The basal forebrain (BF)-hippocampus (HPC) circuit is indispensable for learning and memory, and in vitro models are essential for dissecting its age-related decline. Nonetheless, current culture methods endure brief survival or confine cells to two dimensions, leaving the circuit’s progressive degeneration refractory to long-term investigation. Here, we developed a simple, three-dimensional (3D) compartmentalized co-culture model that mimics the anatomical organization of BF and HPC neurons. Results demonstrate that basal forebrain cholinergic neurons (BFCNs) co-cultured with primary HPC neurons remain viable for more than two months without exogenous growth factors, significantly promoting BFCNs growth, polarity development, and functional maturation. In this system, BFCNs somata were confined within the hydrogel, whereas cholinergic axons extended toward adjacent hippocampal area, reaching 1681.9 ± 351.8 μm by week 5—significantly longer than in BFCNs monocultures. This model can successfully recapitulate age-dependent progressive neuronal degeneration during long-term culture, validating this long-term co-culture as a platform for studying circuit aging and degeneration. Therefore, this low-cost and highly physiological platform provides a new avenue for in-depth investigations into the mechanisms of neurodegenerative diseases.

Neuronal senescence-like states are increasingly implicated in age-related neurodegeneration, yet the neuron-intrinsic regulators that drive these phenotypes remain poorly defined. Guided by prior transcriptomic analysis of aged basal forebrain cholinergic neurons, we investigated Srebf2 and Zmiz1 using primary basal forebrain neuronal cultures with adeno-associated virus-mediated gain- and loss-of-function, quantitative immunocytochemistry, and low-input transcriptomic profiling of fluorescence-activated cell sorting-isolated neurons. Both perturbation strategies produced the expected directional changes in target transcripts. Overexpression of either gene increased the other, whereas knockdown did not elicit reciprocal suppression, indicating asymmetric regulatory coupling. Phenotypically, Srebf2 showed a bidirectional association with senescence-like changes, as both overexpression and knockdown increased p16 and p21, whereas Zmiz1 acted more directionally, with overexpression increasing and knockdown reducing these markers. Transcriptomic profiling revealed broad direction-dependent remodeling, including a set of 55 genes that changed concordantly across perturbation directions. Pathway analysis further showed specialization, with Zmiz1 preferentially associated with an Alzheimer’s disease-related signature and Srebf2 more strongly linked to cholinergic synapse programs. Together, these findings identify Srebf2 and Zmiz1 as coupled but non-equivalent regulators of a neuronal senescence-like program.

There was an error in the original publication