Infection, Genetics and Evolution 23 (2014) 13–19

Contents lists available at ScienceDirect

Infection, Genetics and Evolution

journal homepage: www.elsevier .com/locate /meegid

Multi-locus variable-number tandem repeat analysis (MLVA) revealsheterogeneity of Mycobacterium bovis strains and multiple genotypeinfections of cattle in Ethiopia

http://dx.doi.org/10.1016/j.meegid.2014.01.0211567-1348/� 2014 Elsevier B.V. All rights reserved.

⇑ Corresponding author at: College of Medicine, University of Arizona, 1656,E Mabel St, P.O. Box 245221, Tucson 85719, AZ, USA. Tel.: +1 520 626 6409;fax: +1 520 626 2100.

E-mail address: demelash.biffa@gmail.com (D. Biffa).

Demelash Biffa a,b,⇑, Tone Bjordal Johansen c, Jacques Godfroid d, Adrian Muwonge e, Eystein Skjerve a,Berit Djønne c

a Center for Epidemiology and Biostatistics, Norwegian School of Veterinary Science, P.O. Box 8146, N-0033 Oslo, Norwayb College of Medicine, University of Arizona, 1656, E Mabel St, P.O. Box 245221, Tucson 85719, AZ, USAc Department of Animal Health, National Veterinary Institute, P.O. Box 750 dep, N-0106 Oslo, Norwayd Section for Arctic Veterinary Medicine, Norwegian School of Veterinary Science, Stakkevollveien 9010, Tromsø, Norwaye The Roslin Institute, University of Edinburgh Easter Bush Roslin, Midlothian EH25 9RG, United Kingdom

a r t i c l e i n f o

Article history:Received 5 April 2013Received in revised form 16 January 2014Accepted 20 January 2014Available online 27 January 2014

Keywords:CattleEthiopiaHeterogeneityMLVAMycobacterium bovis

a b s t r a c t

Bovine tuberculosis (BTB) remains a major threat to animal and human health, and obstructsinternational and inter-regional livestock trade in Ethiopia. Many aspects of epidemiology of BTB andits causative agent, Mycobacterium bovis (M. bovis) are not well known. Aims of the study were toelucidate molecular characteristics of M. bovis strains using MLVA typing method. Further aim was todetermine infection pressure associated with occurrence of multiple genotypes in individual infectedcattle. Data and samples were collected in the period July 2006–January 2007 in cattle slaughtered at fiverepresentative abattoirs across the country. Molecular investigation of the isolates was carried out usingmultilocus variable-number tandem-repeat analysis (MLVA) of 28 variable numbers of tandem repeat(VNTR) loci, and the results were compared to spoligotyping. This study is believed to contribute tothe knowledge of molecular genetics and epidemiology of M. bovis in Ethiopia and elsewhere with similarBTB epidemic situation and livestock production settings. Four-hundred and six tissue samples from 337carcasses revealing gross pathologic lesions compatible with BTB were collected from five abattoirs.Fifty-eight isolates obtained from cultured samples were subjected to region of difference (RD) analysisand MLVA typing. RD confirmed all isolates as being M. bovis. MLVA revealed a high heterogeneity ofM. bovis (19 genotypes) and the discriminatory power of MLVA was higher than for spoligotyping(Hunter–Gaston Diversity Index (HGDI) 0.92 vs. 0.82). Adoption of the nine VNTR loci with P3 allelesprovided good differentiation between the isolates. However, differentiation was optimized when MLVAwas combined with spoligotyping (HGDI = 0.99). MLVA confirmed infections with multiple genotypes in38.5% (10/26) of individual animals. In conclusion, the usefulness of MLVA for genotyping M. bovis in highprevalence settings was demonstrated. BTB in Ethiopia is caused by heterogeneous populations ofM. bovis and individual carcasses were often infected with different genotypes, indicating a high infectionpressure perhaps related to the absence of protective immunity conferred by infection.

� 2014 Elsevier B.V. All rights reserved.

1. Introduction

Bovine tuberculosis (BTB), caused by Mycobacterium bovis, is adisease of high economic importance causing serious losses to live-stock industries. It remains a major public health problem in mostAfrican countries where animals and humans live in close proxim-

ity and milk hygiene and meat inspection are not well practiced.The situation has been exacerbated by added burden of HIV/AIDSepidemic (Maho et al., 1999).

Progress made so far towards eradication of BTB has been ham-pered by a lack of precise epidemiological data (Michel et al.,2007a). The availability of such information is critically importantin the effort to control the disease. In Ethiopia, BTB is widely dis-tributed with its existence reported in almost all agro-ecologiczones. It poses a great public health threat owning to the widelyprevailing risk factors linked to socio-cultural practice such as

14 D. Biffa et al. / Infection, Genetics and Evolution 23 (2014) 13–19

consumption of raw milk and meat and living in same house withinfected animals.

With its diverse climate and topographic characteristics, thebreed composition and size of the national herd, Ethiopia is a majorrepository of livestock resources and genetic diversity (Cousinset al., 2004). Livestock are managed under three broad productionsystems namely; subsistence crop-livestock mixed production(small scale); large scale commercial production and transhu-mance mode of production. In most instances, husbandry systemsfollow traditional practices where animals are allowed to grazenatural pastures with no/little extra feed supplements and veteri-nary care. There is no regulation and policy reinforcement regard-ing movement control and this has led to a high degree of stockintermingling and subsequent disease transmission.

Multilocus variable-number tandem-repeat analysis (MLVA)has recently emerged as a genotyping method that is both robustand highly discriminatory for differentiation of isolates in theMycobacterium tuberculosis complex (MTC) (Cosivi et al., 1998a)and is a powerful complement to the existing traditional tools forepidemiological investigation of M. tuberculosis (Corner, 1994;Ayele et al., 2004; Araujo et al., 2005a; Oloya et al., 2007).Themethod is based on PCR analysis of tandem repeated sequencesin the bacterial genome, where specific primers target the flankingsequences of these variable number of tandem repeat (VNTR) re-gions (Diguimbaye-Djaibe et al., 2006; Food Safety Authority of Ire-land (FSAI), 2008). Variations in the number of repeats in each ofthe VNTR loci generate PCR amplicons of varying fragment length,and this size polymorphism forms the basis for the generation ofthe various allele (Food Safety Authority of Ireland (FSAI), 2008).Different research groups have used different combinations ofVNTRs, and published them under the names mycobacterial inter-spersed repetitive units (MIRUs), Queens University of Belfast(QUB), Exact Tandem Repeats (ETRs) and Mtub (M. tuberculosis)which are used as standard nomenclature and reflect agreementsin most of the published works (Corner, 1994; Diguimbaye-Djaibeet al., 2006; Oloya et al., 2007)

Much of the work of finding loci with tandem repeatedstretches of DNA was done in studies analyzing M. tuberculosis(Uitenbroek, 1997). MLVA also has been show to have great poten-tial as a molecular tool for epidemiological investigation of M. bovis(Cosivi et al., 1998a). Usefulness of VNTR loci for typing M. bovismay vary from region to region and appropriate loci need to be sys-tematically chosen for the target region (Cosivi et al., 1998b). Fur-thermore, it is important to expand the number of repeat loci, andto find new combination of markers, in order to achieve high dis-crimination among M. bovis isolates.

The main aim of this study was therefore to explore the useful-ness of MLVA for revealing the molecular and epidemiologicalcharacteristics of M. bovis strains isolated from TB-like lesions inEthiopia. In order to find the best strategy for typing of M. bovis iso-lates from Ethiopia, the discriminatory power of sets of loci usedindividually and in combination with each other as well as withspoligotyping was calculated. Further aim was to determine inten-sity of infection as explained by simultaneous occurrence of multi-ple genotypes in individual infected animals.

2. Materials and Methods

2.1. Study animals

Three breeds of cattle were included in the study. These wereindigenous local zebus, crossbreds (Local zebu x Holstein) and Hol-stein cattle. The zebu cattle were managed under traditional hus-bandry system (especially in pastoral/agro-pastoral areas) thatdepends entirely on natural pastures without extra feed supple-

ments and adequate health services. Pure and crossbred animalsoften belong to the state and private dairy farmers concentratingin central regions of the country where accessibility to commer-cially produced feed, veterinary and artificial insemination servicesare within their reach. They are only salvaged or considered forslaughter, following an age-related decline in milk production orreproductive disorders. Although animals originated from widergeographic regions across the country, for convenience six regionswere identified (based on agro-ecological or farming system) andthese included; South, Southwest, Southeast, East, Addis Ababaarea, and the Central region.

2.2. Collection of samples and isolation of mycobacteria

Details on collection and processing of data and samples, aswell as culturing and identification of mycobacteria and spoligo-typing of the isolates have been previously described (Araujoet al., 2005b). In the study abattoirs (Addis Ababa, Adama,Melge-Wondo, Hawassa and Yabello), a total of 406 samples from337 carcasses were found to reveal gross pathologic lesions com-patible with BTB. All samples were collected and transported toBSL3 laboratory (Oslo, Norway) for detailed mycobacteriologicalanalysis. Field work of the study was carried out in the period July2007–January 2008.

Preliminary isolation of mycobacteria was carried out by cultur-ing each sample separately on egg-based Lowenstein-Jensen (with-out pyruvate), Stonebrinks Medium (with glycerol) and agar-basedMiddlebrook 7H10 following the standard procedure described byOIE (Michel et al., 2007b). A colony from growth with positivemicroscopy (acid fast bacilli) was re-cultured on same sets of med-ia so as to harvest pure culture. Characteristic growth patterns,morphology and colour of the colonies together with positive re-sults for acid fast bacilli based on Ziehl–Neelsen technique wereused for presumptive identification of mycobacteria (Michelet al., 2007b).

Cattle breeds included in the study were Ethiopian indigenouszebus (n = 285); cross (Local x Holstein cattle) (n = 22) and pureexotic cattle (Holstein) (n = 30). Along with sample collection,information linked to geographic origin of animals and organ af-fected were recorded for each individual carcass.

2.3. Identification of M. bovis by genomic region of difference (RD)

All isolates belonging to the MTC were identified by RD deletionanalysis as described by Warren et al. (Collins, 1983). This is a two-step, multiplex PCR reaction method based on genomic regions ofdifference (RD1, RD1mic, RD2seal, RD4, RD9 and RD12) for differen-tiation of members of M. tuberculosis Complex (MTC). The size ofthe respective multiplex PCR amplification products correspondto the presence of the difference M. tuberculosis members (Collins,1983).

2.4. Test for region of difference-Africa-1(RDAf1) deletion

The isolates were tested for presence or absence of RDAf1 as de-scribed by Müller et al. (2009) using three primers; two primerstargeting the flanking regions of RDAf1 (Mb0586c FW andMb0590c Rev) and the third primer hybridizing with the internalregion of RDAf1 (Mb058xc IntRev). M. bovis AF2122/97 (RDAf1present) and M. bovis CHAD491 (RDAf1 deleted), supplied by Vet-erinary Laboratories Agency-Weybridge, UK, were used as controls.Results were interpreted based on the observation that isolateswith Af1 region deleted should generate a PCR product of 531 bpwith primers Mb0586c FW and Mb0590c Rev, while those possess-ing intact Af1 region produce a 349 bp product with primersMb0586c FW and Mb058xc IntRev.

D. Biffa et al. / Infection, Genetics and Evolution 23 (2014) 13–19 15

2.5. Multilocus variable-number tandem-repeat analysis (MLVA)

All M. bovis isolates were analyzed by MLVA typing at Geno-screen, Lille, France. Isolates were heat treated at 95 �C in 20 min,maintained in 400 ll TE-buffer (10 mM Tris-HCl, 1 mM EDTA pH8.0) and transferred to Genoscreen. The isolates were genotypedby PCR amplification of 28 known VNTR markers that includedthe 12 MIRU loci, 3 ETRs loci, 7 QUB loci and 6 additional Mtub loci.Conveniently, all the loci included in the study were collectivelydesignated as Variable Number of Tandem Repeats (VNTRs). Selec-tion of the 28 VNTR markers was made to enhance the discrimina-tory power of the technique and subsequent generation of newinformation regarding genetic diversity of the isolates.

2.6. Data analysis and determination of diversity index

The dataset from the MLVA analysis was validated for the pres-ence of missing, double or migratory alleles as well as non-amplified markers. The dataset was stored in an Excel� file andexported into Bionumerics software package version 4 (AppliedMaths, St.Martens Latem, Belgium) as a character data for analysis.Dendrograms were constructed using the categorical parameterand the unweighted pair-group method using average linkagesalgorithm (UPGMA). A cluster was conveniently defined as a groupof isolates with P 90% genetic similarities. In order to detect anylink between specific clusters and the recorded variables, informa-tion linked to geographic origin of the animal, body system af-fected, husbandry system under which animals were managedand abattoir as well as spoligotype of the different isolates wereincorporated into the dataset and the subsequent cluster analysis.

The discriminatory power of the MLVA method for the wholeset of VNTR loci, of individual loci, combinations of different loci,and combinations of VNTR loci and spoligotyping was establishedby calculating the Hunter–Gaston Diversity Index (HGDI) (Delafos-se et al., 2002), using the VNTR Diversity Confidence Extractorsoftware (V-DICE), developed by Health Protection Agency (HPA)-Bioinformatics, London (http://www.hpa-bioinformatics.org.uk/cgi-bin/DICI/DICI.pl). Likewise, allelic diversity within a locus, ameasure of the variation of the number of repeats at each locus,was computed based on HGDI using a similar approach. V-DICEallows simultaneous calculation of HGDI (with its 95% confidenceinterval), the number of different repeats present at a locus andthe proportion of samples that have the most frequent repeatnumber in each locus. HGDI computation was based on theformula:

1� 1n½n� 1�

Xs

j¼1

x2j ½n=ðn� 1Þ�;

where s represents the number of alleles present, xj is the frequencyof the ith allele at the locus; n is the number of strains and n/n-1 is acorrection factor for bias in small samples. Allelic/locus diversity in-dex can take any number between zero (no diversity) to one (ex-treme diversity); loci with a similar number of repeats in eachsample will have a low diversity, whereas a locus where the numberof repeats is different for nearly every sample will have a high diver-sity index. According to Hilty et al. (2005), HGDI of > 0.25 is consid-ered highly discriminative, 0.11 < HGDI < 0.25 moderately and0.01 < HGDI < 0.11 poorly discriminative.

3. Results

3.1. Deletion analysis and spoligotyping

Fifty-eight isolates, cultured from cattle carcasses revealinggross pathologic lesions characteristic of BTB, were examined by

deletion analysis targeting the RD1, RD4, RD9 and RD12. Allyielded the banding pattern typical of M. bovis. None of the isolates,regardless of the presence or absence of spacer 30 in their spolig-opatterns, were found to have the RDAf1 deletion (data notshown). Spoligotyping performed earlier on 48 of the isolates(Araujo et al., 2005b) together with spoligotyping results of 10additional isolates revealed 17 distinct spoligopatterns.

3.2. Multilocus variable number of tandem repeat analysis (MLVA)

MLVA was carried out on 58 isolates. Six isolates did not pro-duce amplification of the markers after two independent PCRs,and were thus excluded from the analysis. Two isolates from car-cass 332 and two from 321 were considered identical, as theyshowed identical MLVA and spoligo-profiles. One isolate from eachcarcass was therefore excluded from analysis, making the totalnumber of isolates 50.

The MLVA profiles of all isolates, together with carcass identifi-cation number, geographic origin, organ of isolation and spoligo-type, are illustrated in Fig. 1. Based on the categorical parameterand UPGMA, 19 different genotypes named MLVA 1–19 were dif-ferentiated out of the 50 M. bovis isolates (Table 1). Based on 90%genetic similarity, the variants were grouped into eight clusters(C1–C8), with size ranging from including four different profiles(C1 and C7) to only one single profile (C2, C4 and C5) in each cluster(Fig. 1).

Table 4 shows the occurrence of multiple genotypes of M. bovisin pathologic tissue samples of individual cattle carcasses. Ten outof 26 (38.5%) carcasses where M. bovis isolates were found har-boured isolates with different MLVA genotypes. Two carcasses(422 and 321) were infected with 6 distinct combinations of MLVAprofile and spoligotype each. In addition to respiratory and othervital organs, M. bovis was also isolated from the gastrointestinaltract of two carcasses (422 and 645). Except carcass 3224, whoseorigin was linked to the Southeast region, all carcasses harbouringmultiple genotypes originated from the central region or AddisAbaba and its surroundings. M. bovis isolated from carcass 332and 328 were exhibiting four distinct MLVA profiles each, althoughthey belonged to two spoligotypes.

Fig. 2 shows the MLVA profiles of strains belonging to the twomost frequent spoligotypes (SB1176 and SB0133). Twenty isolateswith identical spoligopatterns (SB1176) were differentiated into 12distinct MLVA profiles. Likewise, five MLVA variants were differen-tiated from seven isolates with spoligopattern SB0133.

Different MLVA genotypes were found among the M. bovis iso-lates from the four major cattle management systems in Ethiopia(Fig. 3). The strains isolated from dairy farm holdings (n = 25) weredifferentiated into 13 distinct genotypes, compared to five (sevenisolates), six (eight isolates) and seven (ten isolates) MLVA geno-types from transhumance, small-scale and large-scale farms,respectively (Fig. 3).

4. Discussion

With exception of one study from a Government dairy farm inAddis Ababa area that investigated M. bovis isolates using ETR loci(Central Statistical Authority (CSA), 2006), this is the first studyinvestigating Ethiopian M. bovis isolates using MLVA. One of thelimitations of this study is the relatively small number of M. bovisisolates (n = 50) used to define region-specific optimal panel ofVNTRs. Furthermore, the study takes no account of the evolution-ary lineage relationships between MLVA genotypes. The 50M. bovis isolates were differentiated into 19 distinct genotypes(HGDI = 0.92) that formed 8 clusters at P90% genetic relationship.MLVA 2 was the most dominant genotype representing 20.0% of

Fig. 1. Dendrogram showing genetic relationships of 50 M. bovis isolates from Ethiopian cattle infected with BTB. The analysis was based on MLVA of 28 VNTR loci. Similarityof MLVA profiles was calculated using categorical coefficient, while dendrogram was generated by unweighted pair-group method with arithmetic averages algorithm(UPGMA). Shades are for visual differentiation of different alleles.

16 D. Biffa et al. / Infection, Genetics and Evolution 23 (2014) 13–19

the isolates, all obtained from Addis Ababa and the central region,areas with a high prevalence of BTB(MoH, 2002; Central StatisticalAuthority (CSA), 2006).

None of the current Ethiopian isolates belonged to the Af1 clo-nal complex of M. bovis, as they did not contain the specific RDAf1deletion. This complex is a group of M. bovis strains dominant in

the West African countries of Mali, Nigeria, Cameroon and Chadand is characterized by a specific chromosomal deletion (RDAf1)and the absence of spacer 30 in their spoligopattern (Uitenbroek,1997). This consolidates the findings that the RDAf1 deletion is ab-sent from the M. bovis populations of the East African countriesincluding Ethiopia, Uganda and Tanzania, something that can be

Fig. 2. Differentiation by MLVA of the two dominant spoligotypes of Ethiopian M. bovis isolates; SB1176 (A) and SB0133 (B). The analysis was based on 28 VNTR loci.Similarity of MLVA profiles was calculated using categorical coefficient, while dendrogram was generated by unweighted pair-group method with arithmetic averagesalgorithm (UPGMA). Shades are for visual differentiation of different alleles.

D. Biffa et al. / Infection, Genetics and Evolution 23 (2014) 13–19 17

explained by little livestock trade across these regions of the con-tinent (Uitenbroek, 1997b).

In most instances, studies involving genotypic characterizationof M. bovis by MLVA utilized standard panels of loci consisting of24 or fewer markers (Avery, 2004; Haddad et al., 2004; Diguim-baye-Djaibe et al., 2006). In the present study, four extra loci wereadded to the established set of 24 loci. Two loci (VNTR 2163a and3232) showed high allelic diversity while the remaining two (VNTR1895 and 1982) revealed no allelic diversity. This is sufficient forepidemiological investigation of BTB in view of the fact that atyping method must have a discriminatory index of P90% to beconsidered efficient (Delafosse et al., 2002).

Combination of different typing methods as spoligotyping andMLVA can considerably improve the discriminatory power, andthereby provide better insight into epidemiological and evolution-ary relationships of bacterial isolates. When MLVA was combinedwith spoligotyping, the discriminatory power was higher(HGDI = 99.0%) than for either of the two methods alone, and

differentiated the M. bovis isolates into 39 distinct profiles com-pared to only 19 by MLVA and 17 by spoligotyping.

MLVA clustered the isolates differently from spoligotyping(Figs. 1 and 2) and strains within the dominant spoligotypeSB1176 (40.0% of the total isolates), were further differentiatedinto 12 distinct MLVA genotypes. Furthermore, all the seven iso-lates belonging to spoligotype SB0133, the second most dominantspoligotype, showed five distinct MLVA variants. This demon-strates that isolates with identical spoligopattern can have differ-ent MLVA profiles and vice versa. The difference in clustering ofthe isolates by MLVA method and spoligotyping is explained bythe variation in the site and number of markers targeted, wherein the case of MLVA, multiple loci exhibiting repetitive units areinvestigated compared to spoligotypes that exhibit deleted spacersin the direct repeat region. MLVA differentiated the isolates into 19genotypes (HGDI = 0.92), while spoligotyping differentiated theminto 17 types (HGDI = 0.82). This indicates that for EthiopianM. bovis isolates, MLVA differentiates better than spoligotyping.

01

23

45

60

12

34

56

MLV

A2M

LVA3

MLV

A5M

LVA6

MLV

A11

MLV

A13

MLV

A18

MLV

A1M

LVA2

MLV

A3M

LVA5

MLV

A7M

LVA8

MLV

A9M

LVA1

0M

LVA1

1M

LVA1

2M

LVA1

4M

LVA1

5M

LVA1

9

MLV

A1

MLV

A2

MLV

A3

MLV

A4

MLV

A13

MLV

A14

MLV

A1

MLV

A3

MLV

A5

MLV

A16

MLV

A18

Large scale farm Private dairy farms

Small scale farms Transhumance

Num

ber o

f stra

ins

per

MLV

A ge

noty

pe

MLVA genotypes

Fig. 3. Distribution of MLVA genotypes of M. bovis with respect to major cattlefarming systems in Ethiopia.

18 D. Biffa et al. / Infection, Genetics and Evolution 23 (2014) 13–19

MLVA based on the nine loci containing P3 alleles gave a HGDIof 0.91 (Table 3, panel VI), indicating that sound epidemiologicalinvestigations of M. bovis isolates in Ethiopia can be performedwith as few as nine VNTR loci. Different standard panels of VNTRloci have been tested in order to find the optimal MLVA panel forM. bovis. In the present study, the panel of nine QUB loci gave gooddiscrimination of the isolates, either alone, or in combination withthe set of MIRU or ETR loci (Table 3, panels V, VII and VIII).

The resolution power of the 12 MIRU loci (Table 3, panel XII) forthe current M. bovis isolates (HGDI = 0.72) closely resembles thatobserved from Chad and Ireland (HGDI = 0.75)(Haddad et al.,2004; Diguimbaye-Djaibe et al., 2006). These could suggest a highstability of the marker regardless of the difference in geographicorigin of isolates. In both the Ethiopian and Chadian studies, fiveMIRU loci (MIRU2, MIRU10, MIRU23, MIRU39 and MIRU40)showed no allelic diversity, while three loci (MIRU4, MIRU31 andMIRU26) showed a relatively high allelic diversity (>0.3). By con-trast, a much lower diversity of the 12 MIRU loci (HGDI = 0.38)was reported among M. bovis isolates of Korean origin (Avery,2004).

The combination of MLVA and spoligotyping was able to con-firm infection by multiple genotypes of M. bovis in 10 out of 24 car-casses. Two carcasses were infected with six different genotypeseach, two carcasses harboured four types, two harboured threetypes and the remaining six carcasses were infected with two typeseach. The observation of infection with multiple genotypes contra-dicts the study by Muller et al (Uitenbroek, 1997b) which statedthat establishment of one M. bovis strain in a given animal couldlimit the introduction of new strains. The present observations re-vealed that various M. bovis genotypes could co-exist in a host, per-haps with the possibility of exacerbating severity of pathology asall 10 carcasses showed disseminated tuberculous lesions. Ourfindings reinforce the importance of reducing the infectionpressure in order to avoid the occurrence of animals presenting

disseminated tuberculosis lesions with multiple genotypes, whichcould make them more likely to shed M. bovis and thus to transmitBTB to congeners. Being older could be explained as factor contrib-uting to the occurrence of polyclonal infections since the animalshave had more time to be exposed to multiple infections from dif-ferent sources and harbour and shade infections over a long periodof time.

In Ethiopia, high density linked to expanding large farms forcattle fattening and dairy production leads to a high degree ofstock co-mingling, and hence favour persistence and transmissionof infection in the area. In Addis Ababa and the central area a highnumber of different MLVA genotypes were identified (five andeight genotypes, respectively). Under such circumstances, animalsare exposed to repeated infection with different genotypes, andnine of the ten cattle infected with multiple genotypes were linkedto these regions.

Twelve MLVA variants were only detected in one isolate each,for other genotypes there were distribution overlaps between dis-tant geographic regions. MLVA 1, 3 and 13 were shared betweenAddis Ababa, the central region and the pastoral areas of Borena(Southern region). The most likely explanation would be introduc-tion of infected animals into Addis Ababa and central regions fromthe Borena area through unidirectional trade links. Borena pastoralregion (Southeast and Southwest) has long been the leading sup-plier of beef cattle to expanding feedlot farms and state-owned cat-tle breeding ranches in these regions that serve as a source ofbreeding bullocks to local farmers.

M. bovis strains isolated from dairy farm holdings (n = 25) weredifferentiated into 13 distinct genotypes, compared to five (con-sisting of seven isolates), six (eight isolates) and seven (10 isolates)genotypes of isolates from transhumance, small-scale and large-scale farms, respectively. This is probably due to the common prac-tice of introducing new animals into the farm from various sourcesand the fact that in Ethiopia, regardless of advanced age, dairycows are maintained in farms as long as they remain productive(milk and calf production). Such practice would enhance mycobac-terial-build up in the farm due to continuous circulation of anumber of pathogens within the host population. One study fromstate dairy farm in Addis Ababa area (Holeta) reported that all ani-mals within the farm were infected with the identical strain ofM. bovis, explained by the fact that there was no introduction ofnew animals since its establishment long time ago (Hailemariam,1975).

Some MLVA genotypes have been detected only in specific re-gions (MLVA 2 and MLVA 5 in Addis Ababa and the central region)while others are commonly shared between the regions. This couldsuggest that M. bovis strains have a tendency to form region-specific clonal clusters. This was congruent with observation madein UK (Robinson et al., 2008) and Tanzania (Acha and Szyfres,2001b), and similar clonal characteristic of M. tuberculosis fromEthiopia, Tunisia and The Netherlands was observed by restrictionfragment length polymorphism based on the insertion sequenceIS6110 (Acha and Szyfres, 2001a)

5. Conclusions

MLVA is a robust molecular typing tool that helps elucidate de-tail molecular genetic characteristics of M. bovis and thus couldpowerfully complement the existing traditional tools for epidemi-ological investigations. The use of MLVA typing for genotypic stud-ies revealed relatively high heterogeneity of the current EthiopianM. bovis isolates (19 distinct MLVA genotypes in 50 isolates).Likewise, spoligotyping also differentiated the 50 isolates into 17distinct spoligotypes. The discriminatory power of the MLVAtyping method was higher than spoligotyping. Adoption of VNTR

D. Biffa et al. / Infection, Genetics and Evolution 23 (2014) 13–19 19

loci with P3 alleles (n = 9) (VNTR 2163a, VNTR 1955, VNTR 424,VNTR 577, VNTR 3232, VNTR 2163b, VNTR 2996, VNTR 4052) canprovide improved differentiation between M. bovis isolates fromEthiopia. MLVA based on the VNTR loci was also able to confirmmultiple genotypes infections in individual animals. Further workwith more phylogenetically-robust molecular markers will be nec-essary to clarify the ancestry of the sampled Ethiopian M. bovispopulation.

6. Authors’ contributions

DB: Principal investigator participated in conception and designof the study, conducted field and lab work, data analysis, draftedmanuscript. TBJ, AM: participated in lab work, data analysis, writeup and critical review of the manuscript for important intellectualcontent. JG: data analysis, supervision of the lab work, write up andcritical review of the manuscript for important intellectual content.ES: conceived and designed the experiment, general supervision ofthe research work, acquisition of study fund, general conduct of thestudy, critical revision of the manuscript for important intellectualcontent. BD: supervision of the lab work, data analysis, write upand critical review of the manuscript for important intellectualcontent. All authors have read and approved the final version ofthe manuscript.

Acknowledgments

This work was support by grant obtained from the NorwegianSchool of Veterinary Science through the quota scholarship pro-gram. MLVA analysis was carried out at Genoscreen, Lille, France.We thank Sigrun Nilsen and Vivi Myrann (National VeterinaryInstitute, Oslo) for providing technical assistance during culturing,and identification of mycobacteria. We are also grateful toDr. Akinbowale Jenkins (University of Pretoria, South Africa) forhelp in deletion analyses work.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.meegid.2014.01.021.

References

Acha, P.N., Szyfres, B., 2001a. Zoonoses and Communicable Diseases Common toMan and Animals. Pan American Health Organization (PAHO), Washington, DC,USA.

Acha, P.N., Szyfres, B., 2001b. Zoonoses and Communicable Diseases Common toMan and Animals. Pan American Health Organization (PAHO).

Araujo, C.P., Leite, C.Q., Prince, K.A., Jorge Kdos, S., Osorio, A.L., 2005a. Mycobacteriumbovis identification by a molecular method from post-mortem inspected cattleobtained in abattoirs of Mato Grosso do Sul, Brazil. Mem. Inst. Oswaldo Cruz100, 749–752.

Araujo, C.P., Leite, C.Q., Prince, K.A., Jorge Kdos, S., Osorio, A.L., 2009. Mycobacteriumbovis identification by a molecular method from post-mortem inspected cattleobtained in abattoirs of Mato Grosso do Sul, Brazil. Mem. Inst. Oswaldo Cruz100, 749–752.

Avery, A., 2004. Red Meat and Poultry Production and Consumption in Ethiopia andDistribution in Addis Ababa. International Livestock Research Institute (ILRI),Addis Ababa, Ethiopia.

Ayele, W.Y., Neill, S.D., Zinsstag, J., Weiss, M.G., Pavlik, I., 2004. Bovine tuberculosis:an old disease but a new threat to Africa. Int. J. Tuberc. Lung Dis. 8, 924–937.

Central Statistical Authority (CSA), 2006. Agricultural Sample Survey, 2004/2005(1998 EC). Report on Livestock and Livestock Characteristics. Statistical Bulletin361. Addis Ababa, Ethiopia.

Collins, J.D., 1983. Abattoir associated zoonoses. J. Soc. Occup. Med. 33, 24–27.Corner, L.A., 1994. Post mortem diagnosis of Mycobacterium bovis infection in cattle.

Vet. Microbiol. 40, 53–63.Cosivi, O., Grange, J.M., Daborn, C.J., Raviglione, M.C., Fujikura, T., Cousins, D.,

Robinson, R.A., Huchzermeyer, H.F., de Kantor, I., Meslin, F.X., 1998a. Zoonotictuberculosis due to Mycobacterium bovis in developing countries. Emerg. Infect.Dis. 4, 59–70.

Cosivi, O., Grange, J.M., Daborn, C.J., Raviglione, M.C., Fujikura, T., Cousins, D.,Robinson, R.A., Huchzermeyer, H.F., de Kantor, I., Meslin, F.X., 1998b. Zoonotictuberculosis due to Mycobacterium bovis in developing countries. Emerg. Infect.Dis. 4, 59–70.

Cousins, D.V., Huchzermeyer, H.F.K., Griffin, J.F.T., Bruckner, G.K., Van rensburg,I.B.J., Kriek, N.P.J., 2004. Tuberculosis. In: Coetzer, A.W., Tustin, R.C. (Eds.),Infectious Diseases of Livestock. Oxford University Press, Cape Town, pp. 1973–1993.

Delafosse, A., Goutard, F., Thebaud, E., 2002. Epidémiologie de la tuberculose et de labrucellose bovine des bovins en zone périurbaine d’Abéché, Tchad. Rev. Elev.Méd. Vét. Pays Trop 55, 5–13.

Diguimbaye-Djaibe, C., Hilty, M., Ngandolo, R., Mahamat, H.H., Pfyffer, G.E., Baggi, F.,Hewinson, G., Tanner, M., Zinsstag, J., Schelling, E., 2006. Mycobacterium bovisisolates from tuberculous lesions in Chadian zebu carcasses. Emerg. Infect. Dis.12, 769–771.

Food Safety Authority of Ireland (FSAI), 2008. Zoonotic tuberculosis and food safety,second edition. Dublin, Ireland.

Haddad, N., Masselot, M., Durand, B., 2004. Molecular differentiation ofMycobacterium bovis isolates. Review of main techniques and applications.Res. Vet. Sci. 76, 1–18.

Hailemariam, S., 1975. A brief analysis of the activities of the meat inspection andquarantine division. Ministry of Agriculture, Addis Ababa, Ethiopia.

Hilty, M., Diguimbaye, C., Schelling, E., Baggi, F., Tanner, M., Zinsstag, J., 2005.Evaluation of the discriminatory power of variable number tandem repeat(VNTR) typing of mycobacterium bovis strains. Vet. Microbiol. 109, 217–222.

Maho, A., Mbakasse, R.N., Boulbaye, N., 1999. Causes de saisies aux abattoirs duTchad oriental (Fr.). LRVZ/F. In: Actes des IIIèmes Journées Agro-Sylvo-Pastorales. Laborotoires de recherches veterinaires et zootechniques deFarcha, N’Djamena, Tchad, 29 nov.-3 dec.1997.

Michel, A.L., de Klerk, L.M., Gey van Pittius, N.C., Warren, R.M., van Helden, P.D.,2007a. Bovine tuberculosis in African buffaloes: observations regardingMycobacterium bovis shedding into water and exposure to environmentalmycobacteria. BMC Vet. Res. 3, 23.

Michel, A.L., de Klerk, L.M., Gey van Pittius, N.C., Warren, R.M., van Helden, P.D.,2007b. Bovine tuberculosis in African buffaloes: observations regardingMycobacterium bovis shedding into water and exposure to environmentalmycobacteria. BMC Vet. Res. 3, 23.

MoH., 2002. Regional Tuberculosis case Notification. Ministry of Health (MoH),Federal Democratic Republic of Ethiopia, Addis Ababa, Ethiopia.

Müller, B., Hilty, M., Berg, S., Garcia-Pelayo, M.C., Dale, J., Boschiroli, M.L., Cadmus, S.,Ngandolo, B.N.R., Godreuil, S., Diguimbaye-Djaibé, C., Kazwala, R., Bonfoh, B.,Njanpop-Lafourcade, B.M., Sahraoui, N., Guetarni, D., Aseffa, A., Mekonnen, M.H.,Razanamparany, V.R., Ramarokoto, H., Djønne, B., Oloya, J., Machado, A.,Mucavele, C., Skjerve, E., Portaels, F., Rigouts, L., Michel, A., Müller, A.,Källenius, G., van Helden, P.D., Hewinson, R.G., Zinsstag, J., Gordon, S.V.,Smith, N.H., 2009. African 1, an epidemiologically important clonal complex ofmycobacterium bovis dominant in mali, nigeria, cameroon, and chad. Journal ofBacteriology 191, 1951–1960.

Oloya, J., Kazwala, R., Lund, A., Opuda-Asibo, J., Demelash, B., Skjerve, E., Johansen,T.B., Djonne, B., 2007. Characterisation of mycobacteria isolated from slaughtercattle in pastoral regions of Uganda. BMC Microbiol. 7, 95.

Robinson, P., Morris, D., Antic, R., 2008. Mycobacterium bovis as an occupationalhazard in abattoir workers. ANZ. J. Med. 18, 701–703.

Uitenbroek, D.G., 1997a. SISA-Binomial.Diagnostic statistics. Available from:<http://www.quantitativeskills.com/sisa/distributions/binomial.htm> (accessed01.01.04.).