Antigenic and genetic characterization of rabiesviruses isolated from domestic and wild animals ofBrazil identifies the hoary fox as a rabies reservoir

F. Bernardi,1 S. A. Nadin-Davis,2 A. I. Wandeler,2 J. Armstrong,2

A. A. B. Gomes,3 F. S. Lima,3 F. R. B. Nogueira3 and F. H. Ito1

Correspondence

S. A. Nadin-Davis

nadindaviss@inspection.gc.ca

1Department of Preventive Veterinary Medicine and Animal Health, FMVZ-USP, Av. ProfessorDr Orlando Marques de Paiva 87, 05508-000 Cidade Universitaria, Sao Paulo-SP, Brazil

2Rabies Centre of Expertise, Canadian Food Inspection Agency, Ottawa Laboratory-Fallowfield,3851 Fallowfield Road, Ottawa, Canada K2H 8P9

3Department of Veterinary Medicine - DMV, ‘Centro de Saude e Tecnologia Rural - CSTR’,Federal University of Campina Grande, Caixa Postal 64, 58700-000 Patos-PB, Brazil

Received 31 May 2005

Accepted 4 July 2005

Fifty Brazilian rabies viruses, collected frommany different animal species and several regions of the

country, were characterized by partial sequencing of the central, variable region of the P gene,

a locus useful for sensitive molecular epidemiological studies. Phylogenetic analysis of the

sequences, which included comparison with other rabies strains recovered from throughout the

Americas, identified three main groups of Brazilian viruses, arbitrarily designated BRL-1 to BRL-3.

BRL-1 was found in terrestrial carnivores and clusters with other American strains of the

cosmopolitan lineage. BRL-2 comprised two distinct isolates, recovered from two species of

non-haematophagous bats, that had evolutionary links to insectivorous-bat-derived strains of

North America. BRL-3 consisted of isolates from vampire bats and from livestock species probably

infected via contact with vampire bats. The terrestrial group was further subdivided into three

subtypes: BRL-1a was associated exclusively with dogs and cats, while BRL-1b and BRL-1c were

found exclusively in hoary foxes. These observations strongly support the role of the Brazilian

hoary fox as a rabies reservoir. Screening of representative Brazilian rabies viruses against a

collection of anti-rabies monoclonal antibodies (mAbs) identified a small panel of mAbs that could

be used to discriminate between all Brazilian subgroups as defined by genetic classification in

this study.

INTRODUCTION

The genus Lyssavirus, family Rhabdoviridae, comprises agroup of negative-sense RNA viruses having genomes ofapproximately 12 kb that encode five genes (N, P,M, G andL) and which are all capable of eliciting clinical rabies inmammalian species (Tordo et al., 1998). All members of thisgenus so far recovered in the Americas belong exclusively tothe sero-genotype group 1 that comprises all classical rabiesviruses (Bourhy et al., 1993; Kissi et al., 1995; Badrane &Tordo, 2001; Nadin-Davis et al., 2001, 2002); however, theepidemiology of rabies on the American continent is com-plex. In Canada and the United States, where dog rabies wascontrolled in the 1940s–1950s, the disease remains a signifi-cant public health concern due to its persistence in a varietyof terrestrial- and aerial-wildlife hosts (Krebs et al., 2003).Similarly in Mexico, where intensive urban rabies-control

efforts have substantially reduced both dog and subsequenthuman cases of rabies in recent years, the existence of distinctrabies virus variants associated with specific terrestrial hosts(skunks and foxes) and chiropteran species (the Brazilianfree-tailed bat and the vampire bat) has been recognized(Velasco-Villa et al., 2002). As additional Latin Americancountries strive to reduce human rabies through the controlof the disease in dogs, it is probable that sylvatic rabies,maintained in a variety ofmammalian hosts, will emerge as asignificant problem. Public health authorities will thus needto adjust their control and surveillance efforts in response tothese changes in disease demographics.

To better understand rabies epidemiology, antigenic andgenetic methods of virus characterization are being increas-ingly applied to collections of Latin American rabies viruses.Most antigenic methods of strain discrimination target theviral nucleoprotein, the product of the N gene, which isexpressed in substantial quantities in infected tissues andexhibits sufficient antigenic variation that most strains can

The GenBank/EMBL/DDBJ accession numbers for the nucleotidesequences determined in this work are AY962047–AY962096.

0008-1223 G 2005 SGM Printed in Great Britain 3153

Journal of General Virology (2005), 86, 3153–3162 DOI 10.1099/vir.0.81223-0

be distinguished. In particular, a panel of eight anti-Nprotein monoclonal antibodies (mAbs), which can differ-entiate between 11 distinct rabies virus variants harbouredby a variety of terrestrial and chiropteran hosts, was reported(Diaz et al., 1994; Delpietro et al., 1997). Application of thispanel to rabies virus collections from many Latin Americancountries, including Brazil (Diaz et al., 1994; Roehe et al.,1997; Morais et al., 2000; Favoretto et al., 2002), has identi-fied two major variants, associated with the dog and thevampire bat (Desmodus rotundus), as well as other variantsassociated with several insectivorous bats, including thefree-tailed bat (Tadarida brasiliensis) and the hoary bat(Lasiurus cinereus). In the Ceara state of Brazil, yet anotherdistinct rabies variant associated with a small non-humanprimate, Callithrix jacchus, commonly referred to as thewhite-tufted-ear marmoset, has been reported (Favorettoet al., 2001). In some cases the classification of certain rabiesvirus isolates by this panel can be confounded by non-typicalreactivity patterns not assigned to any known variant, asfound for certain Argentinian rabies viruses (Cisterna et al.,2005). The application of molecular genetic techniques forcharacterization of viral collections can assist in resolvingsuch typing difficulties; moreover, nucleotide sequencingprovides data amenable for prediction of the evolutionaryrelationships between strains. Several studies, targetingeither partial or complete N gene sequences, have beenreported for isolates from Brazil (Ito et al., 2001a, b, 2003;Romijn et al., 2003; Schaefer et al., 2005), Chile (de Mattoset al., 2000; Yung et al., 2002), Colombia (Paez et al., 2003)and Venezuela (de Mattos et al., 1996). Another study onBrazilian rabies viruses targeted the more variable G gene,encoding the viral glycoprotein, and the highly variable G–Lintergenic region (Sato et al., 2004). All these studies againidentified two principal viral types, associated with dog andvampire-bat hosts, and also established the existence ofother reservoirs in various species of insectivorous bats.Several species of rabid frugivorous bats of the genusArtibeus were found to harbour the viral strain normallyassociated with vampire bats, presumably via spillover fromthis reservoir (Shoji et al., 2004).

In Brazil, rabies is still endemic in many parts of the countryand 29 human rabies cases were reported in 2004 (MinisterioDa Saude, 2004). Up until a few years ago, transmission fromdogs was the most frequent means of human exposure, butreports of bat-transmitted rabies are becoming increasinglycommon (Araujo, 2002). Indeed, 22 of the human rabiesdeaths reported in 2004 occurred in the Amazonian state ofPara, where the population reports frequent vampire-bat bites(www.promedmail.org, archive number 20040527.1428). InBrazil, diagnosis of animal rabies is conducted using theWorld Health Organization approved fluorescent antibodytest (FAT) and the mouse inoculation test (MIT), byapproved state laboratories under the supervision of publicor animal health authorities. However, some Northern andNorth-Eastern states lack their own rabies diagnostic service,so specimens must be sent to laboratories located in otherstates (Gomes, 2004), a situation that probably places

significant limitations on rabies surveillance and diagnosisin these areas. Laboratory testing is performed primarily ondomestic animals; wild animals or captive wild animals aretested only sporadically (OPAS, 2001), and so knowledge ofthe role of wildlife in maintaining rabies reservoirs is verylimited.

The objectives of this study were twofold. First, we wished toexamine the molecular epidemiology of vampire-bat rabiesin Brazil to explore whether regional variation in isolatescould be identified and thereby used to monitor diseasespread. Second, we wished to compare isolates taken fromdomestic animals with those obtained from hoary foxes(Dusicyon vetulus). Up until recently, few positive cases ofrabies were diagnosed in Brazilian hoary foxes (Barros et al.,1989), and it was thought that these cases were due toinfection by dog bites. However, numerous individuals ofthis species from the state of Paraiba have recently beendiagnosed as rabid, and the possible reservoir role played bythis wild species was felt to be worthy of further investiga-tion. To achieve these goals, a molecular epidemiologicalstudy of a collection of viruses from Brazil was undertakenby characterization of isolates at their P gene locus, a highlyvariable region of the genome that previously has beenproven useful for sensitive viral typing and phylogeneticstudies (Nadin-Davis et al., 2002, 2003). Finally, we soughtto develop antigenic-typing tools that would allow for therapid discrimination of rabies virus variants identified bygenetic analysis.

METHODS

Rabies viruses. Fifty brain samples, which were diagnosed asrabies-positive by both the FAT (Dean et al., 1996) and the MIT(Koprowski, 1996), were used in these studies. As shown in Table 1and Fig. 1, these isolates came from several distinct geographicalregions of Brazil and from several animal species. For the purpose ofimporting the bovine isolates into Canada, these viruses were pas-saged once in mice and the infected mice brains were used as thesource of the virus. For other species, virus was recovered from theoriginal brains.

RNA extraction and RT-PCR. Total RNA was recovered fromrabies-positive brain tissue using TRIzol reagent, as recommended bythe manufacturer (Invitrogen). RNA (2 mg) was used to synthesizecDNA in a 20ml reaction, as detailed elsewhere (Nadin-Davis, 1998).Reverse transcription was primed with the oligonucleotide RabP-for,sequence 59-CTACTTCTCCGGGGAAACCAGAAG-39, correspondingto bases 1249–1272 of the positive-sense N gene sequence of the PVstrain (Tordo et al., 1986). For amplification of the complete P gene,5 ml cDNA was used together with the Expand high fidelity system(Roche Diagnostics), according to the manufacturer’s specifications,and reverse primer RabP-rev 59-GGRAGCCAYAGGTCRTCGTCAT-39,corresponding to bases 2575–2596 of the negative-sense M gene sequ-ence of the PV strain. Thermal cycling was performed in an AppliedBiosystems 9700 thermal cycler using the following profile: 93 uC hold,2 min, followed by 35 cycles of 93 uC, 10 s; 48 uC, 1 min; 68 uC, 2 min,with a final 5 min hold at 68 uC. For those samples that did not yield adetectable product, a second round of amplification was performedusing internal primers Pseqfor 59-GAGATGGCAGAGGARACTGTA-GATCT-39 (corresponding to bases 1568–1593 of the PV strain) andPseqrev 59-CCTTAACTATGTCRTCAAGRTTCA-39 (corresponding

3154 Journal of General Virology 86

F. Bernardi and others

Table 1. Rabies viruses employed in these studies

GO, Goias; MG, Minas Gerais; MS, Mato Grosso do Sul; MT, Mato Grosso; PB, Paraiba; SP, Sao Paulo;

TO, Tocantins. Dashes indicate that the year of isolation is unknown.

ID no. Species of origin Year of isolation State of origin GenBank accession no.

V902 Vampire bat 1998 SP AY962047

V903 Dog 1989 SP AY962048

V904 Dog 1989 SP AY962049

V905 Equine 1999 SP AY962050

V906 Bovine 1996 TO AY962051

V907 Bovine 1996 TO AY962052

V908 H. velatus 1997 SP AY962053

V909 Cat 1989 SP AY962054

V911 Donkey 2002 PB AY962055

V912 Equine 2001 PB AY962056

V913 Fox 2002 PB AY962057

V914 Fox 2001 PB AY962058

V915 Fox 2000 PB AY962059

V937 Bovine 1999 MG AY962060

V938 Bovine 2001 SP AY962061

V939 Vampire bat 1998 SP AY962062

V941 Equine 2001 PB AY962063

V943 Vampire bat 2000 SP AY962064

V945 Equine 2002 SP AY962065

V947 Bovine 2000 SP AY962066

V948 Equine 1999 SP AY962067

V951 Bovine 1991 SP AY962068

V953 Ovine 1993 SP AY962069

V959 Bovine 2001 SP AY962070

V962 Vampire bat 1995 SP AY962071

V963 Bovine 2001 SP AY962072

V964 Bovine 1998 MG AY962073

V967 Equine 2002 SP AY962074

V968 Equine 2002 SP AY962075

V969 Dog – GO AY962076

V970 Dog – GO AY962077

V971 Bovine 2002 PB AY962078

V972 Dog – GO AY962079

V973 Cat – MS AY962080

V974 Dog 1989 SP AY962081

V977 Cat 1984 SP AY962082

V978 Bovine – MS AY962083

V981 Dog 1987 MG AY962084

V982 Bovine 2001 MT AY962085

V986 Dog 2002 PB AY962086

V988 Equine 2002 PB AY962087

V990 M. molossus 2002 PB AY962088

V996 Fox 2002 PB AY962089

V997 Fox 2002 PB AY962090

V998 Fox 2002 PB AY962091

V999 Fox 2002 PB AY962092

V1001 Fox 2002 PB AY962093

V1002 Fox 2002 PB AY962094

V1007 Bovine – MS AY962095

V1009 Vampire bat 1995 SP AY962096

http://vir.sgmjournals.org 3155

Rabies in Brazil and the role of the hoary fox

to bases 2208–2231 of the PV strain) and the Expand high fidelitysystem. The cycling profile was similar to that used for the firstround, except that an annealing temperature of 50 uC was employed.The amplified products were purified using the Wizard PCR purifi-cation system (Promega).

Nucleotide sequencing and phylogenetic analysis. PurifiedPCR products were sequenced using a Li-Cor 4200L automated sequenc-ing system, a Thermosequenase cycle sequencing kit (AmershamBiosciences) and custom infared (IR)-dye labelled primers (Li-Cor)corresponding in sequence to the nested PCR primers. A 528 bpregion in the central part of the P gene was targeted. Eseq v2 soft-ware was used for base calling and, after manual review and editing,sequences were saved in FASTA format for subsequent alignmentusing CLUSTALX v1.8 (Thompson et al., 1997) and phylogenetic analy-sis using PHYLIP v3.63 (Felsenstein, 1993). Trees were generated bya neighbour-joining algorithm as detailed previously (Nadin-Daviset al., 2002), and presented graphically using TREEVIEW software(Page, 1996).

Antigenic analysis. Antigenic analysis was undertaken essentiallyas described previously using an indirect FAT applied to virus pro-pagated in murine neuroblastoma cell culture (Nadin-Davis et al.,2001). Representative rabies viruses (see Table 2) were grown inmurine neuroblastoma cell culture and tested individually with473 mAbs; most had anti-N specificity but a few were anti-P speci-fic. Those mAbs exhibiting differential reactivities were furtherexamined on all 50 Brazilian rabies samples, and a panel of 10 mAbscapable of differentiating the viruses represented by all phylogeneticclades was assembled.

RESULTS

Phylogenetic studies

The P gene sequences produced during these studies includethe region encoding amino acids 42–218 and, as noted

Fig. 1. Map of Brazil showing the locations of the states fromwhich samples were collected for this study. States are identi-fied according to the abbreviations defined in Table 1.

Table

2.Rea

ctivity

profiles

forselected

mAbstested

againstseveralBrazilianrabiesviruses

mAb

Viral

strain

andan

tigen

usedformAbgeneration*

Isolate,host

speciesan

dgenetic

group

V903

V909

V915

V913

V914

V908

V990

V1009

V905

V907

V911

V943

V963

V971

V982

Dog

Cat

D.vetulusD.vetulusD.vetulusH.velatusM.molossusD.rotundus

Equine

Bovine

Donkey

D.rotundus

Bovine

Bovine

Bovine

BRL-1a

BRL-1a

BRL-1b

BRL-1c

BRL-1c

BRL-2

BRL-2

BRL-3

BRL-3

BRL-3

BRL-3

BRL-3

BRL-3

BRL-3

BRL-3

11DD1

SAD,live

virus

22

22

22

++

+2

22

22

22

2

24FF11

Dog,

Sri

Lanka,

RNP

++

++

++

+++

+++

++

+2

2+

++

+++

++

++

++

++

++

++

+++

++

+

32FF1

Dog,

Lim

a,RNP

22

22

22

2+

++

++

++

++

++

++

++

++

++

32HD2

Dog,

Lim

a,RNP

++

++

++

+++

+++

++

++

++

++

++

++

+++

++

++

++

++

++

++

+++

++

+

M1318

Mid-A

tlanticraccoon,RNP

++

++

++

+++

+++

++

+2

2+

++

+++

++

++

++

++

++

++

+++

++

+

M1386

Cynictis,RNP

(+)

2+

++

+++

++

++

++

++

++

++

++

+++

++

++

++

++

++

M1495

EBLV-2,RNP

++

22

22

++

++

++

++

2+

++

++

++

++

++

++

++

M1590

SHB,RNP

++

++

++

+++

+++

++

++

++

++

++

2+

+2

++

++

++

++

++

+

M1745

ABLV,RNP

++

++

++

+++

+++

++

++

++

2+

++

++

++

++

++

++

++

+++

++

M1748

ABLV,RNP

++

++

++

2+

++

++

++

++

++

++

++

+++

++

++

++

++

++

++

+++

++

+

*RNP,ribonucleoprotein;SAD,StreetAlabam

aDufferin

strain;SHB,silver-hairedbat

strain.

3156 Journal of General Virology 86

F. Bernardi and others

previously (Nadin-Davis et al., 2002), the portions of thegene encoding residues 61–80 and 134–180 were particularlyvariable in a manner reflecting the phylogenetic relation-ships of the isolates. Phylogenetic analysis revealed threemain clades of Brazilian rabies viruses: these may beclassified as terrestrial (BRL-1), which consists of cases indogs, cats and foxes; insectivorous bat (BRL-2), comprisingjust two specimens in two species of non-haematophagousbat; and vampire bat (BRL-3), which includes all cases inherbivores, a single case in a cat (V977), as well as allvampire-bat-derived specimens (see Fig. 2). Within theterrestrial clade, several further subdivisions are verystrongly supported. For example, a group of ten isolates(BRL-1a) all originated from either dogs or cats, while thesix specimens of subgroup BRL-1b, as well as a group ofthree isolates forming subgroup BRL-1c, all representisolates from foxes of the Paraiba region. The segregation

of the two fox clades, labelled ‘fox 1’ (1b) and ‘fox 2’ (1c),from the domestic clade is very highly supported bybootstrap values (1000 and 962, respectively).

Although the viruses from the two non-haematophagousbats clustered together with strong bootstrap support (952),the genetic distance between these two isolates (0?138)would, based on guidelines proposed previously (Nadin-Davis et al., 2002), support the placement of these twoviruses into distinct lineages. However, the investigation ofthe role of these two chiropteran species, Histiotus velatusand Molossus molossus, as rabies reservoirs will requirefurther viral isolations and type determinations from thesespecies.

The BRL-3 viruses represent the variant that circulates inBrazilian vampire bats with frequent spillover into domestic

Fig. 2. Phylogeny of Brazilian rabies virusstrains. Fifty Brazilian isolates were charac-terized by partial nucleotide sequencing(528 bp) of the P gene coding region. Analignment of these data, together with theCVS sequence used as an outgroup, wasgenerated using CLUSTALX. Phylogenies werepredicted from the aligned sequences usinga neighbour-joining algorithm in the PHYLIP

(version 3.63) software package. Bootstrapvalues, determined by using 1000 replicatesof the data, indicate the number of timesthat the clade to the right of the branch ispredicted in the consensus tree. The mainBrazilian groupings referred to in the text areindicated to the right of the tree. Tree bran-ches reflect the genetic distance betweenisolates according to the scale shown at thebottom of the figure. GenBank accessionnumbers are given in parenthesis.

http://vir.sgmjournals.org 3157

Rabies in Brazil and the role of the hoary fox

species, particularly livestock such as bovines and equines.Although this group could be further subdivided into severalsmaller clusters that segregated with moderate to strongbootstrap support, these subdivisions did not exhibit anyobvious trends with respect to either temporal or regionallocalization of subtypes.

To place the Brazilian strains identified in this report into amore global context, selected viral sequences were comparedwith representative viruses from throughout the Americasin another phylogenetic analysis. As shown in Fig. 3, all ofthe BRL-3 viruses clustered within a clade that includedvampire-bat rabies viruses from other Latin Americancountries (Paraguay and Mexico), and Trinidad in theCaribbean. The strongly supported monophyletic nature ofthis clade indicates that all these vampire-bat-derived viruseshave a common origin and a progenitor shared with the

free-tailed-bat-derived strain represented in this tree by theV235.FTB isolate from Texas. Notably there was no strongassociation of any Brazilian vampire-bat-derived isolates toa specific lineage within this clade, further indicating the lackof any clear regional variation within this strain. The twoBrazilian insectivorous-bat-derived isolates (group BRL-2)again grouped together and within a region of the treerepresenting many insectivorous-bat-derived strains recov-ered primarily in Canada. While the Brazilian variants werenot strongly associated with any particular variant fromCanada, it is apparent that these viruses are evolutionarilymore closely related to the North-American-bat-derivedstrains than to the vampire-bat-derived strain. The BrazilianBRL-1 viruses all clustered within a clade representing manyterrestrial strains of the Americas, including dog-derivedisolates from Peru, Paraguay and Mexico, the grey-fox-and coyote-derived strains from Texas in the USA, the

Fig. 3. Phylogeny showing the relationshipsof representative Brazilian rabies viruses toother rabies virus strains of the Americas. A528 bp sequence from the central portion ofthe P gene coding region was used to com-pare 11 representative Brazilian isolates and29 other isolates, all of which were describedpreviously (Nadin-Davis et al., 2002), exceptfor sample 3306.99RAC, which is an isolateof the mid-Atlantic-raccoon-derived strainrecovered from Ontario, Canada, in 1999. Thetree depicts a neighbour-joining analysisemploying two members of the ARCTIC line-age from Canada as an outgroup. Geneticdistances between isolates are reflected inbranch lengths according to the scale at thebottom of the figure. GenBank accessionnumbers are given in parenthesis.

3158 Journal of General Virology 86

F. Bernardi and others

western-Canadian-skunk-derived strain, and an isolaterepresentative of mongoose rabies on the island of PuertoRico. Samples of this clade were previously assigned to agrouping known as the cosmopolitan lineage (Nadin-Daviset al., 2002).

Antigenic discrimination

Of the 473 mAbs tested, 10 were selected based on theirability to differentiate between the Brazilian rabies virusvariants as identified by genetic analysis. Their reactivitieswith representative viruses of the genetic groupings areindicated in Table 2, and the following observations are ofparticular note. mAbs 11DD1 and M1745 differentiate thesingle M. molossus specimen from all others, while bothBRL-2 specimens were exceptional by not reacting to mAb24FF11, 32FF1 reacted to all BRL-3 viruses but not to thoseof the other groups, M1386 reacted weakly if at all to BRL-1aviruses but was positive to all other groups, while M1748reacted with all types except group BRL-1b. Thus, when usedin combination with the other mAbs of the panel, mAbsM1386 and M1748 could be used to discriminate the threesubgroups of clade BRL-1 viruses. Some of these mAbs(M1495, M1590) reacted differentially to certain vampire-bat-derived isolates but these specificities did not correlatewith the genetic relationship of the specimens.

DISCUSSION

The continued characterization of rabies viruses in countriesof the American continent is necessary to more fully definethe extent of virus variation, and to assist in the identifi-cation of all reservoir species involved in maintaining thedisease at the regional level. The situation must be regardedas a continually evolving process in light of evidence thatrabies virus spillover into new hosts can, under specificcircumstances, lead to viral adaptation to the new host,thereby resulting in the emergence of new viral–hostrelationships (Badrane & Tordo, 2001) and even new viralbiotypes (Fooks, 2004).

In this study, phylogenetic analysis of a collection ofBrazilian viruses employed the variable, central region of theP gene. Although the choice of genomic target employed forphylogenetic studies of lyssaviruses does not, in general,greatly impact on the general conclusions of the studies,greater variability within the database provides for a morerobust and sensitive analysis, and hence, due to higherbootstrap values, more strongly supported conclusions. Thecentral portion of the P gene is amongst the more variablecoding regions of the lyssavirus genome (Le Mercier et al.,1997), second only to the region encompassing the 39terminus (coding sense) of the G gene which, together withthe contiguous, non-coding G–L intergenic region, has beenused for molecular epidemiological studies (Nel et al., 1997;Paez et al., 2003, 2005). Similarly, use of the P gene forsensitive and robust phylogenetic studies has been pre-viously reported (Nadin-Davis et al., 2003). An extensivedatabase of lyssavirus P gene sequences is now publicly

available to provide comparative data (Nadin-Davis et al.,2002). Moreover, due to the multifunctional nature of thelyssavirus P protein, including its ability to interact withhost-cell proteins (Poisson et al., 2001), exploration ofstructural variations that may confer some measure of hostadaptation should be continued (Nadin-Davis et al., 2002).

While two principal rabies cycles, maintained by dogs andvampire bats, have been well established in many parts ofLatin America, including Brazil (Ito et al., 2001a), recentreports indicate that the situation is much more complex,particularly with respect to the role played by insectivorousbats, as reported in studies undertaken in Argentina andChile (de Mattos et al., 2000; Cisterna et al., 2005), and alsoin Colombia (Paez et al., 2003). Although this study did notfocus on the role of insectivorous bats as rabies reservoirs,and our sample set included only two rabies isolates fromnon-haematophagous bats (H. velatus and M. molossus),which comprised group BRL-2, these isolates were quitedistinct from all others, and apparently represented twodifferent viral variants. In a continental context, theseisolates were evolutionarily more closely related to speci-mens from North American insectivorous bats than toBrazilian vampire bats. Further isolation of viruses fromthese two chiropteran species will be necessary to establishthese bats as the reservoirs of these rabies variants. However,it should be noted that a group of rabies isolates recoveredfrom species of the genus Histiotus from Chile andArgentina, and which exhibited a distinctive antigenicprofile, formed a monophyletic group that may represent apreviously unidentified reservoir (Yung et al., 2002; Cisternaet al., 2005). The histiotus-derived specimen in this studymay represent this same variant, but unfortunately, sincethe N gene was targeted in those earlier studies, directcomparison between those histiotus isolates and the onereported in this study was not possible. Moreover, a case ofrabies in a M. molossus specimen in Colombia yielded arabies variant that clearly segregated together with otherrabies variants associated with insectivorous bats of theAmericas (Paez et al., 2003), thereby strengthening thepossibility that this bat species may act as a rabies reservoir.

The Brazilian isolates representative of the vampire-bat-derived strain were relatively homogeneous, and clusteredtogether with vampire-bat-derived isolates recovered fromseveral other countries in Latin America and the Caribbean,an observation supporting the concept of a common originof all isolates of this strain despite its extensive geographicalrange. Despite a fairly extensive sampling of this variant,no obvious temporal or spatial trends with regards to theemergence of Brazilian subvariants were identified. Addi-tionally, no consistent differences in reactivity with a largemAb panel could be discerned for viruses of this group,supporting the conclusion that no clear regional variants ofthis strain circulate.

Spillover of a rabies virus strain from its reservoir host toother species is not uncommonly reported, as documented,for example, between wildlife in South Africa (Nel et al.,

http://vir.sgmjournals.org 3159

Rabies in Brazil and the role of the hoary fox

1997), for the raccoon rabies strain to skunks in the UnitedStates (Guerra et al., 2003), and even more recently inNorthern Colombia, where cases of rabies in humans, dogsand grey foxes were due to a single genetic variant (Paezet al., 2005). In addition, there are multiple examples in theliterature that clearly indicate the long-term emergence ofindependent cycles of disease in new reservoir hosts sub-sequent to such spillover events. For example, Bourhy et al.(1999) presented evidence that during the westward move-ment of rabies across Europe during the early 20th century,rabies crossed species from the dog to become established inthe red fox population. Johnson et al. (2003) reported on theapparent recent transmission of rabies virus variants fromdogs to foxes in Turkey, while the existence of two indepen-dent cycles of transmission involving foxes and domesticanimals has been reported within the Federal Republic ofYugoslavia (Stankov, 2001). Indeed, the present situation inNorthern Colombia, which involves dogs and grey foxes,may represent the very initial stages of such a species jump,which, without intervention, might eventually lead to theemergence of a new fox-adapted strain.

Our genetic analysis of all Brazilian isolates of terrestrial-host origin (see Fig. 2) defined onemajor group (BRL-1) forwhich further division into three subgroups was stronglysupported. All isolates recovered from domestic animals,including V986, which came from a dog of Paraiba state,belonged to subgroup BRL-1a, while subgroups BRL-1b andBRL-1c were associated exclusively with hoary foxes.Despite the limited number of isolates studied, these resultsclearly support the existence of genetically distinct strains ofrabies, derived from a common ancestor, that now circulateindependently in dogs and hoary foxes. The phylogeneticanalysis presented in Fig. 3 indicates that the Brazilian BRL-1 rabies viruses cluster within the cosmopolitan lineagebelieved to have been introduced into the Americas duringcolonial times, probably by transportation of infected dogsfrom Europe (Nadin-Davis & Bingham, 2004). Of thespecimens included in this study, the isolate most closelyrelated to the Brazilian viruses came from a dog in Paraguay,suggesting a common origin for the viruses circulating inthese neighbouring countries. Notably the Brazilian fox-derived strain, represented by isolates V997 and V1001 inFig. 3, does not associate closely with the Texas grey-fox-derived strain, represented in this tree by isolate V224.FX,thereby suggesting that these two fox reservoirs haveemerged independently from the progenitor of thecosmopolitan lineage. Our identification of the hoary foxof the Paraiba region of Brazil as a rabies reservoir maintain-ing a viral strain evolutionarily related to the urban rabiesvariant also circulating in Brazil thus appears to mirrorsituations reported elsewhere, particularly in Europe, wheredog to fox transmission, followed by persistence in thewildlife reservoir, has been documented. While it cannot beinferred from our data whether these dog- and fox-derivedstrains were originally urban strains that were subsequentlytransmitted to the sylvatic reservoir, historically transmis-sion from dog to fox appears to be the more common

occurrence. The fact that the hoary fox plays a much moreextensive role in the maintenance and dissemination ofrabies in Brazil than was previously supposed has significantpublic health implications since in this region, where publicawareness of rabies is low and animal vaccination rarelyundertaken, hoary foxes are not infrequently raised as pets.

To maximize the effectiveness of rabies-control program-mes, strain-typing regimens that identify the reservoirsresponsible for disease outbreaks are an essential tool, andseveral methods currently exist. Genetic methods employingeither PCR and nucleotide sequencing (Bordignon et al.,2005), strain-specific RT-PCR and restriction fragmentlength polymorphism analysis (Ito et al., 2003), or multiplexPCR (Sato et al., 2005) have all been applied to collections ofBrazilian isolates. However, to date, these methods discrimi-nate only between dog-related and vampire-bat-related virusvariants, and these methods require specialized technicalfacilities and expertise. Systematic typing by geneticmethodsis time consuming and costly and, even in laboratories ofdeveloped countries, is performed on selected cases only.Routine application of antigenic-typing methods, employ-ing a limited panel of mAbs in an indirect FAT procedure, ismore practical, especially in developing countries. Thus,there is a need for a rational approach to the development ofstrain-typing methods in which genetic characterization ofrepresentative isolates can be used to direct the identifica-tion of mAbs exhibiting reactivity patterns that will differ-entiate between distinct rabies variants. Such an approachhas been described here in which a panel of mAbs, capable ofdiscriminating between the distinct Brazilian viral variantsidentified by genetic methods, was developed. In particular,the ability to discriminate between the fox- and dog-associated variants may be of considerable importance tofuture control efforts, where the roles of both species in thecontinued maintenance of the disease should be clearlyestablished, and the range of the variant associated with foxpopulations needs to be better defined. The mAbs describedheremay be a useful addition to the current CDCmAb panel(Diaz et al., 1994) being employed for strain discrimination.Indeed the utility of the mAbs described in this report toidentify other rabies variants unavailable for these studies(e.g. isolates from the marmoset primate of the Ceara regionthat neighbours Paraiba state, and from other insectivorousbat species) should be the subject of future investigations.

ACKNOWLEDGEMENTS

We thank the Inter-American Institute for Cooperation on Agriculture(IICA) for providing assistance to Dr Fernanda Bernardi to work as acollaborating visiting scientist at the Rabies Centre of Expertise, OttawaLaboratory-Fallowfield, Canadian Food Inspection Agency.

REFERENCES

Araujo, F. A. A. (2002). Raiva humana no Brasil: 1992-2001. (Humanrabies in Brazil: 1992-2001). 90 f. Master in Veterinary Medicine

3160 Journal of General Virology 86

F. Bernardi and others

thesis - Escola de Veterinaria, Universidade Federal de Minas Gerais,

Belo Horizonte, Minas Gerais, Brazil.

Badrane, H. & Tordo, N. (2001). Host switching in Lyssavirus historyfrom the chiroptera to the carnivore orders. J Virol 75, 8096–8104.

Barros, J. S., de Freitas, C. E. A. A. & de Sousa, F. S. (1989). Raivaem animais silvestres no Estado do Ceara particularmente na raposa

(Dusicyon vetulus). Zoonoses Rev Int 1, 9–13.

Bordignon, J., Brasil-dos-Anjos, G., Bueno, C. R., Salvatiera-Oporto, J., Davila, A. M. R., Grisard, E. C. & Zanetti, C. R. (2005).Detection and characterization of rabies virus in Southern Brazil byPCR amplification and sequencing of the nucleoprotein gene. Arch

Virol 150, 695–708.

Bourhy, H., Kissi, B. & Tordo, N. (1993). Molecular diversity of the

Lyssavirus genus. Virology 194, 70–81.

Bourhy, H., Kissi, B., Audry, L., Smreczak, M., Sadkowska-Todys,M., Kulonen, K., Tordo, N., Zmudzinski, J. F. & Holmes, E. C. (1999).Ecology and evolution of rabies virus in Europe. J Gen Virol 80,2545–2557.

Cisterna, D., Bonaventura, R., Caillou, S. & 10 other authors (2005).Antigenic and molecular characterization of rabies virus in

Argentina. Virus Res 109, 139–147.

Dean, D. J., Ableseth, M. K. & Atanasiu, P. (1996). The fluorescentantibody test. In Laboratory Techniques in Rabies, 4th edn, pp. 88–95.

Edited by F. X. Meslin, M. M. Kaplan & H. Koprowski. Geneva:World Health Organization.

Delpietro, H. A., Gury-Dhomen, F., Larghi, O. P., Mena-Segura, C. &Abramo, L. (1997). Monoclonal antibody characterization of rabies

virus strains isolated in the River Plate Basin. Zentrabl VeterinaermedB 44, 477–483.

De Mattos, C. A., De Mattos, C. C., Smith, J. S., Miller, E. T., Papo, S.,Utrera, A. & Osburn, B. I. (1996). Genetic characterization of rabiesfield isolates from Venezuela. J Clin Microbiol 34, 1553–1558.

De Mattos, C. A., Favi, M., Yung, V., Pavletic, C. & De Mattos, C. C.(2000). Bat rabies in urban centers in Chile. J Wildl Dis 36, 231–240.

Diaz, A. M., Papo, S., Rodriguez, A. & Smith, J. S. (1994). Antigenicanalysis of rabies-virus isolates from Latin America and Caribbean.Zentralbl Veterinaermed B 41, 153–160.

Favoretto, S. R., De Mattos, C. C., Morais, N. B., Alves Araujo, F. A.& De Mattos, C. A. (2001). Rabies in marmosets (Callithrix jacchus),

Ceara, Brazil. Emerg Infect Dis 7, 1062–1065.

Favoretto, S. R., Carrieri, M. L., Cunha, E. M. S., Aguiar, E. A. C.,Silva, L. H. Q., Sodre, M. M., Souza, M. C. A. M. & Kotait, I. (2002).Antigenic typing of Brazilian rabies virus samples isolated fromanimals and humans, 1989-2000. Rev Inst Med Trop Sao Paulo 44,

91–95.

Felsenstein, J. (1993). PHYLIP: phylogeny inference package.

(Version 3.52c), Department of Genome Sciences, University ofWashington, Seattle, WA, USA.

Fooks, A. R. (2004). The challenge of new and emerging lyssaviruses.

Expert Rev Vaccines 3, 333–336.

Gomes, A. A. B. (2004). Epidemiologia da raiva: caracterizacao de

vırus isolados de animais domesticos e silvestres do semi-aridoparaibano da regiao de Patos, Nordeste do Brasil. (Epidemiology of

rabies: characterization of viruses isolated from domestic and wildanimals of the semi-arid region of Patos, state of Paraıba, North-

Eastern Brazil). 107 f. Doctorate in Veterinary Medicine thesis -Faculty of Veterinary Medicine and Zootechny, University of Sao

Paulo, Sao Paulo.

Guerra, M. A., Curns, A. T., Rupprecht, C. E., Hanlon, C. A., Krebs,J. W. & Childs, J. E. (2003). Skunk and raccoon rabies in the eastern

United States: temporal and spatial analysis. Emerg Infect Dis 9,1143–1150.

Ito, M., Arai, Y. T., Itou, T., Sakei, T., Ito, F. H., Takasaki, T. & Kurane,I. (2001a). Genetic characterization and geographic distribution of

rabies virus isolates in Brazil: identification of two reservoirs, dogs

and vampire bats. Virology 284, 214–222.

Ito, M., Itou, T., Sakai, T., Santos, M. F. C., Arai, Y. T., Takasaki, T.,Kurane, I. & Ito, F. H. (2001b). Detection of rabies virus RNA isolated

from several species of animals in Brazil by RT-PCR. J Vet Med Sci

63, 1309–1313.

Ito, M., Itou, T., Shoji, Y., Sakai, T., Ito, F. H., Arai, Y. T., Takasaki, T.& Kurane, I. (2003). Discrimination between dog-related and

vampire bat-related rabies viruses in Brazil by strain-specific reverse

transcriptase-polymerase chain reaction and restriction fragment

length polymorphism analysis. J Clin Virol 26, 317–330.

Johnson, N., Black, C., Smith, J., Un, H., McElhinney, L. M., Aylan, O.& Fooks, A. R. (2003). Rabies emergence among foxes in Turkey.

J Wildl Dis 39, 262–270.

Kissi, B., Tordo, N. & Bourhy, H. (1995). Genetic polymorphism in

the rabies virus nucleoprotein gene. Virology 209, 526–537.

Koprowski, H. (1996). The mouse inoculation test. In Laboratory

Techniques in Rabies, 4th edn, pp. 80–87. Edited by F. X. Meslin,

M. M. Kaplan & H. Koprowsky. Geneva: World Health

Organization.

Krebs, J. W., Wheeling, J. T. & Childs, J. E. (2003). Rabies surveill-

ance in the United States during 2002. J Am Vet Med Assoc 223,

1736–1748.

Le Mercier, P., Jacob, Y. & Tordo, N. (1997). The complete Mokola

virus genome sequence: structure of the RNA-dependent RNA

polymerase. J Gen Virol 78, 1571–1576.

Ministerio Da Saude (2004). Programa nacional de profilaxia da

raiva. Casos de raiva humana notificados, e percentual de casos

transmitidos segundo a especie animal. Brasılia, 2004. Brazilian

Ministry of Health annual disease report. http://portal.saude.gov.br/

portal/svs/visualizar_texto.cfm?idtxt=21906

Morais, N. B., Rolim, B. N., Chaves, H. H. M., Brito-Neto, J. & Silva,L. M. (2000). Rabies in tamarins (Callithrix jacchus) in the State of

Ceara, Brazil, a distinct viral variant? Mem Inst Oswaldo Cruz 95,

609–610.

Nadin-Davis, S. A. (1998). Polymerase chain reaction protocols for

rabies virus discrimination. J Virol Methods 75, 1–8.

Nadin-Davis, S. A. & Bingham, J. (2004). Europe as a source of

rabies for the rest of the world. In Historical Perspective of Rabies in

Europe and the Mediterranean Basin, pp. 259–280. Edited by A. A.

King, A. R. Fooks, M. Aubert & A. I. Wandeler. Paris: OIE Press.

Nadin-Davis, S. A., Huang, W., Armstrong, J., Casey, G. A., Bahloul,C., Tordo, N. & Wandeler, A. I. (2001). Antigenic and genetic

divergence of rabies viruses from bat species indigenous to Canada.

Virus Res 74, 139–156.

Nadin-Davis, S. A., Abdel-Malik, M., Armstrong, J. & Wandeler, A. I.(2002). Lyssavirus P gene characterization provides insights into the

phylogeny of the genus and identifies structural similarities and

diversity within the encoded phosphoprotein. Virology 298, 286–305.

Nadin-Davis, S. A., Simani, S., Armstrong, J., Fayaz, A. & Wandeler,A. I. (2003). Molecular and antigenic characterization of rabies

viruses from Iran identifies variants with distinct epidemiological

origins. Epidemiol Infect 131, 777–790.

Nel, L., Jacobs, J., Jaftha, J. & Meredith, C. (1997). Natural spilloverof a distinctly canidae-associated biotype of rabies virus into an

expanded wildlife host range in southern Africa. Virus Genes 15,

79–82.

Organizacion Pan-Americana de la Salud (OPAS) (2001). Boletın:vigilancia epidemiologica de la rabia em las Americas. XXXIII,

pp. 40. Rio de Janeiro: Organizacion Pan-Americana de la Salud.

http://vir.sgmjournals.org 3161

Rabies in Brazil and the role of the hoary fox

Paez, A., Nunez, C., Garcıa, C. & Boshell, J. (2003). Molecular

epidemiology of rabies epizootics in Colombia: evidence for human

and dog rabies associated with bats. J Gen Virol 84, 795–802.

Paez, A., Saad, C., Nunez, C. & Boshell, J. (2005). Molecular

epidemiology of rabies in northern Colombia 1994-2003. Evidence

for human and fox rabies associated with dogs. Epidemiol Infect 133,

529–536.

Page, R. D. M. (1996). TREEVIEW: an application to display

phylogenetic trees on personal computers. Comput Appl Biosci 12,

357–358.

Poisson, N., Real, E., Gaudin, Y., Vaney, M.-C., King, S., Jacob, Y.,Tordo, N. & Blondel, D. (2001). Molecular basis for the interaction

between rabies virus phosphoprotein P and the dynein light chain

LC8: dissociation of dynein-binding properties and transcriptional

functionality of P. J Gen Virol 82, 2691–2696.

Roehe, P. M., Pantoja, L. D., Shaefer, R., Nardi, N. B. & King, A. A.(1997). Analysis of Brazilian rabies isolates with monoclonal

antibodies to lyssavirus antigens. Rev Microbiol 28, 288–292.

Romijn, P. C., Van der Heide, R., Cattaneo, C. A., Silva, R. D. E. C. &Van der Poel, W. H. (2003). Study of lyssaviruses of bat origin as a

source of rabies for other animal species in the State of Rio de

Janeiro, Brazil. Am J Trop Med Hyg 69, 81–86.

Sato, G., Itou, T., Shoji, Y. & 9 other authors (2004). Genetic and

phylogenetic analysis of glycoprotein of rabies virus isolated from

several species in Brazil. J Vet Med Sci 66, 747–753.

Sato, G., Tanabe, H., Shoji, Y., Itou, T., Ito, F. H., Sato, T. & Sakai, T.(2005). Rapid discrimination of rabies viruses isolated from various

host species in Brazil by multiplex reverse transcription-polymerasechain reaction. J Clin Virol 33, 267–273.

Schaefer, R., Batista, H. B. R., Franco, A. C., Rijsewijk, F. A. M. &Roehe, P. M. (2005). Studies on antigenic and genomic properties ofBrazilian rabies virus isolates. Vet Microbiol 107, 161–170.

Shoji, Y., Kobayashi, Y., Sato, G. & 10 other authors (2004). Geneticcharacterization of rabies viruses isolated from frugivorous bat(Artibeus spp.) in Brazil. J Vet Med Sci 666, 1271–1273.

Stankov, S. (2001). Typing of field rabies virus strains in FRYugoslavia by limited sequence analysis and monoclonal antibodies.Med Pregl 54, 446–452.

Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. &Higgins, D. G. (1997). The Clustal_X windows interface: flexiblestrategies for multiple sequence alignment aided by quality analysistools. Nucleic Acids Res 25, 4876–4882.

Tordo, N., Poch, O., Ermine, A., Keith, G. & Rougeon, F. (1986).Walking along the rabies genome: is the large G-L intergenic region aremnant gene? Proc Natl Acad Sci U S A 83, 3914–3918.

Tordo, N., Charlton, K. & Wandeler, A. (1998). Rhabdoviruses: rabies.In Topley and Wilson’s Microbiology and Microbial Infections,pp. 666–692. Edited by L. H. Collier. London: Arnold Press.

Velasco-Villa, A., Gomez-Sierra, M., Hernandez-Rodrıguez, G.,Juarez-Islas, V., Melendez-Felix, A., Vargas-Pino, F., Velazquez-Monroy, O. & Flisser, A. (2002). Antigenic diversity and distributionof rabies virus in Mexico. J Clin Microbiol 40, 951–958.

Yung, V., Favi, M. & Fernandez, J. (2002). Genetic and antigenictyping of rabies virus in Chile. Arch Virol 147, 2197–2205.

3162 Journal of General Virology 86

F. Bernardi and others