable results have been reported Johnson and Walker(1992) using B canis cell wall antigens obtained equiva-lent specificity but lower sensitivity when compared withTAT Mateu-de-Antonio et al (1993) obtained 956 ofspecificity and 938 of sensitivity in an ELISA test usingas antigen an extract prepared from a less mucoid variantof B canis Baldi et al (1994 1997) using as antigen a puri-fied cytoplasmic protein (p18) and Letesson et al (1997)using recombinant cytoplasmic proteins (p15 p17 andp39) all common proteins in the genus Brucella reportedsensitive and specific ELISAs for canine and cattle brucel-losis respectively

In Brazil the acidified antigen test (rose bengal) the 2mercaptoethanol test and the complement fixation test(CFT) are currently used tests for the serodiagnosis ofsmooth brucellae (Brasil 2001) The AGID test is themost widely used in routine serodiagnosis of rough brucel-

lae Presently there is only one commercially availableAGID kit in the country which uses a heat saline solubleB ovis extract as antigen (Schlemper and Vaz 1990 Keidet al 2004) Agglutination tests using whole bacterialcells AGID and ELISA tests with diverse antigen prepa-rations from cell fractions to recombinant proteins havebeen the most quoted tests in reports worldwide for cur-rent serological diagnosis of the infection (Ebani et al2003 Lucero et al 2005 Wanke 2004 Wanke et al2006)

None of the referred serological methods are diagnosti-cally conclusive being the ELISA test still considered themost specific and sensitive Furthermore the use of an anti-gen which is easy to prepare and capable to provide a testwith good sensitivity and specificity would improve sur-veillance studies of canine brucellosis The test would allowas well the identification of potentially infected individualsThe present study describes the application of an antigenprepared from a B canis which was isolated from aninfected dog and used for the standardization of an indirectELISA test


2 Materials and methods

21 Bacterial strain and antigen preparation

A gram negative bacteria was isolated on blood agar(96 h incubation at 37 C with no CO2 requirement) fromthe placenta and fetuses of a bitch with positive AGID testfor B canis that aborted between 42nd and 45th days ofgestation (Vargas et al 1996) The small non hemolyticcolonies isolated presented a white-greyish color with arough morphology as demonstrated by the purple colorof the colonies after the plates being flooded with crystalviolet dye (White and Wilson 1951) No growth occurredon Mac Conkey agar The organism gave positive resultsfor catalase oxidase nitrate reduction urease and negativeresults for citrate Voges Proskauer and methyl red NoH2S production or glucose utilization were observed Thebacteria agglutinated with anti-R sera and with acriflavinwith no agglutination with anti-A or -M monospecific sera(Alton et al 1988) Based on these results the bacteriumwas classified as B canis

For the antigen preparation a heat soluble bacterialextract was obtained following the method described byMyers et al (1972) with minor modifications Briefly theculture was transferred to 5 mL of Brucella Broth (BBLMicrobiological Systems Cockeysville USA) culturedfor 48 h at 37 C and expanded in Roux flasks containing100 mL of Brucella Agar (Difco Laboratories DetroitUSA) under aerobiosis for 48 h at 37 C as described byCarmichael and Bruner (1968) Bacterial cells were har-vested with 50 mL of sterile PBS (phosphate buffered sal-ine 150 mM NaCl 25 mM KCl 15 mM KH2PO49 mM Na2HPO4 AElig 12H2O pH 74) and inactivated by heat(1 h 56 C) The suspension was filtered through sterilegauze and washed three times by centrifugation(3500 middot g 10 min) in PBS The pellet was then re-sus-pended in 10 mL of PBS and autoclaved at 120 C under15 Atmospheres for 20 min The cells were then centri-fuged at 12000 middot g for 20 min at 4 C The supernatantcollected and stored in 200 lL aliquots at 20 C was usedas the ELISA solid phase antigen The protein concentra-tion of the antigen was determined by the method of Lowryet al (1951)

22 Sera

Twenty positive sera from animals with clinical signspositive bacterial isolations and AGID-positive tests forbrucellosis test were used as reference Ten negative serumsamples from adult healthy animals obtained in a kennelwith no history or evidences of canine brucellosis (suchas reproductive disorders associated to fever lymphade-nopathy or other clinical signs) were tested negative bythe AGID and used as negative reference These referencesera were used for the calculation of specificity and sensitiv-ity of the ELISA test Twenty one samples from healthy6 months old animals kept in a research center were used


SM Barrouin-Melo et al Research in Veterinary Science xxx (2007) xxxndashxxx 3

ARTICLE IN PRESS

for calculation of the cut-off for the ELISA tests and asnegative control sera for western blotting assays Thesesera were kindly supplied by The Oswaldo Cruz Founda-tion (FIOCRUZ-Bahia) Two hundred and eighty (280)serum samples of unknown serology status for B canis

were obtained from serum banks of different research insti-tutions These 280 field sera were tested together with thereference sera in a double blind testing for determinationof the kappa values for concordance evaluation betweenthe positive results obtained by ELISA and AGID

23 Agar gel immunodiffusion test ndash AGID test

All sera were tested using a commercial kit (Tecpar-Bra-zil) according to the manufacturerrsquos instructions

24 Indirect ELISA protocol

The indirect ELISA was standardized and performed asdescribed elsewhere (Mateu-de-Antonio et al 1993 Car-penter 1997) with modifications Optimum antigen con-centration as well as peroxidase-conjugated anti-dogimmunoglobulin and sera dilutions were determined byserial titrations and incubations were performed in a humidchamber Briefly flat bottomed polystyrene 96-well micro-titer plates (Corning Inc New York USA) were sensi-tized overnight at 4 C with 50 lLwell of antigen dilutedto 5 lgmL in 005 M sodium carbonate pH 96 The plateswere washed five times with PBS containing 005 (vv)Tween-20 (PBS-T) Each well of the antigen-coated plateswere blocked with 200 lL of 5 skimmed milk in PBS-Tand incubated at 37 C for 1 h After five washes 100 lLof control and test sera samples diluted 11000 in PBS-Tcontaining 1 (vv) of skimmed milk were added to eachwell in duplicate The plates were sealed and incubated at37 C for 1 h After five washing cycles with PBS-T100 lL of anti-dog IgG conjugated with peroxidase(Sigma St Louis USA) at a dilution of 110000 in PBS-T were added to each well and the plates incubated at37 C for 1 h After five washings as described above thecolor reaction was developed by adding 50 lLwell of asolution containing 10 mgmL of o-phenylenediaminedihydrochloride (OPD Sigma St Louis USA) in 005 Mcitrate buffer (pH 40) with 004 (vv) H2O2 The plateswere incubated in the dark for 15 min at room tempera-ture Color development was halted by the addition of25 lLwell of 4N H2SO4 The absorbance measurementswere made at 492 nm using an automatic ELISA platereader (550 Bio-Rad Life Sciences Hercules USA) Sam-ples with an absorbance value equal to or higher than thecut-off value were considered as positives

25 SDSndashPAGE and Western blotting

The antigen was solubilized in sample buffer and sub-jected to sodium dodecyl sulfate polyacrylamide gel elec-trophoresis (SDSndashPAGE) using 12 polyacrylamide gels


and analysed with Coomassie blue (25 brilliant bluein 50 methanol 10 acetic acid) staining For Westernblotting electrophoresed antigen was transferred to nitro-cellulose membranes by using wet procedure Unboundsites on the membranes were blocked with 5 skimmedmilk in PBS-T at 4 C by incubation overnight Themembranes were cut into 05 cm wide strips after beingwashed five times with PBS-T Each of 20 positive and20 negative control sera diluted 1100 in PBS-T with 1of skimmed milk were added to individual strips whichwere incubated at 37 C for 1 h Following the strips werewashed five times with PBS-T and blots developed withperoxidase labeled anti-dog IgG (Sigma St LouisUSA) diluted at 1500 in PBS-T during 1 h at 37 CAfter five washing steps the strips were immersed in asubstrate solution made by a 15 dilution of a 03 4-Cl-a-naphtol in methanol and Tris-saline buffer (005 MTrisndashHCl 02 M NaCl pH 72) with 004 (vv) H2O2and the reaction developed in darkness at room temper-ature for 10ndash15 min A last washing step was performedonce with distilled water

26 Test performance determinations and statistical analysis

The cut-off value was calculated in according to a math-ematical method as described by Frey et al (1998) whichtakes in consideration the upper tail of the t-distributionof negative control sera readings as expressed in the fol-lowing equation

Cut-off frac14 X thorn SDf eth1THORNwhere X is the mean of independent sera readings and SD isthe standard deviation whose calculations were performedusing Prism software (GraphPad Software Inc) The fac-tor f is based on the number of negative controls and theconfidence level (1 a) The value of f used for calculationof the cut-off point in this study was of 2932 and corre-sponded to the desired confidence level of 995(a = 0005) and the sample size of 21 controls since thesera of 21 healthy young dogs were used for this purpose(Frey et al 1998)

Assay-to-assay variation was corrected for statisticalanalysis in accordance with Zwirner (1996) The correctedOD (COD) of each serum sample was calculated by multi-plying its mean OD value of readings by the correction fac-tor (CF) between plates as determined by the followingequation

COD frac14 OD CF eth2THORNwhere CF = nfracMean OD of the positive control serafrom a reference plateMean OD of the positive controlsera from plate of the assay

To calculate the ELISA specificity and the sensitivity 20positive and 10 negative ELISA reference sera were ana-lyzed in comparison with the same serarsquos reactivity by theAGID test The following Eqs (3) and (4) were appliedas described by Ferreira and Avila (1996)


Fig 2 Representative profile of the more consistently recognized bandsby positive sera of B canis antigen preparation on nitrocelluloseimmunoblot The numbers to the right indicate the molecular weight ofthe protein bands


ARTICLE IN PRESS

Specificity

frac14nfracfTrue negative seragfTrue negativethornFalse positive serageth3THORN

Sensitivity

frac14nfracfTrue positive seragfTrue positive serathornFalse negative serageth4THORN

The results obtained by the indirect ELISA were com-pared with those obtained by AGID test using known ref-erence and field samples of unknown serologic status Thereduced difference statistical test was used for comparisonof the proportion of positive results between tests (Sch-wartz 1987) The kappa index of concordance betweenmethods was determined by testing 20 positive and 10negative control sera and the 280 sera from field in dou-ble-blind tests (Armitage and Berry 1994) Mean andstandard deviation (SD) of corrected OD reading valuesof all sera were used for ROC Curve analysis (Grineret al 1981)

3 Results

31 B Canis antigen and standardization of the ELISA

Each Roux flask containing a B canis culture yieldedapproximately 30 mg of bacterial extract as determinedby Lowryrsquos method for protein quantitation This antigenpreparation showed a complex electrophoretic profile withvarious bands in the SDSndashPAGE Coomassie blue-stainedgel (Fig 1) The Western immunoblotting analysis of theantigen using positive and negative reference sera showeda pattern of band recognition more consistently for poly-peptides of 12 18 58 and 65 KDa as represented in Fig 2

Fig 1 Profile of B canis antigen preparation in three different dilutions(lanes b c and d) on a Coomassie blue-stained SDSndashPAGE Lane (a)contains the benchmark pre-stained protein ladder for molecular weight(Invitrogen) as indicated by numbers on the left


In order to determine the optimum concentration ofantigen to be adsorbed to microtiter plates previous assayswere performed using from 1 to 20 lgmL in 005 Msodium carbonate pH 96 Best results were obtained with5 lgmL of bacterial extract in 50 lL (250 ng) of solutionin each well By serial titrations of positive control seraand peroxidase-conjugated anti-dog IgG antibody theoptimal dilutions for the indirect ELISA were establishedas being respectively 11000 and 110000 The cut-off ODvalue as determined by sera samples from 21 healthy6 months old dogs was 0218

32 Analysis of field sera by the indirect ELISA

A double-blind testing performed with the indirectELISA test included 280 field sera from dogs withunknown history of brucellosis the 30 reference positiveand negative sera and the 21 serum samples from younghealthy dogs totalizing 331 samples In this study it wasfound that 72 sera samples (26) presented OD valuesabove the cut-off and were positive to ELISA The ODvalues for the positive samples ranged from 0218 to1115 (mean = 0466 SD = 0238) and for the negativesamples ranged from 0046 to 0217 (mean = 0108SD = 0042) The ratio between the OD value from thehighest positive serum and the mean OD of the negativecontrol sera was 1032 (11150108) and was calculatedaccording to Mateu-de-Antonio et al (1993) and Baldiet al (1994) The distribution of the OD frequencies amongthe field sera are shown in Fig 3

The sensitivity and the specificity of the indirect ELISAdetermined by the comparative testing of positive and neg-ative by the AGID was of 95 and 91 respectively


0

20

40

60

80

100

120

140

160

Mean OD 492 nm

0

20

40

60

80

100

120

140

160

Fre

quen

cy Cut-off 0218

lt005 005 01 015 02 025 03 04 05 06 07 08 09 10 gt10

Fig 3 Frequency distribution of the mean OD values of positive (n) and negative (h) canine field sera by indirect ELISA for detection of antibodyagainst Brucella canis All sera that had an OD above 0218 (cut-off) were considered to be positive


ARTICLE IN PRESS

Assay-to-assay variation as determined by each plate cor-rection factor for repeated assays using the same 30 controlsera was 17

The ability to distinguish positive and negative sera to B

canis antigens by detection of antibody reactivity by AGIDand by indirect ELISA was compared These results wereused to calculate the sensitivity the specificity and theKappa index of concordance between tests The correlationbetween the ELISA and the AGID was highly significantas expressed by the Kappa index of 084 The reduced dif-ference test ( = 164 p lt 005) also showed a significanttest agreement

The overall diagnostic accuracy of the ELISA test wasalso evaluated by the ROC curve (receiver operating char-acteristics-SPSS) analysis applied to the 280 field serausing as standard references the absorbance values of thepositive and negative control sera For the selected cut-offvalue of 0218 the ELISA test had both specificity andsensitivity of 100

4 Discussion

In this study we describe the development and evalua-tion of an indirect ELISA for the serodiagnosis of caninebrucellosis A bacterial whole cell extract was used assolid-phase antigen and sera from culture positive andhealthy negative animals used as reference controls Inthe double-blind analysis of the reactivity of canine fieldsera to B canis antigens the ELISA allowed a clear sepa-ration between positive and negative samples In additionto the relatively good degree of agreement between ELISAand AGID the results indicate that the ELISA is a specificand sensitive test for detection of anti-Brucella antibodiesin canine sera

The procedure used in the present work for obtainingthe antigen preparation was based on the technique


described by Myers et al (1972) with some modificationsThis technique may have accounted for the maintenanceof some intact outer membrane proteins (OMPs) as wellas cytosolic proteins which are highly specific for thegenus Brucella (Cloeckaert et al 1990 Kittelbergeret al 1998 Wanke et al 2002) Among four proteinbands identified more consistently by B canis positivesera by Western blotting in the present study two low-molecular-weight bands of 12 and 18 kDa could be ofparticular interest for further studies A cytoplasmic18 kDa molecular weight Brucella protein has been identi-fied previously as lumazine synthase and described asbeing highly antigenic in serological tests for canine bru-cellosis (Baldi et al 1994 1997 Goldbaun et al 1999)On the other hand low molecular mass protein fractionsof Brucella outer membrane proteins are responsible forconferring good sensitivity in serological tests for diagno-sis of recent infection (Cloeckaert et al 1990 Wankeet al 2002 Lopez et al 2005) These features of the anti-gen preparation may explain the good sensitivity achievedby the ELISA test in our study since the technique usedto extract antigen from bacteria may influence the pri-mary structure of the molecules and have an importantrole in its function

Among various criteria reported for selecting negativeserum groups used for determining the cut-off value theserum collected from healthy dogs before their pubertyunder 6 months of age was chosen This criterion wasbased on the literature which shows a higher probabilityof brucellosis be present in animals at reproductive agesthan in ages before reproductive activity (Moore andGupta 1970 Carmichael and Joubert 1988) The analysisof positive and negative sera provided by research institu-tions as well as the use of field samples made possiblethe comparative evaluation between the AGID test andthe indirect ELISA



ARTICLE IN PRESS

The ELISA is a test that has the advantage of reducinginterference of possible human errors and provides objec-tive and readily measurable results due to the use of cut-off value and internal controls (Baldi et al 1994 Troyet al 1996) Despite the reduced overall sensitivity andspecificity reported for the AGID test (Zwirner 1996)the method was selected for comparison with the indirectELISA test developed in the present study because it isthe official test for serological diagnosis of canine brucello-sis in Brazil

In terms of general sensitivity ELISA techniques aresuperior to other tests namely agglutination and AGIDcommonly used for the serological diagnosis of Brucella

infection (Johnson and Walker 1992 Wanke et al 2002Wanke 2004) High specificity and the capacity of detect-ing low levels of antibodies can be achieved with good anti-gen preparations as shown by the results of the presentstudy in which the standardization of the indirect ELISAtest was performed with sera dilutions of 11000 The agree-ment between AGID and ELISA was significant as deter-mined by the Kappa index of concordance of 084 betweentests The ELISA test specificity was of 91 and its sensi-tivity was of 95 As the AGID results were taken as com-parative standards it is possible that these indexes ofsensitivity and specificity may be underestimated In factthe ROC curve analysis of the test as it was performedshowed 100 indexes for sensitivity and specificity Studiesregarding the reproducibility of the indirect ELISA are cur-rently being carried out

Despite the incidence of human infection by B canis isnot actually known (Wanke 2004) the potential exposureto the bacteria represents a risk particularly for laboratoryor kennel workers as well as dog owners as reported else-where (Godoy et al 1977 Carmichael and Shin 1996Lucero et al 2005) It rises the needing for efficient andinexpensive tests which are appropriate for use in diseasecontrol programs of public health particularly in develop-ing countries The antigen preparation and its applicationin an indirect ELISA as described in the present studywere found to be feasible simple to prepare and of lowcost The bacteria used for antigen preparation wereobtained from a B canis culture isolated from a localinfected animal In spite of efficient ELISA tests beingdescribed elsewhere (Mateu-de-Antonio et al 1993 Baldiet al 1994 1997 Wanke et al 2002) it must be empha-sized that laboratory procedures for antigen purificationthat can improve the quality of immunological tests mayrequire expensive reagents and technology This aspectcan be restrictive for a number of developing countries

Taking into account the advantages of the indirectELISA described herein it can be concluded that the testmight be sufficiently accurate to be recommended as a sero-logical test for either field sera screening or confirmationof clinical canine brucellosis The assay for detecting anti-bodies against B canis antigens may complement or offeran alternative to the available tools for diagnose thedisease


Acknowledgements

We gratefully acknowledge Prof Thereza Martinez forassistance with bacterial techniques and Prof Thereza Cal-mon for statistical counseling We are also indebted to DrMoacir Paranhos (CPqGM-FIOCRUZ-BA) Dr CarlosChavez Olortegui (FUNED-MG) the Laboratorio deDoencas Tropicais (HUPES-UFBA) and the Centro deControle de ZoonosesndashPrefeitura de Sao Paulo for provid-ing the canine serum samples This work was financiallysupported by the Brazilian National Council for Researchand Technological Development (CNPq)

References

Alton JJ Jones LM Angus RD Verger JM 1988 Techniques forthe Brucellosis Laboratory INRA Paris

Armitage P Berry G 1994 Statistical Methods in Medical Researchthird ed Blackwell Source New York p 923

Baldi PC Wanke MM Loza ME Fossati CA 1994 Brucellaabortus cytoplasmic proteins used as antigens in an ELISA potentiallyuseful for the diagnosis of canine brucellosis Vet Microbiol 41 127ndash134

Baldi PC Wanke MM Loza ME Monachesi N Fossati CA1997 Diagnosis of canine brucellosis by detection of serum antibodiesagainst an 18 kDa cytoplasmic protein of Brucella spp Vet Microbiol51 273ndash281

Berthelot X Garin-Bastuji B 1993 Brucelloses canines Le PointVeterinaire 25 125ndash129

Brasil 2001 Departamento de Defesa Animal Informacoes sobre oPNCBET Ministerio da Agricultura Pecuaria e AbastecimentoBrasılia Brasil lthttpwwwagriculturagovbrsdaddaprogramahtmgt

Carmichael LE 1966 Abortion in 200 Beagles JAVMA 149 1126ndash1966

Carmichael LE Bruner DW 1968 Characterization of a newlyrecognized species of Brucella responsible for infectious canineabortions Cornell Vet 58 579ndash592

Carmichael LE Joubert JC 1988 Transmission of Brucella canis bycontact exposure Cornell Vet 78 63ndash73

Carmichael LE Shin SJ 1996 Canine brucellosis a diagnosticianrsquosdilemma Semin Vet Med Surg (Small Animal) 11 161ndash165

Carpenter AB 1997 Enzyme-linked immunoassays In ROSE NR(Ed) Manual of Clinical Laboratory Immunology fifth ed ASMPress Washington DC pp 20ndash29

Cloeckaert A de-Wergifosse P Dubray G Limet JN 1990Identification of seven surface-exposed Brucella outer membraneproteins by use of monoclonal antibodies immunogold labeling forelectron microscopy and enzyme-linked immunosorbent assay InfectImmun 58 3980ndash3987

Ebani VV Cerri D Fratini F Bey RF Andreani E 2003Serological diagnosis of brucellosis caused by Brucella canis NewMicrobiol 26 65ndash73

Ferreira W Avila SLM 1996 Diagnostico laboratorial das principaisdoencas infecciosas e auto-imunes Guanabara Koogan Rio deJaneiro p 302

Frey A Canzio JD Zurakowski D 1998 A statistically definedendpoint titer determination method for immunoassays J ImmunolMethod 221 35ndash41

Godoy AM Peres JN Barg L 1977 Isolamento de Brucella canis emMinas Gerais Brasil Arq Esc Vet Univ Fed Minas Gerais 29 35ndash42

Goldbaun FA Velikovsky CA Baldi PC Mortl S Bacher AFossati CA 1999 The 18-kDa cytoplasmic protein of Brucella



ARTICLE IN PRESS

species ndash an antigen useful for diagnosis ndash is a lumazine synthase JMed Microbiol 48 833ndash839

Griner PF Mayewski RJ Mushlin AI Greenland P 1981Selection and interpretation of diagnostic tests and procedures AnnIntern Med 94 555ndash600

Johnson CA Walker RD 1992 Clinical signs and diagnosis ofBrucella canis infection Comp Cont Educ 14 763ndash772

Keid LB Soares RM Morais ZM Richtzenhain LJ VasconcellosSA 2004 Brucella spp isolation from dogs from commercialbreeding kennels in Sao Paulo State Brazil Braz J Microbiol 35161ndash166

Kittelberger R Diack DS Vizcaıno N Zygmunt MS CloeckaertA 1998 Characterization of an immuno-dominant antigen in Brucella

ovis and evaluation of its use in an enzyme-linked immunosorbentassay Vet Microbiol 59 213ndash227

Letesson JJ Tibor A Eynde G Wansard V Weynants V DenoelP Saman E 1997 Humoral immune responses of Brucella-infectedcattle sheep and goats to eight purified recombinant Brucella proteinsin an indirect enzyme-linked immunosorbent assay Clin Diag LabImmunol 4 556ndash564

Lopez G Ayala SM Escobar GI Lucero NE 2005 Use ofBrucella canis antigen for detection of ovine serum antibodies againstBrucella ovis Vet Microbiol 105 181ndash187

Lowry OH Rosembrogh NJ Farr AL Randall RJ 1951 Proteinmeasurement with the folin phenol reagent J Biol Chem 193 265ndash275

Lucero NE Escobar GI Ayala SM Lopez G 2002 Sensitivity andspecificity of an indirect enzyme-linked immunoassay for the diagnosisof Brucella canis infection in dogs J Med Microbiol 51 656ndash660

Lucero NE Escobar GI Ayala SM Jacob N 2005 Diagnosis ofhuman brucellosis caused by Brucella canis J Med Microbiol 54457ndash461

Mateu-de-Antonio EM Martin M Soler M 1993 Use of indirectenzyme-linked assay with hot saline solution extracts of a variant (M-)strain of Brucella canis for diagnosis of brucellosis in dogs Am J VetRes 54 1043ndash1046


Moore JA Gupta BN 1970 Epizootiology diagnosis and control ofBrucella canis JAVMA 156 1737ndash1740

Myers DM Jones LM Varela-Diaz VM 1972 Studies of antigensfor complement fixation and gel diffusion tests in the diagnosis ofinfections caused by Brucella ovis and other Brucellae Appl Micro-biol 23 894ndash902

Myers DM Varela-Diaz VM 1980 Serological and bacteriologicaldetection of Brucella canis infection of stray dogs in MorenoArgentina Cornell Vet 70 258ndash265

Schwartz D 1987 Methodes statistiques a lrsquousage des medecins et desbiologistes 10 ed Medecine-Sciences Flammarion Paris pp 218ndash223

Schlemper SRM Vaz AK 1990 Inquerito sorologico para brucelosecanina por Brucella canis na regiao do planalto catarinense ndash BrasilRev Bras Med Vet 12 8ndash12

Serikawa T Iwaki S Mori M 1989 Purification of Brucella canis cellwall antigen using immunosorbent assay for specific diagnosis ofcanine brucellosis J Clin Microbiol 27 837ndash842

Troy GC Becker MJ Greene RT 1996 Proficiency testing ofselected antigen and antibody tests for use in dogs and cats JAVMA209 914ndash917

Vargas AC Lazzari A Dutra V Poester FP 1996 Brucelosecanina Relato de caso Ciencia Veterinaria 26 305ndash308

Wanke MM 2004 Canine brucellosis Anim Reprod Sci 82ndash83 195ndash207

Wanke MM Delpino MV Baldi PC 2002 Comparative perfor-mance of tests using cytosolic or outer membrane antigens of Brucellafor the serodiagnosis of canine brucellosis Vet Microbiol 88 367ndash375

Wanke MM Delpino MV Baldi PC 2006 Use of enrofloxacina inthe treatment of canine brucellosis in a dog kennel (clinical trial)Theriogenology 66 1573ndash1578

White PG Wilson JB 1951 Differentiation of smooth and non-smooth colonies of brucellae J Bacteriol 61 239ndash240

Zwirner NW 1996 ELISA In Margni RA Imunologıa e Imunoquı-mica fifth ed Editorial Medica Panamericana Buenos Aires pp 798ndash820



ARTICLE IN PRESS

test (AGID) using antigens prepared either with B canis

or Brucella ovis (Serikawa et al 1989 Baldi et al 19941997 Carmichael and Shin 1996 Lucero et al 2002)Cross-reactions are common occurring with bacteria likePseudomonas aeruginosa mucoid Staphylococcus sppand Bordetella bronchiseptica among others producingfalse positive results (Johnson and Walker 1992 Mateu-de-Antonio et al 1993 Baldi et al 1994 Carmichaeland Shin 1996)

Tests like the counterimmunoelectrophoresis (Myersand Varela-Diaz 1980) the indirect immunofluorescencethe complement fixation test (Schlemper and Vaz 1990Johnson and Walker 1992 Carmichael and Shin 1996)and the enzyme-linked immunosorbent assay (ELISA)(Berthelot and Garin-Bastuji 1993 Baldi et al 1994) haveshown to be effective for the diagnosis of B melittensis B

ovis and B abortus

Several authors have recommended the use of ELISAfor the detection of anti-B canis antibodies in dogs butvariable results have been reported Johnson and Walker(1992) using B canis cell wall antigens obtained equiva-lent specificity but lower sensitivity when compared withTAT Mateu-de-Antonio et al (1993) obtained 956 ofspecificity and 938 of sensitivity in an ELISA test usingas antigen an extract prepared from a less mucoid variantof B canis Baldi et al (1994 1997) using as antigen a puri-fied cytoplasmic protein (p18) and Letesson et al (1997)using recombinant cytoplasmic proteins (p15 p17 andp39) all common proteins in the genus Brucella reportedsensitive and specific ELISAs for canine and cattle brucel-losis respectively

In Brazil the acidified antigen test (rose bengal) the 2mercaptoethanol test and the complement fixation test(CFT) are currently used tests for the serodiagnosis ofsmooth brucellae (Brasil 2001) The AGID test is themost widely used in routine serodiagnosis of rough brucel-

lae Presently there is only one commercially availableAGID kit in the country which uses a heat saline solubleB ovis extract as antigen (Schlemper and Vaz 1990 Keidet al 2004) Agglutination tests using whole bacterialcells AGID and ELISA tests with diverse antigen prepa-rations from cell fractions to recombinant proteins havebeen the most quoted tests in reports worldwide for cur-rent serological diagnosis of the infection (Ebani et al2003 Lucero et al 2005 Wanke 2004 Wanke et al2006)

None of the referred serological methods are diagnosti-cally conclusive being the ELISA test still considered themost specific and sensitive Furthermore the use of an anti-gen which is easy to prepare and capable to provide a testwith good sensitivity and specificity would improve sur-veillance studies of canine brucellosis The test would allowas well the identification of potentially infected individualsThe present study describes the application of an antigenprepared from a B canis which was isolated from aninfected dog and used for the standardization of an indirectELISA test


2 Materials and methods

21 Bacterial strain and antigen preparation

A gram negative bacteria was isolated on blood agar(96 h incubation at 37 C with no CO2 requirement) fromthe placenta and fetuses of a bitch with positive AGID testfor B canis that aborted between 42nd and 45th days ofgestation (Vargas et al 1996) The small non hemolyticcolonies isolated presented a white-greyish color with arough morphology as demonstrated by the purple colorof the colonies after the plates being flooded with crystalviolet dye (White and Wilson 1951) No growth occurredon Mac Conkey agar The organism gave positive resultsfor catalase oxidase nitrate reduction urease and negativeresults for citrate Voges Proskauer and methyl red NoH2S production or glucose utilization were observed Thebacteria agglutinated with anti-R sera and with acriflavinwith no agglutination with anti-A or -M monospecific sera(Alton et al 1988) Based on these results the bacteriumwas classified as B canis

For the antigen preparation a heat soluble bacterialextract was obtained following the method described byMyers et al (1972) with minor modifications Briefly theculture was transferred to 5 mL of Brucella Broth (BBLMicrobiological Systems Cockeysville USA) culturedfor 48 h at 37 C and expanded in Roux flasks containing100 mL of Brucella Agar (Difco Laboratories DetroitUSA) under aerobiosis for 48 h at 37 C as described byCarmichael and Bruner (1968) Bacterial cells were har-vested with 50 mL of sterile PBS (phosphate buffered sal-ine 150 mM NaCl 25 mM KCl 15 mM KH2PO49 mM Na2HPO4 AElig 12H2O pH 74) and inactivated by heat(1 h 56 C) The suspension was filtered through sterilegauze and washed three times by centrifugation(3500 middot g 10 min) in PBS The pellet was then re-sus-pended in 10 mL of PBS and autoclaved at 120 C under15 Atmospheres for 20 min The cells were then centri-fuged at 12000 middot g for 20 min at 4 C The supernatantcollected and stored in 200 lL aliquots at 20 C was usedas the ELISA solid phase antigen The protein concentra-tion of the antigen was determined by the method of Lowryet al (1951)

22 Sera

Twenty positive sera from animals with clinical signspositive bacterial isolations and AGID-positive tests forbrucellosis test were used as reference Ten negative serumsamples from adult healthy animals obtained in a kennelwith no history or evidences of canine brucellosis (suchas reproductive disorders associated to fever lymphade-nopathy or other clinical signs) were tested negative bythe AGID and used as negative reference These referencesera were used for the calculation of specificity and sensitiv-ity of the ELISA test Twenty one samples from healthy6 months old animals kept in a research center were used



ARTICLE IN PRESS




















ARTICLE IN PRESS

Specificity


Sensitivity



3 Results










0

20

40

60

80

100

120

140

160

Mean OD 492 nm

0

20

40

60

80

100

120

140

160

Fre

quen

cy Cut-off 0218

lt005 005 01 015 02 025 03 04 05 06 07 08 09 10 gt10



ARTICLE IN PRESS





4 Discussion








ARTICLE IN PRESS







Acknowledgements


References




















ARTICLE IN PRESS





























ARTICLE IN PRESS




















ARTICLE IN PRESS

Specificity


Sensitivity



3 Results










0

20

40

60

80

100

120

140

160

Mean OD 492 nm

0

20

40

60

80

100

120

140

160

Fre

quen

cy Cut-off 0218

lt005 005 01 015 02 025 03 04 05 06 07 08 09 10 gt10



ARTICLE IN PRESS





4 Discussion








ARTICLE IN PRESS







Acknowledgements


References




















ARTICLE IN PRESS






























ARTICLE IN PRESS

Specificity


Sensitivity



3 Results










0

20

40

60

80

100

120

140

160

Mean OD 492 nm

0

20

40

60

80

100

120

140

160

Fre

quen

cy Cut-off 0218

lt005 005 01 015 02 025 03 04 05 06 07 08 09 10 gt10



ARTICLE IN PRESS





4 Discussion








ARTICLE IN PRESS







Acknowledgements


References




















ARTICLE IN PRESS




























0

20

40

60

80

100

120

140

160

Mean OD 492 nm

0

20

40

60

80

100

120

140

160

Fre

quen

cy Cut-off 0218

lt005 005 01 015 02 025 03 04 05 06 07 08 09 10 gt10



ARTICLE IN PRESS





4 Discussion








ARTICLE IN PRESS







Acknowledgements


References




















ARTICLE IN PRESS





























ARTICLE IN PRESS







Acknowledgements


References




















ARTICLE IN PRESS





























ARTICLE IN PRESS




























Documents

Diagnosis of canine brucellosis by ELISA using an antigen obtained from wild Brucella canis