Table 1.

Key studies investigating mechanisms of neural circuit sculpting by microglia.

mechanism	brain region	model	observations	reference
pruning	developing retinogeniculate system	mouse	labelled microglia show close apposition to synapses and contain pre-synaptic elements pruning targets less-active inputs after pharmacological manipulation of activity pruning of synapses was significantly reduced in CR3 knock-out microglial engulfment of synapses significantly impaired in CR3 and C3 knock-outs	[37]
	motor cortical projection neurons	male SD rats	activation of microglia with LPS reduced GAD+ punctae microelectrode array in motor cortex showed increased power of γ-band suggestive of synchronicity showed increased levels of BDNF, FGF2, Bcl-2 which promote survival pharmacological inhibition of microglia with minocycline prevented effect on GAD+ punctae and reduced neuronal survival	[54]
	organotypic hippocampal slices; cultured hippocampal neurons	mouse	depletion of microglia with clodronate liposomes increased frequency of excitatory postsynaptic currents suggesting increased synaptic density co-incubation of microglia with cultured neurons reduced synapse density and levels of adhesion molecules synCAM-1 and protocadherin	[55]
	hippocampal CA1 neurons and dorsal horn	adult male mice	after spared nerve injury, spine density in CA1 reduced but increased in dorsal horn with impaired long-term potentiation (LTP) in hippocampus but increased LTP in the dorsal horn, this was mirrored by decreased BDNF levels in hippocampus and increased levels in the dorsal horn, TNFα increased in both regions pharmacological inhibition and genetic ablation of microglia prevented upregulation of TNFα and changes in spine density in both regions TNFR1 knock-outs showed no change in spine density after spared nerve injury	[56]
synaptogenesis	motor cortex layer V pyramidal neurons	mouse	tamoxifen-induced Cre system to drive diphtheria toxin receptor expression in CX3CR1+ cells allowing diphtheria toxin-inducible depletion of microglia depletion resulted in reduced spine formation and reducing learning removal of BDNF from microglia broadly phenocopied the original depletion experiment with the exception of the novel object task	[57]
	cultured hippocampal neurons	rat	applied microglia to cultured hippocampal neurons without direct contact resulting in increased synaptic density, showed high levels of IL-10 recombinant IL-10 application increased synaptic density knock-down of IL-10 prevented synaptic formation, as did LPS pre-treatment	[58]
	dorsal root ganglion	adult female SD rats	application of artemin, a glial derived neurotrophic factor (GDNF), after lumbar dorsal root injury failed to regenerate large diameter myelinated afferents but induced regeneration of nociceptive smaller diameter calcitonin gene-related peptide (CGRP+) axons, thereby enhancing recovery of nociceptive behaviour, and also increased isolectin B4 (IB4+) axon regeneration NGF showed differential targeting of regenerating axons with increased density of deeper laminae and no effect on IB4+ axons	[59]
	somatosensory cortex	mouse	in vivo multiphoton imaging of developing cortex in Iba1-EGFP mice, labelled neurons with RFP, used GCaMP6m to visualize calcium transients and lifeact-mCherry to visualize actin showed microglial contact induced calcium transients, F-actin accumulation and filopodial formation which was reduced by pharmacological inhibition and genetic ablation of microglia	[60]
death	Purkinje cells developing cerebellum	mouse	showed microglia with Purkinje cell marker inclusions surrounded by apoptotic Purkinje cells elimination of microglia with clodronate liposomes increased number of Purkinje cells superoxide and hydrogen peroxide scavengers and NADPH oxidase inhibitors reduced Purkinje cell death	[61]
	neuron cultures	mouse	LPS-treated microglia induced neurite beading and death which was also seen by TNFα, this was prevented by MNDAR inhibition and glutamine starvation suggesting excitotoxicity neutralizing antibodies against TNFR1 prevented this increase in glutamate levels and also glutaminase expression in microglia gap junction inhibitors prevented glutamate accumulation suggesting a role for microglial glutamate release	[62]
	cultures, optic nerve, post-mortem brain tissue	SD rats, mouse, human	knock-outs of CSF1, which lack microglia, showed a lack of A1 astrocyte induction LPS-treated microglia induce A1 astrocytes as do IL-1α, TNF, C1q co-culturing A1 astrocytes with retinal ganglion cells results in almost 100% cell death after optic nerve crush, neutralizing antibodies against IL-1α, TNF, C1q to prevent A1 astrocyte formation prevented retinal ganglion cell death; this was also seen in triple knock-out IL-1α/TNF/C1q mice which lack A1 astrocytes A1 astrocytes are found in post-mortem tissue from patients with Alzheimer's, Huntington's, Parkinson's, amyotrophic lateral sclerosis and multiple sclerosis	[63]
survival	layer V cortical neurons	mouse	microglial staining showed accumulation on post-natal tracts in layer V which then dispersed at P14 pharmacological inhibition of microglia with minocycline and genetic ablation increased TUNEL+ (apoptotic) cells in layer V in vitro microglia decreased cell death in a transwell system, detected IGF1 in medium inhibitor of IGF1R phosphorylation and knock-down of IGF1 attenuated survival of these layer V neurons in vivo, and IGF1 was able to rescue cell death in minocycline-treated animals	[64]
	motor cortical projection neurons	male SD rats	activation of microglia with LPS reduced GAD+ punctae microelectrode array in motor cortex showed increased power of γ-band suggestive of synchronicity showed increased levels of BDNF, FGF2, Bcl-2 which promote survival pharmacological inhibition of microglia with minocycline prevented effect on GAD+ punctae and reduced neuronal survival	[54]