Next-Generation Sequencing Combined with Specific PCR Assays To Determine the Bacterial 16S rRNA Gene Profiles of Middle Ear Fluid Collected from Children with Acute Otitis Media

Nex -Gene a ion Sequencing Combined wi h Specific PCR Assays To De e mine he Bac e ial 16S RNA Gene P ofiles o Middle Ea Fluid Collec ed om Child en wi h Acu e O i is Media Saa a Sillanpää, a Lenka K amna, b Sami Oika inen, c Ma kku Sipilä, a Ma kus Rau iainen, a Janne Ai oniemi, d Jussi La anne, a Heikki Hyö y, c,d Ond ej Cinek b Depa men o O o hinola yngology and Head and Neck Su ge y, Tampe e Uni e si y Hospi al and School o Medicine, Uni e si y o Tampe e, Tampe e, Finland a ; Depa men o Pedia ics, 2nd Facul y o Medicine, Cha les Uni e si y in P ague and Uni e si y Hospi al Mo ol, P ague, Czech Republic b ; Depa men o Vi ology, School o Medicine, Uni e si y o Tampe e, Tampe e, Finland c ; Fimlab Labo a o ies, Tampe e, Finland d ABSTRACT The aim o he s udy was o analyze he bac e iome o acu e o i is me- dia wi h a no el modifica ion o nex -gene a ion sequencing echniques. Ou pa ien child en wi h acu e o i is media we e en olled in he s udy, and middle ea fluids we e collec ed du ing 90 episodes om 79 subjec s aged 5 o 42 mon hs (median age, 19 mon hs). The bac e iome p ofiles o middle ea fluid samples we e de e mined by a nes ed-PCR amplifica ion o he 16S RNA gene (V4 egion), ollowed by mass sequenc- ing. The p ofiling esul s we e compa ed o he esul s o specific PCR assays a ge ing selec ed p e alen pa hogens. Bac e iome p ofiling using nes ed amplifica ion o low- olume samples was aided by a bioin o ma ic sub ac ion o signal con aminan s om he ecombinan polyme ase, achie ing a sensi i i y sligh ly lowe han ha o specific PCR de ec ion. S ep ococcus pneumoniae was de ec ed in 28 (31%) samples, Haemophi- lus influenzae in 24 (27%), Mo axella ca a halis in 18 (20%), S aphylococcus spp. in 21 (23%), Tu icella o i idis in 5 (5.6%), Alloiococcus o i idis in 3 (3.3%), and o he bac e ia in 14 (16%) using bac e iome p ofiling. S. pneumoniae was he dominan pa hogen in 14 (16%) samples, H. influenzae in 15 (17%), M. ca a halis in 5 (5.6%), T. o i idis in 2, and S aphylococcus au icula is in 2. Weake signals o P e o ella melaninogenica,Veillonella dispa , and Veillonella mon pellie ensis we e no ed in se e al samples. Fou een samples (16%) we e no explainable by bac e ial pa hogens; no el causa i e agen s we e no de- ec ed. In conclusion, unbiased bac e iome p ofiling helped in depic ing he ue mu ual quan i a i e a ios o ea bac e ia, bu a p esen , i s complica ed p o ocol impedes i s ou ine clinical use. IMPORTANCE Al hough S. pneumoniae,H. influenzae, and M. ca a halis ha e been long es ablished as he mos impo an pa hogens in acu e o i is media using cul- u e and specific PCR assays, he knowledge o hei mu ual quan i a i e ela ions and possible oles o o he bac e ia is incomple e. The ad en o unbiased bac e i- ome 16S RNA gene p ofiling has allowed he de ec ion o nea ly all bac e ia p es- en in he sample, and i helps in depic ing hei mu ual quan i a i e a ios. Due o he di ficul ies in pe o ming mass sequencing in low- olume samples, only a ew bac e iome-p ofiling s udies o o i is media ha e been published, all limi ed o cases o ch onic o i is media. He e, we p esen a s udy on samples ob ained om young child en wi h acu e o i is media, success ully using a s a egy o nes ed PCR coupled wi h mass sequencing, and demons a e ha he me hod can con e quan i a i e in- o ma ion ha dly ob ainable by o he me hods. Recei ed 8 Janua y 2017 Accep ed 3 Ma ch 2017 Published 22 Ma ch 2017 Ci a ion Sillanpää S, K amna L, Oika inen S, Sipilä M, Rau iainen M, Ai oniemi J, La anne J, Hyö y H, Cinek O. 2017. Nex -gene a ion sequencing combined wi h specific PCR assays o de e mine he bac e ial 16S RNA gene p ofiles o middle ea fluid collec ed om child en wi h acu e o i is media. mSphe e 2: e00006-17. h ps://doi.o g/10.1128/mSphe e .00006-17. Edi o Paul D. Fey, Uni e si y o Neb aska Medical Cen e Copy igh © 2017 Sillanpää e al. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0 In e na ional license. Add ess co espondence o Saa a Sillanpää, saa a.sillanpaa@u a.fi. RESEARCH ARTICLE Hos -Mic obe Biology c ossm Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 1 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om KEYWORDS 16S p ofiling, acu e o i is media, bac e iome p ofiling, mass sequencing, nex -gene a ion sequencing Acco ding o adi ional bac e ial cul u e, Haemophilus influenzae,S ep ococcus pneumoniae, and Mo axella ca a halis ha e been es ablished as he main causes o acu e o i is media (AOM) (1, 2). The de ec ion o bac e ial species by cul u e can, howe e , be biased by he a ious g ow h p ope ies o he agen s, and he e o e, he popula i y o molecula es ing is g owing: i can p o ide mo e exac in o ma ion abou he bac e ial e iology o AOM and has also been ins umen al in he sys ema ic de ec ion o no el candida e o ganisms, such as Alloiococcus o i idis (3). Since AOM e ol es ela i ely as and esul s om an acu e inflamma o y p ocess, he p esence o any bac e ium in high quan i y in he middle ea fluid (MEF) is gene ally accep ed as a sign o i s causa i e ole in ha AOM episode. Howe e , causali y is less clea o bac e ial species ha a e ound in lowe quan i ies in he MEF. Occasionally, i uses a e he causa i e agen s, and he con inuous eed o nasopha yngeal flo a h ough he eus achian ube du ing he inflamma o y p ocess may mislead he bac- e iological assessmen . Fu he mo e, he ex e nal ea canal may con amina e he sample, depending on he way he MEF is collec ed upon my ingo omy. O he s (3–11) and we (12) ha e pe o med s udies wi h specific PCR de ec ion assays o MEF bac e ia in AOM: al hough he s udies di e ed in he defini ion o cases, in he spec um o es ed pa hogens, and in he PCR p ime s and p o ocols used, hey mos ly ag eed on he high equencies and quan i ies o H. influenzae and S. pneumoniae, whe eas he equencies o o he bac e ial species ha e a ied widely be ween he s udies. Bac e iome-p ofiling me hods we e he e o e wa an ed o ob aining an unbiased pic u e o he bac e ial flo a and o disco e as-ye -uniden ified bac e ia (7). Mass sequencing has made i possible o cha ac e ize he whole bac e iome by pa allel p ofiling o he 16S RNA gene in he whole bac e ial popula ion. Theo e ically, i s unbiased cha ac e allows he de ec ion o nea ly all bac e ia p esen in he sample, hei axonomic e alua ion, and mu ual ela i e quan ifica ion. This makes i possible o ob ain mo e de ailed in o ma ion on pa hogens in AOM and o iden i y mic obes whose e iological ole could ha e emained dubious using adi ional mic obe-specific me hods. Howe e , he applica ion o mass sequencing in esea ch on o i is media has p o en di ficul , and only h ee s udies o o i is media based on 16S RNA gene p ofiling ha e been published o da e (13–15), all o hem on ch onic o i is media. The pauci y o published s udies has clea ly demons a ed he echnical di ficul y in he amplifica ion o low-quan i y samples wi h p ime s ca ying indices and adap e s o he mass sequencing. The goal o he p esen s udy was o cha ac e ize he bac e ial composi ion o MEF om young Finnish child en wi h AOM and assess he ole o pa hogens no es ab- lished in he disease e iology. Fo his pu pose, we adap ed a p o ocol o sensi i e nes ed PCR coupled wi h mass sequencing, capable o cha ac e izing bac e ial 16S RNA gene p ofiles in samples wi h small quan i ies o bac e ia. RESULTS Samples and hei bac e ial p ofiles. Nine y MEF samples ob ained du ing AOM episodes we e collec ed om 79 child en aged 5 o 42 mon hs (median age, 19 mon hs). Ele en child en con ibu ed wo samples du ing wo sepa a e AOM episodes. The composi ions o he 16S RNA gene p ofiles o he MEF samples a e shown in Fig. 1, along wi h he cul u e esul s. The ela i e p esence and ela i e abundance o di e en bac e ia a e shown in Table 1 and Fig. 2. The 16S RNA gene signal o an indi idual axon in a sample was exp essed as he ac ion o he o e all signal om ha sample. Bac e ia in indi idual samples we e quan ified in o ou ca ego ies, as ollows: (i) he dominan pa hogen (hal o mo e o he sequencing signal wi hin a sample), (i) he sole finding, wi h less han hal o he signal ( he es o he signal being Sillanpää e al. Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 2 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om FIG 1 Bac e ia ound in AOM samples and s eng hs o hei signals. The e ical axis shows indi idual samples: /2, second sample om he subjec ; PERF, sample om pe o a ion; TS, sample om ympanos omy ube; o he samples a e om my ingo omy; c, cul u e; neg., cul u e nega i e; Sp, S. pneumoniae; Hi, H. influenzae; Mc, M. ca a halis; Sa, S aphylococcus au eus; o he , o he bac e ia. The ho izon al (Con inued on nex page) Bac e iome P ofiling in Acu e O i is Media Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 3 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om con aminan signal om he ecombinan polyme ase), (iii) a nondominan pa o a mixed flo a ( he bac e ium being assigned less han hal o eads and p esen wi h o he s in he sample), and (i ) nega i e (no eads o less han 3% o he signal wi hin he sample). The mos equen ly obse ed species was S. pneumoniae, being p esen in 28 (31%) samples, in hal o which i showed a s ong, dominan signal. The second mos equen ly obse ed pa hogen was H. influenzae (24 samples [27%]), also equen ly dominan . Simul aneous s ong posi i i y o hese wo agen s was obse ed only once in ou da a se (S. pneumoniae and H. influenzae in sample 1 o pa ien 7) (Fig. 1). M. ca a halis was p esen in 18 (20%) p ofiles, mos o en as a pa o mixed flo a in a he small quan i ies. Howe e , in fi e samples (5.6%), i was a clea ly dominan pa hogen wi h ew o no o he bac e ia p esen . S aphylococcus spp. we e equen as a genus (21 samples [23%]) bu we e seldom he dominan pa hogen (3 samples). The a iable egion 4 (V4) 16S RNA gene p ofiles we e no ins umen al in u he axonomic classifica ion o s aphylococci, as he sequences o V4 a e iden ical o nume ous S aphylococcus species. F om he p e ious es ing, we knew ha only one o he samples was posi i e o S. au eus among hose wi h S aphylococcus as dominan pa hogen (pa ien 44) (Fig. 1). We he e o e pe - o med Sange sequencing o he V3 o V5 egions o he emaining wo samples wi h dominan S aphylococcus signals (pa ien s 20 and 21) (Fig. 1) and ound s ong signals o S aphylococcus au icula is; his bac e ium was also ound in smalle quan i ies in se e al o he samples by specific PCR. Bac e ia less o en implica ed in AOM e iology. Tu icella o i idis was obse ed in fi e samples (5.6%), o which wo occu ences we e a s ongly posi i e dominan finding, one was a weake signal om a sole bac e ium p esen in he sample, and wo came om mixed flo a. A. o i idis was ound in h ee samples (3.3%), always as a componen o polymic obial flo a: once wi h H. influenzae (pa ien 6), once wi h M. ca a halis and a S aphylococcus sp. (pa ien 23), and once wi h a S aphylococcus sp. (pa ien 45). No o he bac e ia we e no ed as s ong dominan pa hogens, bu upon inspec ion o he weake signals, we ound and confi med P e o ella melaninogenica,Veillonella dispa , and Veillonella mon pellie ensis, mos ly as a componen o a mul ibac e ial flo a (Fig. 1). No candida e o a causa i e agen was ound in 14 samples (16%), ei he because he e emained no bac e ial signal a e sub ac ion o con aminan signal a ising om PCR chemicals (11 ins ances) o because such a signal was weak and could no be unambiguously axonomically assigned (3 ins ances). Compa ison wi h specific PCR. The posi i e esul s o 16S p ofiling showed e y good ag eemen wi h he esul s o specific PCR es s pe o med p e iously o H. in- fluenzae,S. pneumoniae,M. ca a halis, and A. o i idis (12). The PCR es ing was done be o e his mass sequencing was pe o med and was he e o e blinded o he p ofiling esul s. The compa ison is plo ed in Fig. 3. The ag eemen be ween he p ofiling and specific PCR quan ifica ion was especially igh when he pa hogen was dominan . The nega i i y o a bac e ium in he 16S p ofile, howe e , was no an en i ely eliable indica o o i s ue absence: e y small quan i ies o bac e ia o en emained unde- FIG 1 Legend (Con inued) axis shows he nine mos abundan species o gene a ha exceeded 3% o he o al sample sequencing signal in a leas wo AOM samples. The in e sec ions a e he ela i e abundances o he o al sequencing signal as pe cen ages ( ounded o he nea es in ege ). Taxa comp ising ⬍3% o he o al sample sequencing signal we e dis ega ded. The assigna ion o species by 16S RNA gene p ofiling is simplified, since he 16S p ofiling o he V4 egion could no dis inguish be ween closely ela ed species in se e al ins ances, as ollows. (i) S. pneumoniae and he less equen S. pseudopneumoniae; he la e is commonly (mis)iden ified as S. pneumoniae by clinical mic obiology labo a o ies wo ldwide. The sequence o he p ofiled V4 egion o he 16S RNA gene is also closely ela ed o hose o se e al o he s ep ococci. (ii) H. influenzae and he less equen H. haemoly icus; he la e could be excluded in cul u e-posi i e cases by i s be a hemolysis. (iii) M. ca a halis and he less equen Mo axella nonlique aciens; he wo could be di e en ia ed only by classic mic obiological echniques, including di e ences in ypical an ibiog ams. Finally, (i ) he 16S p ofiles in he V4 egion a e iden ical in many S aphylococci; please see he ex o me hods ha disen angled he signals. Sillanpää e al. Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 4 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om ec ed by 16S p ofiling, ye hey we e s ill posi i e in specific PCR es s (e.g., M. ca- a halis wi h a PCR signal beyond cycle 35). Fu he mo e, he use o a iable egion V4 o he 16S RNA gene did no allow de ailed de e mina ion o he species o se e al gene a o impo an pa hogens, so accu a e species iden ifica ion was in e ed om specific PCR es s we had pe o med be o e (12). DISCUSSION The p esen wo k demons a es ha only a limi ed epe oi e o bac e ia can be deemed esponsible o he majo i y o pedia ic AOM cases. The h ee main causa i e TABLE 1 Bac e ia ound in he 16S p ofiles Finding in he bac e ial p ofile a No. o samples posi i e o he species (nⴝ90) No. o posi i e samples % o all samples S ep ococcus pneumoniae 28 31 As a dominan pa hogen b 14 16 Sole finding bu ⬍50% o signal c 3 3.3 Nondominan pa o mixed flo a d 11 12 Haemophilus influenza 24 27 As a dominan pa hogen 15 17 Sole finding bu ⬍50% o signal 3 3.3 Nondominan pa o mixed flo a 6 6.7 Mo axella ca a halis 18 20 As a dominan pa hogen 5 5.6 Sole finding bu ⬍50% o signal 4 4.4 Nondominan pa o mixed flo a 9 10 S aphylococcus spp. 21 23 As a dominan pa hogen 3 3.3 Sole finding bu ⬍50% o signal 6 6.7 Nondominan pa o mixed flo a 12 13 Tu icella o i idis 5 5.6 As a dominan pa hogen 2 2.2 Sole finding bu ⬍50% o signal 1 1.1 Nondominan pa o mixed flo a 2 2.2 Alloiococcus o i idis 3 3.3 As a dominan pa hogen 0 0 Sole finding bu ⬍50% o signal 0 0 Nondominan pa o mixed flo a 3 3.3 O he bac e ia no lis ed abo e 14 16 As a dominan pa hogen 0 0 Sole finding bu ⬍50% o signal 3 e 3.3 Nondominan pa o mixed flo a 11 12 No clea bac e ial finding 14 16 No bac e ium ound 11 12 Unde e mined species, ⬍5% o signal 3 3.3 a The bac e ia o igina ing om he PCR componen s (Taq polyme ase) a e no shown. b A dominan pa hogen was defined as a bac e ium ha makes up hal o mo e o he o al 16S RNA gene p ofile. c Bac e ium occupying 3.0 o 49% o he sequencing signal; no o he bac e ia we e de ec able o e he h eshold 3.0% signal excep he con aminan signal om Taq polyme ase. d Bac e ium occupying 3 o 49% o he sequencing signal; also, o he bac e ia we e p esen in he p ofile a ⬎3.0%. e P e o ella melaninogenica (4% in sample om pa ien 41 and 6% in sample om pa ien 48) and unde e mined Sphingobac e ium (8% in pa ien 72). All h ee samples we e aken by my ingo omy. P e o ella melaninogenica (31% o he p ofile o sample om pa ien 73, 3% in pa ien 22, and 4% in pa ien 50), Veillonella dispa (20% in pa ien 73 and 10% in pa ien 72), Veillonella mon pellie ensis (13% in pa ien 67, 6% in pa ien 73, and 4% in pa ien 2), Lac ococcus lac is (7% in pa ien 15 and 7% in second sample om pa ien 59), Co ynebac e ium ube culos ea icum (4% om pa ien 50 and 3% om pa ien 70, bo h in samples wi h dominan S. pneumoniae), and unde e mined Sphingobac e ium. Bac e iome P ofiling in Acu e O i is Media Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 5 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om bac e ia, S. pneumoniae,H. influenzae, and M. ca a halis, a e complemen ed by he less p e alen T. o i idis,A. o i idis, and S. au icula is. The bac e iome p ofiling se s he bac e ia in a mu ual quan i a i e con ex and, hus, con as s species wi h a likely e iological con ibu ion o o he species p esen in much smalle quan i ies. Al hough we obse ed no no el dominan causa i e agen s, ou esul s may con- ibu e o he knowledge o species only a ely seen in AOM. Fi s , T. o i idis was a clea ly dominan pa hogen in wo samples and he sole abundan bac e ium in ano he sample. So a , he discussion o whe he his agen is a colonize om he ou e ea canal o a causa i e pa hogen has no been esol ed (16)—ou esul s sugges ha in some AOM cases, i could be a ue causa i e agen . Second, ano he such causa i e agen may be S. au icula is, which was p esen in high abundance in wo samples in child en wi hou ympanos omy ubes. The bac e ium was fi s desc ibed in 1983 by Kloos and Schlei e (17), who ound i o be an impo an componen o he flo a o he ex e nal audi o y canal. Despi e se e al anecdo al epo s on se e e in ec ions wi h his bac e ium (e.g., see e e ences 18 and 19), i has been mos ly ega ded only as a cause o o i is ex e na (20). Al hough we canno exclude ha bo h o ou findings o S. au icula is ep esen massi e inad e en con amina ions om he ex e nal audi o y canal, he la ge quan i y in he MEF may a he poin owa d genuine eplica ion in he middle ea . FIG 2 Quan i ies o he bac e ia assessed using he p opo ions wi hin he indi idual sample p ofiles. Negs, coun o samples ha we e nega i e o he bac e ium in he 16S p ofiling, i.e., had a quan i y lowe han 3% o he p ofile signal. Sillanpää e al. Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 6 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om The p opo ion o A. o i idis in ou se o AOM samples was lowe han in some o he ea lie s udies, and his o ganism’s signal was seen exclusi ely as pa o polymic obial in ec ions. The bac e ium was he mos p e alen one in he 16S RNA gene p ofile in one sample only, bu e en hen, wo o he o ganisms we e clea ly p esen . Al hough A. o i idis has a ac ed conside able a en ion in he li e a u e since i was fi s epo ed in o i is media (21), we a e a aid ha some o he la e epo s may ha e o e es ima ed i s equency and abundance, possibly due o echnical limi a ions o he PCR assays used he ein. While ea lie wo ks co ec ly u ilized molecula assays ha confi med he co ec ness o he p oduc by accu a ely de e mining i s leng h on polyac ylamide gels o e ified i s cha ac e by using mel ing analysis, sequencing, o liga ion eac ion (5, 22, 23), some o he la e s udies elied solely only on aga ose gel elec opho esis wi hou any e ifica ion o he inne sequence o he p oduc (3, 4, 24–27); some o hese de ec ion esul s migh be inco ec . Unde less s ingen condi ions, he p ime s (22) end owa d c oss- eac ion wi h human DNA, yielding a agmen o 238 bases (ch 2: 233,742,816–233,743,053 in he GRCh38 assembly), whose size is usually indis inguish- able om he amplicon size o A. o i idis (261 bases) in aga ose gel elec opho esis. A p e ious s udy by Smi h-Vaughan e al. has sugges ed ha he summed abun- dance o h ee majo pa hogens de ec ed by specific PCRs in AOM is much lowe han he o al bac e ial load es ima ed by ano he assay (7). The au ho s called o he use o bac e iome-p ofiling me hods as a possible solu ion o he disc epancy, because hey assumed ha a la ge p opo ion o he o al bac e ial load migh be comp ised o as-ye -uniden ified bac e ia. Mos likely, his is no he case. Fi s , as he au ho s s a ed, ins ead o using MEF samples collec ed unde s e ile condi ions, hey had o eso o es ing ea discha ge samples, which may ha e been s ongly con amina ed wi h ea FIG 3 Compa ison o de ec ion by specific eal- ime PCR and by 16S p ofiling. Ho izon al axis, h eshold cycle o he espec i e specific PCR; e ical axis, p opo ion o he o e all p ofiling signal wi hin he sample. No e ha he V4 sequence o S ep ococcus pneumoniae is nea ly iden ical o hose o se e al u he s ep ococci (e.g., S ep ococcus den isani,S ep ococcus igu inus,S ep ococcus o alis, S ep ococcus mi is, and S ep ococcus in an is). These a e mos likely p esen in se e al samples, deno ed by c osses along he e ical axis o he op le panel: he e, he pneumococcus-specific eal- ime PCR es using he au olysin gene (ly A) is nega i e, bu he weak signal in 16S p ofiling indica es he p esence o hese s ep ococci. Bac e iome P ofiling in Acu e O i is Media Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 7 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om canal flo a. Second, he di e ence may be an a i ac caused by, e.g., he mu ual di e ence in e ficacies o amplifica ion o he ou p ime pai s used o specific PCR de ec ion. Finally, he signal om he ecombinan polyme ase migh infla e he o al bac e ial load in low-abundance samples. In con as , ou app oach o e s qui e accu- a e cha ac e iza ion o he 16S p ofiles, because he se o con aminan species ha o igina ed om Taq polyme ase has been cha ac e ized and sub ac ed, inciden ally p o iding a kind o in e nal quan i a i e s anda d. Thus, he p esen s udy was able o cha ac e ize he whole spec um o bac e ia, and s ill, he p e iously desc ibed pa ho- gens cons i u ed he majo i y o he p ofiles wi hou any suppo o he exis ence o a majo unknown bac e ial causa i e agen . The sensi i i y o 16S RNA gene p ofiling s ands be ween he low sensi i i y o cul u e and he high sensi i i y o specific PCR assays a ge ed o indi idual o ganisms. This can be seen om he esul s o he p esen s udy and is especially p onounced o M. ca a halis. He e, he sensi i i y o specific PCR clea ly supe seded ha o 16S p ofiling. Rega ding heo e ical conside a ions o he po en ial o 16S p ofiling as a diagnos ic ool, al hough i is a he sensi i e in he p esen modifica ion, i could no be elied upon as a p ima y diagnos ic ool in a si ua ion whe e he disease is mos ly caused by a limi ed se o se e al well-known agen s wi h a ailable specific PCR es s. We also obse ed many small-quan i y findings in he 16S p ofiles whe e i has no been cla ified whe he hey eflec ue biological significance o he agen : indeed, he DNA p ofiling may de ec dead bac e ia, al hough s udies in chinchillas ha e shown ha bac e ial DNA in MEF disappea s wi hin 3 days o bac e ial cell dea h (28). Mo eo e , innocuous bac e ia may passi ely en e he middle ea du ing i al in ec ion and can be de ec ed in small quan i ies du ing AOM caused by i al pa hogens. In addi ion, some o he bac e ia de ec ed wi h sequencing migh eflec he no mal flo a o he middle ea (15). In ins ances wi h posi i e cul u e, he iden ified o ganisms we e de ec ed by 16S RNA gene p ofiling as well, wi h he single excep ion o a sample ha was posi i e o S. pneumoniae by cul u e and specific PCR. Tha sample was nega i e o S. pneumoniae in he 16S p ofile, while h ee o he o ganisms we e de ec ed. The li e a u e on 16S p ofiling in o i is media is scan y and limi ed o ch onic o i is media only: Liu e al. analyzed one pa ien wi h ch onic o i is media (13), and Je is- Ba dy e al. analyzed 11 child en wi h o i is media wi h e usion using one ound o PCR and ailed o ob ain su ficien signals in hal o he samples (14). Recen ly, Nee e al. u ilized nes ed PCR in he cha ac e iza ion o bac e iome p ofiles in 24 p edominan ly adul pa ien s wi h ch onic suppu a i e o i is media (71% wi h choles ea oma) and 22 heal hy con ol ea s (15). Thus, he p esen s udy is a significan con ibu ion o he field, being he la ges 16S p ofiling s udy o o i is media and he only one o AOM so a . We ob ained a clea 16S signal om he majo i y o MEF samples and we e able o confi m mos o ou findings by specific PCR. Toge he wi h he wo k by Nee e al. (15), we demons a ed ha nes ed PCR is necessa y o 16S amplifica ion o MEF. While Nee e al. e mina ed he nes ed amplifica ion be o e a alse signal om he Taq polyme ase could eme ge, we used much longe amplifica ion wi h ully de eloped alse signal in nega i e con ols, which was hen sub ac ed om he signals o eal samples. This new modifica ion sol ed he well-known p oblem o samples wi h low bac e ial con en — he ibosomal nucleic acid coming om he Taq polyme ase in he PCRs compe es wi h he ue signal coming om he sample (29, 30). We op imized his me hod o achie e he lowes possible le el o con amina ing backg ound signal using a s a egy inspi ed by Spangle e al. (31), and simila o hei esul s, he op imal chemis y ound was Ho S a polyme ase (Qiagen). An addi ional ad an age o com- pu a ional sub ac ion o he polyme ase-de i ed signal is ha a s able low le el o con amina ion se es as an excellen in e nal con ol in 16S p ofiling. The ela i e quan i y o any ea bac e ium may hen se e as a guide o indica e he ele ance o a pa hogen. The middle ea is s e ile acco ding o specific PCR es s and bac e ial cul u es (32). Howe e , Nee e al. de ec ed bac e ia wi h mass sequencing o samples om heal hy adul s’ middle ea s in up o 43% o cases (15). They showed ha bac e ial loads in he Sillanpää e al. Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 8 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om heal hy middle ea and mas oid ca i y a e low. They de ec ed small amoun s o species om gene a like No osphingobium,S aphylococcus,S ep ococcus,Esche ichia-Shigella, and Bu kholde ia. O hose bac e ia, only S aphylococcus species we e de ec ed in ou s udy, and in 20% o he samples, i was seen in low abundance, which migh eflec i s ole as pa o he no mal mic obio a o he middle ea o as a con aminan om he ou e ea canal. These s udies, howe e , canno answe ques ions abou no mal pedia ic middle ea flo a. This s udy has se e al impo an echnological s eng hs. Fi s , ou p o ocol was able o p o ide eliable signals om he majo i y o he 90 samples, including he low- abundance samples. The use o iplica e eac ions helped subs an ially o ensu e a homogenous signal, which could hen be used o p ofiling. Because o he ela i e quan ifica ion ha is a cha ac e is ic o he 16S p ofiling, we we e able o dis inguish dominan pa hogens p esen in la ge quan i ies om he signals o bac e ia p esen in small quan i ies whose pa hogenic pa icipa ion migh be unlikely. The exis ence o a s able backg ound a ising om he RNA gene con amina ing he ecombinan poly- me ase se ed as an exogenous in e nal con ol o amplifica ion and a compe i i e PCR a ge . We e ified and confi med all p ominen signals om he 16S p ofiling by means o specific eal- ime PCR om he o iginal samples. In mos cases, we used ye ano he le el o e ifica ion by pe o ming Sange sequencing o he ensuing PCR p oduc s. This deg ee o ce ain y is, o ou knowledge, a he excep ional among s udies o bac e ia in AOM. One limi a ion o he desc ibed me hod is i s complica ed p o ocol: o clinical use, a conside able daily coun o samples would be needed o make he 16S sequencing cos e ec i e. This gene ally ende s he me hod unsui able o ou ine clinical p ac ice, whe e ea ly an ibio ic esponse is desi able. In addi ion, he V4 egion o he 16S RNA gene canno disc imina e exac species o se e al clinically ele an o opa hogens. Namely, Haemophilus influenzae has a V4 sequence iden ical o ha o H. haemoly icus, which is nonencapsula ed (i.e., non ypeable) and gene ally conside ed nonpa hogenic. Fu he mo e, S ep ococcus pneumoniae has a V4 sequence iden ical o ha o S ep- ococcus pseudopneumoniae, and e en mo e impo an ly, nea ly iden ical o hose o se e al o he s ep ococci, and he e o e, he pneumococcus-specific PCR es s, includ- ing ou s, ha e long used he au olysin gene (ly A) a he han he poo ly in o ma i e 16S RNA gene. Also, nume ous species o he genus S aphylococcus a e indisc imina e in he V4 egion. Specific PCR assays a e hus needed o such agen s, bu o su p isingly many o ganisms, including hose o po en ial clinical ele ance, no pub- lished p ime sequences a e known. In conclusion, ou wo k has shown he composi ion o mic obial middle ea flo a in AOM in child en, excluded he possibili y o a la ge gap be ween he known agen s and he o al bac e ial load, and demons a ed ha 16S p ofiling by mass sequencing can con e in o ma ion ha dly ob ainable by o he me hods. MATERIALS AND METHODS Pa ien s and hei MEF samples. Child en we e en olled a he Depa men o O o hinola yngology, Tampe e Uni e si y Hospi al, Tampe e, Finland, be ween Sep embe 2010 and Decembe 2011. The diagnosis o AOM was based on he p esence o MEF wi h signs o inflamma ion o he ympanic memb ane, o al e na i ely, o o hea h ough a ympanos omy ube o a spon aneous pe o a ion o he ympanic memb ane and symp oms o acu e espi a o y in ec ion. The MEF specimens we e collec ed a e my ingo omy wi h a s e ile suc ion ip. In child en wi h ympanos omy ubes o spon aneous pe o a ions o he ea d um, MEF specimens we e ob ained om he middle ea by suc ion. The sample se was iden ical o wha had been desc ibed in ou p e ious s udy (12). The s udy p o ocol was app o ed by he E hical Commi ee o he Tampe e Uni e si y Hospi al ( e e ence numbe R10026), and w i en in o med consen was ob ained om all pa icipa ing amilies. P ocessing o he samples, cul u e, and pa hogen-specific PCR. The wo kflow diag am o sample p ocessing is shown in Fig. 4. One aliquo o each MEF sample was ob ained o bac e ial cul u e, and ano he aliquo was immedia ely ozen and s o ed a ⫺70°C un il DNA ex ac ion, specific PCR es s, and nes ed 16S RNA gene mass-sequencing p ofiling we e pe o med. Bac e ial cul u ing, he ex ac ion o nucleic acids, and pa hogen-specific PCR o six candida e pa hogens (H. influenzae,A. o i idis,M. ca- a halis,S. pneumoniae,Pseudomonas ae uginosa, and S aphylococcus au eus) ha e been desc ibed p e iously (12). Bac e ial cul u es we e pe o med ae obically. The esul was lis ed as nega i e i no bac e ial g ow h was seen. O he flo a consis ed o a ypical bac e ia ound in minu e quan i ies. In ou Bac e iome P ofiling in Acu e O i is Media Ma ch/Ap il 2017 Volume 2 Issue 2 e00006-17 msphe e.asm.o g 9 on May 3, 2017 by gues h p://msphe e.asm.o g/Downloaded om