Identification of a specific α-synuclein peptide (α-Syn 29-40) capable of eliciting microglial superoxide production to damage dopaminergic neurons

Mice lacking in gp91 ^phox (gp91 ^phox−/−) and their wild-type (WT) controls with a C57BL/6J background were provided by the Jackson Laboratory. Animal housing, breeding, and experiments were approved by the Institutional Animal Care Committee of the National Institute of Environmental Health Sciences and performed by strictly following the National Institutes of Health guidelines.

To determine whether the tested α-Syn peptide can pass across the blood-brain barrier (BBB) and diffuse into the interstitial tissues in the brain, mice were injected with saline or fluorescein isothiocyanate (FITC)-labeled α-Syn peptide (5 mg/kg) (ABI Scientific, Inc. Sterling, VA) along with Texas red-labeled lectin (30 μl/mouse) (Vector Laboratories, Inc. Burlingame, CA) for 6 h. The brain tissues were sectioned and stained by using anti-ionized calcium-binding adaptor molecule 1 (Iba-1) antibody (Wako Chemicals USA, Richmond, VA) and 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI, a fluorescent dye for nuclei) (Sigma-Aldrich, St. Louis, MO). To examine whether injection of α-Syn peptide leads to microglial activation, mice were injected with saline or unlabeled α-Syn peptide (5 mg/kg) for 24 h and further injected with Texas red-labeled lectin 10 min before sacrifice. Brain tissues were harvested and sectioned (30 μm in thickness) for staining nuclei, Iba-1 (using 019-19741 Ab; Wako Chemicals, Richmond, VA), and major histocompatibility complex II (MHC-II) (using M5/114.15.2 Ab; BioLegend, San Diego, CA) or for biochemical assays (e.g., real-time quantitative PCR (RT-qPCR) and malondialdehyde (MDA)). See below for details.

Dissolved in water at 2 mg/ml, recombinant human α-Syn (rPeptide, Bogart, GA) (endotoxin level: <0.024 EU/μg) was incubated with agitation at 37 °C for 7 days to allow for α-Syn to form aggregates [4, 17]. The concentration of α-Syn aggregates is calculated in terms of the initial concentration of synuclein.

Peptides comprising 12 amino acid residues were chemically synthesized (ABI Scientific, Sterling, VA) according to the amino acid sequence covering the alanine (A) 30-containing loci, which are located within the N-terminal amphipathic region. In the synthesized mutant α-Syn peptides, the alanine (A) 30 was replaced by a P (A30P mutants). The first mutated peptide spans A19 to P30. Shifts occur every two amino acid residues between adjacent peptides. The corresponding amino acid sequences (with the replaced amino acids marked in italics) are AEKTKQGVAEAP for A19-P30 (A30P), KTKQGVAEAPGK for K21-K32 (A30P), KQGVAEAPGKTK for K23-K34 (A30P), GVAEAPGKTKEG for G25-G36 (A30P), AEAPGKTKEGVL for A27-L38 (A30P), and APGKTKEGVLYV for A29-V40 (A30P). In the screening test using O₂ ^• − production assay, raw peptides with a purity of approximately 75 % were used individually or together (referred to as “AP pooled”) to stimulate mouse microglia. The experiments were performed following a double blind strategy. Based on the results of screening tests, specific peptides with a purity >98 % were selected to validate the screening test.

Microglia were prepared as previously described [17]. Briefly, brain cells of mouse pups were collected on postnatal day one (P1) and seeded into poly-d-lysine-coated (Sigma-Aldrich, St. Louis, MO) T175 flasks which were filled with Dulbecco’s modified Eagle medium (DMEM)-F12 (Life Technologies, Grand Island, NY) media containing 10 % fetal bovine serum, 100 U/ml penicillin and streptomycin, 2.0 mM glutamine, 2.0 mM nonessential amino acids, and 1.0 mM sodium pyruvate. Two weeks later, microglia were suspended and harvested by shaking the flasks at 180 rpm for 30 min.

Neuron-glia cultures were established by following the procedure described previously [18]. The cells of mesencephalic tissues were collected from mouse or rat embryos on embryonic day fourteen (E14) and seeded onto poly-d-lysine-coated 24-well plates in DMEM media (Gibco, Grand Island, NY) supplemented with 10 % fetal bovine serum, 10 % horse serum, 2.0 mM glutamine, 2.0 mM nonessential amino acids, 100 U/ml penicillin and streptomycin, 1.0 mM sodium pyruvate, and 5.6 mM glucose. Typically, immunostaining demonstrated that intact neuron-glia cultures were composed of approximately 40 % neurons (NeuN immunopositive cells), 50 % astrocytes (GFAP immunopositive cells), and 10 % microglia (OX-42 immunopositive cells) [18]. To make neuron-astrocyte cultures, the cells of mesencephalic tissues were collected and seeded into the wells of 24-well plates exactly as in the preparation of neuron-glia cultures. Forty-eight hours after seeding the cells, 2 mM l-leucine methyl ester (Sigma-Aldrich, St. Louis, MO), a compound that is capable of removing microglia but has no influence on the growth of other cell types, was added to each well. Six days after the initial cell seeding, neuron-astrocyte cultures were achieved, which usually contain 44 % neurons, 55 % astrocytes, and less than 0.5 % microglia. To prepare reconstituted neuron-glia cultures, an identical number of freshly isolated WT or gp91 ^phox−/− mouse microglia were loaded to rat neuron-astrocyte cultures on day 6 after seeding cells, allowing the newly added microglia to establish cell-cell interactions with neurons and astrocytes. These reconstituted neuron-glia cultures have the same cellular composition as that in the original neuron-glia cultures, including approximately 40 % neurons, 50 % astrocytes, and 10 % microglia. The next day, these reconstituted neuron-glia cultures along with neuron-glia cultures and neuron-astrocyte cultures were treated by adding 150 EU/ml lipopolysaccharide (LPS) (Calbiochem, San Diego, CA) or 1 μM α-Syn peptides. In some experiments, dimethyl sulfoxide (DMSO) or 0.25 mM apocynin (Apo) was added to the cultures 30 min before loading LPS or α-Syn peptides. Seven days after the treatments, all of the cultures were either subjected to dopamine (DA) uptake measurements or immunochemically stained for tyrosine hydroxylase (TH).

O ₂ ^• − produced by mouse microglia was detected based on the superoxide dismutase (SOD)-inhibitable reduction of the tetrazolium salt WST-1 (Santa Cruz Biotech. Dallas, TX) [19]. Briefly, seeded onto 96-well plates overnight, mouse microglia were washed with Hanks’ balanced salt solution (HBSS) without phenol red followed by adding to each well with 50 μl of HBSS with or without 50 U/ml SOD (Sigma-Aldrich, St. Louis, MO), 50 μl of HBSS containing 1 mM WST-1, and 50 μl of HBSS containing the vehicle control or 1 μM of each α-Syn peptide. The absorbance at 450 nm of each sample was immediately measured using a spectrophotometer (Molecular Devices, Sunnyvale, CA) every 2 min for 30 min. The level of O₂ ^• − production was calculated in terms of the difference between the absorbance values in the presence and absence of SOD.

After being stimulated with 150 EU/ml LPS, 250 nM α-Syn aggregates, or 1.0 μM α-Syn peptides, neuron-glia, neuron-astrocyte, and reconstituted neuron-glia cultures were incubated for 20 min at 37 °C with ³H-labeled DA (PerkinElmer Life Sciences, Santa Clara, CA) in Krebs-Ringer buffer (16 mM sodium phosphate, 119 mM NaCl, 4.7 mM KCl, 1.8 mM CaCl₂, 1.2 mM MgSO₄, 1.3 mM EDTA, and 5.6 mM glucose; pH 7.4). Cells were washed with cold Krebs-Ringer buffer and lysed in 1 N NaOH. The radioactivity of the lysates was measured using a liquid scintillation counter (PerkinElmer, Santa Clara, CA) and calculated by subtracting nonspecific DA uptake detected in samples in the presence of 10 μM mazindol (Sigma-Aldrich, St. Louis, MO) from the results of samples that did not include mazindol. Data were presented as the percentage of controls in the same batch of cell cultures [4].

For immunochemical (IC) staining, neuron-glia, neuron-astrocyte, and reconstituted neuron-glia cultures were fixed after stimulation with either 150 EU/ml LPS or 1 μM α-Syn peptides for 7 days. Endogenous peroxidase in the fixed cultures was neutralized with 1 % hydrogen peroxide for 20 min followed by sequentially incubating the cultures together with a TH antibody (Chemicon International, Billerica, MA), biotinylated secondary antibody, and ABC reagents (Dako, Carpinteria, CA). The reaction was revealed by adding 3,3-diaminobenzidine (Thermo Scientific, Waltham, MA) and analyzed using a microscopy coupled with a ×10 objective (Nikon, Tokyo, Japan). The amount of TH-positive cells, which represent dopaminergic neurons, was calculated as the percentage of controls from the same batch of cell cultures [16]. For immunofluorescence (IF) staining of p47 ^phox and CD11b, live microglia were first stimulated by 100 nM phorbol 12-myristate 13-acetate (PMA) or 1 μM α-Syn peptides for 30 min at 37 °C. Cells were further incubated on ice with rat anti-mouse CD16/32 IgG (BD Biosciences, San Jose, CA), which blocks Fc receptors on microglia, followed by immunostaining with an APC-conjugated anti-mouse CD11b Ab (BioLegend, San Diego, CA) for 30 min. After being fixed in 4.0 % paraformaldehyde and permeabilized in phosphate-buffered saline (PBS) containing 0.05 % Triton-X, cells were stained for p47 ^phox using Alexa 488-labeled anti-mouse antibody (Bioss Inc. Woburn, MA) and were further stained by DAPI to visualize the nuclei. To demonstrate the localization of gp91 ^phox and A29-V40 peptide or A19-A30 peptide, freshly isolated WT or gp91 ^phox−/− microglia were fixed in 4 % paraformaldehyde and permeabilized by 0.05 % Triton-X. After preincubation with rat anti-mouse CD16/32 IgG, cells were reacted with mouse anti-gp91 ^phox Ab (Santa Cruz Biotech. Dallas, TX), followed by adding Alexa 568-labeled secondary Ab. Samples were further incubated with FITC-labeled A29-V40 peptide or FITC-labeled A19-A30 peptide along with DAPI and analyzed with a Zeiss 510 confocal microscopy (Carl Zeiss, Pleasanton, CA). For immunofluorescent staining in thick tissues [20], brain slices (30 μm in thickness) were fixed in 4 % paraformaldehyde for 2 h and blocked in PBS containing 0.5 % Triton-X, 5 % fetal bovine serum, and 5 % goat serum at room temperature (RT) for 2 h. The tissues were further incubated with an anti-Iba-1 Ab (Wako Chemicals, Richmond, VA) in cold room overnight. After multiple washes, samples were incubated with an Alexa 647-conjugated goat anti-rabbit secondary Ab (Life Technologies, Grand Island, NY) along with an Alexa 488-labeled rat anti-mouse MHC-II Ab (M5/114.15.2 Ab; BioLegend, San Diego, CA) and DAPI at RT for 2 h. After washes, tissues were imaged by using a Zeiss 510 confocal microscopy (Carl Zeiss, Pleasanton, CA).

Highly aggressively proliferating immortalized (HAPI) cells were stimulated using 0.1 % BSA, 100 nM PMA, or 1 μM each α-Syn peptide and then lysed in a hypotonic lysis buffer (1 mM Tris, 1 mM KCl, 1 mM EGTA, 1 mM EDTA, and 0.1 mM DTT) supplemented with 10 mg/ml of the protease inhibitor mixture (Thermo Scientific, Waltham, MA) and 1 mM PMSF (Thermo Scientific, Waltham, MA). After dounce homogenization (20–25 St, tight pestle A), the cell lysates were centrifuged at 1600×g for 15 min, and the supernatants were further centrifuged at 40,000×g for 2 h. The high-speed centrifugation supernatants, which contain the cytosolic fraction, were then harvested. Meanwhile, the plasma membrane fragments were collected by dissolving the pellets in 1 % nonidet P-40 hypotonic lysis buffer. Proteins in both the high-speed supernatants and the re-suspended pellets were separated by running SDS/PAGE gels and were transferred to polyvinylidene difluoride (PVDF) membranes, followed by immune-blotting using anti-p47 ^phox , anti-p67 ^phox , and anti-gp91 ^phox Abs (Cell Signaling, Danvers, MA). The relative amount of each protein was quantified using the Quantity One program (Bio-Rad, Redmond, WA). The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in whole-cell lysates were immunoblotted by an Ab (Cell Signaling, Danvers, MA) as a loading control [21].

Flag-fused α-Syn peptides (ABI Scientific Inc., Sterling, VA) and cell lysates, which were collected from WT or gp91 ^phox−/− mixed glial cultures or enriched microglia by using IP lysis buffer (Thermo Scientific, Waltham, MA), were mixed in a cold room overnight in the presence of protease inhibitor cocktail (Thermo Scientific, Waltham, MA). Simultaneously, goat anti-gp91 ^phox polyclonal antibodies (Santa Cruz Biotech., Dallas, TX) or control IgGs were conjugated to protein G magnetic beads using a Dynabeads protein G immunoprecipitation kit (Life Technologies, Grand Island, NY) at RT for 10 min. Next, the α-Syn peptides and cell lysate mixtures were incubated with the antibody- or control IgG-conjugated protein G magnetic beads at RT for 30 min. After washing, the beads were boiled in Laemmli sample buffer and the supernatants were resolved using SDS-PAGE. The proteins were transferred to Nitrocellulose membranes (0.2 μm, Bio-Rad Laboratories Inc., Redmond, WA) and immunoblotted using mouse antibodies against Flag (Origene, Rockville, MD), gp91 ^phox (Santa Cruz Biotech. Dallas, TX), or tubulin (Abcam, Cambridge, MA).

Isolation of total RNA from the brain tissues was performed by using TRIzol according to the manufacturer’s instructions (Life Technologies, Grand Island, NY). Quality and quantity of total RNA were measured by using the Agilent 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA). Reverse transcriptase reactions were run on 1 μg of total messenger RNA (mRNA) by using the reverse transcription system (Promega, Madison, WI). Quantitative PCR was performed by using a Light Cycler 480 (Roche, Indianapolis, IN) and the SYBR Green I Master (Roche, Indianapolis, IN) according to the manufacturer’s instructions. Primers for mouse Iba-1 were 5′-GGATTTGCAGGGAGGAAAAG-3′ and 5′-TGGGATCATCGAGGAATTG-3′; primers for mouse MHC-II were 5′-CCG TCA CAG GAG TCA GAA AGG-3′ and 5′-CGG AGC AGA GAC ATT CAG GTC-3′; and primers for β-actin were 5′-CCCAAGGCCAACCGCGAGAAGAT-3′ and 5′-GTCCCGGCCAGCCAGGTCCAG-3′.

Aggregation of FL-α-Syn can be facilitated by the presence of A30P mutation in its apolipoprotein-binding domain, therefore leading to greater neurotoxicity by directly affecting neurons or by activating microglia [16, 22]. To determine the effects of A30P mutation on the ability of α-Syn to induce dopaminergic neuronal damage via microglia activation, we first compared the effects of WT or A30P FL-α-Syn on DA uptake (which has been shown to correlate with the function and structural integrity of dopaminergic neurons [4, 23‐25]) in neuron-glia cultures, which consist of approximately 40 % neurons, 50 % astrocytes, and 10 % microglia [18]. In this study, LPS was used as a positive control because adding 150 EU/ml LPS to neuron-glia cultures consistently causes an approximately 50 % loss of dopaminergic neurons. We found that, as demonstrated previously [16], FL-α-Syn aggregates featuring the A30P substitution are more toxic to dopaminergic neurons than WT, leading to an even lower ability of neurons to take up dopamine (DA), and resulting in fewer surviving TH-positive cells, which represent dopaminergic neurons in the cultures (Fig. 1a, b).

These findings suggest that the regions covering the A30P locus possibly contribute to activation of microglia and the ensuing neuroinflammation-mediated dopaminergic neuronal damage. This hypothesis led us to systematically examine the capacity of peptides corresponding to the amino acid sequences in the proximity of alanine 30 to activate microglia and induce neuronal damage. For this purpose, we synthesized six peptides consisting of 12 amino acids, all of which included the A30P substitution, following the strategy illustrated in Fig. 1c. For each consecutive peptide within this set, the first included amino acid was shifted by two amino acids toward the C-terminus. These peptides with a purity of 75 %, either individually or together, were added to stimulate primary mouse microglia. Superoxide released from microglia was significantly increased (149.4 ± 23.2 % increase) after exposure to the mixture of six peptides harboring the A30P mutation. However, when each of these peptides was added individually to mouse microglial cultures, A29-V40 (A30P, APGKTKEGVLYV) was the only peptide capable of stimulating the release of microglial superoxide (Fig. 1d) approximately 158.0 ± 21.4 % more than the BSA control. In another set of experiments, we tested whether the A30P substitution was critical for generating the elevated microglial response using highly purified peptides (purity >98 %) and found that WT and A30P-mutated A29-V40 peptides were equipotent in activating microglia to release superoxide (128.6 ± 22.2 % and 192.6 ± 47.1 % increase, respectively, compared to the BSA control), suggesting that at the concentration of 1.0 μM, the presence of A30P mutation does not enhance peptide-induced microglial release of superoxide (Fig. 1e). This is different from our previous finding that the increase of superoxide production induced by A30P-mutated FL-α-Syn aggregates (≈250 %) was significantly greater than by its WT form (≈90 %) [16].

Given the finding that A30P-mutated FL-α-Syn induced greater dopaminergic neuronal damage than WT, but WT A29-V40 peptide and its corresponding A30P-mutant activate microglia equally, we compared the effects of A29-V40 peptides with or without A30P on neurons in neuron-glia cultures. Exposure of cultures to either WT or mutated A29-V40 (A30P) peptide led to a reduction in DA uptake (Fig. 2a) and significantly decreased the number of surviving TH-positive cells to a similar level (Fig. 2b). However, unlike what occurred with FL-α-Syn aggregates, the presence of the point mutation A30P in A29-V40 peptide failed to cause additional damage of dopaminergic neurons (Fig. 1a, b and Fig. 2a–c), suggesting that different mechanisms might, respectively, be involved in FL-α-Syn aggregate- and A29-V40 peptide-induced neuronal toxicity. In contrast to A29-V40 peptides, neither WT A19-A30 control peptide nor its A30P mutant, which was not able to induce the release of microglial superoxide (Fig. 1d), impaired the function and viability of dopaminergic neurons (Fig. 2a, b). Moreover, morphological analyses showed that the neurites of surviving neurons were shorter in the cultures treated by A29-V40 peptides than in the vehicle controls. In contrast, the peptide A19-A30, either WT or A30P mutant, did not cause significant changes in neuronal morphology (Fig. 2c). Together, our data suggest that the A29-V40 peptide is a specific α-Syn fragment that is independently capable of inducing dopaminergic neuronal damage and that ROS generation is critical in the process.

Because Nox2 is a key enzyme regulating the release of superoxide from microglia, we examined whether this enzyme contributes to A29-V40-induced dopaminergic neuronal damage. To this end, neuronal loss was analyzed in neuron-glia cultures which were pretreated with the vehicle control DMSO or Apo, a Nox2 inhibitor, followed by adding A29-V40 peptide, either WT or mutated. Consistent with the data shown in Fig. 2, the presence of A29-V40 peptides markedly decreased the value of DA uptake and the number of surviving dopaminergic neurons, compared to that of the controls (Fig. 3a, b). Pretreatment with Apo fully restored the ability to take up DA and the number of surviving dopaminergic neurons in the cultures, which were stimulated by A29-V40 peptides. However, this compound could only partially restore the level of DA uptake and the number of surviving dopaminergic neurons in LPS-treated cultures (Fig. 3a, b), indicating that in addition to Nox2, other mechanisms may also contribute to the LPS-induced damage of dopaminergic neurons. While comparing the responses of WT neuron-glia cultures and those deficient in gp91 ^phox (gp91 ^phox−/− ), the catalytic unit of Nox2, to either WT or mutated A29-V40 peptides, we found that A29-V40 peptides failed to impair the ability to take up DA and cause the loss of dopaminergic neurons in the cultures deficient in gp91 ^phox (Fig. 3c, d). Together, our data suggested that the activity of Nox2 is required to mediate peptide A29-V40-induced dopaminergic neuronal damage.

Next, we determined the source of Nox2 that participated in peptide A29-V40-induced dopaminergic neuronal damage. Since superoxide production by microglia was used as an index while screening the active α-Syn peptides, we hypothesized that the Nox2 activated by peptides A29-V40 originates in microglia but not in neuron and/or astrocytes. To test this hypothesis, we examined DA uptake by neurons and counted surviving dopaminergic neurons in multiple types of cultures, including rat neuron-glia cultures (containing neurons, astrocytes, and microglia), neuron-astrocyte cultures (containing neurons and astrocytes), and reconstituted neuron-glia cultures (established by adding either WT or gp91 ^phox−/− mouse microglia back to rat neuron-astrocyte cultures). Because the A30P substitution did not influence the neurotoxicity of the A29-V40 peptide, we only tested its WT form. As shown in Fig. 4a, whereas peptide A29-V40 reduced DA uptake in WT neuron-glia cultures (the left group of bars), it did not impair DA uptake in neuron-astrocyte cultures that lacked microglia (the left-middle group of bars), meaning that peptide A29-V40 has no direct toxicity to neuron-astrocyte cultures or is incapable of stimulating them to generate toxic products (e.g., O₂ ^·–) to damage themselves. However, once WT microglia had been added back to neuron-astrocyte cultures, peptide A29-V40 re-gained the capacity to damage dopaminergic neurons (the right-middle group of bars). In contrast, adding gp91 ^phox−/− microglia to neuron-astrocyte cultures failed to restore the ability of the A29-V40 peptide to induce neuronal damage (the right group of bars). A similar response pattern was observed by counting the number of surviving dopaminergic neurons in these cultures that were exposed to the peptide A29-V40 (Fig. 4b). Moreover, TH-positive neurons cultured with microglia, either pre-existing or added back after purification, had shorter and slimmer processes after the addition of A29-V40 peptide. This response could not be re-created in neuron-astrocyte cultures or neuron-astrocyte cultures supplemented with gp91 ^phox−/− microglia (Fig. 4c). LPS treatment impaired the function and viability of dopaminergic neurons to a greater extent than A29-V40 peptide in the presence of microglia, but addition of gp91 ^phox knockout microglia only partially restored the loss of neuronal function and viability, suggesting that LPS-induced dopaminergic neuronal damage in our neuron-glia cultures is partially caused by inflammatory mediators (e.g., TNF-α) [26] other than Nox2-derived superoxide (Fig. 4a–c). Although one could argue that some of the observations might be confounded by the fact that mixed species (rat neurons and astrocytes mixed with mouse microglia) were used in the study, it is unlikely that microglia originally from different species cause substantial differences in the neurotoxicity of LPS or peptides, because reconstituted neuron-glia cultures, which were established by loading WT mouse microglia to rat neuron-astrocyte cultures, responded to the stimulation of LPS or A29-V40 peptide very similar to rat neuron-glia cultures (Fig. 4a, b). Moreover, our previous studies showed that superoxide generation by mouse and rat microglia are quite similar. Therefore, our data indicate that the Nox2 activity of microglia accounts for the neurotoxicity caused by the A29-V40 peptide.

Next, we explored direct evidence that A29-V40 activates microglial Nox2. Activation of Nox2 is mediated by the recruitment of cytosolic phosphorylated p47 ^phox and p67 ^phox and activated small GTPase Rac to cell membranes [27]. At the membrane, these subunits, along with gp91 ^phox and p22 ^phox which are normally located on cell membranes, assemble to form an activated multicomponent enzyme complex—Nox2. Therefore, membrane translocation of p47 ^phox and p67 ^phox usually indicates up-regulation of Nox2 activity. Upon stimulation by either PMA, a well-known protein kinase C (PKC) activator, or A29-V40 peptide, both p47 ^phox and p67 ^phox were enriched in the extracts of cellular membranes collected from rat microglia-derived HAPI cells. This response occurred with a simultaneous reduction of p47 ^phox and p67 ^phox in the cytosol. In contrast, A19-A30 peptide failed to increase p47 ^phox and p67 ^phox membrane translocation (Fig. 5a, b). The ability of A29-V40 peptide to activate microglial Nox2 was further confirmed by comparing the amount of superoxide produced by A29-V40 peptide-stimulated WT and gp91 ^phox−/− microglia. Like PMA, A29-V40 peptide markedly accelerated release of superoxide from WT microglia, but not from gp91 ^phox−/− lacking the Nox2 catalytic unit (Fig. 5c). Confocal imaging also revealed PMA- and A29-V40 peptide-induced accumulation of p47 ^phox on WT mouse microglial membranes, co-localized with microglial cell surface CD11b. However, p47 ^phox remained concentrated in the cytosol in A19-A30 peptide-treated cells (Fig. 5d). Intriguingly, none of these stimuli could induce p47 ^phox membrane translocation in gp91 ^phox null microglia (Fig. 5e), suggesting that regardless of the type of stimuli, gp91 ^phox is essential for recruiting and anchoring p47 ^phox onto cell membranes.

As a membrane-permeable compound, PMA can enter the cytosol and directly activate PKC to phosphorylate p47 ^phox and p67 ^phox , eventually leading to Nox2 activation via an “inside to outside” signal pathway [27]. However, A29-V40 peptide and PMA share no similarity in biochemical characteristics, prompting us to speculate that PMA and A29-V40 peptide activate Nox2 through distinct pathways. In this regard, previous studies have reported that some small molecules [28, 29] can directly target and activate the Nox2 catalytic unit gp91 ^phox . Therefore, we tested whether A29-V40 peptide can bind to gp91 ^phox . Supporting this idea, when gp91 ^phox antibody-conjugated magnetic beads were incubated with a mixture of Flag-tagged A29-V40 peptide and WT microglial lysates, both A29-V40 peptide and gp91 ^phox were precipitated, while neither was precipitated by beads conjugated to control IgG. However, the beads failed to pull down A29-V40 peptide from the mixture composed of Flag-tagged A29-V40 peptide and gp91 ^phox−/− cell lysates (Fig. 5f) or precipitate A19-A30 control peptide from the mixture composed of Flag-tagged A19-A30 peptide and WT cell lysates (Fig. 5g), suggesting that A29-V40 peptide is a specific peptide capable of binding to gp91 ^phox . This idea was further confirmed by the observation that FITC-labeled A29-V40 peptide could stick to the cell surface of WT but not gp91 ^phox−/− microglia whereas A19-A30 peptide failed to bind to the cell surface of WT microglia (Fig. 5h).

A major result of Nox2 activation is increased extracellular levels of O₂ ^• −, which is spontaneously converted to H₂O₂, a reaction that is facilitated by SOD [30]. Since, unlike PMA, A29-V40-induced activation of Nox2 occurs via binding to gp91 ^phox , we hypothesized that H₂O₂ may be the membrane-permeable molecule that transmits the signal to the cytosol. To further investigate this hypothesis, we added SOD and catalase, an enzyme with the ability to break down H₂O₂ to H₂O and O₂, to mouse microglial cultures before stimulating cells with PMA (an experimental control), A29-V40 peptide, or H₂O₂ (a positive control) and examined the effects on downstream mediators. A29-V40 peptide markedly increased the level of both phosphorylated p47 ^phox and Erk1/2 in WT microglia while the presence of SOD and catalase attenuated this response. Pretreating WT microglia with SOD and catalase did not affect PMA-induced phosphorylation of p47 ^phox and Erk1/2 (Fig. 6a, c), indicating, as expected, that H₂O₂ does not mediate this response. In gp91 ^phox−/− microglia, A29-V40 peptide was not able to induce phosphorylation of p47 ^phox and Erk1/2 in the presence or absence of SOD and catalase. In contrast, p47 ^phox and Erk1/2 phosphorylation was induced by direct exposure of gp91 ^phox−/− microglia to H₂O₂. Therefore, both the presence of gp91 ^phox and generation of H₂O₂ are necessary for the A29-V40-induced activation of Nox2. Despite the deficiency of gp91 ^phox in microglia, PMA retained the ability to induce Erk1/2 and p47 ^phox phosphorylation in the presence or absence of SOD and catalase (Fig. 6b, d), further confirming that different mechanisms underlie A29-V40 peptide- and PMA-induced microglial Nox2 activation.

As key components in the Ras-Raf-MEK-ERK signal transduction pathway, ERK1 and ERK2 are related to protein-serine/threonine kinases that may cause phosphorylation of p47 ^phox [31]. Therefore, we sought to determine whether Erk1/2 phosphorylation is necessary for p47 ^phox phosphorylation. Pretreatment with the Erk1/2 inhibitor U0126 [32] abolished Erk1/2 phosphorylation in PMA or A29-V40 peptide-stimulated WT microglia to levels even lower than controls. This inhibitor also led to a lower level of p47 ^phox phosphorylation in A29-V40 peptide-stimulated WT microglia but had no effect on PMA-induced p47 ^phox phosphorylation (Fig. 6e, g). Consistent with data shown in Fig. 6b, d, exposure of gp91 ^phox−/− microglia to A29-V40 peptide could neither activate Erk1/2 nor induce p47 ^phox phosphorylation while stimulation by PMA led to phosphorylation of both Erk1/2 and p47 ^phox . This suggests that Erk1/2 activation is upstream of p47 ^phox phosphorylation in A29-V40 peptide-initiated response, but PMA activates Erk1/2 and p47 ^phox independent of gp91 ^phox . Using U0126 to inhibit Erk1/2 activity did not cause any additive suppression of p47 ^phox phosphorylation in A29-V40 peptide-treated gp91 ^phox−/− microglia, compared to the DMSO control. This observation suggests that binding of A29-V40 peptide and gp91 ^phox is the initial event that activates Nox2 to produce superoxide and hence H₂O₂; H₂O₂ then enters cells and induces activation of Erk1/2, upstream of p47 ^phox phosphorylation, which may amplify A29-V40 peptide-induced Nox2 activation. Despite the suppression of Erk1/2 activity in U0126-pretreated gp91 ^phox−/− microglia, PMA still induced p47 ^phox phosphorylation (Fig. 6f, h), indicating that this compound activates Nox2 in an Erk1/2-independent manner.

Next, we examined whether A29-V40 activates microglia in vivo. Six hours after injection (i.v.) of FITC-labeled A29-V40 along with Texas red-conjugated lectin (an endothelial cell marker), the green fluorescent signals in the interstitial tissues of mouse brains were significantly increased compared to saline injection (Fig. 7a). This observation suggests that the A29-V40 peptide can pass across the BBB and diffuse into the interstitium, where it can stimulate microglia. MHC-II is a cell surface marker expressed by activated, but not resting, microglia. In our study, microglia in WT rather than in gp91 ^phox−/− mice, especially those surrounding the capillaries, became immunopositive for MHC-II 24 h after injection of A29-V40 peptide, but not saline or A19-A30 peptide, indicating that A29-V40 peptide activates microglia (Fig. 7b). This finding is consistent with the RT-qPCR analysis which showed higher levels of both MHC-II and Ibl-1 mRNA in the brains collected from A29-V40 peptide-injected but not A19-A30 peptide-injected WT mice. In contrast, neither A29-V40 peptide nor A19-A30 peptide could cause changes in levels of MHC-II and Ibl-1 mRNA in gp91 ^phox−/− brains (Fig. 7c, d). Moreover, administration of A29-V40 peptide to WT rather than to gp91 ^phox−/− brains increased the amount of MDA, one of the products during lipid peroxidation [33], indicating that enhanced redox reactions occurred in these brains (Fig. 7e). Taken together, our data suggest that the A29-V40 peptide is capable of activating microglia and eliciting redox reactions in vivo in a gp91 ^phox -dependent manner.