Microglial-derived C1q integrates into neuronal ribonucleoprotein complexes and impacts protein homeostasis in the aging brain

,


In brief
The microglial-secreted complement protein C1q is internalized by neurons in an age-dependent manner, undergoes RNA-mediated interactions with neuronal ribonucleoprotein complexes, and alters neuronal protein translation and homeostasis in the adult and aging brain; highlighting how temporally regulated neuroimmune interactions impact critical intraneuronal functions.

INTRODUCTION
Glial and neuroimmune interactions contribute to brain development, maturation, homeostasis, and disease through mechanisms that include direct physical contact, response to local cellular cues, and the secretion of factors into the extracellular milieu.The innate immune complement protein C1q is one such protein produced and secreted by microglia that has been ascribed brain-specific functions.2][3][4][5] Intriguingly, C1q protein levels are profoundly upregulated in the aged human and rodent brain, while other complement proteins such as C3 remain low, 6 indicating that there could be classical complement pathway-independent, agespecific brain functions of C1q.
7][8] Adult mice deficient in C1q (C1qKO) have enhanced plasticity in the hippocampus and reduced cognitive and memory decline when compared with age-matched wild-type (WT) animals-effects that were not mediated through the complement pathway (specifically C3) or changes in adult synapse elimination. 6C1q has been detected within synaptic terminals, axons, and dendrites examined in aging primate and rodent cortices by electron microscopy, 7 revealing that age-dependent functions of C1q could be mediated through intracellular interactions within neurons.
Seeking to uncover neuronal functions of C1q across aging, we utilized an unbiased proteomic approach by conducting C1q co-immunoprecipitation from synaptosomes isolated from development, young adult, and adult brain tissue.These experiments revealed age-specific interactions of C1q with ribosomal proteins and RNA-binding proteins (RBPs) and further led to the discovery of the enrichment of C1q within neuronal ribonucleoprotein (RNP) complexes in situ.Many proteins found in RNP complexes have the propensity to undergo liquid-liquid phase separation (LLPS), where optimal conditions and concentrations of specific proteins and/or nucleic acids trigger reversible phase transition within an aqueous environment into a dense droplet, or ''condensate.'' 9Surprisingly, purified C1q protein undergoes RNA-dependent LLPS in vitro, and the co-localization of C1q with neuronal RNP complexes in situ was also dependent on RNA.Loss of C1q led to age-specific alterations in brain protein homeostasis and learning and memory, providing evidence that these interactions may have functional consequences and revealing a unique role of C1q in the adult brain.

Unbiased proteomic analyses uncover age-specific protein interactions between C1q and RBPs
To uncover age-specific C1q protein interactions, we isolated crude synaptosomes from developmental (postnatal day 5), young adult (2-3 months), and adult (1 year) animals and performed C1q immunoprecipitation (C1qIP) using a highly specific monoclonal antibody against C1q 6 followed by mass spectrometry analysis (Figures 1A-1I and S1A).C1q protein interactions changed dramatically across aging, shifting from proteins with known extracellular domains and functions in development to those with largely intracellular activities in adults (Figures 1A-1I).
One such protein, barrier-to-autointegration factor 1 (BANF1), co-immunoprecipitated with C1q in an age-specific manner and was the most significantly enriched protein we detected in both adult datasets (Figures 1J and S1B-S1E).Using proximity ligation assay (PLA), we observed that C1q and BANF1 interactions could be detected in situ in adult mice (Figures 1K, S1F, and S1G) and that these interactions were not observed within microglia (Figure 1L).BANF1 has well-described functions in nuclear assembly and chromatin organization during cell division and both pro-and anti-viral activities in the cytoplasm. 12However, BANF1 has also been detected in proteomic datasets from synapses 13 and RNP complexes, 14 including stress granules. 15nalysis of our adult C1qIP dataset revealed an unexpected enrichment of proteins involved in translation and RNP interactions, including RBPs and ribosomal proteins (Figure 1H).7][18][19][20][21][22][23][24] Given these findings, we next tested whether C1q could be detected within polysome fractions isolated from brain tissue.Not only was C1q present, but it was also enriched in polysomal fractions when compared with unfractionated total lysate (Figures 1M, 1N, S1H, and S1I).Finally, interactions between C1q and the ribosomal protein RPL10a were detected in situ using PLA (Figures S1J and S1K), supporting that these interactions likely occur in vivo.
Microglial-derived C1q co-localizes with neurons in an age-dependent manner Brain C1q protein levels increase dramatically across aging, [6][7][8] and single-cell sequencing and cell-specific knockout of C1q have confirmed that macrophages within the brain are the primary source. 8,25Intriguingly, the sparse C1q labeling of interneurons was previously found to be sensitive to extended tissue fixation. 8o visualize C1q more consistently in situ, we optimized our staining protocol by reducing the length of paraformaldehyde (PFA) post-fixation and replacing serum with bovine serum albumin (BSA).These changes resulted in unexpected robust Figure 1.Unbiased proteomic analyses uncover age-specific protein interactions between C1q and RNA-binding proteins (A) Volcano plot (generated using Genoppi 10 ) representing the fold change (log 2 FC) and adjusted p value (Àlog 10 ) of proteins identified to co-immunoprecipitate with C1q at postnatal day 5 (P5).Data represent two technical replicates performed on crude synaptosomes isolated from WT and C1qKO littermates (n = 3-4 animals pooled per genotype).Significant ''hits'' (labeled in green) were identified with a false discovery rate (FDR) cutoff of p % 0.1.Bait proteins C1qa, b, and c are highlighted in red.(B) Chart representing the top Gene Ontology (GO)-enriched molecular components identified in the P5 C1q-immunoprecipitation dataset from (A).Chart was generated using ShinyGO software 11 comparing significant protein hits to total proteins uncovered in the dataset.(C) Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) network analysis for significant proteins identified in (A).(D) Volcano plot generated as described in (A), representing C1q-immunoprecipitation data collected from 2-to 3-month-old crude synaptosomes.(E) GO enrichment chart as described in (B) from 2-to 3-month-old C1q-immunoprecipitation dataset from (D). (F) STRING network analysis for significant proteins identified in (D).(G) Volcano plot generated as described in (A), representing C1q-immunoprecipitation data collected from 1-year-old crude synaptosomes.(H) GO enrichment chart as described in (B) from 1-year-old C1q-immunoprecipitation dataset from (G). (I) STRING network analysis for significant proteins identified in (G).(J) Venn diagram representing shared significant proteins identified in (A), (D), and (G).Only three proteins, C1q a, b, and c, were common among all three ages.Barrier-to-autointegration factor 1 (BANF1) is highlighted as a common protein identified between 2-to 3-month and 1-year datasets.(K) Representative images of 2-month-old WT and C1qKO hippocampal CA3 region with proximity ligation assay (PLA) targeted to C1q and BANF1 (purple) and DAPI (cyan).Scale bar is 15 mm (633 magnification).(L) Representative image of 2-month-old Cx3Cr1-GFP hippocampal CA3 region with PLA targeted to C1q and BANF1 (purple) and GFP (green).Scale bar is 15 mm (633 magnification).(M) Blots of total brain lysate and subsequent fractions (12 soluble and pellet) of increasing density isolated from a linear sucrose gradient following polysome fractionation.Blots were probed with a-RPL10, a-RPL5, a-FMRP, a-FUS, and a-C1q antibodies.Molecular weight standards are denoted.(N) Quantification of blots of RPL10, RPL5, FMRP, FUS, and C1q normalized to total protein from total brain lysate and polysome fractions represented in (M).Data represent the mean from 3 biological replicates ± SEM.See also Figure S1.
We observed a high concentration of punctate C1q signal within neuronal soma; however, we also observed C1q labeling throughout the neuropil, in dendritic compartments, and presumably in microglia as has been previously reported 3,6 (Figures S2C-S2E).However, when compared with microglia in situ, significantly more C1q protein co-localized with neurons in adult animals across multiple brain regions, including the hippocampus and motor cortex (Figures 2C-2E, S2F, and S2G, respectively).Moreover, neuronal C1q staining was observed using multiple C1q antibodies derived from different species (Figures S2H-S2J).As has been reported, 8 macrophage-specific C1qKO (Cx3cr1-Cre ER ; C1q f/f ) eliminated all detectable signals, while neuronal C1qKO (Synapsin-Cre; C1q f/f ) had no effect (Figures 2F and 2G), confirming that microglia and brain-associated macrophages are the source of neuronal C1q.
To assess neuronal C1q across aging, we quantified co-localization with a pan-neuronal marker (Milli-Mark) and found that co-localization significantly increased from postnatal development into early adulthood and remained consistent throughout aging (Figures 2H and 2I).To test whether C1q interactions with RNP complexes in situ were specific to neurons, we generated mice with neuron-specific, fluorescently labeled ribosomes (''GFP-trap'': Synapsin-Cre; RPL10a eGFP/+ [L10a-eGFP]).We observed that C1q co-localization with eGFP-labeled neuronal ribosomes could be detected across multiple brain regions using both standard confocal microscopy and super-resolution structured illumination microscopy (SIM; Figures 2J-2L and S2K).
These results reveal that C1q partially co-localizes with neuronal RNP complexes in situ in adult mice, raising questions about intraneuronal functions of C1q in the adult and aging brain.

C1q undergoes RNA-dependent LLPS
A common feature of proteins found in RNP complexes is an intrinsically disordered region (IDR), 26,27 and when we more closely scrutinized the amino acid sequence of C1q, we discovered an IDR within the collagen-like domain that was highly predicted to undergo LLPS (ParSe v2 28 ; Figure 3A).We conducted in vitro experiments using purified human C1q protein and RNA isolated from human brain tissue where we observed droplet formation indicative of LLPS when C1q was combined with RNA, either total or polyA-enriched (Figures 3B and S3A), which was prevented by pretreatment with RNase (Figure 3B).The IDR is conserved across human and mouse C1q (Figures 3A and  S3B, respectively), and RNase-sensitive droplets were similarly detected when mouse C1q protein and mouse total brain RNA were combined (Figure S3C).Using timelapse imaging to monitor LLPS dynamics, we observed droplet enlargement and fusion over time (Figures 3C and 3D; Video S1) and found that RNA was necessary not only for the formation of droplets but also for the maintenance of them; addition of RNase resulted in their dissolution (Figures 3E and 3F; Videos S2 and S3).Finally, we determined that droplet formation was dependent on both C1q and RNA concentration (Figures 3G and S3D).
collagenase, which specifically cleaves the collagen-like domain while leaving the globular head domain intact, 29,30 and no longer observed droplet formation (Figure 3H).Additionally, droplet formation did not occur between purified C1q globular heads (kindly provided by G. Andersen) and RNA (Figure 3H), demonstrating the necessity of the collagen-like domain, and IDR, for RNA-dependent C1q LLPS.We also tested whether the individual C1q peptides (a, b, and c), all of which contain IDRs, could undergo LLPS; however, no RNA-dependent droplets were observed with any of the peptides (Figures 3H and S3E).
The canonical role of C1q is as the upstream regulator of the classical complement pathway, and while C1q is predicted to undergo LLPS, other complement proteins, including C3 and C4, are not (Figures S3F and S3G).Using our in vitro LLPS assay, we found no evidence of droplet formation in the presence of RNA for purified human C3 or C4 proteins (Figure 3I).Thus C1q, but not other complement proteins, undergoes RNAdependent LLPS in vitro that requires the intact hexameric complex and IDR-containing collagen-like domain.

RNA is necessary for C1q interactions with neuronal RNP complexes in vivo
Given that C1q is enriched in neuronal RNP complexes and undergoes RNA-dependent LLPS in vitro, we next sought to determine whether C1q:RNP interactions in vivo were also dependent on RNA.In brain-derived isolated polysomes, RNase treatment affected how proteins, including C1q, fractionated across the linear sucrose gradient, resulting in a shift from heavier to lighter fractions (Figures 4A-4D and S4A).In situ, fixed tissue sections were briefly treated (10 min) with RNase prior to immunostaining, which resulted in the specific elimination of cytoplasmic neuronal C1q labeling (Figure 4E).RNase sensitivity of C1q neuronal staining was further observed throughout aging (Figures 4F and 4G).
The use of high serum concentrations (R20%) during immunostaining precluded neuronal C1q labeling to a similar extent as RNase treatment (Figures 2A and 4E, respectively).Serum may contain heat-stable nucleases, 31 so we next tested whether supplementing serum with RNase inhibitor prevented this effect.
We found that neuronal C1q staining was restored (Figure S4B), confirming that the effect of high serum concentrations is through RNase activity.Additionally, brief PFA exposure (10 min) of tissue sections blocked the effect of RNase on neuronal C1q (Figure S4C), demonstrating the sensitivity of this staining to various conditions and highlighting the importance of optimizing reagent and fixation exposure and timing.
To test whether RNA-dependent C1q interactions are specific to the brain or more ubiquitous, we explored another organ with high endogenous C1q protein levels, the liver. 32C1q protein undergoes RNA-dependent droplet formation in the presence of RNA isolated from human liver tissue, demonstrating that, in vitro, C1q can undergo LLPS with RNA from different tissue sources (Figure S4D).However, in situ, we did not observe RNase sensitivity of C1q staining in liver tissue sections (Figure S4E), indicating that RNA-dependent C1q interactions may be unique to neurons in vivo.

Exogenous C1q protein integration into neuronal RNP complexes in vivo is dependent on endocytosis
To directly test whether exogenous extracellular C1q protein can be internalized by neurons and incorporated into RNP complexes in the brain in vivo, we injected purified mouse C1q protein into the lateral ventricles of neuronal GFP-trap C1qKO mice and performed immunolabeling against C1q in isolated tissue (Figures 5A and 5B).1-h following intracerebroventricular (ICV) injection, we detected C1q protein in the ipsilateral hippocampus (Figure 5B), where co-localization of exogenous C1q protein was detected with C1qKO neuronal ribosomes and, similar to WT neurons, was sensitive to RNase treatment prior to immunostaining (Figures 5C and 5D).
Endocytosis-mediated internalization of C1q has been reported previously in T cells, 33 so we next sought to determine whether endocytosis-mediated neuronal internalization of C1q occurs in vivo.We performed ICV injections using a commercial cocktail of cell-permeable dynamin inhibitors to block endocytosis (or their respective negative control compounds) in C1qKO mice 30-min prior to injection with purified mouse (A) Schematic of C1q, with the collagen-like domain containing the intrinsically disordered region (IDR) highlighted in blue.Plot representing the regions along the human C1qa amino acid sequence (x axis) predicted to contain an IDR and undergo LLPS (blue), predicted to contain an IDR but not undergo LLPS (red), or predicted to fold into a stable confirmation (black).Plot predictions were generated using ParSe v2. 28B) Representative images of BSA (200 mg/mL) and total human brain RNA (200 mg/mL), human C1q (200 mg/mL), human C1q (200 mg/mL) and total human brain RNA (200 mg/mL), and human C1q (200 mg/mL) and total human brain RNA (200 mg/mL) pretreated with RNase A (1:1,000).RNA-binding dye SYTO RNASelect (500 nM; green) was added to all conditions, and images were processed using identical settings.Scale bar is 15 mm (203 objective); bright field (BF).(C) Schematic of experimental design for in vitro C1q:RNA LLPS droplet formation assay.(D) Representative still images captured across timelapse imaging at 0 0 , 5 0 , 10 0 , 15 0 , and 20 0 of human C1q (200 mg/mL) and total human brain RNA (500 mg/mL).Scale bar is 10 mm (203 objective); bright field (BF).(E) Schematic of experimental design for in vitro C1q:RNA LLPS droplet RNase sensitivity assay.(F) Representative images captured across timelapse imaging at 0 0 , 1 0 , 3 0 , and 5 0 of human C1q (200 mg/mL) and total human brain RNA (500 mg/mL) following addition of RNase (1:1,000) at T = 0 0 .Scale bar is 10 mm (203 objective); bright field (BF) and SYTO RNASelect (green).(G) Summary diagram of LLPS droplet formation across varying concentrations of human C1q and total human brain RNA.Green circles represent conditions where droplets were observed, and gray circles represent conditions where no droplets were observed.(H) Representative schema and images of human C1q (100 mg/mL) and total human brain RNA (200 mg/mL), human C1q (100 mg/mL) pretreated with collagenase and total human brain RNA (200 mg/mL), human C1q globular heads (100 mg/mL) and total human brain RNA (200 mg/mL), and C1qa peptide (100 mg/mL) and total human brain RNA (200 mg/mL).Scale bar is 15 mm (203 objective); bright field (BF).(I) Representative diagram and images of human C3 or C4 protein (100 mg/mL) and total human brain RNA (200 mg/mL).Scale bar is 15 mm (203 objective); bright field (BF).See also Figure S3 and Videos S1, S2, and S3.
C1q protein (Figure 5E).Pretreatment with the endocytosis inhibitors significantly reduced uptake of exogenous C1q, leading to diminished integration into RNase-sensitive RNP complexes when compared with negative control-treated animals (Figures 5E-5J).We no longer observed a significant effect of RNase treatment on neuronal C1q labeling in tissue from animals that received the endocytosis inhibitors, suggesting that RNA-independent C1q interactions with neurons were not impacted (Figures 5I and 5J).Control experiments utilizing a fluorescently conjugated dextran (10 kDa; Alexa-488), known to be taken up by neurons, 34 similarly demonstrated that pharmacological dynamin inhibition significantly reduced in vivo endocytosis (Figures S5A-S5D).Together, these data demonstrate that exogenous C1q protein can incorporate into neuronal RNasesensitive RNP complexes in vivo in an endocytosis-dependent manner.
Collagen-like domain interactions, but not RNA, mediate C1q neuronal uptake in live acute slices To better visualize uptake by neurons, we generated fluorescently labeled C1q protein (C1q-594) using a maleimide-based commercial conjugation kit (C5 maleimide).Maleimide-conjugated C1q-594 underwent RNA-dependent LLPS in vitro (Figure S6A) and integrated into RNase-sensitive neuronal structures following ICV injection into C1qKO mice (Figure S6B).In live acute brain slices derived from our neuronal GFP-trap mice, bath application of C1q-594 for 1 h (Figure 6A) resulted in neuronal uptake of exogenous C1q and co-localization with GFP-labeled neuronal ribosomes (Figures 6A and 6B).
Cell-surface RNA modified with sialoglycans (glycoRNA) and RBPs may play a role in cell-penetrating protein uptake, 36 so we next tested whether RNA was necessary for C1q internalization.Pretreatment of live acute slices with RNase prior to C1q-594 application did not impact neuronal uptake (Figures 6C  and 6E).Moreover, treatment with RNase following C1q-594 application did not impact C1q labeling of live neurons in acute slices (Figures 6D and 6E), demonstrating that RNase sensitivity of neuronal C1q signal is only observed following fixation and membrane permeabilization.
Similar experiments were performed with fluorescently conjugated C1q protein generated using an amine-reactive commercial kit (Alexa 594 Lightening Link).We discovered that amine conjugation precluded in vitro RNA-driven C1q LLPS (Fig- ure S6C) and in vivo neuronal RNP incorporation following ICV injection in C1qKO mice (Figure S6D).Together, these data reveal how alternative conjugation approaches may have deleterious effects on neuronal C1q localization and LLPS interactions.
We next tested whether altering the collagen-like domain impacted C1q interactions with live neurons, given its importance in mediating RNA-dependent interactions in vitro (Figure 3H).Pretreatment of C1q protein with peptide inhibitor of C1 (PIC), a 15-amino acid peptide that binds to the collagenlike domain of C1q 35 (Figure 6F), resulted in aggregate formation, independent of RNA, and further prevented RNA-dependent C1q LLPS interactions (Figure 6G).Treatment with PIC also significantly reduced internalization of C1q by neurons in live acute slices (Figures 6H-6K), revealing the importance of this domain in neuronal uptake.
Finally, we treated acute live slices with either endocytosis inhibitors or the negative control compounds prior to C1q-594 application (Figure S6E).We observed a significant reduction in C1q uptake by neurons in slices pretreated with endocytosis inhibitors, further supporting our findings that endocytosis in involved in neuronal C1q internalization.Together, these data reveal the importance of the collagenlike domain of C1q in mediating endocytosis-dependent uptake by live neurons and further reveal that RNA, while important in mediating intracellular C1q interactions, is not necessary for internalization.
Macrophage-derived C1q impacts protein translation and fear extinction learning in an age-specific manner RNP complexes occur within multiple cellular compartments and contribute to important biological processes, including RNA homeostasis and transport, proteostasis, and localized translation. 18,37,38Given that C1q is enriched in these structures, we next tested whether C1q altered protein translation across aging.To quantify translation in vivo, we utilized the SUnSet assay, an antibody-based approach that detects the incorporation of the small molecule puromycin into nascent peptides during active translation 39 and has been used previously to assess translation in the brain. 40To ensure the specificity of this assay in vivo, we first injected young adult mice (2-3 months old) with the translation inhibitor anisomycin systemically (intraperitoneal [IP]), as this has been reported to impact neuronal translation 41  (B) Blots of total brain lysate and subsequent polysome fractions from (A) treated with RNase (1:1,000) prior to fractionation.Blots were probed with a-RPL10, a-RPL5, a-FMRP, a-FUS, and a-C1q antibodies.(C) Quantification of blots of RPL10, RPL5, FMRP, FUS, and C1q normalized to total protein from total brain lysate and polysome fractions as shown in (A) and first represented in Figure 1K.Data represent the mean from 3 biological replicates ± SEM. (D) Quantification of blots of RPL10, RPL5, FMRP, FUS, and C1q normalized to total protein from total brain lysate and polysome fractions following RNase treatment as shown in (B).Data represent the mean from 3 biological replicates ± SEM. (E) Representative images of hippocampal immunostaining with a-C1q (purple) and Milli-Mark (neurons; green) in tissue treated ±RNase prior to immunostaining.Scale bar is 500 mm (203 objective).(F) Representative images of CA1 hippocampal pyramidal cell layer from postnatal day (P) 5, P14, P30, 1-year, and 2-year-old WT animals immunostained with a-C1q (purple) ±RNase.Scale bar is 15 mm (633 objective).(G) Quantification of the percentage of neuronal staining (Milli-Mark) that co-localizes with a-C1q immunostaining across ages in (F).Individual data points represent biological replicates; data are represented as the mean ± SEM.Multiple paired t tests.See also Figure S4.

and block
translation-dependent memory formation. 42,431-h following anisomycin (or vehicle) IP injection, we performed ICV injections with puromycin and euthanized animals 1-h later.Quantification of puromycin incorporation, via immunostaining and western blot analysis of brain lysate, revealed a significant reduction in translation following anisomycin treatment (Figures S7A-S7E).
To test whether C1q impacted neuronal protein translation in vivo across aging, puromycin was ICV injected into WT and C1qKO age-and sex-matched littermates and quantified via immunostaining and western blot analysis.To specifically measure neuronal protein synthesis, we used a pan-neuronal marker as a mask to quantify only puromycin co-localized with neurons within the hippocampus and motor cortex.In developmental (postnatal day [P] 5) and young adult (2-3 months) WT and C1qKO animals, we did not detect a significant difference in neuronal translation in vivo (Figures 7A-7D [P5] and 7E-7H [2-3 months]).However, in adult animals (1 year), we detected a significant increase in puromycin incorporation in C1qKO animals when compared with their WT littermate controls (Figures 7I-7L).To further assess the impact of C1q on the brain ''translatome,'' we performed immunoprecipitation against puromycin (Puro-IP) in our injected 1-year-old WT and C1qKO littermates and discovered a subset of differentially regulated proteins undergoing active translation (Figures S7F-S7H).
Global proteomic analysis further demonstrated brain-wide changes in protein content between 1-year-old WT and C1qKO littermates (Figure 7M), revealing an unexpected enrichment of proteins associated with septin complexes in adult WT brain tissue (Figure 7N) and mitochondrial proteins in adult C1qKO brain tissue (Figure 7O; Table S2).While the biological implications of these changes are not yet understood, together these data demonstrate that global loss of C1q leads to brain-wide alterations in translation and protein homeostasis.
[44][45][46][47] To determine whether C1q impacts learning and memory, we utilized our Cx3cr1-Cre ER ; C1q f/f mice to conditionally delete C1q in micro-glia and brain-associated macrophages of young adult mice (6 weeks old). 2 weeks following tamoxifen-induced recombination we assessed the impact of C1q deletion on contextual fear conditioning.Using this associative learning approach, we found that conditional loss of C1q did not impact fear memory acquisition (training; day 1) or retrieval (same context or tone; day 2; Figures 7P and 7Q).However, we observed significant impairments in fear memory extinction in conditional C1q KO animals (novel context, tone; days 3-4; Figures 7P and 7Q).While many factors could contribute to this phenotype, previous data support a role of protein synthesis in fear extinction, 48 and recent evidence suggests that RNA granule trafficking may play a critical role in this process. 49,50Taken together, we demonstrate that an extrinsic immune cell-derived protein, C1q, integrates into neuronal RNP complexes in an age-dependent manner, undergoes RNA-mediated interactions, and impacts neuronal protein translation and homeostasis in the adult brain.

DISCUSSION
Complement-dependent microglia-neuron signaling mediates brain development, maturation, and plasticity in health and disease, 1-6,51-60 yet in the healthy adult brain, the functions of C1q have not been elucidated.Herein we describe an unexpected biophysical property of C1q that mediates unique agedependent intraneuronal interactions that could impact neuronal homeostasis, plasticity, and functions independent of the classical complement pathway.Our data are the first to demonstrate that a non-neuronal secreted immune protein integrates into neuronal RNP complexes and impacts intracellular activities.The specific mechanisms by which C1q is internalized by neurons, how C1q integration into RNP complexes occurs and affects their function, and whether these mechanisms contribute to aging or disease will be interesting avenues to pursue in the future.Importantly, our discovery opens up exciting possibilities exploring how neuroimmune and glial interactions have evolved unique and unexpected downstream biological activities and how temporally regulated RNA and protein interactions may contribute to these functions.(F) Representative hippocampal images of 2-month-old C1qKO mice following ICV injection with negative control cocktail followed by mouse C1q protein, immunostained with a-C1q (purple) and Milli-Mark (neurons; green).Scale bar is 500 mm (203 objective).(G) Representative hippocampal images of 2-month-old C1qKO mice following ICV injection with endocytosis inhibitor cocktail followed by mouse C1q protein, immunostained with a-C1q (purple) and Milli-Mark (neurons; green).Scale bar is 500 mm (203 objective).(H) Representative hippocampal CA3 images as described in (F) ±RNase treatment prior to immunostaining.Scale bar is 15 mm (633 objective).(I) Representative hippocampal CA3 images as described in (G) ±RNase treatment prior to immunostaining.Scale bar is 15 mm (633 objective).(J) Quantification of the percentage of neuronal staining (Milli-Mark) that co-localizes with a-C1q immunostaining from animals described in (H) and (I).Individual data points represent biological replicates; data are represented as the mean ± SEM.One-way ANOVA with Sı ´da ´k's multiple comparisons test.See also Figure S5.
Non-canonical functions of C1q in the adult brain LLPS-driven interactions occur across multiple cellular compartments, contributing to the organization and regulation of molecules within an aqueous environment. 9,37,38Given their transient nature, our current understanding of the dynamics, components, and functions of these structures in vivo is limited.Here we demonstrate RNase-sensitive C1q immunostaining of neurons in situ, providing evidence that altering common technical approaches, such as the duration of aldehyde fixation and serum exposure, may have profound impacts on the durability of RNA-dependent structures.This is further corroborated by studies reporting the impact of extended aldehyde post-fixation on C1q labeling of a subset of GABAergic interneurons 8 and recent work demonstrating the effects of fixation on LLPS structures in cultured cells. 61Our findings highlight the impact of tissue preparation and provide an alternative approach for immunostaining that may enable improved visualization and spatial localization of sensitive RNP complexes and allow for interrogation of their changes under a variety of conditions and models in situ.
Through global proteomic analysis and puromycin-immunoprecipitation we uncovered brain-wide alterations in protein content and active translation (Figures 7M-7O and S7F-S7H; respectively), revealing how C1q could impact critical brain functions in adulthood.In WT animals, for example, we found enrichment of septin complex proteins (Figure 7N), Septins, a family of GTP-binding proteins involved in cytoskeletal regulation and organization, have been implicated in several neural functions, including spine formation, collateral branching and growth of axons, synaptic vesicle release, and the localization of ion channels. 62,63Evidence suggests that, in fungi, septin mRNA and proteins, along with ribosomes, can be transported via endosomes to promote local translation-mediated septin complex formation. 64Whether analogous transport occurs in neurons, if C1q is involved in regulating this process, and how this could impact synaptic function and plasticity will be of interest to examine further.
Our proteomics experiments further revealed an enrichment of mitochondrial proteins in C1qKO mice (Figure 7O).Given that mitochondria fuel local translation during plasticity, 65 and in turn local translation drives mitochondrial biogenesis, 66 future studies should investigate whether C1q impacts synaptic mitochondrial biogenesis through the regulation of local mitochondrial protein translation.Intriguingly, recent evidence suggests that mitochondrial dysfunction in Alzheimer's disease models leads to the accumulation of septins and C1q at synapses. 67ow these interactions impact synaptic health and function across aging will be of interest to further explore.
Adult C1qKO mice have enhanced plasticity in the hippocampus and reduced cognitive and memory decline, 6 and here we report impairments in fear memory extinction learning (Figures 7P and 7Q).Although there could be many factors contributing to these behavioral and physiological changes, we speculate that C1q integration into neuronal RNP complexes could be one mechanism that underlies these phenotypes.We observe an age-specific increase in protein translation in C1qdeficient neurons, indicating that C1q could be acting as a translational repressor.Recent findings demonstrate that another protein containing an IDR, MEG-3, stabilizes LLPS-driven P granules in Caenorhabditis elegans by acting as a surface boundary that regulates droplet interactions with the aqueous environment. 68We propose that C1q may similarly act to stabilize neuronal RNP complexes, helping to maintain their structural integrity and limiting dissolution.These interactions could then contribute to C1q's impact on neuronal protein homeostasis through the regulation of localized protein translation.It will be important to develop new tools to test this hypothesis and to further explore the impact of C1q on local synaptic translation Figure 6.Collagen-like domain interactions, but not RNA, mediate C1q neuronal uptake in live acute brain slices (A) Schematic and representative images of experiments performed in live acute slices with fluorescently labeled C1q.Mouse C1q protein was conjugated to Alexa 594 (maleimide), and bath was applied (1:1,000 of 1 mg/mL) to acute slices from Synapsin-Cre; RPL10a-eGFP; C1q +/À (neuronal GFP-trap) mice for 1 h.Representative images of hippocampal CA3 from acute slices treated with C1q-594 for 1 h.Scale bar is 50 mm (203 objective); C1q-594 (purple) and GFP (neuronal ribosomes; green).(B) Representative single-plane images of hippocampal CA3 as described in (A).Scale bar is 15 mm (203 objective; 53 digital zoom); C1q-594 (purple) and GFP (neuronal ribosomes; green).(C) Schematic and representative images of hippocampal CA3 as described in (A) of slices pretreated with RNase A (1:1,000) prior to C1q-594 incubation.Scale bar is 50 mm (203 objective); C1q-594 (purple) and GFP (neuronal ribosomes; green).(D) Schematic and representative images of hippocampal CA3 as described in (A) of slices treated with RNase A (1:1,000) following C1q-594 incubation.Scale bar is 50 mm (203 objective); C1q-594 (purple) and GFP (neuronal ribosomes; green).(E) Quantification of C1q-594 signal from slices treated as described in (A)-(C).Individual data points represent biological replicates normalized to ''no RNase'' condition; data are represented as the mean ± SEM.One-way ANOVA with Sı ´da ´k's multiple comparisons test.(F) Schematic of C1q interactions with the peptide inhibitor of C1 (PIC).PIC binds to the collagen-like domain containing the intrinsically disordered region (IDR) highlighted in blue. 35G) Representative schema and images of mouse C1q (100 mg/mL) and total mouse brain RNA (200 mg/mL), PIC (100 mg/mL) alone, mouse C1q (100 mg/mL) pretreated with PIC (100 mg/mL), and mouse C1q (100 mg/mL) pretreated with PIC (100 mg/mL) with total mouse brain RNA (200 mg/mL) ±RNase.Scale bar is 15 mm (203 objective); bright field (BF).(H) Schematic of C1q-594 ±PIC treatment of live acute slices isolated from neuronal GFP-trap mice (Synapsin-Cre; RPL10a-eGFP).(I) Representative images of hippocampal CA3 from acute slices as described in (H) treated with C1q-594 for 1 h.Scale bar is 50 mm (203 objective); C1q-594 (purple) and GFP (neuronal ribosomes; green).(J) Representative images of hippocampal CA3 from acute slices as described in (H) treated with C1q-594+PIC for 1 h.Scale bar is 50 mm (203 objective); C1q-594 (purple) and GFP (neuronal ribosomes; green).(K) Quantification of the fluorescence intensity of C1q-594 signal from slices treated as described in (H).Individual data points represent biological replicates normalized to C1q-only treated condition; data are represented as the mean ± SEM.Unpaired t test.See also Figure S6.and subsequent plasticity regulation across aging, independent of the diverse extracellular functions of C1q in the brain, including synaptic elimination.More broadly, the potential effects of cell-extrinsic proteins influencing these mechanisms should be considered.
Spatiotemporal regulation of C1q-mediated neuronmicroglia interactions in health and disease Our data highlight spatiotemporal specificity of neuroimmune activities, demonstrating how shifting molecular interactions may contribute to unique, cell-specific functions across time.We found that C1q protein, when introduced exogenously either through ICV injection or bath application in live brain slices, integrates into RNase-sensitive neuronal RNP complexes in an endocytosis-dependent manner.Intriguingly, C1q internalization by T cells via endocytic mechanisms has been reported previously, which led to altered mitochondrial metabolism and intracellular functions. 33Although cell-surface glycoRNA may facilitate uptake of certain cell-penetrating peptides, 36 we did not observe this to be the case for C1q (Figures 6C-6E).Given that C1q can bind to a wide array of potential receptors known to be expressed by neurons that could facilitate internalization, such as CD93, 69 low-density lipoprotein receptor-related protein 1 (LRP1/CD91), 52 and neuronal pentraxins, 60,[70][71][72] exploring how C1q is internalized by neurons and whether receptor-mediated endocytosis underlies this process will be important to investigate.
Once internalized, the mechanisms by which C1q escapes the endosomal compartment and integrates into neuronal RNP complexes merit further investigation.Some proteins contain a ''protein transduction domain'' consisting of cationic and/or hydrophobic peptides that facilitate endocytosis-mediated internalization and subsequent escape. 73,74Viral proteins, including the HIV protein trans-activator of transcription (TAT), may contain these regions and are thought to undergo endosomal escape via hydrophobic residue insertion, leading to localized membrane destabilization. 73Intriguingly, the globular head domain of C1q is enriched in charged and hydrophobic peptides, 75 so whether these, or additional modifications, contribute to endosomal escape will be of interest to further examine.Moreover, BANF1, an age-specific C1q protein interactor, can interact with several viruses following their endosomal escape, preventing subsequent autophagosome formation and degradation. 76One possibility is that BANF1 interactions with C1q play a role in preventing intracellular degradation processes, enabling subsequent interactions with RNP complexes.Whether BANF1 also protects RNP complexes from autophagic degradation could be one way in which these structures are preserved within the cytoplasm.
8][79] Pathological changes in proteins and RNA may be driven by LLPS interactions, as many of these proteins are involved in stress granule formation, a critical cellular response important for minimizing RNA and protein damage under acute stress conditions.][82][83] Interestingly, C1q was uncovered in two genome-wide screens identifying genes regulating stress granule assembly in U2OS cells, 84,85 revealing an unexpected role of C1q in this process.One potential intracellular function of C1q, then, could be to prevent the spread of toxic proteins through sequestration in stress granule-like structures.Previous studies have shown that C1q binding can prevent cellular damage induced by several toxic proteins, including oligomeric amyloid-b 86 and the toxic form of Prion protein (PRPSC). 87BANF1 was also recently uncovered to play a protective role in preventing pathological tau seeding. 88,89Paradoxically, could the long-term consequence of these interactions lead to increased disruption in local protein homeostasis and ultimately impaired synaptic function?Understanding whether glial-and immune-derived proteins can interact with aggregation-prone proteins within neurons and impact function will be of critical importance.Our findings identify age-dependent intraneuronal functions and RNA-dependent interactions of C1q, elucidating potential roles of glial and immune-derived proteins in temporally and contextually regulated biological processes in health and disease.

Limitations of the study
Our data reveal unique interactions of C1q and uncover agespecific phenotypes in C1q-deficient animals.It is challenging to directly test whether C1q interactions with neuronal RNP complexes underlie these phenotypes, as there are several technical limitations that currently impact our ability to test this in vivo.Given that C1q consists of three unique peptides, modifying the IDR within one or more of the peptides could prevent C1q complex formation and secretion by microglia.Moreover, the age-specific nature of these phenotypes necessitates in vivo applications, yet manipulating microglia by traditional approaches, such as viral targeting, is not currently technically feasible.Further experiments aimed at identifying the mechanisms by which C1q is internalized by neurons could help to elucidate neuron-specific functions of C1q and could be performed using emerging in vivo CRISPR screening approaches.Moreover, such approaches would enable further exploration of how age-dependent C1q:RNP interactions impact behavioral and physiological readouts.Additionally, we employed the SUnSet assay to assess acute translation in vivo; however, puromycin exposure may generate off target effects such as the induction of cellular stress responses. 90An alternative approach utilizing non-canonical amino acid incorporation via the expression of a mutated methionine transferase, 91 for example, could further enable the investigation of C1q-mediated changes in cell-specific translation in vivo across extended time points.

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Beth Stevens (beth.stevens@childrens.harvard.edu).

Materials availability
This study did not generate any new unique reagents.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
All experimental procedures were approved by the Boston Children's Hospital institutional animal care and use committee (IACUC) in accordance with NIH guidelines for the humane treatment of animals.Animals were group-housed in Optimice cages and maintained in the temperature range and environmental conditions recommended by AAALAC.

C1q and puromycin immunoprecipitation (IP)
C1q-IP Crude synaptosome lysates were immediately used for C1q co-immunoprecipitation (co-IP), where 1.5mg of protein was used for each experiment.Anti-C1q (Rabbit monoclonal; Abcam #ab182451) was added to protein lysate at a concentration of 2mg/mg of protein.Protein: antibody mixture was then incubated overnight at 4C with rocking.Protein A/G magnetic beads (Pierce) were used for co-IP by resuspending beads via vortex for 2-3 seconds and removing 40mL per sample.Beads were washed in wash buffer (500mL; 150mM NaCl, 50mM Tris pH7.5, protease and phosphatase inhibitors) and separated using magnetic rack.Protein: antibody mixture was then added to washed beads and incubated for 30' rocking at room temperature.Following incubation, samples were washed three times with 500mL wash buffer and separated using magnetic rack.Following the final wash, for mass spectrometry the beads were resuspended in 10mL of sterile PBS and stored on ice.For Western blot, beads were resuspended in 20mL of 2X Laemmli buffer (Bio-Rad) containing 2-mercaptoethanol (Bio-Rad) and boiled for 5' at 95C.Samples were then centrifuged for 3' at 16100 RCF 4C and places on magnetic rack to remove beads.

Puromycin-IP
Brains isolated from 1-year old WT and C1qKO littermates injected 1-hour prior with puromycin (see below) were dissected out and diced in ice-cold lysis buffer (150mM NaCl, 50mM Tris pH7.5, 1% NP-40) containing 1X protease inhibitor cocktail (Sigma-Aldrich) and 1 mM phosphatase inhibitors sodium fluoride (Sigma-Aldrich) and sodium orthovanadate (Sigma-Aldrich).Diced brains were then homogenized using a mechanical glass homogenizer at $900 rpm.Brain homogenates were rocked for 20-30' at 4C to solubilize proteins and then centrifuged at 16100 RCF for 10' to yield the protein fraction.Brain lysates were used for puromycin or isotype control co-IP where 1mg of protein was used for each experiment.Anti-puromycin (Mouse monoclonal, clone 12D10; EMD Millipore #MABE343) or isotype control (Mouse IgG2a, ThermoFisher #14-4724-82) antibodies were added to protein lysate at a concentration of 2mg/mg of protein.Samples were processed as described above.

C1q immunoprecipitation sample labeling and Mass Spectrometry
After immunoprecipitation, beads were resuspended with 90mL of digestion buffer (2M urea, 50mM Tris HCL) with 2mg of sequencing grade trypsin per sample and incubated for 1H with shaking at 700 rpm at 37C.Supernatant was collected into a fresh tube and beads were washed 2x with 50mL of digestion buffer which was subsequently combined with the supernatant.The supernatants for each sample were then reduced with 2mL of 500mM dithiothreitol for 30' at room temperature and alkylated with 4mL of 500mM iodoacetamide for 45' in the dark.An overnight digestion was then performed by incubating samples with 2mg of sequencing grade trypsin overnight with shaking.Enzymatic digestion was quenched with 20mL of 10% formic acid and desalted on 10mg Oasis cartridges prior to Tandem Mass Tag (TMT, Thermo Scientific 90061) labeling.See Table S1 for TMT labeling schematic.In each TMT cassette, there were two replicates: C1q Ips from WT mice or C1q KO mice.Peptides from each condition were labeled with a unique TMT mass using 25% of the indicated reagents according to the manufacturer's instructions prior to combining all the samples.Orthogonal peptide fractionation was performed using basic reverse phase on a StAGE TIP (ref) to gain deeper coverage of eluted peptides prior to online HPLC separation.Briefly, 3 punches of Empore SDB-RPS material (47mm Phenomenex part # AH0-4048) was loaded into a 200mL pipette tip and washed with 5mM ammonium formate in 2% acetonitrile.Peptides were loaded onto the column and washed once with 5mM ammonium formate in 2% acetonitrile prior to eluting peptides with increasing concentrations of ACN (10, 15, 20.30.40, and 50% all in 5mM ammonium formate).Reconstituted peptides from six final fractions were then separated on an online nanoflow EASY nLC1000 UHPLC system (Thermo Scientific) and analyzed on a benchtop Q Exactive Plus mass spectrometer (Thermo Fisher Scientific).The peptide samples were injected onto a capillary column (Picofrit with 10mm tip opening/ 75mm diameter) in house with 35cm C18 silica material (1.9mm ReproSil-Pur C18-AQ medium, Dr. Maisch GmbH) and heated to 50C in a column heater sleeve (Phoenix-ST) to reduce backpressure during UHPLC separation.Peptides were separated at a flow rate of 200nL/min with a linear gradient to 20% solvent B for 28 minutes followed by a 16 0 linear gradient to 90% solvent B with a hold for 6 minutes with a drop to 50% solvent B for a duration of 10'.The Q Exactive instrument was operated in data-dependent mode acquiring higher-energy collisional dissociation (HCD) tandem mass spectrometry (MS/MS) scans (R=17,500) after each MS1 scan (R=70,000) on the 12 topmost abundant ions using an MS1 ion target of 3e 6 ions and an MS2 target of 5e 4 ions.The maximum ion time utilized for the MS/MS scans was 120 milliseconds; the HCD-normalized collision energy was set to 31; the dynamic exclusion time was set to 20s; and the peptide match and isotype exclusion functions were enabled.
These mass spectra were processed using the Spectrum Mill software package (v7; Agilent Technologies), which includes modules developed for TMT-based quantification.For peptide identification MS/MS spectra were searched against the mouse Uniprot database to which a set of common laboratory contaminant proteins was appended.Search parameters included ESI-QEXACTIVE-HCD scoring parameters, trypsin enzyme specificity with a maximum of 2 missed cleavages, 40% minimum matched peak intensity, +/-20 ppm precursor mass tolerance, and carbamidomethylation of cysteines and TMT labeling of lysines and peptide N termini as fixed modifications.Allowed variable modifications were of oxidation of methionine, N-terminal acetylation, pyroglutamic acid (N-termQ), deamidated (N), pyro carbamidomethyl Cys (N-termC), with a precursor MH+ shift range of -18-64 Da.Identities interpreted for individual spectra were automatically designated as valid by optimizing score and delta rank 1-rank 2 score thresholds separately for each precursor charge state in each LC-MS/MS while allowing a maximum target decoy based FDR of 1.0% at the spectrum level.

Puromycin immunoprecipitation sample labeling and Mass Spectrometry
On-bead trypsin digestion Beads samples were first washed three times with 100 mL of PBS.The beads were then resuspended in 100 mL of 50mM TEAB (triethylammonium bicarbonate) and incubated at 95 C for 5 minutes.After samples returned to room temperature, beads were digested with trypsin for 3 hours at 37 C in a thermomixer shaking at 800rpm.Samples were mounted on a magnetic rack to separate the magnetic beads from solution, which was transferred to a new centrifuge tube and dried in SpeedVac.The dried residue was reconstituted with 80 mL 0.1% TFA and then cleaned up with Pierce C18 spin tips (ThermoScientific, Waltham, MA) prior to LC-MS/MS analysis.

LC-MS/MS Analysis
Samples were analyzed by Q Exactive HF-X High Resolution Orbitrap (Thermo Fisher, Waltham, MA) coupled with Ultimate 3000 nanoLC (Thermo Fischer, Waltham, MA) at Harvard Center for Mass Spectrometry.Peptides were first trapped on a trapping cartridge (300mm x 5mm PepMapÔ Neo C18 Trap Cartridge, Thermo scientific) prior to separation on an analytical column (mPAC, C18 pillar surface, 50 cm bed, Thermo scientific).Peptides were separated using a 125-min gradient (from 1 -35% acetonitrile with 0.1% formic acid) with a flow rate of 350 nL$ minÀ1.The mass spectrometer operated in data-dependent mode for all analyses.Electrospray positive ionization was enabled with a voltage at 2.1 kV.A full scan ranging from 350 to 1400 m/z was performed with a mass resolution of 123104 and AGC target set to 1e6.The top three most intensive precursor ions from each scan were used for MS2 fragmentation (normalized collision energy 30) at a mass resolution of 3.0x104 and AGC of 1e5.The dynamic exclusion was set at 60 s with a precursor isolation window of 1.2 m/z.Data Processing Raw data was submitted for analysis in Proteome Discoverer 3.0 software (Thermo Scientific).The MS/MS Data was searched against the UniProt reviewed Homo sapiens database along with known contaminants such as human keratins and common lab contaminants.Quantitative analysis between samples was performed by label-free quantitation (LFQ).Sequest HT searches were performed using the following guidelines: a 10 ppm MS tolerance and 0.02 Da MS/MS tolerance; Trypsin digestion with up to two missed cleavages; carbamidomethylation (57.021Da) on cysteine were set as static modification; oxidation (+15.995Da) of methionine set as variable modification; minimum required peptide length set to R 6 amino acids.At least one unique peptide per protein group is required for identifying proteins.All MS2 spectra assignment FDR of 1% on both protein and peptide level was achieved by applying the target-decoy database search by Percolator.

Global proteomics analysis of WT and C1qKO 1-year old brain tissue Sample lysis and digestion
Lysis buffer (0.5 M HEPES, 2% SDS, and protease and phosphatase inhibitor cocktail (ThermoFisher, Waltham, MA) and grinding beads were added to each of the six tissue samples and then homogenized with a Tissuelyser LT (Qiagen, Germantown, Maryland).Tissue homogenates were centrifuged at 16,000 g for 20 minutes at 4 C.The supernatant was transferred into a new vial and protein concentration was measured using a NanoDrop Spectrophotometer (ThermoScientific, Waltham, MA).Each sample with 100 mg proteins were reduced with 200 mM TCEP at 55 C for 1 hour and then alkylated with 375 mM iodoacetamide at room temperature for 30 minutes in the dark.Proteins were precipitated using methanol/chloroform/water precipitation method, and then digested with trypsin overnight at 37 C. TMT 6-plex labeling TMTsixplex labeling for digested samples were performed according to manufacturer's instructions (ThermoFisher, Waltham, MA).Briefly, TMT labeling reagents were dissolved with 41mL of anhydrous acetonitrile and then added to the samples in equal volume.
After one hour incubation at room temperature, the reaction was quenched with 8mL of 5% hydroxylamine.Equal amounts of peptides from each sample were combined and dried in a SpeedVac.High pH Reversed-Phase Peptide Fractionation TMT labeled peptides were then fractionated using the Pierce High pH Reversed-Phase Peptide Fractionation Kit (ThermoFisher, Waltham, MA).20 fractions were eluted with a step gradient from 5% ACN in 0.1% TEA -50% ACN in 0.1% TEA using discrete elution solutions.Fractions were evaporated to dryness in a SpeedVac and then resuspended in 0.1% formic acid for LC-MS/MS analysis.

LC-MS/MS Analysis
Samples were analyzed by Q Exactive HF-X High Resolution Orbitrap (Thermo Fisher, Waltham, MA) coupled with Ultimate 3000 nanoLC (Thermo Fischer, Waltham, MA) at the Harvard Center for Mass Spectrometry.Peptides were first trapped on a trapping cartridge (300mm x 5mm PepMapÔ Neo C18 Trap Cartridge, Thermo scientific) prior to separation on an analytical column (mPAC, C18 pillar surface, 50 cm bed, Thermo scientific).The LC gradient was as follows: 2 -25% in mobile phase B (0.1% formic acid in acetonitrile) over 60 minutes, 25% -45% mobile phase B over 15 minutes, 45% -95% mobile phase B over 5 minutes, and finally 95% mobile phase B was held for 10 minutes with the flow rate at 300 nL$ minÀ1.The mass spectrometer operated in data-dependent mode for all analyses.Electrospray positive ionization was enabled with a voltage at 2.1 kV.A full scan ranging from 400 to 1600 m/z was performed with a mass resolution of 123104 and AGC target set to 1x106.The top three most intensive precursor ions from each scan were used for MS2 fragmentation (normalized collision energy 32) at a mass resolution of 3.0x104 and AGC of 1e5.The dynamic exclusion was set at 50 s with a precursor isolation window of 1.2 m/z.Data Processing Raw data was submitted for analysis in Proteome Discoverer 3.0 software (Thermo Scientific).The MS/MS Data was searched against the UniProt reviewed Homo sapiens database along with known contaminants such as human keratins and common lab contaminants.Sequest HT searches were performed using the following guidelines: a 10 ppm MS tolerance and 0.02 Da MS/MS tolerance; Trypsin digestion with up to two missed cleavages; carbamidomethylation (57.021Da) on cysteine, TMT 6-plex tags on peptide N-termini and lysine residue (+229.163Da) were set as static modification; oxidation (+15.995Da) of methionine set as variable modification; minimum required peptide length set to R 6 amino acids.At least one unique peptide per protein group is required for identifying proteins.All MS2 spectra assignment FDR of 1% on both protein and peptide level was achieved by applying the target-decoy database search by Percolator.

Polysome fractionation
Polysome buffer For all gradient preparation and sample resuspension the buffer that was used consisted of 20mM Tris HCl pH 7.4, 150mM NaCl, and 2.5mM MgCl2.Buffer was prepped beforehand and stored at room temperature.

Polysome isolation and fractionation
Polysome fractionation protocol is modified from Fatimy et al. 17 Centrifuges and tools were all prechilled to 4C or on ice prior to use.Isolated brains (two brains pooled per sample) were minced in 10ml of extraction buffer using a scalpel, on ice, then Dounce homogenized sequentially using loose and tight pestles.Homogenate was centrifuged for 15' at 9000g at 4C to remove debris.Supernatant was then collected, and 1% NP-40 was added to concentrate polyribosomes.In polypropylene centrifuge tubes (5ml, Beckman Coulter), 1ml of 60% sucrose solution (prepared fresh in polysome buffer) was added to the bottom of the tube, and 4ml of supernatant was layered on top (two tubes per brain).Tubes were then balanced and ultracentrifuged using a Beckman SW55 Ti Swinging-Bucket rotor at 34000 RPM for 2h at 4C.Following centrifugation, supernatant was removed from each tube and pellets were resuspended in 500ml of polysome buffer supplemented with protease and phosphatase inhibitors.At this step, each biological replicate was split into ±RNase (1:1000; Thermo Fisher) treatment.Resuspended pellets (±RNase, 500ml per sample) were layered on top of the thawed linear sucrose density gradients.Tubes were then balanced and ultracentrifuged using a Beckman SW55 Ti Swinging-Bucket rotor at 34000 RPM for 2h at 4C.Following centrifugation, twelve fractions from the gradient were collected with increasing density into Eppendorf tubes ($416ml per fraction), and the pellet was resuspended in polysome buffer supplemented with protease and phosphatase inhibitors.Each fraction was precipitated overnight in two volumes of 100% molecular grade ethanol (Sigma-Aldrich) at -20C.The following day, fractions were centrifuged at 12000 RPM for 20' at 4C, supernatant was removed, and pellets were dried.Pellets were then resuspended in 2X Laemmli buffer (Bio-Rad) containing 2-mercaptoethanol (Bio-Rad) and boiled for 5' at 95C.

Tissue preparation
Mice were anesthetized with Avertin (240mg/kg, i.p.) and transcardially perfused with PBS.Brains were harvested and cut sagittally down the midline, with one hemisphere prepared for immunostaining and one hemisphere prepared for biochemical analyses.For immunostaining, brains were post-fixed in 4% PFA (Electron Microscopy Sciences) for 1h on ice (or 24h at 4C for Figure S3A), then washed and transferred to 30% sucrose solution for 24-48h at 4C. Tissue was then embedded in a 2:1 mixture of 30% sucrose: OCT (Sakura Finetek) and stored at -80C.For biochemical analyses, brains were immediately fresh frozen in liquid nitrogen and stored at -80C.

Proximity Ligation Assay (PLA) Probe Conjugation
Primary antibodies were conjugated to oligonucleotides using the Duolinkâ PLA Probemaker Conjugation kit (Sigma-Aldrich).20ml of anti-C1q (rabbit monoclonal, Abcam), 20ml of anti-BANF1 (rabbit monoclonal, Abcam), and 20ml of anti-RPL10a (rabbit monoclonal, Abcam) were mixed with 2 ml of Conjugation Buffer.These antibody solutions were then transferred to either PLUS or MINUS oligonucleotides and incubated overnight at room temperature.After addition of 2 ml of Stop Reagent, 24 ml of Storage Solution was added to stabilize the conjugated antibodies.Conjugated primary antibodies were stored at 4C. Tissue Preparation Embedded tissue was sectioned onto X-tra slides (Leica Microsystems) at 15mm using a cryostat, dried, and then washed with TBS.A brief antigen retrieval step was then performed by treating sections with 1% SDS in TBS for 2'.After three washes with TBS, sections were permeabilized for 10' using 0.1% saponin in TBS prior to beginning the Proximity Ligation Assay.

Proximity Ligation Assay
The assay was performed based on manufacturer instructions with slight modifications.Briefly, sections were blocked for 1h at 37C using DuoLinkâ Blocking solution.Primary antibodies were diluted 1:500 in PLA Probemaker Diluent and incubated overnight at 4C.After two washes with Wash Buffer A, sections were incubated with 1X ligase in Duolinkâ Ligation buffer for 30 0 at 37C.Sections were washed twice with Wash Buffer A and then incubated with 1X polymerase in Duolinkâ Amplification buffer for 100'.Following amplification, sections were washed twice with Wash Buffer B and then incubated with TrueBlack Lipofuscin Autofluorescence Quencher (Biotium) for 30 seconds.After three more washes with Wash Buffer B, 0.01X Wash Buffer B was added to sections for 1' and sections were coverslipped using Duolinkâ In Situ Mounting Media with DAPI.Slides were sealed with polish and stored at -20C.To validate IHC, conjugated primary antibodies were detected using AlexaFluor-conjugated secondary antibody (1:200; Invitrogen/Thermo Fisher) in 1%BSA/0.1% saponin (TBS) for 1h at room temperature.Image z-stacks (0.32mm step size, 0.13mm pixel size, 1.02ms pixel dwell time) were acquired on a LSM880 confocal microscope (Zeiss) at 63X (NA 1.4) magnification.

Liquid-liquid phase separation (LLPS) assay
All LLPS assays were performed in buffer consisting of 50mM HEPES, 150mM NaCl, pH 7.5 (LLPS buffer).Specific concentrations of BSA (Sigma-Aldrich), human C1q protein (Quidel), mouse C1q protein (conjugated and non-conjugated; Creative Biolabs), human C1q globular heads (provided by G. Andersen), human C3 protein (Quidel), human C4 protein (Quidel), human C1q a, b, and c peptides (Creative Biolabs), peptide inhibitor of C1 (PIC; Creative Biolabs), human total brain RNA (Takara), human polyA-enriched brain RNA (Takara), human total liver RNA (Takara), and mouse total brain RNA (Takara) are listed in appropriate figure legends.SYTO RNASelect (Molecular Probes, Thermo Fisher) was diluted in LLPS buffer and used at a concentration of 500nM where listed.Proteins and RNA were added in varying concentrations in LLPS buffer to a total volume of 10ml, gently mixed by pipetting up-and-down, and immediately plated onto glass bottom dishes (MatTek) for imaging.For PIC treatment, PIC was first resuspended in sterile water at a concentration of 1mg/ml following manufacturer's instructions.Equal concentrations of C1q and PIC solution were pre-mixed together prior to addition of LLPS buffer and RNA.For RNase A treatment, Rnase A (1:1000, Thermo Fisher) was added in at the same time as C1q protein prior to imaging.All experiments were performed independently using three different lots of C1q protein and RNA.LLPS was imaged using and brightfield and widefield (GFP) live-imaging at 20X (NA 0.8) magnification (0.56mm step size, 0.42mm pixel size, 1.02ms pixel dwell time) on a LSM880 confocal microscope (Zeiss).Timelapse imaging Human C1q protein (200mg/ml, Quidel) and total human brain RNA (500mg/ml, Takara) were combined, added to a glass bottom dish, and immediately imaged in brightfield every 5 seconds for 30' at room temperature.Images were taken on 20X objective of a Zeiss LSM 880 microscope.For Rnase A treatment, human C1q protein (200mg/ml, Quidel) and total human brain RNA (500mg/ml, Takara) were combined with SYTO RNASelect (500nM, Molecular Probes), added to a glass bottom dish, and immediately imaged in brightfield and widefield (GFP) every 5 seconds for 10' at room temperature.Rnase A (1:1000; Thermo Fisher) was added in after initial image was acquired and remained through subsequent imaging.

Concentration Matrix Assay
Each combination of varying concentrations of total human brain RNA (Takara) and human C1q (Quidel) (5mg/ml, 10mg/ml, 50 mg/ml, and 500 mg/ml) were imaged as described above and the presence of droplets was determined.Three independent experiments were preformed using three different lots of C1q protein and RNA.Collagenase treatment Human C1q protein (Quidel) was incubated in Type 3 collagenase (Worthington) (15:1 C1q to collagenase ratio 30 ) in LLPS buffer overnight at 37 C. Control C1q protein underwent the same overnight 37 C incubation in LLPS buffer without collagenase.LLPS was assessed the following day as described above.

Tamoxifen Treatment
Tamoxifen was mixed in corn oil (10mg/ml; Sigma-Aldrich) and vigorously shaken overnight at 37C.Resuspended stocks were stored in the dark at 4C for one week.Cx3Cr1-Cre ER mice received daily intraperitoneal doses (10mg/kg) for five consecutive days beginning at 6 weeks of age and were evaluated two weeks later at 2-months-old for either immunostaining or contextual fear conditioning (see below).No changes in health or behavior were observed in tamoxifen treated animals compared to non-tamoxifen treated animals.

Intracerebroventricular (ICV) injections
Maleimide-based C1q protein conjugation Mouse C1q protein (Creative Biolabs) was conjugated to Alexa-594 using C5 maleimide conjugation kit (Invitrogen) with modifications from manufacturer's instructions.Alexa 594-maleimide was first diluted to 1mM in DMSO and further diluted 1:10 in PBS.Diluted Alexa-594 maleimide was then added to C1q protein at a concentration of 1:10 and incubated for 90 minutes at room temperature with gentle shaking and protected from light.Following incubation, unbound dye was removed using a 10kDa molecular weight cutoff filter (Microcon-10kDa Centrifugal Filter Unit with Ultracel-10 membrane) following manufacturer's instructions.Amine-based C1q protein conjugation Mouse C1q protein (Creative Biolabs) was conjugated to Alexa-594 using Alexa Fluor-594 Lightning Link Conjugation kit (Abcam) following manufacturer's instructions.Injection materials Injections were performed using mouse C1q protein (1mg/ml; Creative Biolabs), unconjugated or conjugated to Alexa-594 as described above.For in vivo endocytosis inhibition, individual components of the Dynamin Inhibitors Toolbox (Abcam), which consisted of the cell-permeable dynamin inhibitors MiTMAB, OcTMAB, and Dynole-34-2, and their respective negative controls Pro-Myristic Acid and Dynole-31-2, were first diluted to 100mM stocks in either sterile water (MiTMAB) or DMSO (all others).Working stocks of 200mM for each compound were then prepared in 0.9% sterile saline, creating a cocktail of either endocytosis inhibitors or negative controls.For in vivo dextran endocytosis experiments, 10kD molecular weight dextran conjugated to Alexa-488 (Thermo Scientific) was diluted at a concentration of 2.5mg/ml in 0.9% sterile saline.For in vivo protein translation analysis, puromycin (5mg/ml; puromycin dihydrochloride (Fisher Scientific)) was diluted in 0.9% sterile saline.Procedure For all procedures, mice were anesthetized using isoflurane (2-4% exposure followed by 1.5% throughout the procedure).A volume of 2ml was injected into the left lateral ventricle (-0.4 mm anteroposterior, 1 mm mediolateral and -2.5 mm dorsoventral) via a pulled glass capillary at a rate of 0.5 ml/min utilizing a Nanoject III (Drummond Scientific).For endocytosis inhibition experiments, animals were first injected with a volume of 2ml into the left lateral ventricle with either the endocytosis inhibitor cocktail or negative control cocktail as described above, and kept under isoflurane anesthesia for 30 min.Following treatment and incubation, animals then received a second ICV injection with either C1q protein or dextran as described.Upon completion of injections (< 15 min), mice were allowed to wake up from anesthesia and returned to their home cage.Animals were sacrificed 1h later for immunohistological and biochemical analyses.No gross toxic effects, overt signs of distress or pain, or any other behavioral changes were observed due to either injected materials or the ICV injection procedure.

Treatments
Maleimide-conjugated C1q-594 (1mg/ml) was freshly prepared as described above and added to slices in modified slicing solution at a concentration of 1:1000 for 1 hour at room temperature.For RNase experiments, RNase A (1:1000) was added in either 10 minutes prior to C1q-594 (pretreatment) or for 10 minutes following 1-hour C1q-594.In both conditions RNase remained for the duration of the experiment following application.For peptide inhibitor of C1 (PIC) experiments, a 1:1 mixture of C1q-594 (1mg/ml) with PIC (1mg/ml) was added to slices in aCSF at a final dilution of 1:1000 for each and incubated for 1 hour at room temperature.For endocytosis inhibition experiments, individual components of the Dynamin Inhibitors Toolbox (Abcam) and their respective negative controls (described above) were added to slices in aCSF at a final concentration of 20mM 15 minutes prior to the addition of C1q-594, and then incubated for 1 hour further.Imaging Acute slices were treated as described above and imaged in fresh modified slicing solution in glass bottom dishes (Mattek).Image z-stacks (0.56mm step size, 0.42mm pixel size, 1.02ms pixel dwell time) were acquired in CA3 with a Zeiss LSM880 microscope and a 20x objective (NA 0.8).Experiments were performed within 4 hours of sacrificing mice.

Contextual fear conditioning behavior
Contextual fear conditioning is a well-established associated learning task that was conducted following the established behavioral paradigm. 99All equipment (Stoelting ANY-Maze), facilities, and training were provided by the Boston Children's Hospital Animal Behavior and Physiology Core.Briefly, for fear memory acquisition (Training Day 1) animals were initially trained to associate a neutral stimulus (tone; 30s, 90db, 2.5kHz) and context (square arena with metal grating) with an adverse stimulus (mild foot shock; 0.5mA, 2s).In the days following training, animals were re-exposed to the neutral stimulus (tone) and a new context (circular arena with no grating).Over the course of four days, animals undergo fear memory extinction, whereby they learn to no longer associate the neutral stimulus (tone) with an adverse stimulus (mild foot shock).Fear exhibited as freezing behavior (% freezing) was monitored and analyzed by an automated video-freeze software (Ethovision XT, Noldus).All behavioral tests were performed on sex-matched, paired littermates concurrently at consistent times of the day across the study blind to genotype.

Proteomic analyses
C1q co-IP datasets were analyzed using Genoppi. 10Non-mouse proteins and proteins with fewer than two unique peptides were removed from further analyses.For C1q-IP experiments, data represent two technical replicates performed on crude synaptosomes isolated from WT and C1qKO littermates (n=3-4 animals pooled per genotype) from postnatal day 5 (P5), 2-3-month-old (2-mo), and 1-year-old (1yr).Significant 'hits' were identified with an FDR cutoff of p %0.1 of comparisons between WT and C1qKO samples.For puromycin-IP, data represent three biological replicates performed on total brain lysate isolated from 1-year-old WT and C1qKO littermates following ICV puromycin injection.Significant 'hits' were identified with a cutoff of p %0.05 of comparisons with isotype control samples.Global brain proteomics data represent three biological replicates performed on brain tissue isolated from 1-year-old WT and C1qKO littermates.Significant 'hits' were identified with an FDR cutoff of p %0.1 of comparisons between WT and C1qKO samples.Protein interactions networks were visualized using STRING network analyses. 100Venn diagram of proteomic data was generated using BioVenn. 95GO enrichment charts were generated using ShinyGO 11 (v0.76.3).Significant 'hits' identified in Genoppi were compared to total proteomic data from each specific dataset.

Immunoblot quantification
All immunoblots were quantified using Fiji (ImageJ, NIH) software. 94Individual antibody blots were normalized to total protein, and further normalized to internal control samples.For polysome quantification, fractions were normalized to unfractionated protein lysate, and data represent the mean of three biological replicates paired ±RNase A. For anti-C1q and anti-puromycin blots, C1qKO animals were normalized to WT sex-matched littermates.

Image analysis
Acquisition parameters (i.e., laser power, gain, and offset) were kept constant among different conditions in each single experiment.For confocal images, co-localization analyses were performed in Imaris software (v9.9.1; Oxford Instruments).Confocal z-stack images (0.42mm step size, 0.21mm pixel size, 1.02ms pixel dwell time) were acquired from four fields of view per section, consisting of hippocampal regions CA1, CA3, and the dentate gyrus, and Layer V of the motor cortex (40X (NA 1.3); LSM880 confocal microscope (Zeiss)).Two sections per animal were imaged.Individual channel images were processed in 3D surface rendering software Imaris and volumes were calculated, masked, and normalized to determine the percentage of co-localization.For co-localization analysis following SIM acquisition, images were processed in Fiji using the plugin ComDet (v0.5.4; https://github.com/UU-cellbiology/ComDet).Particles in each channel were identified with an approximate particle size of 4mm and an intensity threshold of 200, and colocalization was determined based on a maximum pixel distance of 4mm.For quantification of dextran labeling following endocytosis inhibition, C1q-594 labeling of live acute slices, or puromycin immunostaining, images (as described above) were processed in CellProfiler 96 (v4.2.1) using the 'measure image intensity' module.Neuronal labeling (Neuro-Chrom) was used as a mask to quantify dextran or puromycin labeling intensity.Raw values were then normalized to the mean of respective controls.Representative images and immunoblots were adjusted in ImageJ to optimize visual clarity, and images across the same experiment or presented in the same figure panels were adjusted identically.All experiments were performed and analyzed blind to treatment group or genotype.

Statistical analyses
Data points represent biological replicates, consisting of the mean of an individual animal (as detailed in figure legends) ±SEM.All statistical analyses were conducted using PRISM (v9; GraphPad Software).Specific statistical analyses used are described in figure legends, and p-values are listed in graphs.(legend continued on next page)      (B) Representative hippocampal images of 2-month-old C1qKO mice following ICV injection with mouse C1q-594 maleimide (1 mg/mL; purple) 1 h prior immunostained with a-C1q (green) and DAPI (cyan) ±RNase treatment (1:1,000).Scale bar is 100 mm (103 magnification).(C) Schematic of experimental design for in vitro mouse C1q protein labeled with Alexa-594 via amine conjugation (C1q-594 amine) LLPS droplet assay and representative images of mouse C1q-594 amine (100 mg/mL) and total mouse brain RNA (200 mg/mL).Scale bar is 15 mm (203 magnification; bright field [BF] and C1q-594 amine [purple]).(D) Representative hippocampal images of 2-month-old C1qKO mice following ICV injection with mouse C1q-594 amine (1 mg/mL; purple) 1 h prior immunostained with a-C1q (green) and DAPI (cyan) ±RNase treatment (1:1,000).Scale bar is 100 mm (103 magnification).(E) Schematic of experiments performed in live acute brain slices with fluorescently labeled C1q.Mouse C1q protein was conjugated to Alexa 594 (maleimide), and bath was applied (1:1,000 of 1 mg/mL) to acute slices from 2-month-old WT animals for 1 h.Slices were pretreated with a cocktail of endocytosis inhibitors (cell-permeable dynamin inhibitors) or respective negative controls (20 mM) for 15 min prior to the addition of C1q-594.(F) Representative images of hippocampal CA3 from live acute slices as described in (E).Scale bar is 50 mm (203 objective); C1q-594 (purple) and Hoechst (nuclei; blue).(G) Quantification of the fluorescence intensity of C1q-594 signal from slices treated as described in (E).Individual data points represent biological replicates normalized to negative control-treated slices; data are represented as the mean ± SEM.Unpaired t test.

Figure 4 .
Figure 4. RNA is necessary for C1q interactions with neuronal RNP complexes in vivo (A) Blots of total brain lysate and subsequent polysome fractions first represented in Figure 1J.(B)Blots of total brain lysate and subsequent polysome fractions from (A) treated with RNase (1:1,000) prior to fractionation.Blots were probed with a-RPL10, a-RPL5, a-FMRP, a-FUS, and a-C1q antibodies.(C) Quantification of blots of RPL10, RPL5, FMRP, FUS, and C1q normalized to total protein from total brain lysate and polysome fractions as shown in(A) and first represented in Figure1K.Data represent the mean from 3 biological replicates ± SEM. (D) Quantification of blots of RPL10, RPL5, FMRP, FUS, and C1q normalized to total protein from total brain lysate and polysome fractions following RNase treatment as shown in (B).Data represent the mean from 3 biological replicates ± SEM. (E) Representative images of hippocampal immunostaining with a-C1q (purple) and Milli-Mark (neurons; green) in tissue treated ±RNase prior to immunostaining.Scale bar is 500 mm (203 objective).(F) Representative images of CA1 hippocampal pyramidal cell layer from postnatal day (P) 5, P14, P30, 1-year, and 2-year-old WT animals immunostained with a-C1q (purple) ±RNase.Scale bar is 15 mm (633 objective).(G) Quantification of the percentage of neuronal staining (Milli-Mark) that co-localizes with a-C1q immunostaining across ages in (F).Individual data points represent biological replicates; data are represented as the mean ± SEM.Multiple paired t tests.See also FigureS4.

Figure 5 .
Figure 5. Exogenous C1q protein integration into neuronal RNP complexes in vivo is dependent on endocytosis (A) Schematic of experimental design for mouse C1q protein (2 mL, 1 mg/mL) ICV injection into 2-month-old Synapsin-Cre; RPL10a-eGFP; C1q À/À (neuronal GFPtrap C1qKO).(B) Representative hippocampal images of 2-month-old neuronal GFP-trap C1qKO mice following ICV injection with mouse C1q protein immunostained with a-C1q (purple), GFP (neuronal ribosomes), and DAPI (cyan).Scale bar is 500 mm (203 objective).(C) Representative hippocampal CA3 images as described in (B) ±RNase treatment prior to immunostaining.Scale bar is 15 mm (633 objective).(D) Quantification of the percentage of RPL10a-eGFP (neuronal ribosomes) that co-localizes with a-C1q immunostaining ±RNase treatment.Individual data points represent biological replicates; data are represented as the mean ± SEM.Paired t test.(E) Schematic of experimental design for in vivo endocytosis inhibition assay.A cocktail of endocytosis inhibitors (cell-permeable dynamin inhibitors) or respective negative controls were ICV injected (2 mL of 200 mM stocks) into 2-month-old C1qKO animals.30 min following injection, mouse C1q protein (2 mL, 1 mg/mL) ICV injection was performed.(F) Representative hippocampal images of 2-month-old C1qKO mice following ICV injection with negative control cocktail followed by mouse C1q protein, immunostained with a-C1q (purple) and Milli-Mark (neurons; green).Scale bar is 500 mm (203 objective).(G) Representative hippocampal images of 2-month-old C1qKO mice following ICV injection with endocytosis inhibitor cocktail followed by mouse C1q protein, immunostained with a-C1q (purple) and Milli-Mark (neurons; green).Scale bar is 500 mm (203 objective).(H) Representative hippocampal CA3 images as described in (F) ±RNase treatment prior to immunostaining.Scale bar is 15 mm (633 objective).(I) Representative hippocampal CA3 images as described in (G) ±RNase treatment prior to immunostaining.Scale bar is 15 mm (633 objective).(J) Quantification of the percentage of neuronal staining (Milli-Mark) that co-localizes with a-C1q immunostaining from animals described in (H) and (I).Individual data points represent biological replicates; data are represented as the mean ± SEM.One-way ANOVA with Sı ´da ´k's multiple comparisons test.See also FigureS5.

Figure 7 .
Figure 7. Macrophage-derived C1q impacts protein translation and fear extinction learning in an age-specific manner(A) Representative images of P5 WT and C1qKO hippocampal immunostaining with a-puromycin (purple) following ICV injection with puromycin 1 h prior.Scale bar is 500 mm (203 objective).(B) Quantification of the fluorescence intensity of a-puromycin immunostaining in the hippocampi of P5 WT (WT or C1q +/À ) and C1qKO sex-matched littermates.Individual data points represent biological replicates normalized to WT-paired littermates; data are represented as the mean ± SEM.Unpaired t test.(C) Blot of total brain lysate isolated from P5 WT and C1qKO sex-matched littermates as described in (A) probed with a-puromycin antibody.Total protein is shown for reference.(D) Quantification of puromycin normalized to total protein from lysates represented in (C).Individual data points represent biological replicates; data are represented as the mean ± SEM.Unpaired t test.(E) Representative images of 2-to 3-month-old WT and C1qKO as described in(A).(F) Quantification of a-puromycin immunostaining of 2-to 3-month-old WT and C1qKO as described in (B).(G) Blot of total brain lysate isolated from 2-to 3-month-old WT and C1qKO as described in (C).(H) Quantification of puromycin normalized to total protein from lysates represented in (G).Individual data points represent biological replicates; data are represented as the mean ± SEM.Unpaired t test.(I) Representative images of 1-year-old WT and C1qKO as described in (A).(J) Quantification of a-puromycin immunostaining of 1-year-old WT and C1qKO as described in (B).(K) Blot of total brain lysate isolated from 1-year-old WT and C1qKO as described in (C).(L) Quantification of puromycin normalized to total protein from lysates represented in (K).Individual data points represent biological replicates; data are represented as the mean ± SEM.Unpaired t test.(M) Volcano plot (generated using Genoppi 10 ) representing the fold change (log 2 FC) and adjusted p value (Àlog 10 ) of global proteomics data comparing 1-year-old WT to C1qKO brain tissue.C1qa, b, and c peptides are highlighted in red.Values in green represent significant peptides (FDR < 0.01).Three biological replicates per genotype/treatment were used for all proteomic datasets.(N) STRING network analysis and chart representing the top GO-enriched molecular components identified in the 1-year-old global proteomic dataset from (M) of proteins enriched in WT brain tissue.Chart was generated using ShinyGO software 11 comparing significant protein hits to total proteins uncovered in the dataset.(O) STRING network analysis and chart representing the top GO-enriched molecular components identified in the 1-year-old global proteomic dataset from (M) of proteins enriched in C1qKO brain tissue.Chart was generated using ShinyGO software 11 comparing significant protein hits to total proteins uncovered in the dataset.(P) Representative heatmaps of animal movement on days 2 and 4 following contextual fear conditioning.Heatmaps generated in EthoVision software over the course of 3-min novel context/tone exposure.(Q) Quantification of the percentage of freezing behavior animals exhibited following contextual fear conditioning over the course of memory acquisition (day 1: training), retrieval (day 2: same context/no tone and novel context/tone), and extinction (days 3-4: novel context/tone).Data are mean of n = 10 (WT) and 8 (cC1qKO) animals per group.Two-way ANOVA: for WT vs. cC1qKO p = 0.0121 for the combination of genotype 3 time as a significant source of variation and p % 0.001 for time as a significant source of variation; with Sı ´da ´k's multiple comparisons test.See also FigureS7.

Figure S5 .
Figure S5.Characterization of in vivo endocytosis inhibition, related to Figure 5

( A )
Schematic of experimental design for in vivo dextran endocytosis inhibition assay.A cocktail of endocytosis inhibitors (cell-permeable dynamin inhibitors) or respective negative controls were injected (2 mL, 200 mM) into 2-month-old WT mice lateral ventricles.30 min following injection, dextran (10 kDa Alexa-488; 2 mL, 2.5 mg/mL) ICV injection was performed.(B) Representative hippocampal CA3 images of 2-month-old WT mice following ICV injection with negative control cocktail followed by dextran (purple) and immunostained with DAPI (cyan).Scale bar is 15 mm (633 magnification).(C) Representative hippocampal CA3 images of 2-month-old WT mice following ICV injection with endocytosis inhibitor cocktail followed by dextran (purple) and immunostained with DAPI (cyan).Scale bar is 15 mm (633 magnification).(D) Quantification of the fluorescence intensity of dextran in the hippocampi of animals treated as described in(A).Individual data points represent biological replicates normalized to negative control injected animals; data are represented as the mean ± SEM.Unpaired t test.

Figure S6 .
Figure S6.Comparison of maleimide-and amine-based fluorophore conjugation on C1q LLPS interactions, related to Figure 6

TABLE
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
nyyvnb9nh9.1.All remaining data reported in this manuscript will be shared by the lead contact upon request.d This paper does not report original code.d https://imagej.net/software/fiji/