Animals and sera

For this study, we used sera from 10 beagle dogs experimentally infected with B. canis, obtained in a previous work, kindly donated by Efrén Díaz (CENID SAI; INIFAP SAGARPA, México) (^{Tuxpan Galván, 2015}), confirmed by rapid slide agglutination test (RSAT) and bacterial isolation from blood samples. Briefly, animals were inoculated by instillation in the conjunctiva with 5 x 10⁶ CFU/ml of a B. canis field strain, blood samples were taken at different days post infection and cultured in Ruiz-Castañeda medium. Brucella canis identification was performed by routine biochemical tests according to the methodology described by ^{Alton (1988)}.

Sera from 10 clinically healthy dogs were used, whose last vaccination against any agent was at least six months prior to sampling and which gave a negative result in the RSAT (^{Carmichael and Joubert, 1987}). For selectivity tests, we used three sera from dogs with Leptospira interrogans infections confirmed by the microscopic agglutination test, two sera from dogs recently vaccinated against L. interrogans, Canine Adenovirus type 1 and 2, and Canine Distemper, and serum from a positive dog for gastrointestinal infection with Salmonella spp., as confirmed by culture. All these sera came from the sera stock of the Microbiology and Immunology Department, Faculty of Veterinary Medicine at the National Autonomous University of Mexico.

Antigens

The crude antigen (CA) was obtained by sonicating the RM6/66 strain of B. canis using a previously described method (^{de Oliveira et al., 2011}); briefly, 2 g of heat inactivated (75 °C for 1 h) B. canis biomass were taken and placed in 10 mL of a solution containing 10 mM HEPES pH 7.5, and store at -20 °C until use. Bacteria were thus sonicated (Vibra Cell VCX 130 - Sonics) (4 pulses of 1 min at 15 Hz) in presence of protease inhibitor cocktail (Sigma-Aldrich, Saint Luis, Mo). To corroborate cell disruption, a Gram stain was performed. Once the bacterial biomass had been sonicated, it was centrifuged at 700 x g for 30 min at 4 °C in order to obtain the soluble fraction (supernatant).

The purified protein (GroEL) was obtained from the soluble fraction by fast protein liquid chromatography (FPLC) using an anion exchange column (HiTrap Capto Q; GE Healthcare) with liquid chromatography (AKTA Pure; GE Healthcare). The column was buffered with 5 mL of a solution of Tris-HCl (5 mM) and EDTA (10 mM), then the sample was introduced with a flow range of 1 mL/min and washed with the previously described solution. The elution buffer (1 M NaCl) was included in a linear gradient from 0 % to 100 % in nine volumes of 1 mL each. Each fraction was obtained in a volume of 0.5 mL and was resolved by 12 % SDS-PAGE by duplicate and either stained with Coomassie Blue or transferred onto a PVDF membrane for immunoblotting, using sera from infected dogs; further dog anti IgG as secondary antibody (1:10000) and ECL West femto as substrate (ThermoFisher, Carlsbad, California) were used. Subsequently, fractions were concentrated 3 times in an Amicon ultra 50kDa (Millipore, Merck KGaA, Darmstadt, Germany) and resuspended in PBS and two fractions containing two bands each (between 55 kDa and 70 kDa), were selected to resolve them on a gel using isoelectric focusing (IEF) (a total of 20 µg of protein) followed by two-dimensional polyacrylamide gel electrophoresis and the antigenicity of each spot was verified by Western blotting with the aid of sera from B. canis infected dogs. Three spots were identified as GroEL by liquid chromatography-mass spectrometry as described in a previous work, with a coverage of 29, 51 and 21 % for each spot (^{Morales Aguilar, 2016}) (Supplementary material).

iELISA protocol using the sonicated crude antigen

The ELISA using crude antigen (CA-iELISA) was done on 96-well mid-binding polystyrene plates (Nunc-Immuno Micro Well MaxiSorp; Sigma-Aldrich), adding 240 ng of antigen diluted in 100 µL per well in phosphate buffer solution (PBS); this was homogenized at 100 rpm for 5 min and incubated at 37 °C for 17 +/- 1 h. It was washed three times with 200 µL of a solution of PBS + Tween 20 (Tween 20; Sigma-Aldrich) at 0.1 % (PBS-Tween 20). A solution of bovine serum albumin (Bovine serum albumin; Sigma-Aldrich) diluted to 1 % in PBS (100 µL per well) was added, then incubated for 60 min at 37 °C and washed three times with 200 µL of Tween 20 at 0.1 %. The serum diluted in PBS 1:250 in 100 µL was added, then incubated at 37 °C for 60 min, after which it was washed three times with 200 µL of Tween 20 at 0.1 %. A 100 µL solution of the rabbit anti-canine IgG conjugated with horseradish peroxidase secondary antibody (Rabbit anti-Canine IgG (H+L) secondary antibody, HRP; Invitrogen), diluted 1:2000 in PBS, was added (ABTS solution; Roche), incubated at 37 °C for 60 min, and washed three times with 200 µL of Tween 20 at 0.1 %. Finally, 100 µL of ABTS were added (Sigma-Aldrich) to each well in the dark and incubated with the well covered with aluminum foil, at room temperature in orbital agitation at 100 rpm for 25 min. The plates were read using an ELISA reader (Biotek Instruments) at a wavelength of 405 nm using the program Gen5 (https://www.biotek.es/es/products/software-robotics-software/gen5-microplate-reader-and-imager-software/).

iELISA protocol using GroEL

For the technique using GroEL as the capture antigen (GroEL - iELISA), a 96-well plate (Nunc-Immuno Micro Well MaxiSorp; Sigma-Aldrich) was sensitized using 10 ng of purified GroEL protein diluted in sufficient carbonate buffer solution (0.05 M, pH 9.6) to reach 100 µL volume per well. The plate was placed in orbital agitation at 100 rpm for 5 min, covered with parafilm (Parafilm M; Merck) and incubated at 4 °C for 17 ± 1 h, then washed three times with 200 µL of Tween 20 at 0.1 %. 100 µL of serum diluted 1:500 in PBS with bovine serum albumin at 1 g/L (0.1 M, pH 7) was added, covered with parafilm, incubated for 1 h at 37 °C, and washed three times with Tween 20 at 0.1 %. 100 µL per well of a 1:2000 dilution of the horseradish peroxidase goat anti-canine IgG antibody (Goat anti-Canine IgG (H+L) secondary antibody, HRP; Invitrogen) diluted in PBS (0.1 M, pH 7), was added, covered with parafilm and incubated for 1 h at 37 °C, then washed three times with Tween 20 at 0.1 %. Finally, 100 µL of ABTS was added to each well in the dark, covering the plate with aluminum foil and placing it in orbital agitation at 100 rpm for 25 min. The reading was taken on an ELISA reader (Bio-Tek/ELX808; Biotek) at a wavelength of 405 nm using Gen5 software.

In-house validation protocol

The cutoff point was determined using a previously described method (^{Frey et al., 1998}) and calculated using the following equation:

Where X- is the average of the results obtained from a series of negative controls, SD is the standard deviation of that series of results, t is the percentile (1 - α) of a one-tailed Student’s t distribution with (n - 1) degrees of freedom, and n is the number of controls tested. The results are considered positive for B. canis infection if they are above the result of this calculation. Next, precision was evaluated by calculating intermediate precision and repeatability using the supplemental material from a guide, (^{Andreasson et al., 2015}) where the Coefficient of Variation (%CV) is calculated from a series of independent measurements on different days (Intermediate Precision) as well as a series of simultaneous measurements (Repeatability). In addition, the mean and standard deviation of these parameters were calculated, using a positive control and a negative control. Repetitions were done on five different days, five times each. The robustness of the tests was evaluated for positive and negative sera, varying by +/- 5 °C the incubation temperature of the reference sera, using 37 ºC as a reference temperature. Similarly, the incubation time was varied by +/- 5 min compared to the standard; the results were compared using an ANOVA and significant differences were determined by Tukey’s Honestly Significant Difference (HSD) test (^{Andreasson et al., 2015}). Selectivity was evaluated using sera from dogs positive for diseases other than canine Brucellosis. To evaluate the sample stability, 19 aliquots were prepared from a positive control and a negative control and were subjected to a freeze-thaw process for comparison using the following protocol: (^{Andreasson et al., 2015})

Finally, sensitivity and specificity were tested using 10 negative and 10 positive sera, adjusting the results according to the previously described formula to obtain the 95 % Confidence Interval (IC95%) (DasGupta et al., 2001; ^{Jacobson, 1998}), using the following calculation:

A = 2r + 1.96²

C = 2( + 1.96²)

Constructing the confidence interval as (A-B)/C, (A+B)/C.

I-ELISA using sonicated crude antigen

The cutoff point at 0.98 DO was calculated using 10 negative controls. In the precision test, we found that for the negative control the CV was 3.7 % for repeatability and 13 % for intermediate precision, with a mean of 0.5 DO; for the positive control, CV was calculated as 1.1 % for repeatability and 3.8 % for intermediate precision, with a mean of 3.0 DO.

With respect to the ANOVA results for robustness, there were significant differences (p < 0.05) when temperature was varied, however, the Tukey’s HSD test showed that the difference was between results from the incubation of serum at 32 °C and 42 °C, not between either of these and the standard temperature of 37 °C; the ANOVA of the results of tests varying incubation times showed significant differences (p < 0.01), which the Tukey HSD test showed to be between the incubation times of 55 min and 65 min, as well as each of these times compared to the standard 60 min.

The selectivity test allowed the identification of a false positive test using serum from dogs infected with L. interrogans. The sample stability test showed significant variation in the negative sample starting at 4 h at 25 °C, 24 h in refrigeration (4 °C), and the second freeze-thaw cycle, but there was no variation in the sample that was kept frozen (-20 °C) for 30 days. In the positive control aliquots, there was no significant variation during the whole procedure. The test showed a sensitivity of 100 % (CI95% 70 - 100 %) and specificity of 100 % (CI95% 70 - 100 %).

I-ELISA using GroEL

The cutoff point was calculated as 1.228 DO using 10 negative controls. In the precision test, for the negative control the %CV was 4.6 % for repeatability and 12.9 % for intermediate precision, with a mean of 0.7 DO. For the positive control, the %CV was calculated at 8.1% for repeatability and 16.8 % for intermediate precision, with a mean of 1.3 DO. The ANOVA for robustness showed significant differences (p < 0.01) when temperature was varied, which the Tukey HSD test identified to be between the results of sera incubated at 42 °C and the standard temperature (37 °C) and between incubation at 32 °C versus 42 °C. With respect to the tests where incubation time was varied, the ANOVA showed significant differences in results between 55 and 60 min, 60 and 65 min, and 55 and 65 min according to the Tukey HSD test.

The selectivity test allowed the identification of a false positive test of serum from dogs infected with L. interrogans and Salmonella spp. The test of sample stability showed that for negatives, significant differences began to occur at 168 h at 25 °C, one week in refrigeration, and the second freeze-thaw cycle, while for the positives, there were no differences, even after 168 h at 25 °C, 24 h in refrigeration, and 7 freeze-thaw cycles. However, in both cases there was a curve with significant variations during the first hours of sample storage, both at room temperature and in refrigeration. Neither positives nor negatives varied when samples were stored in a frozen state for up to a month. The test showed a sensitivity of 44 % (IC95% 18 - 73 %) and specificity of 67 % (IC95% 35 - 88 %).