Amelioration of amyloid-β-induced deficits by DcR3 in an Alzheimer’s disease model

Hemizygous hAPP transgenic mice (line J20) express an alternatively spliced human APP minigene with the Swedish and Indiana familial AD mutations driven by the PDGF promoter [26]. Hemizygous DcR3 transgenic mice express human DcR3 driven from the CD68 promoter in macrophages/microglia/monocytes [25]. Female DcR3 transgene mice were crossed with male APP transgenic mice to obtain wild-type DcR3 single transgenic, APP single transgenic, and APP/DcR3 double transgenic mice. The littermates of these mice were examined in behavioral tests at 6 months of age and sacrificed for pathological examinations at 6 or 12 months of age.

The water maze consisted of a water pool (122 cm in diameter) containing opaque water and a platform (10 cm in diameter) submerged 1 cm below the water surface. The hidden platform test consisted of 10 sessions over 5 days and each session comprised three 60-s trials with 15-min inter-trial intervals. The platform location remained constant during the hidden platform sessions, and the entry points were changed semi-randomly between days. One day after the final day of hidden platform training section, a probe trial was conducted by removing the platform and allowing mice to explore in the pool for 1 min. The quadrant in which the platform was previously located was defined as the target quadrant, and the proportion of time (as a percentage) that the each mouse spent in the target quadrant was used to measure memory retention. The number of platform crossings and swim speed were recorded and analyzed with the EthoVision video tracking system (Version 3.1 Noldus Wageningen, Netherlands).

During the day 1 training section mice were habituated in a conditioning box (Graphic State 2.101 Contents, Coulbourn Instruments, PA, USA) for 5 min and then received five pairs of an 8-s tone and a 2-s shock (0.4 mA) followed by a 2-min resting interval. On day 2 testing sections, the trained mice were placed back to the same testing box, and their freezing time was scored for 5 min to measure the contextual conditioned fear response. The cued test was conducted 5 min after the contextual test. Mice were habituated for 5 min in a novel-shaped box and then exposed to three 10-s auditory cues followed with a 2-min resting interval. The freezing times of each mouse were scored during all testing sessions.

To detect spontaneous locomotor activity mice were placed in an open chamber (24.32 × 24.32 cm²). Their horizontal movement was detected by a 16 × 16 infrared photo-beam arrays placed 1.5 cm above the bottom of the chamber for 15 min (Version 2.0, TRU Scan Photobeam LINC, Coulbourn Instruments, PA, USA).

The elevated plus-shaped maze consisted of two open arms and two closed arms. All mice were individually placed at the center of the maze and allowed to explore for 10 min. The time spent and distances traveled on each arm were calculated with the Etho Vision video tracking system.

Paraformaldehyde-fixed brains were sliced coronally by using a microtome (Leica SM2010R Heidelberg, Germany) and were stored in cryoprotectant medium (30% glycerol, 30% ethylene glycol in PBS) at -20 °C. For immunohistochemistry (IHC) staining, brain slices were blocked in a TBS-buffered solution containing 1% glycine, 0.4% Triton X-100, 10% FBS (FBL01, Caisson labs, USA), 0.1% sodium azide (13412, Sigma, MO, USA) and 3% serum bovine albumin for 2 h and then incubated for 24 h at 4 °C with anti-Iba1 (019–19741, Wako), anti-YM1 (01404, Stem Cell technology), anti-synaptophysin (04–1019, Millipore), anti-MAP2 (MAB378, Millipore) and anti-Aβ (SIG-39320, 6E10, Covance) to measure the distribution of microglia, M2a activated microglia, pre-synaptic density, neuronal density and the total level of Aβ. After incubation, the slices were incubated for 2 h with Alexa594-labeled (111–585–003, Jackson ImmunoResearch) and Alexa488-labeled secondary antibodies (115–546–003, Jackson ImmunoResearch) at room temperature.

For thioflavin-S staining brain slices were incubated with 0.015% thioflavin-S (T1892; Sigma MO, USA) for 15 min at room temperature. All chemicals unless otherwise stated were purchased from Bio Basic Inc. (Canada).

For immunocytochemistry staining primary cells were fixed with 4% paraformaldehyde to measure the degeneration of primary neurons and the morphology changes of microglia in responses to different treatment conditions. Fixed cells were stained with anti-MAP2 or anti-Iba1 antibody to visualize the structure changes. The stained slices or cells were imaged with a fluorescence microscope (Axio Observer A1; Zeiss Germany) or a confocal microscope (Fluoview FV10i; Olympus USA). Images were analyzed with MetaMorph® Microscopy Automation & Image Analysis Software (Molecular Devices, CA, USA).

To quantify the total or M2a microglia surrounding plaques slices were double stained with 6E10 and anti-Iba1 or anti-YM1 antibodies. Plaque areas were circled to determine the centers. The circles were then enlarged 10 μm in radius from the center, which was considered to be the periphery area for measuring the microglia or secreted YM1 coverage.

For DcR3 measurement up to 500 μl of blood was collected from the facial vein at the submandibular area and was centrifuged at 1,000 g for 15 min to isolate the serum. Serum DcR3 concentrations were measured with a human DcR3 Duo Set (DY142, R&D, USA).

For Aβ measurement the hippocampus of each mouse was homogenized in 5 M guanidine/5 mM Tris (pH 8.0) buffer and diluted with 0.25% casein blocking buffer to a final concentration of 0.5 M guanidine with protease inhibitor (04693116001, Roche, Basel, Switzerland). The levels of total Aβ and Aβ42 were quantified using Aβ ELISA kits (27729 and 27711, IBL, Hamburg, Germany).

For cytokines measurement diluted hippocampus lysates and conditioned media were applied to TNF-α, IL-1β, and IL-6 ELISA kits (555268, 559603, 555240, BD System, NJ, USA). For YM1 measurement, the hippocampus was homogenized in diluting reagent provided by ELISA kit at the concentration of 20 μg tissue/μl, and YM1 concentration were measured using mouse YM1/Chitinase 3-like 3 DuoSet ELISA (DY2446, R&D, USA).

The RNA from the hippocampus and the primary microglia were purified using the Total RNA Mini Kit (Geneaid Taiwan) or TRI reagent (T9424, Sigma, MO, USA), and then immediately reverse transcribed into cDNA by MMLV high-performance reverse transcriptase (RT80125K, Epicentre, WI, USA). The mRNA expression levels were analyzed by using primers (listed in Additional file 1: Table S1) mixed with SYBR Green PCR Master Mix (10476600, Roche, Penzberg, Germany). A StepOnePlus Real-Time PCR System (Applied Biosystem, ABI, MA, USA) was used to monitor the changes of fluorescence intensity from PCR products. GAPDH was used as internal control. The data were analyzed using StepOne software version 2.0.

Cortexes from 12-month-old mice were homogenized with a pestle at the concentration of 1 μg tissue /9 μl HEPES buffer (1% CHAPS 50 mM HEPES, 10 mM EDTA, 150 mM NaCl, pH 7.4). Tissue lysates were centrifuged at 600 × g for 5 mins and supernatants were collected. 200 μl samples were pre-cleared with 50 μl protein G bead (LSKMAGG02, Millipore, Germany) rotating at room temperature for 30 mins. Pre-cleared lysates were incubated with anti-DcR3 (33302, Biolegend, CA, USA), anti-syndecan-1 (10593–1-AP, ProteinTech, IL, USA), anti-glypican-1 (sc-66910, Santa Cruz biotechnology, TX, USA), or anti-Aβ (6E10, SIG-39320, COVANCE, NJ, USA) antibodies at 4 °C overnight, and were then mixed with protein G beads rotating at room temperature for 1 h. The beads were washed with 0.1% Tween 20 in PBS for 20 mins and were eluted by SDS-sample buffer (87.5 mM Tris-HCl, 1% SDS, 30% glycerol, 0.6 M DTT, 180 μM bromphenol blue, pH 6.8) at 95 °C, 10 mins. The eluted samples and input controls were monitored by Western blot.

Proteins were separated via 10% or 15% Tris-glycine SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. The membranes were probed with rabbit anti-PSD95 (3450 Cell Signaling, MA, USA), mouse anti-DcR3 (33302, Biolegend, CA, USA), rabbit anti-syndecan-1 (10593–1-AP, ProteinTech, IL, USA), rabbit anti-glypican-1 (sc-66910, Santa Cruz biotechnology, TX, USA), mouse anti-Aβ (6E10, SIG-39320, COVANCE, NJ, USA), anti-YM1 (01404, Stem Cell technology, Vancouver, Canada), mouse anti-GAPDH (60004-1-Ig, ProteinTech, IL, USA), and mouse anti-actin (MAB1501, Millipore, MA, USA) antibodies. The membranes were washed and probed with the HRP-conjugated goat anti-mouse IgG and goat anti-rabbit IgG (12–349, AP132P, Millipore, MA, USA). Protein signals were developed by using a chemiluminescent substrate ECL detection system (WBKLS0500, Millipore, MA, USA) and quantified by using a luminescence imaging system (LAS-4000, Fujifilm, Japan).

HFIP-treated Aβ1-42 peptides (rPeptide Inc. A–1163–2, GA, USA) were dissolved in 10% DMSO at 100 μM and stored at –80 °C. Before the experiment, the stock was aged at 4 °C for 24 h to generate oAβ or aged at 37 °C for 18 days to generate fAβ. A DcR3-SAS and Vector-SAS stable line were generated by transfecting the human DcR3 gene or control vector into SAS cells. Culture medium was collected after 24 h of DcR3-SAS and Vector-SAS growth for the conditioned medium experiment.

Primary microglia were prepared from postnatal day 0–5 C57B6/J mice. The cortexes were digested with 100 U papain and 400 U DNase I in HBSS buffer at 37 °C for 30 min. Digested cells were passed through a 70-μm cell strainer (Corning NY, USA). Mixed cortical cells were grown in DMEM-F12. After 21 days incubation, microglia were isolated from a 30:37:70% Percoll (P4937, Sigma. MO, USA) gradient and were seeded in 25-T flasks for 24 h.

Primary cortical neuron cultures were prepared from postnatal day 0–1 C57B6/J mice as primary microglia but seeded at a density of 4x10⁵ per well for 7 days in the neurobasal medium.

To obtain conditioned medium (CM) microglial culture were stimulated with the DcR3-SAS medium before, together or after incubating with Aβ for 72 h and their media were collected as pre-, co- or post-treatment Aβ/DcR3-CM. Aβ-CM was collected from microglial culture stimulated with Vector-SAS medium and Aβ for 72 h, and control-CM was collected from microglial culture stimulated with the vector-SAS medium. For the DcR3 immune-depletion control, the DcR3-SAS medium was incubated with anti-DcR3 (33302, Biolegend, CA, USA) and protein G bead (LSKMAGG02, Millipore, Germany) rotating at 4 °C overnight. DcR3 depleted SAS medium was collected for microglia treatment. For the competition assay, microglia were co-treated with DcR3 and 30 μg/ml heparin sulfate (HS) to interfere the DcR3-HSPG interaction for 8 h and then incubated with Aβ for 72 h. These conditioned media were collected and applied to primary neurons for 72 h.

Neuronal survival rate after different CM treatment was assessed using MTT (3006 Biotium Inc., CA, USA) and propidium iodide (PI) staining assays according to the manufacturer’s instructions. For the MTT assay, formazan was solubilized in lysis buffer (10% SDS and 20 mM HCl), and the concentration was determined according to the optical density at 570 nm with a Sunrise™ absorbance reader with Magellan™ data analysis software (Version 6; Tecan Switzerland).

For PI staining neurons were incubated with 10 μg/ml PI in PBS for 20 min and were fixed in 4% paraformaldehyde for immunofluorescence staining with the MAP2 antibody. The staining results were quantified as the ratio of PI⁺ neurons to total MAP2⁺ neurons by using MetaMorph® Microscopy Automation & Image Analysis Software (Molecular Devices, CA, USA).

The cytokines in the primary microglia conditioned medium were detected by using the mouse cytokine array C1000 (AAM-CYT-1000 RayBiotech, GA, USA). Membranes were incubated with control CM, Aβ-CM or Aβ/DcR3-CM (pre-treatment condition) for 16 h and detected with a Biotin-Streptavidin system. Signals were scanned by using a Fujinon LAS-4000 system and quantified by using Multi Gauge V3.0 software (Fujifilm Corporation, Tokyo, Japan). The level of each cytokine in the control group was set as 100.

Purified microglia were seeded at a density of 1x10⁵ cells/well on poly-d-lysine coated coverslips. Attached microglia were treated with oAβ or oAβ + DcR3 for 72 h. Their phagocytic ability was examined by incubating with red fluorescent carboxylated microspheres (F8821 1 μm in diameter, Polysciences Life Technologies, USA) coated with fetal calf serum at 37 °C for 30 min. After three PBS washes, microglia were fixed with 4% paraformaldehyde and stained with anti-Iba1 antibody to visualize the number of engulfed microspheres in the microglia.

Data are presented as the mean ± s.e.m. from at least three independent experiments and were analyzed using Prism software (GraphPad). Differences between data sets were analyzed by unpaired two-tailed Student's t-tests or one-way ANOVA followed by the Bonferroni post hoc test. During multiple contrast analysis, the alpha was set as 0.05 (95% confidence intervals). All the precise numbers of samples and their statistical analysis methods of each experiment are listed in Additional file 2: Table S2. A p value less than 0.05 was considered to be statistically significant.

To investigate the effects of DcR3 on the functional and pathological features of AD transgenic mice overexpressing mutated human APP (line J20) and human DcR3 were crossed to generate APP/DcR3 double transgenic mice. The levels of the full-length APP (FL-APP) in APP and DcR3/APP mice did not change, and nor did the level of DcR3 in DcR3 and DcR3/APP mice (Additional file 3: Figure S1). Because this line of APP mice develops memory impairments and Aβ plaques at the age of 4 months and 5 months, respectively [26, 27], we followed the behavioral changes and Aβ plaque formation in APP/DcR3, APP, DcR3, and wild-type (WT) littermates, respectively, to determine whether DcR3 modulates the pathogenesis of AD at 6 months after birth.

The Morris water maze was used to assess the spatial learning and memory deficits among these four genotypes of mice (Fig. 1a & b). During the 5 days of the hidden platform test the APP transgenic mice spent more time than WT mice to locate the platform, indicating their deficits in memory acquisition. In contrast, the APP/DcR3 double transgenic mice use less time than the APP transgenic mice to reach the platform at last two days (Fig. 1a). At the 6th day of the probe trial, deficits in memory retention were observed in the APP transgenic mice but not in the APP/DcR3 double transgenic mice (Fig. 1b) compared with WT mice. No significant difference in swimming speeds was found among the 4 genotypes of mice (Fig. 1c). This observation suggested that overexpression of DcR3 rescued spatial learning and memory deficits in 6-month-old APP transgenic mice.

Contextual fear conditioning and auditory-cued fear conditioning tests were further applied to evaluate hippocampus-dependent and amygdala-dependent emotional memory respectively. During the training day, impaired learning was observed in the APP transgenic mice but not in the APP/DcR3 double transgenic mice (Fig. 1d). On day 2 of testing, the APP/DcR3 mice displayed a longer freezing time than the APP mice in the contextual fear conditioning test (Fig. 1e), suggesting that DcR3 reversed the hippocampus-dependent fear memory deficits. In contrast, there was no difference in the amygdala-dependent cued fear conditioning test among the 4 genotypes of mice (Fig. 1f). This observation suggests that DcR3 could ameliorate Aβ-induced hippocampus-related memory deficits.

We further examined the spontaneous motor activity and anxiety levels of these mice in the open field test and in the elevated plus maze. Consistent with previous findings [27] the APP mice traveled a longer distance in the open field and spent more time in the open arm of the elevated plus maze. In these two tests, the APP/DcR3 mice also had higher locomotor activity and lower anxiety-like behavior similar to the APP mice (Fig. 1g and h). These observations suggest that DcR3 reverses hippocampus-dependent memory impairment without changing locomotion- and anxiety-related behaviors.

The decline in the cognitive functions in AD patients or mouse models is accompanied by a loss of synaptic markers such as synaptophysin and PSD95 [26, 28, 29]. We thus investigated whether DcR3 prevented the loss of synapses by examining the synaptophysin and PSD95 density (Fig. 2a-d). Compared with WT littermates, significant loss of synaptophysin in the mossy fiber-CA3 pathway was observed in the APP mice as revealed by immunohistochemistry (IHC) staining (Fig. 2a, b & Additional file 4: Figure S2a), while synaptophysin expression level was reversed in the APP/DcR3 mice (Fig. 2a, b & Additional file 4: Figure S2a). Moreover, the synaptophysin intensity in the CA1 area and the dentate gyrus of DcR3 mice was higher than that of WT littermates, suggesting DcR3 is able to upregulate synaptophysin expression even without the presence of Aβ plaques. However, the levels of post-synaptic marker PSD95 and MAP2⁺ neuron had no significant difference among all 4 genotypes of mice (Fig. 2c & d and Additional file 4: Figure S2b & S2c). Because the mossy fiber pathway is critical for memory formation [30], the reversal of synaptophysin loss in this area suggests that DcR3 could preserve synapses to improve spatial memory in the APP mice [31].

Because microglia are the major players in Aβ-induced neurotoxicity we asked whether microglia are involved in DcR3-mediated neuronal protection. To address this question, primary microglia were pre-incubated with recombinant DcR3 or control medium for 8 h, followed by exposing to oligomeric or fibrillar Aβ (oAβ or fAβ) for 72 h (DcR3 pre-treatment condition). These Aβ/DcR3-stimulated conditioned media (CM) were harvested and incubated with primary neuronal cells for 72 h (Fig. 3a). The survival rate of the neuronal cells was determined by MTT assay (Fig. 3b & c) and PI staining (Additional file 5: Figure S3a & S3b). Compare with the Aβ-CM, the survival rate was significantly increased in the Aβ/DcR3-CM-treated neurons. To further observe neurotic dystrophy in these CM-treated neurons, the anti-MAP2 antibody was used to detect the morphology of neurons. More dystrophic neurites and swelling structure (arrow) were observed in the Aβ-CM than Aβ/DcR3-CM treated neurons (Additional file 5: Figure S3c), suggesting DcR3 modulated-CM prevented the loss of synaptic process in response to Aβ-induced stress.

We further examined the protective effect of DcR3 by incubating neurons with DcR3-CM simultaneously or after exposure to Aβ respectively (Fig. 3a, DcR3 co- and post-treatment). Compared with the Aβ-CM, the survival rate was significantly increased in neurons co-treated, but not post-treated, with DcR3-CM (Fig. 3d). These results suggested that DcR3-CM-mediated protection is via preventing but not reversing Aβ-induced neurotoxic effect (Fig. 3b-d & Additional file 5: Figure S3a & S3b). To further distinguish whether the neuroprotective effect of DcR3 was contributed from microglia or neuron, primary neurons were incubated with Aβ-CM and recombinant DcR3 for 72 h (Fig. 3a, neuronal treatment). Under this condition, DcR3 failed to protect primary neuron against Aβ-induced neurotoxicity (Fig. 3e & f). These observations suggest that the DcR3-mediated neuronal protection is via modulating microglia activation, rather than promoting neuronal resistance to Aβ-CM.

To examine the quality of oAβ and fAβ structures western blot was applied to examine the aggregation states of Aβ before, after microglial treatment, and after neuronal treatment. The oligomeric (10-72 kDa) and fibrillar (in the stacking gel) structures remain the major species in oAβ and fAβ group respectively (Additional file 6: Figure S4a-c). To further confirm the importance of DcR3 in neuroprotection, DcR3 in the SAS medium were removed by the anti-DcR3 conjugated protein G beads before treating onto microglia (Additional file 6: Figure S4d). While we treated the neuron with the DcR3-depleted CM, the neuronal survival rate was not returned to normal, which indicated a critical role of DcR3 on modulating innate-related cytokine-induced cytotoxicity (Additional file 6: Figure S4e).

Next, we investigated whether DcR3 improves pathological changes by measuring the amyloid plaque deposition and Aβ levels in mice. Compared with APP mice, less β-sheet amyloid plaques were observed in the hippocampus of APP/DcR3 mice as determined by thioflavin-S staining (Fig. 4a & b). In contrast, total Aβ deposition had no significant difference between APP and APP/DcR3 mice as determined by the 6E10 antibody (Additional file 7: Figure S5a-c). Furthermore, the guanidine-soluble total Aβ and Aβ1–42, which play major synaptotoxic roles in AD [26], were decreased in the hippocampus of APP/DcR3 mice as determined by ELISA (Fig. 4c & d). All the evidence indicated that DcR3 reduces amyloid with β-sheet structure and guanidine-soluble Aβ in APP mice.

We further examined the Aβ phagocytic activity of microglia which are first recruited to amyloid plaques before being activated to engulf Aβ to clear amyloid plaques in vivo [32]. We observed that clusters of Iba1-positive microglia (green) were found adjacent to the amyloid plaques (red) in the hippocampi of the APP mice. Interestingly, DcR3 enhanced microglia recruitment (Fig. 5a). Reconstruction analysis showed a higher percentage of microglia and amyloid plaques co-localization in APP/DcR3 than that in the APP mice (Fig. 5b). This observation supports the argument that DcR3 enhances the recruitment of activated microglia to clear amyloid plaques.

We further asked whether DcR3 directly promotes the Aβ-induced phagocytic ability of primary microglia by in vitro culture system. To determine phagocytic ability oAβ- or oAβ + DcR3-treated primary microglia were incubated with fluorescence labeled microspheres. In the presence of DcR3, the phagocytic activities of the microglia were upregulated significantly (Fig. 5c & d). These observations suggest that DcR3 enhances the phagocytosis of microglia, which may enhance Aβ-clearance ability.

Although no change in microglia survival rate was found there was an obvious difference in the morphology of primary microglia at 72 h after Aβ or Aβ/DcR3 treatment in vitro. The morphology of microglia without any treatment is in fusiform and ramified shape, which are characteristics of resting microglia. After oAβ treatment for 72 h, microglia enlarged and became an amoeboid shape. In oAβ/DcR3 treated microglia, the size of amoeboid shape microglia is smaller than the oAβ group (Additional file 8: Figure S6). These observations demonstrated the potent modulatory effect of DcR3 to attenuate oAβ-induced microglia activation.

Because the change of microglial population can be detrimental or beneficial in neuroinflammation we further investigated the phenotype of microglia activated by Aβ and DcR3 in vivo and in vitro. The expression of cytokines was determined by ELISA, while the markers of type I and type II microglia and components of inflammasome-related proteins were measured by qPCR (Fig. 6a-c and Additional file 9: Figure S7). Compared with APP mice, the expression of TNF-α and IL-1β, which are secreted by activated M1 microglia, were downregulated in APP/DcR3 mice (Fig. 6a & b). In contrast, the expression of YM1 and CCL17, the surface markers for M2a microglia [9], was upregulated by DcR3 (Fig. 6c & Additional file 9: Figure S7a & Additional file 10: Figure S8a-b). Nevertheless, the expression of other inflammatory markers, including the innate immune markers of M2b (IL-6, IL-10), M2c (IL-10, TGF-β, arganise1, CD206) [9, 16], and inflammasome (NLRP3, ASC, IL-18) [11], were similar (Additional file 9: Figure S7b-i). Furthermore, YM1 intensity near the plaques was higher and became more condensed in the APP/DcR3 mice than that in APP mice (Additional file 10: Figure S8a & S8b), indicating that microglia recruited to the plaques are polarized toward M2a-like subtype.

Since the sources of TNF-α and IL-1β in the brain are not only from microglia we confirmed the role of DcR3 to modulate the secretion of cytokines from microglia by using in vitro culture system. The cytokine profiles in the Aβ-CM or Aβ/DcR3-CM (from Fig. 3a, pre-treatment condition) were analyzed by cytokine array (Fig. 6d). Compared with Aβ-CM, lower levels of pro-inflammatory mediators (CD40, TNF-α, IL-1β, IL-12) with higher the levels of the M2a inducer (IL-4) and the M2a marker (YM1) were observed in Aβ/DcR3-CM (Fig. 6d and Additional file 11: Figure S9a). In contrast, the levels of M2b and M2c markers (CXCL2 and IL-10) were not altered by DcR3 (Fig. 6d). All these observations suggest that DcR3 has the potent effect to modulate cytokine secretion by modulating the activation and differentiation of microglia. It is interesting to note that DcR3 did not alter the expression of MMP9, which contributes to plaque clearance in the brain [33], suggesting MMP9-dependent proteolytic degradation of Aβ was not influenced by DcR3 (Additional file 11: Figure S9b). The complete list of all the changes from this cytokines array is listed in Additional file 12: Table S3. Thus, we concluded that DcR3 is able to skew microglia differentiation into IL-4⁺YM1⁺ M2a-like microglia in vivo and in vitro.

Because the interaction between DcR3 and HSPG is critical for modulating macrophage activation in vitro [24] we asked whether DcR3 also interacts with glypicans and syndecans, which are the most abundant HSPGs to modulate myeloid cell differentiation in the brain [22, 34, 35]. We found that DcR3 interacts with glypican-1 (Fig. 7a) and syndecan (Fig. 7b) by co-immunoprecipitation assay. In contrast, DcR3 did not interact with Aβ or APP (Fig. 7c), thus excluding the possibility that the neuroprotective effect of DcR3 is via direct neutralization of Aβ or APP. The interaction between DcR3 and HSPGs suggested that human DcR3 may modulate the activation and differentiation of microglia via interacting with HSPGs in vivo. To further confirm the role of HSPG in DcR3-mediated protection against Aβ-induced neurotoxicity, heparin sulfate (HS) was used to block DcR3-HSPG interaction by a competition assay as described in Fig. 4a [22]. In the presence of HS, DcR3-mediated neuroprotective effect against Aβ was attenuated (Fig 7d), suggesting DcR3-HSPG interaction contributes partially against Aβ-induced toxicity [36].

The importance of microglia-neuron interaction has been implicated in many neuroinflammatory-related disorders [39]. In the pre-plaque AD mouse oAβ and complement C1q initiate complement cascade and recruit microglia via CR3 to eliminate synapses [40]. Manipulation of the innate immune system into alternatively M2 activated microglia has been considered as a promising therapeutic strategy for AD [11, 41]. For example, intracerebral injections of IL-4/IL-13 or IL-33 reverse memory deficits and reduce Aβ plaque load in AD mouse models [41, 42]. In addition, YM1⁺ cells could protect neurons during acute brain injury [43, 44]. We found that DcR3-triggered IL-4⁺YM1⁺ M2a-like microglia contribute to anti-inflammatory response and functional recovery in vitro and in vivo. DcR3 reduces Aβ − induced pro-inflammatory cytokines, TNF-α and IL-1β [45], to prevent severe neuroinflammation and neurodegeneration [5]. Although DcR3 has been shown its’ ability to polarize macrophage differentiation in the periphery [24, 25], this study first presents that DcR3 modulates the complex innate immunity profiles and is able to counter Aβ-induced neuroinflammation in the brain.

In disease-related models microglia activation has dynamic, multidimensional and mosaic signatures [46]. Generally speaking, switching activated microglia from M1 into M2 anti-inflammatory spectrum can reverse inflammatory-related diseases [12, 47]. However, microglia/ macrophage have diverse characteristics in different regions and switch dynamically in response to environmental changes [46, 48]. Studies focusing on macrophage/microglia activation in specific tissues or using particular cytokine stimulation in vitro might not reveal the complex profiles of innate immunity environment [49, 50]. In addition, the M1/M2 microglia classification has been questioned on modulating neurotrophic factors to regulate synaptic plasticity and memory [48, 51, 52]. Therefore, more evidence are needed to better understand the roles of microglia populations toward AD pathogenesis.

To remove plaque via phagocytosis activated microglia are recruited to the site of plaques in AD animal models and patients [32, 53]. This recruitment is important for engulfing amyloids or tissue debris through lysosome-dependent manner or for constituting a barrier to prevents neurotoxic protofibrillar Aβ42 [32, 54, 55]. Although both astrocyte and microglia could be found near the plaques, only microglia have the ability to degrade Aβ [32]. Inefficient phagocytosis ability of microglia has been reported in AD animal models, and down-regulating phagocytosis-related gene expression has been found in AD subjects [56, 57]. This deficit could be caused by Aβ aggregates, reactive oxygen species (ROS), and pro-inflammatory mediators such as TNF-α, IL-1β, IL-6, and IL-18 [58, 59]. Regulating microglia phagocytosis is considered as a detrimental way to prevent tissue damage in the brain [60]. We found that DcR3 induces more microglia activation near plaque regions (Fig. 5a-b) and enhances microglia phagocytosis to remove Aβ in vivo and in vitro (Fig. 4 & Fig. 5c-d). The enhanced phagocytic ability of Aβ + DcR3-stimulated microglia may reverse the deficit of reducing phagocytic cells in AD patients or AD mice models [61, 62]. Altogether, our results suggest a potential beneficial role of DcR3 on modulating microglia into anti-inflammatory phagocytosis status.

We found that DcR3 interacts with mouse surface HSPGs glypican-1 and syndecan-1 (Fig. 7a & b), and DcR3 neuroprotective effect can be shaded by heparin in vitro (Fig. 7d). The HSPGs on the microglia and oligodendrocytes near lesions could modulate the accumulation of senile plaques and neurofibrillary tangles. [63, 64]. Glypican-1 can bind to Aβ aggregates to upregulate ER stress and to stimulate microglia activation that results in enhanced cytotoxicity [34, 65]. Syndecans are involved in the production of Aβ [66], ROS [35], and inflammatory cytokines [35, 67]. Furthermore, deletion of HSPGs accelerates Aβ clearance in APP/PS1 mice, suggesting that inhibition of Aβ-HSPG interaction is able to suppress Aβ-induced neuroinflammation [64]. However, the addition of heparin did not completely reverse DcR3 protection effects, which may due to the multiple roles of heparin such as promoting amyloidogenesis and modulate neuroinflammation [68, 69]. Because DcR3 interacts with HSPGs but not Aβ in our AD mouse model (Fig. 7a-c), DcR3-mediated neuronal protection may compete Aβ binding to HSPG or suppress HSPGs-CD14/TLR4 mediated inflammation [38]. The functions of HSPGs binding domain in DcR3 and its contributions to innate systems requires further analysis.

Several clinical studies have shown contradictory evidence for non-steroidal anti-inflammatory drugs (NSAIDs) in the treatment or prevention of AD [70]. These NSAIDs cannot be used in high doses or for a long period of time because of potential side effects in the cardiovascular or gastrointestinal systems [71]. In all NSAID clinical trials, cognitive performance had no significant improvements, and inconsistent results were reported for indomethacin [72, 73], ibuprofen [74, 75], celecoxib [76], rofecoxib [77, 78], and naproxen treatments [76]. A potential reason for the failure of anti-inflammatory approaches is that different anti-inflammatory cytokines induce diverse responses. For example, one anti-inflammatory cytokine, IL-10, facilitates Aβ aggregation, inhibits microglial phagocytic ability, and causes cognitive dysfunction in APP mice [79, 80]. In contrast, the other anti-inflammatory cytokine, IL-4, triggers microglia phagocytosis to reduce Aβ deposition [15, 81]. Thus, the IL-10 induced microglia may be harmful but IL-4⁺ microglia may be beneficial to AD-related symptoms. For novel treatment approaches, it is important to identify the microglia subtypes that ameliorates AD pathogenesis.

In comparison to other anti-inflammatory treatments DcR3 is a non-cytokine that induces an anti-inflammatory response, neutralizes FasL-induced cell death, and reduces ROS production [82]. Therefore, DcR3 may be an alternative agent for designing the future treatment. DcR3 concentration in our transgenic mouse serum is 268.6 ± 59.2 pg/ml (Additional file 3: Figure S1), which is slightly higher than healthy human (63.7 ± 21.9 pg/ml) but similar to the asthma patients (266.1 ± 60.6 pg/mL) [83]. Because DcR3 is highly expressed in the endometrium during pregnancy, it appears to be a safe natural immunomodulator to suppress neuronal inflammation in the presence of dangerous endogenous signals [84]. These findings suggested that DcR3 may be a safe therapeutic agent for early AD and other neuroinflammation-related diseases.