iPSC lines from healthy and diseased subjects

In the present study one single clone from four independent lines were selected on the basis of different cell sources, genetics, methodologies used for reprogramming, and that differentiations for comparisons (Anatomic and Chambers Protocol) were attempted once except for Ctrl 1 (healthy subject) line which was differentiated twice in the study. Two control lines were obtained from healthy individuals and two of the lines were obtained from disease-affected patients (Table 1). The timing of neuronal maturation was compared using two different protocols for both of the healthy control lines. In addition, the Anatomic protocol was used to compare the functional properties of sensory neurons obtained from three lines for a comparative study for disease modeling. To demonstrate that sensory neurons derived from the Anatomic protocol are amenable to high throughput technologies for disease modeling and drug discovery applications, the commercially available RealDRG™ sensory neurons were characterized by RNAscope, automated patch clamp, calcium imaging, and multielectrode array (MEA) techniques.

Table 1 Clinical phenotype of subjects used in the study

iPSCs were derived from mesenchymal stromal cells (MSCs) (female 69 years) and fibroblasts (repository name CS00iCTR21n1; Male 6 years) of two healthy subjects designated as Ctrl1 and Ctrl2 respectively (Table 1). Ctrl2 has been referred to as a “resistant clone” in this study due to its lower differentiation potential with Chambers protocol in generating peripheral neuronal lineage (Additional file 1: Fig. S1). iPSC-derived from blood cells of a female 9-year-old suffering from IEM a phenotypically similar but different patient to the first reported by [27] and fibroblasts from a female 69-year-old suffering from SFN disorder (repository name UKERi313-R1, [25] were used for disease modelling designated in this study as IEM and SFN, respectively (Table 1). iPSCs generated from blood cells of the IEM patient were found to be heterozygous for the p.Q875E mutation in the Nav1.7 ion channel (Additional file 1: Fig. S2). iPSCs-derived (ANAT001) from cord blood cells of a healthy subject were used to generate commercially available RealDRG™ (Table 1).

Reprogramming to iPSCs and maintenance

Peripheral blood mononuclear cells (PBMCs) from the IEM patient were reprogrammed into iPSCs using the CytoTune-iPS 2.0 Sendai Reprogramming Kit (Thermo Fisher Scientific) containing Sendai virus vectors for OCT4, KLF4, SOX2 and c-MYC (Yamanaka factors). IEM iPSCs were maintained on Geltrex (Gibco) in E8 medium prepared in-house (DMEM-F12, E8 supplement, L-glutamine and HEPES) and passaged with 0.5 mM EDTA in PBS every 3–4 days. Reprogramming of mesenchymal stromal cells from healthy subject (Ctrl1) was performed with the plasmids pCXLE-hSK, pCXLE-hUL, pCXLE- hOCT3/4-shp53 transfected by electroporation [28]. iPSCs were cultured on Vitronectin in E8 medium and passaged with 0.5 mM PBS EDTA every 3–4 days. Fibroblasts from another healthy subject (Ctrl2, resistant clone) were reprogrammed into iPSCs by episomal transduction method [29, 30] and Fibroblasts from SFN patient were reprogrammed retrovirally into iPSCs using the Yamanaka factors [25]. Both cell lines were maintained on Vitronectin (Life technologies) in iPSC Brew medium (Miltenyi biotec) and passaged with 0.5 mM EDTA in PBS every 3–4 days. Reprogramming of Ctrl1, Ctrl2, IEM and SFN donor cells was not part of this study. All iPSC clones with a passage number ranging from 11–34 were used in the study. Karyotypes were found to be normal for all the iPSC clones used in the study, further data available on request [25, 28,29,30]. All iPSC clones used in the study were regularly tested for the mycoplasma contamination.

iPSC-derived sensory neurons from Anatomic (RealDRG™)

For membrane potential FLIPR assay, multielectrode arrays, automated patch clamp and in-situ hybridization studies, Anatomic’s commercially available iPSC-derived sensory neurons (RealDRG™) were used (Table 1). RealDRGs™ were provided as terminally differentiated immature sensory neurons. These neurons were manufactured with scaled-up versions of Anatomic’s Chrono™ Senso-DM kit (Anatomic, cat# 7007). Cord blood cells were reprogrammed using episomal plasmids to generate iPSCs (ANAT001) and were maintained under fully-defined conditions before seeding onto a defined matrix. iPSC clones with a passage number ranging from 10–20 were used in the study. Cultures were fed optimized differentiation formulations daily to produce immature sensory neurons by day 7 post-induction. Day 7 cultures were dissociated and cryopreserved. Lot-specific metrics were recorded including yield, cell number per vial, viability, post-thaw recovery, post-thaw viability, post-thaw morphology, purity, and sterility. Criteria used to determine lots passing quality control included neuronal purity > 95%, verified cell number per vial, post-thaw viability > 70%, and sterility. Lots passing quality control were shipped on dry ice to end users and used according to the manufacturer’s instructions. Additional detail related to differentiation, media compositions, materials used, and bioprocessing steps are proprietary information of Anatomic.

Informed consent

Informed cosent was obtained for each iPSC clone used in the study. Ctrl1 clone: Femoral bone samples for the isolation of MSCs were collected after informed and written consent, and the study was approved by the Ethics Committee of RWTH Aachen University Medical School (EK 252/12) [28]. Ctrl2 clone: The details of the clone are available in the repository CS00iCTR21n1/GM05400, Coriell [29, 30]. SFN clone: Written informed consent from the patient was obtained having Review Board approvals Nr. 4120 UKER, Germany and Nr. 2012/2297 South East, Norway [25]. IEM clone: Reprograming was performed at Uniklinik RWTH Aachen University with written consent from the patient and her parents (ethics committee of the Medical faculty of the RWTH reference number EK243/18). ANAT001 (RealDRG™) clone: Informed consent was obtained to perform reprogramming of cord blood cells.

Differentiation of iPSCs to sensory neurons

Differentiation of all iPSCs (healthy and diseased) was carried out using Chrono™ Senso-DM (and Chambers et al. [6] with modifications designated as “Chambers protocol” in this study.

Anatomic differentiation protocol

The Anatomic differentiation protocol was used for all cell lines investigated in the study. Two variations of the differentiation were compared, one utilizing single cell seeding and one utilizing clump seeding. Ctrl1 iPSCs were differentiated with two different seeding protocols. All other iPSC lines were differentiated with single cell seeding protocol. An optimal seeding density was determined for each cell line with both single cell and clump seeding protocols to maximize yield and efficiency. For clump seeding an ideal iPSC colony size was used ranging between 25 and 50 cells per colony. iPSC cultures with a 60–80% confluence were used for plating the cells as either single cell or clump on Matrix1 (Anatomic, cat# M8001) pre-coated wells. iPSCs were seeded as single cells in a density of 15,000–80,000 cells/cm2 with 10 µM Y-27632 (Abcam Biochemicals, Bristol, United Kingdom). For both single cell and clump seeding protocols, a cocktail of small molecules was added from DIV0 (Days in vitro) through DIV7 of differentiation. Chrono™ Senso-DM 1, 2, 3, 4, 5, 6 and 7 (Small molecules, Anatomic Incorporated, cat# 7007) were added on each day of differentiation starting from DIV0. For each day of differentiation, 0.5 mL of Chrono™ Senso DM (1–7) supplement was added to 4.5 mL of Basecamp (Differentiation basal medium, Anatomic) to create 5 mL of complete differentiation medium that was fed immediately to cultures. Immature neurons generated on DIV7 were then dissociated using Accutase (Sigma cat# A6964), incubated at room temperature for ~ 1 h which ensured isolation of single neurons. 30,000–40,000 cells were then plated onto glass coverslips, coated with PDL 0.1 mg/ml (Sigma cat# P0899) and Matrix 3 (Anatomic, cat# M8003). Neurons were then supplemented with Chrono™ Senso-MM maturation medium (Anatomic, cat# 7008) from DIV7 onwards. Two-thirds medium exchanges were performed thrice weekly.

Chambers differentiation protocol

Ctrl1 cell line was differentiated with the Chambers protocol to compare the potential of differentiation efficiency and generation of sensory neurons to the Anatomic protocol. Ctrl1 iPSCs were differentiated following a previously published protocol with modifications [6, 12]. Briefly, iPSCs were seeded as single cells in a density of 30,000–40,000 cells/cm2 with 10 µM Y-27632 (Abcam Biochemicals, Bristol, United Kingdom). When cells reached 80–90% confluency, usually 24–48 h after plating, neural conversion was induced using dual-SMAD inhibition. LDN-193189 1 µM (Sigma-Aldrich) and SB431542 10 µM (Miltenyi Biotec) were added to the culture medium between DIV0-5. To accelerate neural crest specification and peripheral neuron formation from neural crest cells, three small molecules (3 µM CHIR99021, 10 µM DAPT, and 10 µM SU5402, (all Tocris, United Kingdom) were added between DIV2-10. Between DIV0-5, cells were fed with knockout DMEM/F-12 containing 15% KnockOut serum replacement, 1 mM L-glutamine, 100 µM NEAA, 100 µM β-mercaptoethanol, 100 U/ml penicillin and 100 µg/ml streptomycin (all from Thermo Fisher Scientific). Between DIV4-10, cells were fed with DMEM/F-12, containing 10 ml/l N2 (1X), 20 ml/l B27 (1X) without vitamin A supplements and 100 U/ml penicillin, 100 µg/ml streptomycin (all from Thermo Fisher Scientific). N2/B27 medium was added to basal medium at 25% between days 4–5, 50% between days 6–7 and 75% between days 8–10. The culture medium was changed daily.

Chambers protocol MACS sorting (p75 Neurotrophic receptor)

On DIV10, cells were dissociated using Accutase (Sigma, cat# A6964) and magnetic activated cell sorting (MACS) for CD271 (p75 Neurotrophic receptor) was performed as per the manufacturers protocol (CD271 MicroBead Kit human, MACS Columns and MACS Separators, Neural Crest Stem Cell MicroBeads, human: MicroBeads conjugated to monoclonal antihuman CD271 antibodies all from Miltenyi biotec, MACS buffer 0.5% BSA + 2 mM EDTA). Briefly, cells were dissociated with Accutase for 5 min at 37 °C. The single-cell suspension was then passed through 40 μm cell strainer (Corning, cat# SLS431750) to remove cell clumps. The cell suspension centrifuged at 300 g for 10 min and the cell pellet resuspended into 80 µL of buffer per 107 of total cells. 20 µL of neural crest stem cell microbeads were added (Miltenyi biotec) per 107 total cells and incubated for 15 min in the refrigerator (2–8 °C). Cell suspensions were then applied onto the column and washed 3 times with 500 µL MACS buffer. In the end, magnetically labelled cells were flushed out by firmly pushing the plunger into the column with 1 mL of MACS buffer. 30,000–40,000 cells were seeded onto glass coverslips, coated with 15 µg/ml Poly-L-Ornithine (Sigma, cat# P3655), 10 µg/ml Laminin (Sigma, cat# L2020) and 10 µg/ml fibronectin (Life technologies, cat# 33010018). N2/B27 (Life technologies, cat#17502048 and 17504044) medium supplemented with 20 ng/ml NGF, BDNF, GDNF (all from PeproTech, cat# AF-450-01, 450-02 and 450-10, respectively) and 200 µM ascorbic acid (Sigma, cat# A4544) was used for maturation from day 10 onwards. Medium was changed every 3–4 days. Laminin (500 ng/ml) was added twice weekly in the culture medium.

Immunocytochemistry

For immunostaining, differentiated neurons were seeded onto glass coverslips, coated with PDL/Matrix3 for Anatomic protocol and Poly-L-Ornithine/Laminin/Fibronectin for Chambers protocol. Cells were fixed with 4% paraformaldehyde and permeabilized and blocked with 5% goat serum (Pan biotechnology, cat# P30-1001) or 1% BSA (Sigma, cat # A9418) and 0.1% Triton X-100 (Sigma, cat# T8787) in PBS (Life technologies, cat# 14190-169). iPSC-derived neurons were then stained with anti-peripherin (Santa Cruz Biotechnology, cat# SC 377093) and anti-β-III-tubulin (TUJ-1) (Cell signaling, cat# 5568S) primary antibodies as a marker for peripherin and neuronal identity respectively. Secondary antibodies were goat anti-rabbit IgG Alexa Fluor 594 (β-III-tubulin) and goat anti-mouse IgG Alexa Fluor 488 (Peripherin). Nuclei were counterstained with DAPI (Thermo Fisher Scientific, cat# SC 377093). Fluorescent images were acquired with a LSM 700 confocal microscope (Carl Zeiss). Fluorescent images were obtained on DIV8-35 and DIV11-35 of freshly generated neurons with Anatomic and Chambers protocol respectively for Ctrl1 and Ctrl2 iPSC-derived neurons. For IEM derived cells peripherin and Tuj1 staining was performed on DIV39. For SFN derived cells peripherin and Tuj1 staining on DIV8 and 14 of freshly manufactured neurons was performed.

Manual patch clamp recordings

Whole-cell patch-clamp recordings were performed on all four iPSC-derived neurons on DIV14, 21, 28, 35 with both differentiation protocols. Data was pooled for recordings performed over a period of two days for each time point unless specified. Experiments were performed using a HEKA EPC 10 USB amplifier Patch Master and analyzed using FitMaster v2 X 91 software (all HEKA electronics, Lambrecht, Germany), Igor Pro v6.3.7.2 (WaveMetrics, USA) and GraphPad Prism v9.3.1 (GraphPad Software, Inc., La Jolla, USA). Series resistance was compensated by 30–80%. Currents were low pass filtered at 10 kHz and sampled at 100 kHz. Leak current was subtracted using the P/4 method. The liquid junction potential was corrected for both voltage and current clamp recordings. Glass pipettes (Bio-medical Instruments, Zöllnitz, Germany) were pulled with a DMZ puller (Zeitz Instruments, Martinsried, Germany) to a resistance of 1.0 to 3.5 MΩ for voltage clamp recordings and 1.5 to 4 MΩ for current clamp recordings. All experiments were performed at room temperature.

Current-clamp recordings

Current clamp recordings were performed with extracellular solution containing (in mM): 140 NaCl, 3 KCl, 1 MgCl2, 1 CaCl2, 10 HEPES, 20 glucose (pH 7.4; 300–310 mOsm) and intracellular solution containing (in mM): 4 NaCl, 135 K-gluconate, 3 MgCl2, 5 EGTA, 5 HEPES, 2 Na2-ATP, 0.3 Na-GTP (pH 7.25; 290–300 mOsm). Resting membrane potential (RMP) was recorded immediately after establishing the whole-cell configuration for 4 s. The first action potential (AP) evoked by the square pulse protocol (increments of 10 pA) was used to identify the AP properties and maturity of neurons. All recorded neurons having APs with an overshoot above 0 mV were considered mature. The AP threshold was defined as the potential at which the minimum of the first derivative of the AP (the point of inflection during the depolarization) occurs. Afterhyperpolarization (AHP) was calculated as the minimum potential recorded during the repolarization phase of AP. The number of APs generated and time to 1st AP was determined in response to ramp current stimulus of 500 pA/500 ms. Depolarisation current ramps of 1 nA were given over 100–1000 ms to assess firing in response to slow depolarisation. For calculating the number of APs in response to ramp current injections, APs with an overshoot above 0 mV were counted as APs. After measurement in the current-clamp configuration, the amplifier was switched to the whole cell voltage-clamp mode to measure both sodium and potassium currents. Voltage dependence of activation for both Na+ and K+ currents were recorded with a 500 ms pulse from − 80 mV to 40 mV in 10 mV steps from a holding voltage of − 90 mV.

Voltage-clamp recordings

To isolate voltage-gated sodium currents, experiments were then performed in the presence of K+ and Ca2+ ion channel blockers with extracellular solution containing (in mM): 140 NaCl, 1 MgCl2, 1 CaCl2, 10 HEPES, 1 glucose, 20 TEA-Cl, 1 4-aminopyridine, 0.1 CdCl2 (pH 7.4; 300–310 mOsm) and the intracellular solution containing (in mM): 140 CsF, 10 NaCl, 10 HEPES, 1 EGTA, 5 glucose, 5 TEA-Cl (pH 7.3; 290–300 mOsm). Tetrodotoxin resistant (TTXr) currents were recorded in the presence of 500 nM TTX (Tocris Bioscience) diluted in extracellular solution. Voltage dependence of activation of TTXr currents was determined with a 100 ms pulse from − 90 mV to 40 mV in 10 mV increment steps from a holding voltage of − 120 mV.

To assess the presence of Nav1.8 currents, 1 µM of the selective Nav1.8 blocker A-887826 (provided by Grünenthal) was used on DIV31 of maturation from Ctrl1 iPSC-derived sensory neurons [31]. A-887826 was dissolved in DMSO as 10 mM stock solution and the final maximum concentration of DMSO was 0.1%. Solutions were applied through a gravity-driven perfusion system. A single pulse voltage protocol from a holding potential of –120 mV to –30 mV repeated every 10 s was used. Cells were patched in the presence of 500 nM TTX. After obtaining a stable baseline recording with 500 nM TTX, 1 µM A-887826 was applied to check the percent inhibition of currents. In one cell, washout with extracellular solution (ECS) was performed to obtain the total sodium currents at the end of the experiment.

In situ hybridization

RNAscope in situ hybridization was used to characterize expression of key nociceptor markers in RealDRG™. RealDRG™ iPSC-derived neurons were plated in 8-well chamber slides (Thermo Scientific, cat# 154534) coated with 0.1% PLO (Sigma-Aldrich, cat# P4957)/Matrix 3 (Anatomic) and maintained with Anatomic protocol instructions. In situ hybridization was completed using the RNAscope procedure (multiplex version 1 assay (320851)) utilizing manufacturer’s (Advanced Cell Diagnostics ACD) published protocols [32]. On DIV14 and DIV16, chambers were disassembled from the slide, and the cells were fixed in 10% neutral buffered formalin for 30 min at room temperature. Slides were then washed twice in 1X PBS, and boundaries were drawn around each well using the hydrophobic ImmEdge PAP pen (Vector Labs cat# H-4000). Slides were washed again in 1X PBS before being incubated in protease III reagent (1:30 in 1X PBS) for 10 min at room temperature in a humidity control tray. Slides were washed twice in 1X PBS and then placed in a prewarmed humidity control tray with dampened filter paper to be incubated with probe mixtures for 2 h at 40 °C. DIV14 slides were incubated with Channel 1 NTRK1 (Neurotrophic Receptor Tyrosine Kinase 1, nociceptor marker gene) (ACD cat# 402631), Channel 2 TAC1 (Tachykinin precursor 1 codes for neurokinin A and substance P, neuropeptidergic neuronal marker) (ACD, cat# 310711), and Channel 3 HCN2 (Hyperpolarization-activated cyclic nucleotide-gated 2 ion channels) (ACD, cat# 517021). DIV16 slides were incubated with Channel 1 SCN10A (Sodium ion channel Nav1.8, marker for nociceptors specifically expressed in primary sensory neurons) (ACD, cat# 406291), Channel 2 TAC1 (ACD, cat# 310711), and Channel 3 TRPV1 (Transient receptor potential vanilloid subfamily member 1, nociceptor marker gene) (ACD cat# 451381). Slides also had one well each for positive (ACD, cat# 320861) and negative (ACD, cat# 320871) control probes. Following probe incubation, slides were washed twice in 1X RNAscope wash buffer and incubated in AMP-1 reagent for 30 min at 40 °C. Washes and incubation were repeated for AMP-2, AMP-3, and AMP-4A for 15 min, 30 min, and 15 min, respectively. After amplification, slides were washed in 0.1 M phosphate buffer (PB, pH 7.4) and stained with DAPI (Cayman Chemical, cat# 14285). Slides were then washed twice in 0.1 M PB, air dried, and cover-slipped with Prolong Gold Antifade (Fisher Scientific, cat# P36930) mounting medium. Images were acquired on an Olympus FV1200 confocal microscope using a 40X objective and analyzed using Cellsens software (Olympus).

FLIPR (high-throughput plate reader assays)

For high-throughput fluorescent imaging assays using the FLIPRPenta (Molecular Devices), RealDRG™ were thawed and cultured in Chrono™ Senso-MM Cells were seeded at a density of 1000–10,000 cells/well on black-walled 384-well imaging plates (Corning, cat# CLS3657) pre-coated with Matrix3 and cultured in Anatomic Chrono™ Senso-MM. To record fluorescence responses following stimulation with agonists, cells were loaded with Calcium 4 dye kit (Molecular Devices, cat# R8141) diluted according to the manufacturer’s instructions in physiological salt solution (PSS, composition in mM: NaCl 140, glucose 11.5, KCl 5.9, MgCl2 1.4, NaH2PO4 1.2, NaHCO3 5, CaCl2 1.8, HEPES 10) and incubated at 37 °C for 30 min. Responses were measured every 1 s for 300 s and, analysed using ScreenWorks 5.1.1.86 (Molecular Devices).

Automated Patch clamp recordings

All experiments were performed on Qube384 (Sophion Bioscience A/S). Experiments were executed with single-hole QChips in a format of 48, 120 or 384 simultaneous wells based on the harvest cell numbers. RealDRG™ were thawed and cultured in Chrono™Senso-MM for Automated Patch clamp recordings (APC). Patch clamp recordings were performed on DIV16, 21, 28 and 35 days of maturation. Cells were dissociated using papain (Worthington, cat# LK003150) at 3 units/ml overnight as per the protocol developed by Anatomic. After whole-cell formation controlled with Sophion software, application protocol consisting of both voltage and current clamp protocols were performed at the control condition. Extracellular solution containing (in mM): NaCl 145, CaCl2 2, MgCl2 1, KCl 4, HEPES 10, and Glucose 10, pH 7.4 and intracellular solution containing (in mM): KF 120, KCl 20, HEPES 10, and EGTA 10, pH 7.3 were used for current clamp recordings. To isolate Na+ currents, CsF intracellular solution was introduced by using Qube-384 intracellular solution exchange protocol. CsF internal solution contained (mM): CsF 135, NaCl 10, HEPES 10, EGTA 1.0, pH 7.3 with CsOH. 0.5 µM TTX (Alomone Labs) and 1 µM or 10 µM A-803467 (Millipore-Sigma) were used for Na+ channel characterization. Cells with a membrane resistance (Rm) > 200 MΩ and a cell capacitance Cslow > 2 pF were included in the analysis.

Multi-well microelectrode arrays

Multi-well microelectrode arrays (MEA) were used to evaluate RealDRG™. The day before starting the culture, a 48-well MEA plate (Axion Biosystems, cat# M768-tMEA-48W) was coated with Poly-L-Ornithine (0.01%, EMD Millipore Sigma, cat#A-004-C) and incubated overnight at room temperature. After 3 washes with sterile deionized water, each well was coated with a Matrix 3 (1:50 dilution with dPBS (-/-) from Anatomic, cat#M8003) and incubated for 3 h at 37 °C. RealDRG™ were thawed and cultured in Chrono™ Senso-MM. Cells were resuspended in 2 mL of Chrono™ Senso-MM complete growth medium and counted for a seeding density of 40 k cells/cm2. Immediately prior to seeding, excess Chrono™ Matrix 3 (from plate preparation step) was removed from each well. Cells were added to wells at a total volume of 400 µl in Chrono™ Senso-MM. Chrono™ Senso-MM media was exchanged at 50% from a 400 µl volume every other day and cells were grown at 5% CO2 at 37 °C. MEA electrophysiology data were acquired from an Axion Maestro Classic system at 12.5 kHz sampling rate and processed with a single pole Butterworth bandpass filter (300–5000 Hz). Individual spikes were detected from filtered continuous voltage recordings where exceeding ± 5.5σRMS threshold based on a continuous 1 s data block to estimate σRMS on a per channel basis. Recordings from the MEA plate were performed for 3 consecutive days each week for 4 weeks after plating. During each session, baseline recordings (at 37 °C) were acquired for a duration of 30 min followed by a temperature ramp up to 42 °C with the embedded heating plate. The time to 42 °C was 2.5 min and typical decay back to 37 °C was 3.5 min. For the sake of classifying cells over time in response to temperature ramps, we grouped recordings into different classes: “Consistent responders”, which includes recordings where the mean firing rate (MFR in Hz) showed elevations from baseline to all 3 temperature ramps for a given week; “responders” were those recordings demonstrating an elevation in at least 1 or 2 of the 3 sessions; “negative responders” were recordings demonstrated spontaneous activity, however there was a decrease in the MFR during temperature ramps; and “inactive electrodes” where the recordings failed to demonstrate spontaneous or temperature ramp-evoked spikes.

Statistical analysis

Statistical analysis was performed using GraphPad Prism v9.3.1 (GraphPad Software, Inc.). Two groups were compared by a Mann–Whitney or Multiple t tests. Comparisons between three or more groups were performed using a one-Way or two-Way ANOVA followed by Bonferroni’s, Sidak’s or Tukey's multiple comparisons test. Data are presented as mean ± standard error of the mean (SEM) where P values < 0.05 were considered significant.