Astrocyte-Microglia Signaling Controls Developmental Thalamocortical Synapse Refinement

Abstract

Synapse formation and elimination are two developmental processes that concurrently take place in the neonatal brain. Dysregulation of these two processes have been implicated in the etiology and progression of neurodevelopmental and neurodegenerative diseases. Previous work has found that in mice, the first three postnatal weeks are highly active periods of synapse remodeling throughout the entire brain. Glial cells called astrocytes are highly complex neural derived cells that are born and mature during this period. As they mature, astrocytes instruct the formation of synapses through contact with synaptic components and through the secretion of various synaptogenic factors. Microglia by contrast are the tissue resident macrophages of the central nervous system (CNS). During the first three postnatal weeks, microglia sculpt developing synaptic circuits by engulfment of synaptic components through various phagocytic mechanisms. While the field has steadily grown our understanding of the importance of these two cell types in synapse formation and elimination separately, few studies have addressed the possibility of communication between these two cell types to regulate their respective functions at synapses. Here I used the developing visual thalamocortical circuit as a model system to investigate the molecular cross talk between astrocytes and microglia. To address the impact of this communication on synapse development and function, I focused on one factor called Hevin/Sparcl1 which has previously been shown to be necessary and sufficient for thalamocortical synapse formation and plasticity. Previous studies have shown that Hevin induces thalamocortical synapse formation during the second postnatal week in mouse visual cortex. Hevin orchestrates this process by bridging pre-and post-synaptic cell adhesion molecules, Nrxn1α and Nlgn1B. Curiously, I found that despite high levels of Hevin in the maturing primary visual cortex, thalamocortical synapse numbers decrease even during the time when Hevin expression is at its peak. This refinement process, I determined, was dependent on microglia. Using super resolution microscopy, I found that only a subset thalamocortical synapses have Hevin at their cleft and that loss of Hevin aberrantly enhances microglia phagocytic activity. These initial findings suggested that Hevin likely functioned to spare only specific synapses from microglia mediated elimination. To interrogate this possibility, I used an in vitro microglia culture system to assess the transcriptional responses of microglia to Hevin treatment. Surprisingly, this treatment led to robust transcriptional changes in microglia that were distinct from well described immunological stimulation. This screen implicated Toll-like receptors (TLR) 2 and 4 in this transcriptional response. Further studies using our in vitro culture system showed that proteolytic cleavage of Hevin was required to upregulate TLR2 expression in microglia and that its C-terminus alone was sufficient to upregulate TLR2. Moving in vivo, I found that TLR2 expression is strongly developmentally regulated and highly heterogeneously expressed by microglia in the mouse primary visual cortex. Using overexpression studies in vivo, I also found that microglia strongly upregulate TLR2 in response to Hevin or Hevin’s C-terminus and that these TLR2 high microglia have enhanced phagocytic activity both in normal development and after Hevin/Hevin C-terminal overexpression. These findings indicate that Hevin function is regulated by proteolytic cleavage and suggest that Hevin is a dual signal in synaptic development: both to stimulate synapse formation by neurons and enhance synapse elimination by microglia. I next sought to test the functional relevance of the microglia specific response to Hevin. To do this, I used co-immunoprecipitation studies to identify candidate receptors for Hevin on microglia. I found that Hevin and its C-terminus interacted with both TLR2 and TLR4 but seemed to have a stronger affinity for TLR4. Therefore, I used TLR4 KO mice to test if microglia could still be stimulated by Hevin in vivo. I found that TLR4 KO microglia were no longer responsive to Hevin overexpression and had reduced phagocytic capacity compared to WT microglia. Ultimately, I found that TLR4 KO mice had impaired thalamocortical synapse refinement and impaired circuit plasticity. Taken together, my results identify astrocyte-derived Hevin as a synaptogenic molecule that links thalamocortical synapse formation with synaptic refinement mediated by microglia.