Interplay between homeostatic synaptic scaling and homeostatic structural plasticity maintains the robust firing rate of neural networks

Abstract

AbstractCritical network states and neural plasticity are essential for flexible behavior in dynamic envi-ronments, allowing for efficient information processing and experience-dependent learning. Synaptic-weight-based Hebbian plasticity and homeostatic synaptic scaling were considered the key mechanisms in enabling memory while stabilizing network dynamics. However, the role of structural plasticity as a homeostatic mechanism is less consistently reported, especially under activity inhibition, leading to an incomplete understanding of its functional impact. In this study, we combined live-cell microscopy of eGPF-labeled neurons in organotypic entorhinal-hippocampal tissue cultures with computational modeling to investigate the response of spine-number-based structural plasticity to activity perturba-tions and its interaction with homeostatic synaptic scaling. Tracking individual dendritic segments, we demonstrated that inhibiting excitatory neurotransmission does not monotonically regulate den-dritic spine density. Specifically, inhibition of AMPA receptors with a low concentration of 2,3-dioxo-6-nitro-7-sulfamoyl-benzo[f]quinoxaline (NBQX, 200 nM) significantly increased spine density, while complete AMPA receptors blockade with 50 {micro}M NBQX reduced spine density. Motivated by these findings, we developed network simulations incorporating a bi-phasic structural plasticity rule governing activity-dependent synapse formation. We showed that this biphasic rule maintained neu-ral activity homeostasis under stimulation and permitted either synapse formation or synapse loss, depending on the degree of activity deprivation. Homeostatic synaptic scaling affected the recurrent connectivity, modulated the network activity, and influenced the outcome of structural plasticity. Specifically, it reduced stimulation-triggered synapse loss by downscaling synaptic weights and res-cued silencing-induced synapse loss by upscaling recurrent inputs, thus reactivating silent neurons. Our interaction between these mechanisms offers an explanation for divergent findings in the existing literature. In summary, calcium-based synaptic scaling and homeostatic structural plasticity rules compete and compensate for one another, ensuring efficient and robust control of firing rate home-ostasis. Significance StatementO_LIThis work combined systematic computer simulations and in vitro experiments to explore the in-terplay between homeostatic structural plasticity and synaptic scaling under conditions of activity deprivation. C_LIO_LIWe identified a non-monotonic relationship between neural activity and spine numbers, where par-tial inhibition of synaptic transmissions increased spine density, while complete inhibition reduced it. C_LIO_LIPartial inhibition led to increased spine sizes across all initial spine sizes, whereas complete inhibition selectively increased the size of relatively large spines. C_LIO_LIA biphasic, spine-number-based homeostatic structural plasticity (HSP) rule reconciled previously divergent experimental findings regarding activity-dependent changes in spine density. C_LIO_LIUsing an engineering and complex systems framework, we proposed that the biphasic HSP rule incorporates a negative feedback mechanism and acts as a redundant and heterogeneous mechanism alongside the synaptic-weight-based homeostatic synaptic scaling (HSS) rule. C_LIO_LIBy comparing simulation and experimental results, we demonstrated the necessity of HSP-HSS interplay in maintaining firing rate homeostasis. C_LIO_LIBoth plasticity rules are driven by intracellular calcium concentration, which reflects cumulative neural activity. Thus, we propose that integral feedback control is critical in firing rate homeostasis. C_LI