Effects of Complex I Inhibition on the Architecture of Neural Rosettes Differentiated from Human-Induced Pluripotent Stem Cells

Abstract

Orchestrated changes in cell arrangements and cell-to-cell contacts are susceptible to cellular stressors during central nervous system development. Effects of mitochondrial complex I inhibition on cell-to-cell contacts have been studied in vascular and intestinal structures; however, its effects on developing neuronal cells are largely unknown. We investigated the effects of the classical mitochondrial stressor and complex I inhibitor, rotenone, on the architecture of neural rosettes—radially organized neuronal progenitor cells (NPCs)—differentiated from human-induced pluripotent stem cells. We then analyzed the effects of rotenone on the distribution of cell-contact proteins within neural rosettes. Exposure to rotenone for 24 hours led to a dose-dependent irreversible disruption of the neural rosette architecture and relocalization of the cell-contact proteins ZO-1, β-catenin, and N-cadherin from the rosette center to the pericellular region. Though the levels of nestin and SOX2 remained unchanged, NPCs showed decreased levels of the NPC marker PAX6 and exhibited impaired neurogenesis following rotenone exposure. Our study suggests that complex I inhibition leads to a rearrangement of intercellular contacts with disruptive effects on neuronal development.

Introduction

Neurodevelopment is a complex process involving orchestrated changes in cell morphology and cell arrangements. Disruption of these processes can lead to severe developmental abnormalities. Early in embryonic development, neuroepithelial cells serve as the primary neural stem cells that further give rise to specialized progenitors known as radial glia.^1–3 Radial glia are highly polarized cells that undergo both symmetric and asymmetrical division to temporally regulate the pool of available progenitors, facilitate neurogenesis, and modulate neuronal migration.^1–7 This tightly regulated balance of cell proliferation and differentiation plays a critical role in the development of cortical architecture.

Intercellular contact complexes play an integral role in maintaining cellular polarity and facilitating neural differentiation.^8,9 While there are several types of cell-to-cell contact complexes, neuroepithelia and radial glia rely largely on adherens junction proteins, such as cadherins, as well as the tight junction protein, ZO-1.¹⁰ N-cadherin has unique roles in neurodevelopment, including participating in cellular signaling as well as modulating cellular architecture, cell migration, cell polarization, and asymmetrical cell division during neurulation.^11–13 Interactions between N-cadherin’s intracellular domain and a variety of partners regulate N-cadherin’s adhesive characteristics and its ability to modulate developmentally relevant pathways.¹⁴ One prominent intracellular interaction partner of N-cadherin is β-catenin, a structural component of adherens junctions that also functions as a transcriptional regulator in the canonical Wnt signaling pathway.^15,16

These tightly regulated cellular processes are highly susceptible to environmental factors and cellular stress during neurodevelopment.¹⁷ Several neurodevelopmental disorders are hypothesized to result from the interaction between underlying genetic risk and physiological stressors that result in altered neural differentiation and function.^18,19 One signature stressor in several neurodevelopmental disorders is oxidative stress, which results in elevated levels of reactive oxygen species (ROS).²⁰ Mitochondria serve as a major source of these ROS, which form as by-products of oxidative phosphorylation via the electron transport chain.²¹ Mitochondrial stress can contribute significantly to baseline levels of oxidative stress.^22–24 This combination of mitochondrial stress and oxidative stress creates a feed-forward cycle that has been described in several neurodegenerative and neurodevelopmental disorders.^23–27

Despite the evidence of aberrant neurodevelopment in the presence of oxidative and mitochondrial stress, the mechanisms by which these stressors influence neurodevelopment remain unclear.²⁸ Here, we investigate the impact of rotenone, a mitochondrial complex I inhibitor commonly used to induce mitochondrial dysfunction and oxidative stress, on the cellular architecture of developing neural cells differentiated in vitro from human-induced pluripotent stem cells (iPSCs).²⁹ For this, we differentiated iPSCs to generate neural progenitor cells (NPCs) that organize radially around luminal centers into structures, known as neural rosettes, that resemble the most primitive form of the neural tube and replicate critical aspects of embryonic neurulation.^30–33 We then exposed iPSC-derived NPCs to rotenone and characterized subsequent effects on the neural rosette morphology to better understand the impact of oxidative and mitochondrial stress on neural rosette architecture and cell-to-cell contacts during early neurodevelopment.

Methods

Fibroblast cell culture

Human fibroblast lines were derived from healthy donors recruited as control subjects with institutional review board approval.^34,35 A total of five different lines were used, four males and one female. Fibroblasts were grown in minimum essential medium (MEM Gibco 11090081) containing 15% fetal bovine serum (Sigma-Aldrich F0926), GlutaMAX™ (Gibco 35050061), nonessential amino acids (MEM NEAA Gibco 11140050), and antibiotic–antimycotic (Gibco 15240-062) and passaged at a ratio of 1:4 upon reaching ∼80% confluence. Fibroblasts were used at passage numbers between 10 and 17. For imaging, cells were either plated on glass bottom 96-well (Corning 4580) or 24-well plates (Greiner SensoPlate™ 662892). Rotenone treatments were performed on cells at ∼70% confluency.

Human iPSC culture

Human iPSCs were reprogrammed from healthy donor fibroblasts via synthetic RNA (Stemiotics) as described previously and cultured in Stemflex (Gibco A3349401) in the presence of an antibiotic–antimycotic on Matrigel (Corning 356230) coated surfaces.^36,37 At 75% confluency, iPSCs were split with Versene (Gibco 15040066) and reseeded at ∼20% confluency or at the necessary cell density for specific experiments. iPSCs were used at passage numbers between 13 and 22. All iPSCs were verified to be mycoplasma free, karyotypically normal, express OCT4/NANOG, and form all three germ layers.

Generation of NPCs from iPSCs

NPCs were generated from iPSCs using dual SMAD inhibition adapted from a standard published protocol.^30,38 Briefly, iPSCs (passages 12–15) were split onto Matrigel-coated plates at a density sufficient to reach ∼100% confluency within 1–3 days. At 100% confluency, cells were switched to neural induction media. Neural induction media consisted of a standard base N2/B27 media³⁰ supplemented with 10 μM SB431542 (Sigma-Aldrich P5499), 1 μM dorsomorphin (Peprotech 3014193), and 100 nM LDN193189 (Peprotech 1066208), with media changed every day. Between days 8 and 12, cells were monitored for the formation of a neuroepithelial sheath and split 1:1. Following the initial split, cells were split every 3–4 days and fed daily with N2/B27. Upon visible rosette formation (∼days 13–18), cells were fed every other day with N2/B27 supplemented with fibroblast growth factor (FGF) (20 ng/mL). NPCs used for subsequent experiments were within 22–30 days old, exhibited rosette structure, and exhibited less than 30% neurons as observed through brightfield microscopy.

Rotenone treatments

A stock solution of 50 mM rotenone (Tocris 3616) in dimethyl sulfoxide (DMSO) was used. Prior to treatments, cells were seeded and cultured for a minimum of 48 hours. Following seeding, NPCs, iPSCs, or fibroblasts were treated for 24 hours in their respective growth media containing DMSO, 0.5, 1, 5, 10, or 50 μM rotenone. Cells were either fixed or harvested directly after 24 hours of treatment. For recovery experiments, following a 24-hour treatment, cells were washed twice with Hanks' Balanced Salt Solution (HBSS) and cultured for an additional 48 hours with daily medium changes before being fixed or harvested.

Immunocytochemistry

Cells were fixed with 4% paraformaldehyde diluted in phosphate buffer saline (PBS) for 10–20 minutes at room temperature (RT), followed by three washes in PBS. Cells were permeabilized in 0.1% Triton X-100 diluted in water for 10 minutes and immediately blocked in 5% bovine serum albumin (BSA) diluted in PBS for 1 hour at RT. Primary antibodies were diluted in 1% BSA in PBS before incubating overnight at 4°C. Primary and secondary antibodies are listed in Supplementary Table S1. Following incubation with primary antibodies, cells were washed three times with PBS and then incubated for 1 hour in the dark at RT with the appropriate secondary antibody diluted in 1% BSA. F-actin staining was performed after the secondary antibody staining and before the nuclear staining by incubating the cells with phalloidin (ThermoFisher Scientific A-12379) diluted at 1:550 in PBS for 20 minutes at RT, followed by two PBS washes. Cells were incubated with nuclear stain Hoechst 3342 (Invitrogen H3570) at 1:5000 diluted in PBS for 2 minutes at RT, followed by three washes in PBS. Confocal microscopy was performed using an INCell Analyzer 6000 Cell Imaging System. Single-plane images were acquired using an automated focus. All images were analyzed using ImageJ.

Neural rosette quantification

Light microscopy was performed on a Zeiss Observer A1 inverted phase contrast fluorescence microscope using an LD Plan-NEOFLUAR 20x/0.4Ph2 objective applying differential interference contrast. Images were acquired with an AxioVision (release 4.8.2) controlled camera. NPCs were grown on glass-bottom 96- (Corning 4580) or 24-well plates (Greiner SensoPlate™ 662892) and imaged live or after fixation. To assess the quantity of neural rosettes upon varying treatment conditions, images of 7–9 fields of view were acquired per experiment and condition. The rosette centers per field of view were counted manually in ImageJ.

Western blot

Following treatments, cells were washed with PBS and lysed with radioimmunoprecipitation assay buffer containing protease inhibitor. Protein concentration was measured in each sample using a Pierce BCA Protein Assay (ThermoFisher Scientific 23225). Equal quantities of protein were loaded and were run on an SDS-PAGE gel. Protein was then transferred onto an Immobilon-FL PVDF membrane (Fisher Scientific FiIPFL00010). Following the transfer, the membranes were dried for 1 hour or overnight and then reactivated with methanol prior to use. Membranes were blocked for 1 hour at RT with intercept (TBS) blocking buffer (LI-COR 927–60001) and then incubated in the indicated primary antibody overnight at 4°C. The following day, membranes were washed 3X with TBS-T (tris buffer saline with 0.1% tween-20) and incubated with the secondary antibody for 1 hour in the dark at RT. Prior to imaging, membranes were washed 3X with TBS-T followed by one wash with TBS. Membranes were then imaged using a LI-COR Odyssey DLx Imager. Western blots were analyzed using LI-COR’s Image Studio software. All primary and secondary antibodies utilized for western blotting assays are found in Supplementary Table S2.

RNA extraction, cDNA synthesis, and real-time quantitative polymerase chain reaction

Following treatment, total RNA was extracted using the RNeasy Mini Kit (Qiagen 74106) according to manufacturer instructions. For all samples, 1 μg of total RNA was used for cDNA synthesis with the iScript Reverse Transcription Supermix for real-time quantitative polymerase chain reaction (RT-qPCR) (BioRad 1708840). RT-qPCR was performed with SsoAdvanced Universal SYBR^® Green Supermix (BioRad 1725270) and analyzed on a BioRad CFX96 RT-PCR thermocycler. The following primers were used to amplify human genes: GAPDH: forward primer 5′-ACAGTTGCCATGTAGACC-3′, reverse primer 5′-TTGAGCACAGGGTACTTTA-3′; ZO-1: forward primer 5′-AGGGGCAGTGGTGGTTTTCTGTTCTTTC-3′, reverse primer 5′-GCAGAGGTCAAAGTTCAAGGCTCAAGAGG-3′; CTNNB1 (β-catenin): forward primer 5′-TCTGAGGACAAGCCACAAGATTACA-3′, reverse primer 5′-TGGGCACCAATATCAAGTCCAA-3′; CDH2 (N-cadherin): forward primer 5′-GCGTCTGTAGAGGCTTCTGG-3′, reverse primer 5′-GCCACTTGCCACTTTTCCTG-3′; PAX6: forward primer 5′-CACACCGGTTTCCTCCTTCA-3′, reverse primer 5′- GGCAGAGCGCTGTAGGTGTTT-3′; Nestin: forward primer 5′-TGGAGGCAAAGAGGGTTCAG-3′, reverse primer 5′-TCGGAGAACTCTGTCCCCAG-3′. Relative gene expression was quantified using the Livak method (2-ΔΔCT) with GAPDH as the reference gene and the average DMSO ΔCT value of all the lines serving as the calibrator.

Lactate dehydrogenase assay

Cytotoxicity was assessed by indirectly measuring the lactate dehydrogenase (LDH) activity in the cell culture medium using a colorimetric LDH Assay Kit (Abcam ab65393) according to manufacturer instructions. Four independent NPC lines (day 24 ± 2) and three independent iPSC lines were seeded at a density of 50,000 cells per well, and three independent human fibroblast lines at a density of 12,500 cells per well on flat bottom 96-multiwell plates in their specific culture media. After a recovery period (2 days for NPCs and iPSCs; 3 days for fibroblasts), cells were incubated for 24 hours with medium only or DMSO and the following concentrations of rotenone: 0.5, 1, 5, 10, or 50 μM. The medium solutions (10 μL test sample and 11 μL of lysed cells representing the high control) were incubated for 45 minutes at RT with the reaction solution. The absorbance was measured with a Spectramax plate reader set at 450 nm. The percentage of cytotoxicity was calculated as per the following equation: Cytotoxicity (%)= ([(Test sample − Low control)/(High control − Low control)] × 100).

Seahorse XF Mito Stress test

To study the effects of rotenone on mitochondrial function in the NPCs, the Seahorse XF Cell Mito Stress test (Agilent Inc. 103015-100) was implemented on a Seahorse XF Pro Analyzer with appropriate cartridges/microplates (Agilent Inc. 103793-100). NPCs were thawed (day 25 of differentiation) and recovered in NPC growth media (N2/B27 supplemented with 20 ng/mL FGF) for 5 days. NPCs were subsequently seeded at a density of 150,000 cells per well into Geltrex-coated Seahorse XF microplates (Agilent) and allowed to recover for 48 hours. NPCs were then treated with DMSO, 0.5, 1, 5, 10, or 50 μM rotenone in their respective growth media for 24 hours (six technical replicates per condition). In parallel, Seahorse sensor cartridges were hydrated with XF calibrant solution (Agilent Inc. 103680–100) in a 37°C, non-CO₂ incubator overnight. Cell culture media was removed the following day, and the cells were incubated in standard assay medium for 1.5 hours prior to assay in a non-CO₂ incubator. Standard assay medium consisted of XF DMEM solution, 10 mM glucose, 1 mM pyruvate, 2 mM glutamine, pH 7.4 (Agilent Inc. 103680-100). During incubation, drugs were reconstituted fresh to create 100 μM oligomycin, 100 μM carbonyl cyanide-p-triphenyl (FCCP), and 50 μM rotenone/antimycin A stock solutions in the assay medium. For loading into the hydrated sensor cartridges, drugs were further diluted to create 10X working solutions (i.e., final concentration of 2.5 μM oligomycin, 2.0 μM FCCP, and 0.5 μM rotenone/antimycin A per well). The XF Mito Stress test was run on the loaded sensor cartridge and cell culture microplate using an Agilent XF Pro Analyzer in accordance with manufacturer guidelines. Following completion of the run, the assay medium was removed, and cells were immediately frozen at −80°C overnight. Postassay normalization was conducted using the CyQUANT cell proliferation assay kit (Invitrogen C7026). Briefly, on the day of the assay, CyQUANT cell lysis buffer was diluted 20-fold in sterile PBS and supplemented with GR dye at a 100-fold dilution to increase the linear detection range to 200,000 cells per well. Frozen cells were allowed to come to RT prior to treatment with 200 μL of the cell lysis buffer/GR dye solution for 2–5 minutes in the dark. The well contents were subsequently transferred to an optically clear 96-well microplate (Costar). Measurements were obtained on a fluorescence microplate reader (Varioskan) with excitation = 485 ± 12 nm and emission = 530 ± 12 nm. The DNA content of each well was calculated from a standard curve generated from bacteriophage gamma DNA. Data were normalized to total DNA content (μg/mL) in the Seahorse Wave Pro software and exported to GraphPad Prism for statistical analysis.

Mitochondrial imaging with MitoTracker

NPCs (day 25 of differentiation) were thawed and recovered for 5 days in NPC growth media. Overall, 20,000 cells were seeded in each well of optically clear 96-well microplates. Following a 48-hour recovery, cells were treated with DMSO, 0.5, 1, 5, 10, or 50 μM rotenone in their respective growth media for 24 hours (six technical replicates per condition). NPCs were subsequently incubated with 500 nM MitoTracker CM-H2XRos dye (Invitrogen M7513) in NPC growth media for 30–45 minutes at 37°C, 5% CO₂. Following incubation, the staining solution was replaced with NPC growth media, and plates were imaged immediately on a high-content confocal microscope (Yokogawa CQ1). Single-plane images were autofocused and analyzed using ImageJ. Fluorescence intensity was normalized to the total nuclei count per well. For total nuclei count, wells were fixed in 4% paraformaldehyde diluted in PBS for 10 minutes at RT immediately following MitoTracker imaging. Cells were then washed twice in PBS for 10 minutes, permeabilized in 0.1% Triton X-100 for 10 minutes at RT, then washed once in PBS for 10 minutes and incubated with nuclear stain Hoechst 3342 (1:2,000 in PBS) for 10 minutes at RT in the dark. This was followed by three 5-minute washes in PBS and imaging on the Yokogawa CQ1 high-content confocal microscope.

Neuronal differentiation from NPCs

NPCs (day 30 of differentiation) were seeded into 12-well plates at a density of 1,000,000 cells per well following a 5-day post-thaw period in NPC growth media. On day 33, NPCs were treated with DMSO, 0.5, 1, 5, 10, or 50 μM rotenone in their respective growth media for 24 hours (two technical replicates per condition). Cells were then reseeded into optically clear 96-well microplates at a density of 20,000 cells per well (six technical replicates per condition). Full media changes were conducted carefully at the indicated time points. For the first 2 days of neuron differentiation (corresponding to days 34–36 differentiation), cells were grown in neuron seeding media consisting of N2/B27 supplemented with 10 μM DAPT (STEMCell Technologies 72082), 20 ng/mL BDNF (Peprotech 450-02-50UG), 20 ng/mL GDNF (Peprotech 450-10-10UG), 200 μM ascorbic acid (Sigma Alrich A8960-5G), and 1X CultureOne supplement (Gibco A3320201). Between days 37 and 38, cells received Neuronal Differentiation Media #1, consisting of N2/B27 supplemented with 10 μM DAPT, 10 μM PD0325901 (Fisher/ADOOQ Biosciences A10256-5MG), and 1X CultureOne supplement. On day 39, media was carefully replaced with Neuronal Differentiation Media #2 for 24 h, consisting of N2/B27 supplemented with 5 μM AraC (MedChemExpress HY-13605) and 10 ng/mL BDNF. Between days 40 and 42, cells were treated with Neuronal Differentiation Media #3, consisting of N2/B27 with 10 ng/mL BDNF and 1X CultureOne supplement.

On day 42, cells were fixed with a 10-minute incubation in 4% paraformaldehyde diluted in PBS at RT. Neurons were then washed twice in PBS for 10 minutes and permeabilized with 0.1% Triton X-100 for 10 minutes. Following a 10-minute wash in PBS, cells were blocked in 3% BSA diluted in PBS for 1.5 hours at RT. Primary antibodies were diluted in 1X PBST (phosphate buffered saline-tween, 0.1% Tween-20) and applied to samples overnight at 4°C (Supplementary Table S1). The following day, cells were washed once with PBS for 15 minutes and incubated with secondary antibodies (1:500) diluted in PBST for 1 hour at RT in the dark (Supplementary Table S1). This was followed by a 5- to 10-minute wash in PBS and incubation with Hoechst 3342 (1:2000 in PBS) for 10 minutes at RT in the dark. Cells were subsequently washed 3X in PBS for 5 minutes each and imaged within 24–48 hours on a high-content microscope (GE INCell).

Statistical analysis

Statistical analysis was performed using GraphPad Prism Version 10. Data were first analyzed for normality using GraphPad Prism’s Shapiro–Wilk test. Data that displayed normal distribution were analyzed using a repeated measures one-way Analysis of Variance (ANOVA) or a mixed-effects analysis (for data with missing values) with a confidence interval set at 95%. All parametric analyses were matched by cell line and followed by a Dunnet’s multiple comparisons test with a single pooled variance. Nonparametric data were matched by cell line and analyzed using GraphPad Prism’s Friedman test followed by Dunn’s multiple comparisons test.

Results

Complex I inhibition by rotenone leads to mitochondrial dysfunction

Mitochondrial and oxidative stress are among the most common cellular stressors encountered along neurodevelopment.^20,28,39,40 To model this in vitro, we utilized rotenone, a potent complex I inhibitor previously shown to result in the formation of ROS and mitochondrial stress.^41–47 To confirm that rotenone induces ROS in NPCs, we examined the effects of rotenone exposure in NPCs differentiated from iPSCs that were previously characterized and reprogrammed from healthy individuals.^48,49 NPCs were differentiated from iPSCs using dual SMAD inhibition and treated following rosette formation (DIV 20–28) with a range of rotenone concentrations (0, 1, 5, 10, 50 µM). NPCs were then examined using MitoTracker CM-H2XRos dye, a reduced mitochondrial probe whose fluorescence indicates the presence of ROS.⁵⁰ As expected, we detected a trend toward higher levels of MitoTracker CM-H2XRos dye in the presence of low concentrations of rotenone (0.5 µM); however, at higher concentrations (10 µM), MitoTracker dye levels decreased (Fig. 1A and B). As the accumulation of the MitoTracker probe is partially dependent on the mitochondrial membrane potential,⁵¹ we reasoned this drop may have resulted from decreased potential at higher rotenone concentrations, as has been noted in studies of other cell types.^52–54 To further interrogate the effects of rotenone on mitochondrial stress and dysfunction in these human iPSC-derived NPCs, we performed a Seahorse XF Cell Mito Stress test to measure oxygen consumption rate (OCR) and extracellular acidification rates (ECAR).⁵⁵ These data showed that increasing rotenone concentrations led to alterations in the OCR, without any effects on the ECAR (Fig. 1C and D). Analysis of additional mitochondrial parameters showed that the increasing rotenone concentrations impaired basal, maximum, and spare respiration (Fig. 1E–G). Rotenone also led to significant decreases in the average adenosine triphosphate (ATP) production, though the average coupling efficiency and proton leak were not significantly altered (Fig. 1H–J). Taken together, these data show that rotenone impairs mitochondrial functioning in human NPCs, in accordance with previous findings in other cell types.

FIG. 1.

Rotenone impairs mitochondrial activity in human NPCs. (A) NPCs exhibiting neural rosette architecture were treated with DMSO, 0.5, 1, 5, 10, and 50 μM rotenone for 24 hours and visualized with MitoTracker CM-H₂XRos, a reduced mitochondrial probe. Representative images are shown (n = 3). Scale bar shows 100 μm. (B) Quantification of MitoTracker in NPCs. Average fluorescence intensity of MitoTracker CM-H₂XRos normalized to total nuclei count. Data were analyzed using a Friedman test (P ≤ 0.05) and Dunn’s multiple comparisons test relative to the DMSO control. Error bars indicate standard error of the mean (SEM). (C, D) NPCs exhibiting neural rosette architecture were treated with DMSO, 0.5, 1, 5, 10, and 50 μM rotenone for 24 hours and then analyzed via a Seahorse XF Mito Stress Test. Lines indicate injections of oligomycin (2.5 μM), FCCP (2.0 μM), and rotenone/antimycin A (0.5 μM), respectively. Oxygen consumption rate (C) (pmol/min) and extracellular acidification rate (D) (mpH/min) during Seahorse Mito Stress test (n = 3, 6 replicates per line) are shown. Data were analyzed using a Friedman test (P ≤ 0.05) and Dunn’s multiple comparisons test relative to the DMSO control. Error bars indicate SEM. (E–J) Metabolic measurements calculated from the Seahorse Mito Stress test. Average basal respiration (E), defined as the difference between the last baseline measurement and the nonmitochondrial oxygen consumption rate. Average maximum respiration (F), defined as the difference between the maximum rate measurement following FCCP injection and the nonmitochondrial oxygen consumption rate. Average spare respiratory capacity (G), defined as the difference between the maximum and basal respiration rates. ATP production rate (H), defined as the difference between the last baseline measurement and the minimum rate measurement following oligomycin injection. Average coupling efficiency (%) (I) calculated from (C) and (H), defined as the ATP production rate normalized to the basal respiration rate. Average proton leak (J), defined as the difference between the minimum rate measurement following oligomycin injection and the nonmitochondrial oxygen consumption rate. Data were analyzed using a Friedman test (P ≤ 0.05) and Dunn’s multiple comparisons test relative to the DMSO control. Error bars indicate SEM. P ≤ 0.05 (*). DMSO, dimethyl sulfoxide; NPC, neuronal progenitor cell.

Complex I inhibition disrupts the neuronal rosette architecture

To examine the effect of complex I inhibition on cellular morphology during neural development, we treated NPCs displaying neural rosette structures with a range of rotenone concentrations (0, 1, 5, 10, 50 µM) for 24 hours and visualized them under brightfield microscopy (Supplementary Fig. S1.A–B).^30,56 We found that even the lowest concentration of rotenone (0.5 µM) resulted in fewer neural rosettes, while rotenone concentrations above 5 µM led to a complete disruption of the neural rosette architecture (Fig. 2A and B and Supplementary Fig. S1B). Despite significant dose-dependent effects on the neural rosettes, rotenone at doses up to 10 µM did not result in significant cytotoxicity as measured using an LDH assay (Fig. 2C).

FIG. 2.

Rotenone disrupts neural rosette architecture. (A) NPCs exhibiting neural rosette architecture were treated with DMSO, 5 μM rotenone, and 50 μM rotenone for 24 hours. NPCs were fixed and imaged using brightfield microscopy. Rosette centers are indicated by pink dots. Representative images are shown (n = 4). (B) Neural rosette counts of NPCs treated with DMSO, 0.5, 1, 5, 10, and 50 μM rotenone for 24 hours. Rosette centers were manually counted, graphed, and analyzed using GraphPad Prism. Data were analyzed using a mixed-effects model (P ≤ 0.05) and a Dunnett’s multiple comparison test relative to the DMSO control (n = 4). Error bars indicate standard error of the mean (SEM). P ≤ 0.01 (**) and P ≤ 0.0001 (##). (C) NPCs were treated with DMSO as a control and the following concentrations of rotenone: 0.5, 1, 5, 10, and 50 μM. Positive control is indicated by (+). Following treatment, NPCs were subjected to an LDH assay to assess toxicity. Data were analyzed using a Friedman test (P ≤ 0.01) and Dunn’s multiple comparisons test relative to the DMSO control (n = 5). Error bars indicate (SEM). P ≤ 0.05 (*) and P ≤ 0.001 (#).

Complex I inhibition leads to a relocalization of junctional proteins

During neural differentiation, adherens junction proteins undergo structural rearrangements from a homogenous distribution over the entire cell circumference to concentrated proteins at the neural rosette centers.^57–60 A similar pattern has been described for ZO-1, the main tight junction protein expressed during neurodevelopment.^7,31,59 As these junctional proteins and the actin cytoskeleton are crucial for maintaining neural rosette structure, we sought to determine if rotenone disrupted neural rosette structures via alterations to the actin cytoskeleton and the junctional proteins ZO-1, N-cadherin, and β-catenin.

NPCs were treated with different concentrations of rotenone for 24 hours and then analyzed for localization of ZO-1, N-cadherin, and β-catenin. As expected, under normal conditions, all three proteins and F-actin localized to the luminal center of the neural rosettes (Fig. 3A). Rotenone treatment led to a dose-dependent relocalization of F-actin, ZO-1, N-cadherin, and β-catenin to the cell periphery (Fig. 3A) without an overall effect on total protein levels, although a decreasing trend was observed for N-cadherin (Fig. 3B–E). In addition, we did not detect any changes in the mRNA levels of genes for any of these proteins following rotenone exposure (Supplementary Fig. S2).

FIG. 3.

Junctional protein localization is altered following rotenone exposure. (A) NPCs exhibiting neural rosette architecture were treated with DMSO, 5 μM rotenone, and 50 μM rotenone for 24 hours. NPCs were fixed and imaged for ZO-1, β-catenin, and N-cadherin, shown in red. Hoechst (blue) and phalloidin (green) were used to stain the nucleus and F-actin, respectively. Representative images are shown (n = 4). Scale bar shows 100 μm. (B) NPCs exhibiting neural rosette architecture were treated with DMSO, 0.5, 1, 5, 10, and 50 μM rotenone for 24 hours. Cells were harvested and analyzed by western blot. A representative blot is shown (n = 4). (C–E). Quantification of ZO-1 (C), N-cadherin (D), and β-catenin (E) protein levels from (B). Protein levels were quantified and standardized against a loading control (GAPDH) using LI-COR Image Studio. Fold change compared to the average DMSO control was calculated and plotted in GraphPad Prism. Data were analyzed using a mixed-effects model (P ≤ 0.05) and a Dunnett’s multiple comparison relative to the DMSO control (n = 4). Error bars indicate standard error of the mean (SEM).

Rotenone effects on junctional proteins are specific to neural rosettes

While neural rosettes are a signature structural feature of NPCs, we wanted to examine if rotenone could trigger the relocalization of junctional proteins in cell types that do not classically exhibit a rosette structure. We performed dose–response studies of rotenone exposure in human fibroblasts and iPSCs. While both fibroblasts and iPSCs showed morphological changes in response to rotenone, the localization of junctional proteins in these cell types was not affected (Supplementary Fig. S3). Fibroblasts exhibited altered morphology with thin projections at lower concentrations, ruffling at higher concentrations, and surprisingly near-normal morphology at 50 µM, the highest concentration (Supplementary Fig. S3F). The iPSC colonies displayed only subtle changes with rotenone exposure, although colony formation was disrupted at 5 µM rotenone (Supplementary Fig. S3A). Interestingly, like fibroblasts, iPSCs appeared to be less affected by the highest concentration of rotenone. Protein levels of ZO-1, N-cadherin, and β-catenin were not significantly altered by rotenone in either cell type (Supplementary Fig. S3.)

Rotenone exposure results in lower protein levels of PAX6 in NPCs

Since rotenone selectively impacted the relocalization of junctional proteins and neuronal rosette structure in NPCs, we wanted to determine if NPC identity may also be altered in response to rotenone exposure. Following a 24-hour rotenone treatment, we examined the expression of the NPC markers nestin and PAX6. Nestin exhibited normal localization in the rosettes and did not show any significant change in mRNA or protein levels (Fig. 4A, B, D, F). PAX6 also showed normal localization with rotenone exposure (Fig. 4A). However, protein levels of PAX6 were significantly decreased by rotenone at concentrations as low as 0.5 µM, although mRNA levels were not affected (Fig. 4B, C, E). Further testing showed that the NPC markers SOX2 and FOXG1 were also normally expressed and localized in the setting of rotenone exposure, suggesting that rotenone selectively decreases PAX6 protein levels and alters rosette structure without large-scale effects on NPC identity (Supplementary Fig. S4).

FIG. 4.

Rotenone treatment selectively impacts PAX6 levels. (A) NPCs exhibiting neural rosette architecture were treated with DMSO, 1, 5, 10, and 50 μM rotenone for 24 hours. NPCs were fixed and imaged for NPC markers, PAX6 (red), and nestin (magenta). Hoechst (blue) and phalloidin (green) were used to stain the nucleus and F-actin, respectively. Scale bar shows 100 μm. Representative images are shown (n = 4). (A) NPCs exhibiting neural rosette architecture were treated with DMSO, 0.5, 1, 5, 10, and 50 μM rotenone for 24 hours. Cells were harvested and analyzed by western blot. Representative blot is shown (n = 4). (C, D) Quantification of PAX6 (C) and nestin (D) protein levels from (B). Protein levels were quantified and standardized against a loading control (GAPDH) using LI-COR Image Studio. Fold change compared to the average DMSO control was calculated and plotted in GraphPad Prism. Data were analyzed using a mixed-effects model (P ≤ 0.05) and a Dunnett’s multiple comparison relative to the DMSO control (n = 4). Error bars indicate standard error of the mean (SEM). P ≤ 0.01 (**), P ≤ 0.001 (#), and P ≤ 0.0001 (##). (E, F) NPCs exhibiting neural rosette architecture were treated with DMSO, 1, 5, 10, and 50 μM rotenone for 24 hours. Cells were harvested and analyzed by RT-PCR. Fold change was quantified using the Livak method with GAPDH as a control gene and average DMSO as the reference condition. Data were analyzed using a repeated measures one-way ANOVA (P ≤ 0.05) and a Dunnett’s multiple comparison relative to the DMSO control (n = 3). Error bars indicate standard error of the mean (SEM).

NPCs are highly sensitive to rotenone and are unable to recover beyond a certain threshold

Our data showed that neural rosette morphology was significantly altered by rotenone treatment without large-scale effects on NPC markers. Therefore, we sought to determine if neural rosettes could re-form following a period of recovery after the withdrawal of rotenone. To test this, we exposed neural rosettes to varying concentrations of rotenone for 24 hours, removed rotenone, and then allowed cells to recover for an additional 48 hours. Although there was moderate recovery in the number of neural rosettes in NPCs treated with 0.5 and 1 µM rotenone, rosette count was still half of that seen in the untreated conditions (Fig. 5A and Supplementary Fig. S5). Moreover, NPCs treated with 5, 10, and 50 µM rotenone showed little-to-no recovery (Fig. 5A and Supplementary Fig. S5). While neural rosettes exposed to 5 µM rotenone showed more luminal expression of ZO-1, N-cadherin, and β-catenin following recovery, neural rosettes exposed to 10 and 50 µM rotenone remained disrupted, exhibiting aberrant localization of adherens junction proteins after the recovery period (Supplementary Fig. S6). Nestin protein levels were unchanged; however, protein levels for PAX6 were similarly decreased as previously shown and exhibited minimal recovery (Fig. 5B, F–G). Moreover, a significant trend for decreased protein levels of N-cadherin and β-catenin could now be detected when compared to control conditions (Fig. 5B–E). These results suggest that neural rosettes incur irreversible changes to their cytoarchitecture and identity beyond a certain concentration threshold of rotenone.

FIG. 5.

NPCs are unable to recover following rotenone treatment at select concentrations. (A) NPCs exhibiting neural rosette architecture were treated with DMSO, 0.5, 1, 5, 10, and 50 μM rotenone for 24 hours. NPCs were then allowed to recover for 48 hours and then fixed and imaged for NPC markers, PAX6 (red), and nestin (magenta). Hoechst (blue) and phalloidin (green) were used to stain the nucleus and F-actin, respectively. Scale bar shows 100 μm. Representative images are shown (n = 4). (B) NPCs exhibiting neural rosette architecture were treated with DMSO, 0.5, 1, 5, 10, and 50 μM rotenone for 24 hours and then allowed to recover for 48 hours. Cells were harvested and analyzed by western blot. Representative blot is shown (n = 3). (C–G) Quantification of ZO-1 (C), β-catenin (D), N-cadherin (E), nestin (F), and PAX6 (G) protein levels from (B). Protein levels were quantified and standardized against a loading control (GAPDH) using LI-COR Image Studio. Fold change compared to the average DMSO control was calculated and plotted in GraphPad Prism. Data were analyzed using a mixed-effects model (P ≤ 0.05) and a Dunnett’s multiple comparison relative to the DMSO control (n = 3). Error bars indicate standard error of the mean (SEM). P ≤ 0.05 (*), P ≤ 0.01 (**), and P ≤ 0.001 (#).

NPCs exposed to rotenone show decreased neurogenesis

Since rotenone exposure had significant effects on NPC identity and junctional proteins, we wanted to test whether they would retain their potential to generate neurons. To test this, we treated NPCs with varying concentrations of rotenone for 24 hours and proceeded to induce neuronal differentiation. Following differentiation and sufficient maturation, neurons were fixed and analyzed for neuronal identity using NeuN as well as SOX2 to quantify remaining NPCs (Fig. 6A). The number of NeuN+ nuclei was significantly decreased by rotenone at concentrations as low as 0.5 µM (Fig. 6B). However, to our surprise, there was also a significant decrease in the number of SOX2+ nuclei, suggesting that the remaining cells are neither neurons nor NPCs (Fig. 6C). Moreover, we did not find any glial cells at this early stage of differentiation, consistent with previous studies that show the emergence of glial cells only around day 90.⁶¹ Altogether, these data suggest that the lasting effects of rotenone exposure can have drastic effects on neurogenesis and potentially alter the course of cellular differentiation.

FIG. 6.

NPCs exposed to rotenone have reduced neurogenesis potential. (A) NPCs exhibiting neural rosette architecture were treated with DMSO, 0.5, 1, 5, 10, and 50 μM rotenone for 24 hours and then utilized for neuron differentiation. Neurons were imaged for NPC marker, SOX2 (red), and neuronal marker NeuN (green). Hoechst (blue) was used to stain the nucleus. Scale bar shows 100 μm. Representative images are shown (n = 3). (B, C) Quantification of (A). The average percentage of NeuN+ (B) and SOX2+ (C) nuclei in neuronal cultures. Datapoints represent the average of technical replicates for each cell line. Data were analyzed using repeated measures one-way ANOVA (P ≤ 0.05) with a Dunnett’s test for multiple comparisons relative to the DMSO control. Error bars indicate standard error of the mean (SEM). P ≤ 0.001 (#).

Discussion

In this study, we sought to characterize the impact of oxidative and mitochondrial stress on human iPSC-derived neural rosettes by exposing them to rotenone, a mitochondrial complex I inhibitor. As expected, NPCs exposed to rotenone exhibited impaired mitochondrial function (Fig. 1). We found that rotenone exposure results in significant disruption of the neural rosette architecture that is accompanied by alterations in the organization of key cell-to-cell contacts (Figs. 2 and 3). Rotenone exposure led to a selective decrease in PAX6 protein levels without impacting levels of several other NPC markers (Fig. 4 and Supplementary Fig. S4). We further found that at low doses of rotenone, the effects on neural rosette cytoarchitecture could be partially reversed with the removal of rotenone, whereas exposure to higher concentrations of rotenone led to irreversible effects on the cytoarchitecture of the neural rosettes (Fig. 5). Last, the potential for NPCs to differentiate into cells of neuronal lineage was significantly impacted with all concentrations of rotenone, suggesting that alterations resulting from rotenone treatment could have profound effects on cellular differentiation (Fig. 6).

Several studies have shown that adherens and tight junction proteins play critical roles in maintaining NPC polarity.^10,62–64 Consistent with these studies, NPCs exposed to rotenone in our experiments displayed a rounder morphology instead of an elongated shape and radial organization, suggesting a loss in polarity that may be attributed to the relocalization of junctional proteins (Fig. 3). During neurodevelopment, loss or mislocalization of junctional proteins has been shown to disrupt the balance of cell division, differentiation, and migration.^11,65,66 Loss of N-cadherin has been shown to disrupt the expansion of radial glia and subsequent neural migration.^11,13 Although this process did not impact neurogenesis, it altered the organization of the cortical lamina, a feature also observed in several neurodevelopmental disorders.¹³ Loss of β-catenin in early neurodevelopment has also been shown to have severe consequences, including a loss in cell adhesion and apoptosis.⁶⁵ As we did not observe significant changes to protein levels, it is unclear if mislocalization of junctional proteins would have the same functional effect as ablated expression of these proteins, although we did observe alterations to neurogenesis (Fig. 6).

Mislocalization of N-cadherin has been observed with defects in the Notch signaling pathway as well as the RhoA/ROCK pathways. Although disruption of these pathways has differing effects, both play important roles in rosette formation and neural development.^67,68 Rotenone has been shown to alter and disrupt cell polarity via the RhoA/ROCK pathway, presenting a possible mechanism by which it could similarly influence junctional protein localization.⁶⁹ Rotenone has also been shown to impact microtubule and actin cytoskeleton dynamics.^52,69–72 We did observe an alteration in F-actin localization following rotenone treatment, whereby F-actin signal was located at the cell periphery rather than the center of the rosette (Fig. 3). Previous studies have shown that inhibition of the actin cytoskeleton and microtubule networks disrupts neural rosette structure and leads to a similar relocalization of β-catenin, ZO-1, PARD3, and tubulin as observed with our rotenone-treated neural rosettes.⁷³

Oxidative stress and mitochondrial stressors have been hypothesized to play a role in neurodevelopmental events in a number of neurological and psychiatric disorders.^{22–24,26,39,74,75} Studies in NPCs have found that levels of ROS can modulate cell proliferation and neural differentiation.^76–79 In our study, rotenone exposure significantly disrupted rosette architecture, decreased PAX6 levels, and altered neurogenesis (Fig. 2–4, 6). Other cell types have similarly shown vulnerability to oxidative stress as well as related alterations to adherens junction complexes and the actin cytoskeleton.^80–83 While the effects of ROS on cytoarchitecture have not been directly investigated in neural rosettes, different stressors have been shown to lead to similar alterations in cell junction patterns. Hříbková et al. showed that calcium in particular was critical in maintaining neural rosette structure. Moreover, alterations to neural rosettes via calcium removal or inhibition of the actin cytoskeleton significantly increased apoptosis and decreased neural differentiation. Complex I inhibition can induce alterations to calcium levels and concomitant cytoskeletal changes, suggesting a possible mechanism of rotenone action.^41,69,84,85 Moreover, oxidative stress and mitochondrial dysfunction have also been closely associated with alterations in calcium signaling.^26,75

While the effects of rotenone on mitochondrial function and oxidative stress may lie behind the structural alterations in our neural rosettes, it was surprising that protein levels of the transcription factor PAX6, an NPC marker, were significantly downregulated by rotenone without changes in the protein levels of other NPC markers (Fig. 4 and Supplementary Fig. S4). PAX6 has been shown to regulate neuronal cell polarization, influence the adhesive properties of cortical neurons, and play critical roles in neurogenesis.^86–90 In line with altered neurogenesis, we found that there was a decrease in the number of neurons in cultures differentiated from NPCs exposed to rotenone, though the identity of the surviving cells remains unknown (Fig. 6). Moreover, it is unclear whether the decrease in PAX6 levels is driving disruptions in structure or whether structural changes are altering cellular identity and PAX6 levels. Further research will be needed to elucidate the connection between PAX6 reduction and the loss of cell polarity and repatterning of intercellular connections to better understand how neurogenesis is impacted.

In summary, our results showed that rotenone, a mitochondrial complex I inhibitor, causes a dose-dependent disruption of the neural rosette architecture accompanied by a loss in cell polarity, cytoskeletal rearrangements, and repatterning of cell-to-cell contacts. We further observed a significant decrease in the protein levels of the transcription factor PAX6 and a significant effect on neurogenesis. While the exact molecular mechanisms underlying these changes are not fully understood, our results provide additional information on the nature of cellular effects that result from exposure to mitochondrial and oxidative stress during neurodevelopment.

Author Disclosure Statement

The authors declare no competing interests.

Funding Information

This work was supported by the National Institute of Mental Health Biobehavioral Research Award for Innovative New Scientists (BRAINS) Award R01MH113858 (to R.K.), the National Institute of Mental Health Clinical Scientist Development Award K08MH086846 (to R.K.), the Doris Duke Charitable Foundation Clinical Scientist Development Award (to R.K.), the Ryan Licht Sang Bipolar Foundation (to R.K.), Harvard University Milton Fund Award (to R.K.), the Harvard Stem Cell Institute (to R.K.), a MGH Fund for Medical Discovery (FMD) award (to S.S.), a Deutsche Forschungsgemeinschaft Award VA 1471/1–1 (to M.V.), and Steve Willis and Elissa Freud (to R.K.). Some imaging experiments were performed using the Nikon A1R confocal system in the microscopy core of the MGH Program in Membrane Biology, which is supported by NIH grants DK043351, DK057521, and 1S10 RR031563-01.