Differentiation of the Seven Major Lyssavirus Species byOligonucleotide Microarray

Jin Xi, Huancheng Guo, Ye Feng, Yunbin Xu, Mingfu Shao, Nan Su, Jiayu Wan, Jiping Li, and Changchun Tu

Institute of Military Veterinary, Academy of Military Medical Sciences, Changchun, China

An oligonucleotide microarray, LyssaChip, has been developed and verified as a highly specific diagnostic tool for differentiationof the 7 major lyssavirus species. As with conventional typing microarray methods, the LyssaChip relies on sequence differencesin the 371-nucleotide region coding for the nucleoprotein. This region was amplified using nested reverse transcription-PCRprimers that bind to the 7 major lyssaviruses. The LyssaChip includes 57 pairs of species typing and corresponding control oligo-nucleotide probes (oligoprobes) immobilized on glass slides, and it can analyze 12 samples on a single slide within 8 h. Analysisof 111 clinical brain specimens (65 from animals with suspected rabies submitted to the laboratory and 46 of butchered dogbrain tissues collected from restaurants) showed that the chip method was 100% sensitive and highly consistent with the “goldstandard,” a fluorescent antibody test (FAT). The chip method could detect rabies virus in highly decayed brain tissues, whereasthe FAT did not, and therefore the chip test may be more applicable to highly decayed brain tissues than the FAT. LyssaChip mayprovide a convenient and inexpensive alternative for diagnosis and differentiation of rabies and rabies-related diseases.

Rabies is a worldwide zoonotic disease caused by viruses withinthe genus Lyssavirus, family Rhabdoviridae, with many host

species acting as reservoirs for infection (8). There are 7 majorspecies within the genus: classical rabies virus (RABV), Lagos batvirus (LBV), Mokola bat virus (MOKV), Duvenhage virus(DUVV), European bat lyssavirus type 1 (EBLV-1), European batlyssavirus type 2 (EBLV-2), and Australian bat lyssavirus (ABLV).Recently, 4 new members have been approved by the Interna-tional Committee on Taxonomy of Viruses: Aravan virus(ARAV), Irkut virus (IRKV), Khujand virus (KHUV), and WestCaucasian bat virus (WCBV) (19). Lyssaviruses possess a single-stranded, nonsegmented, negative-sense RNA genome of approx-imately 12 kb that encodes 5 proteins: a nucleoprotein (N), aphosphoprotein (P), the matrix protein (M), a single surface gly-coprotein (G), and an RNA-dependent polymerase (L) (29). Ngene divergence is commonly used for species typing of lyssavi-ruses, since the abundance of N mRNA in infected cells makes itan ideal target for detection and differentiation. Additionally, theN gene has been reported to be the most conservative of all 5 genesbased on DNA polymorphism analysis (34).

The illnesses caused by rabies and rabies-related viruses arevirtually indistinguishable (28). The fluorescent antibody test(FAT), the “gold standard” diagnostic test for rabies virus, is un-able to discriminate between lyssavirus species. Sequence analysisof PCR products has provided the most accurate information, butit is not considered the method of choice to determine the pres-ence of double infections (25). Other species typing methods, suchas conventional gel-based PCR assays for detection of LBV andMOKV lyssaviruses (13) and reverse transcription– quantitativereal-time PCR (RT-qPCR) for detection of RABV, EBLV-1,EBLV-2, and ABLV have been reported (14, 31), but there is stillno available method for the specific detection of DUVV.

Microarray technology has become a rapid and efficientmethod in clinical diagnostics (17). Microarray-based detectionand genotyping of influenza virus (30), hepatitis B virus (21),foot-and-mouth disease virus (2), respiratory viruses, and food-borne pathogens have been reported (22, 26), and the microarray

technique additionally has been widely used for detecting newpathogens.

Recently, Berthet et al. developed a high-density microarray usinga whole-genome amplification (WGA) method for sample prepara-tion (3). Gurrala et al. developed a DNA microarray (LP chip) fordetection and differentiation of lyssaviruses (16), based on randomPCR that labels target DNA (32). Recently, Dacheux et al. reported aresequencing microarray for typing rhabdoviruses (9) with improvedwhole-transcriptome amplification (WTA) (5). De Benedictis et al.developed a pyrosequencing method to type lyssavirus species (11).In our study, the LyssaChip has been developed to specifically differ-entiate the 7 major lyssavirus species based on generic reverse tran-scription–nested PCR (RT-nPCR). It can analyze 12 samples on asingle slide within 8 h, thereby supporting its potential application forthe diagnosis of rabies and rabies-related diseases.

MATERIALS AND METHODSViruses and specimens. Reference strains of the 7 major lyssavirus specieswere kindly provided by the Animal Health and Veterinary LaboratoryAgency, Weybridge, United Kingdom. They were CVS-11 (RABV), RV1(LBV), RV4 (MOKV), RV131 (DUVV), RV9 (EBLV-1), RV1787 (EBLV-2), and RV634 (ABLV). The four newest members of the lyssavirus genus,ARAV, IRKV, KHUV, and WCBV, now considered independent species(19), were not available to us. The LyssaChip analysis results for 111 clin-ical brain tissue specimens submitted by different provinces for laboratoryconfirmation between 2003 and 2010 were compared with standard FAT(10) and RT-nPCR (20) results. Of the 111 specimens, 65 were fromanimals suspected to have rabies (dogs, cows, sheep, pigs, and raccoon),

Received 26 April 2011 Returned for modification 9 July 2011Accepted 15 November 2011

Published ahead of print 21 December 2011

Address correspondence to Changchun Tu, changchun_tu@hotmail.com.

J.X. and H.G. contributed equally to the study.

Copyright © 2012, American Society for Microbiology. All Rights Reserved.

doi:10.1128/JCM.00848-11

0095-1137/12/$12.00 Journal of Clinical Microbiology p. 619–625 jcm.asm.org 619

on October 4, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

and the remaining 46 were from restaurants in Hunan Province from dogsslaughtered for consumption (Table 1).

LyssaChip design and construction. The 12-array concept was usedto construct the slide chip, named the LyssaChip, with each array com-prising a maximum of 15 by 14 oligoprobes in a 210- by 99-�m square(see Fig. 1A). In the chip the clock dial-like layout of specific detectionoligoprobes for typing the 7 major lyssaviruses was designed to readilydifferentiate the species. In addition, positive- and negative-control oli-goprobes (70 nucleotides [nt]) were also added to each array to monitorwhether the hybridization conditions were correct and specific. The se-quence of the positive control was from the pGM-T plasmid, and that ofthe negative control was from Arabidopsis thaliana. To optimize the exci-tation beam of the chip reader to a 532-nm wavelength, a hexachlorofluo-rescein (HEX)-labeled oligomarker (70 nt) with a random sequence wasdesigned and spotted in the center and vertices of each array.

A region within the lyssavirus N gene (371 bp; nucleotides 62 to 442)was the target for probe detection, which is the RT-nPCR target for de-tection of pan-lyssaviruses (20). The oligoprobe sequences used for the 7lyssaviruses were those of Bourhy et al. for RABV (7), Markotter et al. forLBV (23), Sabeta et al. for MOKV (27), Wu et al. for DUVV (34), Amen-gual et al. for EBLV-1 and EBLV-2 (1), and Warrilow et al. for ABLV (33).Specific probes were selected from an extensive alignment comparison of371-bp N sequences of representative strains of the 7 major taxonomiclyssaviruses retrieved from GenBank in order to cover all published lin-eages of lyssaviruses, and sequences were then subjected to a Blastn ho-mology search on the GenBank website. All selected oligoprobes were of alength of 60 to 70 nt with �80% homology and a minimum 25-nt con-tinuously matched region with all targets to be detected and showing thepreferential melting point of 70 � 5°C (32). The 371-bp N gene sequenceswere aligned by using MegAlign of LaserGene software v7.0 (DNASTAR,Madison, WI), and the oligoprobe candidates were designed with Ar-raydesigner software v4.2 (PREMIER Biosoft, Palo Alto, CA). Oligo-probes were commercially synthesized (Sangon, Shanghai, China) andcontained an amino link group at the N terminal. They were diluted to 50pmol/�l with dilution buffer (CapitalBio, Beijing, China) before use. TheMicrogrid II Biorobot system (BioRobotics, Woburn, MA) was used forprinting the oligoprobes onto aldehyde-coated glass slides (CapitalBio).The semifinished slides were immediately fixed at 37°C in 70% humidityfor 2.5 h and then at room temperature for at least 30 h to allow completefixation, after which they were washed with 2% SDS (Promega, Madison,WI), 3% NaBH4 (Sigma-Aldrich) and dried by low-speed centrifugation.

RNA extraction, RT-nPCR, and labeling. Two sets of pan-lyssavirusRT-nPCR primers were used to amplify the nt 62 to 442 fragment of the Ngene (20), with outer primers N127 and N829 for the first round (nucleo-

tides 55 to 899) and inner primers NF371 (labeled with biotin-HEX at the5= end) and NR371 for the second round. Their sequences, along withthose of positive-control (PC) primers, are listed in Table 2.

Samples of brain tissue were prepared as 10% (wt/vol) suspensions byhomogenization in phosphate-buffered saline by using a TissuLyser II(Qiagen, Hilden, Germany). Total viral RNAs were extracted using theautomatic nucleic acid extraction station epMotion 5075 VAC (Eppen-dorf, Hamburg, Germany) with the QIAamp Virus BioRobot MDX kit(Qiagen, Hilden, Germany) as per the manufacturers’ instructions. TheRNA preparations were used immediately for reverse transcription withthe N127 primer. The RT products were then used to amplify the 371-bpN gene fragment by nested PCR using the above primer sets. PCR wasperformed at 94°C for 2 min, followed by 35 cycles of 94°C for 30 s, 56°Cfor 30 s, and 72°C for 40 s. During the second round of amplification, thetargets were labeled with biotin-HEX.

LyssaChip hybridization and analysis. For LyssaChip hybridization,10-�l aliquots of PCR-labeled detection products were mixed with 2 �lPCR-labeled positive control and 10 �l hybridization buffer containing 2�l 99% formamide, 2 �l 50� Denhardt’s solution (Sigma-Aldrich), 2 �l1� SSC (0.15 M NaCl plus 0.015 M sodium citrate), 2 �l 2% SDS (Pro-mega), and 2 �l diethyl pyrocarbonate-water (Sangon, Shanghai, China).The mixture was heated at 95°C for 5 min to denature double-strandedDNA, followed by chilling immediately on ice for 5 min, and then appliedto the array and covered with a plastic coverslip (CapitalBio, Beijing,China) to prevent evaporation of the sample during incubation. Hybrid-ization was performed at 43°C for 3.5 h. After hybridization, the slide waswashed once for 3 min with 2� SSC containing 2% SDS, followed by awash for 5 min with 2� SSC and then drying by low-speed centrifugation.

Fluorescent images of the LyssaChip were obtained by scanning theslides with a GenePix 4000A personal scanner (Axon Instruments, Sunny-vale, CA). The fluorescent signals obtained at 532 nm from each spot wereanalyzed by GenePix Pro 6.0 software (Axon). Median background fluo-rescence was determined from the region surrounding the hybridizationspots and subtracted. Detection signals with a median fluorescence higherthan background level (P � 0.01) were considered positive.

Data statistical analysis. Normalized data obtained with GenePix Pro6.0 were used for statistical analysis with Prism software v5.02 (GraphPadSoftware) to conduct Student’s t tests, a kappa test, McNemar chi-squaretests, 95% confidence interval (CI) determinations, and production ofgraphics.

RT-qPCR differentiation of RABV, EBLV-1, and EBLV-2. LyssaChipresults were validated by standard TaqMan RT-qPCR (31), with reactionmixtures that contained the following: 12 �l 10� PCR buffer, 12 �l 25mM MgCl2, 0.5 �l deoxynucleoside triphosphates (25 nM), 0.3 �l RNasin(5 U), 0.5 �l Moloney murine leukemia virus reverse transcriptase (50 U),0.5 �l Ex-Taq polymerase (5 U/�l; TaKaRa, Dalian, China), 1 �l of 10%(vol/vol) Triton X-100 (Promega, Madison, WI), 1 �l forward primer (20pmol/�l), 1 �l reverse primer (20 pmol/�l), 1 �l each of TaqMan probes(5 �M each), 10 �l template RNA, and nuclease-free water to a finalvolume of 50 �l. The reactions were carried out in an MX3000P multiplexquantitative PCR system (Stratagene, LaJolla, CA). Reverse transcriptionand PCR amplification were performed using the following programs: 1

TABLE 1 Clinical specimens tested in this study

Province wheresample wasobtained

No. of specimens froma:

Dog Cow Sheep Pig Raccoon

Chongqing 31Hunan 57b 2Tianjin 3Shandong 3Guangdong 4Shanxi 4 2Jiangsu 1Hebei 1Guangxi 1Shanghai 1Total 103 3 2 2 1a The specimens from cows, sheep, pigs, and raccoons were from single rabiesoutbreaks.b Of the 57 specimens obtained from dogs in Hunan Province, 46 were of butchereddogs, for which no clinical information was available.

TABLE 2 Primer sets used for LyssaChip sample preparations

Primer Sequences (5=–3=)a

Size ofproduct (bp)

N127 ATGTAACNCCTCTACAATGG 845N829 GCCCTGGTTCGAACATTCTNF371 HEX-ACAATGGAKKCTGACAARATTG 371NR371 CCTGYYWGAGCCCAGTTVCCYTCPC forward HEX-TACCGCGAGACCCACGCTCA 481PC reverse GACGCCGGGCAAGAGCAACTa N � A/G/C/T, K � G/T, R � A/G, Y � C/T, W � A/T, and V � A/G/C.

Xi et al.

620 jcm.asm.org Journal of Clinical Microbiology

on October 4, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

cycle at 42°C for 30 min and 94°C for 3 min, followed by 40 cycles at 94°Cfor 30 s, 55°C for 30 s, and 72°C for 20 s.

RESULTSOligoprobe design and LyssaChip construction. Initially, 78 can-didate oligoprobes were designed by using Arraydesigner softwarev4.2 based on multiple alignment of the 371-bp N gene sequences ofthe 7 major lyssaviruses and then subjected to Blastn search to excludethose with unreasonable GC contents, �80% homology, and lessthan 25 continuously matched nucleotides with target sequences.Fifty oligoprobes that met these parameters were selected and, alongwith 7 control and oligomarker oligoprobes, were double printed and

arrayed on the glass slides as shown in Fig. 1A. Finally, 6 to 10 specificprobes for each virus were selected in order to ensure the detection ofall lineages or subtypes within a species. For instance, a set of 6 probeswas selected to specifically detect different RABV lineages, includingIndian subcontinent, Asian, Africa 2, Arctic-related, Africa 3, andCosmopolitan (7).

Detection and differentiation. Under the above hybridizationconditions, the 7 reference lyssaviruses showed hybridization onlywith their corresponding oligoprobes and no nonspecific hybrid-ization with unrelated oligoprobes (Fig. 1B). The fluorescence in-tensities of all expected hybridizations were significantly higher

FIG 1 The clock dial-like layout of the oligoprobe array printed on the glass slide and its hybridization profile of the 7 major lyssavirus species. (A) Eachmicroarray has a capacity for 210 (15 � 14) probes. The lyssavirus species typing oligoprobes (numbers 1 to 50) and positive- and negative-control probeswere spotted in duplicate on each microarray, except for probe 21, which was spotted in triplicate. The HEX-labeled oligomarker was spotted 8 times: 4at the center and 4 at the vertices. Probe spot numbers: 1 to 6, RABV; 7 to 13, LBV; 14 to 21, MOKV; 22 to 27, DUVV; 28 to 33, EBLV-1; 34 to 43, EBLV-2;44 to 50: ABLV; 51 to 53, positive control; 54 to 56, negative control; 57, HEX-labeled oligomarker. Each slide contained 12 copies of the clock dial-likemicroarrays. (B) Hybridization of the 7 species of lyssaviruses using the LyssaChip. N, negative control; 1, RABV; 2, LBV; 3, MOKV; 4, DUVV; 5, EBLV-1;6, EBLV-2; 7, ABLV.

Differentiation of Lyssaviruses by Microarray

March 2012 Volume 50 Number 3 jcm.asm.org 621

on October 4, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

than the background signal (P � 0.01). These results showed thatthe LyssaChip could specifically detect and differentiate all 7 ma-jor species of lyssaviruses.

To test the sensitivity of LyssaChip, 10-fold serial dilutions ofplasmids bearing the 371-bp N gene fragment of all 7 viruses wereamplified and labeled with HEX by PCR, then hybridized toLyssaChip. The plasmid copy number was calculated using thefollowing formula: [(number of grams/�l of plasmid DNA)/(plas-mid length in bp � 660)] � 6.022 � 1023 � number of molecules/�l. The starting concentration of the plasmid was 2.46 � 109 mol-ecules/�l. The detection limit of all species was 2.46 � 105

molecules/�l, except EBLV-2, for which the limit was 2.46 � 104

molecules/�l (Fig. 2). The results also showed a good linear cor-relation (R2) between template concentration and fluorescence(graph not shown).

To compare the sensitivity of the LyssaChip with those of gel-based RT-nPCR and RT-qPCR, 10-fold serial dilutions of cell-cultured rabies virus, strain SRV9, were tested from 106.2 to 10�0.8

50% tissue culture infective doses (TCID50)/ml. The resultsshowed that the lowest limit of detection with the LyssaChip was0.158 TCID50/ml, which was 10 times more sensitive than RT-nPCR and RT-qPCR.

Cy3 or Cy5 are widely used in the labeling of target sequencesby random primer PCR, but the use of biotin-HEX in labeling isuncommon. At the beginning of the study, Cy3-dCTP (GEHealthcare) and HEX labeling methods were compared, and theresult showed that both labeling methods produced the same flu-orescence density (P � 0.05) (data not shown). However, the la-beling procedure for Cy3 is more complicated and expensive thanthat for HEX, and the latter was therefore used in our samplepreparation.

To evaluate the accuracy and specificity of the LyssaChip, 111laboratory preserved brain specimens (Table 1) were tested. In theassay of 65 rabies-suspected brain tissues, all were LyssaChip pos-itive, with only 64 and 62 positive by RT-nPCR and FAT, respec-tively. Two specimens were FAT negative, and one highly decayedbrain tissue was negative by both RT-nPCR and FAT. Among 46brain tissues of butchered dogs, 3 were positive by all three meth-ods. The assay results of the three methods are summarized inTable 3, which shows that the LyssaChip had 100% sensitivity (CI,94.48% to 100%) and 93.94% specificity (CI, 82.10% to 98.63%)compared with FAT. The LyssaChip had very high consistencywith FAT (� � 0.944) and RT-nPCR (� � 0.981).

Validation of the LyssaChip in the ILT-2010 for rabies virus.The LyssaChip was tested using International Laboratory Test2010 (ILT-2010), a set of 12 samples (one each of RABV, EBLV-1,EBLV-2, and ABLV as positive controls, 1 negative control, and 7blinded samples) received from the European Union ReferenceLaboratory for Rabies (EU-RL) at the French Agency for Food,Environmental and Occupational Health & Safety (ANSES). Thesamples were also tested by FAT, RT-nPCR, and RT-qPCR at thesame time. Results revealed that the 4 methods were 100% consis-tent, but the LyssaChip was additionally able to differentiate be-tween the species within 8 h, with all 12 samples tested on a singlechip slide, while the FAT and RT-nPCR could not differentiatespecies(Table 4).

ARAV, IRKV, KHUV, and WCBV. Since the 4 new lyssavi-ruses recently approved by the International Committee on Tax-onomy of Viruses were not available to us, their detection byLyssaChip could be estimated only by comparison of our probesequences with the corresponding viral genomic regions. Suchcomparisons showed that all 50 probes did not have the minimum25-nt continuously matched sequence essential for hybridization(data not shown). In addition, there were mismatches within the15 nucleotides at the 3= ends of all probes. Therefore, cross-hybridization of our probes with ARAV, IRKV, KHUV, andWCBV seems unlikely.

FIG 2 Sensitivity of detection by LyssaChip. The detection of serially dilutedtemplate by the LyssaChip showed high sensitivity in detecting all 7 viruses,and EBLV-2 in particular.

TABLE 3 Correlation between detection of lyssaviruses in clinicalspecimens by LyssaChip, FAT, and RT-nPCR

Standard methodand result

LyssaChip

Correlation (%)Positive Negative

FATa

Positive 65 0 97.30Negative 3b 43

RT-nPCRPositive 67 0 99.10Negative 1c 43

a The FAT was conducted using fluorescein isothiocyanate-conjugated anti-rabies virusmonoclonal antibody (Fujirebio Diagnostics Inc., Malvern, PA) (10).b The FAT-negative, LyssaChip-positive samples are the highly decayed, BD 12, andCQWX-2 brain tissue specimens (see Discussion).c The highly decayed brain tissue.

TABLE 4 Results of the International Laboratory Test 2010 for rabiesvirus

Specimen

Finding

FAT RT-nPCR RT-qPCR LyssaChipEU-RLdiagnosis

RABV control � � RABV RABV RABVEBLV-1 control � � EBLV-1 EBLV-1 EBLV-1bEBLV-2 control � � EBLV-2 EBLV-2 EBLV-2ABLV control � � � ABLV ABLVNegative control � � � � �Blinded sample 1 � � � � �Blinded sample 2 � � � � �Blinded sample 3 � � � � �Blinded sample 4 � � RABV RABV RABVBlinded sample 5 � � EBLV-1 EBLV-1 EBLV-1aBlinded sample 6 � � EBLV-2 EBLV-2 EBLV-2Blinded sample 7 � � � ABLV ABLV

Xi et al.

622 jcm.asm.org Journal of Clinical Microbiology

on October 4, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

DISCUSSION

For oligonucleotide microarray detection, target cDNA or DNA isusually amplified and labeled by PCR. The sensitivity of such amicroarray is determined by target cDNA amplification. Cur-rently, there are two main approaches. In the first, a commonprimer PCR is used to amplify a genus of pathogens (18), but itsability may be influenced by mismatched bases between primerand template. Additionally, multiplex PCR is used to amplifypathogens of several different genera (6), but it can result in sup-pression effects on some of the primer pairs, leading to false-negative results (12). In the second, the random primer-directedPCR (16, 32) and �29 polymerase-based amplification (3, 5, 9) areused to amplify a wide range of cDNAs. This procedure is widelyused in identification of new pathogens but may yield smalleramounts of target DNA copies and may bias amplification, result-ing in reduced sensitivity and specificity of a microarray (4). Ad-ditionally, random amplification methods are more complex andtime-consuming. In the present study we chose the commonprimer PCR method, i.e., using pan-lyssavirus RT-nPCR to am-plify and label the N gene fragment, since this method was previ-ously shown to detect all 7 major lyssavirus species with high sen-sitivity (20). To avoid primer/target mismatches, 2 sets of primersfor RT-nPCR were located within the most conserved region ofthe N gene of each lyssavirus, and the results showed that theLyssaChip rendered the highest sensitivity compared with theconventional methods, indicating the capability to detect low lev-els of virus particles.

The specificity of a species typing oligonucleotide microarray isdetermined by the oligoprobes. Excellent probes should accu-rately identify target cDNAs and show no cross-hybridizationwith unrelated sequences. Panels of probes were therefore de-signed for each lyssavirus species, since the existence of lineages orsubtypes within a given species makes it impossible to design acommon probe to detect all members within a species. Usually,matching stringency between probe and target is critical for hy-

bridization. It has been reported that, with regard to an oligoprobeof 60 to 70 nt, any mismatch within 15 nucleotides of the 3= endcould significantly diminish the hybridization and produce false-negative results, whereas mismatches of less than 12 nucleotidesscattered throughout a sequence at the 5= end would have limitedimpact on hybridization unless �5 of them were continuous mis-matches (24). However, Honma et al. observed that even a singlenucleotide mismatch near the 3= end of a probe significantly re-duced the hybridization signal (18), which is more consistent withour observations.

Although blinded sample 5 was determined to be EBLV-1 byLyssaChip, the difference in the hybridization of EBLV-1-specificprobes with the EBLV-1 control sample and blinded sample 5, inwhich the double-spotted probe (number 33) hybridized withsample 5 but not with the EBLV-1 control (Fig. 3B) is interesting.The background information provided by EU-RL revealed thatthese two samples were different subtypes of EBLV-1, the controlsample being EBLV-1b and the blinded one EBLV-1a (Table 4).The multiple sequence alignment of all EBLV-1 N gene sequencesavailable from GenBank showed that probe 33 belonged toEBLV-1a and had 1 mismatched nucleotide with the correspond-ing sequence of EBLV-1b at the 13th nucleotide from the 3= end(Fig. 3A). This emphasizes that when designing specific oligo-probes, the possibility of mismatched nucleotides near the 3= endshould be taken into critical consideration. Indeed, this featuremight be used to design probes for differentiating subtypes, e.g.,the number 33 probe could be used for distinguishing between the2 subtypes of EBLV-1 (Fig. 3).

The LyssaChip was tested in detection of RABV in 111 clinicalspecimens and showed comparable specificity with conventionalRT-nPCR and FAT. None of the specimens showed any nonspe-cific hybridization with the probes of the 6 other major lyssavirusspecies. Virus had previously been isolated from 27 of the 68LyssaChip-positive specimens, and the N genes had been se-quenced and found to comprise 4 lineages of RABV (15). This

FIG 3 Sequence analysis of probe 33 (boxed). Twenty-two sequences of the entire 1,353-bp lyssavirus N gene were obtained from GenBank and aligned by usingMegAlign software (version 7.0; LaserGene) with ClustalW. The region corresponding to oligoprobe 33 is shown from the 3= end (left) to 5= end (right). (A) Thetop line is the antisense sequence of probe 33 (70 nt). Sequence alignment revealed that EBLV-1 is divided into 2 lineages, EBLV-1a and EBLV-1b. Probe 33belongs to EBLV-1a, with 98.6% homology to EBLV-1a and 97.1% to EBLV-1b. There was a mismatched nucleotide with the EBLV-1b sequence at position 13from the 3= end of the probe. (B) LyssaChip detection profile of the EBLV-1 control (EBLV-1b) and blinded sample 5 (EBLV-1a).

Differentiation of Lyssaviruses by Microarray

March 2012 Volume 50 Number 3 jcm.asm.org 623

on October 4, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

showed that the LyssaChip can detect different genetic variants ofRABV. With respect to sensitivity, there were differences betweenthe three methods (Table 3): for example, a highly decayed braintissue specimen was FAT and RT-nPCR negative but LyssaChippositive. This specimen was taken from a clinically rabid dog andsubmitted by the Chongqing Center for Animal Disease Controland Prevention for laboratory confirmation. Following its receipt,it was accidentally left in the reception room for a week at roomtemperature. By the time it had been retrieved the specimen haddecayed to a very dark gray color with a strong smell. As tested bythe three methods, this specimen was positive only by LyssaChip,likely indicating that this method is the most sensitive in terms ofdetection with decayed specimens. The explanation of this dis-crepancy is that the sensitivity of FAT and RT-nPCR were signif-icantly decreased since degradation of viral RNA and nucleocap-sids could occur with decay of the tissue, indicating that theperformance of FAT on putrefied specimens is not reliable. Inaddition, two more specimens (HuNDN12 and CQWX-2) werealso FAT negative although they were RT-nPCR and chip positive(Table 3). Sample CQWX-2 was from a dog in a village of WuxiCounty, Chongqing Municipality, that did not show clinical signsbut had bitten five villagers without provocation in a single dayand was thereafter killed for rabies diagnosis. Sample HuNDN12was from a dog which was one of nine culled for safety in a villageof Hunan Province to prevent further possible rabies transmis-sion, as a human rabies case had occurred in the village. Sincethese two dogs were healthy looking with no overt clinical signswhen they were culled, the difference in detection of virus in theirbrain tissues might imply that both dogs were in the early incuba-tion stage, at which time the brain tissue viral load might havebeen too low to be detected by FAT. Calculation of the differenceby the McNemar chi-square test showed statistically significance(P � 0.0001), thereby indicating that the LyssaChip likely had ahigher positive rate. In conclusion, all three methods producedidentical results, but LyssaChip was more sensitive than FAT andeven RT-nPCR in detection of virus in highly decayed specimens.In situations where this might occur, therefore, the LyssaChipcould be more useful. Isolates of lyssaviruses other than RABVwere not tested, since these were not available in China.

The LyssaChip developed in the current study represents arapid, high-throughput, and economical method for the detectionand differentiation of the 7 major lyssavirus species. The entiredetection procedure takes about 8 h, including 1 h for RNA ex-traction, 3 h for RT-nPCR labeling, 3.5 h for hybridization, and 30min for washing and scanning. Conventional RT-nPCR and se-quencing methods are unable to accomplish the same task withinsuch a short time. Additionally, the LyssaChip can assay 12 sam-ples on a single slide and permits the simultaneous processing of atleast 7 to 8 slides at a reasonable cost (approximately $5 [U.S.] persample).

Given its high sensitivity, specificity, speed, and low cost, theLyssaChip may provide an attractive alternative for clinical labo-ratories for detection and differentiation of the 7 major lyssavi-ruses.

ACKNOWLEDGMENTS

This work was funded by the Joint Project of NSFC/Yunnan Province(U1036601), the Special Fund for Agro-scientific Research in the PublicInterest (project 201103032), and the National Key Technology R&D Pro-gram (grant 2010BAD04B01).

REFERENCES1. Amengual B, Whitby JE, King A, Cobo JS, Bourhy H. 1997. Evolution

of European bat lyssaviruses. J. Gen. Virol. 78:2319 –2328.2. Baxi MK, Baxi S, Clavijo A, Burton KM, Deregt D. 2006. Microarray-

based detection and typing of foot-and-mouth disease virus. Vet. J. 172:473– 481.

3. Berthet N, et al. 2008. Massively parallel pathogen identification usinghigh-density microarrays. Microb. Biotechnol. 1:79 – 86.

4. Berthet N, et al. 2010. High-density resequencing DNA microarrays inpublic health emergencies. Nat. Biotechnol. 28:25–27.

5. Berthet N, et al. 2008. �29 polymerase based random amplification ofviral RNA as an alternative to random RT-PCR. BMC Mol. Biol. 9:77.

6. Boriskin YS, et al. 2004. DNA microarrays for virus detection in cases ofcentral nervous system infection. J. Clin. Microbiol. 42:5811–5818.

7. Bourhy H, et al. 2008. The origin and phylogeography of dog rabies virus.J. Gen. Virol. 89:2673–2681.

8. Cliquet F, Picard-Meyer E. 2004. Rabies and rabies-related viruses: amodern perspective on an ancient disease. Rev. Sci. Technol. 23:625– 642.

9. Dacheux L, et al. 2010. Application of broad-spectrum resequencingmicroarray for genotyping rhabdoviruses. J. Virol. 84:9557–9574.

10. Dean DJ, Abelseth MK. 1996. The fluorescent antibody test, p. 88 –95. InMeslin F-X, Kaplan MM, Koprowski H (ed.), Laboratory techniques inrabies, 4th ed. World Health Organization, Geneva, Switzerland.

11. De Benedictis P, et al. 2011. Lyssavirus detection and typing using pyro-sequencing. J. Clin. Microbiol. 49:1932–1938.

12. Elnifro EM, Ashshi AM, Cooper RJ, Klapper PE. 2000. Multiplex PCR:optimization and application in diagnostic virology. Clin. Microbiol. Rev.13:559 –570.

13. Fooks AR, et al. 2009. Emerging technologies for the detection of rabiesvirus: challenges and hopes in the 21st century. PLoS Negl. Trop. Dis.3:530 –542.

14. Foord AJ, et al. 2006. Molecular diagnosis of lyssaviruses and se-quence comparison of Australian bat lyssavirus samples. Aust. Vet. J.84:225–230.

15. Gong W, et al. 2010. Temporal and spatial dynamics of rabies viruses inChina and Southeast Asia. Virus Res. 150:111–118.

16. Gurrala R, et al. 2009. Development of a DNA microarray for simulta-neous detection and genotyping of lyssaviruses. Virus Res. 144:202–208.

17. Hoheisel JD. 2006. Microarray technology: beyond transcript profilingand genotype analysis. Nat. Rev. Genet. 7:200 –210.

18. Honma S, et al. 2007. Development and validation of DNA microarrayfor genotyping group A rotavirus VP4 (P[4], P[6]. P[8], P[9], and P[14])and VP7 (G1 to G6, G8 to G10, and G12) genes. J. Clin. Microbiol. 45:2641–2648.

19. ICTV. 2009. ICTV official taxonomy: updates since the 8th report. http://talk.ictvonline.org/files/ictv_official_taxonomy_updates_since_the_8th_report/m/vertebrate-2008/1208.aspx.

20. Jiang Y, et al. 2009. Development of nested RT-PCR for detection ofanimal rabies virus. Chin. J. Vet. Sci. 29:1003–1007.

21. Kawaguchi K, et al. 2003. Detection of hepatitis B virus DNA in sera frompatients with chronic hepatitis B virus infection by DNA microarraymethod. J. Clin. Microbiol. 41:1701–1704.

22. Kim HJ, et al. 2008. Microarray detection of food-borne pathogens usingspecific probes prepared by comparative genomics. Biosens Bioelectron.24:238 –246.

23. Markotter W, Kuzmin I, Rupprecht CE, Nel LH. 2008. Phylogeny ofLagos bat virus: challenges for lyssavirus taxonomy. Virus Res. 135:10 –21.

24. Palacios G, et al. 2007. Panmicrobial oligonucleotide array for diagnosisof infectious diseases. Emerg. Infect. Dis. 13:73– 81.

25. Pas SD, et al. 2008. Comparison of reverse hybridization, microarray, andsequence analysis for genotyping hepatitis B virus. J. Clin. Microbiol. 46:1268 –1273.

26. Quan PL, et al. 2007. Detection of respiratory viruses and subtype iden-tification of influenza A viruses by GreeneChipResp oligonucleotide mi-croarray. J. Clin. Microbiol. 45:2359 –2364.

27. Sabeta CT, et al. 2007. Mokola virus in domestic mammals, South Africa.Emerg. Infect. Dis. 13:1371–1373.

28. Smith JS. 1996. New aspects of rabies with emphasis on epidemiology,diagnosis, and prevention of the disease in the United States. Clin. Micro-biol. Rev. 9:166 –176.

29. Tordo N, Kouknetzoff A. 1993. The rabies virus genome: an overview.Onderstepoort J. Vet. Res. 60:263–269.

Xi et al.

624 jcm.asm.org Journal of Clinical Microbiology

on October 4, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

30. Townsend MB, et al. 2006. Experimental evaluation of the FluChip di-agnostic microarray for influenza virus surveillance. J. Clin. Microbiol.44:2863–2871.

31. Wakeley PR, et al. 2005. Development of a real-time, TaqMan reversetranscription-PCR assay for detection and differentiation of lyssavirus ge-notypes 1, 5, and 6. J. Clin. Microbiol. 43:2786 –2792.

32. Wang D, et al. 2002. Microarray-based detection and genotyping of viralpathogens. Proc. Natl. Acad. Sci. U. S. A. 99:15687–15692.

33. Warrilow D. 2005. Australian bat lyssavirus: a recently discovered newrhabdovirus. Curr. Top. Microbiol. Immunol. 292:25– 44.

34. Wu X, Franka R, Velasco-Villa A, Rupprecht CE. 2007. Are all lyssavirusgenes equal for phylogenetic analyses? Virus Res. 129:91–103.

Differentiation of Lyssaviruses by Microarray

March 2012 Volume 50 Number 3 jcm.asm.org 625

on October 4, 2020 by guest

http://jcm.asm

.org/D

ownloaded from