Peli1 impairs microglial Aβ phagocytosis through promoting C/EBPβ degradation

Amyloid-β (Aβ) accumulation in the brain is a hallmark of Alzheimer’s disease (AD) pathology. However, the molecular mechanism controlling microglial Aβ phagocytosis is poorly understood. Here we found that the E3 ubiquitin ligase Pellino 1 (Peli1) is induced in the microglia of AD-like five familial AD (5×FAD) mice, whose phagocytic efficiency for Aβ was then impaired, and therefore Peli1 depletion suppressed the Aβ deposition in the brains of 5×FAD mice. Mechanistic characterizations indicated that Peli1 directly targeted CCAAT/enhancer-binding protein (C/EBP)β, a major transcription factor responsible for the transcription of scavenger receptor CD36. Peli1 functioned as a direct E3 ubiquitin ligase of C/EBPβ and mediated its ubiquitination-induced degradation. Consequently, loss of Peli1 increased the protein levels of C/EBPβ and the expression of CD36 and thus, promoted the phagocytic ability in microglial cells. Together, our findings established Peli1 as a critical regulator of microglial phagocytosis and highlighted the therapeutic potential by targeting Peli1 for the treatment of microglia-mediated neurological diseases.


Introduction
Alzheimer's disease (AD) is a rising threat to public health worldwide [1] and is a progressive neurodegenerative disorder with the hallmark of β-amyloid (Aβ) accumulation in the brains [2,3]. Although the etiology of AD is largely unknown, the studies by analyzing the inherited AD families have suggested that multiple mutations in Aβ processing cascade are genetically associated with the disease pathogenesis [4][5][6][7][8]. However, human clinical trials by targeting Aβ production and deposition had failed because of the lack of efficacy in the improvement of cognitive outcomes in AD patients [9][10][11], which implies that the relationship between AD and Aβ plaques is still controversial and arouses the doubt on the validity about amyloid cascade hypothesis. Nevertheless, there are still numerous ongoing potential anti-Aβ antibodies therapies tested in the clinical treatment of AD, such as Gantenerumab and BAN2401. In addition, recent works also suggested that Aβ may impact the Tau hyperphosphorylation and a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 cognitive deficit in AD-like mice, thereby affecting disease symptoms in AD patients as well [12,13]. Moreover, the anti-Aβ antibody Aducanumab has been submitted to FDA for approval, which provided a compelling support for the amyloid cascade hypothesis.
Microglia are the resident macrophages in the brain and can cluster around the amyloid plaques during AD pathogenesis. Previous studies demonstrated that microglial phagocytosis ability was associated with Aβ deposition in mouse AD models [14][15][16][17][18][19]. In contrast, microglial dysfunction is associated with aging in mice and human brains and is relevant with higher levels of Aβ load in mouse AD models [20][21][22][23]. There are also other studies demonstrated that removal of microglia by colony-stimulating factor 1 receptor (CSF1R) inhibitor prevents neuronal loss and improves cognition without affecting Aβ pathology in AD-like mice [24,25]. These studies suggest that the pathological mechanism of AD and the function of microgliamediated Aβ clearance are still a controversial and debated topic in the fields of neurodegenerative disorders. However, little is known about the intrinsic molecular mechanism through which Aβ is engulfed by microglial phagocytosis. We and others have previously revealed that the E3 ubiquitin ligase Pellino 1 (Peli1) is expressed in many kinds of neural cells in murine brain, with the highest expression level in the microglia [26], implying a nonredundant function of Peli1 in central nervous system (CNS) to modulate microglia-related neurological diseases. Indeed, we have identified Peli1 as a crucial modulator of microglia-mediated autoimmune neuroinflammation and viral encephalitis [26][27][28].
Here, we identified Peli1 as an essential negative regulator of microglial phagocytosis through inhibiting the expression of scavenger receptor CD36. Mechanistic studies revealed that Peli1 functioned as the direct E3 ubiquitin ligase of C/EBPβ, a major transcription factor that responsible for CD36 expression, and mediated its degradation. More interestingly, microglial Peli1 is induced in AD brains and thus, further impaired microglia-mediated Aβ phagocytosis during disease pathogenesis.

Peli1 deficiency promoted microglial phagocytic ability
We had previously reported that Peli1 could regulate microglia-mediated CNS inflammation in a multiple sclerosis mouse model [26]. However, because of the high expression of Peli1 in microglia compared with other types of cells in CNS, we speculated that Peli1 could also act as critical regulators in other physiological and pathological process in microglia. To investigate the potential role of Peli1 in modulating microglial phagocytosis, we isolated and cultured the primary microglia from the heterozygous (Peli1 +/− ) and Peli1-deficent (Peli1 −/− ) newborn mice from the same parents (S1A Fig). After challenging the primary microglia with fluorescence-labeled microspheres, we observed a robust phagocytic activity of Peli1-intact cells as determined by flow cytometry (Fig 1A and 1B). Interestingly, loss of Peli1 enhanced microglial phagocytosis of microspheres, as reflected by increased frequencies of fluorescent cells in Peli1-deficient microglia (Fig 1A and 1B). In contrast, Peli1 deficiency does not affect the phagocytic ability of primary cultured astrocytes (Fig 1C and 1D). We further validated the protein levels of Peli1 in microglia and astrocytes with immunoblot, and the data showed that Peli1 was highly expressed in microglia when compared with that in astrocytes (S1B Fig), which supported the previous study [26] and suggested a specific and nonredundant function of Peli1 in microglia to modulate phagocytosis. Subsequent flow cytometry analysis revealed that the phagocytic microspheres-induced mean fluorescent intensity (MFI) in Peli1-deficient microglia was enhanced as compared with Peli1-competent cells (S1C and S1D Fig).
Although Peli1 ablation didn't affect the ability of microglia to phagocytize 1-3 microspheres, Peli1-deficient microglia were more competent to phagocytize more than 4 microspheres (S1E Fig), suggesting that Peli1 negatively regulates microglial phagocytic efficiency. Furthermore, to confirm the function of Peli1 in regulating microglial phagocytosis, we examined the phagocytic ability for the synthetic Aβ 1-42 peptide by using the murine primary microglia. Consistent with the data of microspheres phagocytosis, Peli1 deficiency also enhanced the microglial ability to phagocytize Aβ 1-42 , as reflected by increased MFI of Aβ 1-42containing cells and elevated Aβ 1-42 loading per microglia (Fig 1E-1H). In order to validate that the Aβ 1-42 oligomers were really internalized by phagocytosis rather than endocytosis, we performed the phagocytic assay of Aβ 1-42 peptide with Cytochalasin D to inhibit actin polymerization. As the data show, the differences of Aβ 1-42 phagocytosis between Peli1-sufficient and Peli1-deficient microglia were totally abolished and significantly inhibited after Cytochalasin D treatment (S1F and S1G Fig), implying that the Aβ 1-42 peptide entered microglia via phagocytosis but not endocytosis. The ability of microglial phagocytosis was also essential for the clearance of apoptotic neurons. Therefore, we tested the phagocytic ability of apoptotic neural Neuron-2A (N2A) cells in Peli1-deficient microglia. The data demonstrated that the phagocytosis of apoptotic N2A was not affected after Peli1 was deleted in microglia (S1H and S1I Fig), which further supported the critical and specific role of Peli1 in microglial phagocytosis of Aβ 1-42 peptide.
To exclude the potential possibility of proliferation-induced alteration of phagocytic ability, we examined whether Peli1 modulates microglial proliferation. The results revealed that Peli1 deficiency in primary microglia or its knockdown in BV2 microglial cells does not affect the growth of microglial cells by Ki-67 staining (S2A- S2C Fig). Consistent with the results obtained from primary microglia, we found that reducing Peli1 levels in BV2 cells also promoted the phagocytosis activity, as characterized by increased frequencies of fluorescent cells and enhanced microglial MFI levels when compared with control cells (Fig 1I-1L). In addition, we performed a time course assay of microspheres phagocytosis in BV2 cells. The results suggested that Peli1 knockdown indeed led to increased microspheres phagocytosis after microspheres inoculation for 10, 20, 30, and 60 minutes (S2D and S2E Fig). Furthermore, the percentages of microsphere phagocytosis in Peli1-knockdown BV2 for more than 4 particles per cell were higher, whereas the BV2 percentages for 1 particle per cell were lower than that in control at all of the later time points, implying that Peli1-knockdown BV2 could phagocytize more throughout the whole process of phagocytosis (S2F Fig). Moreover, the difference of percentages between control and Peli1-knockdown BV2 cells that phagocytize more than 4 particles was smaller at 60 minutes than at 10 minutes (S2D and S2F Fig), which revealed that the rates of phagocytosis in both cells were altered with time. Consistent with the result of microspheres phagocytosis, Peli1 knockdown in BV2 cells also enhanced the ability to phagocytize Aβ 1-42 after incubation with Aβ for 12 hours or 24 hours (S2G and S2H Fig).
To further confirm the role of Peli1 in regulating microglial phagocytosis, we reconstituted Peli1-knockdown BV2 cells with full-length (FL) Peli1 or C-terminal-deleted (Peli1ΔC) Peli1 (Fig 1M), which lacked the RING-like domain and thus lost its E3 ligase activity, and assayed the phagocytic activities. The results showed that reconstitution with only FL-Peli1, but not its functional deficient Peli1ΔC, could successfully rescue the effect of Peli1 in suppressing microglial phagocytic ability (Fig 1N and 1O). Collectively, these data identified an intrinsic negative role of Peli1 in regulating microglial phagocytosis.

Enhanced phagocytosis in Peli1-deficient microglia was dependent on C/ EBPβ
It is known that the molecular structures of extracellular substrates are recognized by different types of scavenger receptors that are expressed on the surface of microglia, which then initiate phagocytosis for its clearance function [29][30][31][32][33]. To determine whether enhanced phagocytic efficiency observed in Peli1-deficient microglia was due to the alteration of the expression of these phagocytic receptors, we examined the surface expression of scavenger receptors class A (SRA) and CD36, the major scavenger receptors responsible for the clearance of Aβ in microglia [29,30,34]. The results revealed that Peli1 deficiency or knockdown didn't affect the surface expression of SRA in primary microglia or BV2 cells (S3A-S3D Fig). Interestingly, reducing Peli1 levels promoted CD36 expression on the surface of primary microglia, BV2 cells, and the microglia that directly isolated from the brains of adult mice (Fig 2A-2F). In addition, loss of Peli1 also increased the total protein levels of CD36 in these cells (S3E-S3G Fig), which further validated the negative role of Peli1 in regulating CD36 expression. More intriguingly, CD36 blocking with antibody totally abolished the phagocytic difference between Peli1-competent cells and Peli1-deficient primary microglia or Peli1-knockdown BV2 cells (Fig 2G), implying that elevated CD36 expression in Peli1-deficient or Peli1-knockdown cells contributed to their enhanced phagocytic ability.
To figure out how Peli1 regulates CD36 expression on the surface of microglia, we initially examined its mRNA expression in microglial cell. The results revealed that Peli1 deficiency or knockdown enhanced the mRNA expression of Cd36 in both primary microglia and BV2 cells (Fig 2H), suggesting that Peli1-mediated alteration of CD36 expression was directly modulated at transcriptional level. In order to exclude the potential possibility of other receptors in regulating microglial phagocytosis in Peli1-deficient cells, we further tested the mRNA levels of scavenger receptor B-1 (Srb1), macrophage receptor with collagenous structure (Marco), the receptor for advanced glycation end product (Rage), and triggering receptor expressed on myeloid cells 2 (Trem2), which also mediated Aβ phagocytosis as previous reported [35]. The data showed that Peli1 deficiency did not promote the expression of SR-B1, MARCO, RAGE, or TREM2 in murine primary microglia (S4A Fig), implying the specific role of Peli1 in regulating Cd36 transcription.
The mRNA expression of Cd36 was previously reported to be tightly controlled by 2 major transcription factors, CCAAT/enhancer-binding protein (C/EBP) α and C/EBPβ, which could directly bind to the Cd36 gene promoter and initiate its mRNA transcription [36]. Although C/EBPα was more potent than C/EBPβ to induce Cd36 transcription, our results revealed that Peli1 deficiency in microglia or Peli1 knockdown in BV2 cells did not induced C/EBPα protein expression (Fig 2I). In contrast, Peli1 deficiency or knockdown significantly increased the protein levels of C/EBPβ in both primary microglia and BV2 cells (Fig 2I, S4B and S4C Fig). In addition, we confirmed that increased C/EBPβ protein expression was also observed in Peli1deficent adult microglia as compared with that in wild-type (WT) cells (Fig 2J and 2K). Accordingly, the chromatin immunoprecipitation (ChIP)-quantitative PCR (qPCR) analysis verified that there were indeed more C/EBPβ protein binding in the Cd36 gene promoter of Peli1-deficient primary microglia, adult microglia, and Peli1-knockdown BV2 cells than that in control cells (Fig 2L). These results suggested that Peli1 modulated the transcription of scavenger receptor Cd36 by specifically targeting C/EBPβ protein expression in microglial cells.
To confirm that increased CD36 expression and phagocytic activity in Peli1-deficient microglia is due to elevated expression of C/EBPβ, we reduced Cebpb expression in microglial cells. The results revealed that knockdown of Cebpb compromised the increased phagocytosis of Aβ 1-42 and microspheres in both Peli1-deficient primary microglia and Peli1-knockdown BV2 cells (Fig 2M and 2N which confirmed the dispensable role of C/EBPα in regulating Peli1-mediated inhibition of microglial phagocytosis. Therefore, these data collectively demonstrated that C/EBPβ was important for the enhanced CD36-mediated phagocytosis in microglia with Peli1 absent.

Peli1 mediated the ubiquitination and degradation of C/EBPβ
To dissect the molecular mechanism through which Peli1 regulated C/EBPβ expression, we firstly examined the mRNA levels of Cebpb. The results revealed that reducing Peli1 expression does not alter the Cebpb mRNA levels in BV2 cells (Fig 3A), implying Peli1 might modulate C/EBPβ expression directly at protein level. Indeed, the C/EBPβ protein levels are comparable between control and Peli1-knockdown BV2 cells in the presence of a proteasome inhibitor MG132 (Fig 3B), suggesting Peli1 might regulate the protein stability of C/EBPβ. We and others have previously demonstrated that Peli1 was an E3 ubiquitin ligase mediating protein activation or stability through adding the poly-ubiquitin chains to the substrates [26,[37][38][39]. The increased C/EBPβ protein levels in Peli1-knockdown microglial cells promoted us to investigate whether Peli1 directly regulated the ubiquitination of this transcription factor. The immunofluorescent images showed that Peli1 was associated with C/EBPβ in the nucleus of both murine primary microglia and BV2 cells (Fig 3C), which was confirmed by the coimmunoprecipitated Peli1 when C/EBPβ was pulled down in BV2 cells (Fig 3D). Next, we examined the effect of Peli1 for the ubiquitination status of C/EBPβ in microglial cells, the result revealed that Peli1 deficiency or knockdown inhibited the endogenous C/EBPβ ubiquitination in both primary microglia and BV2 cells (Fig 3E and 3F). In contrast, Peli1 didn't affect the ubiquitination or protein level of C/EBPβ in astrocytes, implying the specific role of Peli1 in the regulation of protein stability of microglial C/EBPβ (S5H Fig). In addition, overexpression of FL-Peli1, but not Peli1ΔC, markedly enhanced C/EBPβ ubiquitination in 293T cells (Fig 3G), which explained why Peli1ΔC failed to rescue the suppressive effect of phagocytosis for Peli1 in BV2 cells (Fig 1N and 1O). Moreover, in vitro ubiquitination assay indicated that FL-Peli1 but not its RING-deletion mutant mediated the ubiquitination of C/EBPβ in the presence of precharged E2 ubiquitin-conjugating enzyme H5a (UbcH5a) (Fig 3H). Together, these results suggested that Peli1 directly associated with C/EBPβ and mediated the poly-ubiquitination and degradation of C/EBPβ in microglial cells.

Peli1 deficiency promoted in vivo Aβ clearance
Because CD36 is critical for the microglial phagocytosis of Aβ, we next examined whether the phagocytosis of physiological Aβ was enhanced in Peli1-deficient microglial cells. To this end, we performed an ex vivo Aβ phagocytosis experiment [29,30,40], in which microglial cells were added to incubate with unfixed brain slices from 10-month-old five familial AD (5×FAD) transgenic mice, an accelerated mouse model mimicking the symptoms of AD [41]. These brain slices contained numerous Aβ depositions, and thus were usually used to measure the microglial phagocytosis of physiological Aβ. As the data show, control BV2 cells efficiently phagocytized Aβ after incubation with brain slices of 5×FAD transgenic mice, as determined by intracellular flow cytometric staining of Aβ (Fig 4A and 4B). Interestingly, reducing Peli1 expression further promoted the phagocytic ability of Aβ from 5×FAD brain slices in both BV2 and primary microglial cells, as reflected by enhanced Aβ MFI and increased frequencies of Aβ + cells (Fig 4A-4F).
To confirm the in vivo Peli1 function in modulating microglial phagocytosis-induced Aβ clearance, we generated WT and Peli1 −/− mice with the 5×FAD background by crossing Peli1 −/− mice with 5×FAD transgenic mice (S6A and S6B Fig), and then analyzed the Aβ deposition in the brains of Peli1 −/− 5×FAD mice in comparison with that of aged-matched Peli1 +/+ 5×FAD mice. Interestingly, we found that Peli1 deficiency does not affect the production of amyloid precursor protein (APP) in the brains of both 6-month-old and 10-month-old 5×FAD mice (S6C-S6E Fig). In 10-month-old mice, we observed a large number of Aβ were deposited in the brains of aged Peli1 +/+ −5×FAD mice (Fig 4G and 4H). Consistent with the ex vivo phagocytosis data, Peli1 deficiency led to an obvious decrease of the Aβ stained-area along with similar staining of Iba1 + microglia in the aged brain slices as compared to that of WT 5×FAD mice (Fig 4G-4I). In addition, Peli1 deficiency also increased C/EBPβ and CD36 proteins levels in the adult microglia that isolated from the brains with 5×FAD background (Fig  4J and 4K, S6F and S6G Fig). Therefore, these data confirmed the negative role of Peli1 in regulating the C/EBPβ degradation, CD36 expression, Aβ phagocytosis and Aβ clearance in 5×FAD mice. The critical role of Peli1 in microglia to curtail Aβ phagocytosis promoted us to ask an intriguing question that whether Peli1 expression was modulated during AD pathogenesis. We then isolated the CD11b + microglia from the brains of age-and sex-matched 8-month-old adult naive and 5×FAD mice and examined the Peli1 expression. Interestingly, the protein levels of Peli1 were elevated in the adult microglia of aged 5×FAD mice when compared with aged naive mice (Fig 4L and 4M) or in the human AD brain when compared with non-AD brain (S6H Fig), suggesting Peli1 was increased in the brain during AD pathogenesis. Accordingly, the expression of C/EBPβ and CD36 were decreased in the adult microglia of 5×FAD mice when compared with aged naive mice (S6I-S6M Fig). Together, these results suggested that the AD brain niche was a potential inducer of microglial Peli1 protein, which may, in turn, have aggravated the disease severity through regulating C/EBPβ degradation and impairing CD36 expression and microglial Aβ clearance.

Discussion
Microglia are critical to maintain the brain homeostasis of healthy individual. Under pathological conditions, microglia functions as a damage sensor to clear cell debris and misfolded or aggregated proteins like Aβ through its abundantly expressed surface scavenger receptors [29,30]. Two scavenger receptors, SRA and CD36, are highly expressed on microglial surface and have been identified to critically regulate microglia-mediated Aβ phagocytosis and clearance [29,30,34]. In the present study, we identified Peli1, an E3 ubiquitin ligase that abundantly expressed in microglia [26], as a negative regulator of CD36 through mediating the ubiquitination and degradation of C/EBPβ. As a consequence, loss of Peli1 increased microglial CD36 expression and thus, promoted in vivo Aβ clearance of AD-like mice.
Published studies have suggested that Peli1 exhibited nonredundant function in microglia [26,27] and have established TLRs signaling as a key pathway for Peli1 induction [26,42,43]. We have previously demonstrated that microglial Peli1 is a pathogenic factor to promote neuroinflammation through mediating toll-like receptors (TLRs) signaling, which, in turn, contributed to the enhanced Peli1 expression in microglia during disease pathogenesis [26]. Other groups have also suggested that TLRs and the downstream kinase TANK-binding kinase 1 (TBK1)/inhibitor of kappa B kinase epsilon (IKKε) are critical for Peli1 induction and activation in macrophages [43,44]. It is known that Aβ could interact with TLRs and then activate the downstream signaling, and the debris from the dead neural cells in AD brains are also strong inducers of TLRs signaling [35,45,46]. Therefore, it is not surprising that microglial Peli1 is strongly induced in AD brains, and it is also reasonable to speculate that the Peli1 induction is mediated through the signaling transduced by TLRs that highly expressed on the surface of microglia during AD pathogenesis.
Microglial phagocytosis may be a double-edged sword in AD. On one hand, reduced microglial phagocytosis and suppressed clearance of apoptotic neurons and protein aggregates may result in the proinflammatory activation of microglia [47,48]. However, on the other side, inappropriate phagocytosis of neurons and synapses in neurodegeneration has been postulated, which contributes to the neuropathology of AD [49][50][51]. In the present study, we found that Peli1 deficiency did not affect microglial phagocytic ability of apoptotic neuron cells (S1H and S1I Fig). A plausible explanation is that different scavenger receptors are involved in the phagocytosis of different substrates [31,52]. In addition, although phagocytosis of Aβ has been implicated in the spreading of toxic oligomers [53], our data showed that no spreading toxic oligomers were observed in the brain of Peli1-deficient mice with 5×FAD background (Fig  4G). These data collectively suggested that Peli1 may act specifically to regulate microglial phagocytic ability to Aβ.
In summary, our study established the E3 ligase Peli1 as a crucial regulator of microglial Aβ phagocytosis in AD. During AD pathogenesis, Peli1 expression was elevated, presumably through the activation of TLRs signaling in microglia, which increased Peli1 protein associated with the transcription factor C/EBPβ and functioned as its direct E3 ubiquitin ligase to mediate the ubiquitination and degradation of C/EBPβ. Consequently, decreased C/EBPβ impaired the transcription of the gene encoding scavenger receptor CD36 and further suppressed the phagocytic ability to Aβ of microglia (Fig 4N). Considering the essential pathogenic role of Peli1 in driving microglia-mediated neurodegeneration like multiple sclerosis and AD, we proposed that microglial Peli1 may be a critical novel therapeutic target that have clinical potential for the neurodegenerative diseases.

Ethics statement
The neuropathology studies on human brain samples were approved by the Ethical Committee from the University of Verona and Azienda Ospedaliera Universitaria Integrata di Verona

Mice
Peli1 +/+ and Peli1 −/− mice were generated by breeding Peli1 +/− mice as previously described [37]. The 5×FAD transgenic mice were purchased from Shanghai Model Organisms Center. In some experiment, Peli1 −/− mice were crossed with 5×FAD transgenic mice to generate Peli1 +/+ and Peli1 −/− mice on 5×FAD background, and the age-matched male mice (10-month-old) were used to examine the Aβ deposition in the brains. All mice were maintained on a C57BL/6 background in a specific pathogen-free (SPF) facility. Mice were sacrificed by cervical dislocation or as indicated for animal experiments.

Viral infection for gene knockdown or reconstitution
BV2 cells were maintained in RPMI 1640 medium (HyClone) containing 10% fetal bovine serum (FBS, Gibco). For gene knockdown experiment, BV2 cells were infected with lentivirus containing shRNA sequence targeting mouse Peli1 that also encoded a puromycin resistance marker or targeting mouse Cebpb or Cebpa that expressed a GFP for flow cytometric sorting. For gene reconstitution experiments, lentivirus particles containing the plasmids encoding mouse FL Peli1 orPeli1ΔC, which also expressed a GFP marker for flow cytometric sorting, were infected into the Peli1-knockdown BV2 cells. Eight hours after infection, virus-containing media were removed and replaced with fresh media, and the cells were selected by puromycin or sorted by flow cytometry 72 hours after infection. The puromycin selected cells or GFP + cells were then applied for phagocytosis assay, immunoblotting, or qPCR analysis.

siRNA transfection
siRNA oligonucleotides were purchased from GenePharma (Shanghai) and transfected into mouse primary microglia or BV2 cells with Lipofectamine RNAiMAX reagent (13778, Invitrogen) following the manufacturer's protocol. In brief, siRNA was dissolved with DEPCtreated distilled water, 0.02 μM siRNA oligonucleotide and 3 μl RNAiMAX reagent were diluted with Opti-ME medium (31985062, Thermo Fisher), respectively, and incubated at room temperature for 5 minutes and then mixed (1:1) thoroughly. For the gene knockdown in primary microglia, the siRNA-RNAiMAX complex was added to the mixed glial cultures, and then the microglial cells were collected for further analysis. For the gene knockdown in BV2 cells, the siRNA-RNAiMAX complex was added to the cell carefully and cultured in cell incubator for 48 hours at 37˚C, and the siRNA-transfected cells were then applied for further experiment.

Immunoblotting and immunoprecipitation
Cells were lysed in RIPA buffer with proteinase inhibitors cocktail (B14001, Bimake), and the lysates were boiled with SDS loading buffer for 5 minutes. The immunoblotting was assessed with SDS-PAGE, followed by blotting with specific primary antibodies as well as horseradish peroxidase (HRP)-conjugated secondary antibodies. For immunoprecipitation (IP), the indicated antibody was added into the cell lysates and incubated at 4 ºC on shaker overnight after preclear, followed by pulling down the desired protein with Protein A/G magnetic beads (HY-K0202, Med Chem Express). The precipitated proteins were boiled with SDS loading buffer and assessed by immunoblotting.

Immunofluorescence
For Peli1 and C/EBPβ colocalization assay, the murine primary microglia and BV2 cells were fixed with 4% paraformaldehyde and permeabilized with Triton X-100, followed by blocking with bovine serum albumin (Roche) at room temperature for 1 hour and incubating with specific primary antibodies at 4 ºC overnight. After washing with PBS for 3 times, the secondary antibodies labeled with different fluorescence were added and incubated for 2 hours, and the fluorescent images were assessed by confocal microscopy (ZEISS Cell Observer). For brain Aβ deposition detection, the brain tissues were fixed with 4% paraformaldehyde overnight and dehydrated with PBS buffer containing 30% saccharose. After blocking, the brain slices were incubated with anti-Aβ, anti-APP, or anti-Iba1 primary antibodies followed by fluorescent labeled secondary antibodies, and images were assessed by microscope (ZEISS Axio Imager A2).
For microscopic analysis of the phagocytosis for Aβ  in primary microglia and BV2 cells, the Rhodamine B-labeled oligomeric Aβ 1-42 (2 μM) were added into the plated cells and incubated for indicated time at 37 ºC in cell incubator. The phagocytic cells were fixed with 4% paraformaldehyde at 4 ºC for 30 minutes, followed by imaging assessed with microscope (ZEISS Axio Imager A1). The integrated density of images was quantified by Image J software (NIH, https://imagej.nih.gov/ij/).
For human brain samples, paraffinated brain autopsy samples from 3 sporadic human AD patients and 3 aged-matched controls were obtained from the MRC London Brain Bank for Neurodegenerative Disease and used for immunofluorescence staining. Temporal cortex sections of male subjects were pretreated with Dako Target Retrevial Solution (pH 9, Dako, S2368) in a water bath at 95˚C for 30 minutes. Brain slices were incubated in blocking buffer for 1 hour at room temperature then with anti-PELI1 antibody overnight at 4˚C. After washing, appropriate fluorophore-conjugated secondary antibodies were added and then were treated with Sudan B Black 0.1% in ethanol 70% for 20 minutes. Nuclei were stained with DAPI. The sections were mounted with Dako medium. Images were acquired with Axio Imager Z2 (ZEISS, Germany).

ChIP-qPCR assay
ChIP assay was operated under the manufacturer's protocol (EZ-ChIP kit, 17-371RF, Millipore). In brief, primary microglia, adult microglia, or BV2 cells were fixed with 1% formaldehyde (F1635-500ML, Sigma-Aldrich) on the rotator for 10 minutes at room temperature and then quenched with 125 mM glycine on ice for another 10 minutes. After washing with PBS twice, cells were lysed with SDS lysis buffer, then cell lysates were sonicated with Covaris E220 for 10 minutes and precleared with Protein A agarose (16-157, Millipore). The sonicated and precleared cell lysates were immunoprecipitated with anti-C/EBPβ antibody (sc-150, Santa Cruz) or isotype IgG as control. After washing with buffers, chromatin was eluted from the Protein A/antibody/protein/DNA complex and reversed crosslinks with 5 M NaCl for 4 hours at 65 ºC. The freed DNA was digested with proteinase K and RNase A and purified with Phenol-Chloroform-Isoamyl Alcohol Mixture (Sangon Biotech). The DNA was then subjected to qPCR with SYBR Green master mix (04913914001, Roche). Sequences of ChIP-qPCR primers on mouse Cd36 promoter were listed on S1 Table. Flow cytometry Cell suspensions of isolated primary microglia, adult microglia, or BV2 cells were stained with the indicated antibodies or challenged with fluorescent microspheres and then subjected to flow cytometric analysis by using a Beckman Gallios flow cytometer as described before [55]. For the intracellular staining of Peli1, C/EBPβ, and Aβ, the cells were fixed and permeabilized by fixation/permeabilization buffer (00-5223-56 and 00-5123-43, Thermo Fisher) before staining the specific primary antibodies and followed by fluorescent-labeled secondary antibodies staining, and then detected by flow cytometer.

Culture and isolation of mouse primary microglia
Primary microglia were cultured and isolated as previously described [56]. Briefly, the brains of newborn mice were dissociated with 0.25% trypsin after removing the meninges, cells were filtered using a 70-μm mesh and then cultured in complete DMEM/Ham's F-12 medium (DF-12, HyClone). After 2 weeks of cultivation with medium changes every 3 days, the adherent astrocytes were removed through mild trypsinization for 30 minutes, and the remaining microglia were collected and were used for further experiments. The purity of the isolated microglia was determined by flow cytometry as more than 97%.

Isolation of mouse adult microglia from brains
Adult mouse microglia were isolated as previously described [57]. In brief, 12-week-old mice were euthanized and perfused with 20 mL cold PBS. Brains were then rinsed, minced, and digested at 37˚C with collagenase IV (0.5 mg/ml; Gibco) and DNase I (10 μg/ml; Roche) in RPMI 1640. Cell suspension was layered onto a Percoll (Amersham) density gradient and isolated by collection of the interface fraction between 37% and 70% Percoll according to the manufacturer's instructions. After intensive washing and cell separation using CD11b MicroBeads (130-049-601, Miltenyi Biotec), the isolated CD11b + adult microglia were then applied for further analysis by flow cytometry using specific antibodies. CD11b + adult microglia could also be isolated by Adult Brain Dissociation Kit (130-107-677, Miltenyi Biotec) companied with MicroBeads separation according to the manufacturer's procedure.

In vitro phagocytosis assay
Mouse primary microglia or BV2 cells were cultured in 12-well plates at a density of 1×10 5 cells/well in RPMI 1640 with 10% FBS, and the microspheres (10 microspheres per cell; 1 μm, internally dyed with Nile red; F8819, Thermo Fisher) or Aβ 1-42 oligomers (2 μM) were added. Cells and microspheres or Aβ 1-42 were incubated together for indicated time points at 37˚C in the cell incubator and were subsequently washed with PBS, followed by fixing with 4% paraformaldehyde for 30 minutes. The cells were then applied for flow cytometric analysis.
For apoptotic neurons phagocytosis assay, the protocol was modified as described before [58,59]. In brief, the N2A cells were digested by 0.25% trypsin and harvested by centrifugation. Collected cells were suspended with DMEM medium and incubated at 60˚C for 1 hour. The N2A cells were then incubated with fixable violet dye to stain the apoptotic cells, followed by flow cytometry to make sure that the percentage of apoptotic cells was more than 95%. After washed with PBS, the stained N2A cells were added to BV2 cells in 6-well plates with DMEM medium containing 0.5% FBS at approximately 60% confluence. After incubation for 24 hours with light avoiding, the BV2 cells were harvested and stained with CD11b antibody to specifically mark the microglial BV2 cells. Percentage of phagocytic dead N2A cells in BV2 was further analyzed by flow cytometric analysis.

Ex vivo Aβ phagocytosis assay
Ex vivo Aβ phagocytosis assays were performed as previously described [29,30]. Briefly, 10-month-old 5×FAD transgenic mice were euthanized and perfused with 20 ml cold PBS. Brains were removed from these mice and snap-frozen in cold isopentane (Sinopharm). The frozen brains sections (10 μm) were prepared and dried for 2-3 hours in the air. After washing with hybridoma serum-free medium (H-SFM; Gbico) containing 1% FBS, brain sections were cultured with 5×10 5 BV2 cells in 1% FBS H-SFM for 18 hours at 37˚C in the cell incubator. The cells were then washed for 3 times with PBS and digested with 10 mM EDTA-PBS buffer containing 2% FBS and were subsequently assessed the intracellular phagocytic Aβ by flow cytometric analysis.

Ubiquitination assay
In vivo and in vitro ubiquitination assay were performed as previously described [60]. For C/ EBPβ in vivo ubiquitination assay, the BV2 cells or 293T cells that transfected with the indicated expression vectors were treated with MG132 for 4 hours and then were lysed with lysis buffer containing protease inhibitors and N-ethylmaleimide (Sigma). The cell extracts were boiled for 5 minutes in the presence of 1% SDS to dissociate the C/EBPβ-interacting proteins and then were diluted with lysis buffer until the concentration of SDS was 0.1% before immunoprecipitation. C/EBPβ was then immunoprecipitated from the cell extracts and was assessed the status of ubiquitination with anti-ubiquitin antibody. For C/EBPβ in vitro ubiquitination assay, the HA-Peli1, HA-Peli1ΔC, and C/EBPβ were translated in vitro by using cell-free quick coupled transcription/translation system (L1170, Promega). The ubiquitination reaction was processed by mixing the translated proteins with precharged ubiquitin-conjugating enzyme E2 UbcH5a (E2-800, Boston Biochem) and incubating at 37˚C for 4 hours according to the manufacturer's protocol. The reaction was then terminated by boiling for 5 minutes in SDS loading buffer, and the samples were subjected to SDS-PAGE and immunoblotting to detect the ubiquitination of C/EBPβ mediated by Peli1.