RT-PCR, nucleotide, amino acid and phylogenetic analysesof enterovirus type 71 strains from Asia
Sunita Singh a, Vincent T.K. Chow a,*, K.P. Chan b, A.E. Ling b, C.L. Poh a
a Programme in Infectious Diseases, Department of Microbiology, Faculty of Medicine, National Uni6ersity of Singapore,5 Science Dri6e 2, Singapore 117597
b Department of Pathology, Singapore General Hospital, Outram Road, Singapore 169608
Received 7 February 2000; received in revised form 17 April 2000; accepted 17 April 2000
Keywords: Asian strains; Enterovirus 71; Phylogenetic analysis; RT-PCR; Sequencing
www.elsevier.com/locate/jviromet
1. Introduction
Enterovirus 71 (EV71) is a major causativeagent of hand, foot and mouth disease (HFMD),
which most commonly affects children and infants(Zheng et al., 1995). Unlike coxsackievirus A16that is more limited in its pathogenicity toHFMD, EV71 can give rise to major complica-tions involving the central nervous system (CNS).Recent studies have shown an aetiological linkbetween EV71 and brainstem encephalitis as acause of pulmonary oedema and death. In this
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204194
disease, there is involvement of the cerebellum,brainstem and diencephalon leading to seizures,coma, truncal ataxia and somnolence (Komatsuet al., 1999; Lum et al., 1998).
EV71 infections have resulted in several out-breaks in Malaysia, Taiwan and Singapore inrecent years. EV71 associated with other en-terovirus infections in Malaysia resulted in 30deaths in 1997 (AbuBakar et al., 1999) while 78deaths were reported in Taiwan in 1998 (Ho etal., 1999). In Singapore, HFMD cases were foundto be sporadic and rarely complicated by CNSinvolvement, unlike the cases occurring in neigh-bouring countries. EV71 is one of the majorcauses of HFMD in Singapore together with cox-sackievirus serotypes A16, A9, B2, B3 andechoviruses. Outbreaks of HFMD were reportedin 1993 (310 cases) and in 1997 (358 cases) char-acterized by mucocutaneous, papulovesicular rashand self-limiting febrile illness (Ministry of theEnvironment, Singapore, personal communica-tion). EV71 infection was reported in 23 patientsin 1997 and 16 patients in 1998, who were suffer-ing mainly from HFMD, myocarditis, coxsackie-like disease, hyperpyrexia, meningoencephalitisand/or acute flaccid paralysis. HFMD mainlyaffects children aged 6 months to 3 years, al-though some cases of infection in adults aged24–36 years have been reported. So far, only onecase of acute flaccid paralysis in a child wasreported.
Analyses of EV71 strains at the molecular levelcould help to explain the varied clinical patternsobserved for EV71 disease, e.g. HFMD in China,Japan and Singapore, and HFMD with CNSinvolvement in USA, Australia, Eastern Europeand recently in Taiwan and Malaysia (Komatsuet al., 1999).
EV71 possesses a single-stranded RNA genomeof approximately 7500 nucleotides of positive po-larity, and belongs to the genus of enterovirusesfrom the family Picornaviridae (Muir et al.,1998). The genome comprises a 5% untranslatedregion (5% UTR), a long open reading frame thatencodes a protein of approximately 2100 aminoacid residues, a short 3% untranslated region (3%UTR) and a polyadenylated tail. The polyproteinis co- and post-translationally cleaved to give
four structural proteins VP4, VP2, VP3 and VP1.Determination of attenuation of virulence, alteredhost range, persistent infection and in vitro celltropism have all been mapped to the capsid-en-coding region (Muir et al., 1998). The capsidregion comprising VP1 and VP4 of coxsack-ievirus B4 is known to encode virulence determi-nants (Caggana et al., 1993). However, little isknown about the virulence determinants of EV71.Comparison of the nucleotide and amino acidsequences of the non-neurovirulent EV71 strainsisolated in Singapore with neurovirulent strainsisolated elsewhere may reveal the pathogenicdeterminants.
In view of the time-consuming and laborioustechnique of virus isolation, the limited supply ofEV71-specific antisera for serotyping, and thealarming outbreaks of EV71-related diseases es-pecially in East Asia, a rapid and precise methodto identify EV71 is required urgently. Reversetranscription-polymerase chain reaction (RT-PCR) is an attractive diagnostic tool particularlyduring outbreaks given that results can be ob-tained within hours, and that serologically unty-pable strains may even be detected. It is knownthat the specific epitopes responsible for serotypicspecificity are clustered mainly in the VP1 region.In a recent study by Oberste et al. (1999), VP1sequences were shown to have a better correla-tion with enterovirus serotypes compared tothose of the 5% UTR or the VP4-VP2 junction.Thus, PCR primers designed from the VP1 regionshould correlate better with serotypic identifica-tion. We describe here the application of newly-designed and highly-specific primers for RT-PCRamplification and direct cycle sequencing of EV71strains to illustrate the genetic relationships andmolecular evolution of EV71 strains from Asia.
2. Material and methods
2.1. Sample collection
Cultured strains of EV71 and other en-teroviruses were isolated by the Virology Labora-tory, Pathology Department, Singapore GeneralHospital. A total of 44 EV71 samples and 30
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204 195
other enteroviruses including coxsackievirus typesA and B, echoviruses as well as serologicallyuntypable strains were studied (Table 1). Theseviruses were isolated from sputum, throat swab,stool, nasal aspirate, rectal swab and vesicularfluid specimens of patients suffering from HFMDand/or other clinical symptoms. All the sampleswere inoculated into HeLa, HEp-2, RD andMRC-5 cell lines, and the virus isolates weretyped by the microneutralization method usingthe Lim Benyesh-Melnick (LBM) antisera pool orthe alternating intersecting pools developed at theNational Institute of Public Health and the Envi-ronment in the Netherlands. The EV71-specificantiserum was from Light Diagnostics, ChemiconInternational, CA, USA. The viruses were har-vested and stored at −70°C. The neurovirulentstrain EV71/7423/MS/87 was a gift from Dr M.A.Pallansch, CDC, Atlanta, GA, USA. The cellcultures of EV71 strains from Malaysia and Japanwere kindly provided by Professor S.K. Lam fromthe University of Malaya, Malaysia.
2.2. Primer design
Two pairs of primers VP1F2/EV71R2 andEV71F4/EV71R5 were designed to amplify targetfragments of 341 bp and 612 bp within the VP1region and 5% UTR of EV71, respectively. Asshown in Table 2, the first eight primers were alsodesigned to amplify and/or sequence the capsidcoding region (VP1, VP2, VP3, VP4) and part ofthe 5% UTR of two local EV71 strains.
2.3. RNA extraction and RT-PCR
The virus-infected cells were lysed using theSDS-urea method, and RNA was purified usingchloroform/isoamyl alcohol extraction followedby ethanol precipitation (Gough, 1988). The driedRNA pellet was reconstituted in sterile distilledwater containing RNasin ribonuclease inhibitor(Promega, Madison, WI, USA) and stored at−70°C prior to use. The yield of RNA wasdetermined by optical density measurements at260 nm. Total RNA (2–5 mg) was mixed with 0.3mg of random hexamer oligonucleotides (LifeTechnologies, Gaithersburg, MD, USA) and incu-
bated at 70°C for 10 min. To this reaction mixturewas added first strand buffer, 0.01 M dithiothrei-tol, 0.5 mM of dNTP mix and 10 U of reversetranscriptase (Life Technologies) in a final volumeof 20 ml. The RT reaction was incubated at 42°Cfor 50 min and heat inactivated at 70°C for 15min. The cDNA was stored at –70°C and used astemplate for subsequent PCR experiments.
With primer pairs EV7116F/EV7116R andEV71F4/EV71R4, long distance PCR catalyzedby KlenTaq2 polymerase with proof-reading ac-tivity (Clontech, Palo Alto, CA, USA), was em-ployed to amplify the complete capsid codingregion and part of the 5% UTR of Singapore EV71strains 13/Sin/98 and 18/Sin/97. A 5 ml aliquot ofthe cDNA reaction product was added to thePCR mixture containing 1×PCR buffer, 0.2 mMeach of the primers, 0.2 mM dNTP mix and 0.4 Uof KlenTaq2 polymerase in a total reaction vol-ume of 50 ml.
The EV71 specificity of the primer pairsVP1F2/EV71R2 and EV71F4/EV71R5 weretested on the enteroviral cDNA samples usingAmpliTaq DNA polymerase (Roche MolecularSystems, Branchburg, NJ, USA). A 5 ml aliquot ofeach cDNA template was used in a total reactionvolume of 50 ml containing 1×PCR buffer, 0.2mM dNTP mix, 0.2 mM each of forward andreverse primers, and 2.5 U of AmpliTaq poly-merase. PCR was carried out at an initial 95°Cfor 1 min, followed by 30 cycles of 95°C for 30 s,55°C for 30 s and 68°C for 3 min. The primer pairVP1F2/EV71R2 that was found to be specificonly for EV71 was then used to amplify all thecDNA samples using KlenTaq DNA polymerase.The PCR products (10 ml each) were subjected toelectrophoresis in a 1.6% agarose gel, with a 100bp DNA ladder (New England Biolabs, Beverly,MA, USA) serving as a molecular mass marker.
Tissue culture infective dose (TCID50) of the7423/MS strain was derived using the microwellplate method and calculated by the formula ofReed and Muench (Lennette and Schmidt, 1969).The sensitivity of the RT-PCR assay was testedby amplifying tenfold serial dilutions of RNAextracted from the virus suspension of knownTCID50.
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204196
a Strains were assigned by the year of isolation, internal lab identifier and the country of origin, i.e., Singapore (Sin), Malaysia(Mal) and Japan (Jap). Strain 47/Mal was isolated from a brain tissue sample of a patient with meningoencephalitis, while strain18/Sin was from a stool sample of a patient with acute flaccid paralysis.
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204 197
Table 2Oligonucleotide primers for RT-PCR and cycle sequencing of enterovirus 71
To generate clean templates for DNA sequenc-ing, PCR was repeated with KlenTaq polymerase,specific DNA bands were excised from the gel,eluted and purified using a DNA gel extraction kit(Qiagen, Germany).
Purified PCR products were cycle-sequenced inboth directions using the ABI PRISM BigDyeTerminator cycle sequencing ready reaction kit(PE Applied Biosystems, Foster City, CA, USA).DNA sequencing was performed using the ABIPRISM 377 DNA Sequencer (PE Applied Biosys-tems). The DNASIS analysis software was usedfor conversion between RNA and DNA se-quences, and for alignment. Predicted amino acidsequences were derived by translating nucleotidesequences using the ORF Finder (NCBI). Thesequence data of our EV71 samples have beendeposited with the GenBank database. Eachbatch of PCRs included a known positive controltogether with a ‘no DNA’ negative control toexclude the possibility of false positive reactiondue to reagent contamination. PCR was also car-ried out with RNA extracted from uninfected RDcells to exclude non-specific results. All PCRswere repeated to exclude false positive resultsarising from cross contamination.
Homology searches were carried out using theBLAST algorithm (http://www.ncbi.nlm.nih.gov/BLAST). The percentage of sequence identity wascalculated with the Blast 2 sequences programme(NCBI) that produces alignment of two given
sequences using the BLAST engine for local align-ment (Tatusova and Madden, 1999). The phyloge-netic pattern of the South East Asian strains ofEV71 was studied on the basis of the tree con-structed from the 341-bp VP1 fragments flankedby specific primers. Multiple alignments and den-drograms were constructed using the TREECONprogramme, version 1.3b (Van de Peer and DeWachter, 1993). The genogroups were generatedby the neighbour joining method rooted to the7423/MS strain. The branched lengths were deter-mined by the maximum likelihood method andthe reliability of the neighbour joining methodwas estimated by bootstrap analysis of 1000 pseu-doreplicate datasets using SEQBOOT analysis.Hydropathic profiles of the predicted proteins ofstrains 13/Sin/98 and 18/Sin/97 were plotted bythe method of Kyte and Doolittle (1982).
3. Results
3.1. Specificity and sensiti6ity of RT-PCR assay
Using primers VP1F2 and EV71R2, RT-PCRgenerated target bands of 341 bp (Fig. 1) for theneurovirulent strain 7423/MS as well as for 1, 8and 34 EV71 strains from Japan, Malaysia andSingapore, respectively. None of the other en-terovirus types that served as negative controlswere amplified by RT-PCR with these primers.The RT-PCR results correlated well with theserotyping data by neutralization assay, thus em-
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204198
phasizing the high specificity of these primers forEV71. However, RT-PCR with primers EV71F4and EV71R5 (Table 2) did not correlate well withserotyping.
The TCID50 of the 7423/MS strain was found tobe 106.5 per inoculum volume of 0.025 ml. RT-PCRcould amplify a minimum equivalent of 101.5 orapproximately 10 virus-infected cells per inoculumvolume as shown in Fig. 1, thereby substantiatingthe sensitivity of RT-PCR with the primers and thepotential applicability for detecting low virus titresin clinical specimens.
3.2. Nucleotide and amino acid sequence analysisof two local EV71 strains
Using primers EV7116F and EV7116R (Table 2),a 2.9 kb fragment was amplified from each of the twoSingapore EV71 strains (Fig. 1). Strain 13/Sin/98was isolated from a case of HFMD, while strain18/Sin/97 originated from a female infant with acuteflaccid paralysis. These EV71 strains were thensequenced and compared with known virulentstrains isolated elsewhere via computational analy-sis of nucleotide and amino acid sequences, and toyield accurate sequence information for designingspecific primers based on highly conserved regionswithin the VP1 gene.
Strains 13/Sin/98 and 18/Sin/97 showed 95%nucleotide identity to each other, reaching as high as
96% in the 5% UTR. The nucleotide sequences ofthese strains showed 92% identity to the neuroviru-lent strain EV71/7423/MS/87, first isolated from aMississippi boy with paralysis in 1987 (Brown andPallansch, 1995). Similarities to the nucleotide se-quences of the other EV71 strains (in the GenBankdatabase) including the prototype EV71/BrCr strainwere found to be less than 85%.
Comparison of the amino acid sequences of strain13/Sin/98, 18/Sin/97 with 7423/MS and BrCr proto-type strains revealed 100% identity of the VP4 region(Fig. 2). Within the VP2 region, there were twovariable sites, i.e. residue 197 (serine in 13/Sin/98 butproline in 18/Sin/97, 7423/MS and BrCr strains) andresidue 315 (where glycine in 7423/MS was replacedby alanine in both local strains and the BrCr strain).There were additional amino acid disparities inthe BrCr strain at positions 114, 163, 195 and 212.
In the VP3 region, leucine at residue 329 (in7423/MS and BrCr) and at residue 538 (in 7423/MSalone) was conservatively substituted with proline(329) in the local strains, and with isoleucine (538)in the local strains as well as the BrCr strain,respectively. Histidine at position 557 of 7423/MSand 13/Sin/98 was replaced by arginine in strain18/Sin/97 and by aspartic acid in BrCr. A compari-son between BrCr and the other three strains re-vealed that glutamic acid replaced glutamine atposition 560, and threonine substituted serine atposition 563.
Fig. 1. Visualization of RT-PCR products of enterovirus 71 by gel electrophoresis. A 100 bp DNA ladder molecular mass markerwas included in lane M. Lanes 1 and 2 represent RT-PCR-amplified 2.9-kb fragments of 13/Sin/98 and 18/Sin/97, respectively. Lanes3 and 4 represent RT-PCR-positive 341-bp fragments of 13/Sin/98 and 18/Sin/97, respectively. Lanes 5 to 11 represent RT-PCRproducts of neat and of ten-fold serial dilutions of EV71/7423/MS strain (with TCID50 of 106.5 per 0.025 ml of inoculum), i.e. neat,10−1, 10−2, 10−3, 10−4, 10−5 and 10−6, respectively. No product was visible at 10−6 dilution (lane 11). Lanes 12 to 15 representRT-PCR-negative results of CA16, CB2, echo 7 and untypable strains, respectively. RT-PCR of RNA from uninfected RD cells(lane 16) and a ‘‘no DNA’’ sample (lane 17) served as negative controls.
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204 199
Fig. 2. Comparison of predicted amino acid sequences ofSingapore EV71 strains 13/Sin/98 and 18/Sin/97 (GenBankaccession numbers AF251358 and AF251359) with those ofstrains BrCr and 7423/MS. Residues 1-926 of the four strainsare aligned, with disparities highlighted in bold letters. Aster-isks denote identity, colons indicate conserved residues whileperiods depict variable regions.
MS and BrCr strains demonstrated numerous dis-parities, including at residues 583, 587, 595, 596,608, 623, 663, 710, 729, 732, 737, 748, 749, 802,809, 811, 814, 840, 847 and 857.
The results showed that the VP4 region dis-played complete homology among the prototypeBrCr, the neurovirulent 7423/MS and the twolocal strains. Relatively few mutations were notedin the VP2 and VP3 regions, whereas the VP1region showed the greatest variation.
3.3. Analysis of the VP1 target region flanked byEV71-specific primers
Twenty out of the 44 PCR products amplifiedby EV71-specific primers were randomly selectedand sequenced, i.e. 13 Singapore, one Japaneseand six Malaysian strains (Fig. 3). Nucleotidesequences of the strains showed 91–93% homol-ogy to the bases coding for the VP1 region of the7423/MS strain.
The predicted amino acid sequences showedalmost complete homology to the 7423/MS strainexcept at position 663 where glutamic acid wasreplaced with lysine in five strains, and at position710 where glutamic acid was substituted by glu-tamine in five strains and by glycine in anotherstrain. In addition, tryptophan at position 736was replaced by glycine in 34/Mal/98. The corre-sponding amino acid sequences of nine strainswere completely identical to those of the 7423/MSstrain.
3.4. Phylogenetic analysis
A phylogenetic tree was constructed based onthe nucleotide sequences of the 341-bp VP1 frag-ments of selected strains studied by Brown et al.(1999), as well as those of Asian strains sequencedby us and those available in the GenBank data-base. These include the prototype BrCr, neuro-virulent 7423/MS, 13 Singapore, 30 Malaysian,31 Taiwanese, one Republic of China, oneJapanese and ten Australian strains. The G10strain of coxsackievirus A16 was included as anoutgroup.
The dendrogram revealed five distinct geno-types (Fig. 4). The genotypes A, B and C were
Interestingly, comparative analysis of the VP1amino acid sequences of the two Singapore, 7423/
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204200
classified as proposed by Brown et al. (1999). Theprototype strain BrCr was the sole member ofgenotype A. The neurovirulent strain 7423/MS,two Taiwanese strains and five Australian strainsrepresented genotype B. Strains of this group (otherthan 7423/MS itself) showed 93–94% identity tothe 7423/MS strain. Sixteen Taiwanese strains, fiveAustralian, one Malaysian and one Chinese strain
Fig. 4. Dendrogram based on nucleotide divergence of the341-bp VP1 gene fragments of Asian strains of EV71. Thestrains available in the GenBank are each indicated by theiraccession number followed by country and year of isolation.
Fig. 3. Alignment of representative VP1 nucleotide sequencesof EV71 strains. Strains 13/Sin/98, 15/Sin/98, 16/Sin/98, 20/Sin/98, 26/Sin/98 and 27/Sin/98 differed only by 9 bases andare exemplified by strain 26/Sin/98. Similarly, strains 28/Mal/98, 33/Mal/98 and 34/Mal/98 differed only by two bases andare represented by 28/Mal/98. Strains 18/Sin/97, 56/Sin/97 and57/Sin/97 differed by 11 bases and are represented singly bystrain 18/Sin/97. Strains 36/Sin/97 and 58/Sin/97 were differentonly by two bases and are represented by 36/Sin/97. Asterisksdenote identity. The nucleotide sequences of these strains havebeen deposited in GenBank under accession numbersAF251223-AF251236 and AF251802-251805.
together represented genotype C which showed81–83% homology to the 7423/MS strain. Thisstudy revealed two new genogroups designated Dand E. Genotype D constituted the biggest clusterin this study, including all the strains from Singa-pore and Malaysia, except one Malaysian strainbelonging to genogroup C. One strain each fromTaiwan and Japan were also included in genogroupD. Strains in genogroup D showed 92–93% nucle-otide identity with the 7423/MS strain. TwelveTaiwanese strains were all distinct and included ina separate genotype designated E. The strains ofthis group had less significant homology to the7423/MS strain than the homology of 85–87% tothe coxsackievirus A16 strain. The genotypes B, C,D and E could be further subdivided into clusterswithin each genotype.
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204 201
As additional confirmation, a dendrogram gen-erated based on 891 nucleotides of the entire VP1region of the Singapore strains 13/Sin/98 and18/Sin/97, together with selected strains reportedby Brown et al. (1999), failed to group these twolocal strains into any of the three genotypes, thusreiterating that they were distinct, belonging to anentirely new genotype D.
4. Discussion
The classical detection of EV71 is by virusisolation via cell culture followed by microneu-tralization tests (Muir et al., 1998). The cultureand identification of EV71 is an intensive andtime-consuming process requiring 2–3 weeks, of-ten hindered by slow progression of cytopathiceffects and viral aggregation (Rotbart et al., 1994;Muir et al., 1998). Reliance on such conventionaltyping schemes during large outbreaks would im-pede the implementation of control measures, anddelay medical treatment options. Many patientswith EV71-related hyperpyrexia, myocarditis andencephalitis are unnecessarily treated with antibi-otics, antiherpes drugs or both while awaitingculture results to exclude bacterial or herpeticinfections. Furthermore, the limited supply ofEV71-specific antisera for neutralization testscould have serious implications during outbreaks.In emergency outbreak situations, a rapid andreliable typing method would justify the isolationof patients for quarantine purposes and the initia-tion of prompt preventive measures, especially ininstitutions associated with children. Availabilityof a rapid typing method would also facilitatedecision-making regarding the administration ofspecific immunoglobins to selected high-risk pa-tients. Rapid diagnosis is of relevance given thepossibility of therapeutic intervention with newanti-enteroviral agents such as pleconaril (Pevearet al., 1999).
The failure to identify EV71 by the neutraliza-tion method has been observed in many laborato-ries, and is of concern especially with regard topoliomyelitis eradication (Muir et al., 1998; VanLoon et al., 1999). Since EV71 may mimic theacute flaccid paralysis of polio, such patients may
be suspected to harbour polioviruses. EV71 isfound to be more difficult than the other en-teroviruses to be recognized serotypically, andmay require treatment with chloroform for opti-mal serotyping (Van Loon et al., 1999). EV71replicates poorly in most cell cultures, no singleline currently supports the propagation of allstrains, and specific antiserum is not widely avail-able (Alexander et al., 1994; Van Loon et al.,1999). The use of virus isolation and serotypingmight therefore be expected to decrease in favourof RT-PCR for the rapid diagnosis of enterovirusinfections. For facilitating early detection of circu-lating enterovirus types, particularly for patientswith aseptic meningitis, RT-PCR constitutes amarkedly more sensitive and specific moleculartyping scheme.
For the direct detection of EV71 in clinicalsamples, this two-step RT-PCR could be easilyimproved upon by adopting a more convenientsingle-step RT-PCR method, coupled with the useof rapid commercial RNA extraction kits. Toillustrate this, we extracted RNA from ten ran-dom samples using the Oligotex direct RNA ex-traction kit (Qiagen), and obtained convincingRT-PCR results (data not shown). Such modifica-tions can enhance the rapidity and simplicity ofthe RT-PCR protocol. The results obtained havecorrelated well with the neutralization assay. Be-cause of the practical problems associated withcell cultures, molecular typing by RT-PCR andsequencing offers a promising approach for differ-entiating clinical isolates of EV71 from otherenteroviruses.
Using a specific primer pair designed for EV71in this study, the correlation between serotyping,RT-PCR amplification and cycle sequencing re-sults showed that this primer pair was sufficientfor amplification and sequencing of the partialVP1 region from all the 44 EV71 strains isolatedfrom various geographic locations. PCR coupledwith sequence analyses of other viruses have beenreported, e.g., for identification of adenoviruses inconjuctival scrapings (Takeuchi et al., 1999) anddengue viruses in serum samples (Seah et al.,1995a,b).
It is essential to understand the epidemiology ofEV71 infection to monitor the emergence of new
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204202
strains and to facilitate early rapid detection ofco-circulating strains during outbreak periods.Carefully optimized PCR assays are at least assensitive as cell culture, while their accessory se-quence data can elucidate biological informationnot obtainable by serotyping methods (Muir etal., 1998). Similar molecular detection schemeshave been developed for typing enterovirusescausing aseptic meningitis where the VP1 se-quences correlated with the serotypes (Oberste etal., 1999). More extensive sequencing studies willdefine the limits of diversity and provide a basisfor improving molecular techniques of strainidentification.
The phylogenetic analysis of EV71 strains aidsthe study of their molecular evolution and theirgenetic relationships with strains from differentoutbreaks. In an earlier investigation undertakenby Brown et al. (1999), only three EV71 isolatesfrom Asia (two from Malaysia and one fromRepublic of China) were studied. In our study, wehave also included strains from Taiwan, Singa-pore and Japan. For routine molecular diagnosticand typing purposes, sequencing of the 341 basesof the VP1 region revealed sufficient data to dif-ferentiate the various EV71 strains from Asia. Itwas observed that some strains clustered in pat-terns similar to those generated by Brown et al.(1999) based on the complete VP1 region, e.g., USstrains were mainly clustered in the majorgenogroups B and C. In their study, theMalaysian strain 0731-MAA-97 (GenBank acces-sion no. AF135911), was observed to be related tobut distinct from strains of genogroup B, andwould hence justify its inclusion in a completelynew genotype designated D together with theother strains from Singapore and Malaysia. Ourstudy thus revealed a new genotype D, whichrepresented all the strains isolated from Singa-pore, Japan and Malaysia except for strain 0756-MAA-97 (GenBank accession no. AF135935)which was placed in genogroup C. It was alsoobserved that there were two major co-circulatinggenotypes of Taiwanese strains. The Taiwanesestrains belonging to genotype C were similar tothe 7423/MS strain, while those of genotype Eexhibited greater similarity to coxsackievirus A16.A few Taiwanese strains were also observed to
belong to genotypes B and D, thus emphasizingthe genetic diversity of Taiwanese strains (Ho etal., 1999).
The results clearly demonstrate the variabilityof EV71 genomic patterns in different geographi-cal locations. It is noteworthy that although thestrains from Singapore were isolated around thesame period as those causing the outbreak inTaiwan, they displayed considerable molecular di-versity and clinical patterns. The local strainsshowed no tendency towards neurovirulence ex-cept for a single case of mild paresis. These weretotally new strains prevalent in Singapore, mainlycausing hand, foot and mouth disease. Other envi-ronmental and host factors such as age, nutrition,genetic constitution, pre-existing immunity maycontribute to the pathogenicity of EV71, culmi-nating in varying clinical expressions.
The outer capsid proteins displayed highestvariation within and between the phylogeneticclusters, implying that evolution is more conspicu-ous in the VP1, VP2 and VP3 regions. En-teroviruses generally exhibit a high degree ofheterogeneity due to the relatively higher rate ofnucleotide misincorporation during viral RNAreplication occurring both in vitro and in vivo(Muir et al., 1998). Chimeric constructs are help-ful for substantiating an association of these re-gions with virulence determination. To elucidatethe molecular basis of virulence, a panel of recom-binant chimeric viruses can be constructed fromvirulent and non-virulent strains to ascertain theirin vitro translatory effects in pathogenicity. Spe-cific mutations thus identified can be analyzed fortheir role in inhibiting neurovirulence, enablingthe mapping of the genetic determinants of viru-lence. Analysis of individual mutations in bothVP1 and VP2 regions revealed that a singleresidue determined the virulent phenotype in cox-sackievirus B4 (Ramsingh et al., 1992; Caggana etal., 1993; Ramsingh and Collins, 1995). Previousstudies also identified five amino acid substitu-tions within the VP1, VP2 and VP4 capsidproteins of the virulent coxsackievirus B4, and thepresence of major T-cell epitopes on VP3 of Theil-er’s virus in demyelination–susceptible SJL/J mice(Yauch et al., 1995).
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204 203
It has been observed that the surfaces of rhi-noviruses and polioviruses contain a series ofremarkably deep crevices forming the so-called‘canyon’, which is formed roughly at the junctionof VP1 with VP2 and VP3. Occurring aroundeach 5-fold vortex in human rhinovirus (HRV),the canyon is the site of receptor attachment. Italso represents a strategy for the viruses to escapethe host’s immune surveillance by protecting thereceptor attachment site in a surface depression(Rossmann, 1994). The amino acid residues liningthe canyon are significantly more conserved thanother surface-exposed residues (Chapman andRossmann, 1993). Site-directed mutagenesis ofHRV14 indicated that modification of severalamino acid residues located in the base of thecanyon affects virus-receptor affinity (Colonno etal., 1988). Furthermore, Wien et al. (1997) havedemonstrated a range of structural consequencesarising from poliovirus receptor mutations involv-ing VP2 residue 142 and VP1 residues 95, 158,160, 166, 226, 228, 239 and 241. Intriguingly, thecorresponding amino acids were conserved in thetwo Singapore (13/Sin/98 and 18/Sin/97), 7423/MS and BrCr strains of EV71. Notwithstandingthis, the residues that showed variations betweenthese four strains (some of which resided in hy-drophobic regions) may influence the canyonstructure and thus virus-receptor attachment.Comparison of the conformational structures ofthe canyons of these strains as predicted by com-puter-aided modelling could provide clues ontheir virus-receptor affinity and other properties.
The availability of EV71 sequence data willenable plasmid constructs incorporating the 5%UTR and the various capsid coding regions to betargeted as candidates for DNA vaccines againstthe neurovirulent enterovirus strains, as well asvalidating their ability to express the desiredproteins and to elicit antibodies against them.Immunodominant epitopes of Theiler’s murineencephalomyelitis viruses have been localized tothe capsid proteins VP1, VP2 and VP3 (Yauch etal., 1995). Mice have been protected against lethalcoxsackievirus B3 infection by DNA immuniza-tion using cDNA constructs of capsid protein(Henke et al., 1998).
Thus, the rapid identification of EV71 throughRT-PCR and direct sequencing of PCR ampliconswill help to elucidate the genotypes involved in anoutbreak and enable prompt implementation ofmeasures that will minimize further transmissionof viruses. The data will also aid in the long-termgoal of developing a DNA vaccine effectiveagainst EV71 in view of its recent association withalarming outbreaks of severe neurologicaldiseases.
Acknowledgements
We are grateful to M.C. Phoon and S.K. Tayfor their excellent technical assistance and to M.Sakharkar for biocomputing advice. Our sincerethanks to Dr M.A. Pallansch and Professor S.K.Lam for providing some EV71 strains for ourstudy.
References
AbuBakar, S., Chee, H.Y., Al-Kobaisi, M.F., et al., 1999.Identification of enterovirus 71 isolates from an outbreakof hand, foot and mouth disease (HFMD) with fatal casesof encephalomyelitis in Malaysia. Virus Res. 61, 1–9.
Brown, B.A., Pallansch, M.A., 1995. Complete nucleotidesequence of enterovirus 71 is distinct from poliovirus.Virus Res. 39, 195–205.
Brown, B.A., Oberste, M.S., Alexander, J.P., Kennett, M.L.,Pallansch, M.A., 1999. Molecular epidemiology and evolu-tion of enterovirus 71 strains isolated from 1970 to 1998. J.Virol. 73, 9969–9975.
Caggana, M., Chan, P., Ramsingh, A., 1993. Identification ofa single amino acid residue in the capsid protein VP1 ofcoxsackievirus B4 that determines the virulent phenotype.J. Virol. 67, 4797–4803.
Chapman, M.S., Rossmann, M.G., 1993. Comparison of sur-face properties of picornaviruses: strategies for hiding thereceptor site from immune surveillance. Virology 195,745–756.
Colonno, R.J., Condra, J.H., Mizutani, S., et al., 1988. Evi-dence for the direct involvement of the rhinovirus canyonin receptor binding. Proc. Natl. Acad. Sci. USA 85, 5449–5453.
Gough, N.M., 1988. Rapid and quantitative preparation ofcytoplasmic RNA from small numbers of cells. Anal.Biochem. 173, 93–95.
S. Singh et al. / Journal of Virological Methods 88 (2000) 193–204204
Henke, A., Wagner, E., Whitton, J.L., Zell, R., Stelzner, A.,1998. Protection of mice against lethal coxsackievirus B3infection by using DNA immunization. J. Virol. 72, 8327–8331.
Ho, M., Chen, E.R., Hsu, K.H., et al., 1999. An epidemic ofenterovirus 71 infection in Taiwan. N. Engl. J. Med. 341,929–935.
Komatsu, H., Shimizu, Y., Takeuchi, Y., Ishiko, H., Takada,H., 1999. Outbreak of severe neurologic involvement asso-ciated with enterovirus 71 infection. Pediatr. Neurol. 20,17–23.
Kyte, J., Doolittle, R.F., 1982. A simple method for displayingthe hydropathic character of a protein. J. Mol. Biol. 157,105–132.
Lennette, E.H., Schmidt, N.J., 1969. Diagnostic Proceduresfor Viral and Rickettsial Infections. American PublicHealth Association, New York, pp. 116–117.
Lum, L.C.S., Wong, K.T., Lam, S.K., et al., 1998. Fatalenterovirus 71 encephalomyelitis. J. Pediatr. 133, 795–798.
Muir, P., Kammerer, U., Korn, K., et al., 1998. Moleculartyping of enteroviruses: current status and future require-ments. Clin. Microbiol. Rev. 11, 202–227.
Oberste, M.S., Maher, K., Kilpatrick, D.R., et al., 1999.Typing of human enteroviruses by partial sequencing ofVP1. J. Clin. Microbiol. 37, 1288–1293.
Pevear, D.C., Tull, T.M., Seipel, M.E., Groarke, J.M., 1999.Activity of pleconaril against enteroviruses. Antimicrob.Agents Chemother. 43, 2109–2115.
Ramsingh, A., Araki, H., Bryant, S., Hixson, A., 1992. Iden-tification of candidate sequences that determine virulencein coxsackievirus B4. Virus Res. 23, 281–292.
Ramsingh, A.I., Collins, D.N., 1995. A point mutation in theVP4 coding sequence of coxsackievirus B4 influences viru-lence. J. Virol. 69, 7278–7281.
Rotbart, H.A., Sawyer, M.H., Fast, S., et al., 1994. Diagnosisof enteroviral meningitis by using PCR with a colorimetricmicrowell detection assay. J. Clin. Microbiol. 32, 2590–2592.
Seah, C.L.K., Chow, V.T.K., Chan, Y.C., 1995b. Semi-nestedPCR using NS3 primers for the detection and typing ofdengue viruses in clinical serum specimens. Clin. Diagn.Virol. 4, 113–120.
Takeuchi, S., Itoh, N., Uchio, E., Aoki, K., Ohno, S., 1999.Serotyping of adenoviruses on conjunctival scrapings byPCR and sequence analysis. J. Clin. Microbiol. 37, 1839–1845.
Tatusova, T.A., Madden, T.L., 1999. Blast 2 sequences—anew tool for comparing protein and nucleotide sequences.FEMS Microbiol. Lett. 174, 247–250.
Van de Peer, Y., De Wachter, R., 1993. TREECON: a soft-ware package for the construction and drawing of evolu-tionary trees. Comput. Appl. Biosci. 9, 177–182.
Van Loon, A.M., Cleator, G.C., Ras, A., 1999. Externalquality assessment of enterovirus detection and typing.Bull. World Health Org. 77, 217–223.
Yauch, R.L., Kerekes, K., Saujani, K., Kim, B.S., 1995.Identification of a major T-cell epitope within VP3 aminoacid residues 24 to 37 of Theiler’s virus in demyelination-susceptible SJL/J mice. J. Virol. 69, 7315–7318.
Zheng, Z.M., He, P.J., Caueffield, D., et al., 1995. Enterovirus71 isolated from China is serologically similar to theprototype E71 BrCr strain but differs in the 5%-noncodingregion. J. Med. Virol. 47, 161–167.
