Direct identification of bovine mastitis pathogens by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry in pre-incubated milk

h tt p : / / w w w . b j m i c r o b i o l . c o m . b r /

Veterinary

Microbiology

Direct

identification

of

bovine

mastitis

pathogens

by

matrix-assisted

laser

desorption/ionization

time-of-flight

mass

spectrometry

in

pre-incubated

milk

Juliana

R.

Barreiro

_,

_Juliano

_L.

_Gonc¸alves

_,

_Rafaella

_Grenfell

_,

_Renata

_F.

_Leite

_,

Luiz

Juliano

_,

_Marcos

_V.

_Santos

a_{UniversidadedeSãoPaulo(USP),FaculdadedeMedicinaVeterináriaeZootecnia,DepartamentodeNutric¸ãoeProduc¸ãoanimal,}

Pirassununga,SP,Brazil

b_{UniversidadeFederaldeSãoPaulo,DepartamentodeBiofísica,SãoPaulo,SP,Brazil}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received9August2017 Accepted16April2018 Availableonline16August2018 AssociateEditor:MilianeSouza

Keywords: Milk Bacteria Mastitis Massspectrometry

a

b

s

t

r

a

c

t

The presentstudy aimedto comparetwoMALDI-TOFidentification methods[(a)direct sampleidentificationafterpre-incubation;or(b)useofbacteriaisolatedonpre-culture)] tostandard,traditionalbenchmicrobiology.Atotalof120quartermilksamplesfrom40 Holsteinlactatingcowswerescreenedbasedonculture-positiveresultsobtainedby microbi-ologicalculture(referencemethod)withthefollowingnumbersofquarterspositivepercow: 4cowswith1,8cowswith2,12cowswith3and16cowswith4infectedquarterspercow. Fordirectidentificationmethod,quartermilksamples(n=120)wereskimmedby centrifu-gation(10,000×g/10min)andpre-incubatedat37◦Cfor12h.Afterpre-incubation,quarter milksamplesweresubmittedtototalbacterialcountbyflowcytometryandfora prepara-tionprotocolforbacterialribosomalproteinextractionfollowedbyMALDI-TOFMSanalysis. ThedirectMALDI-TOFMSidentificationmethodcomparedtomicrobiologicalculture cor-rectlyidentifiedisolatesofcoagulase-negativeStaphylococci(27.2%),Streptococcusagalactiae

(21.8%),Staphylococcusaureus(14.2%),andStreptococcusuberis(5.2%).Thepre-incubation pro-tocolofmilksamples,associatedtothedirectidentificationmethodbyMALDI-TOFMS,did notincreasetheidentificationatspecieslevel(score>2.0)ofpathogenscausingsubclinical mastitisincomparisontothemethodwithoutpreviousincubation.

©2018SociedadeBrasileiradeMicrobiologia.PublishedbyElsevierEditoraLtda.Thisis anopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/ licenses/by-nc-nd/4.0/).

∗ _{Correspondingauthorat:DuquedeCaxiasNorteAvenue225,Pirassununga,SP13635-900,Brazil.Tel.:+551935454240.}

E-mail:[email protected](M.V.Santos).

https://doi.org/10.1016/j.bjm.2018.04.012

1517-8382/©2018SociedadeBrasileiradeMicrobiologia.PublishedbyElsevierEditoraLtda.ThisisanopenaccessarticleundertheCC BY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

Conventionalmicrobiologicalcultureofmilksamplesisbased on biochemical tests for identification of microorganisms causing subclinical mastitis. Different bacterial character-istics are evaluated for identification of a single species, suchasgrowthconditions,colonymorphology,growth char-acteristics in selective medium, carbohydrate fermenting capacity,metabolicandantigeniccharacteristics,and antibi-oticsusceptibility.1_{Theproceduresaretimeconsuming,andit}

maytake2–7daysforthecompletediagnosisofthecausative pathogenoftheintramammaryinfection(IMI).

The mass spectrometry (MS) technique using matrix assisted laser desorption/ionization source (MALDI) and a time-of-flighttypemassanalyzer(TOF)canbeusedforrapid identificationofbacteriaandyeastfromcoloniespreviously culturedonsolidmedium.Suchmethodologyprovidesarapid identification ofmastitis-causing bacteriaby means ofthe extractionofribosomalproteinsfrombacterialcolonies cul-tured on blood agar2 _{or using a direct transfer protocol.}3

However,theMALDI-TOFMSmethodhasbeenusedfordirect identification of bacteria from human blood samples as a rapiddiagnostic tool inhospital laboratories, sinceit does notdependonpreviousbacterialisolationusing microbiolog-icalculture.4,5 _{ThedirectclassificationofGram-positiveand}

Gram-negativebacteriaatthegenuslevelshoweda100% pos-itivepredictivevalue(PPV)whenspectraconsistingofjoint ribosomalproteins(“fingerprints”)wereacquiredfromblood samples.4,6

Urinary tract pathogens have been also successfully directlyidentifiedbyMALDI-TOFMSusingurinesampleswith bacterialcounts>105_{cfu/mL.AccordingtoFerreiraetal.,}7_the

directidentificationofmicroorganismsbyMALDI-TOFMSin urinesamplesshowed91.8%agreementatthespecieslevel and 92.7% at the genus level, when compared to microbi-ologicalculture identification.Theseresultssuggestedthat MALDI-TOFMSallowedtheidentification ofGram-negative bacteria directly from urine samples in a short period of timewhensamplescontainedelevatedbacterialcounts.7_The

MALDI-TOFMSmethodwasalsoappliedtoclinicalsamples ofcerebrospinalfluidandcorrectlyclassifiedthepathogens when bacterial counts in samples were between 104 _and

106_cfu/mL.8

Thescoresusedforthedirect(nonculturebased) identifica-tionMALDI-TOFmethod(direct-MALDI-TOF)inbloodsamples werelowerthanthoseofisolatesobtainedfrombacterial cul-turesobtainedfromhemoculture,indicatingthatthesescores can be influenced by the bacterial count. Previous studies utilizedserialdilutionsandobtainedexcellentidentification spectrawhen thebacterial countinthe blood samplewas ≥106_{cfu/mL,thoughthedirectclassificationof}

microorgan-ismsatthespecieslevel(75.8%)byMALDI-TOFMSwasdone consideringscores≥1.7.9

Thetotalbacterialcountisacriticalfactorforthedirect identification of pathogens that cause mastitis by MALDI-TOFMS.3_{AccordingtoMoussaouietal.,}10_{thepre-incubation}

of blood samples enabled higher precision in the direct identificationofbacteriausingtheMALDI-TOFMSmethod. Recently,ourresearchgrouphasevaluatedthedetectionlimit

ofMALDI-TOFMSfordirectidentification,withoutprevious microbiological culture, of bovine mastitis-causing bacte-ria from milk samples.3 _{Therefore, we suggested that the}

non-culture-basedprotocolcouldbeappliedindiagnostic lab-oratoriesbysubjectingallmilksamplestodirectMALDI-TOF, andthosewithoutapositiveidentificationcouldbe submit-tedtopre-incubationprotocol,beingidentifiedbyMALDI-TOF MScombinedwithstandardbacteriology.However,theeffect ofpre-incubationofquartermilksamplesfromcowsaffected withsubclinicalmastitishasnotbeenevaluatedusingdirect identificationofbacteriabytheMALDI-TOFmethod.Thus,the objective ofthepresentstudy wastocomparetwo MALDI-TOF identification methods[(a)direct sampleidentification afterpre-incubationprotocol;or(b)useofbacteriaisolated onpre-culture]tostandard,traditionalbenchmicrobiology.

Material

and

methods

EthicsapprovalwasobtainedthroughtheEthicalCommittee ontheUseofAnimalsoftheSchoolofVeterinaryMedicine andAnimalScience(UniversityofSãoPaulo,Brazil,protocol number3002/2013)beforethecommencementofthestudy.

Samplecollectionandbacterialidentification

Compositemilksampleswerecollected fromall cowson2 commercialdairieslocatedintheMidwestregionofSãoPaulo State, Brazil.Based upon theseresults, milksamples were asepticallycollectedfromallquartersofpreviously culture-positivecows.Microbiologicalculture(referencemethod)was performedusingproceduresconsistentwithNationalMastitis Councilguidelines.Briefly,10␮Lofmilkpersample(quarter) were inoculated on blood agar plates with5% defibrinated bovine blood.Invertedplates were incubatedaerobicallyat 37◦Cfor48handobservedevery24hforcolonycharacteristics (shape,size,number,andcolor),hemolyticability(presence and type).Gramstain,potassiumhydroxidetest(KOH)and catalasetestswereperformedtodeterminethemorphology anddifferentiationbetweengenera.Specificmicrobiology pro-ceduressuchascoagulase,CAMP,esculin,bileesculinandpyr testwereperformedasdescribedbyOliveretal.11_Atotalof

120quartermilksamplesfrom40cowswerepositive,withthe followingnumbersofquarterspositivepercow:4cowswith 1,8cowswith2,12cowswith3and16cowswith4infected quarterspercow.Allquartermilksampleswerealso submit-tedtothenonculturebasedidentificationMALDI-TOFmethod (directsampleidentification-MALDI-TOF).

Directsampleidentificationofmastitis-causingpathogen byMALDI-TOFMS

Fordirectsampleidentificationofmastitiscausingpathogens, fat wasremoved (skimmed) from 1mLofmilksamplesby centrifugation(10,000×g/10min,followedbyremovalofthe superiorlayeroffat.Afterskimming,milksampleswere agi-tatedinavortexfor30s(KasvibasicK452820,Paraná,Brazil) andsubmittedtopre-incubationat37◦Cfor12hinawater bathwithagitation(Solab,SL155/10,SãoPaulo,Brazil).After theincubationperiod,onealiquotofeachmilksample(40mL)

wastakenandsubmittedfortotalbacterialcount(TBC)usinga flowcytometryequipmentBactoCount(BentleyInstruments, Chaska,MN,USA).

Theremainderofeach ofthemilksampleswas submit-tedtoapreparationprotocolforbacterialribosomalprotein extractionusingtheMALDISepsityper®kit(BrukerDaltonik, Bremen,Germany).Initially,200␮Lofthe“LysisBuffer” solu-tion was added to 1000␮L of milk sample, followed by a centrifugationstepat13,000×gduring2min.After centrifu-gation,thesupernatantwasdiscardedusinga1000␮Lpipette. Next,thepelletwasre-suspendedin1000␮Lofdistilledwater and200␮Lofthe“LysisBuffer”solution,followedbyasecond centrifugationstepat13,000×g.Afterdiscardingthe super-natantusinga1000␮Lpipette,thepelletwasdilutedin1000␮L of“WashingBuffer”solution,followedbyathirdroundof cen-trifugationat13,000×ganddisposalofthesupernatant.

The bacterial pellet was diluted in 1200␮L of a 75% ethanol/watersolution(900␮L/300␮L) inorder toinactivate the bacteria. Thebacterial sediment was then centrifuged andthesupernatantdiscardedbytubeinversion.Next,a sec-ondcentrifugationstepwasdonetoremovetheremaining ethanolpresentinthesampleandthesupernatantdiscarded usinga pipette.Afterpelletdryingatroom temperature, a 70%solutionofformicacidwasaddedinenoughvolumeto coverthepellet(∼30–50␮L)andlysethebacterialcells, fol-lowedbyadditionofthesame volumeof100%acetonitrile (∼30–50␮L).Duringthefinalstageofpreparation,thesamples werecentrifuged(13,000×g/2min)toseparatethesediment ofbacterial cellsfrom thesupernatantcontainingbacterial proteins,consistingmainlyofribosomalproteins(Biotyper3.0 manualpage157;Bruker Daltonik,Bremen,Germany). Vol-umesof1.0␮Lofthebacterialextractwereappliedontosteel targetplates(mtp384TargetPolishedSteel;BrukerDaltonik, Bremen,Germany),followedbydryingatroomtemperature. Thedriedsupernatantwasoverlaidwith1.0␮Lofmatrix solu-tion,consistingof␣-cyano-4-hydroxy-cinnamicaciddilutedin 50%acetonitrileand2.5%trifluoroaceticacid.

Aftertheprotocolforbacterialribosomalprotein extrac-tion,thealiquotsweresubmittedforMALDI-TOFMSanalysis as described by Barreiro et al.2 _{The Bruker Bacterial Test}

Standard (BTS) was used for the mass calibration and instrumentparameteroptimization.Briefly,themassspectra were obtained using an Autoflex III (Bruker Daltonik, Bil-lerica, USA) mass spectrometer and were collected within a2000–20,000m/zmassrange.Spectraldataprocessingwas doneusing theBiotyper 3.0(Bruker Daltonik,Bremen, Ger-many) computer softwarefor microorganism identification (MBTversion6903MPSlibrary).

Theresult was given bymeans ofan algorithm (score) obtainedbythe Biotyper3.0software,inwhich scores<1.7 were consideredas non-reliablediagnoses;scores ≥1.7but <2.0wereconsideredasreliableforgenusidentification;and scores≥2.0werereliableforidentificationofgenusand bac-terialspecies.

Standardprotocolformastitispathogenidentificationby

MALDI-TOFMS

Thebacterialisolates(n=120)werealsosubmittedfor iden-tification byMALDI-TOFMS usingbacterial coloniesgrown

onbloodagar,andusingthelysisprotocoldescribedby Bar-reiroetal.2_{Briefly,thecoloniesweredilutedin1200␮Lofa}

75%ethanol/watersolution(900␮L/300␮L)forbacterial inac-tivation. The bacterial sediment was centrifuged and the supernatantdiscardedbytubeinversion.Asecondroundof centrifugationwasdonetoremoveremainingethanolpresent inthesample,andafterwards,thesupernatantwasremoved using apipette.Allcentrifugationstepswere performedat 13,000×gfor2min.Thepelletwasleftdryingatroom tem-perature,anda70%formicacidsolutionwasaddedinenough volume to cover the pellet (∼30–50␮L) and lyse the bacte-rial cells.Afterthat, the same volumeof100% acetonitrile (∼30–50␮L)wasadded.Duringthefinalstageofpreparation, centrifugationwasdoneinordertoseparatethesediments ofthebacterialcellsfromthesupernatantcontaining bacte-rialproteins,mainlyribosomalproteins(Biotyper3.0manual page 157; Bruker Daltonik, Bremen, Germany). After that, MALDI-TOF MS spectra were obtainedand used for bacte-rialclassificationusingtheBiotyper3.0computerprogramfor microorganismidentification.

Statisticalanalysis

StatisticaltestswereperformedusingMcNemartestonthe pairedproportions12_{andstatisticalsignificancewasdeclared}

atP<0.05.Underthenullhypothesistherewasnodifference betweenstandardmicrobiologicalculture(usedasthe refer-encemethod)andnon-culturedirect-MALDI-TOFmethodof milksamplesafterpre-incubationprotocol.

Results

StandardMALDI-TOFMSprotocolusingbacterialcolonies grownonbloodagarvs.directsampleidentificationafter pre-incubationprotocol

Thereferencemethod,standardmicrobiologicalcultureofthe 120 quartermilk samples,identified Streptococcusagalactiae

(n=32);Streptococcusuberis(n=19);Staphylococcusaureus(n=14) and coagulase-negative Staphylococci, CNS (n=55) (Table 1). Among Streptococcus agalactiae isolates identified by micro-biological culture(n=32), thestandardMALDI-TOF protocol usingbacterialcoloniesgrownonbloodagarcorrectly iden-tified 31 isolates as Streptococcusagalactiae and one isolate as Streptococcus dysgalactiae. For Streptococcus uberis isolates identifiedbymicrobiologicalculture(n=19),MALDI-TOFMS identified14isolatesasStreptococcusuberis,andtheremaining ofisolatesasStreptococcusagalactiae (n=3) andStreptococcus dysgalactiae(n=2).ConsideringStaphylococcus aureusisolates (n=14) identified by microbiological culture, 10 were iden-tified by the standard MALDI-TOF protocol using bacterial coloniesgrownonbloodagarasStaphylococcusaureus,andthe remaining(n=4)asStaphylococcushaemolyticus.Withrespect toCNSidentifiedbymicrobiologicalculture(n=55),the stan-dardMALDI-TOFprotocolusingbacterialcoloniesgrownon bloodagaridentified51CNS(variousspecies,seeTable1),3 isolatesasLactococcusgarvieaeandoneasStaphylococcusspp., identifiedatthegenuslevel(Table1).

Table1–Resultsoftheidentificationofmastitiscausingpathogens(n=120)usingthestandardMALDI-TOFprotocol usingbacterialcoloniesgrownonbloodagarandmicrobiologicalculture.

Microbiologicalculture n StandardMALDI-TOFprotocolusing

bacterialcoloniesgrownonbloodagara

n Divergenceb

Streptococcusagalactiae 32 Streptococcusagalactiae 31

Streptococcusdysgalactiae 1 1

Streptococcusuberis 19 Streptococcusuberis 14

Streptococcusagalactiae 3 3

Streptococcusdysgalactiae 2 2

Staphylococcusaureus 14 Staphylococcusaureus 10

Staphylococcushaemolyticus 4 4 CNSc _{55 Staphylococcuschromogenes 15} Staphylococcushaemolyticus 13 Staphylococcusxylosus 8 Staphylococcusepidermidis 7 Staphylococcuscapitis 3 Staphylococcushominis 3 Staphylococcussaprophyticus 1 Staphylococcussimulans 1 Lactococcusgarvieae 3 3 Staphylococcusspp. 1 1 Total 120 120 14

a _{Identificationatspecieslevel(score>2.0);}

b _{Divergencebetweenmethodologies,numberofisolates;} c _{CNS:coagulase-negativeStaphylococcus.}

PercentageagreementusingstandardMALDI-TOFprotocol usingbacterialcoloniesgrownonbloodagarforthe identifi-cationofpathogenscausingsubclinicalmastitiswasof96.8% forStreptococcusagalactiae;73.6%forStreptococcusuberis;71.4% forStaphylococcusaureusand92.7%forCNS,whencompared tomicrobiologicalcultureresults.

Mastitispathogenidentificationusingdirectsample identificationafterpre-incubationprotocol

In comparison to the reference microbiology method, the direct sample identification after pre-incubation protocol (directMALDI-TOFMSmethod)onlyidentified15(27.2%)of the55isolatesasCNS(Table2).ThisprotocolidentifiedCNSin 9/15milksamples,atthegenusandspecieslevel(score≥2.0), 1/15atthegenuslevel(score≥1.7–1.9),2/15milksampleswere notreliablyidentifiedatthe genusandspecieslevel (score <1.7), and 3/15 were identified as Staphylococcus spp. with scores≥2.0butwithoutspeciessuggestionbyspectraldata processingusingtheBiotyper3.0.Amongtheremainingmilk samplesidentifiedasCNSbymicrobiologicalculture, identifi-cationbydirect-MALDI-TOFmethodwasnotpossiblein24/55 (43.6%;notreliableidentificationbydirect-MALDI-TOF),and fortheothersamples,therewasdisagreementbetween direct-MALDI-TOFandbymicrobiologicalculture.Forexample,15/55 samples(27.2%)previouslyidentifiedasCNSwereidentified usingdirect-MALDI-TOFasLactobacillusspp.,and1/55(1.8%) asStaphylococcusaureuswithascore≤1.7(Table2).

Seven out of 32 (21.8%) milk samples were identified asStreptococcusagalactiaebydirect-MALDI-TOFmethod,1/32 (3.1%)isolatebeingidentifiedatthegenusandspecieslevel (score≥2.0),1/32(3.1%)atthegenuslevel(score≥1.7–1.9),and 5/32 (15.6%)showedtobeindicativeofStreptococcus agalac-tiaebuthadanidentificationscoreof<1.7.Fortheremaining

Streptococcus agalactiae identified by culture (n=25), 12/25 isolates(48%)wereidentifiedasLactobacillusspp.bythe direct-MALDI-TOFmethod,and for13/25(52%)isolatesitwasnot possibletoidentifyanypathogenusingthismethod(Table2). Thedirect-MALDI-TOFmethodcorrectlyidentified1/19 iso-late(5.2%)ofStreptococcusuberis,althoughitwasonlyatthe genuslevel(score≥1.7–1.9).Amongtheremaining Streptococ-cusuberisidentifiedbyculture(n=18),for12/18milksamples there was no identification of any pathogen causing sub-clinical mastitisbythedirect-MALDI-TOF(66.6%),5/18milk samples(27.7%)wereidentifiedasLactobacillusspp.and1/18 sample(5.5%)asStreptococcusagalactiae(Table2).

The methodology of direct identification of pathogens causingsubclinicalmastitisbyMALDI-TOFMSidentified2/14 isolates(14.3%)asStaphylococcusaureus,althoughwithscores <1.7.AmongtheremainingStaphylococcusaureusidentifiedby culture(n=12),for8/12isolates(66.6%)Staphylococcusaureus

wasnotidentifiablewhensubmittedtodirect-MALDI-TOF,and for4/12samples(33.3%),Lactobacillusspp.wasidentifiedby direct-MALDI-TOF,withscores≤1.6(Table2).

Theevaluatedmethodsweresignificantlydifferent, indi-catingdisagreementofidentificationofsubclinical mastitis-causing pathogens between both methods (P<0.05). The direct-MALDI-TOFmethod,whencomparedto microbiolog-icalculture,correctlyidentifiedisolatesofcoagulase-negative

Staphylococci(27.2%),Streptococcusagalactiae(21.8%), Staphylo-coccusaureus(14.2%),andStreptococcusuberis(5.2%).

Discussion

The method using a pre-incubation procedure for a non-culture direct sample-MALDI-TOF method showed low accuracy ofidentification ofmastitiscausing pathogens in our study when compared with the results obtained by

Table2–EfficiencyofthedirectsampleidentificationofmastitiscausingpathogensbyMALDI-TOFMSinquartermilk samples(n=120)vs.microbiologicalculture.

Microbiologicalcultureresults(n) MALDI-TOFMS TBC(×103₎

Isolates(n) Biotypera_score

Coagulase-negative Staphylococcus(55) Unclassifiedsamples(24) 1412.0 Staphylococcusepidermidis(2) Staphylococcushaemolyticus(1) Staphylococcuschromogenes(9) <1.7(n=2)≥1.7–1.9(n=1)≥2.0(n=9) 1979.4 Lactobacillusspp.(15) <1.7(n=9)≥2.0(n=6) 1832.2 Staphylococcusspp.(3) ≥2.0(n=3) 3772.0 Staphylococcusaureus(1) <1.7(n=1) 3657.5 Streptococcus agalactiae(32) Unclassifiedsamples(13) 6776.9 Streptococcusagalactiae(7) <1.7(n=5)≥1.7–1.9(n=1)≥2.0(n=1) 7777.0 Lactobacillusspp.(12) <1.7(n=12) 6634.5

Streptococcusuberis(19) Unclassifiedsamples(12) 3083.3

Streptococcusuberis(1) ≥1.7–1.9(n=1) 5153.0

Lactobacillusspp.(5) <1.7(n=3)≥2.0(n=2) 3454.1

Streptococcusagalactiae(1) <1.7(n=1) 9999.0

Staphylococcusaureus(14) Unclassifiedsamples(8) 4251.3

Staphylococcusaureus(2) <1.7(n=2)

-Lactobacillusspp.(4) <1.7(n=4) 4028.1

a _{Identificationatgenusandspecieslevel,score≥2.0;identificationatgenuslevel,score≥1.7;unreliableidentification,score<1.7.}

microbiological culture. Theidentification ofpathogens by directsample-MALDI-TOFinclinicalsamplesofhumanurine andcerebrospinalfluidwassuccessfullydescribedwhenthe bacterialcountwas >104_–106_{cfu/mL.These bacterial count}

valueswereobtainedinpatientswithurinarytractinfection.7

Similar to the urine samples, the cerebrospinal fluid was evaluatedbydirect-MALDI-TOFforidentificationof microor-ganismsthatcausebacterialmeningitis.8_{Inthepresentstudy,}

we observed TBC varied from 1.4 to 9.9×106_{cfu/mL after}

pre-incubationofmilksamples.However,attainingasimilar identificationpercentageofthepathogens causing subclin-icalmastitisbydirect-MALDI-TOFwasnotpossibleinmilk samplesfrommammaryquarters,asdescribedpreviouslyfor urineandcerebrospinalfluidsamples.8

Thebacterialcountinexperimentallycontaminatedmilk sampleswas a limiting factor that influenced the analytic sensitivityofthedirect-MALDI-TOFmethodforidentification ofpathogensthatcausemastitis.13_{AccordingtoMoussaoui}

et al.10 _{the pre-incubation of blood samples allowed the}

identification ofbacteriabythe direct-MALDI-TOFmethod. Similarly,Ferreiraetal.7_{reportedthatthedirect-MALDI-TOF}

methodwas efficient in bacterial identification in infected urine whenthe bacterial countwas>105_cfu/mL.

Nonethe-less,inthepresentstudy,usingthecutoffofTBCpublished from Barreiro et al.3 _{combined witha 12-h pre-incubation}

protocolofthequartermilksamplesdidnotresultinhigher identificationofmastitis-causingpathogens,sincethe direct-MALDI-TOF method identified ≤20.8% (25/120 isolates) of specificpathogenidentifiedbymicrobiologicalculture.

TheTBCofthemilksamplesfromnaturallyinfectedcows mayhavebeenaninfluencingfactorinthereduced identi-ficationcapacityofthepathogensbythedirect-MALDI-TOF method.Also,the pre-incubation ofthe milksamplesmay havefavoredthegrowthofsymbioticorcontaminantbacteria that could bepresent in milk samples, such as Lactobacil-lusspp.,whichwouldnotallowsufficientbacterialcountfor

the identification ofpathogenscausing subclinicalmastitis bythedirect-MALDI-TOFmethod.Anotherpossible interfer-ingfactorwasthepresenceofmilkcomponents(proteinand fat),evenafterthewashingprotocolusingLysisBuffer,which mayhaveinterferedinspectraacquisitionforthedirect iden-tification of the pathogens causing subclinical mastitis by direct-MALDI-TOFmethod.Milkproteinscouldaffect classi-fication resultswiththe effectiverange between3,000 and 20,000m/z.Whilecomparingmassspectraoriginatingfrom the Standard MALDI-TOF protocol using bacterial colonies grownonbloodagarvs.directsample-MALDI-TOF identifica-tionprotocol,interferencepeaksrangingbetween2,000and 7,000 m/zwere observed foridentification of Staphylococcus aureus(Fig.1).Oneofthemainlimitingfactorsofthe direct-MALDI-TOFmethodisrelatedtothegreatheterogeneityand complexityofthesetofproteinspresentinmilk.Thus,the lowertheconcentrationofthetargetproteinsofidentification, thehighertherisktobemisidentifiedbydirect-MALDI-TOF method,duetothedetectionlimitandtothepossible overlap-pingofmilkproteins,whichpresentsimilarmolecularweight andisoelectricpoints.

Thedirect-MALDI-TOFmethod,whencomparedto micro-biologicalculture,correctlyidentified27.2%,21.8%,14.2%and 5.2%oftheCNS,Streptococcusagalactiae,Staphylococcusaureus

and Streptococcus uberis isolates, respectively. According to La Scolaand Raoult,14 _{the agreementbetween MALDI-TOF}

MS resultsforthe direct identificationof bacteriainblood and inmicrobiologicalcultureswas 59%ofevaluated sam-ples. However, the later study found that for samples in which Streptococcusspp.was isolated,therewaslow agree-mentwiththereferencemethod(4/25,16%).Aftermodifying the preparation protocol for blood samples, there was an increaseof67%intheefficiencyofGram-positive identifica-tion,mainlyforStaphylococcus spp.,thoughforStreptococcus

spp., identificationremained low(4/17, 23%).14 _Webelieved

Intens [a.u.] x105 x105 x104 2.0 1.5 1.5 1.0 1.0 0.5 0.5 4 3 2 1 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 m/z 0.0 0.0

a

b

C

Fig.1–Spectrarangingfrom2000to20,000m/z:(a)StaphylococcusaureusidentifiedbythedirectMALDI-TOFmethodofan experimentallyinoculatedmilksample,(b)Staphylococcusaureusidentifiedinthecolony(ATCC29213)bythestandard MALDI-TOFprotocolusingbacterialcoloniesgrownonbloodagar,and(c)aquartermilksamplewithoutmicrobiological growth.

direct-MALDI-TOFmethodmaybeoccurredbecause MALDI-TOFMSdoesnotidentifyspectrafrommixedculturesreliably. Thecorrectnessofclassificationisbasedonthefalse assump-tion that the “reference ID method” pathogen is the only organismpresent.Despiteitremainadisadvantage,thedirect MALDI-TOFMSmethodmaybeusedasanalternativeto con-ventionalbacterialidentification.

Regardless of the pathogen type studied, we observed the frequency of 36/120 Lactobacillus spp. (30%) identified by the direct-MALDI-TOF MS method combined with the pre-incubationprotocol.Theidentificationfrequencyof Lac-tobacillusspp.mayhaveoccurredduetothepre-incubation protocolofthe milksamples,whichallowedanincreasein thebacterialcountofthesesymbioticmicroorganisms.

Inthepresentstudy,regardingallisolatescorrectly identi-fiedasCNS(15isolates)bythedirect-MALDI-TOFMSmethod, 9/15ofthemwereidentifiedatthegenusand specieslevel (score≥2.0).WithregardtotheStreptococcusagalactiaeisolates, onlytwowereidentifiedatthegenusandspecieslevel(score >2.0).Thelevelofidentificationreliabilitywasevenlowerin

Streptococcusuberis(score≥1.7–1.9)andStaphylococcusaureus

(score<1.7).Therefore,thedirect-MALDI-TOFmethodforthe identification ofthe pathogens causingsubclinical mastitis showed low identification percentages when compared to microbiologicalculture.

In this study, the standard MALDI-TOF protocol using bacterial colonies grown on blood agar identified 88.3% (106/120) of the pathogens causing subclinical mastitis at the genus and species level (score ≥2.0), when compared tomicrobiologicalculture.Theuse ofstandard MALDI-TOF protocolusingbacterialcoloniesgrownonbloodagarforthe identificationofbacteriainclinicalbacteriologyhasbecome more common. Our results are similar to other studies on bacterial identification by MALDI-TOF using bacterial colonies, in which identification percentages at the genus levelwere97–99%,andatthespecieslevel,85–97%.15_Onthe

other hand,thepre-incubationofthequartermilksamples foridentificationofpathogencausingsubclinicalmastitisby the direct-MALDI-TOF method did not display satisfactory results.Furthermore,themilksamplepre-incubationprotocol mayhavefavoredsymbioticandothercontaminantbacterial growth (e.g.Lactobacillusspp.),resultinginlowfrequencyof identificationbythedirectsample-MALDI-TOFmethod.

In an previous study developed by our research group, we observed that a minimum of TBC varying of ≥106 _to

≥108_{cfu/mL would berequiredforsuccessfulidentification}

scores atthe gender and species level, and it also varied according to the mastitiscausing pathogen.3 _{However, the}

presentstudyaimedtoevaluatearapididentification proto-col(withoutculture)usingthepre-incubationofmilksamples followed by directidentification byMALDI-TOF MS. There-fore,milksamplesweresubmittedtoTBCinordertoidentify whetherthemicrobialloadbeforethepre-incubationprotocol waswithintherangeofTBCproposedinthepreviousstudy.3

Thus,itwouldbepossibletousearapidanddirectsample protocolforidentificationofmastitiscausingpathogens. How-ever,wedidnotexpectthatthepre-incubationprotocolwould havefavoredsymbioticandbacterialgrowth(e.g.,Lactobacillus

spp.).Theseresultssuggestthatwecouldhaveincludedthe identificationofmilksamplesafterpre-incubationprotocolby traditionalbenchmicrobiologyandspecificmicrobial count-ingmethods(e.g.standardplatecount–SPC)priortotheuse ofthemilksampleincubationprotocol,whichwerecognize aslimitationsofthepresentstudy.Analternativeapproach thatmayimproveresultswouldbetodevelopsomemethod ofinactivatingsymbioticbacteriapriortopre-incubation.

Conclusions

ThenonculturebasedidentificationMALDI-TOFMSmethod, even after the 12h pre-incubation protocol of the quarter

milksamples,didnotaccuratelyhelptopromotetherapid identification of mastitis pathogens. The pre-incubation protocol ofmilk samples, associatedto the direct identifi-cation (nonculture-based) method by MALDI-TOF MS, did notincreasetheidentificationatspecieslevel(score>2.0)of pathogenscausingsubclinicalmastitisincomparisontothe methodwithoutpreviousincubation.

Conflict

of

interest

statement

Theauthorsdeclarenoconflictsofinterest.

Acknowledgments

We are grateful to the Fundac¸ão de Amparo à Pesquisa do Estado de São Paulo (FAPESP), Brazil,for a scholarship award (2011/14456-0) and research funding (2011/15815-4). TheauthorsaregratefultoDr.KevinL.Anderson(Department ofPopulationHealthandPathobiology,CollegeofVeterinary Medicine,NorthCarolinaStateUniversity)whoreviewedthe Englishwriting.

r

e

f

e

r

e

n

c

e

s

1.ReadMM.TrendsinDNAfingerprintingresearch.NewYork,NY: NovaSciencePublisherInc.;2005.

2.BarreiroJR,FerreiraCR,SanvidoGB,etal.Short

communication:identificationofsubclinicalcowmastitis pathogensinmilkbymatrix-assistedlaser

desorption/ionizationtime-of-flightmassspectrometry.J DairySci.2010;93(12):5661–5667.

3.BarreiroJR,GoncalvesJL,BragaPA,etal.Non-culture-based identificationofmastitis-causingbacteriabyMALDI-TOF massspectrometry.JDairySci.2017;100(4):2928–2934.

4.KokJ,ThomasLC,OlmaT,etal.Identificationofbacteriain bloodculturebrothsusingmatrix-assistedlaser

desorption-ionizationSepsityperandtimeofflightmass spectrometry.PLoSOne.2011;6(8):e23285.

5.SchubertS,WeinertK,WagnerC,etal.Novel,improved samplepreparationforrapid,directidentificationfrom positivebloodculturesusingmatrix-assistedlaser desorption/ionizationtime-of-flight(MALDI-TOF)mass spectrometry.JMolDiagn.2011;13(6):701–706.

6.IlinaEN,BorovskayaAD,MalakhovaMM,etal.Direct bacterialprofilingbymatrix-assistedlaser

desorption-ionizationtime-of-flightmassspectrometryfor identificationofpathogenicNeisseria.JMolDiagn.

2009;11(1):75–86.

7.FerreiraL,Sanchez-JuanesF,Gonzalez-AvilaM,etal.Direct identificationofurinarytractpathogensfromurinesamples bymatrix-assistedlaserdesorptionionization-timeofflight massspectrometry.JClinMicrobiol.2010;48(6):

2110–2115.

8.SegawaS,SawaiS,MurataS,etal.Directapplicationof MALDI-TOFmassspectrometrytocerebrospinalfluidfor rapidpathogenidentificationinapatientwithbacterial meningitis.ClinChimActa.2014;435:59–61.

9.StevensonLG,DrakeSK,MurrayPR.Rapididentificationof bacteriainpositivebloodculturebrothsbymatrix-assisted laserdesorptionionization-timeofflightmassspectrometry. JClinMicrobiol.2010;48(2):444–447.

10.MoussaouiW,JaulhacB,HoffmannAM,etal.Matrix-assisted laserdesorptionionizationtime-of-flightmassspectrometry identifies90%ofbacteriadirectlyfrombloodculturevials. ClinMicrobiolInfect.2010;16(11):1631–1638.

11.OliverSPO,GonzálezRN,HoganJS,etal.Microbiological proceduresforthediagnosisofbovineudderinfectionand determinationofmilkquality.In:Aglobalorganizationfor mastitiscontrolandmilkquality.4thed.Verona,WI:National MastitisCouncil;2004.p.1–40;44–46.

12.BlandJM,AltmanDG.Statisticalmethodsforassessing agreementbetweentwomethodsofclinicalmeasurement. Lancet(Lond,England).1986;1(8476):307–310.

13.BarreiroJR,BragaPA,FerreiraCR,etal.Nonculture-based identificationofbacteriainmilkbyproteinfingerprinting. Proteomics.2012;12(17):2739–2745.

14.LaScolaB,RaoultD.Directidentificationofbacteriain positivebloodculturebottlesbymatrix-assistedlaser desorptionionisationtime-of-flightmassspectrometry.PLoS One.2009;4(11):e8041.

15.PatelR.MALDI-TOFmassspectrometry:transformative proteomicsforclinicalmicrobiology.ClinChem. 2013;59(2):340–342.