Assessment of ethanol-induced toxicity on iPSC-derived human dopaminergic neurons using a novel high-throughput mitochondrial neuronal health (MNH) assay

Excessive ethanol exposure can cause mitochondrial and cellular toxicity. In order to discover potential counteracting interventions, it is essential to develop assays capable of capturing the consequences of ethanol exposure in human dopaminergic (DA) neurons, which are crucial for the development and maintenance of alcohol use disorders (AUD). Here, we developed a novel high-throughput (HT) assay to quantify mitochondrial and neuronal toxicity in human DA neurons from induced pluripotent stem cells (iPSCs). The assay, dubbed mitochondrial neuronal health (MNH) assay, combines live-cell measurement of mitochondrial membrane potential (MMP) with quantification of neuronal branching complexity post-fixation. Using the MNH assay, we demonstrated that chronic ethanol exposure in human iPSC-derived DA neurons decreases MMP and branching complexity in a dose-dependent manner. The toxic effect of ethanol on DA neurons was already detectable after 1 hour of exposure, and occurred similarly in DA neurons derived from healthy individuals and from patients with AUD. We next used the MNH assay to carry out a proof-of-concept compound screening using FDA-approved drugs. We identified potential candidate drugs modulating acute ethanol toxicity in human DA neurons. Among these drugs, flavoxate and disulfiram influenced mitochondrial neuronal health independently from ethanol, leading to amelioration and worsening, respectively. Altogether, we developed an HT assay to probe human mitochondrial neuronal health and used it to assess ethanol neurotoxicity and to identify modulating agents. The MNH assay represents an effective new tool for discovering modulators of mitochondrial neuronal health and toxicity in live human neurons.


Introduction
Ethanol is the most frequently abused drug worldwide and its excessive consumption is a leading risk factor for disability and death (1). High ethanol consumption over time can lead to serious health and social problems, including the development of alcohol use disorder (AUD). AUD is among the most prevalent mental disorders in industrialized countries (2).
Once ingested, ethanol quickly distributes throughout the body and reaches the brain within minutes. Given this rapid and vast spread, ethanol can cause direct organ toxicity. Ethanolinduced neurotoxicity is particularly detrimental given that damaged neurons cannot be replaced. Within the central nervous system, ethanol exposure directly affects dopaminergic (DA) neurons, resulting in increased extracellular dopamine mainly in the striatum (3) higher firing frequency and increased excitability (4), as well as decreased DA synthesis and dopamine D2 receptor availability in AUD patients (5).
At the cellular level, an important role during ethanol intoxication may be played by mitochondria, crucial organelles that safeguard cellular homeostasis. Mitochondria provide energy in form of ATP through oxidative phosphorylation (OxPhos), which turns ADP into ATP by releasing the energy stored in the proton gradient also known as mitochondrial membrane potential (MMP) (6). Mitochondria are also crucial for cell death (7). Selective MMP permeabilization activates caspase-driven apoptotic cell death through opening of the mitochondrial permeability transition pore (mPTP) (8). MMP is thus an essential parameter for viable cells, since it is important for both ATP generation and initiation of apoptosis, and may serve as a marker of cell health (9). Various studies reported ethanol-induced toxicity in mitochondria located in the brain (10), including increased mitochondrial production of free radicals (11), alterations in mitochondrial respiration and MMP (11,12), impairment of ATP production (13,14), and cell death induction (13,15).
The mechanisms underlying ethanol-induced toxicity in human brain cells remain largely unknown. Most investigations are based on animal models, which may not fully recapitulate the human disease state, on post-mortem tissues, which correlates more to an endstage of AUD, or on patient brain imaging, which provides macroscopic data lacking information at the cellular and molecular level (16). Given the lack of suitable human cellbased model systems for the assessment of neurotoxicity and drug discovery, our knowledge of compounds capable of counteracting ethanol toxicity in humans is limited.
Here, we used human iPSCs to investigate the toxic consequences of ethanol in human DA neurons. In order to assess human neurotoxicity in a quantitative and high-throughput (HT) manner, we devised a new assay that we named mitochondrial neuronal health (MNH) assay. The MNH assay is based on high-content imaging (HCI) and combines live-cell measurement of MMP with quantification of neuronal branching outgrowth. Using this assay, we demonstrated the acute and chronic effects of ethanol toxicity on mitochondrial neuronal health in human DA neurons that we derived from control individuals and from subjects with AUD. We also showed that the MNH assay can be used to perform compound screenings to identify drugs influencing mitochondrial neuronal health. Hence, the newly developed MNH assay represents an effective HT tool for analyzing the cellular health of human iPSC-derived neurons and for discovering potential modulatory interventions.

Development of the MNH assay
In order to assess the toxic effects of ethanol in human neurons, we first generated neural progenitor cells (NPCs) from healthy control-derived iPSCs (XM001 line) (17) and human embryonic stem cells (hESCs) (H1 line) (18) using a small molecule-based protocol (19). We then differentiated NPCs into post-mitotic neurons enriched for DA neurons (Fig.   1A). The pluripotent stem cell (PSC)-derived DA neurons expressed the neuron-specific marker MAP2 and DA markers tyrosine hydroxylase (TH) and FOXA2 (Fig 1B). In the differentiated cultures, 75% of cells expressed the pan-neuronal marker beta tubulin-III (TUJ1) and 20% expressed TH (Fig. 1C). These numbers are in accordance with previous protocols (20). We monitored neuronal network activity using micro-electrode array (MEA).
We confirmed that the generated DA neurons were functional and exhibited multiple spontaneous spikes after 4-8 weeks in culture (Fig. 1D).
We next established an HCI-based approach to assess mitochondrial function and neuronal growth capacity in DA neurons. We dubbed the method mitochondrial neuronal health (MNH) assay. To assess neuronal branching capacity, we cultivated DA neurons for 4-8 weeks, re-plated them on the assay plate and kept them for 1 day or 5 days after plating. We then fixed the cultures and stained them with TUJ1 to visualize neuronal arborization (Fig.   1E). Since TUJ1 staining does not allow to discriminate between axons and dendrites, we chose to refer to any projection from the cell body as neurite. In order to assess differences in neurite outgrowth, we compared DA neurons fixed 1 day or 5 days post-plating to NPCs fixed 1day post-plating (Fig. 1F). The MNH assay effectively captured and quantified differences in neuronal arborization. DA neurons at days 5 post-plating showed increased neurite count In order to assess MMP, we implemented an HCI assay that we previously established for iPSC-derived NPCs (21). We measured MMP using the potentiometric fluorescent dye tetramethylrhodamine methyl ester (TMRM), a lipophilic cation that accumulates in the mitochondrial matrix in proportion to the potential of the membrane. We re-plated 4-8-weekold DA neurons on the assay plate and kept them for 5 days before live-staining for MMP and subsequent neuronal branching quantification ( Fig. 2A). Stimulation with the oxidative phosphorylation uncoupler FCCP and the complex III inhibitor Antimycin A (Ant.A) provoked a decrease in the MMP in a dose-dependent manner (Fig. 2B), whereas the ATPase inhibitor Oligomycin led to a dose-dependent increase in the MMP (Fig. 2C). The MMPmodulating effects of the mitochondrial inhibitors are in line with previous works (22,23).
The use of stimulating and inhibiting MMP modulators enabled us to determine the z-factor, a statistical indicator of HT bioassays that is considered excellent if 0.5>x>1 (Zhang et al., 1999). The MNH assay had a z-factor of 0.747, suggesting excellent features regarding reproducibility and robustness (Fig. 2C). Branching complexity did not change upon exposure to FCCP and Ant.A at a concentration of up to 100 µM (Fig. 2D-F), or to Oligomycin at a concentration of up to 200 µM (Fig. 2G-I). Taken together, the MNH assay was able to quantitatively identify changes in neuronal outgrowth capacity and in MMP-specific function.

The MNH assay detects chronic and acute ethanol-induced neurotoxicity in human DA neurons
In alcohol-naïve individuals blood alcohol concentrations (BACs) of 10 -50 mM typically lead to sedation, motor incoordination and cognitive impairment. BACs of  100 mM cause strong sedation and can lead to coma or death in alcohol-naïve individuals. Instead, in chronic alcohol consumers BACs of up to 300 mM have been reported (24), and 100 -200 mM typically lead to sedation, anxiolysis and hypnosis (25,26).
To address the toxic effects of ethanol on human DA neurons, we tested healthy DA neurons after 4-8 weeks of differentiation from iPSCs (XM001 line) and hESCs (H1 line). We exposed healthy DA neurons to different concentrations of ethanol in the media for 7 days (chronic exposure), with full media exchange every other day (Fig. 3A). Chronic exposure to ethanol over 7 days increased MMP at ethanol concentrations of 10 mM-100 mM and caused the MMP to collapse at ethanol concentrations  500 mM (Fig. 3B). Ethanol concentrations higher than 250 mM also strongly impaired neurite outgrowth (Fig. 3C-E).
We next investigated whether the observed changes in mitochondrial neuronal health could be recapitulated in DA neurons after acute exposure to ethanol. We plated 4-8-week-old DA neurons generated form iPSCs (XM001) and hESCs (H1) on the assay plate, kept them for 5 days and then exposed them to ethanol for 1 hour before the assay (Fig. 3F). Acute ethanol exposure led to a dose-dependent reduction of MMP starting at a concentration of 500 mM ethanol in DA neurons derived from iPSCs (Fig. 3G) and hESCs (Fig. 3K) Altogether, using the MNH assay to quantify ethanol-induced neurotoxicity, we determined that acute exposure to ethanol for 1 hour was sufficient to recapitulate the negative effects on mitochondrial neuronal health, which we observed in DA neurons after 7 days of chronic ethanol exposure. Neurotoxicity caused by ethanol ingestion may thus occur very rapidly. Therefore, in order to prevent ethanol-induced neurotoxicity, we might need to identify strategies to counteract the acute consequences of ethanol exposure.

DA neurons from AUD patients recapitulate acute ethanol-induced neurotoxicity
Next, we aimed to address ethanol-induced neurotoxicity in DA neurons derived from individuals diagnosed with AUD according to DSM-5 and ICD-10 (alcohol dependence).
We applied the MNH assay on AUD-DA neurons to quantify changes in MMP and neuronal arborization following 1 hour of acute ethanol exposure (Fig 5B). Nonetheless, the pattern of neurotoxicity detected by the MNH assay in AUD-DA neurons recapitulated the one observed in control DA neurons, suggesting that control neurons can be used to carry out studies aiming at identifying strategies to counteract ethanol-induced neurotoxicity.

MNH-based proof-of-concept compound screening identified modulators of ethanol neurotoxicity
We next sought to determine whether the MNH assay could be used to carry out compound screenings to identify modulators of neurotoxicity. For screening, we used DA neurons derived from control iPSCs (XM001 line). We focused our attention on acute ethanol exposure. Using a library containing 700 FDA-approved drugs (Selleckchem # z65122), we selected 48 compounds that are approved for clinical use in the context of neurological diseases (compound numbers 1-48) (Fig. 6B-D We plated 4-8-week-old DA neurons and cultured them for 5 days. We treated DA neurons overnight (ON) with 0.5 µM of compound for each of the 53 compounds in a proofof-concept experiment. We then exposed DA neurons to 1 M ethanol for 1 hour and performed the MNH assay (Fig. 6A). Total object count, obtained by quantifying Hoechststained nuclei, showed no significant changes in cell number following the compound treatments, suggesting that we could confidently compare treated samples to untreated samples (Fig. 6B). As baseline for comparisons, we used DA neurons treated for 1 hour with 1 M ethanol that were not pre-exposed to any compound These data confirm the results of the screening and suggest that by identifying compounds counteracting neurotoxicity, it may be possible to discover general modulators of mitochondrial neuronal health.

Discussion
In this study, we developed a method -the MNH assay -to assess the cellular health of human neurons derived from human pluripotent stem cells. We employed the MNH assay to dissect the neurotoxicity induced by ethanol on DA neurons. Previous iPSC-based studies assessed alcohol-mediated transcriptional changes in NMDA receptor expression (27), in GABA receptor expression (28), and in the expression of genes involved in cholesterol homeostasis, notch signaling, and cell cycle (29). Ethanol exposure for 1 day or 7 days was found to negatively influence the generation of mature neurons from NPCs (30). However, the effect of ethanol on mitochondrial neuronal function and neurite outgrowth remained to be determined. Using the MNH assay, we showed that acute exposure to ethanol was sufficient to cause loss of mitochondrial neuronal health, similarly to the one happening after 7 days of chronic ethanol exposure. Importantly, acute ethanol-induced mitochondrial and neuronal toxicity occurred in a comparable manner in different DA neurons derived from control individuals and from subjects diagnosed with AUD.
Using the MNH assay, we identified potential modulators of ethanol-induced neurotoxicity. Within a proof-of-concept screening, out of the 53 tested compounds flavoxate, tripelennamine, and acebutolol were found to potentially positively affect mitochondrial health and neurite outgrowth capacity. Conversely, 37 compounds showed a potential worsening effect in the presence of ethanol. Further studies should be carried out to confirm the exploratory results and determine the pharmacokinetic and pharmacodynamic characteristics of the hit candidates. Most drugs of abuse, including alcohol, have significant neurotoxic effects (31). Understanding the level of neurotoxicity caused by ethanol may enable the development of protective drugs to reduce the risk of developing neurodegenerative effects as consequence of excessive alcohol consumption. Interestingly, two drugs commonly employed in subjects with AUD, disulfiram and baclofen, negatively modulated mitochondrial neuronal health in the presence of ethanol. These findings warrant further exploration and raise concerns with respect to the potentially enhanced toxicity of drugs to treat AUD when used in concert with alcohol intake. Among the hits identified by the screening, we discovered possible modulators of mitochondrial neuronal health independent of ethanol. Flavoxate positively affected mitochondrial neuronal health, whereas disulfiram negatively affected mitochondrial neuronal health. In-depth analyses are warranted to confirm their function for human neuronal health. More generally, the findings suggest that the MNH assay might be used for various applications, including disease-specific studies, neurotoxicity studies, and studies aimed to assess the individual neuronal susceptibility to different stimuli (Fig. 7J).
The assessment of possible toxic effects of compounds is of high importance in the drug development process. Neurotoxicity testing commonly relies on in vivo animal studies that are expensive and may not fully recapitulate the toxicity profile of humans (32,33).
Recent studies used MEA to investigate the effect of compounds on spontaneous neuronal activity and network activity (34). Various HT screening methods have also been established (35)(36)(37)(38)(39)(40)(41). Nonetheless, there is a need for additional in vitro systems centered on human iPSC derivatives in order to capture different aspects of human neurotoxicity (42). We suggest the MNH assay as a novel HT assay centered on human neurons that can be further multiplexed with additional imaging-based techniques, including for example the addition of an apoptotic dye to quantify cell death (35,36). iPSC-derived 3D neural cultures may also be employed to assess the neurotoxic potential of drugs and environmental toxicants (43). Therefore, future implementations should aim to adapt the MNH assay to 3D culture systems, such as iPSCderived brain organoids (44).
Taken together, we developed a novel assay that enabled us to capture the ethanolinduced toxicity in human DA neurons and to identify potential small molecule modulators.
We propose the MNH assay as a tool for the evaluation of human neuronal health and for conducting HT drug discovery and drug toxicity studies.  On day 3 after transduction 1 x10 5 cells were plated onto mouse embryonic fibroblast (MEFs) layer or Vitronectin coated plate in repro media 1. From day 4 until day 7, the media was changed with repro media 2 containing basal blood media, NaB and ascorbic acid without cytokines. On day 7 the media was switched to repro media2 and E7 reprogramming media (1:1) containing E6 basal media (Thermo Fisher) with NaB, ascorbic acid and 100 ng/ml FGF2 (PeproTech) and the same was used every day until day 13. From day 13 onwards the cells were cultured in E8 life (Thermo Fisher) until day 20. Emerging colonies were picked three weeks after the transduction and tested for remaining viral RNA expression using RT-PCR by checking the expression of Sendai viral reprogramming particles using primers specific to Sendai virus.

iPSC culture
Control iPSCs were previously generated using episomal plasmids and described as XM001 (17). hESC line H1 was purchased from WiCell and was used in accordance with the (Lonza). We routinely monitored against mycoplasma contamination using PCR. 10 µM ROCK inhibitor (Enzo, ALX-270-333-M005) was added after splitting to promote survival. PSC cultures were kept in a humidified atmosphere of 5% CO2 at 37°C and 5% oxygen. All other cultures were kept under atmospheric oxygen condition. Pluripotency of the generated lines was confirmed following previously published procedures (45) using in vitro embryoid bodies (EB)-based differentiation. We reprogrammed patient PBMCs using Sendai viruses.

SNP karyotyping
Briefly, genomic DNA was isolated using the DNeasy blood and tissue kit (Qiagen, Valencia, CA) and samples were analyzed using the human Illumina OMNI-EXPRESS-8v1.6 BeadChip. First, the genotyping was analyzed using GenomeStudio 1 genotyping module (Illumina). Thereafter, KaryoStudio 1.3 (Illumina) was used to perform automatic normalization and to identify genomic aberrations utilizing settings to generate B-allele frequency and smoothened Log R ratio plots for detected regions. The stringency parameters used to detect copy number variations (CNVs) were set to 75 kb (loss), 100 kb (gain) and CN-LOH (loss of heterozygosity) regions larger than 3 MB.

Differentiation of NPCs and DA neurons
We obtained NPCs and DA neurons using a previously published protocol (19).

Micro-electrode array (MEA) recordings
MEA recordings were conducted using the Maestro system from Axion BioSystems. week-old DA neurons with incubation ON at 37°C and 5% CO2. The next day, staining solution containing 0.5 nM TMRM and 1:10,000 Hoechst was added to the media for 30 min at 37°C, 5% CO2. Further processing was performed according to the MNH assay as described before.

Compound screening -MNH assay
For the proof-of-concept compound screening, 48 compounds were selected from a library of 700 FDA-approved drugs (Selleckchem-z65122) and 5 drugs were added, used in the context of AUD treatment (Table 1)   independent experiments) normalized to the corresponding UT controls exposed to the assay media only.