Infection, Genetics and Evolution 12 (2012) 649–656

Contents lists available at SciVerse ScienceDirect

Infection, Genetics and Evolution

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

Spoligotypes of Mycobacterium tuberculosis complex isolates from patientsresidents of 11 states of Brazil

Harrison Magdinier Gomes a, Atina Ribeiro Elias a, Maranibia Aparecida Cardoso Oelemann a,Márcia Aparecida da Silva Pereira a, Fátima Fandinho Onofre Montes a, Ana Grazia Marsico b,Afrânio Lineu Kritski b, Luciano dos Anjos Filho b, Paulo C. Caldas c, Lia Gonçalves Possuelo d,Patrícia Cafrune d, Maria Lúcia Rossetti d, Norma Lucena e, Maria Helena Feres Saad f,Hebe Rodrigues Cavalcanti f, Clarisse Queico Fujimura Leite g, Rossana Coimbra de Brito h,Maria Luiza Lopes i, Karla Lima i, Maisa Souza i, Rita de Cássia Trindade j, Thierry Zozio k,Christophe Sola k,1, Nalin Rastogi k,⇑, Philip Noel Suffys a,⇑a Laboratory of Molecular Biology Applied to Mycobacteria, Dept. Mycobacteriosis, Oswaldo Cruz Institute, Rio de Janeiro, Brazilb Institute of Thoracic Disease, Federal University of Rio de Janeiro, Brazilc Ministry of Health, Rio de Janeiro, Brazild Federal University of Rio Grande do Sul, Porto Alegre, Brazile Centro de Pesquisa Aggeu Magalhães, Pernambuco, Brazilf Dept. Mycobacteriosis, Oswaldo Cruz Institute, Rio de Janeiro, Brazilg Dept. Pharmacy State University of São Paulo, Araraquara, Brazilh State University of Rio de Janeiro, Rio de Janeiro, Brazili Evandro Chagas Institute, Pará, Brazilj Federal University of Sergipe, Sergipe, Brazilk WHO Supranational TB Reference Laboratory, Institut Pasteur de la Guadeloupe, Abymes, France

a r t i c l e i n f o a b s t r a c t

Article history:Available online 1 September 2011

Keywords:Mycobacterium tuberculosisPopulation structureSpoligotypingBrazilDatabase

1567-1348/$ - see front matter � 2011 Elsevier B.V. Adoi:10.1016/j.meegid.2011.08.027

⇑ Corresponding authors.E-mail addresses: christophe.sola@igmors.u-psud.f

guadeloupe.fr (N. Rastogi), psuffys@ioc.fiocruz.br (P.N1 Present address: Institut de Génétique et Microbio

One of the high tuberculosis (TB) incidence countries in the world, Brazil is characterized by considerabledifferences in TB incidence on regional and state level. In the present study, we describe Brazilian spoligo-types of 1991 Mycobacterium tuberculosis complex (MTC) clinical isolates from patients residents of 11states from different regions of the country, diagnosed between 1996 and 2005. By performing spoligotyp-ing on a large number of M. tuberculosis clinical isolates, one of the main objectives of this study was todetermine the major genotype families causing TB in Brazil and to verify the region-associated genotypedistribution. We observed a total of 577 distinct spoligopatterns, 12.6% of these corresponded to orphanpatterns while 87.4% belonged to 326 shared-types (SITs). Among the latter, 86 SITs (isolated from 178patients) had been observed for the first time in this study, the most frequent being SIT2517 which belongedto the T3-ETH lineage and was exclusively found among patients residents of Belém, the capital of the stateof Pará (n = 8 isolates). Irrespective of shared-type labeling, a total of 19.5% strains were unique (unclu-stered) in our study as opposed to 80.5% clustered isolates (189 clusters, size range from 2 to 205 isolates).The three largest clusters were SIT42 of the Latin-America & Mediterranean (LAM) 9 clade (10.3%), SIT53 ofthe T clade (7.6%), and SIT50 of the Haarlem clade (5.4%). The predominant MTC lineages in Brazil in decreas-ing order belonged to the LAM (46%); the ill-defined T (18.6%); the Haarlem (12.2%), the X (4.7%), the S(1.9%), and the East African Indian (EAI) (0.85%) families. The rest of clades grouped together asMycobacte-rium africanum, Mycobacterium bovis, Beijing, Central Asian (CAS), and the Manu types, represented lessthan 1% of the strains. Finally, about 15% of the isolates showed spoligotype signatures that were not yetclassified among well-defined lineages. In conclusion, we provide hereby a first insight into the populationstructure of MTC isolates in Brazil, showing the predominance of both LAM and T family and the existence ofregion-associated genotypes.

� 2011 Elsevier B.V. All rights reserved.

ll rights reserved.

r (C. Sola), nrastogi@pasteur-. Suffys).logie, Orsay, France

1. Introduction

Human tuberculosis (TB) is generally caused by infection withMycobacterium tuberculosis and is the major cause of adult deathsdue to infectious diseases (Blower et al., 1996). This species is a

650 H.M. Gomes et al. / Infection, Genetics and Evolution 12 (2012) 649–656

member of the M. tuberculosis complex (MTC), a group of closelyrelated slow-growing mycobacteria that includes M. tuberculosis,Mycobacterium bovis, Mycobacterium africanum, Mycobacteriummicroti, Mycobacterium canettii, Mycobacterium pinnipedii andMycobacterium mungi. Although one third of the human populationis infected by M. tuberculosis, only a relatively small proportion ofinfected persons migrate from infection to disease. This could bedue to the long and lasting relationship between the humans pop-ulations and M. tuberculosis strains, as demonstrated by detectionof M. tuberculosis DNA in Egyptian, Peruvian and Andean mummies(Zink et al., 2003; Salo et al., 1994; Konomi et al., 2002) and thedetection of MTC DNA in a North American bison that lived around17,000 years ago (Rothschild et al., 2001). However, despite thislong standing association between certain genotypes of M. tubercu-losis and human populations (Rastogi and Sola, 2007), severe TBepidemic peaks were reported in 18th–19th century Europe(Fletcher et al., 2003). The reemergence of TB in the nineties, par-tially attributed to increasing demography, declining economicalconditions in certain third world countries, and lack of political will– called for immediate global attention, best translated by the Glo-bal Plan to stop TB during 2006–2015 which focuses, among oth-ers, on detection followed by directly-observed treatment ofpatients (Orcau et al., 2011); hence medical interventions are atthe core of the global strategy. However, historically, TB has beenused as a prime example of a ‘‘social disease’’, the control of whichrequires social, economic, scientific and environmental interven-tions (Lönnroth et al., 2009). Indeed, the TB burden follows a strongsocio-economic gradient between countries, within countries, andwithin communities, with the poorest having the highest risk;nonetheless, the causal pathways linking poverty and low socio-economic status to increased risk of tuberculosis are not fullyunderstood. The global TB control mainly focuses on cutting trans-mission through early case detection and effective treatment.

More recently, the global TB control objectives were redefinedto add the research component (http://www.stoptb.org/global/plan), since a better comprehension of the disease as well as thebacterium causing it would help design new approaches to controlTB. In this context, the new innovations to study molecular epide-miology (reviewed in Garcia de Viedma et al., 2011), allow not onlyto comprehend the human demography-influenced macropopula-tion structure of M. tuberculosis, but further allow to boost the localTB control programs thanks to both descriptive as well as interven-tionist molecular epidemiological approaches. According to themost recent Global TB Control Report by WHO, Brazil is one ofthe 22 high burden TB countries in the world with an estimatedincidence of 72/100,000 population (low estimate; high estimatebeing 100/100,000), and a TB/HIV co-infection rate of 12%(http://www.stoptb.org/assets/images/about/tbl_burden.gif). Bra-zil is a Federal Republic, composed of 26 states and a Federal Dis-trict (Brasília), occupying a geographic area of 8514,877 km2, withconsiderable regional differences in TB incidence among states.These differences are partly due to differences in host populationstructure, demographic variables and resulting transmission rates,as well as social determinants such as access to health care system.The country-wise differences in TB incidence are strongly influ-enced by the large size of the country (WHO, 2008), nonethelesswhether the genetic variability of M. tuberculosis strains could bepartly responsible for this phenomenon remains to be investigated.In 2008, a total of 71,059 new TB cases were registered in Brazil,including 6724 cases in the Northern region, 20,164 cases in theNorth-Eastern region, 3115 cases in the Mid-Western region,8764 cases in the South region and 32,266 cases in the South-East-ern region. Since molecular characterization of M. tuberculosis com-plex in developing nations has lead to novel information about thespread of M. tuberculosis (Oelemann et al., 2007), we performedgenotyping of a collection of available strains collected from 11

states that account for 71% of the total Brazilian population of190 million habitants.

Although M. tuberculosis complex represents one of the mostextreme examples of genetic homogeneity, with about 0.01–0.03% synonymous nucleotide variation and no significant traceof genetic exchange among them (Sreevatsan et al., 1997; Gut-ierrez et al., 2005); some recently developed PCR-based tools haveallowed to study both the biodiversity and phylogeographicalassessment of circulating M. tuberculosis clones (Garcia de Viedmaet al., 2011). Among these, the use of spoligotyping led to thedescription of SpolDB4 in 2006, a revolutionary database that al-lowed constructing a first detailed geographic scenario on globaldistribution of M. tuberculosis spoligotypes (Brudey et al., 2006a).We hereby report on a first large nationwide study to describespoligotyping-based genetic diversity and main genotypic lineagesof M. tuberculosis complex clinical isolates in Brazil.

2. Material and methods

2.1. Patients and bacterial isolates

The treatment of TB in Brazil is decentralized and directly takencare of in basic health units located at the municipal district level.Considering the complexity of logistics at the country-wide levelof Brazil, the 11 states chosen in this retrospective study dependedon the available contacts between the collaborating partners insteadof an exhaustive retrospective recruitment of strains/DNAs for spol-igotyping from all states; hence our study sample may be considereda convenience sample. This study describes spoligotyping results ofa total of 1991 M. tuberculosis complex (MTC) strains isolated during1997–2005 from the same number of patients, residing in 11 statesof Brazil, characterized by considerable differences in populationsize and TB incidence (Table 1). The isolates came from residentsof the five political regions of Brazil and included the North region(n = 223, 11.1%), the North-East region (n = 199, 11.4%), the South re-gion (n = 259, 13.1%), the South-East region (n = 1261, 62.3%), andthe Middle-East region (n = 7, 0.35%). For comparison, the total num-ber of new cases of TB in 2008 was 71,059; 74% of which were resi-dents of the 11 states included in the present study as compared tothe remaining 26% of the cases recorded in the other 15 states of thecountry. The population and number of TB cases is presented in Ta-ble 1; it may be summarized to conclude that about 42% of the Bra-zilian population lives in the South-East region and accounts for 45%(n = 32,266) of all new TB cases while the population of the Mid-West region represents 7.3% of the total population and accountsfor 4.4% (n = 3115) of new TB cases.

2.2. Nucleic acid extraction and spoligotyping

The DNA samples or cultures were mostly supplied by govern-mental institutions and when received as cultures, the bacterialDNA was prepared at one of the participating laboratories usingeither the CTAB method (van Soolingen et al., 1991) or bacterial ly-sis by heat shock. Spoligotyping was performed as described previ-ously (Kamerbeek et al., 1997). As shown recently (Abadia et al.,2011), DNA samples prepared by both procedures generated simi-lar spoligotyping results. Note that all genotyping was performedon de-identified DNA samples.

2.3. Computer assisted genotype analysis and comparison withdatabases

Spoligotype patterns as octal codes were entered in the SITVIT2proprietary database of the Institut Pasteur de la Guadeloupe,which is an updated version of the previously released SpolDB4

Table 1The population size and TB incidence (new cases/100,000 inhabitants) among 11 states of Brazil.

Brazilian regions States Population a TB incidence b Number (%) of strains in study c

North Amazonas 3221,939 67.1 37 (1.86%)Pará 7065,573 45.7 186 (9.34%)

North-East Ceará 8185,286 33.8 71 (3.57%)Pernambuco 8485,386 48.1 78 (3.92%)Sergipe 1939,426 38.1 50 (2.51%)

Mid-West Goiás 5647,035 14.7 7 (0.35%)

South-East Minas Gerais 19,273,506 24.0 57 (2.86%)Rio de Janeiro 15,420,375 35.6 1024 (51.43%)São Paulo 39,827,570 73.3 180 (9.04%)

South Paraná 10,284,503 36.5 6 (0.30%)Rio Grande do Sul 10,582,840 26.1 253 (12.71%)

a From:http://www.ibge.gov.br/english/presidencia/noticias/noticia_impressao.php?id_noticia=1065 (Instituto Brasileiro de Geografia e Estatística - IBGE, Populationcount 2007).

b From: http://tabnet.datasus.gov.br/cgi/deftohtm.exe?idb2009/d0202.def MS/SVS - Sistema de Informações de Agravos (Taxa de incidência de TB segundo unidade dafederação, Período 2007).

c Note that for 42 (2.11%) strains collected, the region of origin was not available.

H.M. Gomes et al. / Infection, Genetics and Evolution 12 (2012) 649–656 651

database (Brudey et al., 2006a; available online http://www.pas-teur-guadeloupe.fr:8081/SITVITDemo). At the time of the initialanalysis on April 2, 2009, the database contained genetic patternsof about 72,000 M. tuberculosis clinical isolates from 160 countriesof origin. In this database, Spoligotype International Type (SIT) des-ignates patterns shared by two or more patient isolates, whereas‘‘orphan’’ designates patterns reported for a single isolate. The Bio-numerics software Version 3.5 (Applied Maths NV, Sint-Martens-Latem, Belgium) was used to build minimum spanning trees(MST) based on spoligotyping data obtained in this investigationas reported previously (Helal et al., 2009). MST is an undirectednetwork in which all of the samples are linked together with thefewest possible linkages between nearest neighbors. Using this ap-proach, one considers that all intermediate stages are presentwithin the sample analyzed by including first the individual thatshows the greatest number of possible linkages to other individu-als in the population studied. We used this method to highlight thelinks between spoligotypes and lineages differing by changes ob-served in their direct variable repeats (Helal et al., 2009). Lastly,statistical analysis was performed by comparing categorical vari-ables by a two-tailed Fisher exact test with Yates’ correction whenapplicable (p value of 0.05 or less was considered significant).

3. Results

Spoligotyping showed the predominance of both the LAM andthe T families in Brazil as well as the existence of region-associatedgenotypes. It grouped all the 1991 clinical isolates in 577 distinctpatterns: 251 (43.5%) corresponded to orphan strains (Supplemen-tal Table S1) while the remaining 1740 (87.5%) isolates belonged to326 shared-types (SITs; Supplemental Table S2). Among the latter,86 SITs representing 178 patient isolates had not been reported be-fore, and were assigned new SIT numbers and lineages followingexpert-based interpretations using revised SpolDB4 rules. Themost important new SIT (SIT2517) was composed of eight isolatesfound only in the city of Belém, Para State, North Brazil, and be-longed to the T3-ETH clade. Among the 1991 isolates, 917 (46%)belonged to the LAM family; 370 (18.6%) to the ill-defined Tsuper-family (T1, T2, T3, T4 and T5; defined by default by Brudeyet al., 2006a); 242 (12.2%) to the Haarlem family; 93 (4.7%) as Xfamily; and 31 (1.6%) as S family. The rest belonged to smaller lin-eages and included genotypes characteristic of Beijing, CAS andManu lineages, as well as other MTC subspecies such as M. africa-num and M. bovis, representing less than 1% of the strains.

The top-12 spoligotype clusters grouped 904/1991 (45.4%) ofBrazilian MTC clinical isolates, and six of these belonged to well-defined LAM sublineages: LAM9/SIT42 n = 205, 10.30%; LAM2/SIT17 n = 85, 4.27%; LAM 6/SIT64 n = 73, 3.66%; LAM3/SIT33n = 68, 3.41%; LAM1/SIT20 n = 54, 2.71%; LAM5/SIT93 n = 23,1.15% (Table 2). Other minor LAM sublineages corresponded toLAM4, LAM8, LAM10 and LAM11 comprising 1.5% or less of totalstrains. Nonetheless, differences in the frequency of the differentLAM sublineages among different states were not statistically sig-nificant (Supplemental Tables S1 and S2).

The Haarlem family represented 12.1% (n = 242) of isolates. Inour study, the 242 Haarlem spoligotypes were subdivided intoH1 (n = 74, 30.6%), H2 (n = 7, 2.9%), and H3 (n = 108, 44.6%); a highdiversity was present for H1 and H3 sublineages (18 and 53 differ-ent patterns respectively). The ill-defined T super-family (n = 370,18.6%) was present in our study, and its prototype SIT53 (n = 151,7.6%) was the 2nd most frequent pattern in our study (Table 2);it represented 41% of all T genotype strains and 2/3rd of all T iso-lates came from Rio de Janeiro (absent in Sergipe, Goiás andParaná).

The S family (n = 37, 1.9%) strains – mostly characterized by itsprototype SIT34 (n = 31; characterized by the absence of the spac-ers 9, 10 and 33 to 36) – were most frequent in Rio de Janeiro(n = 27). Interestingly, we also found EAI lineage strains whichare phylogeographically specific of South-East Asia, India andEast-Africa (Brudey et al., 2006a). In our study, it was representedby 17 isolates originating in three states, with a single isolate in Re-cife, Pernambuco (SIT1983) and Rio de Janeiro (SIT11), and 15 inPará state; the latter were subdivided into SIT48/EAI1-SOM(n = 4); SIT702/EAI6-BGD1 (n = 4); SIT2543/EAI6-BGD1 (n = 2);SIT129/EAI6-BGD1 (n = 1); SIT924/EAI (probable ancestor of EAI2-Nonthaburi) (n = 1) and three orphans. Lastly, the ancestral Manulineage strains were represented by all the three sublineages(Manu1, Manu2 and Manu3), and isolated from patients in Rio deJaneiro (n = 6) and in São Paulo (n = 2).

A small group of SITs were found in one region or state of Brazil,e.g., in Rio de Janeiro: SIT18/X2 (n = 13), SIT37/T3 (n = 5), SIT156/Unknown (n = 4), SIT370/T1 (n = 7), SIT1163/T3 (n = 5), SIT746/Un-known (n = 4), SIT176/LAM (n = 7), SIT534/LAM (n = 4), SIT770/LAM9 (n = 8), SIT1106/LAM4 (n = 5); in Pará, SIT2517/T3-ETH(n = 8), and in Pernambuco, SIT2553/X2 (n = 4). On the opposite,we also observed patterns shared between 2 and 3 states, suchas SIT65 (Rio Grande do Sul n = 21; Rio de Janeiro n = 3; São Paulon = 1), SIT119 (Rio Grande do Sul n = 9; Pará n = 1; Pernambucon = 1), SIT157 (Amazonas n = 4; Rio de Janeiro n = 7; São Paulo

Table 2Description of the 12 most predominant clusters representing 904/1991 (45.4%) of Brazilian M. tuberculosis clinical isolates, and their worldwide distribution in the SITVIT2 database.

652H

.M.G

omes

etal./Infection,G

eneticsand

Evolution12

(2012)649–

656

H.M. Gomes et al. / Infection, Genetics and Evolution 12 (2012) 649–656 653

n = 1), SIT240 (São Paulo n = 5; Rio Grande do Sul n = 1), SIT291 (Riode Janeiro n = 7; São Paulo n = 1), SIT462 (Rio de Janeiro n = 5; RioGrande do Sul n = 1); SIT498 (Rio Grande do Sul n = 10; Pará n = 1),SIT535 (Rio de Janeiro n = 13; São Paulo n = 5).

Despite recent controversy around the suitable methods forMTC phylogenetical analyses (Comas et al., 2009), we drew a spol-igotyping-based MST to summarize the probable evolutionaryrelationships between all shared-type patterns observed in ourstudy (Fig. 1; note that hypothetical links were allowed in this tree,and are represented by gray dotted lines). This tree is split in twolarge and opposite branches, the top one for the LAM super-family,and bottom one grouping T, S, X and Haarlem families (Fig. 1A). Inthe zoomed image (Fig. 1B), one may observe two main centralnodes – the upper node representing the SIT42/LAM9, and a middlenode representing SIT53/T connected to H3 and H1 subgroups atthe bottom. The SIT53 node is directly connected to X and S lin-eages (right and left hand sides, respectively). Note that all the pre-dominant LAM sublineages were well represented in the MST;among these the most predominant LAM9 sublineage with its pro-totype SIT42 (n = 205) constituted the central node of the tree. Italso showed the highest diversity with SITs other than its proto-type SIT42. Other LAM sublineages present with their prototypesas the branch nodes were: LAM2/SIT17 n = 85; LAM6/SIT64n = 73; LAM3/SIT33 n = 68; LAM1/SIT20 n = 54; and LAM5/SIT93(n = 23).

4. Discussion

The global TB control mainly focuses on cutting transmissionthrough early case detection and effective treatment. Medicalinterventions are at the core of the global strategy. The influence

Fig. 1. A minimum spanning tree (MST) illustrating phylogenetical relationships betwebranches. The tree connects each shared-type pattern based on degree of differences amolocus. The structure of the tree is represented by branches (continuous vs. dotted lines), apattern, while the length of the branches represents the number of differences betweedenotes the number of spacers of the DR locus that differ between two patterns: continuspacers.

of ethnicity on susceptibility to infection and disease developmenthas been described before and of the most striking examples isinfection with plasmodium, causing malaria (Kwiatkowski, 2000).Recently, it was suggested that different ethnic groups had differ-ent susceptibility for particular M. tuberculosis strains which couldeven be observed in complex population settings (Gagneux et al.,2006). Beside the host genetic background, human migration hasalso been demonstrated to influence the genetic composition andregion-associated genetic variability of parasites, as seen for Myco-bacterium leprae (Monot et al., 2005), and M. tuberculosis (Mokrou-sov, 2007).

In the present study, we provide the first insight into the popu-lation structure of MTC isolates in Brazil, showing the predomi-nance of both the LAM and the T family and the existence ofregion-associated genotypes. Indeed, almost half (46.2%) of theMTC population presented in our study was characterized withLAM genotypes, confirming the predominance of this super-familyin South American TB patients, including in Brazil.

The principal group of isolates (42%) came from the South-East-ern region, the richest and most industrialized Brazilian region andone of the most important economic regions from South America.This region receives migrants from the whole of Brazil and fromother South American countries, as well as tourists from other re-gions. We assume that origin of eventual genetic diversity of M.tuberculosis in Brazil is due to political, economical and social as-pects of the society, as well as a consequence of five centuries ofinterethnic relationships from people worldwide (Parra et al.,2003). In the first decade of the 19th century, the Brazilian portswere opened to all ‘‘friendly nations’’ and this attitude was respon-sible to increase the numbers of immigrants from all parts of theworld (Prado, 1981; IBGE, 2000). Between 1500 (Brazil’s discovery)

en the Brazilian spoligotypes: (A) global tree (B) zoomed image of main nodes andng them as represented by the presence or absence of one or more spacers of the DRnd circles (open vs. closed circles). Each circle represents one individual spoligotypen patterns. The complexity of the lines (continuous, black dotted and gray dotted)ous for one spacer, black dotted for two spacers, and dotted gray for three or more

654 H.M. Gomes et al. / Infection, Genetics and Evolution 12 (2012) 649–656

and 1972, 58% of the immigrants who arrived in Brazil were Euro-peans, 40% Africans and 2% were Asians; between 1551 and 1850(when the slave trade was abolished), approximately 3.5 millionAfricans arrived in Brazil (Alves-Silva et al., 2000).

Knowledge on the distribution of M. tuberculosis genotypes on alarge scale has undergone a tremendous change since the release ofsuccessive versions of the international database of spoligotypes,from SpolDB1 in 1999 to SpolDB4 in 2006, and gradually increasingthe number of SITs, from 53 patterns in SpolDB1 to 1939 patternsin SpolDB4 (reviewed in Brudey et al., 2006a). This database waspresently updated to a version called SITVIT2 (Vanhomwegenet al., 2011) which describes about 2800 SITs from 160 countriesof patient origin. Besides providing an organized and interactivedatabase, this approach also offers an interpretation on worldwidetransmission dynamics, i.e., by definition susceptible to changesdue to new spoligotype input. One such example is the close asso-ciation of certain genotypes such as SIT18 in Brazil that was previ-ously reported in low frequency only in Great Britain, TheNetherlands and the United States (Brudey et al., 2006a). Despitethe reported occurrence of SIT18 found specifically in the Rio de Ja-neiro area (after 1997), its frequency in the world continues to below (Supplemental Table S2). SIT18 is linked to the X family of pre-sumed Anglo-Saxon ancestry (Brudey et al., 2006a). Despite anoverall low frequency of the X lineage in Brazil (4.7%), it is well-dis-tributed in six out of 11 states investigated: Pará, Pernambuco,Ceará, Rio Grande do Sul, Rio de Janeiro and São Paulo.

The LAM family predominated in South America with approxi-mately 45% in SpolDB4 and this is similar to the frequency of46.2% of LAM isolates observed presently. However, a state-wisedifference in frequency of LAM was observed; being highest inthe state of Sergipe, (81.6%) and may be due to the fact that thiswas the first region that was colonized by the Portuguese and sec-ondly by the state of Ceará (70%), an area that included three of the‘‘capitanias’’, originally granted by the Portuguese crown. The low-est frequency of LAM strains was observed in São Paulo state (35%).Although the LAM frequency is high in Latin-America, regional dif-ferences are observed, ranging between 65% in Venezuela to 17.4%in French Guiana (Brudey et al., 2006a). The predominance of theLAM super-family, as observed in Brazil, was also reported in Por-tugal, being 51% (David et al., 2007). It is reasonable to accept thatthis is due to the exclusive relationship between Portugal and Bra-zil during three centuries (1500–1808 AD), although sublineagessuch as LAM1 and LAM9 are observed twice as much in the presentstudy as compared to in Portugal. In this context, one would expectthe LAM family to be highly prevalent also among other former col-onies of Portugal such as Angola, Mozambique, Guinea-Bissau,Cabo Verde, São Tomé e Príncipe and East Timor, as well as regionssuch as Goa, Diu, Damão, Galiza e Macau. Although spoligotypesare not available from all the regions, available data in SITVIT2database shows that the proportion of LAM strains varies from28% in Mozambique to 16% in Guinea-Bissau. For more detailedinformation on spoligotypes that are shared between Brazilianand Portuguese isolates, please refer to the Supplemental Table S2.

A particular aspect of the LAM family strains is that they are un-ique to harbor a significant proportion of strains with a specificdeletion named RDRio characterized by a single long sequence poly-morphism (>26.3 kb) with deletion or modification of 10 genes(Lazzarini et al., 2007); interestingly the RDRio/LAM strains consti-tute a predominant clade causing TB in Rio de Janeiro – LAM1 andLAM2 being exclusively of the RDRio genotype, LAM4, LAM5, LAM6and LAM9 being composed of both RDRio and wild type (WT) geno-types, and LAM3 composed exclusively of WT isolates. Since RDRio

genotype is present in 30% of the isolates from Rio de Janeiro, andis suspected of being more virulent and able to cause disease morefrequently (Lazzarini et al., 2007), it will be worthwhile to investi-gate the presence of this genotype in the rest of Brazil. Although we

did not look for the RDRio deletion in the present study, but assum-ing that LAM1 (SIT20) and LAM2 (SIT17) are exclusively composedof the RDRio genotype and correlating it to the distribution of Bra-zilian spoligotypes in the present study (with the exception ofParaná, LAM1 and LAM2 isolates were present within a frequencyof 2–8% in all states), we may extrapolate that the RDRio genotypeis indeed widely spread throughout Brazil.

Regarding evolutionary modern strains belonging to the Euro-American lineage which comprises LAM, Haarlem, X and T clades(Gagneux and Small, 2007), we further looked on their distribu-tion pattern in Brazil (this study) as compared to Portugal (Davidet al., 2007). Regarding the ill-defined T family, and Haarlemfamily of European descent that was described in 1999 in TheNetherlands (with strains sharing the absence of spacer 31 butsubdivided as sublineages H1, H2 and H3 based on minor addi-tional variations: Kremer et al., 1999; Brudey et al., 2006a); weobserved similar frequencies of both T1 and H1 sublineages asopposed to the H3 sublineage which was more frequent inPortugal.

Considering the stable association between M. tuberculosis andits human host (Gagneux et al., 2006), a group of SITs found in Bra-zil at very low frequency and simultaneously uncharacteristic ofthe South American region in SITVIT2 database (Supplemental Ta-ble S2) could represent imported genotypes due to internationalmigratory flux. We therefore concentrated on genotypes that areobserved in a very low frequency in the present study, but werenot reported in Brazil or Latin-America in previous studies, andare phylogenetically specific of regions other than South America.As can be seen through the Supplemental Table S2, the genotypesBeijing/SIT1, EAI3-IND/SIT11, EAI1-SOM/SIT48, Manu2/SIT54 andManu2/SIT1690, LAM11-ZWE/SIT59, LAM11-ZWE/SIT2173,LAM10-CAM/SIT2550, and T5-RUS1/SIT803 are among some ofsuch examples.

Other examples include strains that are not phylogeneticallyspecific of South American region in SpolDB4 (Brudey et al.,2006a), but for whom some strains might have established them-selves in South America in an isolated manner at lower proportionsdue to recent migratory events (as seen through the updated SIT-VIT2 database). Such examples include Africanum/SIT32, EAI6-BGD1/SIT129, EAI6-BGD1/SIT702, EAI/SIT924, and EAI3-IND/SIT1983. Lastly, there are also some extreme examples of newly ar-rived genotypes in Brazil that have undergone evolutive pressurewith subsequent loss of some spacers, resulting in new spoligotypepatterns of lineages with well-documented specificity for distantcontinents. Such patterns are not yet documented in the country/continent of origin of the lineage, but are exclusively present inBrazil (Supplemental Table S2). Some examples include: T3-ETH/SIT2517 (n = 8), T-Tuscany/SIT2523 (n = 3), T5-RUS1/SIT2533(n = 2), T3-ETH/SIT2542 (n = 5), EAI6-BGD1/SIT2543 (n = 2), CAS1-Delhi/SIT2545 (n = 2), T3-ETH/SIT2546 (n = 2).

Other significant intercontinental matches were observed withNorth America and Europe, one explanation could be the historicalrelationship of these regions with Brazil right from the beginningof colonization (Table 2; Supplemental Table S2). The migratoryflux to North America and Europe from African and Asian countriesin recent decades and continued commercial and familial travel be-tween Brazil, North America and Europe could also indirectly ex-plain for some rarely observed genotypes in our study, e.g.,Beijing/SIT1, EAI3-IND/SIT11, EAI1-SOM/SIT48, Manu2/SIT54 andManu2/SIT1690, LAM11-ZWE/SIT59, LAM11-ZWE/SIT2173,LAM10-CAM/SIT2550 and T5-RUS1/SIT803), that are phylogeneti-cally specific of regions other than South America. Although thesestrains most probably represent imported cases of MTC, one cannotfully exclude the remote possibility that some isolates could be theremnants of strains undergoing extinction that were previouslypresent in Brazil due to a past undocumented association between

H.M. Gomes et al. / Infection, Genetics and Evolution 12 (2012) 649–656 655

M. tuberculosis and its autochthonous host subpopulation in a dis-tant past.

The incomplete information we had regarding patients did notallow us to define whether clusters harbored outbreak strainsbut because most clusters included strains from different states(with the exception of EAI isolates found in significantly highernumbers in Belém, Pará state), one may assume that most of thespoligotyping clusters were not made up of outbreak strains. TheEAI genotypes, characterized by the absence of spacers 29–32and 34, were observed in 21 isolates, 16 were found in Pará, fourin Rio de Janeiro and one in Pernambuco. Among various EAI sub-types described in SpolDB4 (Brudey et al., 2006a), EAI1-SOM/SIT48,EAI6-BGD1/SIT129, EAI6-BGD1/SIT702, EAI/SIT924 and EAI6-BGD1/SIT2543 were found in a high proportion in the Pará state.As shown in the Supplemental Table S2, some of these SITs havebeen reported in neighboring French Guiana (SIT129 at 16% andSIT924 at 23.5%), suggesting transmission between citizens fromboth regions. Indeed, there is an important influx of Brazilians fromBelém to Cayenne in French Guiana for economical reasons, whilepeople from French Guiana travel frequently to Belém for tourism(Brudey et al., 2006b).

Among the spoligotypes that were observed in the presentstudy but related as being uncommon in South America, we men-tion four isolates with the ancestral Manu2/SIT54 genotype (threein Rio de Janeiro and one in São Paulo) and the LAM11-ZWE/SIT59spoligotype (three in Rio de Janeiro and one in Minas Gerais). InSpolDB4, the Manu2 genotype was specific for India, whileLAM11-ZWE was predominant in Zimbabwe. Unfortunately, wehave no information regarding eventual contact between the Bra-zilian patients with these isolates or with foreign patients.

Lastly, despite the fact that nearly 3.5 million people of Africanwere forced to migrate during three centuries, the low frequency inour study strains with phylogeographical specificity for Africa, e.g.,LAM10-CAM/SIT2550 (M. tuberculosis) or the AFRI lineage (M. afri-canum), is noteworthy. We also found a relatively well spreadT5_Madrid2/SIT58 genotype which is highly prevalent in Spain(Brudey et al., 2006a); it was found in four states (one in Pará, fourin Rio de Janeiro, three in Rio Grande do Sul and one in São Paulo).Additionally, three isolates of T5_Madrid2/SIT1227 were found inRio de Janeiro. Lastly single isolates T1-RUS2/SIT280 and T-Tus-cany/SIT159 were also found in Rio de Janeiro, and could be dueto patient origin or contact with foreigners; nonetheless detailedpatient records were not available to conclude precisely.

Complementing genotyping of MTC clinical isolates to conven-tional epidemiological studies helps define bacterial populationstructure with a better understanding of the importance of circu-lating clones on TB transmission dynamics. Hence the results ofthis investigation which provide with a first insight in the popula-tion structure of the Brazilian MTC clinical isolates are particularlyimportant in the context of recent Brazilian health sector reformsfor TB control. Indeed, the new national plan on TB control nowimplements the WHO approved DOTS strategy (directly-observedtreatment, short-course) for TB control in 5500 municipalitiesthrough decentralized implementation via out-patient clinics(Kritski and Ruffino-Netto, 2000). A better knowledge of circulatingMTC strains and their genotypes, and any specific relationships be-tween specific MTC clones and host subpopulations, as well asvariables such as drug-resistance, virulence characteristics etc.,would be beneficial both for researchers and the national TB con-trol program. Nonetheless the clustering rate observed in our study(at almost 80%) is certainly too high since further differentiation ofTB clusters as defined by spoligotyping was not systematically per-formed using secondary markers. The next steps of this ongoingwork will now extend to MIRU-VNTRs based typing to draw epide-miological conclusions (Supply et al., 2006; Rastogi and Sola,2007).

Acknowledgments

We sincerely thank Angela Werneck, José Roberto Lapa e Silva,Júlia Salen, Luís Cláudio de Oliveira Lazzarini and Marcílio Fernan-des Baliza for their contribution to this study. This investigationwas partly supported by the Fiocruz-Pasteur research grant to Na-lin Rastogi and Philip Noel Suffys.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.meegid.2011.08.027.

References

Abadia, E., Zhang, J., Ritacco, V., et al., 2011. The use of microbead-basedspoligotyping for Mycobacterium tuberculosis complex to evaluate the qualityof the conventional method: Providing guidelines for Quality Assurance whenworking on membranes. BMC Infect Dis. 11, 110.

Alves-Silva, J., da Silva Santos, M., Guimarães, P.E., et al., 2000. The ancestry ofBrazilian mtDNA lineages. American Journal of Human Genetics 67, 444–461.

Blower, S.M., Small, P.M., Hopewell, P.C., 1996. Control strategies for tuberculosisepidemics: new models for world problems. Science 273, 497–500.

Brudey, K., Driscoll, J.R., Rigouts, L., et al., 2006a. Mycobacterium tuberculosiscomplex genetic diversity: mining the fourth international spoligotypingdatabase (SpolDB4) for classification, population genetics and epidemiology.BMC Microbiology 6, 23.

Brudey, K., Filliol, I., Ferdinand, S., et al., 2006b. Long-term population-basedgenotyping study of Mycobacterium tuberculosis complex isolates in the Frenchdepartments of the Americas. Journal of Clinical Microbiology 44, 183–191.

Comas, I., Homolka, S., Niemann, S., Gagneux, S., 2009. Genotyping of geneticallymonomorphic bacteria: DNA sequencing in Mycobacterium tuberculosishighlights the limitations of current methodologies. PLoS ONE 4 (11), e7815.

David, S., Ribeiro, D.R., Antunes, A., et al., 2007. Contribution of spoligotyping to thecharacterization of the population structure of Mycobacterium tuberculosisisolates in Portugal. Infect Genet Evol. 7, 609–617.

Fletcher, H.A., Donoghue, H.D., Holton, J., et al., 2003. Widespread occurrence ofMycobacterium tuberculosis DNA from 18th-19th century Hungarians. AmericanJournal of Physical Anthropology 120, 144–152.

Gagneux, S., DeRiemer, K., Van, T., et al., 2006. Variable host-pathogen compatibilityin Mycobacterium tuberculosis. Proc Natl Acad Sci U S A. 103, 2869–2873.

Gagneux, S., Small, P., 2007. Global phylogeography of Mycobacterium tuberculosisand implications for tuberculosis product development. Lancet Infect Dis 7,328–337.

García de Viedma, D., Mokrousov, I., Rastogi, N., 2011. Innovations in the molecularepidemiology of tuberculosis. Enfermedades Infecciosas y Microbiologia Clinica29 (Suppl. 1), 8–13.

Gutierrez, M.C., Brisse, S., Brosch, R., 2005. Ancient origin and gene mosaicism of theprogenitor of Mycobacterium tuberculosis. PLoS Pathogens 1 (e5), 55–61.

Helal, Z.H., Ashour, M.S., Eissa, S.A., Abd-Elatef, G., Zozio, T., Babapoor, S., Rastogi, N.,Khan, M.I., 2009. Unexpectedly high proportion of ancestral Manu genotypeMycobacterium tuberculosis strains cultured from tuberculosis patients in Egypt.Journal of Clinical Microbiology 47, 2794–2801.

IBGE – Instituto Brasileiro de Geografia e Estatistica., 2000. Brasil: 500 anos depovoamento Centro de Documentacaoe Disseminacao de Informacoes,Ministerio do Planejamento Rio de Janeiro, Brazil, p. 231.

Kamerbeek, J., Schouls, L., Kolk, A., et al., 1997. Simultaneous detection and straindifferentiation of Mycobacterium tuberculosis for diagnosis and epidemiology.Journal of Clinical Microbiology 35, 314–907.

Konomi, N., Lebwohl, E., Mowbray, K., 2002. Detection of mycobacterial DNA inAndean mummies. Journal of Clinical Microbiology 40, 4738–4740.

Kremer, K., van Soolingen, D., Frothingham, R., et al., 1999. Comparison of methodsbased on different molecular epidemiological markers for typing ofMycobacterium tuberculosis complex strains: interlaboratory study ofdiscriminatory power and reproducibility. Journal of Clinical Microbiology 37,2607–2618.

Kritski, A.L., Ruffino-Netto, A., 2000. Health sector reform in Brazil: impact ontuberculosis control. Int J Tuberc Lung Dis. 4, 622–626.

Kwiatkowski, D., 2000. Genetic susceptibility to malaria getting complex. CurrentOpinion in Genetics and Development 10, 320–324.

Lazzarini, L.C., Huard, R.C., Boechat, N.L., Gomes, H.M., Oelemann, M.C., Kurepina, N.,Shashkina, E., Mello, F.C., Gibson, A.L., Virginio, M.J., Marsico, A.G., Butler, W.R.,Kreiswirth, B.N., Suffys, P.N., Lapa ESilva, J.R., Ho, J.L., 2007. Discovery of a novelMycobacterium tuberculosis lineage that is a major cause of tuberculosis in Riode Janeiro, Brazil. Journal of Clinical Microbiology 45, 3891–3902.

Lönnroth, K., Jaramillo, E., Williams, B.G., et al., 2009. Drivers of tuberculosisepidemics: the role of risk factors and social determinants. Social Science andMedicine 68, 2240–2246.

Monot, M., Honoré, N., Garnier, T., et al., 2005. On the origin of leprosy. Science 308,1040–1042.

656 H.M. Gomes et al. / Infection, Genetics and Evolution 12 (2012) 649–656

Mokrousov, I., 2007. Towards a quantitative perception of human-microbial co-evolution. Front Biosci. 12, 4818–4825.

Oelemann, M.C., Fontes, A.N., Pereira, M.A., et al., 2007. Typing of Mycobacteriumtuberculosis strains isolated in community health centers of Rio de Janeiro City,Brazil. Memorias do Instituto Oswaldo Cruz 102, 455–462.

Orcau, A., Caylà, J.A., Martínez, J.A., 2011. Present epidemiology of tuberculosisPrevention and control programs. Enfermedades Infecciosas y MicrobiologiaClinica 29 (Suppl. 1), 2–7.

Parra, F.C., Amado, R.C., Lambertucci, J.R., et al., 2003. Color and genomic ancestry inBrazilians. Proc Natl Acad Sci U S A. 100, 177–182.

Prado, P., 1981. Retrato do Brasil: ensaio sobre a tristeza brasileira. São Paulo:Instituição Brasileira de Difusão Cultural. pp. 157.

Rastogi, N., Sola, C., 2007. Molecular evolution of the Mycobacterium tuberculosiscomplex, p. 53–91. In: Palomino, J.C., Leao, S., Ritacco, V. (Eds.), Tuberculosis2007: from basic science to patient care. Amedeo Online Textbooks, http://www.tuberculosistextbook.com/index.htm

Rothschild, B.M., Martin, L.D., Lev, G., et al., 2001. Mycobacterium tuberculosiscomplex DNA from an extinct bison dated 17, 000 years before the present.Clinical Infectious Diseases 33, 305–311.

Salo, W.L., Aufderheide, A.C., Buikstra, J., et al., 1994. Identification of Mycobacteriumtuberculosis DNA in a pre-Columbian Peruvian mummy. Proc Natl Acad Sci U SA. 91, 2091–2094.

Sreevatsan, S., Pan, X., Stockbauer, K.E., et al., 1997. Restricted structural genepolymorphism in the Mycobacterium tuberculosis complex indicates

evolutionarily recent global dissemination. Proc Natl Acad Sci U S A. 94,9869–9874.

Supply, P., Allix, C., Lesjean, S., et al., 2006. Proposal for standardization of optimizedMycobacterial Interspersed Repetitive Unit-Variable Number Tandem Repeattyping of Mycobacterium tuberculosis. Journal of Clinical Microbiology 44, 4498–4510.

Vanhomwegen, J., Kwara, A., Martin, M., Gillani, F.S., Fontanet, A., Mutungi, P.,Crellin, J., Obaro, S., Gosciminski, M., Carter, E.J., Rastogi, N., 2011. Impact ofimmigration on the molecular epidemiology of tuberculosis in Rhode Island.Journal of Clinical Microbiology 49, 834–844.

van Soolingen, D., Hermans, P.W., de Haas, P.E., et al., 1991. Occurrence and stabilityof insertion sequences in Mycobacterium tuberculosis complex strains:evaluation of an insertion sequence-dependent DNA polymorphism as a toolin the epidemiology of tuberculosis. Journal of Clinical Microbiology 29, 2578–2586.

World Health Organization., 2008. Global tuberculosis control surveillance,planning, financing: WHO report. http://www.who.int/tb/publications/global_report/2008/download_centre/en/index.html, Geneva, World HealthOrganization.

Zink, A.R., Sola, C., Reischl, U., et al., 2003. Characterization of Mycobacteriumtuberculosis complex DNAs from Egyptian mummies by spoligotyping. Journalof Clinical Microbiology 41, 359–367.