Microgel and star gum in Alzheimer's disease coordinate the clearance of synapses via the complement C1q-dependent pathway

Chronic neuroinflammation, manifested by glial abnormalities, elevated pro-inflammatory cytokine levels, and loss of synapses, is one of the main pathological features of
AD.
One pathway through abnormal overactivation in AD mouse models and in the brains of AD patients leading to neuronal damage and synaptic loss is the classical complement pathway (CCP).

Patients with AD have abnormally elevated
CCP factors in the brain and cerebrospinal fluid (CSF).
Human genetic association studies also support the involvement of this complement pathway in the pathogenesis of AD
.
Dysregulation of complement pathways may play a role
in a variety of central nervous system disorders.

To support the neurotoxic effects of complement, pharmacological or genetic suppression of the complement pathway improves neurodegeneration and synaptic loss
in mouse models of AD, MS, frontotemporal dementia, and neuroinvasive viral infections.
During development, microglia improve neuronal circuitry
by engulfing excess synapses.
In addition to microglia, in the adult brain and disease, astrocytes also show clearance of synapses
during development.
In contrast to microglia, astrocytes phagocytic synapses do not appear to be dependent on C1q
under physiological conditions.
Although reactive astrocytes have synaptic toxicity in different central nervous system diseases, including AD, HD, PD, and MS, their molecular mechanisms remain largely unclear
.

Recently, the journal Nature Aging published a research paper entitled "Complement C1q-dependent excitatory and inhibitory synapse elimination by astrocytes and microglia in Alzheimer's disease mouse models".
Through integrated multiomics analysis, the authors' team found that astrocytes and microglia were exposed to and cleared synapses through the Cq1 complement pathway in AD mouse models, and C1q deletion had a neuroprotective effect in TauP301S transgenic mice, suggesting that inhibition complement is an attractive strategy
to improve neurodegenerative diseases such as AD.

C1q deletion reduces neurodegeneration in P301S mice

To investigate the role of C1q in neurodegeneration in P301S mice, the authors' team performed gene ablation on C1q, using magnetic resonance imaging (MRI), behavioral and pathological analysis, and transcriptomic and synaptic proteomic analysis in male mice (Figure 1a).

C1q-deficient (C1qKO) mice have a reduced rate of brain volume growth during maturation in a gene-dose-dependent manner compared to wild-type (WT) mice (Figure 1b,c), but no difference in brain volume change in C3KO mice was observed (Supplementary Figure 1a), suggesting that C1q may have CCP-independent physiological functions
in the brain.
In P301S mice, brain volume decreased significantly between 6 and 9 months, reflecting neurodegeneration (Figure 1c, d).

On the P301S; In C1qKO mice, brain volume loss was mild at 9 months of age, and there was no significant difference in brain volume from C1qKO mice (Figure 1c, d).

The hippocampus is the brain region most affected by Tau pathology and glial cells, and the authors observed that P301S mice reduced hippocampal volume even at 6 months and further shrank between 6 and 9 months, while at P301S; Hippocampal volume loss in C1qKO mice is delayed, with significant protective effect at 6 months (Figure 1e,f).

In contrast to the protection provided by homozygous C1q knockout (C1qKO), P301S; C1qHet mice are similar to P301S mice, meaning that a greater than 50% reduction in C1q is required to prevent TauP301S neurodegeneration
.

To assess the behavioral consequences of C1q deletion, the authors assessed the motor activity of 9-month-old mice in open areas (Figure 1g).

As expected, P301S mice showed hyperactivity, which is thought to be caused
by hippocampal injury.
Although in the absence of P301S transgenetics, the C1q genotype had no effect on exercise capacity, P301S; Overactivity of C1qKO mice was rescued compared to P301S; C1qHet mice do not (Figure 1g).

Figure 1 C1q deletion reduces neurodegeneration in P301S mice

The team then analyzed the brain histopathology of 9-month-old mice
.
The C1q immune reactivity of the P301S brain is significantly enhanced, while that of P301S; C1qHet brain reduces C1q immunoreactivity by about 50%, P301S; C1q immunoreactivity of the C1qKO brain was not detected (Supplementary Figure 1b, c).

The P301S brain is characterized by strong phosphorylated Tau immunoreactivity, measured by increased area of Iba1+ and glial fibrous acidic protein (GFAP), respectively (Supplementary Figure 1d-f).

Compared to P301S mice, P301S; There was no difference in these histopathological readings from C1qHet from Mice P301S, P301S; C1qKO has a slight decreasing tendency compared to P301S mice (Supplementary Figure 1d-f).

The authors observed a trend towards reduced copper aminostaining, which mirrored P301S; C1qKO Hippocampal damaged neurons and increased density of neuronal markers NeuN (Supplementary Figure 1e-h).

P301S hippocampal RNA-seq shows upregulation of many genes, including multiple markers of activated microglia and astrocytes (Additional Figure 2); However, C1q deletion has no effect on these major transcriptome changes in the P301S hippocampus (Supplementary Figure 2).

Overall, these results suggest that C1q deletion reduces P301S brain degeneration and normalizes behavior, but has no significant effect
on Tau pathology, glial cell hyperplasia, or the extent of glial cell transcriptional changes.
Thus, the protective effect of C1q deletion appears to play a role
downstream of Tau pathology and overall glial responses.

Supplementary Figure 1 Brain Hcar2-induction, APPPS1 and PS19 mouse models of 5xFAD mice

Hcar2 expression and validation of the Hcar2 antibody ab81825

C1qKO attenuates proteome changes in synapses of P301S mice

The authors' team next investigated changes
in synaptic protein composition in the P301S and C1q genotypes.
The authors isolated the hippocampal postsynaptic density (PSD) component from male mice aged 6 and 9 months, respectively, to detect early and late disease changes (Figure 2a).

These moieties are highly enriched in proteins of PSD and postsynaptic membranes; They also contain components of presynaptic terminals, cross-synaptic adhesion molecules, and some glial cell-specific proteins that may reflect the close interaction of glial cells with synapses
.
Therefore, the author team used the terms PSD and synaptic score
interchangeably throughout the study.
Multiple tandem mass tag (TMT) proteomics detected a total of 7101 proteins in PSD fragments of 6-month-old mice and 4175 proteins in PSD fragments of 9-month-old mice (Figure 2b and Supplementary Table 1).

While the authors previously detected fewer proteins in their analysis of 9-month-old female mice of P301S using label-free proteomics, almost all of them were also found in the current study (Figure 2b).

Synaptic fragments from C1qKO mice showed only a small amount of differentially expressed (DE) protein at both ages compared to WT (Figure 2c and Supplementary Figure 3a,c).

In the 6-month-old P301S synaptic fragment, the authors found 108 downregulated proteins and 68 upregulated proteins (2.
5% of the total protein; Figure 2 c, e).

At 9 months, 253 proteins were down-regulated and 434 proteins were up-regulated, accounting for 16.
5% of the total protein (Figure 2c, e).

In contrast, the P301S; The C1qKO synaptic fragment showed only 17 DE protein decreases and 19 DE protein increases (0.
5% of total protein) at 6 months, compared to 79 DE protein decreases and 224 DE protein increases (7% of total protein) at 9 months (Figure 2c, f).

In comparison P301S; At synapses at C1qKO and P301S at 9 months, the authors consistently identified many DE proteins (Figure 2c and Supplementary Figure 3b), and with C1q deletion, Tau-dependent changes decreased (Supplementary Figure 3d).

C1q deficiency does not affect Tau levels in synaptic parts of the P301S brain (Figure 2e, f).

Thus, C1q deletion attenuates the pathologically induced age-dependent changes in Tau, although C1q deletion has little effect in non-transgenic mice (Figure 2d and Supplementary Figure 3d).

Since the deletion of C1q did not significantly alter the transcriptome changes in the P301S brain (Supplementary Figure 2), the changes in the synaptic proteome are most likely caused
by local protein changes in the synapses.

Figure 2 Proteome changes in which C1q deletion attenuates P301S synapses

The authors evaluated the effects of the P301S transgene and C1q deletion on different functional classes of proteins in the synaptic proteome (Supplementary Figure 3e).

Compared to WT synapses, many core PSD proteins in P301S, such as glutamate receptors, scaffold proteins, synaptic adhesion molecules, and certain presynaptic active zone proteins, tend to increase at 6 months but decrease at 9 months (Supplementary Figure 3e).

This may reflect compensatory synaptic changes early in the disease, overcome by synaptic damage at a later stage
.
Using the synaptic gene ontology tool SynGO, the authors' team confirmed that the downregulated DE protein in 9-month-old P301S PSDs is significantly rich in synaptic tissue and typical presynaptic and postsynaptic proteins (Supplementary Figure 3f).

Although the P301S; The synaptic function of the C1qKO synaptic fragment is similar in quality, but its changes are not as significant as those of P301S PSDs (Supplementary Figure 3f).

Analysis of the KEGG pathway of P301S synaptic DE protein showed significant enrichment of "metabolic pathway" (mainly containing mitochondrial proteins such as Echs1, Maob and Atp8), "fatty acid degradation" (such as Adh5, Hadha and Hadh), "peroxisomes" (such as Dhrs4, Ehhadh and Abcd1), and "peroxisome proliferator-activated receptor (PPAR)" signals (such as Cpt1a and Cpt2).
"Alzheimer's disease" (e.
g.
, Adam10 and APOE) and ionic homeostasis (e.
g.
, Slc4a4 and Aqp4) are particularly prominent at 9 months (Figure 2g).

On the P301S; In the C1qKO synapse, the increase in these pathways was relatively weak, with no significant increase at 6 months, and the change at 9 months was similar to the change in P301S PSDs at 6 months (Figure 2g).

At the same time, most of the pathways added in the 9-month-old P301S synapse compared to P301S were in P301S; Reduction in C1qKO (Supplementary Figure 3b, c).

It is worth noting that the "metabolic pathway" is the most significantly increased pathway in the 9-month-old P301S synapses, in P301S; There was no significant induction in C1qKO (Figure 2g).

We also note that annexin is one of the most inducible proteins in the 9-month-old TauP301S synapse and is found in P301S; C1qKO is partially normalized (Figure 2d and Supplementary Figure 3d).

Analysis of DE protein reduction in P301S synapses highlighted "glutamate synapses" (such as Shank1 and Grin2a), "endocytosis" (such as Vps4a and Rab11Fip2), "axonal guidance" (such as Ntng1 and Smad2), "MAPK-" (such as Mapk1 and Map4k4), "Ras-" (such as Ksr1 and Syngap1), "phosphatidylinositol-" (such as Dgkb and Dgkq) and "wnt - Signal" (Apc2 and Dvl3) and "Ampk-Signal" (PPP2r5c and Prkaa2) pathways (Figure 2g).

P301S at 6 months old; In the C1qKO synapse, no pathway was significantly reduced, and only "glutamatergic synapse", "endocytosis", and "ras signaling" were significantly reduced at 9 months (Figure 2g).

In the P301S PSD fragment, different proteins of the "actin cytoskeleton" pathway were significantly increased (e.
g.
, ezrin and gelsolin) or decreased (e.
g.
, Baiap2 and Pak6).

In the synaptic portion, at least some of the increased actin regulatory proteins are expressed by glial cells that may be associated with synapses
.

Because synaptic protein mutations lead to multiple neurological and neuropsychiatric disorders, the authors' team investigated whether the DE protein was enriched with genetic signals
in genome-wide association studies (GW As) for related traits and diseases.
For the 1000 proteins with the most up- and down-regulated in P301S and WT synapses, the authors tested polygenic signals for 752 personality traits using stratified linked imbalance (LD) score regression (methods), mainly from the UK Biobank, and selected the GWAS study
。 After classifying traits into 23 categories or domains, the authors found 9-month-old P301S and P301S; C1qKO synaptic downregulated proteins are significantly enriched in cognitive, mental, and activity domains, including educational achievement, intelligence scores, cognitive performance, and conceptual interpolation (Figure 4a).

In contrast, enrichment of these traits is limited in upregulated proteins from 9-month-old P301S PSD fragments or 6-month-old DE proteins (Supplementary Figure 4b).

This indicates 9-month-old P301S and P301S; C1qKO synaptic protein downregulation is associated
with human cognitive function and behavior.
Supplementary Figure 3 Changes in the C1q-dependent synaptic proteome of P301S mice

Supplementary Figure 4 P301S and P301S; Genetic enrichment analysis of differentially rich proteins identified by the synaptic fraction of C1qKO

C1q-dependent elevation of glial cell protein at P301S synapses

The authors note that many typical astrocyte-specific proteins such as Aqp4, Mlc1, and Slc1a4 increase in a c1q-dependent manner in the 9-month-old P301S synaptic fragment (Figure 2d).

While contamination of astrocyte proteins is possible, the authors also considered whether copolymerization of astrocytes proteins with synaptic preparations caused close interaction
between astrocytes and synapses.
The authors generated pseudosomal single-cell RNA sequencing (scRNA-seq) data from the P301S hippocampus and found that most of the 55 highly upregulated proteins were primarily expressed by glial cells rather than excitatory neurons (Figure 3a).

In contrast, the protein that decreases the most is mainly produced by excitatory neurons (Supplementary Figure 5a).

In addition to Aqp4 and Mlc1, many other upregulated DE proteins are selectively or predominantly expressed by astrocytes (e.
g.
, cluster proteins, Slc1a3, Sdc4, AHNAK, ezrin, GFAP, and Thbs4) (Figure 3a).

A small amount of upregulated protein is mainly expressed by microglia (e.
g.
, Gpnmb and Myo1f) (Figure 3A).

The authors hypothesize that the surge in glial protein in the 9-month-old P301S synaptic fraction may reflect an increase in its secretion and subsequent accumulation at the synapse, and/or an increase
in contact of glial protrusions with damaged synapses.
Consistently, many of the increased glial proteins are either localized to the plasma membrane (e.
g.
, Slc16a1 and Aqp4), cytoskeleton (e.
g.
, ezrin), or extracellular/secretory (clusterin and Thbs4; Supplementary Figure 5b), and is known to be present in astrocytes processes
.

Of note, the increase in glial protein in the synaptic fraction of P301S is C1q-dependent (Figure 3b).

This opposes indiscriminate contamination of PSD preparations
with glial-derived proteins.
P301S; The relative decrease in glial protein in the C1qKO synapse appears to be due to a specific change in the binding of glial protein to the synaptic moiety, rather than the overall abundance of glial protein, as P301S versus P301S; There was no significant difference in the expression of the corresponding gene in the C1qKO brain (Supplementary Figure 5c).

The proteins that increase most in the P301S synapse are the astrocyte-specific mitochondrial proteins Echdc3 and Sfxn5 (Figure 3a,b).

Many mitochondrial proteins increase in an age- and C1q-dependent manner in P301S synaptic fragments (Supplementary Figure 5d, e).

Energy metabolism differs between central nervous system cell types, with a significant increase in the "metabolic pathway" in P301S at 9 months, but P301S; The C1qKO synapse is not elevated (Figure 2g).

Although mitochondria can be transferred from astrocytes to neurons under pathological conditions, the authors deduce that mitochondria in the synaptic part may arise from glial processes
that have close physical contact with synapses.
Historically, the 20 mitochondrial proteins with the largest increase in the P301S synapse were mainly expressed by astrocytes (e.
g.
, Maob, Cpt1a and Tst, Slc25a18), while significantly reduced mitochondrial proteins had a wide range of expression patterns, including stronger neuronal production (e.
g.
, Wasf1) (Supplementary Figure 5f).

Similarly, peroxisome proteins increased in a C1q-dependent manner in 9-month-old P301S synaptic fragments are also predominantly expressed by astrocytes (Supplementary Figure 5g).

Supplementary Figure 5 Abundance of synaptic glial protein and expression of its corresponding genes

Next, the authors' team used immunoelectron microscopy (IEM) to determine whether the physical connection of astrocytes to synapses was altered
in P301S mice.
They identified astrocyte processes by immunolabeling of the astrocyte-specific glutamate transporter EAAT2/Glt1 and quantified the length of astrocyte processes in contact with hippocampal dentate gyrus (DG) and CA1 region synapses (Figure 3c and Supplementary Figure 6a).

The length of synapse-associated astrocyte protrusions in the DG and CA1 regions of P301S mice increased approximately twofold compared to WT mice (Figure 3d and Supplementary Figure 6b).

There was no change in the average synaptic perimeter of P301S mice, meaning that astrocytes were exposed to a larger proportion of the synaptic membrane (Figure 3e and Supplementary Figure 6c).

Of note, the degree of astrocytes-synaptic association is significantly associated with the presynaptic percentage labeled by C1q3 (Figure 3f).

The authors quantified the spatial contact of excitatory synapses (Homer1 puncta) with the surface of GFAP+ astrocytes in the CA1 region of the hippocampus using confocal microscopy (Figure 3g).

Since the loss of one copy of C1q has no effect on the pathology of P301S mice (Figure 1 and Supplementary Figure 1), the authors set P301S and P301S in this analysis; C1qHet brains are grouped to increase statistical power
.
Although there was no difference between C1qKO and WT hippocampus, we observed a significant increase in the surface GFAP–Homer1 association of the P301S hippocampus, compared with P301S; C1qKO is significantly lower compared to P301S (Figure 3h).

The cytoplasmic astrocyte marker protein S100b was used to visualize astrocyte volume, confirming the C1q-dependent increase in astrocytes–Homer1 association in P301S mice (Supplementary Figure 6d, e).

In summary, the authors' analysis of synaptic proteomics data, IEM, and immunohistochemical (IHC) measurements suggests that astrocytes can increase interaction
with synapses in a C1q-dependent manner during the disease stage of synaptic phagocytosis and loss.

Figure 3 Glial protein is elevated at the P301S synapse and normalized by C1qKO

Supplementary Figure 6 iEM and iHC analysis of astrocytes-synaptic interactions

Elevated glial proteins in AD brain synapses

The authors wanted to know if the glial protein in the synaptic part of AD patients was also elevated
.
A comparison with recently published data on the synaptic proteome of the superior temporal gyrus (BA 41/42) in AD patients showed a significant positive correlation between changes in human AD and control groups and between the synaptic proteome of 9-month-old P301S and WT mice (Figure 4a).

Of note, glial proteins elevated in the P301S synaptic fragment, including complement factors C1q and C4, astrocyte marker proteins MLC1 and GFAP, microglia GPNMB and AHNAK, and annexin are the most elevated proteins in AD synaptosomes (Figure 4a).

The authors reasoned that elevated C4 levels and CCP activation could be detected in the patient's CSF, which could be a useful biomarker for complement activation
.
In CSF in AD patients, the total C4 concentration and the processing (lysis and activation) C4 concentration increased significantly (Figure 4b).

Similar results were observed in CSFs from independent patient cohorts, with a trend toward elevated total C4 and a significant increase
in processed C4.
In contrast, complement factor B (a component of the alternative complement pathway) was not significantly altered in AD CSF (Supplementary Figure 7).

In contrast, complement factor B (a component of the alternative complement pathway) was not significantly altered in AD CSF (Figure 4b and Supplementary Figure 7).

Activated Bb sublevels were very low in the control group and AD CSF, but did show an elevated trend in AD CSF (Figure 4B and Supplementary Figure 7).

Therefore, upregulation of glial proteins in the P301S synaptic proteome also appears in AD and may be related to
the pathophysiology of the disease.
Elevated levels of C4 (and C3) in cerebrospinal fluid in AD patients are consistent with
the role of CCP in neurodegeneration in Alzheimer's disease.

Fig.
4 The glial protein in the synaptic part of human AD is increased, and C4 in AD CSF is increased

Glial C1q-dependent synaptic elimination in P301S mice

Based on proteomics, IEM, and IHC data from the authors' team, hypothesis is that astrocytes may interact with synapses in a C1q-dependent manner during synaptic phagocytosis
.
To analyze astrocyte and microglia phagocytic synapses, the authors' team immunostained GFAP+ astrocytes, Iba1+ microglia, Lamp1+ lysosomes, and the excitatory postsynaptic marker Homer1 (Figure 5a).

Since inhibitory synapses are also affected in AD, we also immunolabeled
the inhibitory postsynaptic marker gephyrin.
The number of Homer1 and gephyrin spots in microglia and astrocyte lysosomes was measured in the same image by confocal microscopy and 3D reconstruction of the CA1 region of the hippocampus (Figure 5a).

Notably, using GFAP or S100B, the authors essentially identified the same astrocyte Lamp1+ structural population (Supplementary Figure 8a).

As expected in previous studies, the microglial lysosomes in the P301S hippocampus contain excitatory synapses, with a approximately 10-fold increase in Homer1 points compared to the WT control (Figure 5b).

Compared to P301S, P301S; The microglia phagocytosis of Homer1 in the brain of C1qKO is significantly reduced (Figure 5b).

Notably, the authors also found a significant portion of Homer1 spots in astrocytes lysosomes, which increased 5- to 10-fold in the hippocampal region of P301S (Figure 5c).

P301S; The proportion of Homer1 spots in lysosomal bodies of C1qKO brain astrocytes was significantly reduced, indicating that astrocyte phagocytosis of the excitatory structure of P301S mice is at least partially dependent on C1q (Figure 5c).

Gephyrin spots are also present in microglia and astrocytes lysosomes (Figure 5d, e).

As with Homer1, in the P301S hippocampus, gephyrin uptake is elevated by microglia and astrocytes and is partially dependent on C1q (Figure 5d, e).

However, in healthy brains, phagocytosis of excitatory and inhibitory synapses is not affected by C1q loss, as WT and C1qKO hippocampals are equally low in glial lysosomal Homer1 and gephyrin (Figure 5b-e).

The number of phagocytic synaptic junctions corresponds to volume changes in lysosomes and microglial lysosomes across genotypes (Supplementary Figure 8b).

Notably, in all genotypes, more Homer1 spots were consistently found within astrocytes than microglial lysosomes (Figure 5f).

In contrast, gephyrin puncta levels were lower in astrocytes lysosomes and microglial lysosomes of the P301S hippocampus (regardless of C1q genotype) and similar to gephyrin puncta levels in astrocytes and microglia from non-transgenic WT animals (Figure 5g).

。 This tendency to engulf gephyrin puncta is particularly pronounced in the P301S brain, where intense gephyrin puncta immunoreactive accumulation is often observed rather than astrocyte lysosomes (Supplementary Figure 8c).

One possibility is that this strong immune response may reflect the removal
of dendritic fragments containing many inhibitory synapses.
with P301S; Synaptic phagocytosis was consistently reduced in C1qKO brains, and excitatory and inhibitory synaptic loss was improved in C1q-deficient P301S mice (Figure 5h).

Finally, the authors tested whether astrocyte phagocytotic synaptic structures require C3, a central complement component
downstream of C1q.
Significant increase in phagocytosis of Homer1 and gephyrin by microglia and astrocytes in P301S mice, P301S; Metaphasis mean fractional reduction of Homer1 and gephyrin in C1qKO mice (Supplementary Figure 8d-g).

However, only the spotted gel algalin in astrocytes lysosomes is reduced in P301S; Statistical significance was achieved in C1qKO versus P301S mice (Supplementary Figure 8g), possibly due to high animal-to-animal
variability.
Just like in the C1q experimental cohort, the authors' team found more Homer1 spots in astrocytes lysosomes in each brain they analyzed, and gephyrin puncta in microglial lysosomes in P301S mice in the C3 experimental cohort (Supplementary Figure 8h, i).

Next, the authors analyzed the P301S; C3 labeling
of excitatory synapses in the C1qKO cohort.
The percentage of C3+ Homer1 tumors in the hippocampus of P301S was significantly increased compared to WT, compared with that of P301S; C1qKO was comparable to the WT hippocampus and significantly reduced compared to the P301S brain (Figure 5i, j).

P301S vs P301S; There was no significant difference in the total number of C3 punctate protrusions in the C1qKO brain (Figure 5k), suggesting that the absence of C1q particularly affected C3 deposition at synapses
.
Overall, the data suggest that C3 acts downstream of C1q activation, CCP promotes astrocytes and microglia to eliminate synapses
.

Figure 5 Astrocytes and microglia eliminate excitatory and inhibitory synapses in P301S mice in a C1q-dependent manner

Supplementary Figure 8 P301S mouse astrocytes and microglia pairs

Complement-dependent phagocytosis of excitatory and inhibitory synapses

Astrocytes compensate for impaired microglial phagocytosis

Microglia-specific loss or mutation of TREM2 function increases the risk of AD, while loss of TREM2 in mouse models of AD has profound effects on microglial function and inhibits its activation, migration to Aβ plaques, and phagocytic activity
.
To examine whether microglial dysfunction may affect astrocytes' clearance of synapses, the authors analyzed the effect
of Trem2 deletion in the AD mouse model TauPS2APP (which binds Aβ amyloid and Tau pathology).
。 At 17 months of age, when plaque, Tau phosphate, dystrophic axons and glial hyperplasia are present, the authors are interested in the TauPS2 APP and TauPS2APP; Trem2KO brain sections were immunostained with previously established protocols and imaged the CA1 region of the hippocampus with and without amyloid plaques (identified by the presence of Lamp1+ malnutrition axons) (Figure 6a).

In the TauPS2 brain, microglia and astrocytes engulf more Homer1 and gephyrin puncta near the plaque compared to plaque-free regions (Figure 6b-e).

in the TauPS2APP; In the Trem2KO brain, microglia near the plaque engulfed synapses significantly less compared to TauPS2 APP mice (Figure 6b,c), while astrocytes phagocytosed Homer1 puncta were not affected by Trem2 deficiency (Figure 6d).

It is worth noting that compared to the TauPS2 APP brain, in the TauPS2APP; In Trem2KO, the effect of gephyrin puncta in astrocytes near the plaque is significantly increased (Figure 6e).

As in P301S mice, astrocytes lysosomes contain more Homer1 puncta compared to microglial lysosomes (Figure 6f), while gephyrin puncta is more abundant in microglial lysosomes (Figure 6g)
in TauPS2 APP mice.
The overall effect of trem2 deficiency in TauPS2 mice was an increased proportion of Homer1 and gephyrin in astrocytes and microglial lysosomes (Figure 6f, g).

Therefore, Trem2 is necessary for microglia near plaques to effectively phagocytose synapses, and astrocytes can compensate, at least to some extent, for the impaired
phagocytosis of microglia that inhibit synapses.

Fig.
6 Astrocytes compensate for Trem2 deficiency TauPS2APP

Impaired phagocytosis of microglia with inhibitory synapses in mice

Through in-depth analysis of proteome data, the authors' team discovered a novel role
of astrocytes and microglia on excitatory and inhibitory synaptic clearance through complement-dependent mode.
Indicates the concomitant/coordinated role
of astrocytes and microglia in synaptic phagocytosis in pathophysiological processes.
This study also found that astrocytes and microglia may sense different complement molecules
at synapses.
Overall, the findings of the authors' team advance the understanding of the underlying mechanisms of complement-mediated synaptic elimination and neuronal damage, identify the unexpected division of inhibitory synapses and excitatory synaptic phagocytosis in microglia and astrocytes, and open new avenues
for potential treatments for AD.