Application Note
Functional characterization of healthy and Alzheimer’s disease-related 3D neurospheres formed using human iPSC-derived glutamatergic neurons, GABAergic neurons, and astrocytes
Krishna Macha, Carole Crittenden, Oksana Sirenko | Molecular Devices, LLC
Rebecca Fiene, Scott Schachtele, Coby Carlson | FUJIFILM Cellular Dynamics, Inc.
Introduction
Genetic background of Alzheimer’s disease
The allelic composition of apolipoprotein (ApoE) in an individual can greatly impact risk of developing Alzheimer’s Disease (AD). For example, an estimate of developing AD by age 85 is ~65% in people with two copies of the E4 allele (ApoE4/4) as compared to only 10% in people with both wild type E3 alleles (ApoE3/3). This striking difference identifies ApoE4 as a major genetic risk factor for AD and highlights its importance in disease pathogenesis. ApoE is a lipid transport protein and the human ApoE3 and ApoE4 isoforms differ only at one amino acid residue (position 112). It is unclear how ApoE4 perturbs the intracellular lipid state, but there are reports that ApoE 4/4 can disrupt the cellular lipidomes of human induced pluripotent stem cell (iPSC)-derived astrocytes. In this application note, we demonstrate methods for modeling AD in iPSC-derived neurons engineered to contain combinations of ApoE allelic variation (E2/2, E3/3, and E4/4).
Neural 3D organoids from human induced pluripotent stem cells (iPSC) are a rapidly developing technology with great potential for understanding brain development and neuronal diseases. A promising parallel approach is to assemble similar structures as 3D spheroids or “neurospheres” by using defined combinations of fully differentiated human iPSC-derived cells in tri-culture, including glutamatergic neurons (Gluta), GABAergic neurons (GABA), and astrocytes (ASC). In this study, we chose to model AD by incorporating allelic variants of the ApoE gene (2/2, 3/4, and 4/4) to create disease-specific “neurospheres”.
3D neurospheres were formed by combining 20–25,000 cells in ultra-low attachment (ULA) plates and their morphology was tracked over time using live cell imaging. The neurospheres’ diameters ranged from 400–500 μm. Functional performance was tested via calcium oscillation assay on FLIPR® Penta High-Throughput Cellular Screening System for its fast kinetic recordings capability. Notably, the calcium-sensitive dye used contains a background fluorescence masking technology that enables sensitive detection of neurospheres in 384-well format. The calcium oscillation patterns were analyzed for metrics including peak count, amplitude, and width. Distinct differences in pattern and decrease in baseline oscillation frequencies were observed between wild type (ApoE 3/3) and ADspecific neurospheres, specifically for those containing the ApoE 4/4 allelic variant. Within each group, the calcium kinetics and patterns were highly consistent.
For pharmacological characterization, a panel of select compounds was used to show both the appropriate responses to GABA, AMPA, and NMDA, as well as changes to neuroactive compounds. Some drugs previously shown to affect AD phenotypes (memantine and donepezil) decreased calcium peak amplitude and altered other metrics as visualized by ScreenWorks software and analyzed by Peak Pro 2. Additionally, the neurospheres’ viability was analyzed using confocal fluorescence imaging for cell organization and expression of various neural markers, including TUJ1 and GFAP. Taken together, this biological system of 3D neurospheres assembled from human iPSC-derived cell types paired with high-content imaging and detailed analysis of calcium oscillations demonstrates a promising tool for disease modeling and compound testing.
Methods
3D spheroids were formed using human iPSC-derived cell types ((FUJIFILM Cellular Dynamics) including iCell® GlutaNeurons, iCell GABAneurons, and iCell Astrocytes 2.0. Briefly, cryopreserved vials were thawed and mixed in desired ratios following the iCell NeuroSpheres protocol, provided by Fujifilm Cellular Dynamics, (e.g., 90% neurons (~70:30 Gluta: GABA) and 10% astrocytes), and then 25,000 cells/well in complete BrainPhys™ medium were plated into 384-well ULA plates (either black Corning #4516 with clear bottom or clear S-Bio #MS-9384UZ). After 2 days, cells formed compact spheroids and were then maintained with media changes every 2–3 days until at least day 21. On the day of assay, cell spheroids were loaded with 2X concentration of FLIPR Calcium 6 dye indicator (Molecular Devices) and incubated for 2 hours. We used a high-speed EMCCD camera on the FLIPR Penta instrument (Molecular Devices) to measure the patterns and frequencies of spontaneous calcium waveforms from 3D neurospheres. Baseline recordings were acquired for ≥10 min, and then plates were dosed with drugs for 30–90 min. Peak analysis was accomplished with ScreenWorks Peak Pro 2 software (Molecular Devices), allowing characterization of both primary and secondary peaks, as well as complex calcium oscillation patterns. High-content imaging was done by the automated ImageXpress Micro Confocal Imaging system (Molecular Devices) and was used to capture 3D structures of the neurospheres for viability evaluation.
Figure 1. Schematic diagram of the process workflow. (1) iPSC-derived cells are thawed and combined in ratios of approx. 90% neurons and 10% astrocytes into (2) ULA 3D spheroid-forming plates. (3) Neurospheres form within 24–48 hours and (4)–(5) maintained in culture with regular media for >21 days. (6) Cells are assayed on the FLIPR Penta or imaging instruments.
Table 1. Commercially available iCell neural products used in this application note.
Table 2. Calculations for generating 3D neurospheres using iCell neural products. Neurospheres can be generated via calculating direct proportions of cells (Base on # of cells). When using iCell GlutaNeurons it may be of interest to adjust cell proportions based on the percent glutamatergic neuron population indicated on the lot-specific product certificate of analysis (Corrected for % of Gluta).
Results
Formation and characterization of 3D neurospheres
“iCell NeuroSpheres” were composed of glutamatergic and GABAergic neurons together with iPSC-derived astrocytes. Cells were maintained in co-culture with complete BrainPhys™ medium until day of assay as indicated.
Figure 2 shows the time-course of neurosphere formation and images of micro-tissues stained with Hoechst nuclear dye and TUJ1 neuronal marker. Figure 3 demonstrates recordings of calcium oscillation patterns for control samples, also for microtissues treated with indicated neuro-active substances. Calcium oscillation activities were captured by the FLIPR® Penta High-Throughput Cellular Screening System and analyzed by the Peak Pro 2 Software. Neurospheres responded to compounds in dose-response manner with changes in amplitude, frequencies and patterns of calcium oscillations. EC50s and directions of changes in the peak counts are shown in the table of Figure 3.
figure 2. Formation of iCell NeuroSpheres. (A) 3D neurospheres can be formed in various ULA plate types, including the faCellitate BIOFLOAT™ 384-well option demonstrating the time course of sphere formation. Cell culture is straightforward, and cells can survive for many days. (B) Self assembly of iPSC- derived neurospheres can be performed in high-throughput format including Corning 384w B/C plate, and results in uniform neurosphere formation (only one 3D structure per well); (C) Neurospheres can be stained with calcium indicator dyes for function testing or with antibodies for neural-specific markers, such as TUJ1.
Figure 3. Functional testing of iCell NeuroSpheres. (A) Ca2+ oscillations of 3D neurospheres were recorded by kinetic imaging using a FLIPR Penta instrument. Traces (in red) represent fluctuations in fluorescent intensities recorded with the calcium-sensitive dye Calcium 6 (60 min post-dose). Waveform patterns affected by neuroactive compounds are listed in the table to the right. (B) Representative time-lapse images of 3D neurospheres loaded with Calcium 6 dye were captured with an interval of 0.4 sec using the ImageXpress Micro Confocal Imaging system.
ApoE GABA neurons for AD disease modeling
For the modeling of AD, we used mutated GABAerdic neurons. We analyzed AD-related phenotypes through FLIPR Penta measurement of calcium oscillations from 3D neurospheres assembled with GABAergic neurons from four different ApoE allelic variants, including 2/2, 3/3 (control), 3/4, and 4/4 (in addition to 70% glutaneurons and 10% astrocytes). Representative patterns for calcium oscillation activities of different mutated phenotypes are presented in Figure 4. The presence of GABAergic neurons with ApoE4/4 mutation showed the most distinct change in functional activity. Apparent decrease in oscillation frequency and increase in amplitude in comparison to control ApoE 3/3 WT phenotype was observed. Other mutations did not demonstrate significant differences.
With the ApoE 4/4-containing disease model showing the strongest difference from control cells (ApoE 3/3), we compared the treatment of neurospheres to a select group of neuroactive compounds. The experiment demonstrated that the expected responses were observed with AMPA, Baclofen, and GABA.
Compounds used for the treatment of AD disorder
Previous work published by Strong et. al. (DOI: 10.1101/2022.05.04.490442) had similarly illustrated the functional deficits caused by the ApoE4 allele in GABA neurons could be measured in 3D spheroids using FUJIFILM Cellular Dynamics iPSC-derived cell types. Importantly, they demonstrated that various clinically- approved compounds used to treat the symptoms of AD could reverse the calcium waveform phenotypes generated in vitro.
Memantine is a clinically useful treatment and is believed to act via the blockade of current flow through NMDA receptors, which are broadly involved in synaptic function in the brain. In rodents, memantine antagonizes native NMDA receptors with a micromolar potency.
Donepezil is a clinically-approved acetylcholinesterase inhibitor used to increase cortical acetylcholine levels. Symptoms of AD are believed to be related to cholinergic deficit, particularly in the cerebral cortex.
EUK-134 is a synthetic superoxide dismutase/catalase mimetic that is thought to prevent excitotoxic neuronal injury. This compound and other mimetics have also been shown to inhibit beta amyloid plaque production in animal models.
Figure 5. Ca2+ oscillations of WT (ApoE 3/3) and genetically modified (GABA-neurons contain ApoE 4/4 mutation) neurospheres determined by kinetic calcium imaging using the FLIPR Penta instrument and analyzed using Peak Pro 2 software and their corresponding peak counts were reported in parenthesis (ApoE status, peak number). Neurospheres were treated with compounds for 45 min and 90 min for two experiments, respectively. Modulations of the activity phenotypes were observed in two independent experiments with different concentrations of the compounds, where the increased concentration of Donepezil is the most impacted by increased concentration.
Treatment microtissues containing ApoE 4/4 with memantine resulted in the increased frequencies of calcium oscillations, bringing the patterns closer to WT (ApoE 3/3) phenotype. Data from two independent
experiments showed in the Figure 5. Notably, memantine also increased oscillation activity with other tested phenotypes (data not shown).
Comparison of ULA plate type
With the increased availability of biologically relevant cell types (i.e., human iPSC-derived) that can be incorporated into 3D neurospheres, there is concurrent development around cell culture technologies (i.e., ULA plates) for the formation of such 3D structures. We compared two popular plates: Corning 4516 (Figure 6A) which is black with a clear bottom vs. S-bio PrimeSurface U-bottom (Figure 6B) which is clear. Both plate types were applicable for the formation of 3D neurospheres. Comparison data is presented in Figure 6.
![Neurosphere sizes (areas from the transmitted light [TL] image)](./media_1c968118e12dc6b4dedae9ed1ed1d7bffeb61d433.jpg?width=750&format=jpg&optimize=medium)
Figure 6. Neurosphere sizes (areas from the transmitted light [TL] image) were determined from images taken in TL and compared across different plates (A–B). (D) Ca2+ oscillations of WT and genetically modified (GABA-neurons containing ApoE 4/4 mutation) neurospheres were recorded by (C) kinetic calcium imaging (ApoE 3/3) using the FLIPR Penta instrument. Data show consistent calcium oscillation amplitude measurements between Corning and S-bio ULA plates. (D) Representative calcium traces from WT (ApoE 3/3) and ApoE 4/4 neurospheres recorded from the S-bio ULA plate.
Summary
- In vitro 3D neurospheres, generated using commercially available iPSC-derived neural cells, present a useful cell–based assay for assessment of neurotoxicity, neuro-active effects of various neuromodulators, and disease modeling.
- This assay platform shows promise for the evaluation of compound effects and early detection of neurotoxicity in vitro due to its ease of use, consistency across wells and assay plates, simplicity of data analysis, and ultimate biological relevance.
- Analysis of kinetic calcium imaging provides reliable and accurate read-outs for the functional neural activity and can be used for evaluation of phenotypic changes and compound effects.
- Decreased calcium oscillation frequency (Peak Count) was observed for neurospheres formed with GABAergic neurons containing ApoE 4/4 alleles which was reversed with drugs used for treatment of AD disorder.
- This biological system of 3D neurospheres paired with a detailed analysis of calcium oscillations demonstrates a promising and informative tool for disease modeling and phenotypic functional evaluation that could be used for compound testing.