Veterinary Microbiology, 24 (1990) 73-80 73 Elsevier Science Publishers B.V., Amsterdam

Diagnosis of bovine brucellosis by enzyme immunoassay of milk*

P. Kerkhofs L, Y. Botton 2, P. Thiange 2, P. Dekeyser 1 and J.N. Limet 3 ~ National Institute for Veterinary Research, Groeselenberg, 99, B- 1180 Brussels (Belgium)

2Centre de dkpistage des maladie du bktail, chaussbe de Marche, 604 B-5101 Erpent (Belgium) 31nternational Institute for Cellular and Molecular Pathology, avenue Hippocrate, 74, B- 1200 Brussels

(Belgium)

(Accepted 12 December 1989)

ABSTRACT

Kerkhofs, P., Botton, Y., Thiange, P., Dekeyser, P. and Limet, J.N., 1990. Diagnosis of bovine bru- cellosis by enzyme immunoassay of milk. Vet. Microbiol., 24: 73-80.

Enzyme-immunoassays using lipopolysaccharide as antigen were developed for the detection of bovine IgG 1, IgG2 or IgA Brucella antibodies (Ab) in milk. The results of these tests were compared with those of the milk ring test (MRT) by analyzing 3212 bulk milk samples from farms located in regions where brucellosis is prevalent. Among the 105 herds detected by ELISA and/or MRT, 29 infected herds were detected by ELISA only. The 40 MRT-positive herds were also ELISA positive. Five herds became infected during the study and were detected by ELISA 15 days to 6 months prior to detection by MRT.

The ELISA IgG1 titration (IgG 1 ELISA) detected 92.8% of the herds found positive in the three ELISA assays. The concomitant use oflgA ELISA raised the sensitivity to 100% but slightly decreased the specificity. The IgG2 ELISA did not improve the diagnosis. The sensitivity of MRT and lgG1 ELISA was compared by testing successive dilutions in negative milk of 110 individual MRT positive milks samples. On average, IgG 1 ELISA was 22 times more sensitive than MRT.

INTRODUCTION

The milk ring test (MRT), developed by Fleischauwer in 1937, is com- monly used for brucellosis screening in dairy herds. The coloured Brucella suspension (Alton et al., 1977) is mainly concentrated in the cream layer when milk contains antibodies (Ab). In the absence of Ab, it is homogene- ously distributed in the milk under the white cream layer. However, MRT will be less and less able to detect the antibodies of one positive cow in the bulk milk of a refrigerated tank, due to increasing number of lactating cows

*This work was supported by the Institut pour l'Encouragement de la Recherche Scientifique dans l'Industrie et l'Agriculture (IRSIA).

0378-1135/90/$03.50 © 1990 Elsevier Science Publishers B.V.

74 P. KERKHOFS ET AL.

per farm. Refrigerated tanks are now capable of containing the milk of 50, 100, and even 200 cows. Moreover, the efficacy of the MRT seems to be due more to the frequency of testing than to its sensitivity (Fensterbank, 1986 ).

Forschner and Bringer (1986) described an ELISA for the detection of bo- vine Brucella Ab in bulk milk. This test succeeded in detecting milk from one infected cow diluted 100 fold in normal milk. In this article, we described immunoassays for independant titration of IgG 1, IgG2 and IgA Brucella Ab in bulk milk. IgA Ab, the most active in the coloured ring formation, are mainly produced in the mammary gland and are therefore more specific for udder infection. IgG 1 Ab, less active in the ring formation, however, are detectable in the milk earlier after infection (Sutra et al., 1986; Sutra and Dubray, 1987 ). The assays developed were evaluated for specificity with 1457 bulk milks from brucellosis free herds. The sensitivity was estimated by the titration of 3212 bulk milks from regions where brucellosis is prevalent and dilutions in nega- tive milk of 110 MRT positive milk samples.

MATERIALS AND METHODS

Milk Ring test (MRT)

This test was described by Alton et al. ( 1977 ). Briefly, 50/zl ofhematoxylin coloured antigen (IFFA-Merieux, Lyon) was mixed with 1 ml milk in a dis- posable polystyrene tube of 6 mm in diameter and incubated for 1 h at 37 ° C. A negative result corresponds to a white cream layer upon a uniformly col- oured milk. The degree of a positive reaction depends on the colour intensity of the cream layer and the proportion of the coloured antigen remaining in the milk.

ELISA

Buffers. Glycine buffer saline (GBS): 0.17 M NaC1, 0.1 M glycine adjusted to pH 9.2 with NaOH containing 400 mg/1 NAN3. GBS-EDTA-Tw: GBS buffer containing 50 mM EDTA and 0.1% Tween 80, pH 9.2. GBS-EDTA-cholate: GBS buffer containing 50 mM EDTA and 2% sodium cholate, pH 9.2. Cit- rate-phosphate buffer: 0.051 M NazHPO4, 0.024 M citric acid, pH 5. The washing solution was NaCI-Tw.: 0.15 M NaCI, 0.01% Tween 20.

Antigen. The lipopolysaccharide (LPS) of Brucella abortus biovar 3 (field strain ) was obtained by phenol extraction of heat-killed and freeze-dried bac- teria according to Westphal and Jann (1965) and purified as described by Moreno et al. (1979).

Anti-bovine immunoglobulin Mab. Mab 1C8, 3H3 and 16.35 were used sepa- rately. They recognize the heavy chain of IgG1, heavy chain of IgG2 and

DIAGNOSIS OF BOVINE BRUCELLOSIS BY ENZYME IMMUNOASSAY OF MILK 7 5

weakly those o f l g G l , and IgA respectively. The Mab 1C8 and 3H3 were pro- duced and provided by professor Depelchin and Dr. Letesson of "Centre d ' Immunologie Appliqu6e ~ l'61evage, "IRSIA", Namur (Letesson et al., 1985 ). Mab 16.35 was a gift from Dr. van Zaane of CDI, Lelystad (Van Zaane, 1987).

These Mabs were conjugated to periodic acid activated peroxidase (Nak- ane and Kawao'i, 1974) and stored at - 20 ° C in 50% glycerol.

Assay. Polystyrene microtitration plates (Greiner labortechnic, Stuttgart) were coated by incubation with 100/A of a 1 m g / m l LPS solution in 5-fold water-diluted GBS. The min im um incubation time for coating was 3 h at 37°C. The coated plates may be stored for several months either at 4°C or washed, dried and packed in a plastic bags with a dessicant capsule.

After vigorous stirring, milk samples were 2-fold diluted in GBS-EDTA- cholate. Plates were washed four times with NaC1-Tw by using a 12-channel manual immunowash (Nunc) , and 50/ t l of each milk dilution were dis- pensed in a well. After 1 h incubation at room temperature, the plates were washed five times. Fifty #1 of conjugated antibody diluted in GBS-EDTA-Tw containing 2% fetal calf serum were dispensed in the wells and the plates were incubated for 1 h at room temperature in the dark. After five washings with NaC1-Tw, the amount of bound conjugate was revealed by dispensing 100/~1 of 0.4 m g / m l ortho-phenylene-diamine solution and 2 mM H202 in citrate- phosphate buffer. The colour development was stopped after a max imum of 20 min incubation in the dark by the addit ion of 25/~1 of 2 M H2SO 4 in each well. The light absorbance was measured at 492 nm and 620 nm. The tests were carried out in duplicate.

Origin of the samples

Brucellosis free herds. These herds were selected in a part of the country where the prevalence of brucellosis is very low. The bulk milk samples we used were negative to the MRT and all the animals of the corresponding herds were negative to the serum agglutination test (SAT) and rose bengal plate test (RBPT).

Herds from regions where brucellosis is prevalent. 3212 bulk milk samples were tested. SAT and RBPT were performed on serum of all the corresponding animals (Alton et al., 1977 ). Positive sera were retested in the complement fixation test (CFT) . Brucella organisms were isolated from milk samples or cotyledons and abortion products (Alton et al., 1977) on at least one animal in 51 of the 69 farms which presented positive sera.

The animals of 5 herds had been vaccinated and 13 herds included animals with positive serology but no culture had been performed.

76 P. KERKHOFS ET AL.

MRT positive milk samples. To evaluate the sensitivity of the ELISA, MRT and IgG 1, ELISA were carried out in parallel on serial dilutions of 110 MRT positive individual milk samples. The serial dilutions ( 1/3 to 1/2187 ) were done in milk from uninfected cows shown to be negative for both tests. Milk samples were stored either frozen at - 2 0 ° C or at 4 °C after addition of 0.2% formaldehyde.

RESULTS

Characteristics of the three ELISAs Calibration curves of the three assays are presented in Fig. 1. By using a 12-

channel manual immunowash or three washings of 3 min after immersion in the NaC1-Tw, the coefficient of variation within a plate was less than 10%.

The optical densities observed for 1457 milk samples from uninfected farms were compared to those of successive dilutions ( 1 / 1000 to 1 / 128 000) of a strongly positive milk sample for IgA ELISA. For the IgG1 and IgG2 ELISA, a reference serum with a CFT value of 1200 IU was used as reference. The optical densities observed for the MRT negative milk samples obtained from brucellosis free farms were generally low (less than 0.060) and 98.9, 98.8, and

o£

10 3 lO 4 10 5

D I L U T I O N ( I / X )

Fig. 1. S tandard curves of the three assays. Mean absorbance values and s tandard deviat ions were calculated f rom data ob ta ined with four different plates run on the same day.

DIAGNOSIS OF BOVINE BRUCELLOSIS BY ENZYME IMMUNOASSAY OF MILK 77

97.9% of the samples were lower than the value corresponding to the mean plus three standard deviations for IgG 1, IgG2 and IgA respectively. However, cut-off values equal to the mean plus eight to fifteen standard deviations have been used to ensure a 100% specificity

Field application of ELISA; comparison with MRT Among the 3212 milk samples tested, 105 samples were positive to one or

more of the three ELISAs and MRT. Only 40 samples were positive to both ELISA and MRT. Herds were classified on the basis of the serological exam- ination of their animals, i.e. the results of the SAT, CFT and RBPT on serum. When available, bacteriological results were taken into account; all culture positive herds were also serologically positive.

Table 1 presents the ELISA and MRT results in function of this classifica- tion. MRT detected only 58.0% of the herds with a positive serology whereas the three ELISAs combined detected all of them. The 40 MRT positive herds were also ELISA positive. Twenty-nine infected herds were positive in ELISA and negative in MRT. Five of them were recently infected and were discov- ered earlier by ELISA than by MRT. ELISA results were confirmed by MRT fifteen days later (1/5 ), or by serology (4/5) . The serological tests became clearly positive from 15 days to 6 months later.

Twelve false positive reactions were observed in the MRT and not in ELISA; and one false positive reaction was observed in both tests. ELISA was, how- ever, found positive in 24 herds without serological or bacteriological evi-

TABLE 1

Results of the MRT and ELISA in function of the serology of the herds taken as index of the infection

Positive Negative Sensitivity Specificity serology serology (%) (%)

MRT + 40 13 58.0 M R T - 29 3199 99.6

EL IgGl + 64 9 92.8 EL IgGl - 5 3203 99.7

EL IgG2+ 34 4 49.3 EL I g G 2 - 35 3208 99.9

EL IgA+ 34 16 49.3 EL I g A - 35 3196 99.5

EL Ig + 69 24 100 E L I g - 0 3188 99.3

Serological tests were SAT, CFT and RBPT. EL Ig, combination of the results of the three ELISAs.

78 P. K E R K H O F S E T AL.

dences of infection (0.7%). The specificity and sensitivity of the ELISA were 99.3 and 100%, respectively, compared to 99.5 and 58.0% for MRT. Values for each assay are given in Table 1.

IgGl ELISA was clearly the most efficient assay. It detected the greatest number of culture-positive infected herds and herds with positive serology, not confirmed by the isolation of Brucella. The five MRT negative infected herds discovered during the study were IgG1 positive, but five milk samples from infected herds were only IgA ELISA positive.

The IgA ELISA slightly improved the sensitivity of detection but detected the greatest number (16) of farms with negative serology. The 3H3 Mab that essentially detects IgG2 gave very few false positive responses (4), but missed 51% of infected herds.

Comparison of the sensitivity of lgG1 ELISA and MRT Serial dilutions of I 10 positive individual milk samples with negative milk

were tested by both MRT and ELISA. Fig. 2 illustrates the ratio of the last dilution positive by ELISA compared to MRT. Only six milk samples exhibit the same titre in both tests. The mean titre ratio was 22 and the highest in- crease in sensitivity observed by ELISA for 12 samples was 81. No milk was IgG 1 ELISA negative and MRT positive. The highest dilution found positive was 1/2187, for only one sample in MRT and for 11 samples in ELISA.

The increase of sensitivity was not significantly influenced by the method of preservation.

5 0

4O

..d t l zs

< 30

20

z 10

0

1 3 9 27 81

ELISA t i t r e / MRT t i t r e

Fig. 2. Comparison of end points of ELISA and MRT on 110 positive individual milk samples. The mean value of the ratio of the ELISA titre to MRT titre was 22.

DIAGNOSIS OF BOVINE BRUCELLOSIS BY ENZYME IMMUNOASSAY OF MILK 79

DISCUSSION

In a previous paper we described an ELISA for the titration of Brucella antibodies in serum using LPS as antigen (Limet et al., 1988 ). LPS is consid- ered the immunodominan t antigen (Moreno et al., 1987 ) and LPS antibod- ies are implicated in all classical tests used for brucellosis diagnosis. Impor- tant cross reactions are observed mainly with Yersinia LPS but no specific antigen has yet been clearly identified (Corbel, 1985 ).

The present work shows that the ELISA detected Brucella Ab in milk more efficiently than the MRT and therefore facilitates the earlier identification of infected herds. By analyzing 3212 samples, we located by ELISA six culture positive and five newly infected herds presenting a negative MRT. All in- fected herds positive to the MRT, were positive to ELISA. ELISA is able to analyze a milk sample unsuitable for the MRT (i.e. too little cream, too many cells, colostrum, clotted milk, frozen milk).

IgM Ab were not assayed owing to their very weak specificity. The IgA ELISA presented the highest number of false positive results but some MRT positive milk samples are only IgA ELISA positive. IgG2 Ab seems to be very specific but in milk they are a poor indicator of infection. Therefore, the IgG 1 ELISA seems to be the most suitable assay. The ideal assay protocol would probably be a simultaneous detection of IgG 1 and IgA Ab followed by a con- firmation of the positive samples using each test individually.

A few samples of milk from uninfected herds gave absorbance value above the mean plus 3 standard deviations. A better knowledge of the nature of the interaction involved in such "false positive" reactions may enable us to de- vise an even more sensitive test.

The correlation between the IgG 1 ELISA of bulk milk and serology of in- dividual cows was better than that between the MRT and serology. Sutra and Dubray ( 1987 ) explained the discordances between MRT and serological tests by the fact that MRT and serological tests do not detect the same antibody classes. For a positive MRT, IgA (produced following the udder infection) or serum IgM must be present in adequate quantities, IgG 1 Ab is not directly implicated in the ring formation. Although IgG2 and the IgGA play an active role in the MRT (Sutra et al., 1986), in ELISA, IgG1 Ab detects more in- fected herds. Heck et al. (1980) also observed a high positive correlation be- tween serological tests and ELISA on milk samples.

The assay developed by Forschner and Bringer ( 1986 ) detected the milk of an infected animal diluted 100 fold in negative milk. Our results suggest that the increase in sensitivity observed varies largely from one sample to another. Some positive milk samples may only be diluted 3-fold whereas others are still positive after a 1/2187 dilution. In our comparison, ELISA was on an average 22 times more sensitive than MRT.

8 0 P. KERKHOFS ET AL.

ACKNOWLEDGEMENTS

We thank P. Matholet and N. Tuyttens for providing the milk samples and corresponding health status and Prof. A. Depelchin and D. Van Zaane for providing us monoclonal antibodies. We also thank A. MacMillan for reading the manuscript; R. Wijffels, and M. Mammericks for their useful help in this work, and A. de Boeck, J. Van Broeck and P. Michel for their technical assistance.

REFERENCES

Alton, G.G., Jones, L.M. and Pietz, D.E., 1977. Laboratory Techniques in Brucellosis, 2nd Edn. Monograph Series No. 55. World Health Organisation, Geneva, 167 pp.

Corbel, M.J,. 1985. Recent advances in the study of Brucella antigens and their serological cross- reaction. Vet. Bull., 55 (12) 927-942.

Fensterbank, R., 1986. Brucellosis in cattle, sheep and goats: diagnosis, control and vaccination. Rev. Sci. Tech. Off. Int. Epiz., 5 (3): 605-618.

Forschner, E. and BiJnger, I., 1986. Detection of antibodies against brucellosis in farm bulk milk samples by ELISA. Dtsch. Tier~irzt. Wochenschr., 93: 269-273.

Heck, F.C., Williams, J.D., Pruett, J., Sanders, R. and Zink, D.L., 1980. Enzyme-linked immu- nosorbent assay for detecting Ab to Brucella abortus in bovine milk and serum. Am. J. Vet. Res., 41: 2082-2084.

Letesson, J.J., Lostrie-Trussaert, N. and Depelchin, A., 1985. Production d'anticorps mono- clonaux specifiques d'isotypes d'immunoglobulines bovines. Ann. M6d. V6t., 129:131-141.

Limet, J.N,. Kerkhofs, P., Wijfels, R. and Dekeyser, P., 1988, Le diagnostic S6rologique de la brucellose bovine par ELISA. Ann. M6d. V6t., 132: 565-575.

Moreno, E., Munro, W.P., Jones, L.M., Schurig, G.B. and Berman, D.T., 1979. Purification and characterisation of smooth and rough lipopolysaccharides from Brucella abortus. J. Bacte- riol., 138: 361-369.

Moreno, E., Borowiak, D. and Mayer, H., 1987. Brucella lipopolysaccharides and polysacchar- ides. Ann. Inst. Pasteur/Microbiol., 138:102-105.

Nakane, P.K. and KawaoY, 1974. Peroxidase-labelled antibody, a new method of conjugation. J. Histochem. Cytochem., 22:1084-1091.

Sutra, L. and Dubray, G., 1987. Interpr6tation des discordances entre le ring test et les epreuves serologiques dans le depistage de la brucellos bovine. Bull. Soc. Vet. Prat. France, 71: 565- 574.

Sutra, L., Caffin, J.P. and Dubray, G., 1986. Role of milk immunoglobulins in the Brucella milk ring test. Vet. Microbiol., 12: 359-366.

Van Zaane, D., 1987. Five years monoclonal antibodies at the Central Veterinary Institute. Tijdschr. Diergeneeskd., 112: 264-270.

Westphal, O. and Jann, K., 1965. Bacterial lipopolysaccharides: extraction with phenol-water and further application of the procedure. In: R.L. Whistler (Editor),Methods in Carbohy- drate Chemistry, Vol. 5. Academic press, New York, NY, pp. 89-95.