Please eat (only part) of me: synaptic phosphatidylserine cues microglia to feast

Microglia, the brain's tissue‐resident macrophages, contribute to the developmental elimination of extranumerary synapses and to pathologic synapse loss in mouse models of neurodegeneration. Two papers published in The EMBO Journal reveal that phosphatidylserine (PS) is a neuronal cue for microglial synapse elimination.

New studies implicate exposed phosphatidylserine in developmental synapse elimination via both complement‐ and TREM2‐dependent mechanisms.

Synapse elimination, or pruning, is the normal loss of select synapses during development and learning. Pruning is activity‐dependent, such that “stronger” synaptic partners are preserved while others are lost; this is critical for proper neural circuit refinement (Goda & Davis, 2003). This process may go awry in neurodegeneration and other neurological diseases—abnormal synapse loss is believed to lead to or worsen such conditions. Microglia play an important role in the elimination of extranumerary synapses during both development and disease (Lee & Chung, 2019). Further, a recent study highlights the importance of microglia‐mediated synapse loss during “normal” forgetting in rodents (Wang et al, 2020). We know several molecular mechanisms regulating microglial synapse pruning: The classical complement component C1q tags electrically weak excitatory synapses for C3 receptor‐dependent elimination, the loss of CX3CR1‐CX3CL1 signaling leads to an abundance of hippocampal synapses with functional consequences, and several phagocytosis‐linked receptors such as the Alzheimer risk gene Trem2 mediate normal synapse elimination (reviewed in Lee & Chung, 2019). In addition, astrocytes, the most abundant glial cells in the central nervous system (CNS), also contribute to the refinement of neuronal circuits through both the formation (Allen & Eroglu, 2017) and receptor‐mediated elimination of synapses (Lee & Chung, 2019). Despite this wealth of knowledge, there is much more to understand about how glia sculpt neural circuits.

Two recent papers describe phosphatidylserine (PS) as a new neuronal cue for microglial synapse elimination (Li et al, 2020; Scott‐Hewitt et al, 2020; Fig 1). PS is a ubiquitous cytoplasm‐facing phospholipid that “flips” out on the surface of apoptotic cells as an important “eat me” signal for phagocytes. The two studies use a fluorescently labeled probe to show that PS is exposed at select retinogeniculate and hippocampal synapses and that these synapses are preferentially engulfed by microglia. In the visual system, PS is exposed on projections from the ipsilateral eye, which in mice are preferentially removed during a process called eye‐specific segregation. Scott‐Hewitt et al, 2020 demonstrate that C1q knockout mice have increased presynaptic PS and decreased eye‐specific segregation in the lateral geniculate nucleus, suggesting that C1q‐dependent pruning is mechanistically linked to PS exposure. They demonstrate in vitro that loss of known PS receptor TREM2 phenocopies cloaking of exposed PS with annexin V, with decreased microglial synapse engulfment. Li et al identify a previously unknown PS receptor. GPR56, also known as ADGRG1, is an adhesion receptor with previously established roles in myelination through its expression by oligodendrocyte precursor cells (Giera et al, 2015) and cortical patterning through interactions with collagens (Li et al, 2008). The authors find that the GPCR autoproteolysis‐inducing (GAIN) domain of the S4 splicing isoform of GPR56 binds directly with PS to mediate microglia‐mediated synapse pruning during development. Together, these studies uncover presynaptic PS exposure as an upstream mediator of both complement‐ and TREM2‐dependent microglial synapse elimination and identify a new receptor underlying this process.

Figure 1. Microglial synapse pruning is triggered by locally exposed phosphatidylserine.

Phosphatidylserine (PS), normally cytoplasm‐facing, is locally exposed in a subset of synapses during development. Microglia detect exposed PS at presynapses through TREM2 and GPR56. Synapses with exposed PS are then preferentially engulfed by microglia during eye‐specific segregation. This mechanism interacts with previously described complement‐mediated targeting of synapses in as yet undefined ways.

What are the neuronal mechanisms that lead to PS exposure at synapses? While well demonstrated in dying cells, there is also precedent for PS exposure as a flag for phagocytosis of subcellular structures. Diurnal expression of PS at photoreceptor outer segments facilitates their engulfment by retinal pigmented epithelium (Ruggiero et al, 2012). How PS exposure is induced selectively at synapses promises to unlock further clues about how these circuits develop. One clue, especially in the visual system, could be circadian variations. While these cycles are well described for outer segment turnover, there is growing evidence for them in the function of neurons and microglia (Griffin et al, 2019). Whether such rhythms regulate PS exposure and whether astrocyte‐mediated synapse elimination is also affected by these processes remain open questions.

An insight into the upstream mechanism regulating PS exposure is found in a recent preprint (preprint: Park et al, 2020). CDC50a functions as part of the flippase required to maintain PS cytoplasmically (Segawa et al, 2018). Its loss by Thy1‐Cre‐mediated deletion results in constitutive external leaflet expression of PS by Thy1‐expressing neurons (preprint: Park et al, 2020). Interestingly, the authors found that constant PS exposure leads to increased loss of post‐synapses selectively on inhibitory neurons. This is in contrast to the other two studies which demonstrated the role of PS at presynaptic elimination. This discrepancy may suggest directions for future study—such as whether neuronal activity modulates PS exposure. Blocking activity in the visual system in one eye leads to selective pruning of these “weakened synapses” with the expansion of inputs from the normally active eye; this is also true if activity is instead driven in one eye versus the other. When activity is inhibited bilaterally, however, normal pruning is blocked, suggesting there may be some punishment cue from electrically active synapses or that there is a need for electrical “competition” (Lichtman & Colman, 2000). One such proposed signal is retinal astrocyte‐derived TGFβ which causes C1q deposition at presynapses in the visual system. Another potential candidate is IL33 which when released by astrocytes promotes microglia‐mediated synapse elimination during normal development (Lee & Chung, 2019). The identification of these or other signals as upstream of PS exposure at synapses may further tie activity to glia‐mediated sculpting of neuronal circuits.

The complementary studies from the Piao, Stevens, and Matteoli groups fill a major gap in our understanding of synapse elimination by identifying PS as an “eat (part of) me” signal that tags synapses for elimination by microglia. They convincingly demonstrate how this process lies upstream of known mechanisms underlying microglia‐mediated synapse loss. These findings identify a likely central node that may link how neural activity interacts with glia‐mediated circuit sculpting. Future work on how PS is exposed selectively at synapses, how this is linked with neuronal activity, and whether this mechanism is dysfunctional in disease promises to greatly advance our understanding of neural circuit development.

The EMBO Journal (2020) 39: e105924