TOWARD PRECISE POPULATION-WIDE PLASTICITY IN HUMAN BIOLOGICAL NEURONAL NETWORKS

Human stem cell–derived neural models cultured on high-density microelectrode arrays (MEAs) allow for read/write access to populations of neurons with single-cell resolution. Robust multicellular models that form mature synaptic networks and can be controlled via on-chip stimulation have the potential to facilitate discoveries related to neural computation, synaptic plasticity, memory formation, and biohybrid computing. We defined a long-term (>3-month) maturation protocol for iPSC-derived NGN2-ASCL1 induced cortical neurons co-cultured with human astrocytes, enabling MEA recordings up to four months post-induction. We developed a simple algorithm to detect isolated neuronal units based on spike timing and the spatial oversampling of neurons on high-density arrays. We found that spiking activity in isolated units from long-term co-cultures shifted from an oscillatory state with prolonged down-states to asynchronous, irregular activity with fewer silent periods compared to the two-month time point. These long-term cultures exhibited increased spontaneous firing rates and spike amplitudes, reduced percentages of spikes within network bursts, and normalized coefficients of variation of the interspike intervals. We then tested the efficiency of single-electrode, on-chip stimulation to drive low-latency evoked spikes in isolated units. We found that iPSC-derived induced neurons exhibited significantly increased response probabilities to single-electrode stimulation at time points greater than 90 days post-induction. Additionally, this increased responsiveness over time correlated with morphological changes in the axon initial segment. Finally, we developed a parallel paired-stimulation paradigm to modulate spike timing–dependent plasticity across multiple neuronal pairs, paving the way for population-wide bidirectional control of synaptic weights in biological neuronal networks.