Molecular and Cellular Probes (1991) 5, 2 29-240

PCR technique as an alternative method for diagnosis and molecular epidemiology of rabies virus Debora Sacramento, Herve Bourhy and Unite de

la

Rage, Institut Pasteur, 25, rue du Docteur Roux,

Noel Tordo* 75724

Paris Cedex 15, France

(Received 13 December 1990, Accepted 22 January 1991) We have investigated the PCR amplification technique of viral nucleic acids as an alternative protocol for diagnosis and epidemiological studies of rabies virus . A primer set mapping in the nucleoprotein cistron allowed a specific and sensitive amplification of infected brain material, fulfilling the diagnosis requirements . One hundred samples checked by Southern or dot-blot analysis using both radioactive and non-radioactive probes showed identical results in parallel with routine techniques . For molecular epidemiological studies we selected another set of conserved primers flanking the highly evolutive pseudogene (tP gene) region . This set was found to be efficient for all tested fixed rabies virus strains or wild rabies virus isolates as well as the rabies-related Mokola virus . We describe a progressive characterization of the strain that could be extended from rapid typing by a limited panel of restriction enzymes, to the ultimate identification of the nucleotide sequence by an original direct sequencing technique of amplified segments . KEYWORDS : Lyssaviruses, rabies, PCR, diagnosis, typing, molecular epidemiology . INTRODUCTION The two main problems for rabies survey are the detection of infected samples and the typing of infecting strains for epidemiological studies . Both human rabies post-exposure treatment and sanitary prophylactic measures depend on these laboratory results . During the last 30 years significant advances have been made in the diagnosis of rabies virus .' The histological diagnosis of Negri bodies in the infected brain' has been replaced by the direct detection of viral nucleocapsid inclusions on brain smears with a fluorescent polyclonal antibody . This fluorescentantibody test (FAT) has become a basic method for routine rabies diagnosis . In addition, viral isolation either by newborn mouse intracerebral inoculation test (MIT) or by the rabies tissue culture infection test (RTCIT) is also performed to allow further analysis of the etiological agent . RTCIT has reduced viral isolation from 5-11 to only 1 day, particularly when highly susceptible murine neuroblastoma cell cultures (N2a)

are used . Such a rapidity promoted the RTCIT from a simple isolation technique to a confirmatory test for diagnosis . More recently, a method based on the ELISA technique, the rapid rabies enzyme immunodiagnosis (RREID) was developed .' Each of these methods has its merits and disadvantages . For example, FAT is rapid and inexpensive but requires the availability and maintenance of u .v. microscopes, and suitably trained personnel . MIT is neither rapid nor ethical, whereas the RTCIT is fast but expensive and necessitates technical expertise in cell culture . Trials in typing the rabies strains were originally performed by seroneutralization in mice' or in vitro .' However, only an approximate distinction could be obtained between rabies and rabies-related viruses . The . availability of monoclonal antibodies has made possible the definition of more precise antigenic differences distinguishing serotypes and even subtypes .~' 2

* Author to whom correspondence should be addressed .

0890-8508/91/030229 + 12 $03 .00/0

229

1991 Academic Press Limited

230

D. Sacramento et at.

The recent application of cloning and sequencing techniques to rabies genes has permitted substantial advances in the understanding of the molecular structure and mechanisms of the virus ." , " The rabies virus belongs to the Lyssavirus genus of the Rhabdoviridae family and consists of an unsegmented negative stranded RNA genome encoding successively the N, M1, M2, G, `I' and L genes ."" Up to now, however, few investigations have attempted to apply these powerful techniques for diagnostic purposes . A single report describing hybridization analysis of dotblotted samples, with a complex mixture of rabies specific cDNA probes, has been published . 20 However, this study lacks the required statistical import, being restricted to a small panel of preselected samples . A preliminary investigation of the polymerase chain reaction (PCR) technique was also recently presented in experimentally infected mouse brain with a laboratory rabies strain ." In this paper, we present a complete investigation of the PCR technique for the purpose of rabies diagnosis, from the systematic analysis of 100 samples in blind and parallel with classical diagnosis methods . In addition, the PCR proves to be a powerful tool for molecular epidemiology studies, allowing analysis of the infecting strain without prior cell culture adaptation .

MATERIALS AND METHODS

Cells and viruses Baby hamster kidney cells (BHK-21) were grown in Eagle's MEM supplemented with 2% NaHCO3 and 10% foetal calf serum . Confluent monolayers were infected at a multiplicity of one particle forming unit per cell with the following rabies strains : Louis Pasteur Virus (PAS), Pasteur Virus (PV-11), Challenge Virus Strain (CVS-11), Pitman-Moore (PM), Evelyn-Rokitnicki-Abelseth (ERA), Avirulent Orsay 1 (AvO-1), as well as the rabies-related Mokola virus (MOK 5) . The origin and passage history of the strains has been described elsewhere .22,23 Tordo and Sacramento (submitted) . After 1 h at 35 ° C, the inoculum was removed and the cells incubated for 48-72 h at 37 ° C.

contamination . Samples (approximately 2 g) were isolated in the north and east of France between October 1989 and May 1990 and were stored deep frozen at -80 ° C in Eppendorf tubes .

Oligodeoxynucleotides Oligodeoxynucleotides were synthesized in DNA Synthesizers (Applied Biosystems, Pharmacia) . Two sets of specific primers were selected in very stable regions of the genome to allow suitable amplification of any rabies and rabies-related strains (see Fig . 6) : N1 (+) sense : (587) 5'-TTT GAG ACT GCT CCT TTT G-3' (605) N2 (-) sense: (1029) 5'-CC CAT ATA GCA TCC TAC-3' (1013) G (+) sense : (4665) 5'-GAC TTG GGT CTC CCG AAC TGG GG-3' (4687) L (-) sense : (5543) 5'-CAA AGG AGA GTT GAG ATT GTA GTC-3' (5520) The first set N1-N2, mapping in the nucleoprotein cistron, was employed for diagnosis purposes . The second set (G-L), flanking the non-coding `Y gene allowed analyses of the evolutionary relationships between strains and also molecular epidemiological studies .

Probes To obtain specific probes, DNA fragments internal to the amplified segments, were prepared from previously described plasmids containing rabies or Mokola gene insertions ."" The N probe was a Rsa I-Pvu II fragment in position 665-908 in the PV strain rabies genome. The `I probe was either a Bgi II-Hin dill fragment in position 5083-5341 in the PV strain rabies genome or a Hpa 11-Nin dll fragment in position 4933-5406 of the Mokola genome . The probes were either 32 P-labelled using the Multiprime labelling kit (Amersham) or non-radioactively labelled with digoxigenin-dUTP using the DNA labelling and detection kit non-radioactive (Boehringer) . In both cases, the protocols for probe labelling, hybridization conditions and immunological detection were as defined by the manufacturers' instructions .

Collection of animal samples

RNA extraction

One hundred internal brain samples collected by the National Reference Centre for Rabies, Pasteur Institute (Paris, France) were obtained via the occipital foramen 24 using a disposable plastic pipette to avoid

After removing culture medium, cells were washed three times in PBS and covered with extraction buffer containing 1 % SDS, 1 % NP40, 1 mm EDTA pH = 8 . 0, 50µg ml -1 dextran sulphate . The cell lysate was

PCR amplification of rabies genes extracted with an equal volume of phenol-hydroxyquinoline25 and centrifugated at 13,000 g for 30 min at room temperature . Identical successive extractions with phenol-chloroform (50/50) and chloroform were performed until material at the aqueous-organic interface disappeared . In this aim, one phenol, two phenol-chloroform and one chloroform extraction were usually sufficient . The aqueous phase was then adjusted to 0 . 3 M in sodium acetate pH=5 . 2, and the total RNA was precipitated at -20° C by adding two volumes of ethanol and pelleted by centrifugation. After three washings in 70% ethanol, the pellets were dried and resuspended in pyrolysed water at 1 µg µl - ' as estimated by measuring the absorbance at 260 nm . Identical procedures were employed for brain samples, except that material was homogenized in the extraction buffer using a plastic pestle .

cDNA synthesis Total brain or cellular RNA (1 µg) was hybridized to 100 ng (approximately 15 pmoles) of either N2, N1 or G primer and reverse transcribed for 90 min at 42 ° C in 10 µl containing 200 units of Moloney murine leukemia virus reverse transcriptase (Bethesda Research Laboratories, buffer supplied), 1 mm of each deoxynucleotide and 4 units of RNasin (Promega Biotec) . The reaction mixture was then diluted 10-fold in TE buffer (10 mm Tris-HCI pH =8 . 3, 1 mm EDTA) .

231

Direct analysis of PCR-amplified products Aliquots (2 to 10µl) of each amplification mixture were analysed either directly or after digestion by suitable restriction enzymes on 1 . 2% to 1 . 5% agarose gel containing ethidium bromide and then photographed . Barn HI, Hin dil, Hin dill, Pst I and Rsa I restriction enzymes were used .

Southern blot analysis of PCR-amplified products with 32 P-labelled probes One tenth (10 µD of each amplification mixture was separated on 1 . 2% agarose gel containing ethidium bromide, photographed and transferred onto ZetaProbe nylon membranes (Bio-Rad) using 0 . 4 N NaOH . The membranes were pre-incubated for 2 h at 65 ° C in 0 . 5 M NaH 2 P0 4-Na2 HPO, pH =7 . 5, 1 mm EDTA, 7% SDS, 1 % Bovine Serum Albumin (BSA) and hybridized overnight at 65 ° C in the same solution containing 32 Plabelled probes (approximately 2 x 10 5 cpm ml -1 ) . After three successive washings for 15 min at 65 ° C in 40 mm NaH 2 PO g Na 2 HPO, pH = 7 . 5, 1 mm EDTA, 1 % SDS, the blots were dried and exposed to Kodak XAR5 film with intensifier screens at -70 ° C for 1-4 h .

Dot-blot analysis of PCR-amplified products with non-radioactive probes One tenth (10 µp of amplified product was diluted in 90 pi of TE, denaturated at 95 ° C for 10 min, chilled on

PCR reaction Diluted RNA/cDNA hybrid (10 RD was amplified in 100lal of reaction mixture containing 1 µM of each primer, 200 µM of each dNTP, 10 mm Tris-HCI pH =8. 3, 50 mm KCI, 1 . 5 mm MgCl 2 , 10% dimethylsulphoxide, 0 . 01 % gelatin and 2 units AmpliTaq DNA polymerase (Perkin-Elmer Cetus), covered by 100 .tl of paraffin oil . The amplification was realized in a Thermal Reactor (Hybaid) and in an Intelligent Programmable Heat Block Cycling Reactor (Hybaid) for the N1-N2 and G-L primer sets, respectively . For the N1-N2 primer set the program was 5 cycles (denaturation 94 ° C, 60 s; annealing 45 ° C, 90 s ; pause 50 ° C, 20 s ; and elongation 72 ° C, 90 s) followed by 30 cycles where denaturation and elongation were reduced to 30 s and to 60 s, respectively . For the G-L primer set the program was 30 cycles (denaturation 94 ° C, 50 s ; annealing 45 ° C, 90 s; elongation 72 ° C, 120 s) . The ultimate elongation was always completed at 72 ° C for 10 min .

ice and promptly filtered through a Hybond-N nylon membrane (Amersham) with a multiwell vacuum filtration unit (Biodot Apparatus BioRad) . The membrane was then dried, u .v . irradiated for covalent fixation and hybridized overnight with a non-radioactive specific probe labelled with digoxigenin-dUTP . After washing, the presence of digoxigenin was revealed by immunological detection with a peroxidase conjugated anti-digoxigenin antibody .

Nucleotide sequencing One third (30 µl) of PCR reaction mixture was purified by electrophoresis on a 0. 7% NuSieve GTG agarose gel (FMC) in 1 x TAE buffer ." The amplified fragment (about 80 µl) was cut out from the gel, placed in an Eppendorf tube and heated 95°C for 10 min (10µl was sufficient for a single 4-track sequencing reaction) . Annealing and sequencing were performed with T7 Sequencing Kit (Pharmacia) following the manufacturer's protocol except that 50 ng of either G or L

232

D. Sacramento et al.

primers replaced the universal primer and that all steps were done at 37 ° C to avoid agarose polymerization . The four sequencing reactions were lyophilized in a Speed Vac Concentrator (Savant) up to 13 µl, denaturated at 90 ° C for 3 min, immediately snap-cooled and loaded onto a sequencing gradient gel (60 cm x 0 . 1 mm) from 0 . 5 x TBE-5% acrylamide to 5 X TBE-7% acrylamide . 25 The bottom and top buffer were 1 x and 0 . 5 x TBE, respectively . The migration was carried out for approximately 5 h at 50 W (between 1600 and 2400 V) . The gel was fixed in aqueous 10% ethanol, 10% acetic acid, dried on Whatman filter paper and exposed to Kodak XAR-5 film overnight at room temperature .

RESULTS

Analysis of brain sample PCR products by agarose gel electrophoresis One hundred brain samples wer° checked for rabies infection by PCR using either N1 or N2 to prime N cDNA and then both primers for amplification . Figure 1 shows the result obtained with N1 cDNA priming, that produced best results . Several samples exhibited an evident amplified band of about 443 by after ethidium bromide gel electrophoresis (Fig. 1a) . Others showed multiple amplified products forming diffuse bands often encompassing the 443 by position (no . 41, 47, 67, 68, etc . on Fig . 1a) . The degree of amplification appeared very variable, probably depending on the amount of initial template . In order to differentiate between specific and non-specific amplified products, a Southern blot of the gel was hybridized with an internal rabies N probe (Fig . 1b) . Only 22 clearly hybridizing samples, were found positive after 1 h exposure. As would be expected, the N1-N2 primer set was also efficient in amplifying the 443 by fragment from a Mokola virus plasmid encompassing the N gene (+ control in Fig . 1a) . This proved its competence for diagnosis of rabies and rabies-related infections . However, the Mokola band was hardly detected with the internal rabies N probe (Fig. 1b) .

detection was estimated by comparison with serial dilution of single stranded M13 DNA containing the rabies targeted region . The limit appeared to be between1-10 pg corresponding to 2 . 3 X 10 5-2 . 3 x 106 single-stranded molecules . The quantity of doublestranded amplified products then ranged from 1 . 15 x 106 (sample no . 50) to 1 . 15 x 10 8 (sample no . 48) molecules .

PCR amplification of the 'I` gene Total RNA was prepared from BHK-21 cell cultures infected with different fixed strains of rabies virus (AvO-1, CVS-11, PM, ERA, PV-11, PAS) or with Mokola virus (MOK-5) . Total RNA was also extracted from animal brains infected with wild strains of rabies virus recently isolated in France . PCR amplification of the genomic cDNAs obtained by G priming was then performed using the G-L primer set . The results presented in Fig . 3 show a faithful amplification of the expected fragment, approximatively 879 by long, in any of the tested rabies or Mokola virus infections as well as in the positive DNA control . As a first insight towards specificity, the Mokola band appears somewhat heavier than rabies ones, in perfect correlation with the size deduced from the genomic sequence (894 bp, unpublished results) . However, bands also appear in the negative cDNA controls . One of these (from the uninfected brain) was particularly surprising in comigrating with the specific product . To clarify this question, a Southern blot of the gel was hybridized with either a rabies or a Mokola internal P probe . The former detected only the fixed or wild rabies strains as well as the positive DNA control of rabies origin . The latter specifically hybridized to the Mokola products and to the size marker of Mokola origin . This demonstrated the non-specific nature of the negative control amplification, the absence of contamination, and the high sensitivity and efficiency of PCR with the G-L primer set . In addition, the differential detection with rabies and Mokola internal probes could be regarded as a preliminary approach in typing unknown infecting strains .

Direct sequencing of amplified fragments Dot-blot assay In order to simplify the diagnosis procedures, agarose gel and 32 P-labelling were replaced by dot-blot and digoxigenin-labelling, respectively . Immunological detection of the hybridizing probe took only 1 h, forming a brown-coloured precipitate . The same 22 samples, were found positive (Fig . 2) . The sensitivity of

The G-L amplified fragment of the fixed and wild rabies strains was excised from 0 . 7% NuSieve agarose gel and directly sequenced using either G or L primers . An example of the high quality of the sequence obtained is given in Fig. 3d showing part of the wild rabies strain 'I' gene structure obtained with the L primer. When compared to homologous data

(a )

(b)

233

Fig. 1 . PCR amplification of the N1 cDNA of 100 brain samples from different animal species using the N1-N2 primer set . (a) 1/10th of the amplified products were analysed by ethidium bromide agarose gel electrophoresis . (b) Southern blot of the same gel hybridized with the internal '2 P-labelled rabies N probe . One positive (+ ; 1 pg of pBR 322 vector containing the Mokola N gene) and one negative (- ; cDNA of uninfected brain) controls were submitted to amplification . Unsuccessful amplification of water (wY demonstrates that no contamination was artificially introduced during the experiment . Arrows indicate the position of the expected 443 by amplified band by comparison with size markers (mw), flanking the gel .

12 ?4 36 48 60 72 84

Fig. 2. PCR amplification of the N1 cDNA of 100 brain samples from different animal species using the N1-N2 primer set. One tenth of the amplified products were analysed by dot-blot hybridization using the internal digoxigenin-labelled rabies N probe and revealed by immunological detection . Dilution series of M„ vector containing the rabies N gene (1 ng, 100 pg, 10 pg and 1 pg) are presented as positive patterns, as well as a negative water control .

D . Sacramento et al.

234

Virus strains Controls . . . . . . . .. . . . . . . . . . . ... . . . .. . . . . . . . . . . . . . . . . . . .

(a)

8. R H A 8 2 1 H mw 0 M K

A

v D N A

M P O W W A P K R R S V

E C 0 R P V i A M S I

MW

(b)

(c)

PCR amplification with the G-L primer set of total C primed cDNA from : (i) fixed rabies virus strains (AvO-1, CVS-11, PM, ERA, PV-11, PAS) and Mokola virus (MOK-5) grown on BHK-21 cell cultures ; (ii) two french wild rabies virus strains MR) isolated from infected animal brains . One positive (+DNA ; 1 pg of pBR 322 vector containing the rabies T gene) and three negative (-BHK; -Brain ; -H 2O for cDNA of uninfected BHK-21 cells, uninfected brain and water, respectively) controls were submitted to amplification . (a) analysis of amplified products by ethidium bromide agarose gel electrophoresis . Arrows indicate the position of the expected amplified band (879 bp) by comparison with size markers (mw) corresponding to pBR 322 vector containing a Mokola virus insertion . (b, c) Hybridization of the corresponding Southern blot with rabies and Mokola internal 32 P-labelled 'P probes, respectively . (d) Autoradiogram of a sequencing reaction of the G-L PCR fragment from WR using the L primer . The sequence presented encompasses the 'P stop and the L start transcription signals a previously characterized ." Fig . 3.

available from cloned cDNA templates, a perfect similarity was observed suggesting that exogenous errors introduced by the Taq polymerase are negligible in front of the endogenous variations between rabies isolates . Indeed substitution error frequency was estimated to 10 -3 -10 - ' in unsegmented negative stranded RNA viruses ."

Further trials in typing Previous analyses of several rabies and Mokola virus sequences have allowed classification of the strains into homologous groups (Tordo and Sacramento,

submitted) . The fixed rabies strains, PAS, PV-11 and ERA (group I) and CVS-11, AvO-1 and PM (group II) delineate two separate groups . HEP, although related to group II, generates an independent group Ill . The french wild strains MR) and the Mokola virus WOK) form two additional distinct IV and V groups . From the nucleotide sequence of the 'F gene, 16,19,27 also Tordo and Sacramento, submitted; this study), an enzyme restriction map of each isolate was deduced (Fig. 4) . Using a panel of only four major restriction enzymes Bam HI, Hin dll, Hin dill and Pst I, it is easy to distinguish the five groups by simple observation of the cleavage of the G-L amplified fragment in ethidium bromide agarose gels . A unique sensitivity to

PCR amplification of rabies genes

Group

235

Strain Hind M

I

T I PV, PAS RSO I

Top I (580)

Top



I (290) +

Born Toq I

(290)

HI 1

HI l

Sam

III

HEP



II

CVS, AvOl, PM



Q

MOK

PSI I ( t Hin d m I

0

100 200 300 400 500 600 700 800 900 Bass pairs

Fig . 4.

Endonuclease digestion pattern in agarose gel of the C-L amplified fragment with 1=Bam HI, dll, 3 = Hin dill, 4=Pst I, 5 = Rsa I . The sensitivity to an enzyme is symbolized by a star . The restriction map of the amplified region is schematized in most fixed or wild rabies strains (including HEP and SAD, not checked in the present paper) as well as in the Mokola virus . The Taq I cleavages differentiating between ERA and SAD strains are shown in grey boxes .

2 = Hin

Hin dill,

Bam HI or Hin dll characterized members of groups I, III and V, respectively . In contrast, members of groups II and IV are sensitive to a couple of enzymes, Barn HI-Pst I and Bam HI-Hin dill, respectively . Once the analysed isolate was classified into one of the five main groups, a more precise analysis could be carried out . For example, within group I, additional digestion with Rsa I differentiates the ERA-

SAD strains, that are cleaved, from the PV-11-PAS ones, that are not . Furthermore, a large 580 by Taq I fragment observed in ERA is cut into two 290 by bands for SAD . Figure 4 shows the cleavage pattern of the G-L amplified segment obtained with the four major enzymes on Mokola virus and on all rabies strains cited above, except HEP and SAD the data of which came from other studies ."', " The distinction between PV-11-PAS and ERA strains by Rsa I digestion is also shown . DISCUSSION We previously reported the cloning and sequencing of the genome of several rabies strains of which only

the PV strain was characterized to completion .'" Similarly, we have cloned the whole genome of the rabies-related Mokola virus 18 and determined the structure of the first 6800 nucleotides of the 3' end (unpublished results) . Rabies and Mokola viruses are the prototypes of the two most divergent serotypes of the Lyssavirus genus in the Rhabdoviridae family, serotypes 1 and 3 respectively ." Therefore, their comparative sequence analysis shown in Fig . 5 is illustrative of the maximal diversity of this viral genus. As previously noted '13,11,29 the homology is not randomly located but instead the different genes are subjected to unequal selective pressures . In general, protein coding regions appear more conserved than non-protein coding ones, with the exception of the atypical M1 gene, that appears highly variable . The degree of homology decreases from the N to M2 and G genes . Similar observations were already determined by comparison between different rabies strains29 or between rabies virus and vesicular stomatitis virus ." The more divergent region is obviously the G-L interregion encoding the rabies Y pseudogene. 13,16

D. Sacramento et al .

236

Li

/

6 primer L

Molecular epidemiology

5

primer G

0

I

4

0

. . . . . . . . . . . . . . . . . . .. . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

0

2

3

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . .. . . . . . . . . . . . . . . . 'I I

2

MI

primer N2

I N

I ;

Diagnosis primer NI

I

Mil

N I

is

M2

I

.

1

2

3

4

G

L 5

6

Kb

Rabies PV

Fig. 5 . The nucleotide sequence of the PV strain of rabies virus is compared with its counterpart in the Mokola virus, over more than 6 . 5 kb . The computer program uses a double filter . First, the sequences are compared by windows of 20 nucleotides, and a diagonal trace appears when at least 70% of identical residues are in the same position (not shown) . The results are processed through a second filter (window size, 40 nucleotides ; % conserved residues, 60%), in order to select the major diagonals (homologies) of the graph . The convenient genomic regions selected either for diagnosis or for molecular epidemiology are indicated . Thick and thin dotted lines mark the limits of the genes and of the coding regions, respectively . Only the COOH end of each protein is indicated, because at the scale used, the thin dotted line corresponding to the NH, end is melted with the flanking thick line as the result of the very limited 3' untranslated region of these genes .

PCR is convenient for rabies diagnosis For diagnosis purposes, the target viral transcript should be present in substantial quantity (sensitivity) and be as invariant as possible to be detected in every isolate (ubiquity) . The N and mRNA fulfils both criteria because it is highly conserved (Fig . 5) and produced at first and in highest amounts during infection, as a result of the decreasing efficiency of transcription from the 3' to the 5' genome extremity ." Therefore, we first tried to amplify cDNA to NmRNA (not shown), then cDNA to N gene (Figs 1 and 2), to compare their respective efficiencies . We were highly surprised to observe constantly that the amplification level was better in this latter case . We propose that this unexpected difference is due to the protective effect of nucleocapsid structures for the viral genome, while the messengers are more access-

ible to nucleases . However, a more precise description of the viral transcription and replication mechanisms in brain are necessary before firmly supporting such an hypothesis . As amplification efficiency is inversely related to the sequence length," we chose a set of primers separated by a short distance (about 400 residues) . Since there was no real difficulty in finding conserved sequences of about 20 residue long between rabies and Mokola nucleoprotein genes, we selected the N1-N2 pair of primers, primarily because of their intrinsic composition, both being equally rich in G and C residues . Both primers were synthesized according to the Mokola sequence, N1 (19-mer) and N2 (17-mer) showing 2-3 and 0-2 mismatches with the known rabies sequences, respectively (Fig . 6) . This choice of template-primer heterogeneity was deliberate to ensure an efficient amplification, even in

PCR amplification of rabies genes

Primer Ni --- N gene + sense F

E

T

A

P

--- Primer N2 - sense

F

V

587

237

G

C

Y M

G

605

PV

TTT GAL ACa GCC CCr ITT G

EFA-SAD

TTT GAG ACa (mac CCC TPf G

1013 CAT aT 003 ATA

i C

1029 CC

I

CAT OCT AM ATA TAC CC

ERR-SAD

CAa CCT ACG ATg TAC GC

CIS-AV01-E 4

TTc GAL ACa ca:a CCT TIT G

CJS Av01-SAD

IEP

ITCCGAGACI'cm ccrTIT GI

MX

587

ICAT

605

CCT Am ATA T9c cc

1013 F

E

T

A

P

V

G

C

Y

M

W gene Primer

Primer G + sense

Iv

GAC TIG GGT CIC

G

L

- sense

N W G D L G L P ~'~4687 4665 IGCCTIT cOST CTC CCG AIC 005 GrA

EFA-SAD

MX 1029

F

D

Y

N

L

N

S P

L

5520

5543

I=ATG

TfA GAG TIO AGA GGRAAC(

RI

AK TM GG CTG ATG TTA GAG TIG AGA 5Th AOC

CVS-AV01-EM

GAC CTG GGT CTC CCG AAC TCL GG

IEP

GAC cTG GGT CTC OCG AM TOG GG

EFA-SAD

. . . CVS-VA

1-fM

FEF GAC CTG GGT CTg CCt CAt TCL GG

MX 4674

4696

CTG ATG TRI GAc 5544

D

L

G

L

P

H

W

G

ACA GGA AAC M

D

Y

N

L

N S

5567 P

MX

L

Fig . 6 . Comparison of the N1, N2, G and L primer structure in all the available rabies strains and in Mokola virus . Data were obtained from Tordo et al.'-17 for PV, from Anilionis et al ." for ERA, from Conzelmann et al." for SAD, from Poch et al . 29 for AvOl, from Morimoto et al ." for HEP, and from personal unpublished results for MOK . The numbering is respective to the PV rabies strain and Mokola virus genome . The synthesized primers are boxed .

extreme conditions that could naturally occur with the wild rabies strains . Our data clearly showed an effective amplification of the 443 by target region, even when a low temperature (45° C) was used for annealing (Figs 1 and 2) . The direct

inary attempts to compare the sensitivity of non-

observation of the amplified segment after migration in ethidium bromide agarose gel appears insufficient for diagnosis, since commigration of several non-specific bands was observed, particularly in highly damaged samples . However, probing of the corresponding Southern blot with a specific 32P-labelled internal rabies N probe, definitively discriminates between specific and non-specific amplifications . Interestingly, this probe has

trials, for over 100 french field samples, we found a perfect concordance (100%) with the three methods usually performed in rabies diagnosis, FAT on brain smears, RTCIT and RREID' : the same 22 and 78 samples were respectively found positive and negative whatever the technique used . This definitely proves that PCR techniques could be an alternative protocol for diagnosis, even for highly divergent lyssaviruses like Mokola virus . We avoided the prob-

difficulties to detect the positive control Mokola fragment, although present in sufficient amounts . Such differential hybridization could also be seen as a first means to typing, although more powerful techniques are detailed below . Alternatively, dot-blot analysis of the amplified products with a non-radioactive digoxigenin labelled probe can be used . This second technique is faster, clearly more simple and can detect as little as 1 pg of amplified products . A further simplification by direct dotting without filtration is currently under investigation . In addition, digoxigenin labelled probes are stable for several months, whereas 32 P-labelled ones must be Prelimleast every 2 weeks . changed at

radioactive and radioactive labelled probes on dotblots suggest that they are very similar at a comparable extent of revelation . Performing both techniques in systematic blind

lem of contamination by working on internal brain samples, handling samples and reagents with gloves, aliquoting PCR reagents in hoods, storing mixtures in a designated freezer and using Microman (Gilson) for dispatching products . The ultimate precaution is obtained by separating work areas for setting up reactions. The National Reference Centre for rabies of Paris receives daily suspected samples early in the morning (day 1) . The results of FAT (2-3 h) and RREID (4 h), used as a reference for diagnosis, are generally available early afternoon . By comparison, RNA extraction (2 h), cDNA synthesis (1 . 5 h), PCR amplification (3 h)

238

D . Sacramento et al.

and dot-blotting are completed on day 1 . After over-

rabies sequence, allowing the amplification of an 879

night hybridization, the immunological detection takes 2-3 h and the results are available in the morning of day 2, in synchronism with the results of the RTCIT (18-20 h) . Therefore, the PCR appears as a possible technique for post-mortem diagnosis especially as simplified andquicker protocols are under investigation. In this context, we are in the process of system-

by band . The true specificity of the selected G-L primer set is evidenced by the amplification of the Mokola'P gene (serotype 3), highly divergent from the rabies 'P gene (serotype 1), such that internal probes are unable to show reciprocal hybridization on Southern blots (Fig . 3) . Consequently, we succeeded in amplifying every fixed or wild rabies strains of serotype 1 that we

atically using PCR for the statistical improvement of the method over a larger number of samples . The true future of PCR would reside in its ability to surpass the sensitivity of routine post-mortem techniques or in its suitability for intra-vital diagnosis on animal secretions, when laboratory diagnosis remains inefficient.

Molecular epidemiology For epidemiological studies, in contrast with diagnosis purposes, the important point is to find sensitive criteria to differentiate virus isolates . In that goal, highly variable genomic areas are the most suitable . For this reason, we have focused our attention on the remnant rabies 'P gene a non-protein coding region greatly susceptible to mutations 13' 16 (also, Tordo and Sacramento, submitted) . We preferred this region than the highly variable M1 gene because it is more likely to represent the natural evolution of the virus outside any external selective pressure . Envisaging efficient amplification of the 'P region of every rabies isolate, we looked for conserved primers in the flanking genes. Due to the variability of the proximal G gene, notably in the transmembrane and cytoplasmic peptides, 16 a suitable sequence could only be located upstream from the transmembrane peptide . This 23-mer showed up to five mismatches between rabies and Mokola viruses (Fig . 6) . However, its five last 3' nucleotides, in which template complementarity is so important for efficient amplification, promised to be invariant. The first three code for a tryptophan, a highly conserved amino-acid during evolution 32 encoded by a unique codon (TGG) . The last two correspond to the two first nucleotides of a conserved glycine codon . So, the highly variable 'wobble' position is avoided . At the opposite side of the 'P gene, we selected the closest conserved region in the L gene as the reverse primer . Our previous comparative analysis of the rhabdoviral and paramyxoviral L proteins designates a DYNLNSPL peptide as the most NH2 conserved motif between unsegmented negative strand RNA polymerases ." We used the corresponding 24-mer oligonucleotide as the basis for the L primer . It shows only two differences between rabies and Mokola sequences . The G-L primers were synthesized according to the PV strain

checked . Nowadays, rabies virus strains are distinguished by antigenic differences using a panel of selected monoclonal antibodies against the nucleocapsid- or the glycoproteins ."' However, this analysis would be fastidious and time consuming to be applied to every isolate since requiring previous cell adaptation . In contrast, we have shown that the G-L amplified segment finely and rapidly characterizes the original rabies isolates either by restriction enzyme analysis (Fig. 4) or ultimately, by dideoxy sequencing technique (Fig. 3) . The former method is a powerful technique for typing rabies strains by visual inspection of the digestion products on agarose gel and is easily realized the day of sample reception . For the direct sequence analysis, an additional day is required to purify the amplified segment on NuSieve GTG agarose gel, and to perform and run sequencing reactions . Results are available in the morning of the 3rd day and can be immediately analysed by suitable computer programs . Among the various direct sequencing protocols described, we have chosen a simple and fast method, adapted from Kretz et al." The only modifications were the percent of agarose gel 07% instead of 1%) and the T7 sequencing kit (Pharmacia instead of USB) . Under these conditions, the protocol is highly efficient for direct sequencing of any double-stranded PCR product . It has unlimited applicability, even when using an internal sequencing primer, different from those employed for amplification . The fidelity was also repeatedly verified by comparing the results to that obtained with cloned cDNA templates .", ", " It must be noted that the purification on NuSieve agarose gel can also be required before typing studies, when non-specific bands commigrate with the specific product, resulting in a confused restriction pattern . This alternative is, however, rarely necessary, a modification in the setting of the PCR apparatus being generally sufficient to avoid such a problem .

CONCLUSION The PCR technique described here is a rapid, sensitive and specific method and can be used as an alternat-

PCR amplification of rabies genes ive protocol for the routine diagnosis as well as a powerful tool for the molecular epidemiology of lyssaviruses . It is efficient, even on highly degraded samples . We have paid particular attention in simplifying the protocols, to avoid personal laboratory procedures and to work with easily available commercial kits . The true force of this technique is that it does not require previous cell adaptation, shown to be a source or selector of mutations ." Therefore, it is more likely to preserve the natural situation and invites a worldwide epidemiological survey resulting in a more complete picture of the global distribution of rabies virus .

ACKNOWLEDGEMENTS We are grateful to Professor P. Sureau in whose laboratory this work was carried out. We thank Tri Nguyen for many helpful discussions and Brian Lockhart for critically reading this manuscript. The authors would also like to thank the technical assistance of Pascal Cozette . This work was supported by a grant from the Comite Consultatif des Applications de la Recherche de I'Institut Pasteur .

REFERENCES 1 . Bourhy, H ., Rollin, P . E ., Vincent, J . & Sureau, P. (1989) . Comparative field evaluation of the fluorescent-antibody test, virus isolation from tissue culture, and enzyme immunodiagnosis for rapid laboratory diagnosis of rabies . Journal of Clinical Microbiology 27, 519-23 . 2 . Negri, A . (1903). Beitrag zum studium der atiologie der Tollwut . Zentralblatt fur Bakteriologie, Parasitenkunde, lnfektionskrankheiten and Hygiene 43, 507-28. 3 . Perrin, P., Rollin, P . E. & Sureau, P . (1986) . A rapid rabies enzyme immuno-diagnosis (RRIED) : a useful and simple technique for the routine diagnosis of rabies . Journal of Biological Standardization 14, 217-22 . 4 . Johnson, H . N . (1973) . The virus neutralization index test in mice . In Laboratory Techniques in Rabies (Kaplan, M. M. & Koprowski, H ., eds) pp . 94-7 . Geneva : World Health Organization . 5 . Smith, J . S ., Yager, P . A. & Baer, G . M . (1973) . A rapid tissue culture test for determining rabies neutralizing antibody . In Laboratory Techniques in Rabies (Kaplan, M . M. & Koprowski, H ., eds) pp . 354-7 . Geneva: World Health Organization . 6 . Wiktor, T. J ., Koprowski, H . (1978) . Monoclonal antibodies against rabies virus produced by somatic cell hybridization : detection of antigenic variants . Proceedings of the National Academy of Sciences, USA 75, 3938-42 . 7 . Wiktor, T. J ., Flamand, A . & Koprowski, H . (1980) . Use of monoclonal antibodies in diagnosis of rabies virus infection and differentiation of rabies and rabiesrelated viruses. Journal of Virological Methods 1, 3346 . 8 . Flamand, A ., Wiktor, T . J . & Koprowski, H . (1980) . Use of hybridoma monoclonal antibodies in the detection of

239

antigenic differences between rabies and rabies-related virus proteins . I . The nucleocapsid protein . Journal of General Virology 48, 97-104. 9 . Flamand, A ., Wiktor, T . J . & Koprowski, H . (1980) . Use of hybridoma monoclonal antibodies in the detection of antigenic differences between rabies and rabies-related virus proteins . II. The glycoprotein . Journal of General Virology 48, 105-9 . 10 . Dietzschold, B ., Rupprecht, C . E ., Tollis, M., Lafon, M ., Mattei, J ., Wiktor, T . J . & Koprowski, H . (1988) . Antigenic diversity of the glycoprotein and the nucleocapsid proteins of rabies and rabies-related viruses : implications for epidemiology and control of rabies . Reviews of Infectious Diseases 10, S785-98 . 11 . Vincent, J ., Bussereau, F . & Sureau, P . (1988) . Immunological relationships between rabies virus and rabiesrelated viruses studied with monoclonal antibodies to Mokola virus . Annales de I'Institut PasteurlVirologie 139,157-73 . 12 . Smith, J . S . (1989) . Rabies virus epitopic variation : use in ecologic studies . In Advances in Virus Research (Maramorosch, K ., Murphy, F . A . & Shatkin, A . J ., eds) pp . 215-53 . San Diego : Academic Press . 13 . Tordo, N ., Poch, O . (1988) . Structure of rabies virus . In Rabies (Campbell, J. B . & Charlton, K . M., eds) pp . 25-45 . Boston : Kluwer Academic Publishers . 14 . Wunner, W . H ., Larson, J . K ., Dietzschold, B. & Smith, C . L . (1988) . The molecular biology of rabies viruses. Reviews of Infectious Diseases 10, S771-84. 15 . Tordo, N ., Poch, 0 ., Ermine, A . & Keith, G . (1986). Primary structure of leader RNA and nucleoprotein genes of the rabies genome : segmented homology with VSV . Nucleic Acids Research 14, 2671-83 . 16. Tordo, N ., Poch, 0 ., Ermine, A ., Keith, G . & Rougeon, F . (1986) . Walking along the rabies genome : is the large G-L intergenic region a remnant gene? Proceedings of the National Academy of Sciences, USA 83, 3914-8 . 17 . Tordo, N ., Poch, 0 ., Ermine, A ., Keith, G . & Rougeon, F . (1988) . Completion of the rabies virus genome sequence determination : highly conserved domains along the L (polymerase) proteins of unsegmented negative-strand RNA viruses . Virology 165, 565-76 . 18 . Bourhy, H ., Tordo, N ., Lafon, M . & Sureau, P . (1989) . Complete cloning and molecular organization of a rabies-related virus : Mokola virus . Journal of General Virology 70, 2063-74 . 19 . Conzelmann, K . K ., Cox, J . H ., Schneider, L. G . & Thiel, H . J . (1990). Molecular cloning and complete nucleotide sequence of the attenuated rabies virus SAD B19, Virology 175, 485-9 . 20 . Ermine, A ., Tordo, N . & Tsiang, H . (1988) . Rapid diagnosis of rabies infection by means of a dot hybridization assay . Molecular and Cellular Probes 2, 75-82 . 21 . Ermine, A., Larzul, D ., Ceccaldi, P . E ., Guesdon, J . L. & Tsiang, H . (1990) . Polymerase chain reaction amplification of rabies virus nucleic acids from total mouse brain RNA . Molecular and Cellular Probes 4, 189-91 . 22 . Clark, H . F ., Wiktor, T . J . (1972) . Rabies virus . In Strains of Human Viruses (Majer, M . & Plotkin, S . A ., eds) pp . 177-82 . Basel : Karger. 23 . Lafon, M., Bourhy, H . & Sureau, P . (1988) . Immunity against the European bat rabies (Duvenhage) virus induced by rabies vaccines : an experimental study in mice . Vaccine 6, 362-8 . 24 . Barrat, J ., Halek, H . (1986) . Simplified and adequate sampling and preservation techniques for rabies dia-

240

D. Sacramento et al .

gnosis in Mediterranean countries . Comparative Immunology, Microbiology and Infectious Diseases 9, 10 . 25 . Maniatis, T ., Fritsch, E . F . & Sambrook, J. (1982) . Molecular Cloning : a Laboratory manual . 1st edn, pp . 1-545 . New York : Cold Spring Harbor . 26 . Steinhauer, D . A ., de la Torre, J . C . & Holland, J . J . (1989) . High nucleotide substitution error frequencies in clonal pools of vesicular stomatitis virus . journal of Virology 63,2063-71 . 27 . Morimoto, K ., Ohkubo, A . & Kawai, A . (1989) . Structure and transcription of the glycoprotein gene of attenuated HEP-Flury strain of rabies virus . Virology 173, 465-77 . 28 . Bourhy, H ., Sureau, P . & Tordo, N . (1990) . From rabies to rabies-related viruses . Veterinary Microbiology 23, 115-28 . 29 . Poch, 0., Tordo, N . & Keith, G . (1988) . Sequence of the 3386 3' nucleotides of the genome of the AvOl strain rabies virus: structural similarities of the protein regions involved in transcription . Biochimie 70, 1019-29 . 30 . Banerjee, A . K . (1987). Transcription and replication of rhabdoviruses . Microbiological Reviews 51, 66-87 . 31 . Saiki, R . K ., Scharf, S ., Faloona, F ., Mullis, K . B ., Horn,

32 .

33 .

34 .

35 .

36.

G . T ., Erlich, H . A . & Arnheim, N . (1985) . Enzymatic amplification of ~-globin genomic sequences and restriction site analysis for diagnosis of sickle cell anemia . Science 230, 1350-4 . Dayhoff, M. O . (1979) . Atlas of Protein Sequence and Structure . Washington D .C . : National Biochemical Research Foundation . Poch, 0 ., Blumberg, B . M ., Bougueleret, L . & Tordo, N . (1990) . Sequence comparison of five polymerases (L proteins) of unsegmented negative-strand RNA viruses : theoretical assignment of functional domains . journal of General Virology 71, 1153-62 . Kretz, K . A ., Carson, G . S. & O'Brien, ) . S . (1989) . Direct sequencing from low-melt agarose with SequenaseR . Nucleic Acids Research 17, 5864 . Meyerhans, A ., Cheynier, R ., Albert, J ., Seth, M ., Kwok, S ., Sninsky, J ., Morfeldt-Manson, L ., Asjo, B . & WainHobson, S . (1989) . Temporal fluctuations in HIV quasispecies in vivo are not reflected by sequential HIV isolations . Cell 58, 901-10 . Anilionis, A ., Wunner, W . H . & Curtis, P . J . (1981) . Structure of the glycoprotein gene in rabies virus . Nature (London) 294, 275--8 .