Pon1 Deficiency Promotes Trem2 Pathway–Mediated Microglial Phagocytosis and Inhibits Pro-inflammatory Cytokines Release In Vitro and In Vivo

Paraoxonase 1 (PON1) plays an anti-inflammatory role in the cardiovascular system. Levels of serum PON1 and polymorphisms in this gene were linked to Alzheimer’s disease (AD) and Parkinson disease (PD), but its function in the neuroimmune system and AD is not clear. To address this issue, we used Pon1 knockout rats previously generated by our lab to investigate the role of Pon1 in microglia. Knockout of Pon1 in rat brain tissues protected against LPS-induced microglia activation. Pon1 deficiency in rat primary microglia increased Trem2 (triggering receptor expressed in myeloid cells 2) expression, phagocytosis, and IL-10 (M2-phenotype marker) release, but decreased production of pro-inflammatory cytokines such as IL-1β, IL-6, and IL-18 especially TNF-α (M1-phenotype markers) induced by LPS. Pon1 deficiency in rat primary microglia activated Trem2 pathway but decreased LPS-induced ERK activation. The phagocytosis-promoting effect of Pon1 knockout could be reversed by administration of recombinant PON1 protein. The interaction between PON1 and TREM2 was verified by co-immunoprecipitation (co-IP) using rat brain tissues or over-expressed BV2 cell lysates, which might be involved in lysosomal localization of TREM2. Furthermore, Pon1 knockout also enhanced microglial phagocytosis and clearance of exogenous Aβ by an intrahippocampal injection and decrease the transcription of cytokines such as IL-1β, IL-6, and TNF-α in vivo. These results suggest that Pon1 knockout facilitates microglial phagocytosis and inhibits the production of proinflammatory cytokines both in vivo and in vitro, in which the interaction between Pon1 and Trem2 may be involved. These findings provide novel insights into the role of PON1 in neuroinflammation and highlight TREM2 as a potential target for Alzheimer’s disease therapy.


APOE Apolipoprotein E TNF-α
Tumor necrosis factor-α IL-1β Interleukin-1 β IL- 6 Interleukin-6 IL- 12 Interleukin-12 IL- 18 Interleukin-18 IL- 10 Interleukin-10 iNOS Inducible nitric oxide synthase co-IP Co-immunoprecipitation KO Knockout SDS-PAGE Dodecyl sulfate, sodium salt -polyacrylamide gel electrophoresis HRP Horseradish peroxidase RT-PCR Reverse transcription-polymerase chain reaction DSS Disuccinimidyl suberate Background PON1 is a member of the multi-gene paraoxonase (PON) family comprised of three members, PON1, PON2, and PON3, located on chromosome 7 in humans [1]. Human PON1 is an esterase that displays both paraoxonase and arylesterase activities for hydrolyzing and detoxifying toxic organophosphorus compounds, interaction with various drugs, and antioxidant activity by hydrolyzing oxidized phospholipids in high-density lipoprotein (HDL) and lowdensity lipoprotein (LDL) [2][3][4][5][6]. PON1 expression can be modulated by diet, lifestyle, and pharmaceuticals [7], but also by genetic polymorphisms in the promoter and coding regions [8][9][10]. PON1 is now considered to be a major antioxidant factor for eliminating oxidized HDL [5]. Levels of serum PON1 and polymorphisms in the PON1 gene were linked to atherosclerosis, coronary artery disease, and stroke [11][12][13][14]. PON1 is synthesized mainly in the liver and secreted into the blood where it associates predominantly with HDL [2]. However, immunohistochemical studies and RT-PCR have revealed the presence of PON1 in a wide variety of tissues including the brain, spinal cord, heart, and kidney in humans [15], mice [16], and rats [17]. Whether this is due to local synthesis or to PON1 being transported from the liver to other tissues by HDL is unclear. Although PON1 is present in brain tissue, its role in neurodegenerative diseases is controversial. Many studies on Parkinson's disease (PD) suggest that the MM55 PON1 genotype is a susceptibility marker for PD and is associated with exposure to organophosphates in the environment [18]. Neither L55M nor Q192R PON1 shows association with the pathogenesis of PD, however [19]. In dementias, associations with beta-amyloid levels, senile plaque accumulation, and cholinesterase activity suggest an involvement of the PON1 gene in the pathogenesis of AD and response to treatment [20]. AD and other dementias are associated with decreased PON1 activity [21,22], but the majority of meta-analyses failed to find any association between PON1 polymorphisms and the development of neurodegenerative diseases such as AD and PD [23,24]. PON1 might be a potential player in the pathogenesis of neurological disorders, but the precise pathogenic links and the mechanisms are still unclear.
In previous studies, we established a Pon1 knockout (KO) rat model and found that Pon1 KO impaired T cell development in the thymus [25], suggesting that PON1 might be a novel immune regulator. As the resident macrophages of the central nervous system (CNS), microglia strongly influence the pathological response to a stressor by cordoning off brain lesions, phagocytizing cellular debris, and releasing cytokines, chemokines, and growth factors [26,27]. However, the role of PON1 in microglial phagocytosis is unknown. In the present study, we demonstrated that PON1 inhibited microglial phagocytosis by down-regulation of triggering receptor expressed in myeloid cells 2 (TREM2). Knockout of Pon1 negatively regulated the production of pro-inflammatory cytokines induced by LPS or Aβ but facilitated Trem2-dependent microglial phagocytosis in vivo and in vitro.

Human Brain Samples and Animals
Human brain tissues were obtained from the Human Brain Bank as previously described [28]. Pon1 −/− rats were generated previously in our lab [25]. Knockout (KO) rats were genotyped by PCR using the primers, 5′ AAG CGG GTG CTG AAG ACT and 5 ′ACT GCT GGC TCC TTC TCA . A 521-bp fragment of WT and a 179-bp fragment of the Pon1 KO gene were amplified with 30

Survival Analysis
The cumulative percent mortality was calculated every 12 h for LPS-treated rats during the 48-h period after injection.
Survival was determined at the point where no further loss of animals occurred. Kaplan-Meier curves were generated using GraphPad Prism 7.

Histochemical Staining and Analysis
Rats were anesthetized with pentobarbital, and the brains were carefully dissected out. Routinely, brain tissues were fixed in 10% formalin and embedded in paraffin blocks. After heat-mediated antigen retrieval with Tris/EDTA buffer pH 9.0 was performed, sections were stained with antibody to the microglia marker, ionized calcium-binding adaptor protein-1 (Iba1) (1:200, ab178847, abcam). Stained sections were scanned using a Panoramic III scanner (3D Histech, Hungary), and digital images were obtained. Images were analyzed using Image J software according to the protocol reported previously by Young and Morrison [29]. The sections were evaluated by an observer blinded to the rat genotypes.

Serum
At the indicated time points after LPS injection, blood samples were taken and allowed to clot for 30 min at room temperature. Serum was obtained after centrifugation at 1500 × g for 10 min at 4 °C.

Cell Culture and Treatment
Microglial cells were isolated from brains of 1-3-day-old neonatal WT or Pon1 −/− rats according to a previously described procedure with modifications [30]. Briefly, after enzymatic digestion and mechanical dissociation, mixed glial suspensions were filtered through a 70-μm cell strainer, and cells were collected by centrifugation, resuspended in DMEM/F12 supplemented with 10% FBS (Gibco), and cultured in 75-cm 2 flasks (Nunc, Roskilde, Denmark) in a humidified atmosphere of 5% CO 2 /95% air at 37 °C. After 7-10 days, the top layer microglia cells were obtained by shaking off the astrocytic layer of the glial cultures and harvested by centrifugation and seeded onto 6-or 12-well plates for further experiments. Microglia were treated with LPS (Sigma) at a concentration of 200 ng/mL or recombinant human PON1 protein that was produced with True-ORF clone, RC210356 in human embryonic kidney cells HEK293T (TP310356, Origene). Control cells were treated with DPBS. Cells for immunofluorescence analysis were seeded onto glass coverslips in 12-well plates. Cells for RT-PCR and western blot were seeded onto 6-well plates.

Cytokine and Chemokine Measurement
Rat serum was collected, and the level of IL-10 and TNF-α in serum was determined using ELISA kits (R1000 and RTA00, R&D Systems). The minimum detected dose is typically less than 10 pg/ml. The intra-assay CV value was less than 5.5%, and inter-assay CV value was less than 9.9%. Cell culture supernatants were obtained by centrifugation, and cytokines were detected with a 23-plex rat cytokine panel (Luminex).

Western Immunoblotting
Total protein lysates from rat brain tissues or cultured cells were prepared as previously described [28]. After SDS-PAGE, the proteins were transferred to nitrocellulose (Millipore), and the membranes were incubated overnight with antibodies against PON1 (1:500, ab24261, abcam) (antibody against p-DAP12 was designed and prepared by Genscript in USA, and other antibodies used are listed in Table S1). After incubation with the appropriate secondary antibody for 1 h at room temperature, antibody binding was detected with an HRP-conjugated immunoglobulin G (Santa Cruz) using a chemiluminescence detection system (Santa Cruz). For quantitative analysis, the PON1 level was normalized to GAPDH using Image J software.

RNA Isolation and Quantitative RT-PCR
Total RNA was extracted using Trizol (Invitrogen, UK) and treated with RNase-free DNase I to remove any contaminating genomic DNA. First-strand cDNA was synthesized from 2 ug of total RNA using random hexamer primers according to the Superscript III reverse transcriptase manufacturer's protocol (Invitrogen, USA). Detection of PON1 and TREM2 mRNA was carried out by RT-PCR, using GAPDH for normalization. The primers for human PON1 were 5'-TTG GGT TTA GCG TGG TCG TAT-3' and 5'-TCC AAC CCA AAG GTC TCC TG -3'; for human GAPDH 5'-AAC GGA TTT GGT CGT ATT G-3' and 5'-GCT CCT GGA AGA TGG TGA T'. The primers for rat Pon1 were 5'-TAA AGG AAT CGA AGC GGG TGC-3' and 5'-CGG TGG ACG AGG AGT CTG G-3'; for rat Gapdh 5'-TAT CGG ACG CCT GGT TAC -3' and 5'-TGC TGA CAA TCT TGA GGG A-3'. For rat Trem2: r-Trem2-F 5'-CTC TCC ACG TGT TTG TCC TGT-3'; r-Trem2-R 5'-TCA TCT GTG ATG ACC GTG CT-3'. The primer sequences of seven cytokines were listed in supporting Table S2. Twenty-four cycles of RT-PCR were performed using the following conditions: denaturation at 94 °C for 30 s; annealing for rat Pon1 and Trem2 or human PON1 and TREM2 at 60 °C for 30 s; annealing at 55 °C for GAPDH for 30 s; and extension at 72 °C for 30 s.

RNA Sequencing
RNA sequencing was done by Novogene (Beijing, China). Briefly, total RNA was isolated from rat primary microglia treated with or without LPS, and RNA-Seq libraries were prepared by standard protocols. After cluster generation, the libraries were sequenced on an Illumina Hiseq platform, and 125 bp/150 bp paired-end reads were generated. Differential gene expression analysis was performed using the DESeq2 R package (1.16.1). The p values were adjusted using the Benjamini and Hochberg method, and p < 0.01 was the criterion for differential expression. Gene ontology (GO) enrichment analysis was performed by function "sota" in the "clValid" package, and GO terms with p < 0.01 were considered significantly enriched by differential expressed genes.

Immunofluorescence
Primary microglia were seeded on coverslips overnight and then fixed using 4% paraformaldehyde or methanol (Trem2 and Lamp1). Then the cells were permeabilized in 0.5% Triton X-100 for 5-10 min and then blocked for 30 min in normal goat serum (ZSGB, Beijing, China) at room temperature. Immunofluorescence staining was performed using the antibodies listed in Table S1 and appropriate secondary antibodies. Images were captured using a laser scanning confocal microscope (SPF Leica) and processed with Photoshop version 7 (Adobe Systems Inc.).

Co-immunoprecipitation Assays
To determine the interaction between PON1 and TREM2, co-IP assays for both proteins from gene-transfected cells and endogenous proteins were performed.
For endogenous proteins, total brain lysates were prepared with the same IP buffer as above. One aliquot was used for IP of the target protein, while the other served as the input control. PON1 antibody (1:100, ab126597, abcam) was added to the lysate and incubated overnight at 4 °C on a tube roller-mixer at low speed. The following day, 50 μL of protein A/G agarose (Pierce 20,241, Thermo) was added to the lysates and incubated at room temperature for 2 h. The precipitates were collected and analyzed as above.

Phagocytosis
Microglial phagocytosis of TAMRA labeled beta-amyloid (AS-60488, 1 mM, Anaspec) and FITC-dextran (53,557-1G, 1 mg/ml, Sigma) was quantified by confocal microscopy after incubation of microglial cells at 37 °C. At the indicated time points, the Aβ-or dextran-containing medium was removed and cells were washed, fixed with 4% PFA, and incubated with DAPI (Thermo Fisher) for staining the nuclei.

Disuccinimidyl Suberate Crosslinking
Binding of Pon1 and ApoE was examined by cross-linking. The crosslinking reagent disuccinimidyl suberate (DSS) was added at a concentration of 5 mM to mechanically fractured rat brain tissues. After 30 min at room temperature, a quench solution was added to a final concentration of 20 mM Tris (pH 7.5) and incubated for 15 min at room temperature. SDS-sample buffer was added to the reaction mixture and heated at 90 °C for 5 min. Western blots were exposed to the corresponding antibodies for the detection of Pon1, ApoE, and Trem2 (Table S1).

Stereotactic Injections
Before Aβ1-42 (ChinaPeptides, Shanghai) was used, it was firstly incubated in sterile saline at 37 °C for 2 days to allow the formation of different-sized oligomers. Three-monthold WT and Pon1 −/− rats, weighing 250-300 g, were randomly divided into four groups: 0-day, 3-day, 7-day, and 14-day models. The rats were anesthetized and maintained with 1.5-2% isoflurane in oxygen, mounted in a stereotactic frame (RWD, Shenzhen, China), and injected with 4ug/1ul Aβ1-42 into each side of the hippocampus at coordinates (AP: − 3.5 mm, ML: 2.0 mm, DV: 3.0 mm) using a Nanoliter injector 2000 (World Precision Instruments Inc., Sarasota, FL, USA). The injection was performed within 5 min and following the injection, the needle remained in the target location for 5 min before it was gradually withdrawn.

Statistical Analyses
All experiments were performed at least three times, and all samples were tested in triplicate. The data are shown as the mean ± S.D. unless otherwise noted. Student's t test was used for statistical analysis when only two groups were tested. Two-way analysis of variance (ANOVA) and Tukey's multiple comparisons tests were used to compare multiple groups. Log-rank (Mantel-Cox) test was used for comparison of survival curves. In all cases, a two-tailed p value less than 0.05 was considered statistically significant. Statistical analyses were performed using Graph Pad Prism version 7.0 (Graph Pad Software, La Jolla, CA, USA).

PON1 Is Expressed in Both Human and Rat Brain Tissues and Rat Primary Microglia
Given the important function of PON1 in macrophages [31], we proposed that PON1 might play a role in brain macrophages and microglia. PON1 mRNA was detected in both human and rat brain tissues, and the level in rat liver was approximately 3.8 times higher than in rat brain (Fig. 1A, B). The absence of PON1 protein in Pon1 KO rats was confirmed by western blot of total proteins from brain tissues and primary microglia (Fig. 1C, D). PON1 was also detected in HM1900 and BV2 microglia cell Rat tissues  The expression of PON1 protein in liver and brain tissues (C) from WT rats (n = 3) and Pon1 −/− rats (n = 3) or in primary microglia (D) isolated from 3-day-old WT rats (n = 3) and Pon1 −/− rats (n = 3) was compared by western blot. Double immunofluorescence staining of microglia using antibody against Pon1 (green) and Iba1 (red) were performed on brain sections (E) of WT rats (n = 4, magnification × 200, scale bar = 100 μm, arrows indicate co-localization) or in primary microglia (F) isolated from WT and Pon1 −/− rat (n = 4 each group, magnification × 630, scale bar = 50 μm). G Double immunofluorescence staining of Pon1 (green) and Map2 (red) or of Pon1 (green) and Gfap (red) were done in primary neurons or astrocytes from WT rats (n = 6, magnification × 630, scale bar = 25 μm). The nuclei were stained using DAPI (blue). Student's t test was used for statistical analysis and a two-tailed P value < 0.05 was considered statistically significant. * p < 0.05 lines (Fig. S1). To identify the location of Pon1 protein in microglia, we stained brain tissue sections from WT rats at 2 months of age. Immunofluorescence analysis indicated that Pon1 protein was expressed in a small number of Iba1-positive microglia (Fig. 1E). Unlike the results from brain tissue sections, immunofluorescence staining of primary microglia from neonatal rats showed Pon1 colocalized with Iba1 in almost all microglia. Pon1 immunostaining was not observed in microglia from Pon1 −/− rat (Fig. 1F). Pon1 also co-localized with Map2 in cultured neurons and with Gfap in cultured astrocytes (Fig. 1G).

Pon1 Deficiency Reduces LPS Lethality
Exposure to bacterial lipopolysaccharide (LPS) causes death in rodents [32][33][34]. Intraperitoneal injection of 20 mg/kg LPS into WT rats resulted in death of 100%  Fig. 2A). The appearance of WT and Pon1 −/− rats at 1 h and 6 h after injection of 5 mg/kg LPS is shown in Fig. 2B. TNF-α is considered to be the cytokine responsible for LPS lethality [32][33][34]. So, the level of TNF-α in the peripheral blood of PON1 −/− rats (n = 4) and WT rats (n = 4) was determined up to 3 h after injection of a lethal dose of LPS (20 mg/kg), and the results indicated that both WT and KO rats showed an increase in TNF-α starting at 30 min after injection, peaking at 90 min and declining between 120 and 180 min (Fig. 2C). TNF-α levels in Pon1 −/− rats declined significantly faster than in WT rats (p < 0.01, 2-way ANOVA), indicating that the absence of PON1 protects against the toxic effects of LPS in rats possibly by reducing TNF-α level in serum. As showed in Fig. 2C, the difference between WT and Pon1 −/− rats with LPS treatment is obvious or significant in the last timepoint (at 180 min), and further, we paid attention to the release of pro-inflammation cytokines at late acute phase, so we choose a timepoint longer than 3 h and a low dose to measure the alteration of cytokine release. ELISA assays were performed, and the results confirmed that the absence of Pon1 reduced the increase of TNF-α level (Fig. 2D, p < 0.05, n = 4/group) and increased the release of IL-10 ( Fig. 2E, p < 0.05, n = 4/ group) induced by LPS (5 mg/kg, 6 h).

Pon1 Deficiency Inhibits LPS-Induced Microglial Activation In Vivo
Pon1 inhibits monocyte-to-macrophage differentiation, and Pon1 KO caused obvious morphological changes of macrophages in mice [35]. Iba1 immunohistochemistry were performed, and the results showed that the microglia in brain sections from Pon1 −/− rats were larger (Fig. 3A-C) and had fewer and shorter branches (Fig. 3D, E) than those from WT rats without LPS treatment. The microglia from WT rats with LPS treatment showed M1-like activation and polarization with amoeboid shape with no long branches (Fig. 3A, B). In contrast, the microglia from Pon1 −/− rats still had larger cell bodies and longer process lengths than those from WT rats after LPS stimulation.

Pon1 Deficiency in Rat Microglia Decreased LPS-Induced Cytokine Levels
Rat primary microglia cells were isolated to explore the effect of Pon1 loss on cytokine release in vitro. Microglia cells were identified by Iba1 or Cd11b immunofluorescence staining (Fig. S2). To determine whether Pon1 KO altered cytokine release from microglia, we quantified the levels of IL-1β, IL-5, IL-6, IL-18, TNF-α (M1-phenotype markers), and IL-4, IL-10 (M2-phenotype marker) in cell supernatants following LPS administration (Fig. 4A). The results indicated that IL-1β, IL-6, and TNF-α were all significantly decreased (48.4%, 17.1%, and 26.6% respectively) but in Pon1 −/− microglia compared to WT without LPS treatment. In contrast, a significant 4.5-fold increase of IL-10 was observed in Pon1 −/− microglia compared to WT. Further, the increase of pro-inflammatory IL-5, IL-18, and TNF-α induced by LPS was smaller in Pon1 −/− microglia than in WT, while the increase of anti-inflammatory IL-10 in Pon1 −/− microglia was larger than in WT. Interestingly, here LPS increased not only pro-inflammatory genes but also anti-inflammatory genes expression including IL-4 and IL-10, which is consistent with previous report in rat primary microglia cells [36,37]. On another hand, IL-4 and IL-10 responded to Pon1 KO with LPS treatment or not very differently, suggesting the different role of these two cytokines in microglia responses to Pon1. Additional five cytokines of IL-7, IL-12, MCP-1, VEGF, and G-CSF showed no significant difference between the two groups (Fig. S3). Inducible nitric oxide synthase (iNos), a key inflammatory mediator, was also significantly increased in WT and Pon1 −/− microglia after LPS stimulation; but the increase of iNos in Pon1 −/− microglia was smaller than in WT (Fig. 4B, C). LPS induced iNos expression or IL-1β production by activating ERK, JNK, and p38 in murine macrophages [38] or microglia cells [39,40]. In accordance with the cytokine results, the increase in p-Erk/Erk in Pon1 −/− microglia was less than in WT (Fig. 4D), whereas the p-p38 and p-JNK were not changed after LPS treatment (Fig. S4), which might explain the differences in cytokine release. These data suggest that Pon1 KO resisted to sustained pro-inflammatory activation in rat microglia.

Pon1 Deficiency Enhances Phagocytosis in Rat Primary Microglia
To determine whether Pon1 deficiency affected endocytosis, we measured the internalization of fluorescently labeled Aβ peptide, previously reported to be endocytosed by microglia [41]. The results demonstrated that Pon1 depletion significantly enhanced intracellular levels of fluorescent Aβ (Fig. 5A, B). Consistent with the enhanced uptake, we found a similar effect using fluorescently labeled dextran (Fig. 5C, D). These results show that Pon1 depletion in microglia increased the overall phagocytic activity. A 24-h treatment with 200 ng/ml LPS inhibited the phagocytic ability of WT and Pon1-depleted microglia (Fig. 5A-D)

Pon1 Deficiency and LPS Treatment Changed the Transcriptomic Profiles in Rat Primary Microglia
Twelve microglia samples treated with or without LPS (200 ng/ml) were subjected to RNA-seq. Differential expression analysis demonstrated that Pon1 knockout or in combination with LPS treatment caused the transcriptional alteration of a number of genes associated with nervous system development, neurogenesis, generation of neurons, neuron differentiation, glial cell differentiation, and gliogenesis, etc. (Fig. 6A-C, Table 1). All the genes associated with glial cell differentiation (Fig. 6D, E) and most genes related to nervous system development were downregulated, and only several genes associated with nervous system development were upregulated in Pon1 −/− or Pon1 −/− LPS microglia cells compared with WT or WT LPS microglia cells, respectively (Fig. S5A,  B). Eight genes of Sox10, Olig2, Olig1, Ptprz1, Sox8, Gpr37l1, Reln, and Erbb3 associated with glial cell differentiation were decreased significantly in both Pon1 −/− and Pon1 −/− LPS microglia cells compared with WT or WT LPS microglia cells, respectively (Fig. 6D, E), suggesting that these genes may be involved in Pon1 KO-mediated microglia phenotype alteration. LPS treatment resulted in similar changes in gene profile in Pon1 KO and WT rats ( Fig. S6A-D, fold change > 2, p < 0.01). The genes with the most significant variation were classified in four categories by GO analysis. The two categories of up-regulated genes induced by LPS treatment were associated with the defense response, the response to lipopolysaccharide, the response to molecule of bacterial origin, immune system processes, and multi-organism processes (Fig. S6A brain sections from each rat were used to count the number of Iba 1-labeled microglia and the average cell size was quantified and compared (mean ± SEM, sixty cells from 6 rats for each group). D Quantification of the number of projections and E lengths of projections of microglia were performed using Image J (sixty cells from 6 rats for each group). *p < 0.05, ***p < 0.001 indicates significance; NS indicates no significance. Unpaired t tests and post hoc tests were used for statistical analysis induced by LPS treatment were associated with immune system processes, the dynein complex, leukocyte differentiation, extracellular matrix, chromosome segregation, and the proteinaceous extracellular matrix (Fig. S6C, D, p < 0.01). These results demonstrate that Pon1 KO changed the transcriptome profiles of microglia, which might be partly responsible for the phenotypic changes of microglia from Pon1 KO rats.

Pon1 Deficiency Up-regulated Trem2 Signal in Rat Primary Microglia
To investigate the mechanism underlying Pon1-mediated phagocytosis and cytokine release, we analyzed the protein expression of genes that are known to be involved in LPS response and phagocytosis. As shown in Fig. S7, we found a significantly increased nuclear translocation of P65 after  (Fig. 7A, B) and the phosphorylated tyrosine kinase, p-Syk (Fig. 7A, C), were significantly increased in Pon1 −/− microglia compared to WT with LPS treatment or not, but Trem2 was decreased with LPS treatment in WT but not Pon1 −/− microglia, similar to the results of the phagocytosis assay. Expression of the guanine nucleotide exchange factors, Vav2 (Fig. 7A, D) and Vav3 (Fig. 7A, E), and actin-related protein, Arp2 (Fig. 7A, F), was increased in Pon1 −/− microglia compared to WT, which could be affected by LPS. The data suggested that Pon1 deficiency enhanced phagocytosis by activation of the Trem2 signaling pathway and up-regulation of the actin cytoskeleton. As shown above, Trem2 protein levels were significantly increased in Pon1 −/− microglia compared to WT (Fig. 7A, B) even though Trem2 mRNA was not altered by Pon1 deficiency (Fig. 7G). However, when human PON1 recombinant protein was added to Pon1 −/− microglia, the protein level of Trem2 and p-Dap12 was decreased dependent on the PON1 protein concentration (Fig. 7H-J), suggesting the possible role of Pon1 in Trem2 pathway.

Pon1 Protein Interacts with Trem2 and Promotes Its Lysosome Localization in Microglial Cells
Co-immunoprecipitation (co-IP) was employed to identify the mechanism of TREM2 regulation by PON1. The results confirmed that there was a direct interaction between endogenous Pon1 and Trem2 proteins in rat brain tissues (Fig. 8A, B) and between over-expressed human PON1 and TREM2 proteins in BV2 cells (Fig. 8C, D). Immunofluorescence assays were performed to verify the co-localization of Trem2 and the lysosomal marker Lamp1 in primary microglia. Trem2 showed a highly increased, clustered distribution in Pon1 −/− microglia compared to WT (Fig. 8E, F ). After administration of recombinant human PON1 protein, the distribution of Trem2 was decreased but the co-localization of Trem2 with Lamp1 in both WT and Pon1 −/− microglia were increased (Fig. 8E-G). ApoE is a novel ligand for Trem2 [42], and we used DSS crosslinking to test whether the binding of ApoE and Trem2 was disrupted by Pon1 in rat brain tissues. The results indicated that the crosslinking between ApoE and Trem2 was increased in Pon1 −/− rat brain tissues relative to WT (Fig. 8H-J), suggesting that Pon1 might compete with ApoE for binding to the Trem2 receptor on microglia. Taken together, the results suggested that Pon1 interacts with Trem2 and promotes its localization in lysosomes. Conversely, Pon1 deficiency increases Trem2 and causes clustering and distribution in the microglial cells, which might promote Trem2 mediated phagocytosis.

Pon1 Deficiency Facilitated Microglial Aβ42 Clearance In Vivo
Further, we investigated the effects of Pon1 on microglial Aβ clearance in a rat model of AD produced by an intrahippocampal injection of amyloid-β1-42 (Aβ1-42) (Fig. 9A). The results indicated that Pon1 knockout enhanced the clearance of Aβ1-42 (Fig. 9B, C). Aβ1-42 injected into the hippocampus of WT and Pon1 −/− rats induced microglial activation, proliferation, accumulation, and phagocytosis, and the amount of Aβ1-42 then gradually decreased along with the time increase ( Fig. S8 and Fig. 9C). Pon1 knockout significantly increased the number of microglia cells around the injection sites and Aβ1-42 uptake of single microglia cell, finally reduced the deposit of Aβ42 inside hippocampus neurons (Fig. S8 and Fig. 9D-G). On another hand, Pon1 knockout decreased Aβ-induced increase of cytokines including IL-1β, TNF-α, iNOS, IL-10, and IL-6, and increased the expression of Mrc-1, one receptor mediated endocytosis (Fig. 9H). In contrast with LPS, Aβ1-42 injection increased the transcription of Trem2 in both WT and Pon1 −/− rats along with the time increase, and the protein level of Trem2 in Pon1 −/− rats was still higher than WT rats (14 days post-injection) (Fig. S9). These results suggest that Pon1 knockout with high Trem2 level may facilitate microglial clearance of exogenic Aβ in rat brains.

Discussion
The expression of PON1 in brain has long been controversial. Pon1 immunostaining showed the presence of protein in white matter areas of mouse [16] and rat brains [43], and the Human Protein Atlas program has just detected weak PON1 protein expression in human glioblastoma. PON1 mRNA was also detected in homogenates of the frontal cortex of AD and healthy controls [15], as well as in the frontal cortex of rat brain [17]. In our study, PON1 mRNA was detected in human prefrontal lobe and rat brain tissues, and PON1 protein was expressed in rat brain and microglia. Thus, it is inferred that both local synthesis and HDL delivery from liver contribute to the wide distribution of PON1 in brain [43,44], but its role in brain disease is unclear.
Our previous research revealed that Pon1 KO impaired T cell development in rat thymus [25], suggesting that PON1 functions as a novel immune regulator. In the present study, we found that microglia from Pon1 KO rats were larger in cell body and had fewer and shorter branches (Fig. 3). Pon1 KO reduced LPS-induced microglia activation, promoted phagocytosis (Fig. 5), and inhibited the production of specific cytokines (Fig. 4). These data were similar to previous reports on Pon1 KO mice in which macrophages were larger, contained larger cytosolic compartments, and appeared more granular than control C57BL/6 macrophages and recombinant PON1 inhibited monocyte-to-macrophage differentiation [35]. Interestingly, we found that Pon1 KO protected rats from LPS-induced lethality and decreased the LPS-induced TNF-α release in serum (Fig. 2), these data are similar to the previous data of Van Oosten et al., who showed that apolipoprotein E (ApoE) protection from LPS lethality and inhibited the LPS-induced increase in TNF-α [34]. Thus, it seemed that Pon1 had an opposite effect on LPS endotoxemia and on the release of TNF-α by macrophages or microglia compared to ApoE. ApoE has been suggested as a putative ligand for the recently identified immune receptor, TREM2, which is mainly expressed on microglia and promotes anti-inflammatory responses and phagocytosis [45][46][47]. ApoE binding increased phagocytosis by primary microglia dependent upon TREM2 expression [42]. Our results demonstrated the interaction of Pon1 with Trem2 and the up-regulation of Trem2 by Pon1 KO. Thus, we proposed that Pon1 might affect microglial phagocytosis by competing with ApoE for binding to the Trem2 receptor on microglia. ApoE was previously reported to bind to PON1 with high affinity, and the stabile PON1 complex was involved in anti-atherogenic activity [48]. Although we found that Trem2 was cross-linked with ApoE in cortical homogenates in the absence of Pon1 (Fig. 8H-J), further investigations are needed to elucidate the complicated associations of Pon1, ApoE, and Trem2. Considering that the reported activity of PON1 was affected by the complex formed with HDL, ApoA or amyloid A, and ApoE but not apoA-I is the major apolipoprotein constituent of HDL produced by astrocytes and microglia in the CNS [2,[49][50][51], the possibility of the interaction of TREM2 with PON1-HDL (ApoE) complex or other kind complex cannot be excluded.
In this study, we focused on the role of Pon1 and Pon1 KO-induced microglial phagocytosis. We demonstrated for the first time that the absence of Pon1 in microglia increased Trem2 protein levels, which was reversed by administration of PON1 recombinant protein, in parallel with changes in phagocytic activity. Trem2 inhibited the LPS-induced pro-inflammatory responses and its up-regulation improved cognitive impairment and pathology in a mouse model of AD [52][53][54][55]. Pon1 KO showed similar effects to Trem2 on microglia, promoting phagocytosis and inhibiting the production of certain cytokines. These results suggested that Pon1 regulated the activation of microglia through Trem2 signaling. We verified the interaction between TREM2 and PON1 proteins, and showed that the interaction could promote the internalization of TREM2 into lysosomes possibly for potential degradation as we speculated (Fig. S10), but additional research needs to be done to confirm this.
LPS triggers inflammatory responses in humans and other mammals. Many papers reported that LPS promoted phagocytosis [56][57][58], while some studies showed an inhibitory effect of LPS on phagocytosis [59][60][61]. We demonstrated that LPS significantly enhanced the production of TNF-α in microglia and inhibited Pon1 KO-induced phagocytosis. Previous studies reported that TLR4 activation by LPS impaired the phagocytic capacity of microglia [59]. TNF-α inhibited phagocytosis induced by activated macrophages [60,61], suggesting that the level of TNF-α was negatively correlated with the phagocytic ability of macrophages. LPS or PON1 down-regulated Trem2 (Fig. 6A,H), which might also be responsible for the reduction in phagocytosis seen after exposure to LPS. This result is consistent with a previous report that LPS down-regulated Trem2 mRNA in BV2 cells [62]. Also, as Fig. 5 showed, LPS inhibited the endocytosis of amyloid beta and dextran in WT microglia. Interestingly, the post hoc tests showed that only dextran uptake is inhibited by LPS in Pon1 KO cells, and there was no significant difference between Pon1 −/− and Pon1 −/− LPS groups in terms of amyloid beta application. Dextran was usually used for probing fluid-phase endocytosis of macropinocytic or micropinocytic processes [63], whereas the internalization of Abeta peptides in microglial cells was reported through a scavenger receptor-mediated phagocytosis [41,64]. Thus, the results suggested that Pon1 KO resisted to the inhibition of LPS on Abeta phagocytosis but not dextran pinocytosis. We just observed a small number of Iba1-positive microglia expressed Pon1 protein in rat brain tissues (Fig. 1E). In view of this, we used primary microglia to determine the expression of Pon1 in Iba1-positive microglia. Cell immunofluorescence staining did indicate that Pon1 co-localized with Iba1 in almost all microglia (Fig. 1F). The reasons for the difference may be the complexity of multiparameter immunostaining [65] and the limitations of primary Fig. 8 Interaction between Pon1 and Trem2. Co-immunoprecipitation (Co-IP) was used to determine the interaction between Trem2 and Pon1 in WT rat brain tissues and BV2 cells over-expressing PON1. Western blots showed co-IP of endogenous Pon1 and Trem2 by anti-PON1 antibody (A) or by anti-TREM2 antibody (B) in rat brain tissues (n = 6). Flag-PON1 or empty vector was co-transfected with Myc-TREM2 into 1.0 × 10 7 BV2 cells. The tags were then switched (Flag-TREM2 and Myc-PON1) and the co-transfection repeated. BV2 lysates were immunoprecipitated with anti-FLAG M2 beads and western blots showed co-IP of PON1 and TREM2 (C and D). Next, WT and Pon1 −/− microglia were incubated with 1 ng/ul human PON1 recombinant protein for 15 min, and immunofluorescence staining was performed to determine the distribution of Trem2 and the co-localization of Trem2 and Lamp1 proteins (E). DAPI was used to stain cell nuclei. (4 × zoom of × 630 original magnification, scale bar = 2 μm). Trem2 expression was quantified using Image J (F), and the co-localization Trem2 with Lamp1 was quantified from a number of images using Cell Profiler Analyst Mac (G). DSS crosslinking was performed on mechanically fractured rat brain tissues, and western blotting was used to detect Pon1 (H), ApoE (I), and Trem2 (J) proteins in control and cross-linked samples. The crosslink products between ApoE and Trem2 (I and J) were quantified using Image J. *p < 0.05, **p < 0.01, and ***p < 0.001 indicate significance; NS indicates no significance. Unpaired t tests and post hoc tests were used for statistical analysis microglia model [66]. Thus, we have tried to investigate the effects of Pon1 on microglial Aβ clearance using a rat model (Fig. 9). Fortunately, by and large, the results from in vivo study supported the results from the study in primary microglia model. These results suggested that Pon1 knockout promoted microglial phagocytosis of Aβ in primary microglia and microglial clearance of exogenic Aβ in rat brains.

Conclusion
In conclusion, the present study offers evidence that Pon1 knockout with high Trem2 level promotes microglial phagocytosis and decreased the production of proinflammatory cytokines both in vivo and in vitro, in which the interaction between Pon1 and Trem2 might play an important role. These findings provide insights into the role of PON1 in neuroinflammation and illustrate the potential of TREM2-directed therapeutics in neurodegenerative diseases.
Author Contribution LFZ and LZ designed the study and wrote the paper, LZ, WD, YWM, LB, CXS, XZ, and JWL performed the experiments. All the authors have read and approved the final manuscript.