Bid Promotes K63-Linked Polyubiquitination of Tumor Necrosis Factor Receptor Associated Factor 6 (TRAF6) and Sensitizes to Mutant SOD1-Induced Proinflammatory Signaling in Microglia

Visual Abstract


Introduction
Amyotrophic lateral sclerosis (ALS) is a fatal, progressive neurodegenerative disease, characterized by the selective death of upper motoneurons in the cerebral cortex and lower motoneurons in the brainstem and anterior horn of the spinal cord (Ferraiuolo et al., 2011). Mutations in the superoxide dismutase 1 (SOD1) gene, linked to chromosome 21q22.1, account for ϳ12-20% of the familial cases of ALS (Rosen et al., 1993;Renton et al., 2014). The cytotoxicity of mutant SOD1, a 153 aa cytosolic protein, is mainly described as a toxic gainof-function mechanism (Tu et al., 1996). Mutations in SOD1 lead to protein misfolding and aggregate formation (Rotunno and Bosco, 2013) and have been shown to induce endoplasmic reticulum (ER) stress and proteasomal dysfunction, which trigger several stressactivated disease processes in ALS (Kikuchi et al., 2006;Kieran et al., 2007). In addition, mutant SOD1 G93A has been shown to have marked misfolding effects on wild-type SOD1 (Grad et al., 2011). Of note, it has become increasingly evident that non-cell autonomous toxicity mechanisms contribute to mutant SOD1induced motoneuron degeneration (Liu et al., 2009;Grad et al., 2011). Activation of both astrocytes and microglia are implicated in ALS pathogenesis with glial cell crosstalk contributing to the burden of inflammation (Hensley et al., 2006;Evans et al., 2013;Brites and Vaz, 2014). Studies have identified microglia to be phe-notypically neuroprotective at disease onset, however a proinflammatory activation state is soon acquired upon disease progression resulting in a chronic inflammatory pathology (Liao et al., 2012). Microglia expressing mutant SOD1 display elevated responses to inflammatory stimuli (Xiao et al., 2007;Sargsyan et al., 2009), and mutant SOD1 G93A secreted from motoneurons activates microglia and induces neurotoxicity .
Accumulating evidence demonstrates that mutant SOD1 induces increased Toll-like receptor (TLR) expression in ALS (Liu et al., 2009;Casula et al., 2011). TLRs are the master regulators of the cellular innate immune response (Scheffel et al., 2012), and are key mediators of the initiation and propagation of the inflammatory cascade in response to bacterial, viral or microbial nucleic acids, known as pathogenassociated-molecular-patterns (Kielian, 2006). TLR activation initiates several cascades of intracellular pathways, one of which leads to NF-B activation. NF-B is a transcription factor and the major cytoplasmic facilitator of inflammatory stimuli. Recent evidence suggests that SOD1 G93A activates the transmembrane receptors TLR2 and TLR4 in a CD14-dependent manner .
Previous studies observed increased levels of the Bcl-2 family member Bid (BH3-interacting domain death agonist) in the spinal cords of SOD1 G93A transgenic mice (Guégan et al., 2002;König et al., 2014). Bid is involved in the permeabilization of the mitochondrial outer membrane during death receptor activation, which leads to apoptosis . Recent research has suggested an immunoregulatory role of Bid (Mayo et al., 2011, Yeretssian et al., 2011König et al., 2014). An interaction between Bid and the innate immune receptor nucleotide-binding and oligomerization domain was suggested, however this study did not demonstrate attenuated LPS-induced inflammation in the absence of Bid (Yeretssian et al., 2011). Reduced phagocytic functioning was revealed in bid-deficient microglia (Mayo et al., 2011), and Bid was shown to associate with the IKK complex upstream of NF-B, specifically NEMO in intestinal epithelial cells (Yeretssian et al., 2011) and astrocytes (König et al., 2014). NEMO/IKK␥ is the regulatory subunit of the IKK complex [IKK␣, IKK␤, and IKK␥/NEMO (NF-B essential modulator)], activation of which is central to NF-B activation. We therefore set out to investigate the role of Bid as a positive regulator of mutant SOD1-induced TLR signaling in microglia, with a focus on Bid promotion of TLR4-NF-B pathway activation.

Mixed glial-cell culture and microglial isolation
All procedures involving animals were conducted under a license from the Department of Health and Children, as well as the Health Products Regulatory authority in Ireland. Procedures were reviewed by the Ethics Committee of the Royal College of Surgeons in Ireland. Mixed glial cultures were prepared from the cortices of P0 -P2 WT and bid Ϫ/Ϫ mice of mixed sexes on a C57BL/6 background. bid Ϫ/Ϫ mice were generated in the laboratory of Dr Andreas Strasser, WEHI, Melbourne, Australia (Kaufmann et al., 2007). The cortices were dissected and the meninges were removed before incubation in Trypsin-EDTA at 37°C for 10 min. DMEM-F12/L-glutamine (Gibco, Life Technologies) containing penicillin-streptomycin (1%; Sigma-Aldrich) and fetal bovine serum (10%; Sigma-Aldrich) was added to the cortices before trituration and passage through a 40 m nylon cell strainer (BD Falcon). The cells were centrifuged at 300ϫ g for 6 min and the pellet was resuspended in DMEM-F12 containing Pen/ Strep and FBS. The cells were plated at a density of ϳ4 cortices/T75 flask, and treated with M-CSF (10 ng/ml; R&D Systems), and GM-CSF (20 ng/ml; R&D Systems) from 1 day post-dissection to promote microglial prolifer-ation (Suzumura et al., 1990). Microglia were isolated using the shaking method. The media containing the detached and floating microglia was collected and centrifuged at 800 ϫ g for 6 min. The pelleted isolated microglia were resuspended in DMEM-F12 (containing FBS and Pen/Strep) and plated at a density of 3.5 ϫ 10 5 cells/well in a 6-well plate. The remaining cells were passaged and cultured in the absence of M-CSF and GM-CSF to obtain astrocyte cultures.

Mixed primary motoneuron preparation
Mixed primary cultures enriched for motoneurons were prepared from C57BL/6 murine WT E12 embryos by the dissection of the ventral horn of the spinal cord and subsequent purification, as previously described (Sebastià et al., 2009). In brief, the ventral horns were incubated with 0.025% trypsin in Neurobasal media for 10 min, followed by gentle dissociation in the presence of 0.1 mg/ml DNase1. The dissociated motoneurons were centrifuged at 300 ϫ g for 3 min and resuspended in complete Neurobasal media supplemented with 2 mM GlutaMAX, 2% horse serum, 2% B27, GDNF (Promega, Catalog #2781; 2 ng/ml), CNTF (R&D Systems, Catalog #557-NT-10; 1 ng/ml), 100 U/ml penicillin and 100 g/ml streptomycin. The cells were seeded at a density of 0.25 ϫ 10 6 cells/well in a 24-well plate pre-coated with polyornithine and laminin and cultured at 37°C in 5% CO 2 . Motoneurons were harvested for Western blotting at 7 DIV.

Generation of SOD1 G93A conditioned media
NSC-34 cells were reverse transfected with 5 g/well CFP or SOD1 G93A -CFP plasmid using Lipofectamine 2000 (1:2, DNA-Lipofectamine; Invitrogen) in Opti-MEM (Invitrogen) at a density of 4 ϫ 10 5 cells/well in a 6-well plate. After 4 h of incubation with the transfection mix, the cells were washed once with serum containing DMEM and incubated overnight in fresh DMEM containing full serum. The cells were then washed with serum free DMEM and incubated in the latter for 3 d before conditioned media was collected and centrifuged at 800 ϫ g for 5 min to remove cellular debris. Conditioned media was stored at -20°C or concentrated on 10 kDa size exclusion columns (10K, Amicon Ultra, UFC501096, Millipore) centrifuged at 4°C at 14,000 rpm for 45 min, boiled in Laemmli buffer for 5 min, and ran on an acrylamide gel for quantification of SOD1 content in the media.

Overexpression of SOD1 G93A in microglia
BV-2 cells were transfected using electroporation (Amaxa Nucleofector II, Lonza). Briefly, the cells were harvested and a density of 5 ϫ 10 6 cells/5 g plasmid was used for each cuvette. The cells were washed with RPMI (centrifuged at 90 ϫ g for 10 min) and the pellet was resuspended in 100 l RPMI including 5 g pcDNA3-CFP (13030; Addgene), pcDNA-SOD1wt-CFP, or pcDNA-SOD1 G93A -CFP and transfected via electroporation (program "A-023"). The transfected cells were carefully triturated in 1 ml of RPMI and placed in the incubator for 1 min before being plated on 6-well plates. Cells were incubated for 8 h and lysed using Buffer RLT (Qiagen RNA easy kit) for mRNA extraction, or at 24 h in RIPA buffer for protein analysis. Primary microglia were transfected with CFP or SOD1 G93A -CFP (0.35 ϫ 10 6 cells/5 g plasmid) using Lipofectamine (2 l/g plasmid) and OptiMEM. The cells were incubated with the transfection mix for 2.5 h before being replaced with serum containing media. The cells were lysed in RLT buffer 24 h post-transfection for RNA extraction and qPCR analysis.

Co-immunoprecipitation and pull-down experiments
Overexpression of tumor necrosis factor receptor associated factor 6 (TRAF6) was performed by transfection of the pCMV5-FLAG-wt-TRAF6 vector in BV-2 cells. In short, BV-2 cells were transfected with FLAG-wt-TRAF6 (Addgene, 21624) or pCMV5-FLAG (1.5 g/3 ϫ 10 5 cells) using Lipofectamine. The cells were stimulated with LPS for 1 or 4 h (1 g/ml BV-2, or 100 ng/ml primary glia and astrocytes). Cells were lysed in RIPA buffer (Tris 50 mM, NaCl 150 mM, SDS 0.1%, sodium-deoxycholate 0.5%, Triton X-100 or NP-40 1%, plus 1:100 Protease Inhibitor, Sigma-Aldrich). Co-immunoprecipitation or pull down experiments were performed using DynaBeads Protein G [35 l of Dynabeads/sample (100 -400 g protein), Life Technologies, 10007D], and a magnetic rack (Life Technologies). The beads were washed in RIPA buffer and incubated with 5 g primary antibody (in 200 l PBS for 30 min at room temperature). The beads were washed three times for 5 min in RIPA buffer and equal amounts of protein (100 g) were incubated rotating for 2 h at room temperature (in a total volume of 750 l). The protein was eluted from the beads and denatured in RIPA buffer plus 1ϫ Laemmli buffer by incubating for 10 min at 70°C. The tubes were placed on the magnetic rack and the denatured supernatant was collected for gel electrophoresis.

siRNA transfection
An siRNA targeting Bid, sequence ACACGACUGU-CAACUUUAU, was designed using an algorithm optimized for siRNA selection (Reynolds et al., 2004), and obtained from Sigma-Aldrich. Briefly, BV-2 cells were transfected with 100 M siRNA/3 ϫ 10 5 cells using Lipofectamine. Optimal silencing of bid was determined by qPCR analysis to be 48 h post-transfection. An siRNA consisting of a scrambled nucleotide sequence was used as the control in siRNA experiments.

Western blot
Media was aspirated from the cells and the wells were washed gently with PBS and lysed with 80 l/well of 6-well plate RIPA buffer for Western blot analysis. The cell lysates were incubated on ice for 15 min and centrifuged at 4°C at 13,000 rpm. The supernatants were used to determine the protein concentration by BCA assay (Micro BCA protein determination kit, Thermo Scientific). Laemmli buffer was added to each sample and the samples were boiled for 5 min and loaded onto 10%, 12%, or 15% polyacrylamide gels as appropriate. Semi-dry transfer was performed on to PVDF membrane for 1.5 h at 18 V. Membranes were exposed to Ponceau S and blocked in 3% milk for 1 h. Primary antibodies in 3% milk were incubated either overnight at 4°C or 2 h at room temperature. The membranes were washed in TBS-Tween 20 (0.1%), and placed in 3% blocking solution containing the appropriate secondary antibody (peroxidase-conjugated anti-mouse IgG, anti-rabbit IgG, or anti-goat IgG, Sigma-Aldrich; 1:5000, as appropriate) for 2 h at room temperature. The membrane was washed three times for 5 min in TBS-Tween 20, exposed to ECL Chemiluminescent Reagent (Millipore) for 5 min and imaged on a LAS-3000 Imager (Fuji). Quantification of protein levels were calculated from optical density measurements from Western blot experiments and normalized to respective loading control (␣-Tubulin, ␤-actin, or GAPDH).

Dual-luciferase assay
BV-2 cells were cotransfected by reverse transfection with renilla and firefly luciferase reporter gene vectors (renilla luciferase plasmid: firefly luciferase plasmid, 1:12) in Opti-MEM (Sigma-Aldrich; 1 g plasmid per well/100 l) using X-tremeGENE HP Reagent (Roche; 2 l/g plasmid). Following 5 h incubation with transfection mix the cells were washed and incubated with BV-2 growth media (RPMI ϩ 10% FBS, 1% L-glutamine and 1% Pen/Strep) overnight. BV-2 cells were treated with 100 M Bid Inhibitor (BI-6C9, Sigma-Aldrich), resuspended in DMSO, at a concentration 100 M 30 min prior to treatment with LPS (1 g/ml; Sigma-Aldrich, Catalog #L4391). BV-2 cells were lysed in 1ϫ Passive Lysis Buffer (Promega) and luminescence was measured using Dual-Luciferase Assay Kit (Promega). HEK293 and HEK294-TLR4 cells were similarly cotransfected with constitutive renilla (phRL-TK) and firefly B-luciferase reporter gene vectors [1:12 in OptiMEM using Lipofectamine (2 l/1 g plasmid)] by reverse transfection for 3 h before the cells were allowed to recover overnight in full serum media. After being washed gently with serum-free media both HEK293 and HEK293-TLR4 cells were stimulated with CFP or SOD1 G93A -CFP conditioned media for 8 or 24 h before being lysed in passive lysis buffer, with luminescence measured using the method as above.

Proximity ligation assay
The in situ proximity ligation assay (PLA) was carried out using the Duolink system (DUO92101-1KT, Sigma-Aldrich). This technique allows the identification of proximity between epitopes of proteins in a complex in vitro by using hybridization between oligonucleotides linked to secondary antibodies when bound to primary antibodies against two specific proteins in close proximity Յ40 nm in situ (Soderberg et al., 2006). Here we used mouse anti-TRAF6 (1:100) and rabbit anti-Bid (1:500; Abcam, ab62469) as primary antibodies to determine close proximity. BV-2 cells were plated in a 96-well plate and transfected with either pCMV-Myc-FLAG or pCMV-FLAG-TRAF6wt (Plasmid 21624, Addgene) at a concentration of 0.075 g/2 ϫ 10 4 cells. Twenty-four hours posttransfection the cells were exposed to serum free media 2 h prior to treatment with either vehicle (1ϫ PBS) or LPS (1 g/ml). The cells were immunostained as described above. Controls included combinations of mouse anti-TRAF6 (1:100) plus rabbit anti-HA-tag (1:500; Santa Cruz Biotechnology, sc805), and rabbit anti-Bid plus mouse anti-IRF2 (1:500; Santa CruzBiotechnology, sc101069). PLA probes were diluted in primary antibody diluent (1% horse serum and 0.3% Triton x-100 in PBS) at a ratio of 1:10 and incubated for 1 h at 37°C. The ligation stock was diluted (1:5) with high-purity H 2 O and ligase 1:40. Twenty microliters per well of ligation-ligase solution was incubated with the cells at 37°C for 30 min. The cells were incubated with the amplification solution (1:5 with highpurity H 2 O and 1:40 polymerase) for 100 min at 37°C. The cells were washed three times with PBS and Hoechst was used as a nuclear stain (1:1000 in PBS, Hoechst 33342). The wells were imaged in order to detect amplification dots representing TRAF6-Bid close-proximities and analysis was performed using ImageJ software.

Statistical analysis
Statistical analysis was performed using GraphPad Prism software or MATLAB software (v2014b, Math-Works) as applicable. Results are represented as mean Ϯ SEM. Statistical significance was determined using the tests as detailed in the respective figure legends and Table 1 (‫ء‬p Յ 0.05).

bid ؊/؊ microglia show reduced NF-B activation upon TLR4 stimulation
To investigate the involvement of Bid in microglial immune responses we examined Bid protein levels in resting WT mixed motoneurons, astrocytes, and microglia. We found higher levels of Bid protein in astrocytes and microglia ( Fig. 2A,B; astrocytes 5.9-fold Ϯ 2.78, microglia 6.2-fold Ϯ 2.08), verifying previous findings (König et al., 2014). To investigate the effects of Bid deficiency on SOD1 G93Ainduced microglial tlr4 mRNA expression we generated WT and bid Ϫ/Ϫ primary microglia cultures and transiently transfected the microglia with CFP or SOD1 G93A -CFP. tlr4 expression was higher in SOD1 G93A -overexpressing wildtype microglia compared with bid-deficient microglia ( Fig.   2C; 2.9-fold Ϯ 1.2 increase vs 0.8-fold Ϯ 0.9 decrease). tlr4 mRNA levels did not significantly increase in wild-type microglia overexpressing SOD1 G93A compared with CFPtransfected microglia, but increased TLR4 protein levels were detected in SOD1 G93A -overexpressing BV-2 cells (Fig. 1D,F).
To additionally investigate the effect of a lack of Bid on TLR4-induced proinflammatory signaling in microglia both WT and bid Ϫ/Ϫ microglia were stimulated with the well characterized TLR4 agonist LPS . Of note, both TLR2 and TLR4 signal to NF-B via the MyD88-dependent pathway . Wild-type and bid Ϫ/Ϫ microglia were stimulated with LPS for 4 h and lysed for Western blot analysis of the cytokine IL-1␤ and TNF␣ levels. IL-1␤ was used as a proinflammatory marker as IL-1␤ has been shown to be rapidly secreted in response to LPS (Dinarello, 1997) and to be critical in the NF-B induction of COX-II-derived prostaglandins in the CNS (Laflamme et al., 1999). TNF␣ is a proinflammatory cytokine secreted in response to LPS (Sawada et al., 1989), and microglial secretion of TNF␣ is shown to induce motoneuron death in vitro (He et al., 2002). Analysis revealed substantially reduced proinflammatory cytokine IL-1␤ and TNF␣ production (Fig.  2D,E) in bid-deficient microglia compared with wild-type cells 4 h post stimulation with LPS. Because microglia are the first line of immune response in CNS pathogenesis we investigated the effect of biddeficiency on microglial TLR4-NF-B activation following acute LPS stimulation. LPS-induced activation of the IKK complex leads to IB␣ degradation, in a phosphorylationinduced and proteasome-dependent manner (Alkalay et al., 1995;Brown et al., 1995;Yang et al., 2003). Wildtype and bid Ϫ/Ϫ microglia were stimulated with LPS for 5 min, fixed in 3% formaldehyde and stained with anti-CD11b and anti-pIKK␣/␤. CD11b is a cluster of differentiation molecule that is highly expressed in cells of the innate immune system, including macrophages and microglia, and is involved in mediating inflammation by regulating migration and adhesion properties. CD11b was used as a microglial marker and the fluorescence intensity of pIKK␣/␤ was measured on CD11b-expressing cells. Overall decreased immunofluorescence of phosphorylated IKK␣/␤ was noted in LPS-stimulated Bid-deficient microglia compared to wild-type CD11b-positive cells (Fig. 3A). Western blot analysis of microglia stimulated acutely with LPS, for 5, 15, 30, min or 1 h, showed significantly less phosphorylation of IKK␣/␤ and less phosphorylated p65 in bid Ϫ/Ϫ microglia compared with WT upon LPS stimulation from 5 min to 1 h (Fig. 3B,C). The degradation profile of the NF-B inhibitor IB␣ was measured as an indicator of TLR4-NF-B activation (Fig.  3D,E) and demonstrated that bid-deficiency delayed signal-induced degradation of IB␣ noted at 30 min and 1 h post LPS stimulation, confirming previous results (Mayo Additionally, the phosphorylation of p65 was significantly induced in wild-type microglia but not in bid-deficient microglia 1 h post LPS stimulation (Fig. 3F,G; 3.7-fold Ϯ 1.3 increase vs 2.7-fold Ϯ 0.8 increase). NF-B transactivation potential, as measured by Bresponse element-dependent luciferase expression, was also significantly reduced when Bid was inhibited using 100 M of the small molecule Bid inhibitor (BI-6C9) prior to LPS stimulation (24 h) in BV-2 cells ( Fig. 3H; BI-6C9 ϩ LPS decreased 3.00 Ϯ 1.858-fold compared with DMSO ϩ LPS). COX-II levels were assessed in Bid-depleted BV-2 cells following stimulation in paracrine with SOD1 G93A . BV-2 cells were transfected with siRNA targeting Bid followed by stimulation with SOD1 G93A in cMedia. Western blot analysis determined reduced COX-II levels in SOD1 G93A -treated BV-2 cells when Bid was depleted compared with control siRNA-transfected BV-2 cells exposed to the same SOD1 G93A stimulus (Fig. 3I,J; 1.6-fold Ϯ 0.8 increase vs 0.8-fold Ϯ 0.6 decrease).

Reduced levels of Peli1 in bid ؊/؊ microglia upon TLR4 activation
A number of E3 ubiquitin ligases regulate NF-B signaling and activation. Peli1 is responsible for catalyzing the K63linked polyubiquitination of the Interleukin-1 receptorassociated kinase (IRAK) complex (Moynagh, 2009) and has been described as a key regulator of CNS inflammatory pathogenesis (Song and Qian, 2013). Wild-type and bid Ϫ/Ϫ microglia were stimulated with LPS acutely as above, and lysed for Western blot analysis. Overall, Peli1 levels were significantly lower in bid-deficient microglia at early time-points post LPS stimulation (from 5 min to 1 h post-treatment) compared with wild-type microglia ( Primary mixed motoneurons were harvested at 7DIV, and purified astrocytes and isolated microglia were lysed 2 d post-plating to allow time for the cells to become quiescent. Bid protein levels were examined, by Western blot, in purified astrocytes and microglia at rest compared with motoneurons. Bid protein levels were assessed by measuring the optical densities of each sample relative to that of the respective loading control (B; GAPDH or ␤-actin, experiment was repeated 3 times with similar results). C, tlr4 mRNA expression in wild-type and bid Ϫ/Ϫ primary microglia overexpressing CFP or SOD1 G93A -CFP, analyzed by qPCR. Samples were normalized to internal control gapdh (n ϭ 4 -6 wells pooled from 2 separate experiments; p ϭ 0.027, one-way ANOVA, Bonferroni's multiple-comparison post hoc test). D, IL-1␤ levels were analyzed by Western blot in wild-type and bid Ϫ/Ϫ microglia 4 h post-LPS treatment. E, TNF␣ levels were analyzed in wild-type and bid-deficient microglia 4 h post-LPS stimulation. Wild-type and bid Ϫ/Ϫ microglia were stimulated with vehicle, LPS (100 ng/ml) for 4 h before being harvested with RIPA lysis buffer and used for Western blot analysis. A, Primary wild-type and bid Ϫ/Ϫ microglia were stimulated with LPS for 5-30 min in serum-free media before being fixed in 3% paraformaldehyde and stained with anti-phosphorylated IKK␣/␤ (pIKK␣/␤) and anti-CD11b. Immunohistochemistry analysis of anti-pIKK␣/␤ mean fluorescence on CD11b-positive cells is depicted. Scale bar, 50 M. B, C, Primary wild-type and bid Ϫ/Ϫ microglia were treated with LPS for 5, 15, 30 min, or 1 h in serum-free media before being lysed for Western blot analysis of pIKK␣/␤ protein levels (n ϭ 3 pooled from 3 separate experiments; p ϭ 0.014, 3-way ANOVA, Tukey post hoc test). D, E, Wild-type and bid Ϫ/Ϫ microglia were stimulated with LPS for 5 min to 2 h in serum-free media, lysed in RIPA buffer and IB␣ levels were analyzed by Western blot (n ϭ 3-4 wells from 3-4 separate experiments). F, G, Wild-type and bid Ϫ/Ϫ microglia were stimulated with LPS for 1 h before being lysed in RIPA buffer. pp65 levels were assessed by Western blot (n ϭ 7 wells pooled from 6 separate experiments; p ϭ 0.0162, one-way ANOVA, Tukey's post hoc test). H, BV-2 cells were cotransfected with NF-B-RE-luciferase and renilla-luciferase plasmids for 24 h and subsequently treated with LPS for 24 h. The cells were lysed in passive lysis buffer and NF-B activation was quantified by dual luciferase assay (represented as relative B-dependent firefly activity, n ϭ 6 -16 wells pooled from 2 separate experiments, 2 outliers removed, Grubbs test followed by Kruskal-Wallis and Dunn's multiple-comparison post hoc test). I, J, COX-II levels in Bid-depleted BV-2 cells stimulated with CFP or SOD1 G93A cMedia. BV-2 cells were transfected with an siRNA targeting Bid ("siBid") or a scrambled control siRNA ("siControl"), and stimulated with cMedia 48 h post-siRNA transfection, when Bid levels were optimally reduced. Twenty-four hours post-cMedia treatment the cells were lysed in RIPA and COX-II levels were measured. Dashed line indicated irrelevant lanes spliced out. Quantification of optical density was normalized to anti-␣-tubulin for each Western blot (n ϭ 3, from 3 separate experiments).

Bid promotes TRAF6-mediated polyubiquitination
Previous studies have identified Bid and IKK␥/NEMO association in response to inflammatory stimuli in intestinal epithelia (Yeretssian et al., 2011), as well as astrocytes.
Here we identified an interaction with an upstream mediator of NEMO activation, TRAF6 (Goh et al., 2012). TRAF6, a member of the TRAF family of intracellular signaling adaptor proteins, is critical for K63-linked ubiquitination of both Peli1 and the IRAK complex (Conze et al., 2008;Goh et al., 2012;Moynagh, 2014). We found that Bid and TRAF6 coimmunoprecipitated in BV-2 cells and WT glia (Fig. 5A-C). This interaction between TRAF6 and Bid was additionally investigated in astrocytes by coimmunoprecipitation. Here we observed that TRAF6 immunoprecipitated with Bid in wild-type astrocytes upon LPS stimulation for 1 h, an interaction that was not seen in bid-deficient astrocytes (Fig. 5D). Additionally, we used proximity-ligation assays to explore the association of Bid with TRAF6 proteins within microglia in situ. Here, a significant increase in Bid and TRAF6 close-proximities was detected upon LPS stimulation in BV-2 cells (Fig. 5E,F; 1.9-fold increase Ϯ 1.24).
Ubiquitination is a post-translational modification that involves the formation of ubiquitin linkage chains and leads to a variety of biological processes depending on which of the seven lysine residues of ubiquitin (K6, K11, K27, K29, K33, K48, K63) are covalently conjugated to form a polyubiquitin chain (Ikeda and Dikic, 2008). In addition, recent studies have identified a hybrid K63/linear (K63/M1)-linked ubiquitin chain formation in IKK complex signaling and subsequent NF-B activation (Emmerich et al., 2013). As the polyubiquitination of TRAF6 is essential for the positive regulation of downstream proteins of the TLR-NF-B pathway, such as NEMO, TAK1, and Peli1 we next assessed the levels of TRAF6 polyubiquitination in wild-type glial cells overexpressing Bid (Fig. 6A). Higher levels of TRAF6 polyubiquitination were present in wildtype glia compared with bid-deficient cells (Fig. 6A,B; 0.75-fold Ϯ 0.41 decrease). Overexpression of Bid promoted TRAF6 polyubiquitination in glial cells, however, basal levels of Bid also showed increased TRAF6 polyubiquitination compared with bid-deficient glia (Fig. 6B,D). In addition, the impact of the absence of Bid on TRAF6 K63-linked autoubiquitination was investigated. Wild-type and bid Ϫ/Ϫ glial cells overexpressing Bid and/or Ubiquitin-HA were stimulated with LPS for 1 h and TRAF6 was immunoprecipitated from these samples. Subsequent Western blot analysis of immunoprecipitated samples revealed an increased K63-linked polyubiquitination of TRAF6 in LPS stimulated wild-type microglia overexpressing ubiquitin-HA compared with bid Ϫ/Ϫ glia overexpressing ubiquitin-HA (Fig.  6C,D; 0.67-fold Ϯ 0.3 decrease). Next, the levels of total K63-linked ubiquitin chains were assessed in wild-type and bid-deficient microglia upon acute stimulations with LPS for 30 min and 1 h (Fig. 6E,F; 2.1-fold Ϯ 0.42 increase vs 1.5-fold Ϯ 0.36 increase 1 h post-LPS stimulation). The microglia were lysed and analyzed by Western blot, using an anti-K63-linked ubiquitin chain specific antibody. Lower levels of total K63-linked ubiquitin chains were observed when Bid was absent in microglia.

Discussion
In this study, we demonstrate that Bid positively mediates microglial TLR4-NF-B signaling, highlighting a nonapoptotic role of Bid in regulating immune responses in the CNS. Bid promotes the K63-linked polyubiquitination of TRAF6 upstream of IKK and absence of Bid attenuates the TLR4-induced NF-B proinflammatory microglial response.
There is an emerging emphasis on the role of the innate immunity in driving neurodegenerative disease progression (Nguyen et al., 2002;, with mounting evidence implicating a central role of microglia in ALS pathogenesis (Beers et al., 2006;Di Giorgio et al., 2007;Glass et al., 2010). TLR2 and TLR4 mediate ALS-linked mutant SOD1 toxicity, as extracellular mutant SOD1 G93A activates microglia in a CD14-dependent manner Lee et al., 2015). Here we show increased TLR2 and TLR4 levels following overexpression of SOD1 G93A in BV-2 cells. Moreover, the data presented demonstrates increased COX-II levels in response to SOD1 G93A , elicited through both autocrine and paracrine stimulation, further consolidating the previous findings presenting TLR2, TLR4, and NF-B activation in mutant SOD1 treated microglia (Frakes et al., 2014;Lee et al., 2015). A lack of induction of Bid in microglia upon mutant SOD1 or LPS stimulation was observed (data not shown), indicating that the constitutively high levels of microglial Bid are sufficient for the signaling cascade to respond to TLR4-induced stimulation. Recent studies proposed that Bid positively regulates the immunological profile of macrophages and epithelial cells, and Bid-deficiency resulted in decreased proinflammatory cytokine mRNA levels in response to LPS in microglia (Mayo et al., 2011), and attenuated NF-B activation in astrocytes (König et al., 2014). These findings are of critical importance as elevated levels of Bid are seen both in the spinal cord and activated astrocytes in the SOD1 G93A transgenic mouse model (Guégan et al., 2002;König et al., 2014), suggesting increased Bid levels may contribute to the chronic inflammation evident in ALS disease progression. Interestingly, TNF␣ levels increase concurrent with disease progression in mutant SOD1linked ALS pathogenesis (Hensley et al., 2003), and here we found reduced levels of TNF␣ in LPS-induced biddeficient microglia. We have demonstrated that Bid is highly expressed in unstimulated microglia, and that microglial bid depletion attenuates SOD1 G93A -induced tox-A B C D E F Figure 5. Bid associates with TRAF6 in microglia and astrocytes, as shown by coimmunoprecipitation and PLA. A, Coimmunoprecipitation of Bid and TRAF6 in BV-2 cells. BV-2 cells were stimulated with LPS (1 g/ml) for 15 min or 1 h and Bid was immunoprecipitated. Negative controls included anti-Bid immunoprecipitation from bid-deficient mixed glia lysates, and IgG immunoprecipitation from all samples. Cells were lysed in RIPA buffer and analyzed for TRAF6 content after immunoprecipitation of Bid. B, Coimmunoprecipitation of Bid and TRAF6 in BV-2 cells overexpressing TRAF6-FLAG. BV-2 cells were stimulated with LPS (1 g/ml) and lysed in RIPA buffer. FLAG was detected by Western blotting and represents TRAF6FLAG immunoprecipitated with Bid in BV-2 cells. An IgG immunoprecipitation was included as a negative control. C, Coimmunoprecipitation of Bid and TRAF6 in WT and bid-deficient primary mixed glia stimulated with LPS for 1 and 4 h (100 ng/ml). Samples were lysed post-LPS stimulation and Bid was immunoprecipitated from the lysates. The samples were analyzed for TRAF6 content by Western blotting. IgG immunoprecipitation was carried out as an additional negative control. D, Coimmunoprecipitation of Bid and TRAF6 in wild-type and bid Ϫ/Ϫ astrocytes. Purified astrocytes were stimulated with LPS (100 ng/ml) for 1 and 4 h, and lysed in RIPA buffer for Bid immunoprecipitation. The samples were analyzed for TRAF6 by Western blotting. E, Representative images of PLA and phase contrast in TRAF6-FLAG overexpressing BV-2 cells immunostained with anti-Bid and anti-TRAF6 (n ϭ 2 wells/condition, 4 fields of view per well). Negative control representative images of PLA in TRAF6-FLAG overexpressing BV-2 cells immunostained with anti-TRAF6 and anti-HA-tag (n ϭ 1 well/condition, 6 fields of view-LPS, 1 field of view ϩ LPS), or immunostained with anti-Bid and anti-IRF2 (n ϭ 1 well/condition, 7 fields of view). Scale bar, 10 m. F, Quantification of PLA interactions in BV-2 cells. BV-2 cells were transfected with TRAF6-FLAG or empty FLAG vector and stimulated with LPS for 1 h. The cells were fixed with 3% paraformaldehyde, incubated with anti-Bid and anti-TRAF6 and PLA was quantified (significant increase of PLA dots in TRAF6 transfected versus control transfected cells and vehicle vs LPS treated cells, two-way ANOVA). Negative controls included immunostaining with anti-Bid plus anti-IRF2, and anti-TRAF6 plus anti-HA-tag.
icity. The role of microglia in SOD1-mediated toxicity has been supported by studies which demonstrate that accumulation of mutant SOD1 in either motoneurons alone (Pramatarova et al., 2001) or astrocytes (Gong et al., 2000) may not be sufficient to efficiently cause motoneuron degeneration, with the addition of SOD1 G93A microglia to co-cultures shown to induce motoneuron death . Additionally, a recent study shows that NF-B is activated concurrently with disease progression in SOD1 G93A mice, and that inhibition of NF-B in microglia rescued motoneurons from cell death in SOD1 G93A murine cells in culture, and delayed motor function deficits in SOD1 G93A mice (Frakes et al., 2014).
The therapeutic potential of attenuated TLR4 signalling in ALS is highlighted in studies demonstrating reduced motoneuron degeneration in SOD1 G93A mice lacking tlr4 (Lee et al., 2015), and in rescued SOD1 G93A stimulated motoneuron death in vitro using TLR4 inhibitors (De Paola et al., 2016). We demonstrate that Bid has a specific role in TLR4 signaling, as LPS primarily elicits its response via TLR4 and is a potent agonist of TLR4-mediated NF-B activation Bottcher et al., 2003), inducing the pro-inflammatory M1 activated microglial phenotype (Chhor et al., 2013). Acute LPS stimulations reveal decreased phosphorylation of IKK␣ and IKK␤, required for activation of the IKK complex and further signaling to IB␣ (Wang et al., 2001). IB␣ sequesters NF-B subunits in the cytosol inhibiting their nuclear translocation, and bid-deficient microglia showed protracted IB␣ degradation kinetics, indicating delayed TLR4 signaling (Mayo et al., 2011). Our data demonstrates decreased NF-B activation levels in Bid-inhibited LPS-stimulated BV-2 cells. Constitutive activation of NF-B in WT microglia induces both gliosis and motoneuron cell death (Frakes et al., 2014), and therefore ablating this activation may ameliorate both of these hallmarks of ALS aetiology, further highlighting a potential neuroprotective role for bid depletion in microglia. Figure 6. Bid promotes K63-linked polyubiquitination of TRAF6. A, TRAF6 polyubiquitination levels in wild-type and bid Ϫ/Ϫ glia. TRAF6 was immunoprecipitated and TRAF6 polyubiquitination was assessed in WT mixed glia overexpressing Bid-dsRed plus Ubiquitin-HA (HA-Ub), and wild-type glia overexpressing ubiquitin-HA plus dsRed. bid Ϫ/Ϫ glia were transfected with ubiquitin-HA plus dsRed. Immunoprecipitation of IgG was used as an immunoprecipitation control. Single membrane cut as represented by dashed line. B, Quantification of TRAF6-linked ubiquitin chains, as measured by OD in wild-type and bid Ϫ/Ϫ glia transfected with ubiquitin-HA plus dsRed (n ϭ 2 pooled from 2 separate experiments). C, Wild-type and bid Ϫ/Ϫ glia were transfected with the following plasmids, Bid plus dsRed, ubiquitin-HA plus dsRed or Bid plus ubiquitin-HA plasmids for 24 h and subsequently stimulated with LPS for 1 h. The cells were lysed in RIPA buffer. TRAF6 was immunoprecipitated from the transfected cell lysates and the samples were analyzed by Western blot. The membrane was exposed to an anti-K63-linked ubiquitin-specific antibody, which identified the K63-linked ubiquitination of TRAF6. D, Quantification of TRAF6 K63-linked ubiquitin chains, as measured by optical density in wild-type and bid Ϫ/Ϫ glia transfected with ubiquitin-HA plus dsRed (n ϭ 2-3 from 2-3 separate experiments). E, Representative Western blot of total K63-linked ubiquitination levels in WT and bid-deficient microglia upon LPS stimulation. F, Quantification of OD analyses of total K63-linked polyubiquitination in wild-type and bid Ϫ/Ϫ microglia following LPS stimulation (n ϭ 3 cultures from 3 separate experiments; p ϭ 0.050, one-way ANOVA, Dunn's multiple-comparison post hoc test). Wild-type and bid Ϫ/Ϫ microglia were treated with LPS (100 ng/ml) for 30 min or 1 h before lysis in RIPA buffer and K63-linked ubiquitination levels were analyzed probing with anti-K63-linked ubiquitin-specific antibodies in Western blotting experiments.

A B C D E F
TLR-NF-B signaling is tightly regulated by a number of E3 ubiquitin ligase proteins. It has previously been reported that the E3 ubiquitin ligase Peli1 regulates the NF-B pathway and reduced Peli1 levels are shown to impair TLR signaling . Here we identified decreased levels of Peli1 upon acute LPS stimulation in bid Ϫ/Ϫ microglia compared with wild-type. As Peli1 facilitates the K63-linked ubiquitination of the IRAK complex (Butler et al., 2007;Ordureau et al., 2008), and subsequent downstream signaling to NF-B, reduced Peli1 levels in bid-deficient microglia may suggest accelerated Peli1 degradation. We propose this to be a process by which Bid regulates TLR-induced NF-B activation.
Many studies to date have shown that the TRAF6 RING domain is essential for the activation of IKK in TLR signaling by either activation of TAK1 (Lamothe et al., 2007;Walsh et al., 2008) or the Peli1-IRAK complex (Conze et al., 2008). Additionally, inhibiting the formation of Peli1-IRAK-TRAF6 interactions was identified to prevent the degradation of the NF-B inhibitor IB␣ (Choi et al., 2006). We propose that an interaction between Bid and the E3 ubiquitin ligase protein TRAF6 modulates the NF-〉 signaling pathway in microglia. Together with previous studies, our data implies a potential cell-type-specific interaction between Bid and different signaling proteins essential for NF-B activation. Interestingly the Bid-TRAF6 interaction we observe in microglia suggests that Bid may interact specifically with E3 ubiquitin ligase proteins, which regulate NF-B activation. Ubiquitination of multiple proteins, forming mainly K63 and M1 chains (for review, see Iwai et al., 2014), is required for NF-B activation, in order for the IKK complex to become phosphorylated with subsequent K48-linked proteasomal degradation of IB␣, resulting in the release NF-B into the cytosol. TRAF6 activates NF-B via facilitating the recruitment of the TAK1/TAB1/TAB2 complex (Lamothe et al., 2007) and by directly forming ubiquitin chains with NEMO (Gautheron et al., 2010) and IRAK-1 (Qian et al., 2001). We have demonstrated that Bid promotes the autoubiquitination of TRAF6, specifically K63-linked ubiquitination upon LPS stimulation. TRAF6 is autoubiquitinated by K63linked chains, preventing the K48-linked degradation of TRAF6, and providing docking sites for an array of proteins that mediate NF-B activation (Chen, 2005). The requirement of TRAF6 autoubiquitination for TRAF6mediated ubiquitination of target proteins is debatable, with some studies arguing TRAF6 autoubiquitination to be unnecessary (Walsh et al., 2008;Wang et al., 2010) and other studies demonstrating that TRAF6 autoubiquitination is essential for TRAF6-mediated NEMO K63-ubiquitination and downstream NF-B activation (Lamothe et al., 2007). Interestingly, it has been reported that unanchored polyubiquitin chains generated by TRAF6 can activate both TAK1 and IKK complexes (Xia et al., 2009).
Additionally, TLR4 is of increasing interest in ALS therapeutics, with another agonist of TLR4, high-mobility group box 1 (HMGB1), shown to be secreted from motoneurons and to have increased reactivity in astrocytes and microglia concurrent with ALS disease progression (Lo Coco et al., 2007;Lee et al., 2015). Of note, TNF␣ secretion from NF-B-activated glial cells provides a death receptor agonist which may contribute to the rate of motoneuron death (Aebischer et al., 2013). We propose that the Bid-potentiated polyubiquitination of TRAF6 positively regulates TLR4-induced NF-B signaling by promoting Peli1-IRAK-TRAF6 interactions and supporting proinflammatory signal propagation to IB␣ and NF-B. Furthermore, as NF-B is a complex therapeutic target, a cell-specific approach may prove more effective. Depletion of microglial Bid offers a potential avenue for the regulation and attenuation of the TLR4-and TRAF6mediated inflammatory response in ALS pathogenesis.