For better or worse, complement proteins tag synapses for elimination by microglia. A paper published July 13 in Nature Neuroscience identifies a neuronal protein that keeps this process in check. Researchers led by Gek-Ming Sia at the University of Texas in San Antonio report that neurons spit out SRPX2, a protein that directly binds the complement protein C1q and halts the complement cascade. The researchers found that SRPX2 limits synaptic pruning by microglia during brain development in mice. They proposed that it could play a part much later in life, in neurodegeneration, where complement-mediated pruning has been implicated in the destruction of synapses. Autoantibodies to SRPX2 have been found in the cerebrospinal fluid of Alzheimer’s disease patients (Lim et al., 2019). 

  • SRPX2 binds C1q, inhibiting the classic complement cascade.
  • Without SRPX2, microglia engulf more synapses.
  • The visual and somatosensory systems seem most sensitive.

In the developing brain, the complement system tags weakened or redundant synapses for consumption by microglia (Dec 2007 news; Schafer et al., 2012). However, overzealous pruning has been implicated in driving synaptic destruction in neurodegenerative disease (Aug 2013 news; Nov 2015 conference news; Jul 2019 conference news). Though scientists have identified numerous complement inhibitors outside of the brain, they have thus far detected none within the CNS.

Previously, the researchers hypothesized that sushi repeat protein X-linked 2 (SRPX2) could fit the bill. Secreted by neurons, the protein enhances synaptic signaling, and variants in its gene have been tied to language and social behavior disorders (Sia et al., 2013; Chen et al., 2017; Soteros et al., 2018). Furthermore, sushi domains are common among complement inhibitors (Gialeli et al., 2018). 

Co-first authors Qifei Cong and Breeanne Soteros started by investigating whether SRPX2 interacts with C1q—the complement protein that binds synapses and sets off their pruning. In lysates from mouse brain, the researchers detected SRPX2 and C1q in complexes. In cell culture experiments, they found that SRPX2 directly bound C1q, but only when cell membranes were present. This suggested that SRPX2 likely binds C1q  immobilized on the cell surface.

Complement-mediated elimination of synapses sculpts neuronal circuitry during brain development, particularly in the somatosensory cortex and the dorsal lateral geniculate nucleus (dLGN), a region in the thalamus that receives visual input from the retina. Neurons in these regions expressed SRPX2, while microglia, astrocytes, and oligodendrocytes did not. Using immunohistochemistry, Cong and colleagues spotted SRPX2 mingling with C1q, sometimes in excitatory pre-synapses. SRPX2 knockout mice produced normal levels of C1q but an excess of C3, a protein in the complement cascade downstream of C1q that is required to trigger microglia to engulf synapses. This suggested that SRPX2 blocks the function of C1q, halting the cascade.

Neuronal Defender. SRPX2 mRNA (red) spotted in neurons (left), but not microglia (Iba), astrocytes (GFAP), or oligodendrocytes (Olig2) of the dorsal lateral geniculate nucleus (top) and layer 4 of the somatosensory cortex (bottom). [Courtesy of Cong et al., Nature Neuroscience, 2020.]

What were the consequences of taking the brakes off the cascade? Using mice devoid of SRPX2, C3, or both, the researchers pieced together the role of the complement inhibitor in the development of retinal axons and the somatosensory cortex. In both, SRPX2 transiently kept synaptic pruning in check, guiding the growth of neural circuitry. Without SRPX2, microglia excessively thinned dendritic spines. C3 was required for pruning in SRPX2 knockouts, directly implicating the complement cascade. SRPX2 knockout mice had fewer synapses in their somatosensory cortices and this persisted beyond the pruning stage.

Sia told Alzforum that his lab is now exploring if SRPX2 protects synapses during aging, or during neurodegenerative disease. He thinks SRPX2 expression might track with that of C1q, which spikes during development and then creeps up again with age.

Shane Liddelow of New York University would not be surprised if SRPX2 rose and fell with complement. He proposed that weak SRPX2 expression or secretion might hasten synaptic deterioration during aging and disease. Along those lines, Sia wondered whether SRPX2 could turn out to be a resilience factor, shielding against synaptic damage inflicted by Aβ plaques or tau tangles, for example.

“Since the classical complement pathway, which is initiated by C1q, seems to be overactivated in schizophrenia, Alzheimer's disease, and other CNS diseases, it is intriguing to speculate that expression or activity of SRPX2 or other endogenous complement inhibitors might regulate synapse loss and neuronal injury in these diseases,” wrote Borislav Dejanovic of the Broad Institute of MIT and Harvard. He added that it will be interesting to see if other sushi-domain-containing proteins expressed in the brain also inhibit complement, and to investigate which types of neurons and synapses are affected by these pathways.—Jessica Shugart


  1. This paper describes a novel complement inhibitor, SRPX2, expressed by neurons, that directly binds C1q and suppresses complement activation, C3 deposition on synapses, and subsequent C3-mediated synaptic elimination. They used CRISPR technology to develop an SRPX2 KO mouse model and examined complement, dendritic spines, and synapses in two brain regions: the retinogeniculate pathway in dorsal lateral geniculate nucleus (dLGN) and Layer 4 of the somatosensory cortex (L4 SS) up to age postnatal day P90. Interestingly, they found both spatial- and temporal-dependent effects of SPRX2 deletion. In dLGN, C3 levels were elevated at P4 and P10 but returned to WT levels by P30. In L4 SS, C3 levels were similar to WT at P30 but elevated at P60 and returned to WT levels by P90. C3 levels were unaffected in Layers 2 and 4 of SS. Microglial engulfment was increased and synapse number decreased at peak C3 elevation (i.e., P4 and P10 in dLGN and P60 in L4 SS). Although the C3 levels returned to normal by P90 in L4 SS, the synapse number remained low, suggesting a possible long-term effect of over-pruning during this apparent critical window during brain development. The authors also double-crossed the SPRX2 KO with C3 KO mice and showed that C3 KO dominated, suggesting that SRPX2 is upstream of C3. Synaptic pruning increased in SPRX2 KO mice and reduced in C3 KO and SPRX2-C3 double KOs.

    This new data is very consistent with published reports, including our own, showing regional differences in the effects of constitutive complement deletion. We and others have shown that C3 depletion protects against age- and AD-associated synapse loss and cognitive decline in WT and amyloid Tg mouse models. Similar results have been shown in complement-deficient tau models.

    What’s exciting is that this novel complement inhibitor, which clearly plays an important role during brain development, may also play a role later in life. It will be very interesting to determine whether SPRX2 levels are reduced with aging and/or neurodegeneration in human and animal model brains, thereby allowing classical complement activation leading to more C3 deposition on synapses and binding to CR3 on microglia to induce synaptic elimination. If so, elevating SPRX2 levels therapeutically may suppress complement-induced microglia-mediated synaptic pruning that occurs in aging (Shi et al., 2015) and/or neurological diseases, including Alzheimer’s, Parkinson’s, and schizophrenia. However, the authors point out that they do not yet know whether SPRX2 inhibits other molecules or pathways. More research is needed, but this work suggests a new potential therapeutic target (or pathway).

    Recently, Drs. Wei Cao and Hui Zheng reported that Type 1 Interferon drives C3-mediated microglial synaptic elimination in Alzheimer’s disease (Roy et al., 2020). They demonstrate the presence of nucleic acid (NA), likely derived from neurons, in a subset of amyloid plaques in both human AD brain and amyloid mouse models, that are surrounded by reactive microglia of the neurodegenerative phenotype (MGnD). They hypothesize that the NA+ plaques stimulate an IFN-induced innate immune response that promotes the conversion of microglia to the MGnD phenotype, which then phagocytose local synapses via IFN-C3-mediated synaptic elimination. Interestingly, hippocampal injection of RNA-containing amyloid caused a significant increase in complement gene expression, including C3, and a downregulation of negative regulators of complement (Prelp and Cd55). They convincingly demonstrate that Type 1 IFNβ signaling, like SPRX2, is upstream of C3.

    Taken together, it is clear that innate immunity plays a large role in synaptic health during brain development, aging, and neurological disorders. These papers help to identify drivers and repressors of these signaling pathways that ultimately end with complement-mediated synaptic engulfment. Further understanding of signaling, molecules including cell specificity, brain region and timing of effect, will hopefully help to inform drug development for treatment of developmental and neurodegenerative diseases.


    . Complement C3-Deficient Mice Fail to Display Age-Related Hippocampal Decline. J Neurosci. 2015 Sep 23;35(38):13029-42. PubMed.

    . Type I interferon response drives neuroinflammation and synapse loss in Alzheimer disease. J Clin Invest. 2020 Apr 1;130(4):1912-1930. PubMed.

  2. This unique paper deserves a broad readership because of the clever strategies the authors used in determining the role of inhibition of the complement system (via SRPX2) in regulating C1q activity, C3 levels and, in turn, preventing synaptic elimination via microglia. This is truly a well-done study. It involves SRPX2, C3, and double-knockout mouse lines coupled with excellent tools/techniques (e.g., electrophysiology, retinogeniculate synapse elimination assay in the dLGN in vivo, etc.) that very well support the authors’ conclusions on the SRPX2-C1q-C3 pathway in selective synapse elimination at specific time points during development and in specific brain regions.

    While the determination of C3 levels supports their conclusions, it would be important to determine the levels of both C3a/C3b (in addition to total C3 levels) in SRPX2 KO mice, which would further prove the increased activity of C1q in SRPX2 KO mice.

    A note regarding Figure 4: While the evidence presented strongly argues for differences in the number of functional synapses in SRPX2 KO cells, there are a couple of explanations that need further exploration. First, the reduction in AMPAR currents without a change in NMDA current may suggest the presence of more silent synapses, rather than a reduction in the total number of synapses. However, the non-significant finding for NMDARs is potentially due to a small number of neurons that had very large NMDAR currents, while the majority did show diminished amplitudes. A complementary method that could be used would be to assess the frequency and amplitude of miniature excitatory postsynaptic currents (mEPSCs) for both NMDA and AMPA. A reduction in frequency of mEPSCs for both receptor types, without a corresponding change in paired pulse ratio between wild-type and SRPX2 KO neurons, would provide further supporting evidence for the hypothesized reduction in the total number of synapses via increased complement-mediated phagocytosis.

    The authors nicely discuss the potential relevance of SRPX2 and other complement inhibitors’ role in neurological/psychiatric conditions. To complement what is observed, it is important to test if SRPX2 overexpression would do the opposite and reduce C3 levels at P4/P10 and prevent elimination of synapse. Also, it is important to determine which complement inhibitor (if any) may play a regulatory role in synapse elimination at P30 and in adult dLGN.

    Perhaps extending this study to hippocampal synaptic plasticity, the role of microglial C3Rs and any regulators of microglial complement receptors, would be important to investigate in the context of AD/ADRD. 

Make a Comment

To make a comment you must login or register.


News Citations

  1. Paper Alert: Does the Complement Devour Synapses?
  2. Curbing Innate Immunity Boosts Synapses, Cognition
  3. Microglia Control Synapse Number in Multiple Disease States
  4. Do Microglia Finish Off Stressed Neurons Before Their Time?

Paper Citations

  1. . Putative autoantibodies in the cerebrospinal fluid of Alzheimer's disease patients. F1000Res. 2019;8:1900. Epub 2019 Nov 11 PubMed.
  2. . Microglia sculpt postnatal neural circuits in an activity and complement-dependent manner. Neuron. 2012 May 24;74(4):691-705. PubMed.
  3. . The human language-associated gene SRPX2 regulates synapse formation and vocalization in mice. Science. 2013 Nov 22;342(6161):987-91. Epub 2013 Oct 31 PubMed.
  4. . Next-generation DNA sequencing identifies novel gene variants and pathways involved in specific language impairment. Sci Rep. 2017 Apr 25;7:46105. PubMed.
  5. . Sociability and synapse subtype-specific defects in mice lacking SRPX2, a language-associated gene. PLoS One. 2018;13(6):e0199399. Epub 2018 Jun 19 PubMed.
  6. . Novel potential inhibitors of complement system and their roles in complement regulation and beyond. Mol Immunol. 2018 Oct;102:73-83. Epub 2018 Jun 7 PubMed.

Further Reading


  1. . SRPX2 mutations in disorders of language cortex and cognition. Hum Mol Genet. 2006 Apr 1;15(7):1195-207. Epub 2006 Feb 23 PubMed.

Primary Papers

  1. . The endogenous neuronal complement inhibitor SRPX2 protects against complement-mediated synapse elimination during development. Nat Neurosci. 2020 Jul 13; PubMed.