Introduction

Brucellosis is a bacterial zoonosis of economic and public health impact in many countries, especially in the Mediterranean area. It is caused by Brucella spp. and affects both domestic and wild animal species, besides humans. In animals, symptoms comprise abortion, infertility and reduction of milk production [1]. In humans, brucellosis manifests as a febrile, debilitating illness that, untreated, might evolve into chronic conditions (hepatomegaly, endocarditis, spondylitis, orchitis, arthritis, neurobrucellosis…) [2]. Currently, twelves species of Brucella have been identified [3], three causing proved zoonotic and economic threats: B. melitensis, B. abortus and B. suis which preferentially infect small ruminants, cattle and swine, respectively [4]. Classically, these species are subdivided in biovars: B. melitensis includes three biovars (1, 2 and 3), B. abortus eight (1, 2, 3, 4, 5, 6, 7 and 9), and B. suis five (1–5) [5][6]. The European Union has declared Belgium as Officially Free of Brucella melitensis (ObmF) in 1991 (Directive 91/68/EEC), and Officially Free of Bovine Brucellosis (OBF) (Commission Decision 2003/467/EC) in 2003. Various monitoring programmes, including serologic surveys in sheep and diagnosis of cattle abortions, are implemented to help preserve this status. Through these measures, early detection of B. abortus outbreaks in cows was achieved in 2010 and 2012 [7]. Brucella melitensis has never been isolated from animals in Belgium. However, other Brucella spp. circulate in the country: wild boars host B. suis biovar 2 [8] and marine mammals strains have been isolated on animals stranded on the Belgian coastline (unpublished data).

Human brucellosis is a mandatory notifiable disease in Belgium. Despite the Brucella-free status and the continuous monitoring, human brucellosis cases are reported yearly [9]. Applied case definition for human brucellosis is that recommended by the European Centre for Disease Prevention and Control [10]. Briefly, a confirmed case will meet both clinical (fever and one other generic symptoms of brucellosis) and laboratory criteria (isolation of the strain from clinical specimens and/or presence of Brucella specific antibodies in serum detected by agglutination tests or ELISA). In case of isolation, the identification of the bacterial species and its biovar is essential for epidemiological follow-up and control of the disease. Classical biotyping is the gold standard to investigate phenotypic characteristics [11]; this method becomes uncertain in case of appearance of mutants within the same species or biovar or the presence of strains with atypical reactions to microbiological tests [12]. Molecular genotyping methods assessing polymorphisms in the omp2A and omp2b genes [13] or by the sequencing of rpoB for example are helpful to confirm the species and the biovars of Brucella [14]. It is known that some human cases can emerge in ObmF countries with the source of contamination retrieved in infected endemic wildlife [15]. In this context, deeper molecular epidemiological tools become particularly interesting to trace-back the infection to identify the geographical source. To fulfill the objective of sub-species discrimination between strains, Variable Number Tandem Repeats (VNTR) has been investigated in Multi-Locus VNTR Analysis (MLVA) by various scientific groups since 2003 [6][16][17]. This method has proved its efficacy to obtain epidemiological information on Brucella strains, access to public databases and sharing of information between countries. In this study, MLVA was used to characterize Belgian human isolates collected during the national reference activity from 1996 to 2015.

Methods

Results

A total of 37 Brucella strains with an average of roughly two strains annually were isolated in Belgium during the reference activity between 1996 and 2015. Colony morphology, staining, growth characteristics, and slide agglutination with monospecific anti-Brucella sera were used to characterize all isolates (Table 1). These standard bacteriological procedures classified the bacterium at the species and the biovar levels. Globally, biotyping identified 28 strains as B. melitensis biovar 3, three strains as B. melitensis biovar 1, three as B. melitensis biovar 2, one as B. melitensis rough, and two as B. abortus biovar 1. The rough strain could not be typed (strain 16) and two smooth B. melitensis biovar 3 strains showed an atypical susceptibility to dyes and a delayed agglutination with anti-M serum (strain 32 and 37). Susceptibility to phages was tested in the latter strains which confirmed the identification of atypical B. melitensis biovar 3 strains.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Results of the Belgian human Brucella strains—classical biotyping.

https://doi.org/10.1371/journal.pone.0174756.t001

All isolates were further characterized by MLVA (Table 2). The 16 VNTRs were split into three panels (1, 2A and 2B) [17] because of the difference in their diversity index. Three MLVA analyses were run depending on the panels considered (MLVA8 = panel 1, MLVA11 = panel 1 + panel 2A, MLVA16 = panel 1 + panel 2A + panel 2B). A comparison of the Belgian isolates with the strains present in the public repository by MLVA8 and MLVA11 [19] allowed the identification of the associated genotypes (Table 2). Fourteen genotypes were identified with respect to the MLVA8; five were new and received numbering as “184- to 188”. These genotypes were all single-locus variants of known genotypes. Nineteen genotypes were identified with MLVA11; eight as new, numbered “361- to 368”, seven as single-locus variants and one as a two-locus variant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. MLVA16 genotype of the Belgian human Brucella strains.

https://doi.org/10.1371/journal.pone.0174756.t002

MLVA11 data were used to draw a minimum spanning tree (MST) of Brucella melitensis strains comprising the human Belgian isolates and the human strains isolated worldwide [19] (Fig 1). Three main clusters were derived, corresponding to the origin of the isolates: an East Mediterranean, a West Mediterranean and an American cluster. Belgian human strains did not cluster within a homogenous cloud, indicating the diversity and suggesting that these isolates were imported. A more detailed link with the geographical origin of some Belgian isolates could be established (Fig 1 and Table 3). The strains isolated from patients infected in Italy clustered separately from the other strains, and demonstrated genotypes described for Italian strains (genotype 51 for MLVA8 and genotypes 91 and 96 for MLVA11). The strains from patients with a history of contamination in Turkey had genotypes comparable to those described for Turkish strains in the database (mainly genotypes 42 and 43 for MLVA8 and genotypes 104, 116 and 125 for MLVA11). The only strain isolated from a patient traveling from Peru felt within the Americas cluster and showed the same genotype as Peruvian strains present in the public database. The strain of a patient traveling from Morocco had a genotype close to an African strain in the database. MLVA8 and 11 results suggested a possible origin of the infection for 12 out of 16 Belgian strains for which information about the patient or stay abroad was unknown (Table 3). For example, strains 8, 9, 10, 13, 14, 17, 19, 20, 21 and 32 had typical genotypes of Turkish strains (42 and 43 for MLVA8 and 116 and 125 for MLVA11), strain 16 demonstrated the same genotype as Italian strains (51 for MLVA8 and 96 for MLVA11) and strain 15 had a typical genotype of Chinese strains (45 for MLVA8 and 115 for MLVA11). Little information can be drawn regarding the Brucella abortus strains isolated in Belgium. One of them had the same genotypes (28 for MLVA8 and 82 for MLVA11) as hundreds of strains isolated all over the world, mainly from cattle. The other strain had a new genotype.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Minimum spanning tree obtained with MLVA-11 data of Belgian human isolates together with human strains worldwide (public databases).

Each strain is represented by a circle. Large circles regroup strain with identical genotypes dimensionally proportional to the number of entries. The lines, whose length is drawn on the basis of the number of mutations (number of variant VNTR), connect the different genotypes. The numbers aside the lines indicate the number of variant VNTRs. The grey areas connect genotypes with maximum three-locus variants. The East Mediterranean group is colored in blue, the America group in red, the West Mediterranean group in green, and the Belgian strains in yellow. Arrows describe origins for some Belgian isolates.

https://doi.org/10.1371/journal.pone.0174756.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. MLVA11 predicted geographical origin of the Belgian human Brucella strains.

https://doi.org/10.1371/journal.pone.0174756.t003

The complete MLVA16 was used to generate a dendrogram (S1 Fig) revealing that the 37 Belgian isolates clustered in 33 different genotypes. This analysis confirmed that the isolates came from multiple geographical origins but did not provide additional epidemiological information.

In certain contexts, MLVA16 can be used to cluster the strains regarding to the biovars, [6] [20]. For B. melitensis this seems to be not effective [17] [21]. Despite the use of three levels of analysis (MLVA8/MLVA11/MLVA16), it was not possible to cluster the Belgian B. melitensis strains according to their biovar. In some cases the three described biovars (bv1; bv2 and bv3) were found within the same cluster (S1 Fig).

The HDGI for the different MLVA16 loci were calculated and confirmed the high discriminatory power of MLVA (S1 Table).

Discussion

Brucellosis is a worldwide zoonotic disease that affects various domestic and wild animals but some Brucella species are occasionally able to infect humans, considered as accidental hosts. The transmission occurs by aerosol, consumption of raw dairy products or direct contact with contaminated animals. Human brucellosis is characterized by a febrile debilitating disease that can become a chronic infection accompanied with serious complications if not adequately treated. The impact of the infection on health, the mode of transmission and the zoonotic aspect of the disease have driven European countries to impose a compulsory declaration for each confirmed case to the reference centres. In Belgium, like in other OmbF and OBF countries, human brucellosis is limited to few cases every year and its incidence is low. The majority of the human cases diagnosed in Belgium are due to travel in endemic countries and to immigrations because of the absence of animal control measures in many parts of the world. Epidemiological surveillance in Brucella-free countries remains therefore essential. In Belgium, some Brucella species (B.suis, B. ceti and B. pinnipedialis) reside in wildlife and it was recently shown that clustered outbreaks in domestic animals have occurred. Brucella species are genetically close to each other [22] and, despite their host preference, some species are able to cross the host specificity barrier [23][24]. This zoonotic characteristic inevitably justifies passive surveillance of brucellosis in humans. Classical biotyping is the gold standard for species and biovar attribution of the isolates but the information lacks resolution. In some cases, appearance of mutants that generate genetic diversity is responsible for atypical susceptibility to dyes or delayed agglutination with anti-Brucella sera and thus hinders an accurate identification of the species and biovar [12]. Molecular genotyping based on single gene or multilocus sequence typing to identify locus polymorphisms or other mutations in genes have also been developed to understand the phylogeny of the genus Brucella and try to address the question of species and biovars [25][26]. These methods are not exhaustive when sub-typing for epidemiological trace-back is necessary. Since 2002, when the first complete genomic sequences of Brucella melitensis and suis [27][28] were released, a number of tools have been developed to differentiate the isolates according to DNA repetitive sequences. A first application of microsatellites fingerprinting to Brucella strains was made in 2003 [16] and was based on nine VNTRs that allowed discrimination of some isolates beyond the biovar level [29]. Afterwards, other sets of VNTR markers with a different speed of evolution were identified that demonstrated higher discriminatory power [6][30]. The group of Al Dahouk modified the set of VNTR markers of Le Fleche and showed their ability to cluster Brucella melitensis strains in 3 main geographical groups. In our study, the complete VNTR panel for Brucella was analyzed in strains isolated from humans in Belgium from 1996 to 2015 to improve their characterization and evaluate their origin. When compared with biotyping, we could assess the ability of VNTR in the identification of the biovars of Brucella.

Previous studies showed that MLVA was able to discriminate biovars of Brucella species such as B. suis and B. abortus [6]. However, it was also shown that the technique has its limitations with exceptions [31] and that it is not possible to discriminate biovars of the B. melitensis species [17]. Our study confirmed that the high genetic diversity regarding MLVA in this species did not allow clustering it according to the biovars. Therefore, classical biotyping and MLVA characterization remain complementary tools for Brucella strain surveillance. VNTRs were interesting to understand the epidemiology of the isolated strains. The study of MLVA11 was effective to establish phylogeographic relations of the Belgian strains with strains present in public databases. For instance, the single strain belonging to the Americas cluster was isolated in a patient that travelled to Peru. It shared the same genotype as 62 strains coming from Peru described by various groups [32][33]. Six isolates displayed the exact MLVA11 genotypes of dozens of strains coming from Italy described previously [6][12][17]. While the Italian origin for 5 of these Belgian isolates was known, thanks to the databases, it is suspected that the 6^th patient also contracted the disease in this country. These 6 strains belonged to the West Mediterranean cluster. A last isolate belonging to this cluster was a new genotype based on a single locus variant with Italian strains and was isolated from a patient with a travel history in Spain. The East Mediterranean cluster contained the majority of the strains isolated in Belgium. Nine of them shared the same MLVA11 genotype (genotype 43) with dozens of strains coming from Turkey as previously described [17] [34]. Few strains with the same MLVA11 genotype come from Syria and Irak [17]; this genotype was also observed in strains coming from Lebanon [35]. Seven other strains isolated in Belgium shared the same genotype with a great panel of strains isolated in China [21] and Turkey [17][34] (genotype 42). Information about the genotypes present in countries geographically located between Turkey and China is missing but this genotype 42 seems to be highly represented in the Middle East [17][35] and has already been isolated in countries like Kazakhstan [17][36]. We hypothesize that this particularly stable genotype [21] is largely spread from Turkey to China and that MLVA11 is not variable enough to distinguish strains from this region of the world. Our hypothesis is sustained by an MST analysis performed on 446 B. melitensis MLVA15 profiles of strains belonging to the East Mediterranean group [37]. In this study, part of the Turkish genotypes clustered with most of the genotypes observed in China and Kazakhstan, suggesting a common origin between strains from the Asian continent and the East Mediterranean area [37].

When examined with MLVA16 which includes panel 2B composed of five unstable satellites with a high mutation rate, Belgian isolates were found highly diverse; 33 genotypes were distinguished among the 37 isolated strains. They apparently came from all around the Mediterranean region except one strain with a genotype clustering together with strains from Peru. MLVA16 did not add to the information obtained already by MLVA11 that strains isolated in Belgium were coming from a large panel of areas in the world. It seems that increasing the number of markers is particularly useful in case of outbreak investigation in endemic countries. For highly genetic diverse strains such as in the context of imported cases, MLVA11 is in general sufficient to trace the geographical origin of infection. Accessibility of MLVA16 information in public repositories about human strains isolated in Belgium (and worldwide) could help endemic countries to better investigate a possible link and the spread with their cases.

Classical biotyping revealed that the human cases in Belgium are mainly Brucella melitensis, a species never isolated in animals in Belgium. This is in agreement with the fact that B. melitensis is much more common than B.abortus in many countries of Eastern Europe and the Middle East [38]. More precisely, most of the strains belonged to the biovar 3. This is consistent with the fact that the majority of patients travelled to or originated from Mediterranean countries and the Middle East where this biovar is predominant [11]. Since 1996 only two human isolates in Belgium revealed to be B. abortus, both belonging to biovar 1. One of the two patients was coming from Ecuador. To our knowledge, B. abortus is the only species isolated in humans in this country [39]. We do not have any information about the second patient but it is worthwhile to note that Brucella abortus biovar 1 was not enzootic in Belgium during the years 1980–2000. It has only been isolated in the late nineties from cattle imported from Portugal where this biovar was prevalent. Taken together, this information convincingly indicate that the majority of human brucellosis cases in Belgium is associated with travel to Brucella endemic countries as observed in neighbouring countries [40].

Our study is the first characterization of human Brucella isolates in Belgium. It highlights the genetic diversity of the strains consistent with the imported nature of the infections and the use of MLVA as an important tool to improve the understanding of the phylogeography of Brucella.

Supporting information

Acknowledgments

The authors thank Dr. Imberechts H. for critical review of the manuscript and Dr. Melzer F. for scientific assistance. We acknowledge Dr. Muyldermans G., Dr. Lernout T., Prof. Van den Berg T., Dr. Knapen K. and Dr. Mertens V. for their support in the NRC activity. Gratitude is also due the laboratory staff of the Bacterial diseases in livestock unit of CODA-CERVA, Brussels, for their work in the complementary serological activity of the NRC, and the dispatch service of CODA-CERVA. This work was possible also thanks to the collaborative attitude of the first line clinical laboratories.