LETTER
Isolation and characterization of a bat SARS-likecoronavirus that uses the ACE2 receptor
Xing-YiGe
1
*
,Jia-LuLi
1
*
,Xing-LouYang
1
*
,AlekseiA.Chmura
2
,GuangjianZhu
2
,JonathanH.Epstein
2
,JonnaK.Mazet
3
,BenHu
1
,Wei Zhang
1
, Cheng Peng
1
, Yu-Ji Zhang
1
, Chu-Ming Luo
1
, Bing Tan
1
, Ning Wang
1
, Yan Zhu
1
, Gary Crameri
4
, Shu-Yi Zhang
5
,Lin-Fa Wang
4,6
, Peter Daszak
2
& Zheng-Li Shi
1
The2002–3pandemiccausedbysevereacuterespiratorysyndromecoronavirus(SARS-CoV)wasoneofthemostsignificantpublichealtheventsinrecenthistory
1
.AnongoingoutbreakofMiddleEastrespira-torysyndromecoronavirus
2
suggeststhatthisgroupofvirusesremainsakeythreatandthattheirdistributioniswiderthanpreviouslyrecog-nized. Although bats have been suggested to be the natural reservoirsof both viruses
3–5
, attempts toisolate theprogenitor virusof SARS-CoV from bats have been unsuccessful. Diverse SARS-like corona- viruses (SL-CoVs) have now been reported from bats in China,Europe and Africa
5–8
, but none is considered a direct progenitor ofSARS-CoVbecauseoftheirphylogeneticdisparityfromthisvirusandtheinabilityoftheirspikeproteinstousetheSARS-CoVcellular receptor molecule, the human angiotensin converting enzyme II(ACE2)
9,10
.Herewereportwhole-genomesequencesoftwonovelbatcoronavirusesfromChinesehorseshoebats(family:Rhinolophidae)inYunnan,China:RsSHC014andRs3367.ThesevirusesarefarmorecloselyrelatedtoSARS-CoVthananypreviouslyidentifiedbatcoro-naviruses, particularly in the receptor binding domain of the spikeprotein. Most importantly, we report the first recorded isolation of a live SL-CoV (bat SL-CoV-WIV1) from bat faecal samples in VeroE6cells,whichhastypicalcoronavirusmorphology,99.9%sequenceidentitytoRs3367andusesACE2fromhumans,civetsandChinesehorseshoe bats for cell entry. Preliminary
in vitro
testing indicatesthatWIV1alsohasabroadspeciestropism.Ourresultsprovidethestrongest evidence to date that Chinese horseshoe bats are naturalreservoirs of SARS-CoV, and that intermediate hosts may not benecessaryfordirecthumaninfectionbysomebatSL-CoVs.Theyalsohighlighttheimportanceofpathogen-discoveryprogramstargeting high-risk wildlife groups in emerging disease hotspots as a strategy for pandemic preparedness.
The 2002–3 pandemic of SARS
1
and the ongoing emergence of theMiddleEastrespiratorysyndromecoronavirus(MERS-CoV)
2
demon-strate that CoVs are a significant public health threat. SARS-CoV wasshowntousethehuman ACE2 moleculeasitsentryreceptor,andthisisconsideredahallmarkofitscross-speciestransmissibility
11
.Thereceptorbinding domain (RBD) located in the amino-terminal region (aminoacids318–510)oftheSARS-CoVspike(S)proteinisdirectlyinvolvedin binding to ACE2 (ref. 12). However, despite phylogenetic evidencethat SARS-CoV evolved from bat SL-CoVs, all previously identifiedSL-CoVshavemajorsequencedifferencesfromSARS-CoVintheRBDoftheirSproteins,includingoneortwodeletions
6,9
.ReplacingtheRBDofoneSL-CoVSproteinwithSARS-CoVSconferredtheabilitytousehumanACE2andreplicateefficientlyinmice
9,13
.However,todate,noSL-CoVshavebeenisolatedfrombats,andnowild-typeSL-CoVofbatorigin has been shown to use ACE2.Weconducteda12-monthlongitudinalsurvey(April2011–September2012)ofSL-CoVsinacolonyof
Rhinolophussinicus
atasinglelocationinKunming,YunnanProvince,China(ExtendedDataTable1).Atotalof117analswabsorfaecalsampleswerecollectedfromindividualbatsusing a previously published method
5,14
. A one-step reverse transcrip-tion (RT)-nested PCR was conducted to amplify the RNA-dependentRNApolymerase(RdRP)motifsAandC,whichareconservedamong alphacoronaviruses and betacoronaviruses
15
.Twenty-seven of the 117 samples (23%) were classed as positive by PCRandsubsequentlyconfirmedbysequencing.Thespeciesoriginof all positive samples was confirmed to be
R. sinicus
by cytochrome bsequence analysis, as described previously
16
. A higher prevalence wasobserved in samples collected in October (30% in 2011 and 48.7% in2012)thanthoseinApril(7.1%in2011)orMay(7.4%in2012)(ExtendedData Table 1). Analysis of the S protein RBD sequences indicated thepresence of seven different strains of SL-CoVs (Fig. 1a and ExtendedDataFigs1and2).InadditiontoRBDsequences,whichcloselymatchedpreviouslydescribedSL-CoVs(Rs672,Rf1andHKU3)
5,8,17,18
,twonovelstrains (designated SL-CoV RsSHC014 and Rs3367) were discovered.Their full-length genome sequences were determined, and both werefound to be 29,787 base pairs in size (excluding the poly(A) tail). TheoverallnucleotidesequenceidentityofthesetwogenomeswithhumanSARS-CoV(Tor2strain)is 95%,higherthan thatobservedpreviously forbatSL-CoVsinChina(88–92%)
5,8,17,18
orEurope(76%)
6
(ExtendedData Table 2 and Extended Data Figs 3 and 4). Higher sequence iden-tities were observed at the protein level between these new SL-CoVsand SARS-CoVs (Extended Data Tables 3 and 4). To understand theevolutionary origin of these two novel SL-CoV strains, we conductedrecombination analysis with the Recombination Detection Program4.0 package
19
using available genome sequences of bat SL-CoV strains(Rf1, Rp3, Rs672, Rm1, HKU3 and BM48-31) and human and civetrepresentative SARS-CoV strains (BJ01, SZ3, Tor2 and GZ02). Threebreakpointsweredetectedwithstrong
P
values(
,
10
2
20
)andsupportedbysimilarityplotandbootscananalysis(ExtendedDataFig.5a,b).Break-points were located at nucleotides 20,827, 26,553 and 28,685 in theRs3367 (and RsSHC014) genome,and generated recombination frag-mentscoveringnucleotides20,827–26,533(5,727nucleotides)(inclu-ding partial open reading frame (ORF) 1b, full-length S, ORF3, E andpartial M gene) and nucleotides 26,534–28,685 (2,133 nucleotides)(including partial ORF M, full-length ORF6, ORF7, ORF8 and partialNgene).PhylogeneticanalysisusingthemajorandminorparentalregionssuggestedthatRs3367,orRsSHC014,isthedescendentofarecombinationoflineagesthatultimatelyleadtoSARS-CoVandSL-CoVRs672(Fig.1b).The most notable sequence differences between these two new SL-CoVsandpreviouslyidentifiedSL-CoVsisintheRBDregionsoftheirSproteins.First,theyhavehigheraminoacidsequenceidentitytoSARS-CoV (85% and 96% for RsSHC014 and Rs3367, respectively). Second,there are no deletions and they have perfect sequence alignment withthe SARS-CoV RBD region (Extended Data Figs 1 and 2). Structural
*
These authors contributed equally to this work.
1
CenterforEmergingInfectiousDiseases,StateKeyLaboratoryofVirology,WuhanInstituteofVirologyoftheChineseAcademyofSciences,Wuhan430071,China.
2
EcoHealthAlliance,NewYork,NewYork10001, USA.
3
One Health Institute, School of Veterinary Medicine, University of California, Davis, California 95616, USA.
4
CSIRO Australian Animal Health Laboratory, Geelong, Victoria 3220, Australia.
5
College of Life Sciences, East China Normal University, Shanghai 200062, China.
6
Emerging Infectious Diseases Program, Duke-NUS Graduate Medical School, Singapore 169857.
2 8 N O V E M B E R 2 0 1 3 | V O L 5 0 3 | N AT U R E | 5 3 5
Macmillan Publishers Limited. All rights reserved
©201
3
and mutagenesis studies have previously identified five key residues(aminoacids442,472,479,487and491)intheRBDoftheSARS-CoVS protein that have a pivotal role in receptor binding
20,21
. Although allfive residues in the RsSHC014 S protein were found to be differentfrom those of SARS-CoV, two of the five residues in the Rs3367 RBDwere conserved (Fig. 1 and Extended Data Fig. 1).Despite the rapid accumulation of bat CoV sequences in the lastdecade, there has been no report of successful virus isolation
6,22,23
. Weattempted isolation from SL-CoV PCR-positive samples. Using anoptimized protocol and Vero E6 cells, we obtained one isolate whichcausedcytopathiceffectduringthesecondblindpassage.Purifiedvirionsdisplayed typical coronavirus morphology under electron microscopy (Fig. 2). Sequence analysis using a sequence-independent amplifica-tion method
14
to avoid PCR-introduced contamination indicated thattheisolatewasalmostidenticaltoRs3367,with99.9%nucleotidegenomesequence identity and 100% amino acid sequence identity for the S1region. The new isolate was named SL-CoV-WIV1.TodeterminewhetherWIV1canuseACE2asacellularentryreceptor,we conducted virus infectivity studies using HeLa cells expressing ornot expressing ACE2 from humans, civets or Chinese horseshoe bats.WefoundthatWIV1isabletouseACE2ofdifferentoriginsasanentry receptorandreplicatedefficientlyintheACE2-expressingcells(Fig.3).This is, to ourknowledge, the first identification of a wild-type batSL-CoV capable of using ACE2 as an entry receptor.Toassessitscross-speciestransmissionpotential,weconductedinfec-tivityassaysincelllinesfromarangeofspecies.Ourresults(Fig.4andExtended Data Table 5) indicate that bat SL-CoV-WIV1 can grow inhuman alveolar basal epithelial (A549), pig kidney 15 (PK-15) and
Rhinolophus sinicus
kidney (RSKT) cell lines, butnot in human cervix (HeLa),Syriangoldenhamsterkidney(BHK21),
Myotisdavidii
kidney (BK),
Myotis chinensis
kidney (MCKT),
Rousettus leschenaulti
kidney (RLK) or
Pteropus alecto
kidney (PaKi) cell lines. Real-time RT–PCR indicated that WIV1 replicated much less efficiently in A549, PK-15and RSKT cells than in Vero E6 cells (Fig. 4).
0.020.01
Tor2BJ01SZ3GZ02Rf1HKU3Rp3Rs627Rm1
3367 SHC014
SARS CoVSL-CoVRf1Rp3Rm1HKU3BJ01GZ02SZ3Rs672Tor2
SHC014 3367
442472479487491YLNTYYLNTYYLNTYYLKSYSFNNYWPRAHS—SVYS—SIYS—SVYRLASF
0.2
Human SARS CoV Tor2Human SARS CoV BJ01Human SARS CoV GZ02Civet SARS CoV SZ3
Bat SL-CoV Rs4087-1 Bat SL-CoV Rs4110 Bat SL-CoV Rs4090 Bat SL-CoV Rs4079 Bat SL-CoV Rs3367 Bat SL-CoV Rs4105 Bat SL-CoV RsSHC014 Bat SL-CoV Rs4084 Bat SL-CoV Rs3267-1 Bat SL-CoV Rs3369
Bat SL-CoV Rf1Bat SL-CoV HKU3-1Bat SL-CoV Rm1Bat SL-CoV Rs672Bat SL-CoV Rp3
Bat SL-CoV Rs3262-1 Bat SL-CoV Rs4092 Bat SL-CoV Rs4075
Bat SL-CoV Rs3262-2Bat SL-CoV Rs4085Bat SL-CoV Rs3267-2
Bat SL-CoV Rs4108 Bat SL-CoV Rs4081 Bat SL-CoV Rs4096 Bat SL-CoV Rs4087-2 Bat SL-CoV Rs4097 Bat SL-CoV Rs4080
Bat SARS-related CoV BM48-31Bat CoV HKU9-1
ab
948580926499999890100100521001001001001008610010010010071869551976898
Figure 1
|
Phylogenetic tree based on amino acid sequences of the S RBDregion and the two parental regions of bat SL-CoV Rs3367 or RsSHC014.a
, SARS-CoV S protein amino acid residues 310–520 were aligned withhomologousregions of bat SL-CoVs using the ClustalWsoftware.A maximum-likelihood phylogenetic tree was constructed using a Poisson model withbootstrapvaluesdeterminedby1,000replicatesintheMEGA5softwarepackage.TheRBDsequencesidentifiedinthisstudyareinboldandnamedbythesamplenumbers. The key amino acid residues involved in interacting with the humanACE2moleculeareindicatedontherightofthetree.SARS-CoVGZ02,BJ01andTor2wereisolatedfrompatientsintheearly,middleandlatephase,respectively,of the SARS outbreak in 2003. SARS-CoV SZ3 was identified from
Pagumalarvata
in2003collectedinGuangdong,China.SL-CoVRp3,Rs672andHKU3-1were identified from
R. sinicus
collected in China (respectively: Guangxi, 2004;Guizhou, 2006; Hong Kong, 2005). Rf1 and Rm1 were identified from
R. ferrumequinum
and
R. macrotis
, respectively, collected in Hubei, China, in2004. Bat SARS-related CoV BM48-31 was identified from
R. blasii
collected inBulgaria in 2008. Bat CoV HKU9-1 was identified from
Rousettus leschenaultii
collected in Guangdong, China in 2005/2006 and used as an outgroup. Allsequencesinboldanditalicswereidentifiedinthecurrentstudy.Filledtriangles,circles and diamonds indicate samples with co-infection by two differentSL-CoVs.‘–’indicatestheaminoaciddeletion.
b
,Phylogeneticoriginsofthetwoparental regions of Rs3367 or RsSHC014. Maximum likelihood phylogenetictrees were constructed from alignments of two fragments covering nucleotides20,827–26,533 (5,727 nucleotides) and 26,534–28,685(2,133 nucleotides) oftheRs3367 genome, respectively. For display purposes, the trees were midpointrooted. The taxa were annotated according to strain names: SARS-CoV, SARScoronavirus;SARS-likeCoV,batSARS-likecoronavirus.ThetwonovelSL-CoVs,Rs3367 and RsSHC014, are in bold and italics.
RESEARCH LETTER
5 3 6 | N AT U R E | V O L 5 0 3 | 2 8 N O V E M B E R 2 0 1 3
Macmillan Publishers Limited. All rights reserved
©201
3
Toassessthecross-neutralizationactivityofhumanSARS-CoVseraagainst WIV1, we conducted serum-neutralization assays using nineconvalescent sera from SARS patients collected in 2003. The resultsshowedthatsevenofthesewereabletocompletelyneutralize100tissueculture infectious dose 50 (TCID
50
) WIV1 at dilutions of 1:10 to 1:40,furtherconfirmingthecloserelationshipbetweenWIV1andSARS-CoV.Our findings have important implications for public health. First,theyprovidetheclearestevidenceyetthatSARS-CoVoriginatedinbats.Ourpreviousworkprovidedphylogeneticevidenceofthis
5
,butthelack of an isolate or evidence that bat SL-CoVs can naturally infect humancells, until now, had cast doubt on this hypothesis. Second, the lack of capacity of SL-CoVs to use of ACE2 receptors has previously beenconsideredasthekeybarrierfortheirdirectspilloverintohumans,suppor-ting the suggestion that civets were intermediate hosts for SARS-CoVadaptationtohumantransmissionduringtheSARSoutbreak
24
.However,the ability of SL-CoV-WIV1 to use human ACE2 argues against thenecessity of this step for SL-CoV-WIV1 and suggests that direct bat-to-humaninfectionisaplausiblescenarioforsomebatSL-CoVs.Thishasimplicationsforpublichealthcontrolmeasuresinthefaceofpoten-tialspilloverofadiverseandgrowingpoolofrecentlydiscoveredSARS-like CoVs with a wide geographic distribution.Our findings suggest that the diversity of bat CoVs is substantially higherthanthatpreviouslyreported.Inthisstudywewereabletodemon-stratethecirculationofatleastsevendifferentstrainsofSL-CoVswithinasinglecolonyof
R. sinicus
duringa 12-monthperiod. Thehigh geneticdiversity of SL-CoVs within this colony was mirrored by high pheno-typic diversity in the differential use of ACE2 by different strains. ItwouldthereforenotbesurprisingiffurthersurveillancerevealsabroaddiversityofbatSL-CoVsthatareabletouseACE2,someofwhichmay have even closer homology to SARS-CoV than SL-CoV-WIV1. Ourresults—in addition to the recent demonstration of MERS-CoV in aSaudi Arabian bat
25
, and of bat CoVs closely related to MERS-CoV inChina,Africa,EuropeandNorthAmerica
3,26,27
—suggestthatbatcoro-naviruses remain a substantial global threat to public health.Finally,thisstudy demonstratesthe publichealthimportance of path-ogendiscoveryprogramstargetingwildlifethataimtoidentifythe‘knownunknowns’—previously unknown viral strains closely related to knownpathogens.Theseprograms,focusedonspecifichigh-riskwildlifegroupsand hotspots of disease emergence, may be acritical partof future globalstrategies to predict, prepare for, and prevent pandemic emergence
28
.
HeLa-hACE2HeLa-cACE2HeLa-bACE2HeLaDAPIFITCCy3Merged
0 12 24 48 0 12 24 48 1
×
10
3
1
×
10
4
0 12 24 48 0 12 24 48
T C I D
5 0
m l
– 1
1
×
10
3
1
×
10
4
1
×
10
5
1
×
10
6
T C I D
5 0
m l
– 1
1
×
10
3
1
×
10
4
1
×
10
5
1
×
10
6
T C I D
5 0
m l
– 1
1
×
10
3
1
×
10
4
1
×
10
5
1
×
10
6
1
×
10
7
T C I D
5 0
m l
– 1
Time afterinfection (h)128 µm
Figure 3
|
Analysis of receptor usage of SL-CoV-WIV1 determined by immunofluorescence assay and real-time PCR.
Determination of virusinfectivity in HeLa cells with and without the expression of ACE2. b, bat;c, civet; h, human. ACE2 expression was detected with goat anti-humanACE2antibody followed by fluorescein isothiocyanate (FITC)-conjugated donkey anti-goat IgG. Virus replication was detected with rabbit antibody against theSL-CoV Rp3 nucleocapsid protein followed by cyanine 3 (Cy3)-conjugatedmouse anti-rabbit IgG. Nuclei were stained with DAPI (4
9
,6-diamidino-2-phenylindole). The columns (from left to right) show staining of nuclei (blue),ACE2 expression (green), virus replication (red), merged triple-stainedimages and real-time PCR results, respectively. (
n
5
3); error bars representstandard deviation.
200 nm
Figure 2
|
Electron micrograph of purified virions.
Virions from a 10-mlculture were collected, fixed and concentrated/purified by sucrose gradientcentrifugation. The pelleted viral particles were suspended in 100
m
l PBS,stained with 2% phosphotungstic acid (pH7.0) and examined directly using aTecnai transmission electron microscope (FEI) at 200kV.
LETTER RESEARCH
2 8 N O V E M B E R 2 0 1 3 | V O L 5 0 3 | N AT U R E | 5 3 7
Macmillan Publishers Limited. All rights reserved
©201
3