Isolation and Molecular Characterization of Brucella Isolates in Cattle Milk in Uganda

Brucellosis is endemic in livestock and humans in Uganda and its transmission involves a multitude of risk factors like consumption of milk from infected cattle. To shed new light on the epidemiology of brucellosis in Uganda the present study used phenotypic and molecular approaches to delineate the Brucella species, biovars, and genotypes shed in cattle milk. Brucella abortus without a biovar designation was isolated from eleven out of 207 milk samples from cattle in Uganda. These isolates had a genomic monomorphism at 16 variable number tandem repeat (VNTR) loci and showed in turn high levels of genetic variation when compared with other African strains or other B. abortus biovars from other parts of the world. This study further highlights the usefulness of MLVA as an epidemiological tool for investigation of Brucella infections.


Introduction
The genus Brucella has ten recognized species with more than 90% DNA homology [1,2]. These species cause brucellosis that is of economic and public health importance in terrestrial and aquatic animals and humans [1,3]. Species of B. melitensis, B. abortus, some B. suis biovars, B. canis, B. ceti, and B. inopinata are zoonotic and in humans the infection causes a debilitating disease with relapsing fever and flu-like symptoms with multiple organ involvement [4][5][6]. In cattle Brucella causes abortions, placentitis, orchitis, mastitis, and prenatal death [5,7].
Brucellosis in cattle is almost exclusively caused by B. abortus [8], but B. melitensis and B. suis have been implicated in some herds [9,10] making the vaccination of cattle using vaccines targeting only B. abortus less effective in preventing brucellosis in cattle and transmission to humans [11,12]. Brucella biovars and genotypes are known to be regionally restricted in their distribution [13] but the evolution of international travel and trade and changing ecosystems have led to introduction of new biovars and genotypes into regions and hosts where they were not previously found [3]. A study done in Uganda in 1958 isolated B. abortus biovar 3 from a human patient [14] and in the neighboring Kenya B. melitensis biovar 1 and B. abortus biovar 3 have been isolated from cattle [9]. Phylogenetic analysis of B. melitensis isolates in Kenya showed a high degree of homology with isolates in Israel and the B. abortus isolates closely resembled that isolated in Uganda. In a related study in Egypt, B. abortus, B. suis, and B. melitensis were isolated from cattle and all had a high level of phylogenetic variability within each species although the isolates used in these studies were few [15]. The above findings indicate a complex epidemiology of brucellosis in cattle in the region and call for refined diagnostic methods beyond phenotypic typing.
High resolution phenotypic and molecular approaches have been developed for Brucella speciation, biotyping, and epidemiological trace-back [16,17]. To date, advanced molecular technologies have not been widely used in low income countries where brucellosis is endemic in livestock and humans [7,18]. Thus, information on the prevailing Brucella species, biovars, and genotypes/strains in such areas of endemicity may shed new light on the epidemiology of Brucella infection and the species and biovars circulating. Besides this generic scientific rationale for undertaking such investigations, increased understanding of the Brucella epidemiology is critical for refining control of brucellosis in resource weak countries where the same measures as in high income countries cannot be applied.
In northern and eastern Uganda where this study was performed, there was a considerable mixing of livestock species during years of insurgency in the 1990s, presenting ideal conditions for inter-and intraherd transmission of diseases such as brucellosis. Indeed, high herd and individual animal seroprevalences of up to 27% and 7.5%, respectively, were recorded in a recent survey in the region [19]. This high prevalence may pose a severe threat to public health as previous studies around the capital of Uganda suggest milk or milk products from cows as a major source of Brucella infection in humans [20,21]. The present study aimed at isolating and molecular-typing Brucella from cattle milk in northern and eastern Uganda for better understanding of its epidemiology.

Study Design and Collection of Samples.
Milk samples were collected from 207 lactating cows in urban and periurban areas of Gulu and Soroti towns of northern and eastern Uganda from May 2011 to March 2012 for isolation of Brucella. A total of 110 individual cow milk samples were collected from 72 herds in Gulu and a total of 97 individual cow milk samples were collected from 33 herds in Soroti. These herds were part of the 166 herds whose animals had been screened for brucellosis and were within a radius of 15 Km in both Gulu and Soroti towns, with the two towns being 200 Km apart. The cattle from which milk samples were taken had been screened for Brucella antibodies and in total 17 of the 207 cows were seropositive. In both towns, the seropositive herds from which the milk samples were taken were near each other. The numbers and selection of households and animals included are described in detail previously [19]. Midstream milk samples were collected from all quarters with 10-20 mL collected from each teat into sterile 100 mL falcon tubes. The samples were transported chilled to Makerere University College of Health Sciences microbiology laboratory, Kampala, Uganda, kept at 4 ∘ C, and cultured within three days.

Sample Preparation, Brucella Culturing, and Biotyping.
The milk was centrifuged at 3000 ×g at 4 ∘ C for 15 minutes and the pellet and supernatant were plated on both Farrel and Centro de Investigación y Tecnología Agroalimentaria (CITA) selective media supplemented with calf serum [22]. Briefly, Farrel's medium was prepared from Brucella medium base (Oxoid, UK), sterilized at 121 ∘ C for 15 minutes, and supplemented with Brucella selective supplement (Oxoid, UK) according to the manufacturer's instructions. The CITA medium was prepared according to De Miguel et al. [22]. Inoculated plates were incubated at 37 ∘ C for 8 days in a 5-10% carbon-dioxide incubator and read every 24 hours from day three of incubation for colony growth. Resultant colonies were subcultured and biotyped based on their colony morphology, serum and carbon-dioxide requirement for growth, hydrogen sulphide production, urease activity, oxidase test, and growth in presence of dye basic fuchsin and agglutination of anti-Brucella IgG monospecific sera A (Animal Health and Veterinary Laboratories Agencies, Weybridge, UK), according to the OIE Terrestrial Manual [8]. Representative colonies are stored at −80 ∘ C in 20% glycerol for long-term storage.

Genomic DNA Extraction and Real-Time PCR Detection.
The colonies that conformed to all the above phenotypic characteristics of Brucella were subjected to genomic DNA extraction using the Norgen bacterial genomic DNA isolation kit (Norgen Biotek Corp., Ontario, Canada). The extracted Brucella genomic DNA was used to run a real-time multiplex PCR assay with oligonucleotide primers, probes, reaction mixture, and PCR conditions according to Probert et al. [23]. Amplification and real-time fluorescence detection were done on the Rotor-Gene 3000 real-time PCR machine (Corbett Research-Corbett Life Sciences, Mort Lake Australia). Three positive and two negative controls were included in each run. When the cycle threshold (CT) value of the samples was ≤40, samples were evaluated as positive. This real-time multiplex PCR was designed to detect Brucella at both the genus and species levels for B. abortus and B. melitensis since it has the genus specific probe and the species specific probes for the two Brucella species commonly infecting cattle.

Brucella Species Confirmation.
Positive samples on realtime PCR were analysed further for Brucella speciation using the Bruce-ladder multiplex PCR assay kit (Ingenasa, Spain). The oligonucleotide primers, reaction mixture, and PCR conditions were performed according to the manufacture's conditions in conformity with López-Goñi et al. [16]. However amplification was done in a MyCycler thermal cycler (BioRad).  Figure 1). However the triplex realtime PCR was unable to detect the Brucella species involved using its B. melitensis probe ( Figure 2) and B. abortus probe ( Figure 3).

Discussion
Here we present for the first time phenotypic and molecular characterization of Brucella isolates from cattle milk in Uganda. These results contribute to better understanding of geographical transmission patterns of Brucella in cattle in Uganda and are important if specific control measures are to be implemented in the future. In Uganda, most of the milk is marketed unprocessed through the informal milk marketing linkages, thus acting as a potential source of human brucellosis infections. All UG Ba-m isolates were from Brucella seropositive cattle conforming to the well-known fact that Brucella infected lactating female animals shed the bacteria in their milk since the organism relocates to the udder from the pregnant uterus upon delivery [27]. This has public health implications in the region since most of the milk is consumed unpasteurized. One third of the seropositive cattle were not shedding the Brucella in the milk suggesting that these animals either had cleared the infection or were chronically sick, thus not shedding the bacteria as shown in a study by Capparelli et al. [28]. All UG Ba-m isolates being from seropositive cattle suggest an active infection. Notably, Brucella was not isolated from milk from any of the seronegative cows. This suggests that seroconversion precedes shedding of the Brucella in milk, and thus serological tests can be sufficient in predicting possible shedders.
The phenotypic characteristics of all UG Ba-m isolates matched those of B. abortus biovars 1, 3 or the former biovar 7. All isolates being B. abortus conform to the wellknown fact that B. abortus is the predominant species in cattle [14]. Furthermore, all 11 UG Ba-m isolates having the same phenotypic profile suggest that they belong to the same biovar attesting to the suggestion of regional predominance of certain Brucella biovars in Africa, for instance, predominance of B. abortus biovar 6 in nomadic cattle in Western Sudan [29], B. abortus biovar 3 and B. melitensis biovar 1 in cattle in Kenya [9], and B. melitensis biovar 3 in ruminants in Egypt [15]. However the numbers of isolates in these studies were too few to make a solid basis for generalization.
Detection of all the 11 UG Ba-m isolates as Brucella by the Brucella genus specific probe in the triplex real-time PCR with strong signals confirmed them as Brucella. The inability of the B. melitensis probe in the triplex real-time PCR to detect the isolates as B. melitensis suggested that they are not B. melitensis. The inability of the B. abortus species specific probe in the triplex real-time PCR to detect the isolates as B. abortus suggested that they are not B. abortus biovars 1, 2, 3, 4, 5, 6, and 9 but could be the former B. abortus biovar 7 since the primers used were not targeting B. abortus biovar 7 [23,30].
All the 11 UG Ba-m isolates were confirmed as B. abortus by the Bruce-ladder multiplex PCR. The inability of the triplex real-time PCR to detect these isolates at its B. abortus species specific probe contrary to the Bruce-ladder multiplex PCR suggests that these isolates belong to the former Brucella abortus biovar 7. This suggestion is based on the fact that the triplex real-time PCR was not designed to detect the former Brucella abortus biovar 7 with its B. abortus species specific probe (detecting only B. abortus biovars 1, 2, 3, 4, 5, 6, and 9), and the Bruce-ladder multiplex PCR was designed to detect all the B. abortus biovars including the former B. abortus biovar 7 [16,23,30]. This suggests that a diphasic PCR protocol involving the triplex real-time PCR by Probert et al. [23] and the Bruce-ladder multiplex PCR by López-Goñi et al. [16] could be used to replace the risky procedure of identifying the former B. abortus biovar 7 using the conventional biotyping methods by detecting particularly its agglutination with monospecific anti-sera A and M and its growth in both basic fuchsin and thionin dyes. This protocol needs, however, to be tested on all the former B. abortus biovar 7 isolates.
The evidence adduced using a combination of phenotypic and molecular approaches designated all UG Ba-m isolates as atypical B. abortus without a biovar designation.    isolates were monomorphic at molecular analysis, which could be due to the isolates being from a small geographical region of approximately 15 Km radius per region (data not shown here) and having been collected in a short time frame making it possible for all the isolates to be from a common source as animals mix in the grazing grounds. The genetic monomorphism observed is partly congruent with that observed in five B. melitensis biovar 1 isolates obtained from bovine milk in central Kenya by Muendo et al. [9], although their finding was in a different Brucella species. Our results are further supported by findings by Garin-Bastuji et al. [31] who found similar monomorphism in isolates in Mongolia isolated 5 years apart in the same region. The genetic monomorphism exhibited at the minisatellite and microsatellite loci that are otherwise polymorphic even in highly genetic homogenous species like Brucella suggests that there is one or very few circulating strains of Brucella in this region of Uganda attesting to the regional predominance of Brucella biovars and strains.
The closest known strain for the 11 UG Ba-m isolates was a B. abortus strain 07-994-2411 isolated from cattle in the neighboring Kenya in 1963. This strain has no biovar designation but was formerly known as B. abortus biovar 7, before biovar 7 was suspended from the Brucella nomenclature (International Committee on Systematic Bacteriology, 1988). B. abortus biovar 7 was suspended from Brucella nomenclature because the reference strain (63/75) was thought to be a mixture of B. abortus biovars 3 and 5. A recent study by Garin-Bastuji et al. [31] proposed the reintroduction of B. abortus biovar 7 in the approved list of bacterial names having identified B. abortus strains from Turkey, Mongolia, and Kenya that perfectly matched the former B. abortus biovar 7 characteristics. B. abortus biovar 7 can be differentiated from other B. abortus biovars by its ability to agglutinate with anti-A and anti-B monospecific sera. B. abortus biovar 7 has smooth colonies, does not require carbon dioxide for growth, produces hydrogen sulphide, is oxidase and urease positive, does not agglutinate with monospecific anti-sera R, grows in the presence of dyes thionin and basic fuchsin at a concentration of 20 g/mL, and is lysed by phages Tbilisi (Tb), Weybridge (Wb), Izatnagar 1 (LZ 1 ), and R/C. A genetic difference in the UG Ba-m isolates at 5 loci out of 16 polymorphic loci examined compared to the known closest related strain (07-994-2411) from Kenya in a period of half a century could be a result of mutations and proves the ability of MLVA to differentiate strains from different localities, a finding congruent with that of Verger et al. [2]. The observed genetic similarity between the Kenyan strain and the 11 UG Ba-m isolates compared to other isolates from distant places could be due to the cross-border transmission of Brucella in cattle that could have been facilitated by cattle rustling across the pastoral Karamoja subregion of Uganda and Kenya over the past years.

Conclusions
In conclusion, our findings suggest B. abortus without biovar designation (atypical B. abortus) as a cause of brucellosis in cattle in northern and eastern Uganda. The Ugandan isolates exhibited a single MLVA-16 pattern and show in turn high levels of genetic variation when compared with other African strains, highlighting the usefulness of MLVA as an epidemiological tool for investigation of Brucella infections. Furthermore, the ability of a diphasic PCR protocol involving the triplex real-time PCR by Probert et al. [23] and the Bruce-ladder multiplex PCR by López-Goñi et al. [16] to detect B. abortus could be used to replace the procedure of identifying the former B. abortus biovar 7 using the conventional biotyping methods.

Ethical Approval
This study entailed collection of milk samples from farmers' cattle. Ethical clearance was obtained from the Ethical Review Committee of the College of Veterinary Medicine, Animal Resources and Biosecurity, Makerere University. The farmers were informed of the study and their verbal consent was sought prior to commencement of data collection.