ContentslistsavailableatScienceDirect
Journal
of
Virological
Methods
jou rn al h om ep ag e :w w w . e l s e v i e r . c o m / l o c a t e / j v i r o m e t
Analysis
of
sensitivity
and
rapid
hybridization
of
a
multiplexed
Microbial
Detection
Microarray
James
B.
Thissen
a,
Kevin
McLoughlin
b,
Shea
Gardner
b,
Pauline
Gu
a,
Shalini
Mabery
a,
Tom
Slezak
b,
Crystal
Jaing
a,∗aPhysical&LifeSciencesDirectorate,LawrenceLivermoreNationalLaboratory,Livermore,CA94551,UnitedStates
bComputationsDirectorate,LawrenceLivermoreNationalLaboratory,Livermore,CA94551,UnitedStates
Articlehistory:
Received16August2013
Receivedinrevisedform14January2014
Accepted21January2014
Availableonline3March2014
Keywords: Microarray Pathogen Detection Virus Clinical Environmental
a
b
s
t
r
a
c
t
Microarrayshaveproventobeusefulinrapiddetectionofmanyvirusesandbacteria.Pathogendetection microarrayshavebeenusedtodiagnoseviralandbacterialinfectionsinclinicalsamplesandto evalu-atethesafetyofbiologicaldrugmaterials.AmultiplexedversionoftheLawrenceLivermoreMicrobial DetectionArray(LLMDA)wasdevelopedandevaluatedwithminimumdetectableconcentrationsforpure unamplifiedDNAviruses,alongwithmixturesofviralandbacterialDNAsubjectedtodifferentwhole genomeamplificationprotocols.Inadditiontheperformanceofthearraywastestedwhenhybridization timewasreducedfrom17hto1h.TheLLMDAwasabletodetectunamplifiedvacciniavirusDNAata concentrationof14fM,or100,000genomecopiesin12Lofsample.Withamplification,positive iden-tificationwasmadewithonly100genomecopiesofinputmaterial.Whentestedagainsthumanstool samplesfrompatientswithacutegastroenteritis,themicroarraydetectedcommongastroenteritisviral andbacterialinfectionssuchasrotavirusandE.coli.
Accuratedetectionwasfoundbutwitha4-folddropinsensitivityfora1hcomparedtoa17h hybridiza-tion.Thearraydetected2ng(equivalentconcentrationof15.6fM)oflabeledDNAfromaviruswith1h hybridizationwithoutanyamplification,andwasabletoidentifythecomponentsofamixtureofviruses andbacteriaatspeciesandinsomecasesstrainlevelresolution.Sensitivityimprovedbythreeorders ofmagnitudewithrandomwholegenomeamplificationpriortohybridization;forinstance,thearray detectedaDNAviruswithonly20fgor100genomecopiesasinput.Thismultiplexedmicroarrayis anefficienttooltoanalyzeclinicalandenvironmentalsamplesforthepresenceofmultipleviraland bacterialpathogensrapidly.
©2014TheAuthors.PublishedbyElsevierB.V.
1. Introduction
Rapid detection and characterization of bacterial and viral
pathogens is important for clinical microbiological diagnostics,
publichealth,drugandfoodsafety,environmentalmonitoringand
biodefense.Variousdetectiontechnologiesbasedonnucleicacid
signatureshaveemergedinthepastfewyears,includingTaqMan
PCRandLuminexbeadbasedsystems.Whilethesetechnologies
are abletoidentify selectedpathogens at thespecies or strain
level rapidly,theydo not have thecapability toprovide broad
∗ Correspondingauthorat:LawrenceLivermoreNationalLaboratory,P.O.Box808,
MailstopL-452,Livermore,CA94551,UnitedStates.Tel.:+19254246574.
E-mailaddress:jaing2@llnl.gov(C.Jaing).
spectrumdetectionaboutknownornovelorganisms.
Character-izationofknown,emerging,engineered,orunknownpathogens
requiresaplatformthathasthecapacitytoassesswholegenome
sequencecontentfromavarietyofpathogensveryrapidly.While
sequencing providesthe most in depth information to
charac-terizeamicrobial genome,thecosts,labor,and timeassociated
withlibrarypreparation,sequencing,bioinformaticanalysis,and
data storage may be prohibitive when analyzing many
iso-lates in a standard laboratory setting. To meet this need, a
pan-microbial microarray was developed, the Lawrence
Liver-moreMicrobialDetectionArray(LLMDA)(Gardneretal.,2010).
The originalversion ofthis array contains 388,000probes
cov-ering all currently sequenced viruses and bacteria, and can
detect any of these organisms within 24h. This microarray
uses long (50–65bp) oligonucleotide probes to enable
detec-tion of novel, divergent species with homology to sequenced
organisms.TheLLMDAhasbeenrecentlyusedtoidentifya
con-taminating pig virus from a rotavirus vaccine (Victoria et al.,
2010),avaccineusedworldwideininfantstopreventrotavirus
0166-0934©2014TheAuthors.PublishedbyElsevierB.V.
http://dx.doi.org/10.1016/j.jviromet.2014.01.024
Open access under CC BY-NC-ND license.
Table1
Probesummaryfor72KLLMDAarray.
Numberof probes Average numberof probesper sequence Targetsequences
48,893 20 Allviralfamiliesexcept Orthomyxoviridaeandfamily unclassifiedcompleteviralgenomes andsegments
7777 10 SegmentsintheOrthomyxoviridae family
2972 10 Familyunclassifiedviralgenomesand completesegments
7864 15 Bacterialgenomesandplasmids 3410 – RandomcontrolswithGC%andlength
distributionmatchedtotargetprobes
70,916 Total
infection. This array has also been used to detect viral infec-tions from various human clinical samples (Erlandsson et al., 2011).
Recent development of multiplexed microarrays makes
screeningoflargenumbersofsamplesmorecost-effectivewith
higher throughput(Palka-Santiniet al., 2009; Suo et al., 2010;
Bierbaumet al., 2012).These multiplexed microarrays are low
tomediumdensityandarefocusedonarelativelysmallnumber
ofpathogens.A modifiedversion oftheLLMDAwasdeveloped,
using the 4×72K Roche NimbleGen microarray format that
couldserveasabroadspectrumandfastdetectiontechnologyto
analyzebacterialandviralpathogens.Thisarraycontainsatotalof
70,916probes,designedtodetect2200viralspecies(38,000target
sequences) and 900 bacterial species (3500 target sequences).
Atargetwasdefined asaseparately accessioneddraftgenome,
finishedchromosome,plasmidorviralsegmentsequenced.
Aseriesofexperimentswiththe4×72Karraywasperformedto
estimatetheminimumdetectableconcentrationsofpureviral
tem-platesandmixturesofviralandbacterialtemplates,bothwithand
withoutamplification,usingvariousrandomamplification
tech-niques.Thesensitivityandspecificityofthemicroarraywasalso
exploredwhenitwashybridizedformuchshortertimesthanthe
standard17h.Inadditionthearraywastestedusingknownand
unknownclinicalsamplestoevaluateitsutilityindetecting
infec-tiousagentsfromhumansamples.
2. Materialsandmethods
2.1. Probedesignforthe4×72KLLMDAarray
Probedesignforthe388KversionoftheLawrenceLivermore
MicrobialDetectionArray(LLMDA)wasdescribedinGardneretal.
(Gardner et al., 2010). For the4×72K version, theinitial pool
ofcandidateprobeswasdown-selectedtohaveatotalof70,916
probes,asindicated inTable1,ratherthan the388,000probes
selectedforthepreviousversion.Eachviralgenomeorsegment
targetwasrepresentedby10–20probesonthe72Karray,in
con-trasttothe50 probespertargetinthe388Kversion.Thesame
processwasusedtodown-selectfromthecandidatepoolof
oli-gosaswasdescribedin(Gardneretal.,2010).Thesamebacterial
probeswereusedasonthe388Kversion,with15probesper
tar-get.Thearrayswerecustomorderedusingthe4-plex72Kformat
fromRocheNimbleGen.
2.2. Nucleicacidextractionfrombacterialandviralsamples
Humanadenovirus type 7 strainGomen (Adenoviridae) and
respiratorysyncytialvirus(RSV)werepurchasedfromtheNational
VeterinaryServicesLaboratory(Ames,IA)andgrownatLawrence
LivermoreNationalLaboratory(LLNL).Forpurificationof
adenovi-rusDNA,asolutionof10mLvirusculture,1mL5.5%TritonX-100,
and460L0.5MEDTAwasaddedtoa50mLtubeandvortexed
vig-orously.Followingvortexing,1.28mL10%SDSand540L10U/mL
proteinaseKwasaddedtothemixtureandincubatedat55◦Cfor1h
withmixingevery10min.Afterincubation,420L5MNaCland
13.7mL25:24:1phenol/chloroform/isoamylalcoholwasaddedto
thetubeandinverteduntilthemixturewascompletely
homoge-nous.Oncehomogenous,thesolutionwascentrifugedat10,000×g
for10minandtheupperaqueouslayerwastransferredtoanew
tube.Atotalof27.4mLof100%ethanolwasaddedtothe
aque-ouslayerand incubatedat−20◦C for1h. Followingincubation,
thesolutionwascentrifugedatmaximumspeedfor10minand
thesupernatantwasdiscarded.Thepelletwaswashedoncewitha
solutionof70%ethanoland150mMNaCl.Theliquidwasremoved
andpelletwasallowedtoair drybeforebeingre-suspendedin
nucleasefreewater.
ForextractionofRNAfromRSVvirus,asolutionof2mLofvirus
cultureand 6mLTRIzol LSReagentfromLifeTechnologies was
mixedvigorouslyandincubated15minatroomtemperature.
Fol-lowingincubation,1.6mLchloroformwasaddedtothecapofthe
tube,shakenvigorouslyfor15s,andincubatedatroom
tempera-turefor15min.Thesolutionwascentrifugedat4◦Cfor15minat
3000×gandtheaqueouslayerwastransferredtoanewtube.70%
isopropylalcoholwasaddedtotheaqueouslayer,mixedby
invert-ingseveraltimes,andincubatedfor10minatroomtemperature.
Thesolutionwascentrifugedatmaximumspeedfor10minat4◦C
andtheliquidwascarefullypouredoff.Theremainingpelletwas
washedwith70%icecoldethanol.Theethanolwaspouredoffand
thepelletwasairdriedbeforebeingre-suspendedinRNasefree
water.
EbolaZaire95andRiftValleyfever(RVF)virusstrainZH501
RNAwereobtainedfromNationalBioforensicAnalysisand
Coun-termeasuresCenter(NBACC).Copynumberquantitatedvaccinia
ListerDNAwaspurchasedthroughAdvancedBiotechnologiesInc.
(Columbia,MD).BacillusanthracisAmesDNAwasobtainedfrom
internalsourcesatLawrenceLivermoreNationalLaboratory(LLNL)
andnucleicacidswereextractedusingtheEpicentreDNA
extrac-tionkitaccordingtomanufacturer’sprotocols.
Extractednucleicacidsamplesfromfourpediatricstoolsamples
fromchildrenwithacutegastroenteritisandenrolledintheNew
VaccineSurveillanceNetwork(NVSN)wereprovidedbyDr.Mike
BowenfromtheCentersforDiseaseControlandPrevention(CDC).
2.3. Nucleicacidamplification
Thenucleicacidsampleswereamplifiedusingfourdifferent
protocols: a protocol developed by Wanget al., 2003, referred
tointhisstudyasRandomPCR1;thesameprotocolwith
mod-ified primers (Wong et al., 2007), referred to as Random PCR
2;the TransPlex® Whole TranscriptomeAmplification Kit (Cat.
WTA1)fromSigmaAldrich(St.Louis,MO) (Iscoveetal.,2002);
andQuantiTectWholeTranscriptomeKit(Cat.207043)from
Qia-gen(Valencia,CA)(Berthetetal.,2008;Erlandssonetal.,2011).
OnengofEbolaZaire95(9.34×107genomecopies)RNA,1ngof
vacciniaLister(4.76×106genomecopies)DNA,1ngofB.anthracis
Ames(1.77×105genomecopies)DNAand1ngofRVF(1.48×108
genomecopies)RNAweremixedandthemixturewasamplified
usingthefourprotocolsrespectively.
2.4. Samplepreparationforlimitofdetectiontesting
Vaccinia Lister DNA was 10-fold serially diluted from
100,000copiesto100copiesforlimitofdetectiontestingwithout
also10-foldseriallydilutedfrom10,000copies to10copiesand
amplifiedusingtheRandomPCR2protocol,andtheentire
ampli-fiedproductwashybridizedtothearray.Copynumberquantitation
wasperformedbytheAdvancedBiotechnologiesInc.(Columbia,
MD)byrealtimePCRusingtheRocheLightCycler.
RSVwastiteredatLLNLandtissuecultureinfectiousdose50
(TCID50)wasdetermined.TheviralRNAwas10-foldseriallydiluted
from 1,000,000TCID50 to 1000TCID50 and reverse transcribed
intocDNA.Double-strandedcDNAwassynthesizedaccordingto
theInvitrogenSuperScript® Double-StrandedcDNASynthesisKit
instructions.RSVRNAwasalso10foldseriallydilutedfrom100
TCID50to0.1TCID50andamplifiedusingRandomPCR2
amplifi-cation.Theentireamplifiedproductwaslabeledandhybridizedto
the4-plexarrays.
2.5. Samplelabelingandmicroarrayhybridization
The72K4-plexDNAmicroarraysweresynthesizedbyRoche
NimbleGen.Fourdifferentsamplescanberunoneachofthe
subar-rayssimultaneouslytosavetimeandcost.Amplifiedorunamplified
DNAorcDNAsampleswerefluorescentlylabeledusingtheRoche
NimbleGenOne-ColorDNALabelingKitaccordingtothe
recom-mendedprotocols.Thelabeledsamplewaspurifiedafterlabeling,
andhybridizedusingtheNimbleGenHybridizationKittothe4-plex
72Karrayaccordingtomanufacturer’sinstructions.
For limit of detection of adenovirusGomen strain DNA,the
concentrationoffluorescentlylabeledDNAwasmeasuredwitha
Qubitfluorometerandthenumberofnanograms/genome
equiva-lentwasdeterminedbasedonGenBankchromosomalandplasmid
genome sizes. The inputquantity to the labelingreaction was
250ng(6.52×109copies).Followinglabeling,approximately16g
(4.16×1011copies)ofCy3-labeledDNAwasobtained.Thelabeled
samplewasseriallydiluted2foldfrom2g(5.22×1010copies)to
0.5ng(1.3×107copies)andhybridizedtothearrays.
Themicroarrayswereallowedtohybridizefor 17hor1hat
42◦C, and washed using the NimbleGen Wash Buffer Kit (Cat.
05584507001) as described (Gardner et al., 2010).Microarrays
were scanned on an Axon GenePix 4000B 5M scanner from
MolecularDevices(Sunnyvale,CA).
2.6. Microarraydataanalysis
MicroarraydatawereanalyzedusingtheCompositeLikelihood
Maximizationmethod(CLiMax)asdescribed(Gardneretal.,2010).
Thetargetdatabaseforthe4-plex72Karraywasbuiltatalater
datethanthatusedforthe388KversionoftheLLMDA;inallother
respects,theanalysiswasthesameas reported(Gardneretal.,
2010).Probeswithsignalintensityabovethe99thpercentileof
therandomcontrolintensitieswereconsideredpositive,forthe
purposeofdataanalysis.
3. Results
3.1. Comparisonofdifferentprotocolstoamplifyamixtureof
pathogens
Fourdifferentrandomamplificationprotocolswereperformed
onamixtureofbacterialandviralnucleicacids,containingDNA
orRNAfromB.anthracisAmes,RiftValleyfevervirusstrainZH501,
vacciniavirusstrainLister,andEbolavirusstrainZaire95,to
evalu-atetheefficienciesofthedifferentprotocols.Allfouramplification
productswerecorrectlydetectedbymicroarraytothespecieslevel
(Table2).TheloglikelihoodisestimatedfromtheBLAST
similar-ityscoresoftheprobestoeachofthepossibletargetsequences,
togetherwiththeprobesequencecomplexityandothercovariates
derivedfromtheBLASTresults.EbolavirusZaire95wasdetected
Table2
Microarrayanalysisofamixtureofviralandbacterialtemplatesafterdifferent
randomamplificationprotocols.
Amplificationmethod Detectedvirusesandbacteria Log-oddsscore RandomPCR1 Bacillusanthracis 474.5
Cowpoxvirus 216.5 Variolavirus 15.0
RVFvirus 126.2
EbolaZaire95 111.1 RandomPCR2 Bacillusanthracis 473.3 Vacciniavirus 210.3 Variolavirus 13.6
RVFvirus 124.9
EbolaZaire95 110.0 SigmaTransPlex Bacilluscereus 132.9 Vacciniavirus 223.1 Variolavirus 18.8
RVFvirus 136.6
EbolaZaire95 115.4 QiagenWTA Bacillusanthracis 504.1 Cowpoxvirus 229.8 Variolavirus 17.2
RVFvirus 137.2
EbolaZaire95 109.8
Resultsfromthemicroarraydatawereanalyzedusingthecompositelikelihood maximizationmethod.Theconditionallog-oddsscoresforthesamplesareshown. TheloglikelihoodisestimatedfromtheBLASTsimilarityscoresoftheprobestoeach ofthepossibletargetsequences,togetherwiththeprobesequencecomplexityand othercovariatesderivedfromtheBLASTresults.Adetailedlistofdetectedtarget sequencesisprovidedinSupplementaryFile1.
atthestrainlevel.Allthreesegments,S,LandMoftheRVFgenome weredetectedonarraysusingallfourmethods.RandomPCR1and 2andtheQiagenWTAkitdetectedtheB.anthracischromosome atthespecieslevel.ArrayanalysisoftheSigmaTransPlex prod-uctgaveahigherscoretoastrainofthecloselyrelatedB.cereus speciesratherthantoB.anthracis.Orthopoxviruseswerecorrectly detectedbyallfourarrays;theRandomPCR2andSigmaTransPlex arrayscorrectlyidentifiedvacciniavirusasthetopscoringspecies, whiletheRandomPCR1andQiagenWTAarraysidentifiedcowpox virusasthetophit.Allfourarraysassignedahigherexplanatory scoretoamixtureofvacciniaorcowpoxwithvariolavirusthan tovacciniaorcowpoxalone.Ofthefourprotocolstested,Random PCR2performedslightlybetterintermsofaccuracy.Inthisstudy, theRandomPCR2protocolwasusedtodeterminethedetection limitofamplifiedpathogensonthemicroarray.
3.2. Detectionsensitivityofvacciniavirus
VacciniavirusListerDNAquantitiesof100,1000,10,000and 100,000genomecopiesper12Lsamplevolumewereappliedto the4×72Karraytoanalyzeitslimitofdetection.Table3shows
thefractionofprobesspecifictovacciniavirusListerstrainthat
had intensitiesgreaterthanthe99thpercentileof thenegative
controlsatdifferentviralconcentrations.Twoindependent
biolog-icalreplicateexperimentswereperformed.Theaverageofthetwo
replicatesresultedinmorethan70%oftheprobesbeingabovethe
detectionthresholdat100,000copiesofviralDNA,ora
concentra-tionof14fM.
VacciniavirusListerDNAquantitiesof10,100,1000and10,000
genomecopieswereamplifiedusingtheRandomPCR2method,
and theamplification productswere hybridizedtothe4×72K
MDAarrayinordertoanalyzeitslimitofdetectionwhen
com-bined with PCR amplification. Two replicate experiments were
performed.ThearraywasabletodetectvacciniaListerstrainwith
only100genomecopiesasinput,with72and82% of
Table3
DetectionofunamplifiedandamplifiedVacciniaListerviralDNAonthe72KLLMDAarray.
Vacciniavirus Nonamplified Amplified
DNAcopynumberinreaction 100 1000 10,000 100,000 10 100 1000 10,000
%ofVacciniaListerstrainprobesdetectedin eachreplicate
0% 0% 0% 79% 5% 82% 100% 100%
0% 0% 20% 68% 14% 72% 100% 100%
ThesensitivityofthearrayatdifferentVacciniaListervirusDNAconcentrationswascalculatedusingthepercentageofprobesspecifictoVacciniaListerstrainvirusthatwere detectedonthearray.Tworeplicateexperimentswereperformedandthepercentagedetectedoutof39VacciniaListerstrainspecificprobesisreportedforeachreplicate.
3.3. DetectionsensitivityofRSV
RSVcDNAquantitiesof1000,10,000,100,000and1,000,000 TCID50 per 12L sample volume were applied to the 4×72K
array.Table4shows theresponse ofthearrayat differentRSV
concentrations.Tworeplicateexperimentswereperformed.The
arraywasabletodetectRSVcDNAat100,000TCID50,withan
aver-ageof90%ofRSV-specificprobesabovethedetectionthreshold.
RSVcDNAquantitiesof0.1,1,10and100TCID50wereamplified
byrandomPCRandappliedtothe4×72KMDAarray.Tworeplicate
experimentswereperformed.ThearraydetectedRSVcDNAfrom
theproductof1TCID50,withanaverageof75%ofprobesabovethe
detectionthreshold.
3.4. Detectionofadenovirusafter1hand17hhybridization
Afterlabelinghumanadenovirustype7GomenstrainDNA,
2-foldserialdilutionsofthelabeledDNAwereperformedfrom2g
to0.5ngandthedilutedDNAwashybridizedtothearraysfor17h.
TheDNAconcentrationsthusrangedfrom3.9fMto15.6pM.The
responseofthearraytotheadenovirusisshowninTable5.Atall
concentrationstested,90–100%oftheprobesspecifictothe
ade-novirusstraintestedwereabovethedetectionthreshold.CLiMax
analysisofthesearraysidentifiedthetype7Gomenstrainasthe
tophitatallconcentrationsexceptthesmallest(3.9fM),atwhich
adenovirustype3strainNHRC1276receivedthehighestscore.
To determine thespeed of detection using this microarray,
hybridizationsofadenovirustype7GomenstrainDNAwerealso
performedfor only1h. SeriallydilutedDNA quantitiesranging
from2g to0.5ng of DNA werehybridizedtothearray. Over
70%ofthespecificprobeswereabovethethresholdwith2ngof
DNA,correspondingto15.6fMconcentration.CLiMaxanalysis
pos-itivelyidentifiedtheadenovirustype7Gomenstrainwith8ngof
DNAafter1hhybridization,comparedto1ngofDNAwith17h
hybridization.
3.5. Detectionofvirusesandbacteriafromhumanfecalsamples
Four acute gastroenteritis samples received from the CDC’s
NVSNwereanalyzedonthe72K4-plexmicroarray.These
sam-pleswerepreviouslyanalyzedusingenzymeimmunoassays(EIA)
or real-time reverse transcription (RT)-PCR to determine the
presenceofalimitednumberofknownentericpathogens(Bowen,
unpublishedresults).Sample 4496was negativeby EIAor PCR
assays.Rotaviruswasdetectedinsamples4551,4912and4949
using EIA or PCR assays. Table 6 is a summary of the viruses
andbacteriadetectedonthemicroarray.Rotavirusfromsample
4551wasnotdetectedwhenalltargetsequenceswereusedfor
analysis,butitwasdetectedwhenviraltargetsequencedatabase
only wasused in theanalysis. For sample 4949, rotaviruswas
detectedinbothreplicateswhenviraltargetsequencedatabase
onlywasusedintheanalysis,howeveronlyoneoftworeplicates
detectedrotaviruswhenalltargetsequenceswereusedinthe
anal-ysis.
Themicroarrayresultsconfirmedtherotavirusfindingsfrom
thethreesamplesthatwerepreviouslytestedpositiveforrotavirus
attheCDC.Themicroarrayalsodetectedahumancalicivirusthat
wasnotpreviouslydetectedbyPCRorEIAattheCDC.Additionally,
variousbacterialpathogensincludingE.coliF11,Salmonella
enter-icaTyphi,Enterococcus faecium,Enterococcus faecalis,Bacteriodes
intestinalisweredetectedfromthefoursamples.Thesebacteria
oftencausenosocomialinfections,gastrointestinalproblemsor
uri-narytractinfections.
4. Discussion
Microarrays,alongwithPCRandDNAsequencingareeffective
methodsformicrobialdetectionanddiscoveryusingnucleicacid
samples.Whilemicroarraysarenotassensitiveorinexpensiveas
PCRassays,theycanqueryhundredsofthousandstoseveral
mil-lionregionsofDNAinparallel(dependingonplatformandformat),
comparedtoatmostafewtensofregionsinthelargestmultiplexed
PCRassays.PCRassaysaretoolimitedforbroad-spectrum
micro-bialanalysis.Severalgroupshaveappliedmicroarraytechnology
topathogen detection for clinical diagnostics, food safety
test-ing,environmentalmonitoringandbiodefense(Wangetal.,2002;
Palaciosetal.,2007;Wongetal.,2007;Palka-Santinietal.,2009; Jaingetal.,2011;McLoughlin,2011).Onemainlimitationforthe
wideruseofpathogendetectionmicroarraysisthecostoftheassay,
whichisatleastanorderofmagnitudehigherthanPCRassays.
Amultiplexedpathogendetectionmicroarraywasdevelopedand
wasamodifiedversionofthepreviouslyreportedLawrence
Liver-moreMicrobialDetectionArray(LLMDA)(Gardneretal.,2010)to
furtherimprovetheefficiency,cost-effectivenessandthroughout
ofpathogendetectionmicroarrays.
Inthisstudy,fourdifferentamplificationprotocolswere
ana-lyzed: two random PCR based protocols; the Qiagen whole
transcriptionamplificationprotocol,whichusesaphi29DNA
poly-merasebasedapproach;andtheSigmaTransPlexkit,whichalso
usesPCRbasedamplification.Whenfourdifferentprotocolswere
appliedtoa mixtureof fourviraland bacterialpathogens, RVF
andEbolaviruseswerecorrectlyidentifiedatspeciesresolution
byallprotocols,B.anthraciswascorrectlyidentifiedbytwo
proto-cols,Vacciniaviruswascorrectlyidentifiedbytwoprotocols.The
RandomPCR2amplificationprotocolcorrectlyidentifiedallfour
species.
Thedetectionlimitofthearraywasevaluatedbyusingserial
dilutionsofaDNAvirus,vacciniavirusLister,andanRNAvirus,
RSV,withvariousmethodsusedtoquantifytheamountsofviral
DNA or RNA. Without whole genome amplification,the
detec-tionlimit forRSV is 100,000TCID50.Thisissimilarin rangeto
othermolecularassayssuchastherealtimePCRassay,multiplex
ligation-dependentprobeamplification(MLPA)andadual
prim-ingoligonucleotidesystem(DPO)wherethelowestconcentration
detectedforRSV was105 TCID
50 (Bruijnesteijn vanCoppenraet
etal.,2010).ForvacciniavirusListerDNAwithoutrandom
ampli-fication,thearraydetected100,000genomecopiesorabout20pg
ofviralDNAinthe12Lsamplevolume,correspondingto14fM
Table4
DetectionofunamplifiedandamplifiedRSVcDNAonthe72KLLMDAarray.
RSV Nonamplified Amplified
TCID50valuesinreaction 1000 10,000 100,000 1,000,000 0.1 1 10 100
%ofRSVprobesdetectedineachreplicate 5% 9% 91% 100% 40% 73% 100% 100%
5% 11% 89% 100% 40% 77% 100% 100%
ThesensitivityofthearraytodetectRSVviruscDNAatdifferentRSVTCID50concentrationswascalculatedusingthepercentageofprobesspecifictoRSVvirusthatwere
detectedonthearray.Tworeplicateexperimentswereperformedandthepercentageof30RSVspecificprobesforeachreplicateisreported.
ThelimitofdetectionforvacciniavirusDNAwasfoundtobe 100copieswhenrandomwholegenomeamplificationwasused. Thisdetection limit is similar in range toPCR based technolo-gies. A studyusingreal time PCRassays toidentifyvariants of vacciniavirusdetectedthevirusat10fg oraround 50genome copies of vaccinia DNA (Trindade et al., 2008). The difference
in sensitivity between PCR and microarrays is partially due to
the random amplification, using random primers, instead of a
viralspecificPCRamplification.Palka-Santinietal.hasreported
that specificmultiplexed PCR amplificationprior tomicroarray
hybridizationcanincreasethedetectionlimitofpathogensbya
factorof100–1000(Palka-Santinietal.,2009).Itispossiblethat
thesensitivityofthemicroarraywillincreaseifa targeted
mul-tiplexed amplification is applied to the samples prior to array
hybridization.
The detection limit for the RSV virus is around 1 TCID50
whencoupledwithwholegenomeamplification.Thissensitivity
issimilartomultiplexedPCRbasedLuminexassays.Agroupthat
developedaLuminexbeadassaypaneltodetect20human
respi-ratoryvirusesreportedadetectionlimitof1TCID50forRSVtypeA
and10TCID50forRSVtypeB(Mahonyetal.,2007).
Previousevaluation oftheoriginal design ofthis microarray
hasdemonstratedthatthearraydetectedvirusesfromavariety
ofhumanclinicalsamplessuchasurine,feces,serum,skinlesion,
cerebralspinalfluid,trachealaspirate(Erlandssonetal.,2011).In
14 outof14 clinicalsamples,theexpectedviruswasdetected:
Herpessimplexvirus1,Herpessimplexvirus2,Human
papillo-mavirus16,Humanpapillomavirus6/16/53/61,BKpolyomavirus
(BKV),JCpolyomavirus,RotavirusA,Astrovirus,Sapovirus,Dengue
1,HepatitisCvirusinduplicate,enterovirusandRSV.Thisstudy
alsodemonstratedthatthemicroarraydetectedBKpolyomavirus
positiveurinesamplescontaining≥1000copies/mL(oran
equiva-lentof5viralcopiesinaPhi29-amplificationreaction)(Erlandsson
etal.,2011).Inanotherstudy,thearraywasusedtodetect
Kaposi-sarcomaassociatedherpesvirus, orhumanherpesvirus8 from
bladdercancersamples(Paradˇziketal.,2013).Thiswasthefirst
indicationoftheassociationbetweenherpesvirus8withbladder
cancer, whichfurtherdemonstrated thepotentialofthe
micro-bialdetectionarraytechnologytoidentifypathogensthatmight
belinkedtocancerandotherdiseases.
Inthisstudy,theperformanceandsensitivityofthe4-plex
ver-sionoftheLawrenceLivermoreMicrobialDetectionArray(LLMDA)
onthedetectionof viruseswasevaluated usinghumanclinical
samples.IncollaborationwiththeCDC,acutegastroenteritis
sam-plesfromtheNewVaccineSurveillanceNetworkwereappliedto
the4-plexmicroarray whichidentified rotavirus,norovirusand
severalenterobacteria(E.coliF11,SalmonellaentericaTyphi,
Entero-coccusfaecium,Enterococcusfaecalis,Bacteriodesintestinalis)from
Table5
DetectionofAdenovirusDNAafter17hror1hrhybridizationonthe72KLLMDAarray.
AdenovirusDNA 0.5ng 1ng 2ng 4ng 8ng 15.6ng 125ng 1000ng
%ofadenovirusGomenstrainprobesdetectedat1h 10% 30% 80% 90% 100% 100% 100% 100% %ofadenovirusGomenstrainprobesdetectedat17h 100% 100% 100% 100% 100% 100% 100% 100% 500pgto1gofadenovirusGomenDNAwasanalyzedandthe%ofprobesspecifictoadenoviruswascalculated.Notallconcentrationshigherthan15.6ngareshown.The percentageofprobesspecifictoadenovirusGomenstrain(67totalprobes)wasusedinthecalculation.
Table6
LLMDAanalysisofhumanfecalsamplesfromacutegastroenteritiscases.Thearraydatawasanalyzedagainstbothviralandbacterialtargetsequencedatabase.Additionally, arraywasalsoanalyzedagainstviraltargetsequencedatabaseonlytoconfirmthevirusesdetected.Allsampleswereperformedinduplicateandallvirusesorbacterialisted weredetectedintwooutoftwosamplesexceptwhennoted.AdetailedlistoftargetsequencesdetectedislistedinSupplementaryfile2.
SampleID LLMDAviralresults AvgLogodds LLMDAbacterialresults AvgLogodds 4496 HumancalicivirusHu/NLV/GII/MD145-12/1987/US 207.6 Salmonellaentericasubsp.enterica
serovarTyphistr.CT18
568.3 TTvirusstrainTTVCHN1,completegenome 170.0
4551 RotavirusAstrainS2genotype G2P[4]segment2a
7.5 EscherichiacoliF11 677.1
Salmonellaentericasubsp.enterica
serovarTyphistr.CT18
79.4
Enterococcusfaecium,Glued
fragmentsofsequence630811
69.3
Stenotrophomonasmaltophilia
K279a
58.7
4912 HumanrotavirusAisolate
rj8200/04NSP4gene
151.7 EscherichiacoliF11 807.7
EnterococcusfaecalisTX0104 215.2
4949 RotavirusAstrainCC598VP7geneb 79.2 Bacteroides
intestinalisDSM17393
262.8
aOnlyoneoutoftworeplicatesdetectedrotaviruswhenanalyzedagainstviralonlytargetdatabase.Rotaviruswasnotdetectedwhenanalyzedusingalltargetsequence
database.
bTwooftworeplicatesdetectedrotavirusAwhenanalyzedagainstviralonlytargetsequences.Onlyoneoftworeplicatesdetectedrotaviruswhenanalyzedagainstall
thesamples.TheNVSNbeganpopulation-basedactive
gastroen-teritissurveillanceamongU.S. children<3years ofagein2006
(Payneetal.,2008,2011).CurrentsurveillanceeffortsutilizePCR
andimmunoassaystodetectrotavirusandalimitednumberofviral
pathogens,butdonotprovideacomprehensiveanalysisofother
entericpathogenspresentinsamples.Thedatasuggestedthat
bac-terialandviralco-infectionsarecommoninacutegastroenteritis
patients.Thevirusconcentrationsinthesesamplesappeartobeat
alowlevelsincethehumancaliciviruswasnotdetectedbyPCRor
EIA.Rotavirussequencesfromtwosamplesweredetectableonly
whenthedatawasanalyzedagainstviralsequencesonly.Thisis
likelyduetothehigherlevelofentericbacteriaconcentrationsin
thefecalsamples.Moresensitiveviraldetectionfromhumanfecal
samplescouldpotentiallybeaccomplishedusingatargetedsample
amplificationprotocoltoselectivelyamplifyviralfamiliesof
inter-est,whichcouldprovideadvantagesoverwholegenomerandom
amplification.
Themultiplexedpathogendetectionarrayisausefultoolfor
rapidviralandbacterialpathogendetection.Thecostfora1-plex
microarrayisinthehundredsofdollarspersample,whilethecostof
a4-plexmicroarrayisonly25%ofthe1-plexarray.Thoughthecost
isprobablynotgoingtobeatlevelscomparabletoPCR,thisassay
canreplacehundredsofPCRreactionswithjustonereaction,
there-forethe4-plexmicroarrayincreasestimeandefficiencyinrapid
diagnosisofclinicalinfectionsandsurveillanceofenvironmental
pathogens.
5. Conclusion
Inanefforttoimprovecost-effectiveness,speedand
through-put,amultiplexedversionoftheLawrenceLivermoreMicrobial
DetectionArraywasdevelopedandsensitivitytestingusinga
num-berofviralagents and humanclinicalsampleswasperformed.
Theseresultshavedemonstratedthatthearrayissensitivetodetect
viralinfectionsandhavethepotentialforrapidscreeningof
bac-terialandviralinfectionsorenvironmentalsamples.Multiplexed
microarrayspresentanewopportunityforhigh-throughputand
cost-effectivescreeningofthousandsofmicrobialspecies.Aswith
anytechnologybasedonnucleicaciddetection,thecapabilities
ofmicroarraysarelimitedbythegenomesequenceinformation
availableatthetimeofdesign.Furtheradvancesinarray
technol-ogysuchasautomatedarraysampleloadingandimagescanning,
fasterhybridizationtimes,andlabel-freemethodstodetect
probe-targetbindingwillbroadentheapplicationsofmicroarrayseven
further.Ultimately,microarrayscouldbeadvancedinto
point-of-caredeviceswhichdeliverresultsinlessthananhour.
Acknowledgements
ThisworkperformedundertheauspicesoftheU.S.Department
ofEnergybyLawrenceLivermoreNationalLaboratoryunder
Con-tractDE-AC52-07NA27344.
AppendixA. Supplementarydata
Supplementarydataassociatedwiththisarticlecanbefound,
intheonlineversion,athttp://dx.doi.org/10.1016/j.jviromet.2014.
01.024.
References
Berthet,N.,Reinhardt,A.K.,Leclercq,I.,Ooyen,S.v.,Batéjat,C.,Dickinson,P.,
Stam-boliyska,R.,Old,I.G.,Kong,K.A.,Dacheux,L.,Bourhy,H.,Kennedy,G.C.,Korfhage,
C.,Cole,S.T.,Manuguerra,J.-C.,2008.Phi29polymerasebasedrandom
ampli-ficationofviralRNAasanalternativetorandomRT-PCR.BMCMol.Biol.9, 77.
Bierbaum,S.,Königsfeld,N.,Besazza,N.,Blessing,K.,Rücker,G.,Kontny,U.,Berner,
R.,Schumacher,M.,Forster,J.,Falcone,V.,Sand,C.,Essig,A.,Huzly,D.,Rohde,
G.,Neumann-Haefelin,D.,Panning,M.,2012.Performanceofanovelmicroarray
multiplexPCRforthedetectionof23respiratorypathogens(SYMP-ARIstudy). Eur.J.Clin.Microbiol.Infect.Dis.31,2851–2861.
BruijnesteijnvanCoppenraet,L.E.S.,Swanink,C.M.A.,vanZwet,A.A.,Nijhuis,R.H.T.,
Schirm,J.,Wallinga,J.A.,Ruijs,G.J.H.M.,2010.Comparisonoftwocommercial
molecularassaysforsimultaneousdetectionofrespiratoryvirusesinclinical samplesusingtwoautomaticelectrophoresisdetectionsystems.J.Virol. Meth-ods169,188–192.
Erlandsson,L.,Rosenstierne,M.W.,McLoughlin,K.,Jaing,C.,Fomsgaard,A.,2011.
Themicrobialdetection arraycombinedwith randomPhi29-amplification usedasadiagnostictoolforvirusdetectioninclinicalsamples.PLoSONE6, e22631.
Gardner, S., Jaing, C., McLoughlin, K., Slezak, T., 2010. A microbial
detec-tion array (MDA) for viral and bacterial detection. BMC Genomics 11, 668.
Iscove,N.N.,Barbara,M.,Gu,M.,Gibson,M.,Modi,C.,Winegarden,N.,2002.
Rep-resentationisfaithfullypreservedinglobalcDNAamplifiedexponentiallyfrom sub-picogramquantitiesofmRNA.Nat.Biotechnol.20,940–943.
Jaing,C.,Gardner,S.,McLoughlin,K.,Thissen,J.,Slezak,T.,2011.Detectionof
adventi-tiousvirusesfrombiologicalsusingabroadspectrumMicrobialDetectionArray. PDAJ.Pharm.Sci.Technol.65,668–674.
Mahony,J.,Chong, S.,Merante, F.,Luinstra,K.,Sinha,T., Petrich,A.,Lisle, C.,
Yaghougian,S.,Janeczko,R.,2007.Developmentofarespiratoryviruspanel
testforthedetectionoftwentyhumanrespiratoryvirusesusingmultiplexPCR andafluidmicrobead-basedassay.J.Clin.Microbiol.45,2965–2970.
McLoughlin,K.,2011.Microarraysforpathogendetectionandanalysis.BriefFunct.
GenomicProteomic,342–353.
Palacios,G.,Quan,P.-L.,Jabado,O.,Conlan,S.,Hirschberg,D.,Liu,Y.,2007.
Panmi-crobialoligonucleotidearrayfordiagnosisofinfectiousdiseases.Emerg.Infect. Dis.,13.
Palka-Santini,M.,Cleven,B.,Eichinger,L.,Kronke,M.,Krut,O.,2009.Largescale
mul-tiplexPCRimprovespathogendetectionbyDNAmicroarrays.BMCMicrobiol.9, 1.
Paradˇzik,M.,Buˇcevi ´c-Popovi ´c,V., ˇSitum,M.,Jaing,C.,Degoricija,M.,McLoughlin,
K.,Ismail,S.,Punda-Poli ´c,V.,Terzi ´c,J.,2013.AssociationofKaposi’s
sarcoma-associatedherpesvirus(KSHV)withbladdercancerinCroatianpatients.Tumor Biol.,1–6.
Payne,D.C., Staat,M.A., Edwards,K.M.,Szilagyi, P.G.,Gentsch,J.R., Stockman,
L.J.,Curns,A.T.,Griffin,M.,Weinberg,G.A.,Hall,C.B.,Fairbrother,G.,
Alexan-der,J.,Parashar,U.D.,2008.Active,population-basedsurveillanceforsevere
rotavirus gastroenteritis in children in the United States. Pediatrics 122, 1235–1243.
Payne,D.C.,Staat,M.A.,Edwards,K.M.,Szilagyi,P.G.,Weinberg,G.A.,Hall,C.B.,
Chap-pell,J.,Curns,A.T.,Wikswo,M.,Tate,J.E.,Lopman,B.A.,Parashar,U.D.,2011.
Directandindirecteffectsofrotavirusvaccinationuponchildhood hospitaliza-tionsin3UScounties,2006–2009.Clin.Infect.Dis..
Suo,B.,He,Y.,Paoli,G.,Gehring,A.,Tu,S.-I.,Shi,X.,2010.Developmentofan
oligonucleotide-basedmicroarraytodetectmultiplefoodbornepathogens.Mol CellProbes24,77–86.
Trindade,G.d.S.,Li,Y.,Olson,V.A.,Emerson,G.,Regnery,R.L.,Fonseca,F.G.d.,Kroon,
E.G.,Damon,I.,2008.Real-timePCRassaytoidentifyvariantsofVacciniavirus:
ImplicationsforthediagnosisofbovinevacciniainBrazil.J.Virol.Methods152, 63–71.
Victoria,J.G.,Wang,C.,Jones,M.S.,Jaing,C.,McLoughlin,K.,Gardner,S.,Delwart,
E.L.,2010.Viralnucleicacidsinlive-attenuatedvaccines:detectionofminority
variantsandanadventitiousvirus.J.Virol.84,6033–6040.
Wang,D.,Coscoy,L.,Zylberberg,M.,Avila,P.C.,Boushey,H.A.,Ganem,D.,DeRisi,J.L.,
2002.Microarray-baseddetectionandgenotypingofviralpathogens.Proc.Natl.
Acad.Sci.99,15687–15692.
Wang,D.,Urisman,A.,Liu,Y.-T.,Springer,M.,Ksiazek,T.G.,Erdman,D.D.,Mardis,E.R.,
Hickenbotham,M.,Magrini,V.,Eldred,J.,Latreille,J.P.,Wilson,R.K.,Ganem,D.,
DeRisi,J.L.,2003.ViraldiscoveryandsequencerecoveryusingDNAmicroarrays.
PLoSBiol.1(2),e2,http://dx.doi.org/10.1371/journal.pbio.0000002.
Wong,C.,Heng,C.,WanYee,L.,Soh,S.,Kartasasmita,C.,Simoes,E.,Hibberd,M.,
Sung,W.-K.,Miller,L.,2007.Optimizationandclinicalvalidationofapathogen