Encoding directional firing patterns in tissue-engineered neural networks by modulating endogenous plasticity pathways through optogenetic stimulation.

Introduction: Here, we demonstrate a method for encoding directional firing patterns in engineered neural networks by modulating endogenous neural plasticity pathways using spatiotemporally patterned optogenetic stimulation. Neural networks in the brain and the spinal cord control a wide range of activities from locomotion to sensing and memory. A key aspect of these networks that determines their function is the well-defined map of directional connectivity between the neurons that make up the network. One of the major drawbacks of neural networks engineered in vitro to model neural development and disease is the lack of defined directional connectivity. In neural networks that are formed through spontaneous self-organization of neural populations, neurons tend to form synaptic connections in random directions, leading to amorphous networks with limited functionality. Here, we explore an approach for encoding directional connectivity in engineered neural networks by evoking spike timing-dependent plasticity (STDP). STDP is a form of Hebbian plasticity where synapses are strengthened or pruned depending on the sequence of firing of the pre- and post-synaptic neurons (Fig. 1). We hypothesize that applying spatiotemporally patterned external stimulation to an otherwise randomly interconnected neural network can lead to the emergence of directional firing patterns through STDP. To demonstrate, we form neural networks on microelectrode arrays (MEA) using stem cell-derived optogenetic neurons and apply patterned optical stimulation. Our results show that this approach can successfully encode directional connectivity in engineered neural networks.

Materials and

Methods: We obtained optogenetic neurons by directed differentiation of optogenetic mouse embryonic stem cells. Noggin, bFGF, and FGF-8 were used for neural induction, followed by retinoic acid for neural differentiation, producing populations of glutamatergic neurons expressing AMPA and NMDA receptors (verified by RT-qPCR). Neurons were then embedded within a fibrin matrix and seeded in a 4 mm-diameter well on top of MEAs for recording and optical stimulation (Fig. 2). Samples were cultured on the MEAs for 4 days to allow the formation of an initial network. Starting on day 5, we implemented a daily training regimen using a pattern illuminator (Fig. 3) to apply patterned optical stimulation. Neural networks were stimulated with a 100 µm-wide beam moving diagonally across the sample once per second for 1 hour each day.

Results, Conclusions, and Discussions: Fig. 4 shows a comparison of the firing patterns of a stimulated neural network (after the 3-day training regimen) and a control (untrained) sample. Each row of the raster plot represents data recorded from one electrode on the MEA, and the rows are ordered with respect to the direction of the diagonal sweep pattern. The vertically aligned network bursts in the control sample indicate synchronous firing, i.e., no directional propagation. In the trained sample, however, the network bursts appear skewed, with the bursts propagating through the network in the direction of applied stimulation pattern. To quantify this apparent directional propagation, we plot the timepoint corresponding to the local maxima of the burst recorded at each electrode against the position of the respective electrode on the MEA (Fig. 5). The slope of the best-fit line is the propagation delay along the direction of stimulation. Zero slope indicates near-simultaneous bursting across the network (no discernable directionality), and positive values indicate successful encoding of the applied stimulation direction within the neural network. Comparing the network responses on the first, second, and third days of training, we find that the directionality is encoded gradually throughout the training regimen (Fig, 5). Interestingly, at the start of stimulation on the third day, the network already exhibits a positive slope, showing that the encoded pattern from the previous day’s training has persisted, suggesting permanent modulation of the internal synaptic connectivity map. Overall, our results demonstrate that spatiotemporally patterned external stimulation is a viable approach for encoding directional firing patterns in engineered neural networks. Further advances may enable the development of networks with defined connectivity maps mimicking various neural circuits of the brain and the spinal cord, thus serving as functionally relevant in vitro models for neural development and disease.

Acknowledgements (Optional): This work has been published as a pre-print: Pagan-Diaz, Gelson J., et al. "Engineering 3D neuronal networks with directional endogenous neuronal plasticity pathways." bioRxiv (2023): 2023-05. DOI: 10.1101/2023.05.17.540876