Seroprevalence and characterization of Brucella species in cattle slaughtered at Gauteng abattoirs, South Africa

Abstract Background Brucellosis is an infectious and contagious zoonotic bacterial disease of both humans and animals. In developing countries where brucellosis is endemic, baseline data on the prevalence of brucellosis, using abattoir facilities, is important. Objectives The aim of this study was to determine the seroprevalence of antibodies against Brucella in slaughter cattle at Gauteng province, South Africa and to characterize isolates of Brucella spp. Methods In this cross‐sectional study, un‐clotted blood samples with corresponding organ tissue samples were collected from slaughtered cattle. Serological [Rose Bengal test (RBT), complement fixation test (CFT) and indirect ELISA (iELISA)], molecular (PCR) and bacteriological methods were used to detect Brucella antibodies and Brucella spp. from 200 slaughtered cattle in 14 abattoirs. Results The RBT revealed a seroprevalence of brucellosis as 11.0% (22 of 200) and iELISA confirmed 5.5% (11 of 200). The estimated seroprevalence from RBT and iELISA was 5.5% while RBT and CFT was 2.0% (4 of 200). Brucella melitensis (n = 6) and B. abortus (n = 5) were isolated from 11 cattle tissues (5.5%) as confirmed to species level with AMOS PCR and differentiated from vaccine strains with Bruce‐ladder PCR. Seven of the 11 isolates originated from seropositive cattle of which five were biotyped as B. abortus bv 1 (n = 2) and B. melitensis bv 2 (n = 1) and B. melitensis bv 3 (n = 2). Conclusions This is the first documentation of B. melitensis in cattle in South Africa. The zoonotic risk of brucellosis posed by Brucella‐infected slaughter cattle to abattoir workers and consumers of improperly cooked beef cannot be ignored.


| INTRODUC TI ON
Brucellosis is an infectious and contagious zoonotic bacterial disease of humans and a wide range of domestic animals and wildlife, particularly the ruminants (Corbel, 2006;Radostits, Gay, Hinchcliff, & Constable, 2006;Smirnova et al., 2013) and some marine animals (Foster, Osterman, Godfroid, Jacques, & Cloeckaert, 2007;Scholz & Vergnaud, 2013). The Brucella species are Gram-negative, non-capsulated, facultative intracellular, non-spore forming, cocco-bacilli bacteria (Godfroid, 2012;Seleem, Boyle, & Sriranganathan, 2010;Smirnova et al., 2013). Brucella spp. infecting farm animals include B. abortus, B. melitensis, B. suis and B. ovis (Godfroid, Nielsen, & Saegerman, 2010;Smirnova et al., 2013). The Brucella species are known to have host preferences, although there could be cross-infection with other hosts. Brucella abortus has a host preference for the cattle but it can cause infection in other hosts including humans (undulant fever). Brucella melitensis, has a host preference for sheep and goats and it is the most pathogenic of the Brucella spp. that causes infection in humans (Malta fever). Brucella canis and B. suis has host preferences for dogs and pigs, respectively, and can cause brucellosis in humans (Alton, 1990;Carmichael, 1990;Godfroid et al., 2005;Pappas, 2013). Brucellosis has been eradicated or well controlled in developed countries (Pappas, Papadimitriou, Akritidis, Christou, & Tsianos, 2006). However, in many of the low and middle income countries (LMICs) such as in Africa, South and Central America, Middle East, Asia, Mediterranean Basin and the Caribbean, brucellosis is still common and high in occurrence both in the animal and human populations (Adesiyun & Cazabon, 1996;Godfroid et al., 2005;Olsen & Palmer, 2014;Pappas et al., 2006). Brucellosis may have existed in South Africa as an ancient disease as suggested by a paleopathological analysis study on the fossil of the late Pliocene hominin species (D'Anastasio, Zipfel, Moggi-Cecchi, Stanyon, & Capasso, 2009). It had been suspected in 1898 that goats may have been the source of suspected cases of undulant fever which was also called "camp fever" in 40 patients around the Kimberley area of South Africa, an area where diamond mines were operating (Strachan, 1932;Van Drimmelen, 1949). Serological diagnosis, as well as cultural isolation, which identified B. melitensis, were relied upon for diagnosis from 1902 to 1911 (Strachan, 1932;Zammit, 1905) from samples including human blood, goat serum and milk samples in South Africa (Strachan, 1932). Brucella melitensis outbreaks have been documented in sheep in 1965 (Van Drimmelen, 1965), in goats in 1989 that was identified as B. melitensis bv 1 (Ribeiro, Herr, & Chaparro, 1990), 1994 (Reichel, Nel, Emslie, & Bishop, 1996), 2000 (Emslie & Nel, 2002(DAFF, 2015. Brucella abortus infection in South Africa was reported in 1913, when contagious abortion was observed to spread across the country in cattle (Van Drimmelen, 1949). In South Africa, B. abortus bv 1 predominantly cause infection in cattle (90%) and B. abortus bv 2 to a lesser extent (Godfroid, Bishop, Bosman & Herr, 2004). Mauff (1980) reported fives cases of brucellosis associated with by-products (condemned meat and unborn calves) at a new abattoir in South Africa. These were confirmed cases based on serological tests and culture and it was reported that the affected individuals did not wear any protective clothing at the by-product facility (Mauff, 1980). Despite this risk and threat of exposure of humans to brucellosis at abattoirs, the only published report of an abattoir-based study on bovine brucellosis was in 1984 where a prevalence rate of 1.5% was reported for cattle sampled at Cato Ridge abattoir in KwaZulu-Natal province (Bishop, 1984). In the Eastern Cape province, a 9.2% prevalence rate of B. abortus (of which 0.8% B. abortus S19 vaccine strain) was isolated from cattle, 2.9% B. melitensis from sheep and 6.3% B. melitensis from goats using different samples (blood, milk and lymph nodes) followed by species specific confirmation using PCR (Caine, Nwodo, Okoh, & Green, 2017 (OIE, 2016). These control measures amongst others are instituted to prevent a spillover of the disease to other domestic animals and wildlife in areas close to the wildlife parks (Simpson et al., 2018).
In South Africa, an estimation of over 3,476,000 of cattle were slaughtered from September 2015 to August 2016 in the abattoirs (DAFF, 2016). These abattoir facilities can also be used to monitor disease control policies, detect newly introduced disease agents and to assess intervention programmes, such as brucellosis vaccination, and most importantly, abattoir survey may also facilitate early intervention to mitigate the epidemic loss of animals (Alton, Pearl, Bateman, McNab, & Berke, 2015;Fasina et al., 2015;Kaneene, Miller, & Meyer, 2006). As such, an abattoir surveillance study on brucellosis can generate baseline data on the occurrence of the disease among the animal population, especially when the animals come from various farms to be processed into wholesome meat products for human consumption (Alton et al., 2015;Fasina et al., 2015;Kaneene et al., 2006). However, abattoir data may not produce reliable prevalence estimates because the population of slaughtered cattle tends not to correctly represent the target population which may affect the validity of results from such facilities.
Animals and carcasses of Brucella-infected slaughtered animals can be a source of infection to susceptible abattoir workers, as these workers may be exposed to infection through direct contact with infected animal's secretions or blood, or indirectly through the consumption of raw meat or undercooked meat (Corbel, 2006). It has been documented in the literature that the risk of carcass contamination by bacteria, including Brucella spp., increases with the throughput, i.e. the number of animals slaughtered (Sadler, 1960).
Due to the predominant voluntary nature of the brucellosis testing scheme in South Africa, known and unknown brucellosis infected cattle can be slaughtered at abattoirs. The aim of this study was to determine the seroprevalence of antibodies against Brucella and characterize Brucella spp. from slaughtered cattle from abattoirs.

| Study area, study design and sample size
A cross-sectional study was conducted to determine the seroprevalence of brucellosis and to detect, isolate and characterize Brucella Fourteen abattoirs in the Gauteng Province (Figure 1), which were operational and consented to participate in the study, were randomly selected from 28 operational abattoirs identified. These abattoirs were categorized into high throughput (n = 7) slaughtering more than 20 cattle per day and low throughput (n = 7) where 20 or less cattle are slaughtered daily. Among the selected abattoirs for the study, 13 operated as multi-species and 1 was mono-species facilities.

| Sample size and collection
In this cross-sectional study, a systematic simple random sampling method was used to determine the sample size and power. Sample size was estimated at 50.0% expected frequency, with 5.0% acceptable margin of error and a design effect of 1 and clusters equals to 1 in Epi-Info 7 version 10. At a 95.0% confidence level, the sample size of 200 animals was achieved. A systematic sampling method was used at selected abattoirs. For all selected animals, blood samples were collected as follows: Unclotted blood was collected from 200 cattle at the point of slaughter using sterile 50 ml cups with approximately 5 ml of the blood aliquoted into yellow-capped vacutainer tubes.
Corresponding tissue samples comprising of lymph nodes, spleen and liver were collected from each of the cattle. The lymph nodes of each animal were pooled and comprised the retropharyngeal, parotid, submandibular and mesenteric lymph nodes. The excised tissues were processed according to set laboratory protocols F I G U R E 1 Locations of abattoirs sampled in the Gauteng Province of South Africa in a bio-safety laboratory level 2 hood. The homogenized tissues were used for DNA extraction (PCR) and bacterial isolation.
Demographic data comprising of the animal species, breed, sex, age, the farm origin and the identity of the abattoir where the animal was slaughtered were collected for each of the animals slaughtered. The age of the animals was determined using the dental formula as described by Eubanks (2012). The vaccination status of the animals could not be ascertained.

| Serological test methods on serum samples
Serological procedures were carried out using three different procedures including the RBT, the CFT and the indirect enzyme-linked immunosorbent assay (iELISA). For this study, because RBT is unable to differentiate between antibodies produced in response to vaccination or natural exposure, its results were not used as stand-alone percentage estimates but only as part of a serial test in a combination with either CFT or iELISA results.
Rose Bengal test was conducted as described by OIE (2009). The diagnostic sensitivity and specificity for the RBT have earlier been confirmed to be 100.0% and 75.0%, respectively, based on previous validation studies (Nielsen et al., 2005;Stemshorn et al., 1985).

| Bacterial isolation from tissue samples
Homogenate (200 μl) from each tissue (lymph nodes, spleen and liver) was inoculated onto Farrell's and modified CITA media, respectively, and incubated at 37°C with 5.0% CO 2 . Plates were observed for bacterial colony growth for 10 days. Brucella organisms were identified presumptively by morphology using the Stamp's modified Ziehl-Neelsen staining method (OIE, 2009).
Morphologically identified Brucella colonies were purified by subculturing on the sheep blood agar media. Mixed or contaminated cultures were subjected to serial dilution 1:1,000 using buffered peptone water and inoculated onto the selective media until purified colonies were obtained. Pure Brucella cultures were biotyped at ARC-OVR, South Africa according to the methods described by .

| Molecular detection
Genomic DNA was extracted from cultures of each homogenized tissue (lymph nodes, spleen and liver) and thereafter, the cells of Brucella cultures were purified using the set protocol according to Isolate II Genomic DNA kit by Bioline (South Africa).
Detection of DNA and isolation procedure for Brucella spp. isolation was performed on all animals, regardless of their serological status. Genus-specific 16S-23S rRNA interspacer region (ITS) PCR was used to amplify Brucella region and B. abortus strain 544 and B. melitensis Rev 1 served as positive controls as described by Keid et al. (2007) for detection Brucella DNA in tissue samples (lymph nodes, spleen and liver). The process amplified a 214 bp fragment using primers (ITS66: ACATAGATCGCAGGCCAGTCA and ITS279: AGATACCGACGCAAACGCTAC). Primers were used at a final concentration of 0.2 μM with 1× DreamTaq Green PCR Master Mix (ThermoFisher Scientific, South Africa) and 2 μl DNA in a 15 μl PCR reaction. The PCR cycling condition consisted of 95°C for 3 min, followed by 35 cycles at 95°C for 1 min, 60°C for 2 min, 72°C for 2 min and a final extension at 72°C for 5 min. PCR products were analysed by electrophoresis using a 2.0% agarose gel stained with ethidium bromide and viewed under UV light.
The multiplex AMOS PCR assay that identifies and differentiates B. abortus, B. melitensis, B. ovis and B. suis was conducted as described (Bricker & Halling, 1994;Weiner, Iwaniak, & Szulowski, 2011) using DNA extraction from cultures. Four species-specific forward primers were used at a final concentration of 0.1 μM with 0.2 μM reverse primer IS711 (Table 1) with 1× MyTaq™ Red PCR Mix (Bioline, South Africa) and 2 μl of template DNA in 25 μl PCR reaction. PCR cycling condition was initial denaturation at 95°C for 5 min followed by 35 cycles at 95°C for 1 min, 55.5°C for 2 min, 72°C for 2 min and a final extension step at 72°C for 10 min. PCR products were analysed by electrophoresis using a 2.0% agarose gel stained with ethidium bromide and viewed under UV light.
A multiplex Bruce-ladder PCR assay to identify and differentiate between vaccine strains and field isolates of Brucella spp. was conducted as described (García-Yoldi et al., 2006;Lopez-Goñi et al., 2008;Weiner et al., 2011) (Table 2). Eight species-specific forward and reverse primers were used at a final concentration of 6.25 μM with 1× MyTaq™ Red PCR Mix (Bioline, South Africa) and 2 μl of template DNA in a 25 μl PCR reaction. The PCR cycling condition included an initial denaturation cycle at 95°C for 5 min followed by 25 cycles at 95°C for 30 s, at 64°C for 45 s, and at 72°C for 3 min and a final extension step at 72°C for 10 min. PCR products were analysed by electrophoresis using a 2.0% agarose gel stained with ethidium bromide and viewed under UV light.

| Statistical analysis
Data collected were managed using the Microsoft Excel version 2007. The R software (RCoreTeam, 2013) was used to analyse the data and to conduct descriptive analysis, and Epi-Info 7 version 10 was used to conduct analyses of frequency with 95% confidence interval, and calculated odds ratio and Chi-Square tests to examine differences between groups.
Animal prevalence was determined by the number of positive animals divided by the total number of animals sampled.

| Ethical approval
Ethics approval for the study was obtained from ARC-OVI Animals Ethics Committee (AEC12-16), University of Pretoria Animal ethics Committee (V089-16). Section 20 approval was granted according to Act 35 of 1984 by the Directorate of Animal Health, South Africa.
Of the 25 ITS-PCR positives, 56.0% (14/25) were from seropositive cattle (RBT or ELISA), while 44.0% (11/25) were from seronegative cattle. The serology positivity, PCR detection and isolation rates among the cattle population according to sex and age with the chisquare measure of association is shown in Table 3. Type of abattoir is an indication for carcass contamination that can be detected by molecular and bacteriological methods.
Five of the 11 Brucella isolates were purified and biotyped. Two were determined to be B. melitensis bv 3, one as B. melitensis bv 2 and the remaining two as B. abortus bv 1. The five Brucella isolates were from iELISA-positive cattle.

| D ISCUSS I ON
In this study, antibodies as well as the pathogen, were detected by a polyphasic approach using serology (RBT, CFT and iELISA), PCR and culturing. The brucellosis estimated seroprevalence and culture prevalence (gold standard) among the slaughter cattle at the Gauteng abattoirs is at 5.5%. Brucella abortus bv 1 was isolated in this study, which was expected since B. abortus bv 1 is the most dominant biovar causing infection in cattle (Coetzer, Thomson, & Tustin, 1994). For the first time B. melitensis bv 2 and 3 were also isolated from cattle tissue. In the past B. melitensis bv 1 has been isolated from goats (Reichel et al., 1996) and humans (Wojno et al., 2016), with S19 is unknown as the vaccine producer of S19 (Onderstepoort Biological Products) only indicate overall total vaccine sold. There was also a period when B. abortus S19 vaccine was unavailable (DAFF, 2017). There is no compulsory testing to monitor B. melitensis in sheep and goats in South Africa (DAFF, 2017). Brucella melitensis outbreaks reported in South Africa in sheep and goats have mostly been noticed only when human brucellosis was detected in individuals associated with sheep and goats (Emslie & Nel, 2002).
The detection of B. abortus bv 1 has been documented to be the common species and biotype infecting the cattle population in South Africa (Coetzer et al., 1994;Gradwell, 1977) and in other countries especially in the Southern African region and the Caribbean Africa (Kolo et al., 2018). The occurrence of B. melitensis in cattle in this study may be as a result of rearing cattle together with sheep, goats and wildlife on the same farm or sharing of grazing land with sheep or goats (Radostits et al., 2006;Verger, Garin-Bastuji, Grayon, & Mahé, 1989). This has huge implication to infection of other cattle in the herd and most importantly spill over to other farms and other species around that geographical areas (Godfroid et al., 2014).

Brucella melitensis bv 2 and 3 have never been isolated in livestock
in the country. This has significant implications for brucellosis control in South Africa as the brucellosis scheme only focus on testing bovine high risk herds. As this study report B. melitensis bv 2 and 3 to be present in cattle, there is a high probability of its presence in the sheep and goat populations. B. melitensis has been isolated from sable antelope (DAFF, 2015) and biotyped as B. melitensis bv 3 (unpublished results). Brucella melitensis was isolated and reported in humans in the Western Cape province in 2015 (Wojno et al., 2016) and biotyped as B. melitensis bv 1 (van Heerden unpublished data). In future studies, the impure culture could be purified by picking multiple colonies resulting from the same animal and speciating the pure cultures using PCR followed by biotyping.
The possibility exists that a higher frequency of false positives and tests occurs when RBT is used relative to other tests which led to the recommendation that the RBT should not be used as a standalone test to determine the seroprevalence of brucellosis (OIE, 2009).
Since the current study was unable to confirm the vaccination status of the slaughter animals tested, the possibility of vaccine-induced antibody production, unrelated to natural exposure to Brucella spp.
by both tests cannot be ignored. However, molecular (Bruce-ladder PCR) and isolation methods were used to differentiate vaccine strains from our isolates ( Figure 3) and no vaccine strain was identified.
It is of diagnostic significance that Brucella DNA was detected using ITS-PCR in 11 seronegative cattle along with 14 seropositive cattle, an overall prevalence of 12.5% (25/200). The ITS-PCR (genus-specific) is a good option as a screening tool for Brucella DNA as the PCR is capable of detecting very small amount of DNA in the tissue samples even if as little as 3.8 fg of Brucella DNA mixed with 450 ng of host DNA (Keid et al., 2007). This is significant because the application of PCR to detect Brucella DNA in animal tissues can be used to diagnose brucellosis in immunecompromised animals that are unable to seroconvert following exposure and infection, and in animals where the Brucella organism, which is an intra-cellular organism, is localized in the tissues, such that serology may not be able to diagnose the infection from the serum samples (Radostits et al., 2006). However, the ITS-PCR sensitivity and specificity should be validated in South Africa especially to ensure the specificity of this PCR, as it could react with Brucella-like organisms in this region that could not be tested in the initial validation by Keid et al. (2007). The PCR method used in our study has proven to be fast, safe and does not require specialized laboratory as required in the bacteriological methods. Based on the detection rate of 92.0% in the lymph node tissues in our study, we recommend the pooling of the lymph nodes initially to assay for Brucella DNA. The strategic application of bacteriology, serology and PCR was shown to result in increased detection frequency of brucellosis in cattle.

| RECOMMENDATIONS
It is recommended that abattoir-based study be conducted at pro-

| LI M ITATI O N S
The use of serology alone can only be presumptive because other pathogens can cross-react to yield false-positive results. In our study, many of the RBT-positive sera were negative on iELISA.
Major inferences could not be made on serological data generated because of the unavailability of important variables such as the vaccination history. This is because some serological tests are unable to distinguish between antibodies produced in response to exposure to vaccination or natural exposure to Brucella spp.
Although the data generated from abattoir studies may be invaluable to surveillance (passive and active), the animals sampled may not be representative of the animal population in the province, based on the sample size used and the movement of livestock from other provinces in the country to Gauteng province where the study was conducted from other provinces in the country. A trace back study to the farm sources from where abattoir-tested seropositive cattle originated was not conducted due to the lack of cooperation by the farmers who were apprehensive of potential quarantine, with associated economic losses, due to anticipated enforcement of control measures.

| CON CLUS IONS
In conclusion, this study has provided data that B. melitensis may be circulating in the cattle population in South Africa. Brucella melitensis bv 2 and 3 have never been isolated in the cattle population in the country. This suggests a potential risk to abattoir workers and to a lesser extent to the consumers of raw or under cooked meat and meat products. The study also provided a current data on the prevalence (isolation, serology and PCR) of brucellosis in slaughter cattle in Gauteng province, South Africa. The seroprevalence of brucellosis detected in slaughter cattle in this study emphasizes the importance of using abattoirs for passive and active surveillance of diseases of public health and economic importance. Our study also confirmed that as a diagnostic strategy, it is imperative to institute more than one diagnostic method or test for the diagnosis of brucellosis in animals.

ACK N OWLED G EM ENTS
The authors acknowledge Gauteng Department of Agriculture and Rural Development for funding this project. We also acknowledge our collaborators; Agricultural Research Council-Onderstepoort Veterinary Research (ARC-OVR), Faculty of Veterinary Sciences, University of Pretoria and all abattoirs that participated in the study.

CO N FLI C T O F I NTE R E S T
There is not conflict of interest with regards to this research.

AUTH O R S' CO NTR I B UTI O N S
FK, carried out all of the experimental work and manuscript preparation. AA, FF and HvH played equal roles as the study leaders and in carrying out data analysis and manuscript preparation. BD, CK, IM, AG and AP were involved in the generation and analysis of the biochemical tests and amplicon data. All authors read and approved the final manuscript.