Optimised PDMS Tunnel Devices on MEAs Increase the Probability of Detecting Electrical Activity from Human Stem Cell-Derived Neuronal Networks

ORIGINAL RESEARCH published: 31 Oc obe 2017 doi: 10.3389/ nins.2017.00606 F on ie s in Neu oscience | www. on ie sin.o g 1Oc obe 2017 | Volume 11 | A icle 606 Edi ed by: Michele Giugliano, Uni e si y o An we p, Belgium Re iewed by: Anja Kunze, Mon ana S a e Uni e si y, Uni ed S a es Sophie Pau o , USR3505 Ins i u des Technologies A ancées en sciences du Vi an (ITAV), F ance Ma c Heuschkel, High School o Landscape Enginee ing and A chi ec u e, Uni e si y o Applied Sciences Wes e n Swi ze land, Swi ze land *Co espondence: Me i Ris ola [email p o ec ed] Susanna Na kilah i [email p o ec ed] †These au ho s ha e con ibu ed equally o his wo k. Special y sec ion: This a icle was submi ed o Neu al Technology, a sec ion o he jou nal F on ie s in Neu oscience Recei ed: 25 July 2017 Accep ed: 17 Oc obe 2017 Published: 31 Oc obe 2017 Ci a ion: Toi anen M, Pelkonen A, Mäkinen M, Ylä-Ou inen L, Sukki L, Kallio P, Ris ola M and Na kilah i S (2017) Op imised PDMS Tunnel De ices on MEAs Inc ease he P obabili y o De ec ing Elec ical Ac i i y om Human S em Cell-De i ed Neu onal Ne wo ks. F on . Neu osci. 11:606. doi: 10.3389/ nins.2017.00606 Op imised PDMS Tunnel De ices on MEAs Inc ease he P obabili y o De ec ing Elec ical Ac i i y om Human S em Cell-De i ed Neu onal Ne wo ks Ma ia Toi anen1†, Anssi Pelkonen1†, Mee i Mäkinen1, Lau a Ylä-Ou inen1, Lassi Sukki2, Pasi Kallio2, Me i Ris ola 1*†and Susanna Na kilah i 1*† 1Neu oG oup, BioMediTech Ins i u e and Facul y o Medicine and Biosciences, Uni e si y o Tampe e, Tampe e, Finland, 2Mic o and Nanosys ems Resea ch G oup, BioMediTech Ins i u e and Facul y o Biomedical Sciences and Enginee ing, Tampe e Uni e si y o Technology, Tampe e, Finland Measu emen o he ac i i y o human plu ipo en s em cell (hPSC)-de i ed neu onal ne wo ks wi h mic oelec ode a ays (MEAs) plays an impo an ole in unc ional in i o b ain modelling and in neu o oxicological sc eening. The p e iously epo ed hPSC-de i ed neu onal ne wo ks do no , howe e , exhibi epea able, s able unc ional ne wo k cha ac e is ics simila o oden co ical cul u es, making he in e p e a ion o esul s di icul . In ea lie s udies, mic o unnels ha e been used bo h o con ol and guide cell g ow h and ampli y he axonal signals o oden neu ons. The aim o he cu en s udy was o de elop unnel de ices ha would acili a e signalling and/o signal de ec ion in en i e hPSC-de i ed neu onal ne wo ks con aining no only axons, bu also soma a and dend i es. The e o e, MEA-compa ible polydime hylsiloxane (PDMS) unnel de ices wi h 8 di e en dimensions we e c ea ed. The hPSC-de i ed neu ons we e cul u ed in he unnel de ices on MEAs, and he spon aneous elec ical ac i i y o he ne wo ks was measu ed o 5 weeks. Al hough he unnel de ices imp o ed he signal- o-noise a io only by 1.3- old a bes , hey signi ican ly inc eased he pe cen age o elec odes de ec ing neu onal ac i i y (52–100%) compa ed wi h he con ols (27%). Signi ican ly highe spike and bu s coun s we e also ob ained using he unnel de ices. Neu onal ne wo ks inside he unnels we e amenable o pha macological manipula ion. The esul s sugges ha unnel de ices encompassing he en i e neu onal ne wo k can inc ease he measu ed spon aneous ac i i y in hPSC-de i ed neu onal ne wo ks on MEAs. The e o e, hey can inc ease he e iciency o unc ional s udies o hPSC-de i ed ne wo ks on MEAs. Keywo ds: human plu ipo en s em cells, mic oelec ode a ay, neu onal ne wo k, unnel de ice, in i o model INTRODUCTION Analysis o neu onal ne wo k ac i i y in i o is a pi o al pa o mode n b ain disease modelling, neu opha macological es ing, and neu o oxicological sc eening (Johns one e al., 2010; Valdi ia e al., 2014). In i o neu onal ne wo ks de i ed om human plu ipo en s em cells (hPSCs) can eplace animal-de i ed models and be e p edic esponses in humans (Ca anaugh e al., 2014; Hunsbe ge e al., 2015; Pei e al., 2016). Fu he mo e, hei ac i i y can be measu ed using Toi anen e al. PDMS Tunnels Inc ease MEA Ac i i y mic oelec ode a ays (MEAs) (Johns one e al., 2010; Jones e al., 2011; Mose , 2011). Fo example, he e ec o a neu o oxin on he MEA-ac i i y o a hPSC-de i ed ne wo k can be obse ed be o e any mo phological changes (Ylä-Ou inen e al., 2010). Howe e , analyses o MEA da a om hPSC-de i ed ne wo ks can be e y challenging due o low pe cen age o elec odes (o en <20%), o en i e a ays, ha de ec neu onal ac i i y (Ylä-Ou inen e al., 2010; Tukke e al., 2016). E en when he hPSC-de i ed ne wo ks p oduce obus , measu able ac i i y, he necessa y di e en ia ion and unc ional de elopmen can be e y slow, aking up o se e al mon hs (Odawa a e al., 2016). The a iable and slow de elopmen o neu onal ne wo k ac i i y on MEA appea s o be cha ac e is ic o all neu onal cul u es o human o igin; in p ima y human neu ons, which do no equi e p e-di e en ia ion, he eme gence o elec ical ac i i y can ake nea ly 40 days, whe e as in co esponding a neu ons he same de elopmen happens in only 10 days (Napoli and Obeid, 2016). The e o e, i is clea ha new app oaches a e needed o acili a e he analysis o hPSC-de i ed neu onal ne wo k unc ions. Mic oenginee ed polydime hylsiloxane (PDMS) de ices can be used o answe speci ic ques ions on mechanisms o neu al unc ion and pa hology (Taylo e al., 2005; Sco e al., 2013; Ren e al., 2015), and hey can also be used o acili a e he analysis o elec ical unc ion o hPSC-de i ed ne wo ks. PDMS de ices consis ing an open chambe (o “well”) can guide he ne wo k o g ow mo e densely on op o he measu ing elec odes, and hus acili a e he de elopmen and de ec ion o neu onal ac i i y (K eu ze e al., 2012). PDMS mic o unnel de ices, on he o he hand, inc ease he de ec ed ac i i y by ampli ying ex acellula elec ical signals de ec ed by he MEA (Fi zGe ald e al., 2009; Wie inga e al., 2010; Wang e al., 2012). Acco ding o a gene ally accep ed heo y in he ield his occu s because signal ampli udes measu ed by MEA in unnels a e in luenced by a de i a i e o Ohm’s law (U =R I), whe e he esis ance o he medium inside he unnel inc eases as he unnels heigh (h) and wid h (w), i.e., c oss sec ion (A) dec eases and leng h (l) inc eases (R =ρ(l / A)). The inc eased esis ance, in u n, mani es s as highe po en ial di e ences du ing elec ical ac i i y o he measu ed cells, which ansla es o an inc eased signal- o-noise a io (SNR) in he MEA eco dings. Howe e , he mic o unnels p o iding he bes ampli ica ion (c oss sec ions ≤100 µm2,h ≤5µm) a e designed o encompass only neu i es and no neu onal soma a o en i e neu onal ne wo ks, and o en equi e a cus om made elec ode a ay (Fi zGe ald e al., 2008; Dwo ak and Wheele , 2009; Hong e al., 2017). These neu i e-encompassing mic o unnels a e use ul o analysing ce ain pa ame e s such as he speed o signal conduc ion along axons, bu he e is a need o la ge unnel de ices which can p o ide obus MEA da a om en i e neu onal ne wo ks con aining also dend i es and cell soma a. In his s udy, he objec i e was o de elop unnel de ices ha a e compa ible wi h a comme cially a ailable MEA pla o m, and a e able o house en i e hPSC-de i ed neu onal ne wo ks and concomi an ly possess su icien ly small ea u es o ampli y he ex acellula signals on MEAs in compa ison o s anda d cul u es. The e o e, hPSC-de i ed neu onal ne wo ks we e cul u ed on MEAs in unnel de ices wi h di e en dimensions. The spon aneous elec ical ac i i y o he neu onal ne wo ks in he unnels was measu ed up o 5 weeks and compa ed o da a om s anda d MEA con ols. We obse ed ha while he unnels p o ided li le o no imp o emen o signal de ec ion, hey inc eased he measu ed ne wo k ac i i y conside ably. Thus, he use o he unnel de ices sol ed one o he main p oblems in s udying hPSC-de i ed ne wo ks using MEAs, which is he low pe cen age o ac i e elec odes. MATERIALS AND METHODS P oduc ion o MEA-Compa ible PDMS Tunnel De ices Cus om PDMS unnel de ices and SU-8 moulds o he de ices we e ab ica ed using apid p o o yping me hods (Du y e al., 1998). The ou e diame e o he PDMS de ices (Figu e 1A) was 15 mm and he heigh was app oxima ely 3 mm o be compa ible wi h he MEA a ay (60MEA200/30iR-Ti MEAs, Mul iChannelSys em [MCS], Ge many) and ampli ie (MEA2100, MCS). The PDMS de ices con ained wo cell pla ing a eas ha we e in e connec ed by unnels and a e e ence elec ode well. The designed unnel dimensions we e a ied as p esen ed in Table 1 and Figu e 1. The cell cul i a ion a ea in on o he unnels is co e ed by a PDMS lid. The di e en PDMS unnels we e aligned on op o he MEA elec odes as illus a ed in Figu es 1B–F. PDMS (Sylga d 184, Dow Co ning) de ices we e ab ica ed using me hods desc ibed by Pa k e al. (2006). The moulds we e ab ica ed om SU-8 3050 (Mic o Resis Technology GmbH) on op o a silicon wa e . A 15-mm-diame e punching ool was used o punch indi idual de ices ou o PDMS shee s. A 3-mm-diame e manual punching ool was used o c ea e inle s (cell pla ing a eas; Figu e 1A) a a dis ance o 100– 1,000 µm om he unnel mou h, and an opening o he e e ence elec ode. The use o ci cula cell supply inle s and he need o ha e equal unnel leng hs c ea ed a lid a ound he punching hole. Thus, he MEA elec odes placed in on o he unnels we e co e ed by he PDMS lid loca ed ei he 43 o 105 µm abo e he MEA su ace depending on he unnel heigh . He ea e , hese elec odes loca ed unde nea h he PDMS lid bu ou side he unnels a e e e ed as ou side elec odes. The dimensions o he unnel de ices we e cha ac e ised using bo h ligh mic oscopy and p o ilome y. A Zeiss Axio Image .A1m (Ca l Zeiss AG) was used o inspec he mould o po en ial aul s. A B uke Dek ak XT s ylus p o ilome e (B uke Co po a ion) was used o measu e he heigh s o he mic os uc u es om he mould. Acco ding o he measu emen s, he mould heigh s we e 43 ±7µm and 105 ± 15 µm. The a ia ion in heigh s was caused by he sligh bending o he silicon wa e s by he spinne acuum du ing spin-coa ing, which caused he ea u es o be hicke in he middle o he moulds. F on ie s in Neu oscience | www. on ie sin.o g 2Oc obe 2017 | Volume 11 | A icle 606 Toi anen e al. PDMS Tunnels Inc ease MEA Ac i i y TABLE 1 | Tunnel dimensions and numbe o unnels, elec odes and MEAs. w(µm) h(µm) l(µm) Numbe o MEA pla es Elec odes pe unnel, ou side a ea o MEA well To al no analysed unnels, ou side a eas o MEA wells To al no analysed elec odes To al no ac i e elec odes pe week 12345 6-well MEA con ol 5 – 9 30 – – 270 – 75 107 120 146 101 1-well MEA con ol 4 – 59 4 – – 236 – – 1 18 39 42 Ou side 43 – 13 8 12 13 8 – 156 96 29 87 112 119 118 Ou side 105 – 21 15 12 21 15 – 254 180 9 74 150 157 154 100 43 1,000 5 4 4 6 4 5 24 16 13 18 24 24 24 100 105 1,000 7 5 4 9 5 7 36 20 2 14 24 34 31 500 43 1,000 1 – 11–12 2 – 2 23 – 3 2 12 21 23 500 105 1,000 2 – 11–12 4 – 3 46 – 1 6 34 40 44 750 43 1,000 3 2 15–16 4 2 2 63/32 8 24 40 47 48 750 105 1,000 7 5 15–16 9 5 6 142/80 7 48 71 92 91 1,500 43 1,000 5 2 31 5 2 4 155/62 11 69 127 131 127 1,500 105 1,000 7 5 31 7 5 3 217 155 4 39 167 111 137 The unnel wid h (w), heigh (h) and leng h (l) a e p esen ed in he i s h ee columns. “Ou side” e e s o elec odes ou side he ac ual unnels bu unde nea h he PDMS lid (Figu e 1). The second numbe a e he backslash indica es he co esponding numbe in he pha macological expe imen (Figu e 6), and he hi d numbe a e he second backslash indica es n in he Cy oSpec e ne wo k o ien a ion analysis (Figu e 3E). P epa a ion o PDMS Tunnel De ices and MEAs o Cell Cul u e The MEAs we e always cleaned be o e use acco ding o manu ac u e ’s ins uc ions (washed wi h 1% Te gazyme [Sigma- Ald ich], insed wi h dis illed H2O and au ocla ed). MEAs we e coa ed wi h 0.05% polye hylenimine as p e iously desc ibed (Ylä-Ou inen e al., 2010). To make he unnels hyd ophilic and hus amenable o coa ing, he PDMS de ices we e ea ed wi h oxygen plasma in a PICO plasma sys em (Diene elec onic) o 3 min a 50 W. They we e manually aligned unde a mic oscope on he MEA elec odes and e e sibly bonded o he MEAs, i.e., hey could s ill be manually emo ed. Mouse laminin (20 µg/ml; Sigma-Ald ich) was pipe ed in o he PDMS unnel de ices on MEAs h ough bo h cell pla ing a eas (Figu e 1A). Cell cul u e con ol pla es (4-well pla e, Nunc, The mo Fishe Scien i ic, Inc.) we e coa ed wi h 20 µg/ml o 10 µg/ml mouse laminin in wells wi h o wi hou co e slips (Ø =13 mm, VWR), espec i ely. The MEAs and he con ol pla es we e incuba ed wi h he laminin solu ions a +4◦C o e nigh as p e iously desc ibed (Ylä-Ou inen e al., 2010). Neu al Di e en ia ion and Cell Cul u e The human emb yonic s em cell (hESC) line Regea 08/023 and he human induced plu ipo en s em cell (hiPSC) line 04311.WT we e used in he expe imen s. BioMediTech has app o al om he Finnish Medicines Agency (FIMEA) o pe o m esea ch wi h human emb yos (Dn o 1426/32/300/05). The e a e also suppo i e s a emen s om he egional e hical commi ee o Pi kanmaa Hospi al Dis ic o he de i a ion, cul u ing, and di e en ia ion o hESCs (R05116) and hiPSCs (R08070). This s udy was ca ied ou in acco dance wi h he ecommenda ions o FIMEA and Pi kanmaa Hospi al Dis ic wi h w i en in o med consen om all subjec s who p o ided cell ma e ial. All subjec s ga e w i en in o med consen in acco dance wi h he Decla a ion o Helsinki. The imeline o he expe imen s is shown in Figu e 2. The hESCs and hiPSCs we e di e en ia ed in o neu al cells o 8–10 weeks in neu osphe e cul u es in di e en ia ion medium (NDM) consis ing o 1:1 Dulbecco’s Modi ied Eagle’s Medium/F12:Neu obasal Medium supplemen ed wi h 2 mM Glu aMax, 1x B27 supplemen , 1x N2 supplemen (all om Gibco In i ogen), 25 U/ml penicillin/s ep omycin (Lonza G oup L d) and, in his neu osphe e di e en ia ion s age, 20 ng/ml basic ib oblas g ow h ac o (bFGF, R&D Sys ems) as p e iously desc ibed (Lappalainen e al., 2010) wi h o wi hou low-dose nal exone LDN193189 (100 nM; S emcell Technologies, Inc.). The neu osphe es con aining p e-di e en ia ed neu al cells we e manually dissec ed in o small cell agg ega es (Ø∼50– 200 µm). App oxima ely 15 small agg ega es (50,000–150,000 cells in o al) we e pla ed in bo h cell pla ing a eas o he PDMS de ices (i.e., bo h ends o he unnels) as close o he unnels as possible o ensu e unnel coloniza ion, and simila ly on con ol pla e wells. The pla ing p ocedu e was iden ical o each de ice. The PDMS de ices on he MEAs we e subme ged in he cell cul u e medium (1 ml). The cells we e main ained in a humidi ied incuba o a 37◦C and 5% CO2. Hal o he medium was changed h ee imes a week. A e 1 week in adhe en cul u e, 4 ng/ml bFGF and 5 ng/ml b ain-de i ed neu o ophic ac o (BDNF) (Gibco In i ogen) we e added o he medium. Immunocy ochemis y The con ol cells we e ixed a e 14 days in adhe en cul u e, and immunocy ochemical (ICC) s aining was pe o med as p e iously desc ibed (Lappalainen e al., 2010) o e i y he neu al iden i y o he cells. P ima y an ibodies, abbi an i-be a-III Tubulin (β- ub) (1:2000; GenSc ip ) and abbi F on ie s in Neu oscience | www. on ie sin.o g 3Oc obe 2017 | Volume 11 | A icle 606 Toi anen e al. PDMS Tunnels Inc ease MEA Ac i i y FIGURE 1 | Designs o he unnel de ices. The PDMS unnel de ices (A) consis o wo cell pla ing a eas ha a e in e connec ed by unnels, and a e e ence elec ode well. The wo unnel heigh s a e p esen ed in he lowe igh co ne in (A). The ed dashed lines indica e he c oss sec ion o a unnel, in his case 100 µm wide. The numbe s in (A) p o ide he unnel wid hs (w) and heigh s (h) in µm. Tunnels wi h di e en dimensions we e aligned on MEAs (B–F). Da k g ay indica es PDMS bonded o he MEA su ace, ligh g ey ep esen s cell cul i a ion a eas in he unnels and a eas unde he PDMS lid, while he ed dashed lines indica e unnel pe ime e s. The numbe s in (B–F) p o ide he unnel wid hs (w) and leng hs (l) in µm. FIGURE 2 | Timeline o he expe imen s. Neu ons we e di e en ia ed in neu osphe e cul u e o 8–10 weeks be o e adhe en cul u e in unnel de ices on MEAs and in s anda d cell cul u e wells (con ol cells). MEA measu emen s we e pe o med wice weekly o e 5 weeks, and a pha macological es was pe o med a he end o he cul u e. Neu i e o ien a ion analysis was pe o med om he phase con as images aken a he 5 h week. Immunocy ochemis y was pe o med on con ol cells a e 2 weeks in adhe en cul u e o bo h cell lines. polyclonal an i-Mic o ubule-Associa ed P o ein 2 (MAP2) (1:400; Millipo e), we e used oge he wi h seconda y an ibodies, Alexa 488 an i- abbi and Alexa 568 an i-mouse (bo h 1:400; Molecula P obes). In addi ion, he nuclei o he cells we e s ained wi h 4′,6-diamidino-2 phenylindole (DAPI), which was included in he moun ing medium (Vec ashield Moun ing Medium wi h DAPI, Vec o Labo a o ies). The cells we e imaged wi h a luo escence mic oscope (Olympus IX51, Olympus Co po a ion). Phase Con as Mic oscopy and Neu i e O ien a ion Analysis The cells we e imaged wi h a phase con as mic oscope (Nikon Eclipse TE2000-S, Nikon Co po a ion) once pe week o ollow he neu onal mo phology, mig a ion and ne wo k o ma ion. The cell cul u e con ol pla es we e used as a no mal mic oen i onmen cell con ol o e alua e neu onal iabili y, mo phology and mig a ion. The con ol pla es we e ollowed o 2 weeks. The o ien a ion o he neu i es was analysed using Cy oSpec e 1.2 so wa e (h p://www. u . i/cy ospec e) F on ie s in Neu oscience | www. on ie sin.o g 4Oc obe 2017 | Volume 11 | A icle 606 Toi anen e al. PDMS Tunnels Inc ease MEA Ac i i y (Ka asalo e al., 2015). The so wa e u ilises spec al analysis and calcula es he o ien a ions o image componen s, in his case neu i es (Hyysalo e al., 2017), and desc ibes hei a iance wi h a ci cula a iance alue, which is 1 when he componen s a e andomly aligned, and 0 when all componen s a e comple ely unidi ec ional. The so wa e was used in he mixed componen mode and spec al esolu ion/noise was se o balanced. Wa eleng h se ings (componen size) we e se o 1µm (minimum) and 30 µm (maximum). The o ien a ion o neu i es inside he unnels was analysed using phase con as images ob ained a week 5 a e pla ing, and compa ed o images o he eely g owing ne wo ks on cell cul u e con ol pla es. The elec odes and hei acks we e excluded om he image analysis by using a cus om MATLAB sc ip which au oma ically de ec ed he elec odes and he acks and eplaced hei pixel alues wi h local mean in ensi ies compu ed om he co esponding egions o each image, and by using Cy oSpec e’s componen size il e ing. MEA Measu emen s MEA measu emen s we e pe o med wi h an MEA sys em consis ing o a il e ampli ie MEA2100, so wa e MC_Rack and empe a u e con olle s TC02 se a 37◦C (all om MCS). The elec ical ac i i y o he neu onal ne wo ks was measu ed wice a week o 5 weeks (Figu e 2). The du a ion o each eco ding was 10 min and he sampling a e was 25 kHz. To analyse he noise, signal ampli ude and SNR in he unnel de ices, he MEA da a we e compa ed o ea lie eco dings om he same 60MEA200/30iR-Ti MEAs wi h no PDMS unnel de ices, e e ed o as he 1-well MEA con ol. The 1-well MEA con ol is an open olume sys em wi h no liquid olume es ic ions. To analyse he de elopmen o ne wo k ac i i y, he da a we e compa ed o eco dings om 60-6wellMEA200/30iR-Ti MEAs (MCS), in which he indi idual wells we e sepa a ed using SpikeBoos e de ices (BioMediTech) (K eu ze e al., 2012), and is e e ed o as he 6-well MEA con ol. The dimensions o he cell cul u ing a eas on he SpikeBoos e de ices a e he same as he cell pla ing a eas on he unnel de ices, and he 6-well MEA con ol can be conside ed a pa ially es ic ed olume sys em. The combina ion o 6-well MEAs and SpikeBoos e is he mos used MEA se up in ou labo a o y, and i ypically p o ides he bes ne wo k ac i i y de elopmen . All used MEAs had he same su ace ma e ial (Si3N4), elec ode ma e ial (TiN), elec ode diame e (30 µm) and elec ode- o-elec ode dis ance (200 µm). Pha macological es ing wi h e odo oxin (TTX; 1 µM, Toc is Bioscience) was pe o med a he end o he s udy (Figu e 2). MEA ac i i y was measu ed o 5 min a e addi ion o esh medium o he MEA and a e addi ion o TTX o he medium whe e he PDMS de ice was subme ged. TTX and he equipmen used o handling i we e s o ed, handled and disposed acco ding o ins i u ional sa e y egula ions (BioMediTech ins i u e and Facul y o Medicine and Biosciences, Uni e si y o Tampe e). Signal Analysis and S a is ics Spikes we e de ec ed om he MEA da a using MATLAB (The Ma hWo ks, Inc.) wi h a cus om-made analysis p og am based on Qui oga e al. (2004). Analysis was pe o med sepa a ely o each elec ode (modi ied om Qui oga e al., 2004). Fi s , he ol age signal was il e ed (200–3,000 Hz band pass). Nex , he noise was calcula ed as he median (md) o he absolu e alues om he il e ed eco ding di ided by 0.6745. Signal alues which exceeded i e imes his noise alue we e conside ed as spikes. Bo h nega i e and posi i e spikes we e de ec ed. Spikes la ge han 500 imes noise we e emo ed as a e ac s. Fo spike wa e o m analysis, 0.8 and 1.76 ms o ol age signal was clipped be o e and a e he la ges absolu e alue o he spike om he il e ed da a. The de ec o dead ime be ween wo wa e o ms was 1.48 ms. The peak- o-peak ampli udes we e measu ed as he di e ence be ween he highes and lowes ol age alues in he s o ed wa e o ms. A peak- o-peak md was ob ained om all wa e o ms om one channel o iden i y a single alue pe channel. SNR was calcula ed by di iding he md peak- o-peak spike ampli udes by he co esponding noise alues. An elec ode was ega ded as an ac i e elec ode (measu ing neu onal ac i i y) i mo e han 2 spikes we e eco ded in a minu e (spike equency 0.033 Hz). The h eshold was de e mined by measu ing he spike a es om emp y MEAs and MEAs wi h TTX-silenced neu onal cul u es (da a no shown). Pe cen age o ac i e elec odes was calcula ed o each unnel and con ol well sepa a ely and elec odes unde nea h he PDMS de ices we e excluded om he analysis. Da a om he elec odes a he unnels mou h (unde he ed dashed line in Figu es 1B–F) we e no included in analyses because hey could be conside ed nei he ou side no unnel elec odes. Bu s s (clus e s o spikes) we e de ec ed sepa a ely o each elec ode using a me hod based on Kapucu e al. (2012) which de ines bu s s using he cumula i e mo ing a e age o in e -spike in e als. The numbe o epea s (n) in di e en analyses a e p esen ed in Table 1. S a is ical analyses we e pe o med in SPSS (IBM). The MEA da a we e ound o ha e a non-no mal dis ibu ion, and he e o e he nonpa ame ic K uskal-Wallis es wi h Dunn’s pos hoc es was used o de e mine whe he he e we e s a is ically signi ican di e ences among he di e en unnels and con ols. The da a om he neu i e o ien a ion analysis (Cy oSpec e esul s) we e ound o be no mally dis ibu ed and hus we e analysed by uni a ia e analysis o a iance wi h Bon e oni’s pos hoc es . A p- alue less han 0.05 was conside ed signi ican . RESULTS Neu onal Ne wo k Cul u es in Tunnel De ices A e cell pla ing, he neu ons s a ed o mig a e and elonga e neu i es in o he unnels. The i s neu i es and neu ons en e ed he unnels as ea ly as 3 days a e pla ing, and ypically by 2–3 weeks he neu ons had o med a ne wo k co e ing app oxima ely he en i e a ea inside he unnels. Examples o ne wo k g ow h om he na owes (w=100 µm) and wides unnels (w=1,500 µm) a e shown in Figu es 3A,B, espec i ely. Occasionally, he e e sible PDMS-MEA bonding led o pa ial de achmen o some o he PDMS de ices om F on ie s in Neu oscience | www. on ie sin.o g 5Oc obe 2017 | Volume 11 | A icle 606 Toi anen e al. PDMS Tunnels Inc ease MEA Ac i i y FIGURE 3 | Ne wo k de elopmen in unnels and immunocy ochemis y images o neu onal cul u es. Ne wo k de elopmen in an h=43 µm, w=100 µm unnel (A) and an h=43 µm, w=1,500 µm unnel (B). The elec ode- o-elec ode dis ance is 200 µm. Red dashed boxes in (A,B) show he loca ion o he co esponding sub igu es on he igh , whe e soma a a e indica ed by black and neu i e bundles by whi e a owheads, d, days. Rep esen a i e MEA aces om he same week a e shown unde each unnel image. The elec odes om which he aces we e ob ained a e ma ked wi h whi e do s. Immunocy ochemis y images o wo cell lines: (C) he hiPSC line 04311.WT and (D) he hESC line 08/023 g owing on cell cul u e con ols pla es. The cell nuclei (DAPI, blue) and neu onal ma ke s (mic o ubule-associa ed p o ein 2 [MAP2] and class III ubulin [β- ub], g een) we e s ained. (E) Neu i e o ien a ion analysis con i med signi ican ly smalle ci cula a iances o he ne wo ks in unnels han in eely g owing open cul u es. nopen cul u e is 8 and no he wise is 2–7 (Table 1). S a is ical di e ence be ween he g oups was analysed using uni a ia e analysis o a iance, and he * symbols indica e signi icance based on Bon e oni’s pos hoc es s. eely g owing open cul u es (*0.05 >p≥0.01; **0.01 >p≥0.001; ***p>0.001). he MEAs. The cul u es wi h insu icien PDMS-MEA bonding we e excluded om he expe imen s. The neu onal na u e o he used cells was e i ied by immunocy ochemical s aining o known neu onal ma ke s (Figu es 3C,D). Cell iabili y in he unnels was good, and no signi ican cell dea h o de achmen was obse ed wi h phase con as mic oscopy du ing he 5-week cul u e pe iod. The unnels con ained neu i es and cell soma a mig a ed in o he unnels. Towa d he end o he cul u e he neu i es ended o o m hick bundles ha we e ypically nex o he PDMS walls ega dless o he unnel wid h. All unnels F on ie s in Neu oscience | www. on ie sin.o g 6Oc obe 2017 | Volume 11 | A icle 606 Toi anen e al. PDMS Tunnels Inc ease MEA Ac i i y a ec ed neu onal ne wo k de elopmen by causing neu i e o ien a ion (Figu es 3A,B) compa ed wi h andom neu onal ne wo ks wi hou unnels (Figu es 3C,D). This obse a ion was e i ied by he neu i e o ien a ion analysis, which showed ha he mean ci cula a iance alues in he ne wo ks inside unnels we e signi ican ly smalle han in eely g owing ne wo ks [F(3, 31) =6.63, p=0.001; Figu e 3E]. The unnel heigh had no signi ican e ec on he neu i e o ien a ion [F(1, 31) =2.94, p= 0.097].The neu i es in he na owes (w=100 µm) unnels we e he mos unidi ec ional (Figu es 3A,E,). MEA Signal De ec ion Inside Tunnel De ices To de e mine whe he he di e en unnel dimensions a ec ed he signal de ec ion on MEA, we calcula ed he noise alues, md peak- o-peak signal ampli udes, and om hese, he co esponding SNRs in each ac i e elec ode using a cus om- made MATLAB algo i hm (Figu e 4). Examples o he neu onal signals a e shown in he na ow unnel (w=100 µm) and wide unnel (w=1,500 µm) (Figu es 3A,B). Da a om s anda d 1-well MEAs se ed as he ele an con ol o hese pa ame e s because he unnel de ice da a we e ob ained using he same indi idual MEAs. The 1-well MEA con ol da a did no con ain measu emen s om he i s week a e pla ing, and week 1 da a we e he e o e omi ed om his analysis bu a e p esen ed in Supplemen a y Figu e 1. The ou side-g oup e e s o he elec odes ou side he unnels bu unde he PDMS lid (Figu e 1), and hus ha e he same has he unnel elec odes. The md noise alues we e highe in he elec odes inside he PDMS unnels in compa ison o he 1-well MEA con ol (md 1.3 µV; Figu es 4A,B). In he h=43 µm unnels, noise was signi ican ly inc eased in he elec odes inside w=100 µm, w=750 µm, and w=1,500 µm unnels (md 1.8, 2.0, and 2.0, p<0.001; Figu e 4A). The 6-well MEA con ol had a compa a i ely high noise le el (md 3.1 µV). In he h=105 µm, he unnels noise was signi ican ly inc eased in w=500 µm, w =750 µm, and w=1,500 µm unnels (md 1.9, 1.9, and 1.8; p <0.001; Figu e 4B). In he h=105 µm de ices, he noise was pa icula ly high a he i s week a e pla ing (Supplemen a y Figu e 1), p obably due o sys em s abiliza ion and/o p o ein adso p ion o he elec ode su ace. Howe e , inclusion o he da a in he analyses had no signi ican e ec on he esul s. In gene al, noise was signi ican ly highe in he h=43 µm de ices compa ed wi h he h=105 µm de ices (p<0.001; Figu e 4A s. Figu e 4B). These esul s sugges ha PDMS unnels can inc ease noise in MEA eco dings depending on he unnel dimensions. In ag eemen wi h he inc eased noise alues, he md peak- o-peak signal ampli udes we e inc eased in elec odes inside unnel de ices when compa ed wi h he 1-well MEA con ol (md 13.3 µV; Figu es 4C,D). In he h=43 µm unnel de ices, he ampli udes we e signi ican ly inc eased in he ou side elec odes as well as he elec odes in w=100 µm, w=750 µm and w=1,500 µm unnels (md 18.4, 18.9, 20.1, and 19.5 µV, espec i ely; p<0.001; Figu e 4C). The 6-well MEA con ol had ela i ely high signal ampli udes (md 24.3 µV). In he h= 105 µm unnels, he ampli udes we e signi ican ly inc eased in FIGURE 4 | Noise in ac i e elec odes, signal ampli ude and signal- o-noise a io. The noise was calcula ed o ac i e elec odes in h=43 µm(A) and h= 105 µm unnel de ices (B) and con ols (6 and 1-well MEAs). The unnel wid h o con ol g oup is indica ed on he x-axes. The median signal ampli ude in ac i e elec odes in h=43 µm(C) and h=105 µm de ices (D) and con ols was also calcula ed. The signal- o-noise- a io (SNR) in h=43 µm(E) and h= 105 µm de ices (F) was calcula ed om he noise and median signal ampli ude in ac i e elec odes. n(ac i e elec odes) is 61–547 (weekly numbe s o ac i e elec odes a e in Table 1). Please no e ha he low numbe o h=43 µm, w=500 µm unnels (2) may a ec he esul s. S a is ical di e ences be ween g oups we e analysed using he K uskal-Wallis es , and he * symbols indica e signi ican di e ences using Dunn’s pos hoc es s. he 1-well con ol (*0.05 >p≥0.01; **0.01 >p≥0.001; ***p>0.001). F on ie s in Neu oscience | www. on ie sin.o g 7Oc obe 2017 | Volume 11 | A icle 606 Toi anen e al. PDMS Tunnels Inc ease MEA Ac i i y he elec odes in w=500 µm, w=750 µm, and w=1,500 µm unnels (md 18.2, 20.0, and 18.0 µV; p<0.001; Figu e 4D). In gene al, signal ampli udes we e signi ican ly highe in he elec odes in he h=43 µm de ices compa ed wi h he h= 105 µm de ices (p<0.001; Figu e 4C s. Figu e 4D). These esul s sugges ha PDMS unnel de ices can inc ease MEA signal ampli ude depending on he de ice dimensions. SNR was ound o be highe in he elec odes in unnel de ices compa ed wi h he 1-well MEA con ol (md 8.0, Figu es 4E,F). In he h=43 µm unnel de ices, SNR was signi ican ly inc eased in he ou side elec odes unde he PDMS lid as well as elec odes in w=100 µm, w=750 µm, and w=1,500 µm unnels (md 10.0, 9.9, 9.1, and 9.9, espec i ely; p<0.001; Figu e 4E). SNR was lowes in he 6-well MEA con ols (md 7.7). In he h=105 µm unnel de ices, SNR was signi ican ly inc eased in he ou side elec odes as well as elec odes in w=100 µm, w=500 µm, w=750 µm, and w=1,500 µm unnels (md 9.7, 9.8, 9.6, 10.1, and 9.7, espec i ely; p<0.001; Figu e 4F). The unnel heigh had no signi ican e ec on SNR (p=0.244; Figu e 4E s. Figu e 4F). In summa y, he SNRs eco ded om inside he unnel de ices we e e y simila , hus making he di e ences in noise and signal ampli udes be ween di e en designs i ele an in e ms o MEA signal de ec ion. Howe e , i appea s ha a PDMS unnel de ice on he MEA in gene al imp o es SNR. Tunnel De ices Inc ease Spike and Bu s Ac i i y on MEA To assess whe he he unnel de ices could a ec he spike and bu s ac i i y on MEA, we analysed he pe cen age o ac i e elec odes, spike coun and bu s coun in ac i e elec odes using ou cus om-made MATLAB algo i hm (Figu e 5). We compa ed he ac i i y da a om unnel elec odes o he 6-well MEA con ol. The s a is ical analyses be ween he unnel de ices and con ols we e pe o med sepa a ely each week because he ac i i y in he unnel elec odes inc eased d ama ically o e ime. The pe cen age o ac i e elec odes inc eased in he unnel de ices (Figu es 5A,B). A week 1, he pe cen age o ac i e elec odes was highes in he 6-well MEA con ols. Howe e , while he pe cen age in he 6-well MEA con ols dec eased om 46 o 27% o e 5 weeks, he pe cen age in he unnel de ices s eadily inc eased, su passing he 6-well MEA con ols 2–4 weeks a e pla ing and inally eaching 52–100% a week 5. The pe cen age o ac i e elec odes was especially high in he w=100 µm and w=500 µm unnels, eaching 80–100% by week 5. The pe cen ages we e also highe in he h=43 µm unnels compa ed wi h he h=105 µm unnels (69–100% s. 52–85% a week 5; Figu e 5A s. Figu e 5B). The smalles unnels (h=43 µm, w=100 µm) we e bes in e ms o he pe cen age o ac i e elec odes, eaching 100% as ea ly as week 3 (Figu e 5A). The e we e signi ican ly mo e spikes pe ac i e elec ode in he unnel de ices compa ed wi h he 6-well MEA con ols (Figu es 5C,D). The md spike coun in he 6-well MEA con ols ne e eached highe han 125 o e 10 min (week 4). The md spike coun was highes , wi h 3,449 spikes o e 10 min, in he smalles unnels (h=43 µm, w=100 µm) a week 5 (p<0.001; Figu e 5C). The ou side elec odes in he h=43 µm unnel de ices also had high spike coun s, wi h md eaching 2,715 a week 5. The spike coun s be ween he ou side elec odes and he w=100 µm unnels did no di e signi ican ly du ing any week. The spike coun s we e also signi ican ly inc eased in highe (h=105 µm) de ices compa ed wi h he 6-well MEA con ols (Figu e 5D). The maximal spike coun in he h=105 µm de ices was eached a week 4, when he md spike coun in he w =500 µm unnels was 992 (p<0.001). The md spike coun was nea ly as high in he w=100 µm unnels o he co esponding ime poin (964). In gene al, he spike coun was highe in he h =43 µm unnel de ices compa ed wi h he h=105 µm de ices. Fo example, in he w=100 µm unnels, he di e ence was signi ican a weeks 4 (2,886 and 964, espec i ely; p=0.013) and 5 (3,449 s. 907, espec i ely; p=0.002; Figu e 5C s. Figu e 5D). Taken oge he , he spike coun da a showed ha he w=100 µm unnels and ou side a ea o he h=43 µm unnel de ices we e he bes o inc ease he amoun o measu ed ne wo k ac i i y. As wi h he spike coun s, he bu s coun s we e also inc eased in he unnel de ices compa ed wi h he 6-well MEA con ols. In he 6-well MEA con ols, he md bu s coun was highes a week 5, wi h 6 bu s s o e 10 min (Figu es 5E,F). In he smalles unnels (h=43 µm, w=100 µm), he bu s coun eached a maximum, 373, a week 4 (p<0.001; Figu e 5E). The bu s coun s we e also e y high in he ou side elec odes o he h= 43 µm unnel de ices, wi h 250 bu s s a week 5 (p<0.001). The bu s coun s be ween he ou side elec odes and he w= 100 µm unnels did no di e signi ican ly du ing any week. The bu s coun s we e also signi ican ly inc eased in he highe (h =105 µm) unnels compa ed wi h he 6-well MEA con ols, achie ing a maximum o 108 a week 4 in he w=500 µm unnels (p<0.001; Figu e 5F). The md bu s coun was also high in he w=100 µm unnels a he same week, wi h 80 bu s s. Howe e , he numbe o bu s s in he h=105 µm unnels was gene ally less han in he h=43 µm unnels. Fo example, in he w=100 µm unnels, he di e ence was signi ican a weeks 4 (80 and 373, espec i ely; p=0.006) and 5 (56 and 368, espec i ely; p=0.002; Figu e 5E s. Figu e 5F). The bu s coun da a sugges ed ha he MEA ac i i y was highes in w= 100-µm-wide unnels and he ou side a ea o he h=43-µm- high unnel de ices, which is consis en wi h he spike coun esul s (Figu es 5C,D). Howe e , conside ing also he pe cen age o ac i e elec odes (Figu es 5A,B), i is no he a ea ou side he unnels, bu pa icula ly he smalles unnels (h=43 µm, w= 100 µm), ha had he g ea es abili y o inc ease he amoun o measu ed ne wo k ac i i y. MEA Ac i i y Inside Tunnel De ices Can Be A ec ed by Pha macological T ea men To es whe he he MEA signals inside he unnel de ices o igina ed om neu onal ac i i y, we measu ed he MEA ac i i y inside he de ices be o e and a e TTX ea men (Figu es 6A,B). TTX inhibi s he unc ion o neu onal ol age- ga ed sodium channels and, he e o e, blocks he elec ical F on ie s in Neu oscience | www. on ie sin.o g 8Oc obe 2017 | Volume 11 | A icle 606 Toi anen e al. PDMS Tunnels Inc ease MEA Ac i i y FIGURE 5 | Pe cen age o ac i e elec odes, spike coun and bu s coun . The pe cen age o ac i e elec odes was calcula ed om he MEA da a o he h=43 µm (A) and h=105 µm unnel de ices (B) as well as he con ols (6 and 1-well MEAs). To al no analysed elec odes was 23-270 (Table 1). The spike coun in he ac i e elec odes o e 10 min was also analysed o he h=43 µm(C) and h=105 µm unnel de ices (D) and con ols. The numbe o bu s s o e 10 min was analysed om he spike da a o he h=43 µm(E) and h=105-µm-high unnel de ices (F) and he con ols. n(ac i e elec odes) was 1-167 (Table 1). Please no e ha he low numbe o h=43 µm, w=500 µm unnels (2) may a ec he esul s. S a is ical di e ences be ween g oups each week we e analysed using he K uskal-Wallis es , and * symbols indica e signi ican di e ences using Dunn’s pos hoc es s. he 6-well con ol (*0.05 >p≥0.01; **0.01 >p≥0.001; ***p>0.001). ac i i y o neu ons. A e addi ion o egula medium, he pe cen age o ac i e elec odes was be ween 26% (h=105 µm, ou side elec odes) and 94% (h=43 µm, w=100 µm). A e he addi ion o TTX o he medium, he pe cen age o ac i e elec odes d opped be ween 5% (h=105 µm, w=1,500 µm) and 0% (e.g., h=43 µm, w=100 µm). The e ec o TTX on MEA ac i i y in unnel de ices was p omp and clea , showing ha he measu ed ac i i y was o neu onal o igin and ha he unnel de ices can be used in pha macological expe imen s. F on ie s in Neu oscience | www. on ie sin.o g 9Oc obe 2017 | Volume 11 | A icle 606