To form ORN precursor spheroids, 96 well plate (Thermo Scientific) was coated with 100 µl of 3 µM REP diluted in PBS and incubated for one hour in a humidified (37℃) incubator and 5% CO₂. After one hour, REP solution was aspirated and 1×10⁶ cells/ml dissociated ORN precursor suspension was plated on the 96 plate. After 24 hours, ORN precursor spheroids were formed and the cultivation time dependent ORN spheroids were transferred to the 25 µM laminin-coated coverslip substrate by pipetting and differentiated for 72 hours. Before calcium imaging, coverslip substrate was combined with magnetic chamber (Live Cell instrument, Seoul, Korea) and media was replaced with KREBS ringer buffer (115 mM NaCl, 5.9 mM KCl, 2.5 mM CaCl₂·2H₂O, 1.2 mM MgCl₂, 1.2 mM NaH₂PO₄, 10 mM HEPES sodium salt, 10 mM D-glucose) followed by washing with PBS. Then, cells were dyed with fluo3-AM (fluo3-AM, life technologies, Waltham, USA) according to the manufacturer's instruction. Fluorescence images of calcium indicator were monitored using a confocal microscope (Zeiss LSM 700; Carl Zeiss, Oberkochen, Germany). To stimulate ORNs, a 100 µM isovaleric acid (IVA), 2-isobutyl-3-methoxypyrazine (IBMP), citralva, and a 100 µM odorant mixture containing IVA, IBMP, and citralva were applied to the magnetic chamber by using tubing connected syringe. Intracellular calcium concentrations were calculated from the background corrected ratio of fluorescence at 506/526 nm using a two-point calibration scheme and the equation:

[Ca]i=Kd(F0/Fs)(R−Rmin)/(Rmax−R)(1)

where R is the fluorescence ratio F₅₀₆/F₅₂₆, R_min and R_max are the fluorescence ratios of 1 µM fluo-3 standards at the limiting low (0 µM Ca²⁺) and high (10 mM Ca²⁺) concentrations, respectively, K_d is the calcium dissociation constant of fluo-3, and F₀/F_s is the ratio of fluorescence intensities when excited at 562 nm at low and high Ca²⁺ concentrations, respectively.

Primary culture of ORNs was performed as previously described with some modifications [16]. Zero to one-day old rat pups were decapitated and olfactory epithelium (OE) was dissected and immediately placed in modified Eagle's medium containing D-valine (MDV; Welgen Inc., Worcester, USA) containing 4.8 gam/liter of HEPES buffer. The turbinates were transferred into fresh MDV to minimize contamination and centrifuged at 1000 rpm for 5 min. After the supernatant was decanted, the tissue was minced to achieve fragments of approximately 0.5 mm, resuspended in MDV, and centrifuged at 1000 rpm for 5 min. The minced tissue was then placed in 30 ml of MDV containing 1% (w/v) BSA, (SIGMA, St. Louis, USA), 1 mg/ml hyaluronidase (SIGMA), 50 µg/ml DNase (SIGMA), 2 mg/ml collagenase D (Roche Diagnostics GmbH, Mannheim, Germany), and 5 mg/ml dispase II (Roche), and incubated for 1hr at 37℃. Following incubation, cells were passed through 150 µm wire mesh and nylon mesh filters (70 µm, 40 µm and 10 µm. Small Parts, Miami, FL). After centrifugation at 1,200 rpm for 5 min, the cell pellet was resuspended in plating medium [MDV containing 10% (v/v) fetal bovine serum (FBS; GIBCO, USA), 5% (v/v) Nu-serum (BD Biosciences, Franklin Lakes, NJ, USA), 10 µM cytosine arabinoside (ara C), and 25 ng/ml nerve growth factor (NGF, Collaborative Research)]. Cultures were placed in a humidified 37℃ incubator containing 5% CO₂. The cells were fed with modified Eagle's medium containing D-valine, 15% FBS, gentamicin, kanamycin, ara C (SIGMA).

Before plating dissociated ORN precursors, 96 well plates (Thermo Scientific, Waltham, USA) were coated with 100 µl of 3 µM TGPG[VGRGD(VGVPG)₆]₂₀WPC (REP) diluted in PBS and incubated for one hour in a humidified (37℃) incubator and 5% CO₂. REP was prepared as previously described [17]. After one hour of incubation in the humidified incubator, REP solution was aspirated and dissociated ORN precursors suspensions (1×10⁶ cells/ml) were plated and stored in the humidified incubator. Spheroids were generated within 24 hours and fresh MEM (Welgene, Gyeongsan, Republic of Korea) containing 0.5% FBS (GE Healthcare lifesciences, Brussel, Belgium), 0.5% penicillin streptomycin (Thermo Fisher, Waltham, USA), kanamycin (SIGMA), gentamycin (SIGMA) was changed every 24 hours.

Media was aspirated and cells were washed with PBS. Cells were fixed with a 1:1 mixture of acetone and methanol. Cells were permeabilized by incubation in 0.05% Triton X-100 (Triton® X-100; SIGMA, St. Louis, USA) for ten min and blocked with 4% normal donkey serum for one hour. Cells were then incubated in a primary antibody diluted solution overnight at 4℃. Primary antibodies were anti-NST (used at 1µg/ml, abcam, Boston, USA), anti-neural cell adhesion molecule (AB5032; used at 1:100, abcam), antimicrotubule-associated protein 2 (used at 1:200, 4542, Cell signaling Technology, Danvers, USA), anti-(sex determining region Y)-box 2 (SOX2; used at 1:200, sc-17320, Santa Cruz, Dallas, USA), anti-nestin (used at 1:200, MAB353, Millipore Corp. Burlington, USA), anti-Ki-67 (used at 1:200, D3B5, Cell Signaling Technology). Incubated cells were washed three times with PBS and fluorescently labeled with secondary antibodies for 2 hours at room temperature. Secondary antibodies were donkey anti-mouse IgG H&L Alexa Fluor ®488 (ab150105; abcam), goat anti-rabbit IgG H&L Alexa Fluor ®568 (ab175471; abcam), and donkey anti-goat IgG H&L Alexa Fluor ®488 (ab150129; abcam). After washing with PBS, cells were immersed in mounting medium containing DAPI (VECTOR, Burlingame, USA) and Hoechst (Invitrogen, Carlsbad, USA) and sealed with a coverslip. Fluorescence images were obtained using a confocal microscope (Zeiss LSM 700; Carl Zeiss, Oberkochen, Germany).

For atomic force microscopy (AFM) image inspection of REP and laminin, glass coverslips were coated with REP (3 µM) and laminin (25 µg/ml). To coat REP on the coverslips, REP solution (3 µM) was poured on the coverslip and incubated for one hour in 37℃. Next, coverslips were washed three times with warmed autoclaved distilled water (37℃) and dried in a hood. To prepare laminin-coated coverslips, laminin (25 µg/ml) diluted with autoclaved distilled water was poured onto the coverslips and stored at 37℃ overnight. Coverslips were then washed three times with autoclaved distilled water and dried using a critical point dryer (EM CPD 300; Leica, Wetzlar, Germany). AFM (XE-150; Park Systems, Suwon, Republic of Korea) was operated in non-contact mode with a scanning rate of 0.5 Hz. For scanning electron microscopy (SEM), a coverslip was coated with REP (3 µM) and laminin (25 µg/ml) prepared in diluted autoclaved distilled water for an hour at 37℃. Next, coverslips were washed three times with autoclaved distilled water (37℃) and dried in a hood. To prepare laminin coated coverslips, laminin (25 µg/ml) diluted in autoclaved distilled water was poured onto the coverslip and stored at 37℃ overnight. Then, coverslips were washed three times with autoclaved distilled water and dried using the critical point dryer (EM CPD 300; Leica, Wetzlar, Germany). Before SEM inspection, coverslips were coated with Osmium in the Osmium plasma coater (Vacuum Devices Inco., Mito, Japan). SEM (Carl Zeiss, Oberkochen, Germany) was operated under chamber pressure at 6.5⁻⁶ mbar.

Single well MEA (M64-GLx; Axion bio systems, Atlanta, USA) was used in all experiments. MEA has 64 platinum black microelectrodes in an 8×8 configuration. Single electrode has 30 µm diameter with 200 µm spacing (center to center) and four ground electrodes (2 stimulation & 2 recording grounds) were equipped. Average impedance was 25 kΩ at 1 kHz and butterworth filter (0.1~3,000 Hz) setup was used. Before plating ORN precursor spheroids, MEA was sterilized three times with 95% ethanol. MEA plate wells were pretreated with 25 µM laminin prior to seeding ORN precursors. After 72 hours of ORN precursor seeding for differentiation, the media was changed to HEPES buffer. Odorants (citralva, IBMP, and IVA) were diluted in HEPES buffer to modify concentrations. In the entire experiment, odorant stimulations were performed with a 60 sec; between odorant stimulation, buffer solution was applied with 30 sec of interval. After three to five odorant stimulations, stabilization time was given for 5 min following a change to fresh media considering the desensitization of the ORNs.

Principal component analysis (PCA) was performed as follows:

XOdor, Trial=xCH1,xCH1,xCH1,⋯ ⋯,fCH64(2)

where Odor represents the odorants used (ie, citralva, IBMP, IVA), Trial is the number of trials, and CH_n is the number of MEA electrodes. Data matrix, X, was from 64 MEA channels based on the results. Transformation was defined by a set of p-dimensional vectors of weights w_(k)=(w₁,…,w_p)_(k) that mapped with a new vector of principal component scores given by t_k(i)=x_(i)·w_(k). Each variable of t was the maximum possible variance from x, and weight vector w constrained to be a unit vector. We found kth component by subtracting k−1 principal components from X.

Xk=X−∑s=1k−1Xw(s)w(s)T(3)

and calculated the weight vector by extracting the maximum variance from this new data matrix.

ws=arg maxw=1Xkw2=arg maxwTwkTXkwwTw(4)

Final principal components decomposition of X can be given as,

T=XW(5)

and we generated dimension-reduction transformed data using principal component k=1 to k=3. Transformed data y was extracted from matrix T, because we only used k=1 to k=3 data and we can define data y as,

y(n)=(y1,y2,y3)(n)(6)

Data y can be mapped in a 3D space, so we can define distance between odorant p and odorant q as a Euclidean distance:

d=p−q=(y1(p)−y1(q))2+(y2(p)−y2(q))2+(y3(p)−y3(q))2(7)

Distance d represents data similarity and a higher d value means less similarity.

Total RNA was obtained from ORNs and ORN spheroids using Trizol (Trizol® Reagent; Life technologies, Waltham, USA) and complementary deoxyribonucleic acid (cDNA) was synthesized by reverse transcription. Quantitative real-time RT PCR was performed using the SYBR Green PCR master mix kit (QuantiTect® SYBR® Green PCR Kit; QIAGEN, Hilden, Germany) on the Rotor-Gene Q (QIAGEN, Hilden, Germany) under the following conditions: 40~45 cycles of 95℃ for 30 sec, 60℃ for 30 sec, and 72℃ for 30 sec. Primers of the following sequences were obtained from GenoTeck (GenoTeck, Daejeon, Republic of Korea) and Bionics (Bionics, Daejeon, Republic of Korea): cyclophilin A (forward: AGC ACT GGG GAG AAA GGA TT and reverse: AGC CAC TCA GTC TTG GCA GT); MAP2 (forward: TGT TGC TGC CAA GAA AGA TG and reverse: ACG TGG CTG GAC TCA ATA CC); neuron specific enolase (NSE, forward: GTG GAC CAC ATC AAC AGC AC and reverse: TGA GCA ATG TGG CGA TAG AG); neuron-specific class III beta-tubulin (forward: TGA GGC CTC CTC TCA CAA GT and reverse: CTC ACG ACA TCC AGG ACT GA); adenylyl cyclase III (forward : TCC TGT GTT GTG CAT ACG CT and reverse: ACG TTA GCC AGG ATC TCC CT); olfactory marker protein (OMP, forward : GAA GCA GGA TGG TGS GSS GC and reverse: ATG AGG TTG GTG AGG TCG CG); neural cell adhesion molecule (NCAM, forward: AAA GGA TGG GGA ACC CAT AG and reverse: TAG GTG ATT TTG GGC TTT GC); (sex determining region Y)-box 2 (SOX2, forward: AGG GCT GGG AGA AAG AAG AG and reverse: TTG CTG ATC TCC GAG TTG TG); Ki67 (forward: GCC CAT CAC CAC AGA GAT TT and reverse: CAG TCT TCA GGG GCT CTG TC); nestin (forward: GAG GAA GCA TCG AAC TCT GG and reverse: CAG CTT TAG CTT GGG ATT GC). The house keeping gene cyclophilin A was used as an internal standard.

Calcein assay reagent (calcein red-orange, AM; Invitrogen, Carlsbad, USA) was used to evaluate call viability. ORN precursor spheroids were gently washed three times with Dulbecco's phosphate-buffered saline (DPBS; gibco® by life technologies™, Waltham, USA), treated with 3 µM calcein-AM solution and incubated at 37℃ for 30 min. Cell viability within each well was quantified by measuring optical density at 517 nm with a fluorescence reader (Gemini EM; Molecular Devices, San Jose, USA).

In this study, spheroid culture of the ORN precursors using REP was initially suggested. REP is an amphiphilic biosynthesized elastin-like polypeptide which modulates cellular morphology by activating cell adhesion proteins [17,18,19,20]. We have modified the dissociated neonatal rat ORN precursors based on the previously developed method [16] to study biological characteristics of ORNs and ORN-based odorant measurement [21,22,23]. In the previous results by Ronnett, dissociated rat ORN precursors were cultured on a laminin-coated plate and differentiation of ORNs was verified using immunocytochemical assays [16]. Additionally, physiological characteristics of the ORNs were verified by monitoring intracellular cAMP levels upon odorant stimulations, suggesting that the dissociated ORN precursors were appropriate for this study [24].

To identify optimized conditions for spheroid formation, diameters and morphology of spheroids depending on the seeding populations and REP concentrations were observed by measuring diameters of ORN spheroids. The obtained ORN precursors were plated on either REP or laminin-coated cell culture plates. ORN precursors cultured on the laminin-coated plate differentiated with neurite outgrowth, whereas, ORN precursors on the REP-coated plate aggregated and formed spheroids within 24 hours (Fig. 1A). After 24 hours of cell seeding, the average diameters of spheroids on the 1 µM REP, 3 µM REP, 5 µM REP coated plates were 24.2±0.7 µm, 62.6±2.2 µm, 56.2±2.8 µm, respectively. On day three, the average diameter of spheroids on the 3 µM REP-coated plate increased significantly to 77.5±3.3 µm and the diameter of spheroids on the 5 µM REP increased significantly to 69.4±3.6 µm on day five. ORN spheroids on both 3 µM and 5 µM REP-coated plates maintained their average diameter (60 to 70 µm) for ten days, however, the average diameter of ORN spheroids on 1 µM REP-coated plates was roughly 20 µm with poor spheroid morphology and unstable cellular aggregation (Fig. 1B). As a result, the relatively small diameter of spheroids in the 1 µM REP-coated plates might have low neuronal population after differentiation, and be limited in their ability to generate electrical signals following odorant stimulation. Therefore, 3 µM REP was used for remaining experiments.

To characterize the effect of REP on spheroid formation, surface characteristics of REP-coated coverslips was inspected using atomic force microscopy (AFM) and scanning electronic microscopy (SEM). REP formed tissue-like porous structures and attached laminin matrices formed network-like structures (Fig. 1C). Additionally, surface topography and roughness was measured by AFM. Compared to the laminin, REP formed hundreds of nanoscale height structures and had 148.3±15.9 nm roughness, which is 23.8 times higher than those grown on the laminin-coated surfaces (Fig. 1D). Therefore, disparate topographical characteristics of REP compared to laminin might affect the spheroid formation of ORN precursors plated on the REP.

Based on previous studies characterizing spheroid formation using synthetic peptides, cell-synthetic peptides and cell-cell interactions may be involved in cellular aggregation [17,20,25]. Taking previously reported results into account, VG(VGVPG)₆ domains provide the hydrophobic environment enhancing cellular coupling and binding of RGD ligands to integrin on the cellular membrane to activate signal cascades which lead to the expression of cell adhesion molecules and ECM [20,25]. Together with our results, porous structures of REP with relatively high roughness may affect cellular attachment; topographical characteristics may be a factor in the formation of spheroids. Next, to assess the viability of ORN spheroids depending on the cultivation period, the calcein assay was performed. OD_517nm values of dissociated ORN spheroids dyed with calcein-AM were measured using a fluorescence reader and entire OD_517nm values were normalized by OD_517nm values of the day one ORN spheroids. On day three, significantly increased fluorescence intensity was measured (1.05±0.03 fold), and after day three, viability gradually decreased (0.87±0.01 fold) on day ten (Fig. 1E). Decreases in the fluorescence intensity might represent a declined viability due to the environmental conditions and cell loss during media change. High viability of ORN spheroids during cultivation is necessary for the development of a cell-based biosensor bioelement; our results revealed sufficient viability for ten days.

To quantify stemness of the ORN precursor spheroids, mRNA levels of ORN stemness markers were analyzed by quantitative polymerase chain reaction (qPCR). SRY (sex determining region Y)-box 2 (SOX2), Ki67, and nestin, were used as measurement of stemness, and neuron-specific enolase (NSE) was used to verify that there was no neuronal differentiation of ORN spheroids. Normalized relative mRNA levels of SOX2, Ki67, and nestin were 8.4±0.8 folds, 8.6±0.8 folds, 11.7±1.1 folds, respectively on day three. Expression levels of these three markers were slightly increased on day five. On day ten, these levels decreased below day three values, however these differences were not statistically significant. NSE expression on days three, five, and ten were roughly 0.1 fold (Fig. 1F). For further identification of stemness of the ORN precursor spheroids, we performed immunocytochemistry (ICC) using stemness markers (ie, nestin, Ki67, and SOX2). We found that nestin was expressed on the surface of the ORN spheroids, Ki67 was co-labeled within the nucleus and SOX2 was expressed in the cell body. Those expression characteristics were not significantly different, however expression modality declined depending on the cultivation time (Fig. 1G).

Next, we differentiated ORN spheroids and monitored their developmental characteristics using qPCR and ICC. Differentiation of ORN precursor spheroids was accomplished by transferring spheroids onto laminin-coated plates and cultivating them for 72 hours (Fig. 2A). Once the spheroids were transferred to the laminin-coated plates, the ORN precursors that formed spheroids migrated radially and displayed neurite outgrowth. (Fig. 2B).

To investigate the differentiation of ORN precursors, transcript levels of ORN developmental markers (ie, ac3, omp, map2, and ncam) were analyzed by qPCR. Adenylyl cyclase III (AC3) is an important protein expressed in ORNs and plays a critical role in the olfactory signal transduction cascade [4]. Olfactory marker protein (OMP) is also a protein found in fully matured ORNs, and both microtubule associated protein 2 (MAP2) and NSE are expressed in differentiated neurons. Thus, these markers are regarded as neuronal maturation markers [26,27,28]. Neural cell adhesion molecule (NCAM) was used as a neuron marker. The expression levels of ac3, omp, map2, and nse on day three were 6.3±0.7, 6.3±0.7, 8.7±0.5, 9.1±0.8 fold, respectively (Fig. 2C). On day five, expression levels of ac3 and ncam increased and expression levels of omp and map2 decreased. And on day ten, expression level of all markers decreased below the level of day three. Expression level of nestin at three, five, and ten days in the laminin group was roughly 0.1 fold, indicating no ORN spheroid differentiation.

To further examine differentiation of ORN precursor spheroids, cells were immunostained with βIII-tubulin (NST), NCAM, and MAP2 which are expressed in ORNs. Immunostaining of ORNs using antibodies against NST and NCAM proteins revealed clear expression in the neurites and soma of ORNs. In addition, MAP2 expression was observed in only one neurite of a bipolar neuron, suggesting that ORN precursors spheroids could differentiate into bipolar neurons which is the intrinsic characteristics of in vivo ORNs (Fig. 2D).

Next, we verified the physiological characteristics of ORNs using calcium imaging analysis to facilitate a comparison to in vivo ORNs. To activate ORNs, we used an odorant mixture of three odorants reported to activate the main signal pathway of ORNs [4]. The ORN precursor spheroids were transferred to laminin-coated plates, and the plated ORNs on the laminin-coated plates were stimulated with the odorant mixture. When the ORNs were stimulated with the odorant mixture, ORN activity was measured by assessing changes in the fluorescent intensity (Fig. 3A and 3B).

To further investigate the physiological characteristics of the ORNs, we analyzed the net-odor evoked calcium concentration and response rate based on the results of calcium imaging. Netodorant evoked calcium concentration is the absolute value of elevated calcium concentration in the odorant-stimulated ORNs and calculated using the formula described in Materials and Methods. Computed average of net odor-evoked concentrations in the ORNs cultivated for three, five, and ten days as spheroid were 195.7±24.0 nM, 205.3±24.9 nM, 205.4±22.3 nM, respectively (Fig. 3C). However, these values were not statistically significant among the three different groups. And response rate was calculated as a ratio of the number of activated cell population and whole number of cell population in a confocal microscope software window. Response rate can be used as an indicator for qualification of ORNs as a bioelement for the development of an electronic nose. Calculated response rates depending on the spheroid cultivation time (ie, three, five, and ten days) were 29.9±2.5%, 27.6±3.0%, 28.1±4.2%, respectively (Fig. 3D). Likewise, there was no significant difference among three different spheroid cultivation time groups suggesting that the physiological characteristics of the ORNs differentiated from spheroids have reliable activation rates upon odorant stimulation for up to ten days.

To understand physiological characteristics of ORNs under the single odorant stimulation, we monitored changes of intracellular calcium level of ORNs using 100 µM IBMP, 100 µM IVA, and 100 µM citralva. Fig. 3E shows a confocal microscope software window with color modification. Three cells of interest were monitored those changes in fluorescence intensity upon the odorant stimulations. Cell a was activated upon all odorant stimulations, cell b was activated by IVA, and cell c was activated by both IVA and citralva (Fig. 3E).

To extend cell storage capacity, we cryopreserved ORN spheroids for 35 days and tested physiological characteristics after thawing. ORN spheroids on day five were cryopreserved using 10% DMSO and 20% FBS solution and thawed after 35 days on the laminin-coated plate for differentiation (Fig. 3F). Compared to the day five ORN spheroid physiological characteristics, net odor-evoked calcium concentration and response rate decreased to 72.5% and 72.3%, respectively (Fig. 3G and 3H). Although the physiological level was declined by 30%, in consideration of this extended storage period, cryopreservation of primary culture ORNs showed sufficient storage capacity in view of their potential to be developed as a bioelement of a bioelectronic nose.

To utilize the ORN spheroid culture for developing a bioelectronic nose, we verified their capabilities to detect odorant types and concentrations by plating ORN spheroids on a commercialized MEA. ORN spheroids were differentiated onto the MEA followed by laminin coating. An illustration of experimental protocol is shown in Fig. 4A. We hypothesized that the electrical signals generated in the ORNs upon odorant stimulations and intensity of field potentials measured by MEA produced specific patterns depending on the types and concentrations of the given odorants. Therefore, analyzing the patterns of the signal from the ORNs enables the distinction of varying types and concentrations of odorants (Fig. 4B).

First, to demonstrate whether odorant concentrations can be detected using the ORN spheroids plated MEA, ORNs were stimulated by 1 µM, 10 µM, and 100 µM IBMP for two times and generated field potentials were measured. Measured field potentials were illustrated for displaying concentration dependent similarity of signal patterns. In the 1 µM IBMP stimulation, measured field potentials occurred primarily at the following electrodes [electrode position (x,y): (2,4), (4,4), (4,7), (5,2), (7,5)] with the following values (29.9 µV, 25.8 µV, 30 µV, 29.8 µV, 29.8 µV, respectively). And in the 100 µM IBMP stimulation, field potentials measured at the same electrodes were 23.8 µV, 58.5 µV, 37 µV, 37.1 µV, 37.3 µV, respectively (Fig. 4C). Waveforms of field potentials are shown in Supplementary Fig. 1. In the second session, field potentials were measured at similar position of electrodes. Interestingly, at electrode number (4,4), the biggest intensity of field potentials was measured like first session, and at other electrode positions monitored in the first session also measured increased field potentials depending on the IBMP concentration. And based on the results, dose-response curves of field potentials at five electrode positions were constructed depending on the IBMP concentration (Fig. 4D) and these shape of curve were similar to ligand-olfactory receptor interactions reported in many studies [29,30,31]. Additionally, we found significant increase of averages of field potentials in both sessions and the average intensities in first and second sessions were 12.02±2.92 µV in 1 µM IBMP, 14.52±3.42 µV in 10 µM IBMP, 27.89±2.59 µV in 100 µM IBMP and 5.17±2.75 µV in 1 µM IBMP, 14.64±3.78 µV in 10 µM IBMP, 27.66±2.87 µV in 100 µM IBMP, respectively (Fig. 4E). The correlation coefficient of the first and second patterns were 0.57 at 1 µM, 0.75 at 10 µM, and 0.84 at 100 µM. As a result, average field potential in the second session decreased by 30% compared to the first session. However, an increase in field potential depending on IBMP concentration was still observed, suggesting that different concentrations of odorant measurement using the ORN-based biosensor can be performed with reproducibility by analyzing signal patterns.

Next, to test capabilities of odorant discrimination, 100 µM IBMP, 100 µM citralva, and 100 µM IVA were used to stimulate ORNs. Using another setup of MEA which different distribution of ORNs were plated, patterns of field potentials depending on types of odorants were observed upon three times of odorant stimulations. In the first stimulation of 100 µM IVA, the highest field potential intensity was measured at electrode position (3,5). During three times of IVA stimulations, similar patterns of field potentials were produced (Fig. 4F). In the second odorant stimulation of 100 µM citralva, considerably different patterns of field potentials were produced and population of the responded ORNs were increased by 2.5 times compared to those of in the IVA stimulation. Likewise, in the IBMP stimulation, different patterns of field potentials were observed with different position of electrodes. Waveform data are shown in Supplementary Fig. 2. Average intensity of field potentials through three odorant stimulations showed odorant dependent characteristics. Following three odorant stimulations with IVA, 25.57±6.04 µV, 43.53±4.81 µV, and 30.25±5.55 µV field potentials were measured. Following stimulation with citralva, 39.67±4.54 µV, 43.25±4.06 µV, and 34.84±3.50 µV field potentials were measured. These two odorants showed increased and decreased field potentials in the second and third stimulations. However, IBMP showed a gradually decreasing field potential in the three times of stimulations (Fig. 4G). Based on the changes of average field potential in the three odorant stimulations, odorant dependent modality of desensitization was found. Next, discrimination of measured patterns of field potentials were verified by performing PCA (Fig. 4H). PCA results showed that patterns collected from the same odorant had similar component values, meaning that the developed ORN-based biosensor had capability of discriminating IVA, citralva, and IBMP.

All waveforms measured in entire MEA experiments were classified into three types depending on polarity and order of peaks. First type of waveform (Supplementary Fig. 3) has only positive peak which is known as representative recorded signals from olfactory supporting cells [32]. Second type of waveform (Supplementary Fig. 3) has first negative peak and following positive peak which is usually measured waveforms near somatic ORNs [33]. And third type of waveform (Supplementary Fig. 3) has first positive peak and following negative peak which is reported as a representative dendritic ORN waveform. In conclusion, we found that both dendritic and somatic ORN type of waveforms were contributed to the activation by odorants, however supporting cell type of waveforms were not affected to the generation of meaningful signals because supporting cells do not express ORs, thus there were no interactions with odorants. Therefore, based on the classified three type of waveforms, we calculated that 46.7% was dendritic ORN type, 32% was somatic type, and 21.3% was ORN supporting cell type in the entire waveforms. And we can predict that efficiency of field potential measurement (78.7%) from the ratio of dendritic and somatic ORN type of waveforms and supporting cell type of waveforms.