Building an astrocyte network in vitro to model network level interactions, metabolism and functions, in health and disease

Glial cells, named after the Greek word for 'glue', were once considered to be merely passive support for neurons(1, 2), but recently there has been a shift from this 'neuron-centric' view, towards an increasing recognition glial cell's key roles(3-6). In particular, astrocytes (ACs) are now considered fundamental in multiple CNS functions, including blood-brain-barrier interactions, inter-cellular signalling, metabolic homeostasis, and maintaining synaptic connections (7-14).

In fact, ACs are vital for brain metabolism and to maintain normal neuronal function(15, 16) (Figure 1A), as they (i) contact capillaries with specialised processes, modulating glucose transport and other molecules into the extracellular space(9); (ii) buffer ions to maintain homeostatic gradients(17); and (iii) store glycogen, which can then be converted to lactate and used for energy by neurons(18). Moreover, they form part of the "tripartite synapse" with neurons, reuptaking extra-synaptic glutamate to prevent excitotoxicity (19) and replenishing neurotransmitter stores in neurons.

Additionally, ACs have been implicated in numerous neurodegenerative disorders (NDDs; 20-22), such as ALS (23-25). NDDs are generally progressive and incurable, with limited therapies (26, 27) and some - like ALS - have clear, AC-dependent non-cell-autonomous phenotypes (28, 29). This has been often linked to the loss of metabolic and homeostatic support by ACs (8, 30, 31). Therefore, metabolic support from ACs represents an attractive target for future therapies (28, 29, 32-34).

In vivo animal models and ex vivo human primary tissue have been crucial in understanding the pathological mechanisms of neurodegeneration (35, 36). However, a unifying constraint of these models is their inherent complexity and, in the case of animal models, the pathological divergence from human disease, often limiting their translational potential (37, 38) In vitro models provide a perfect complementary reductionist model for mechanistic experiments and molecular pathway dissection. In particular, the advent of iPSCs (39, 40) has revolutionised the field by allowing us to generate iPSCs from any patient and recapitulate the disease in cultures by differentiating them into the desired CNS cell types (24, 41-44).

iPSC-based modelling has shown considerable promise for modelling NDDS; for example, in ALS, they have efficiently recapitulated existing disease phenotypes(24, 42) and discovered new pathological features(23, 45) However, there is still a considerable gap between in vitro and in vivo systems, particularly when considering cellular networks and circuitry. In vitro 3D cultures (46, 47) and organoids (48), can be used to obtain more physiological models, but at the expense of ease-of-manipulation and control of cellular components (49, 50). Instead, bioengineering techniques, such as micro-fluidics and micro-patterning, can be used to enhance conventional culture systems, better mimicking in vivo environments, in vitro (51, 52).

ACs form functional networks in vivo(53-55), physically connected by intercellular gap junctions composed of connexins allowing for the intercellular passage of ions, metabolites and secondary messengers(56, 57) Furthermore, intercellular calcium waves (ICW), a characteristic of AC networks, depends on these gap junctions for propagation over proximal regions(58). AC networks are dynamic (19) and consist of hundreds of ACs, each maintaining their own non-over-lapping domains (59). Moreover, changes in AC ICWs and alterations in connexins have been reported in neurodegeneration (60, 61). However, current modelling platforms to cannot recapitulate AC networks in vitro, which limits our understanding of how AC network properties contribute to or protect from neurodegeneration.