Application Note
Use of MEA and AI to Assess Seizurogenic Potential of Compounds
Combining microelectrode array (MEA) electrophysiology with AI-based network analysis on the NANOSTACKS™ Brain MPS to detect seizurogenic activity in pharmaceutical compounds.
Download PDFOverview
Overview
Seizurogenic liability — the potential of a compound to trigger abnormal neuronal electrical activity — is a critical safety concern in CNS and non-CNS drug development. Standard cytotoxicity assays cannot detect sub-lethal functional neurotoxicity, including network hyperexcitability that underpins seizure risk.
This application note describes a workflow combining NANOSTACKS™ Brain MPS, microelectrode array (MEA) electrophysiology, and AI-based pattern recognition to identify compound-induced changes in neuronal network activity as an indicator of seizurogenic potential. The approach provides functional, mechanistically informative data not available from viability assays alone.
Study Design
Experimental Design
Model establishment
NANOSTACKS™ Brain MPS assembled with human neuronal co-culture. Cells differentiated to functional network maturity prior to compound exposure (typically 14–21 days).
Baseline MEA recording
Spontaneous network activity recorded using multi-well MEA plates. Baseline parameters established: mean firing rate (MFR), burst frequency, synchrony index.
Compound exposure
Test compounds applied at multiple concentrations. MEA recordings acquired at defined timepoints post-exposure (acute: 30 min–4 h; sub-chronic: 24–72 h).
AI-based network analysis
Network activity patterns analysed using machine-learning classification trained on known seizurogenic and non-seizurogenic compound profiles. Burst pattern features extracted and scored.
Model Configuration
Model Configuration and Cell Types
NANOSTACKS™ Brain MPS — composition
- Glutamatergic and GABAergic neurons (iPSC-derived or SH-SY5Y differentiated)
- Astrocytes — support network formation and modulate excitability
- Microglia — immune component; respond to compound-induced neuroinflammation
- Optional: brain microvascular endothelial cells for BBB modelling
- MEA-compatible 24-well plate format
See the Brain MPS product page for full configuration options.
Readouts & Methods
Readouts and Methods
MEA electrophysiology
- Mean firing rate (MFR)
- Burst frequency and duration
- Synchronised network events (SNA)
- Inter-burst interval (IBI)
- Network burst amplitude
AI / ML analysis
- Supervised classification (seizurogenic vs non-seizurogenic)
- Burst pattern feature extraction
- Dose–response network scoring
- Comparison to reference compound training set
Viability (parallel wells)
- LDH release (cytotoxicity)
- Calcein-AM / PI live-dead imaging
- Neurite integrity (fluorescence)
Neurochemical
- Neurotransmitter release (ELISA or LC-MS/MS)
- Cytokine panel (IL-6, TNF-α) for neuroinflammation
Key Findings
Key Findings
- 1NANOSTACKS™ Brain MPS maintained spontaneous neuronal network activity for >21 days, enabling both acute and sub-chronic seizurogenicity assessment.
- 2Known positive controls (4-aminopyridine, picrotoxin) produced characteristic burst pattern changes detectable within 30–60 minutes of exposure.
- 3AI-based network scoring correctly classified seizurogenic compounds with sensitivity and specificity superior to single-parameter MEA analysis.
- 4Sub-lethal concentrations — below cytotoxicity thresholds — produced distinct network hyperexcitability signatures, demonstrating functional sensitivity beyond standard viability assays.
- 5The multi-well format enabled dose–response profiling of 4–6 compounds per plate, increasing experimental throughput relative to single-chip MEA systems.
Significance
Why It Matters
Seizurogenic risk is a regulatory requirement for CNS compounds and an increasingly important consideration for non-CNS drugs with off-target neural effects. Current regulatory guidance (ICH S7B) recommends functional neuronal network assays as part of safety pharmacology packages.
The NANOSTACKS™ MEA workflow provides a human-relevant, functional assessment that aligns with New Approach Methodology (NAM) frameworks under the FDA Modernization Act 2.0. By combining network electrophysiology with AI-based pattern analysis, it enables detection of seizurogenic compounds that pass standard cytotoxicity screening.
For platform background, see What is NANOSTACKS™? and What is MPS?.
Platform
Platform Used
This application note uses NANOSTACKS™ — Revivocell's modular microphysiological systems platform. NANOSTACKS™ uses stackable inserts in standard SBS 24-well plates, with orbital shaking providing physiological flow without external pumps or microfluidic hardware. The platform is MEA-compatible, enabling direct integration with multi-well electrophysiology systems.
Data & Figures
Experimental Data & Figures
Figures included in the full application note:
- Fig. 1 — Coculture setup: schematic of NANOSTACK™ positioned above MEA neurons layer
- Fig. 2 — AI seizurogenicity probability scores for 9 reference compounds across 5 concentrations (Acetaminophen, Carbamazepine, DMSO, DZ, Kainic acid, NMDA, PTZ, Paroxetine, Theophylline, Varenicline)
View all figures and raw data outputs in the embedded PDF below.
Related Resources
Related Resources
Full Application Note