Vandana Deora1 | John D. Lee1,2 | Eduardo A. Albornoz1,5 | Luke McAlary3,4 | Cyril J. Jagaraj6 | Avril A. B. Robertson5,7 | Julie D. Atkin6 | Matthew A. Cooper5 | Kate Schroder5 | Justin J. Yerbury3,4 | Richard Gordon1,2 | Trent M. Woodruff1
Abstract
Microglial NLRP3 inflammasome activation is emerging as a key contributor to neu- roinflammation during neurodegeneration. Pathogenic protein aggregates such as β-amyloid and α-synuclein trigger microglial NLRP3 activation, leading to caspase-1 activation and IL-1β secretion. Both caspase-1 and IL-1β contribute to disease pro- gression in the mouse SOD1G93A model of amyotrophic lateral sclerosis (ALS), suggesting a role for microglial NLRP3. Prior studies, however, suggested SOD1G93A mice microglia do not express NLRP3, and SOD1G93A protein generated IL-1β in microglia independent to NLRP3. Here, we demonstrate using Nlrp3-GFP gene knock-in mice that microglia express NLRP3 in SOD1G93A mice. We show that both aggregated and soluble SOD1G93A activates inflammasome in primary mouse microglia leading caspase-1 and IL-1β cleavage, ASC speck formation, and the secre- tion of IL-1β in a dose- and time-dependent manner. Importantly, SOD1G93A was unable to induce IL-1β secretion from microglia deficient for Nlrp3, or pretreated with the specific NLRP3 inhibitor MCC950, confirming NLRP3 as the key inflammasome complex mediating SOD1-induced microglial IL-1β secretion. Microglial NLRP3 upregulation was also observed in the TDP-43Q331K ALS mouse model, and TDP-43 wild-type and mutant proteins could also activate microglial inflammasomes in a NLRP3-dependent manner. Mechanistically, we identified the generation of reactive oxygen species and ATP as key events required for SOD1G93A-mediated NLRP3 acti- vation. Taken together, our data demonstrate that ALS microglia express NLRP3, and that pathological ALS proteins activate the microglial NLRP3 inflammasome. NLRP3 inhibition may therefore be a potential therapeutic approach to arrest microglial neu- roinflammation and ALS disease progression.
KEYWORDS:glia, IL-1β, innate immunity, motor neuron disease
1 | INTRODUCTION
Amyotrophic lateral sclerosis (ALS) is an adult-onset, progressive neu- rodegenerative disease caused by the deterioration of motor neurons within motor cortex, brain stem, and spinal cord (Lee et al., 2017; Paez-Colasante, Figueroa-Romero, Sakowski, Goutman, & Feldman, 2015; Sibilla & Bertolotti, 2017). Current therapies for ALS are inade- quate. For example, the most widely used ALS therapeutic riluzole, a glutamate release inhibitor, prolongs survival by only 2-3 months on average (Geevasinga et al., 2016; Knibb et al., 2016). Hence, there is an urgent need to further identify the mechanisms driving ALS pathol- ogy, in order to identify novel therapeutic targets that could signifi- cantly alter the progression of the disease. Although the etiology of ALS is still unknown, emerging evidence indicates that the innate immune system could be a major contributor that drives the activation of macrophages/microglia (Brites & Vaz, 2014). Activation of these cells leads to the production of proinflammatory neurotoxic cytokines such as IL-1β that further drive motor neuron death (Meissner, Molawi, & Zychlinsky, 2010).Mutations in copper-zinc superoxide dismutase-1 (SOD1) were the first identified mutations for familial forms of ALS (Rosen et al., 1993), and are the most studied. The SOD1G93A mutation was used to create the SOD1G93A mouse (Gurney et al., 1994), which is the most widely used ALS model. The mutation decreases the folding stability of SOD1, which induces the formation of protein aggregates (Alexianu, Kozovska,& Appel,2001; Pasinelli & Brown, 2006; Pramatarova, Laganiere, Roussel, Brisebois, & Rouleau, 2001). These protein aggregates play an important role in SOD1-mediated ALS pathogenesis (Watanabe et al., 2001), and wild-type SOD1 aggregates have also been recently linked to sporadic ALS pathology (Forsberg et al., 2019; Pare et al., 2018). A major hallmark of protein aggregates in ALS patients is the presence of TDP-43 (Neumann et al., 2006), which is thought to translocate from the nucleus to the cytoplasm as part of the disease pathogenesis (Barmada et al., 2010).
Mutations in TDP-43 (e.g., TDP-43Q331K) also lead to familial forms of ALS (Sreedharan et al., 2008), and transgenic TDP-43Q331K mice have increased microglial activation and motor neuron degeneration (Lee et al., 2018).ALS protein aggregates are potent immune response triggers in microglia (Meissner, Molawi, & Zychlinsky, 2010; Roberts et al., 2013; Zhao et al., 2010). A key component of the innate immune system activated by protein aggregates and recently linked to neurodegenera- tive disease is the NLRP3 inflammasome (Codolo et al., 2013; Gordon et al., 2018; Heneka et al., 2013; Parajuli et al., 2013), an intracellular signaling complex. Inflammasome activation requires a priming signal for the upregulation of NLRP3 and cytokine precursors pro-IL-1β and pro-IL-18, followed by an activation step, which involves the recruitment of the inflammasome adapter, apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC), activation of the caspase-1 protease, and the cleavage and release of IL-1β and IL-18. Inflammasome activation can also lead to a form of inflammatory cell death, known as pyroptosis (Patel et al., 2017; Tschopp & Schroder, 2010). Microglial expression of NLRP3, and inflammasome component upregulation and activation is observed in Alzheimer’s disease and Parkinson’s disease (Gordon et al., 2018; Heneka et al., 2013). Inflammasome activation and upregulation of NLRP3 inflammasome components have also been observed in ALS patients and mouse models of ALS (Bellezza et al., 2018; Johann et al., 2015), with caspase-1 and IL-1β playing key roles in ALS pathology (Meissner, Molawi, & Zychlinsky, 2010). Despite this, a recent study suggested that microglia in SOD1G93A mice, and human ALS patients do not express NLRP3 (Johann et al., 2015). A separate study also surpris- ingly indicated that SOD1G93A-mediated caspase-1 activation and IL- 1β production in microglia occurs independently of NLRP3 (Meissner, Molawi, & Zychlinsky, 2010).
Given that microglial NLPR3 inflammasome activation is emerging as a key therapeutic target for neurodegenerative diseases, this study, therefore, aimed to revisit these prior findings by examining NLRP3 microglial expression in ALS models, and SOD1, and TDP- 43-mediated inflammasome activation in primary mouse microglia. We utilized Nlrp3-deficient microglia, as well as wild-type cells treated with the specific NLRP3 inhibitor, MCC950, to demonstrate that both soluble and aggregated recombinant SOD1G93A specifically activates the NLRP3 inflammasome to induce caspase-1 activation, ASC speck formation, and IL-1β secretion. TDP-43Q331K also activated the micro- glial inflammasome in a NLRP3-dependent manner. Furthermore, expression of the Nlrp3 locus was observed in microglia in the SOD1G93A mouse ALS model using a Nlrp3-GFP knock-in mouse. These studies, therefore, confirm a role for microglial NLRP3 inflam- masomes in ALS, and suggest that therapeutic targeting of NLRP3 may be an amenable neuroprotective strategy for ALS.
2 | MATERIAL AND METHODS
2.1 | Ethical statement
All experimental procedures were approved by the University of Queensland Animal Ethics Committee, and complied with the policies and regulations regarding animal experimentation and other ethical matters. They were conducted in accordance with the Queensland Government Animal Research Act 2001, associated Animal Care and Protection Regulations (2002 and 2008), and the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes, 8th Edition (National Health and Medical Research Council, 2013).
2.2 | Animals
Transgenic hSOD1G93A ALS mice (B6-Cg-Tg (SOD1-G93A) 1Gur/J) were obtained from Jackson laboratory (Bar Harbor, ME) and were bred on C57BL/6J background to produce hSOD1G93A mice and wild- type control mice. These hSOD1G93A mice carry a high copy number of the mutated allele of the human SOD1 gene. Female hSOD1G93A and wild-type mice at end-stage disease (P155- P170) were used in this study (Lee et al., 2013). Nlrp3-GFP gene knock-in mice (Nlrp3tm1Tsc; Guarda et al., 2011), were backcrossed to hSOD1G93A mice to generate hSOD1G93A × Nlrp3-GFP reporter mice.Female hSOD1G93A × Nlrp3-GFP mice were examined at end-stage disease. Transgenic TDP-43Q331K (Line 103) mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and were bred on a C57BL/6J background to produce TDP-43Q331K and wild-type (C57BL/6J) con- trol mice. TDP-43Q331K mice express a myc-tagged, human TDP-43 cDNA modified to have an ALS-linked glutamine to lysine residue mutation at position 331, under the direction of the mouse prion pro- tein promoter. The prion protein promoter ensures that the transgene expression is directed primarily to the central nervous system, and is very low in other peripheral tissues (Arnold et al., 2013). Female wild- type and TDP-43Q331K mice at 16 months of age were used in this study (Lee et al., 2018).
2.3 | Immunohistochemistry
Fluorescence double-labeling immunohistochemistry was performed to localize NLRP3(i.e.,visualization of the Nlrp3 locus via GFP) in hSOD1G93A × Nlrp3-GFP reporter mice with specific cell-type markers for astrocytes and microglia as previously described (Lee et al., 2013; Lee et al., 2018).Briefly, sections were rehydrated in phosphate- buffered saline (PBS) (pH 7.4) for 10 min and then blocked in PBS con- taining 3% bovine serum albumin (BSA) for 1 hr at room temperature. Sections were incubated overnight at 4。C with combination of primary antibodies (GFP [1:500; ab13970] with GFAP [1:1,000; ab7260] and CD11b [1:500; ab8878]). All primary antibodies were diluted in PBS (pH 7.4) containing 1% BSA. Sections were washed 3 × 10 min with PBS prior to incubation with an appropriate Alexa conjugated second- ary cocktail: Alexa 555 goat anti-rat, Alexa 555 goat anti-rabbit, and Alexa 488 goat anti-chicken (Invitrogen, Eugene, OR). All secondary antibodies were diluted in PBS (pH 7.4) containing 1% BSA (1:11,000 for Alexa 555 and 1:600 for Alexa 488). Following 3 × 5 min washes in PBS, all sections were mounted with Prolong Gold Anti-Fade medium containing DAPI (Invitrogen, Eugene, OR). Sections with no primary antibodies were used as negative controls for all immunohistochemistry experiments to give a measure of nonspecific background staining.
2.4 | Realtime quantitative PCR
Total RNA was isolated from spinal cord of wild-type, SOD1G93A, and TDP-43Q331K mice using RNeasy Lipid Tissue extraction kit according to manufacturer’s instructions (QIAGEN, CA). Total RNA was purified from genomic DNA contamination using Turbo DNAse treatment
(Ambion, NY), then converted to cDNA using AffinityScript cDNA syn- thesis kit according to manufacturer’s instructions (Agilent Technolo- gies, CA). Commercially available gene-specific Taqman probes for NLR family, pyrin domain containing 3 (Nlrp3; Mm00840904_m1), PYD and CARD domain containing (Pycard; Mm00445747_g1), caspase 1 (Casp1; Mm00438023_m1) and interleukin 1 beta (Il1b; Mm004342 28_m1) were used to amplify target gene of interest (Applied Biosystems, MA). Relative target gene expression to geometric mean of reference genes glyceraldehyde-3-phosphate dehydrogenase (Gapdh; Mm99999915_g1) and beta actin (Actb; Mm02619580_g1) was determined using this formula: 2−ΔCT where ΔCT = (Ct (Target gene)- Ct (Gapdh and Actb)), as per our previous studies (Lee et al., 2013; Lee et al., 2018). Final measures are presented as relative levels of gene expression in SOD1G93A and TDP-43Q331K mice compared with expression in wild-type controls. Probe sets were tested over a serial cDNA concentration for amplification efficiency. No reverse transcrip- tion and water as no template control was used as negative controls. All samples were run in triplicate and were tested in three separate experiments.
2.5 | Primary microglia culture
Primary microglia cultures were prepared from C57BL/6J and NLRP3 knockout (Nlrp3−/−) mouse pups at postnatal Day 0-1. Mouse brains were collected and their meninges were removed before washing with sterile ice-cold Dulbecco’s modified Eagle’s medium/F-12 nutrient mixture (DMEM/F12, GIBCO, Waltham, MA) supplemented with 10% heat-inactivated fetal bovine serum, 50 U/ml penicillin, 50 μg/ml streptomycin, 2 mM L-glutamine, 100 μM nonessential amino acids, and 2 mM sodium pyruvate (Invitrogen, Eugene, OR). The brains were then incubated with 0.25% trypsin (Sigma, St. Loius, MO) at 37。C for 30 min with gentle agitation. An equal volume of DMEM/F12 com- plete media was added to stop the trypsinization and the brain tissues were washed three times in the same medium. The tissue was then triturated gently to prepare a single cell suspension and then passed through a 70 μm nylon mesh cellstrainer to remove tissue debris and aggregates. The cells were resuspended in DMEM/F12 complete medium and were seeded into T75 flasks. Cells were then incubated in a humidified CO2 incubator at 37。C. The media was changed after 4-5 days and the mixed glial cells were allowed to grow to the desired confluence. Separation of microglial cells were performed at 14-16 days with a column-free magnetic separation system as described in previous reports (Deora, Albornoz, Zhu, Woodruff, & Gordon, 2017; Gordon et al., 2011).
2.6 | Preparation and purification of recombinant SOD1 proteins
Human wild-type and mutant SOD1 protein in their native conforma- tions were produced in Escherichia coli using a plasmid containing both human SOD1 and theyeast copper chaperone for SOD1 as previously described (McAlary, Aquilina, & Yerbury, 2016). BL21 (DE3) E. coli grown in 2× LB were induced with the addition of IPTG to 0.5 mM and cultured at 18。C overnight in the presence of 3 mM CuSO4 and 200 μM ZnSO4 to ensure that native SOD1 was generated. Protein purification was performed by heat denaturation, ammonium sulfate precipitation followed by size exclusion chromatography (Superdex 75: GE Healthcare, UK), and anion exchange using a Q-Sepharose anion exchange column (GE Healthcare, UK) eluted with a salt gradi- ent of 0-125 mM NaCl pH 7.5.
2.7 | SOD1 aggregation and validation assays
Soluble SOD1 protein was diluted to 2 mg/ml with 1× PBS (pH 7.4) and then mixed with 100 mM DTT, 10 mM EDTA in PBS solution (pH 7.4) to give a final concentration of 1 mg/ml SOD1, 50 mM DTT, 5 mM EDTA. The aggregation vessel was laid on its side on an orbital shaker set at 200 rpm at 37。C. After 72 hr of shaking visible aggre- gate structures formed, and were separated from the solution by cen- trifugation and multiple washes in PBS (Roberts et al., 2013). Samples were then tested for aggregate formation using a ThT assay by incu- bating with ThT (16 μg/ml) for 30 min at room temperature and fluo- rescence intensity measured at 490 nm. Aggregate formation was further confirmed using transmission electron microscopy as previ- ously described (Anderson & Webb, 2011).
2.8 | TDP-43 protein generation
Codon optimized 6× Histidine tagged TDP-43WT, TDP-43A315T, and TDP-43Q331K cDNA sequences were cloned into LIC-A bacterial expression vector (Eschenfeldt, Lucy,Millard, Joachimiak, & Mark, 2009). Each construct was transformed and overexpressed in the E. coli. strain SHuffle (C3029J; New England Biolabs, Ipswich, Massa- chusetts) containing the constitutively expressing bacterial chaperone DsbC, which promotes disulfide bond formation and facilitates protein folding. The cells were harvested and resuspended in lysis buffer (20 mM Tris-HCL pH 8.0, 300 mM NaCl) with freshly added protease inhibitors (Roche 1169749800) and lysozyme at a working concentra- tion of 0.1 mg/ml. The lysates were further sonicated (10 cycles; pulser: 30%; power: ~40-50%) and centrifuged at 13,000 rpm and 4。C for 30 min. The pellets containing recombinant TDP-43 were resuspended in solubilization buffer (6 M Guanidinium hydrochloride, 20 mM Tris- HCL pH 8.0, 300 mM NaCl, 20 mM Imidazole). The samples were then incubated at room temperature on a rotary mixer for at least 2 hr, cen- trifuged at 13,000 rpm for 30 min at 4。C and the supernatant con- taining recombinant TDP-43 was then subjected to affinity chromatography on a Ni-NTA column (pre-equilibrated with solubiliza- tion buffer). The columns were washed with wash buffer (4 M Gua- nidinium hydrochloride [GuHCL], 20 mM Tris-HCl pH 8.0, 300 mM NaCl, and 20 mM imidazole), and further washes were applied at decreasing concentrations of GuHCL (4 M > 2 M > 1 M > 0.5 M > 0 M) to refold the denatured TDP-43 proteins on the column. Finally, each recombinant protein was eluted with elution buffer (20 mM Tris-HCl pH 8.0, 300 mM NaCl, and 250 mM imidazole) and its purity was assessed by SDS-PAGE analysis using Coomassie Brilliant Blue. The concentration of purified TDP-43 in each fraction was determined.
2.9 | ALS protein microglial treatments
After separation of primary microglia, cells were seeded at a density of 1*106 cells/ml in plates coated with poly-D-lysine. The cells were serum starved overnight, and were primed with ultrapure LPS (200 ng/ml) for 3 hr. The microglia were then stimulated with soluble and aggregated SOD1 protein for 4-48 hr, and supernatant and cell lysates collected for determination of inflammasome components using a mouse IL-1β ELISA kit (R&D Systems, Minneapolis, MN), and western blotting. Detection of cleaved caspase-1 p20 and cleaved IL-1β were observed in the supernatant through western blotting after concentration of pro- tein using methanol precipitation. For inhibitor assays, compounds were added 30 min prior to the treatment with ALS proteins. Unstimulated and LPS-primed cells were used as controls.
2.10 | OSU-03012 NLRP3 inhibitor (MCC950) synthesis
MCC950 was synthesized in 10 steps using published methodology and then converted to the monosodium salt using one equivalent of NaOH. The final compound was >98% pure by LC-MS and NMR, all spectra (final product and intermediates) were in agreement with liter- ature precedent (Urban et al., 2003).
2.11 | Western blot assays
Microglial cell samples were lysed using a modified RIPA buffer con- sisting of 10 mM Tris (pH 7.4), 100 mM NaCl, 1 mM EDTA 1 mM EGTA, 0.1% SDS, 0.5% sodium deoxycholate, 1% triton-X, 10% glyc- erol, phosphatase inhibitors (1 mM NaF, 20 mM Na4 P2O7, 2 mM Na3VO4) and protease inhibitors. The supernatant and microglial cell lysates were separated using Bio-Rad 4-20% precast gels and then transferred onto nitrocellulose membranes. Membranes were blocked in Odyssey blocking buffer (Licor, Lincoln, NE) for 1 hr and were incu- bated overnight with the following antibodies; mouse anti-caspase-1 (p20; AG-20B-0042-C100, Adipogen, San Diego, CA), goat anti- cleaved IL-1β (m118; sc-23460, Santa Cruz Biotechnology, Dallas, TX),mouse anti-NLRP3 (cryo-2, AG-20B-0014-C100, Adipogen), mouse anti-GAPDH (NB300-221, Novus Biologicals, Centennial, CO) and rabbit anti-ASC (AL177; AG-25B-0006, Adipogen) at 4。C temper- ature. All primary antibodies were diluted in Odyssey blocking buffer. Primary antibodies were detected with IRDye® 680RD donkey anti- goat IgG, donkey anti-mouse IgG, donkey anti-mouse IgG, and IRDye® 800CW donkey anti-mouse IgG for 1 hr at room temperature. These secondary antibodies were detected using a Licor Odyssey CLx Imag- ing System and Image Studio Lite was used for Western blot analysis.
2.12 | Immunocytochemistry
Primary microglia were fixed with 4% paraformaldehyde in PBS for 15 min. Microglia were then blocked with 2% BSA in PBS with 0.1% Triton-X and 0.05% Tween-20 for 1 hr at room temperature. Follow- ing this, cells were incubated with ASC and NLRP3 antibodies over- night at room temperature. These cells were washed three times with PBS prior to incubation with an appropriate Alexa conjugated second- ary antibody for 1 hr at room temperature. All primary and secondary antibodies were diluted in 2% BSA in PBS with 0.1% Triton-X and 0.05% Tween-20. Following three washes in PBS, the cells were incu- bated with DRAQ5 before being mounted with Prolong Gold Anti- Fade medium (Invitrogen) for imaging.
2.13 | Lactate dehydrogenase (LDH) assay
Supernatants from LPS-primed microglia were collected after 48 hr of incubation of soluble and aggregated SOD1G93A protein (100 μg/ml), or 1 hr after nigericin (20 μM) incubation. LDH release was quantified using an LHD assay kit (Sigma) as per manufacturer protocol (Rayamajhi, Zhang, & Miao, 2013). Cytotoxicity was calculated as a percentage of LDH in total cell lysates.
2.14 | Reactive oxygen species (ROS) assay
To determine the level of ROS, primary microglial cells were plated on black 96-well plate and primed for 3 hr with LPS (200 ng/ml). Microglia were then stimulated with both soluble and aggregated form of SOD1G93A for 48 hr at 37。C. The microglia were stained with 5 μM of CellROX® Deep Red Reagent (C10422; Life Technologies, Carlsbad, CA) for 30 min at 37。C. Finally, the microglia cells were washed with PBS and the fluorescence measured at 644/665 nm.
3 | RESULTS
3.1 | NLRP3 inflammasome components are
upregulated in the SOD1G93A mouse model of ALS
NLRP3 is predominantly expressed in myeloid cells (Guarda et al., 2011), and are upregulated in microglia in mouse models, and human post-mortem brains from Alzheimer’s disease and Parkinson’s disease (Gordon et al., 2018; Heneka et al., 2013). Surprisingly however, a recent study indicated that microglia from ALS patients and SOD1G93A ALS mice, do not express NLRP3 (Johann et al., 2015). We therefore revisited this finding by examining NLRP3 inflammasome components and cytokine targets in the SOD1G93A mouse model of ALS. First, we demonstrated that spinal cord tissue from end-stage SOD1G93A mice show significantly upregulated expression of inflammasome pathway genes Il-1β (17-fold increase), Nlrp3 (eight- fold increase), Pycard (the gene expressing ASC; four-fold increase), and Caspase-1 (two-fold increase), compared to age-matched wild- type mice (Figure 1a-d). We also confirmed expression of the Nlrp3 locus in both microglia and astrocytes of SOD1G93A ALS mice by uti- lizing a mouse line endogenously expressing GFP under the control of Nlrp3 regulatory elements. These Nlrp3-GFP knock-in mice were bac- kcrossed to hSOD1G93A mice, and spinal cord tissue examined for GFP expression at end-stage disease. We identified strong GFP signal on both GFAP-positive astrocytes, as well as CD11b-positive microglia (Figure 1e-h). These results demonstrate that microglia
FIGURE 1 NLPR3 is expressed by microglia and is upregulated in the hSOD1G93A ALS mouse model. (a-d) mRNA expression of (a) Il-1β,
(b) Nlrp3, (c) Pycard, and (d) Casp1 in the spinal cord of hSOD1G93A mice (SOD1) relative to age matched wild-type mice (WT) at 155-170 days of age. Data are expressed as means ± SEM (n = 6, *** p < .001, unpaired Student's t test). Panels e-h show double labeling of Nlrp3-GFP (green in e and g) with cellular markers (red) for microglia (CD11b; f) and astrocytes (GFAP; h) in the ventral lumbar spinal cord of hSOD1G93A × Nlrp3-GFP reporter mice at 155-170 days of age. Nlrp3-GFP colocalizes with CD11b-positive microglia (e and f, yellow arrows) and GFAP-positive
astrocytes (g and h, yellow arrow). Scale bars for all panels = 10 μm [Color figure can be viewed at wileyonlinelibrary.com]
FIGURE 2 Soluble and aggregated SOD1G93A dose-dependently increase IL-1β secretion in primary microglia. (aand b) Transmission
electron microscopy images for soluble and aggregated SOD1G93A protein (black arrow indicating amyloid-like structures; Scale bar = 100 nm). (c) ThT fluorescence intensity for soluble and aggregated SOD1G93A protein (n = 4, *** p < .001, one-way ANOVA with post hoc Dunnett's test).
(d-f) Soluble SOD1G93A increases IL-1β production at 50, 100, and 200 μg/ml concentrations determined by ELISA and western blotting of
supernatants after 48 hr incubation (n = 3-4, ** p < .01 and *** p < .001, one-way ANOVA with post hoc Dunnett's test). (g-i) Aggregated
SOD1G93A produces higher concentrations of IL-1β at 48 hr compared to soluble forms at 50, 100, and 200 μg/ml, determined by ELISA and western blotting of supernatants (n = 3, ** p < .01 and *** p < .001, one-way ANOVA with post hoc Dunnett's test). (j) Aggregated SOD1G93A alone and BSA do not generate IL-1β, and boiling soluble SOD1G93A significantly reduces IL-1β (n = 3-4, *** p < .001 compared to LPS; ##
p < .01 compared to SOD1G93A, one-way ANOVA with post hoc Dunnett's test). (k) Soluble and aggregated SOD1WT protein (100 μg/ml)
increases IL-1β production, and (l) soluble and aggregated SOD1G93A (100 μg/ml) are unable to generate IL-18 from LPS-primed microglia after 48 hr (n = 4, * p < .05, ** p < .01, *** p < .001, one-way ANOVA with post hoc Dunnett's test). ATP (5 mM) was used as a positive control for inflammasome activation. All data are expressed as mean ± SEM [Color figure can be viewed at wileyonlinelibrary.com]
FIGURE 3 SOD1G93A induces ASC speck formation and generates IL-1β in primary microglia through caspase-1 activation. (a) Soluble and
(b) aggregated SOD1G93A (100 μg/ml) increases IL-1β production at 24 and 48 hr determined by ELISA (n = 4, ** p < .01 and *** p < .001, one-way ANOVA with post hoc Dunnett's test). (c andd) Pretreatment of microglia with the caspase inhibitor Ac-YVAD-cmk (20 μM) reduces IL-1β production in response to soluble and aggregated SOD1G93A at 48 hr (n = 4-6, *** p < .001, one-way ANOVA with post hoc Tukey's multiple comparison test).
(e) Double immunolabeling of NLRP3 (red) and ASC (green) in control microglia (LPS primed only), and microglia treated with soluble or aggregated SOD1G93A (white arrows indicate ASC speck formation; scale bar = 5 μm). (f) LDH release (marker of pyroptosis) was measured in primary microglia treated with soluble or aggregated SOD1G93A (48 hr incubation), or the positive control, nigericin (20 μM, 1 hr incubation; n = 3, *** p < .001, one- way ANOVA with post hoc Dunnett's test). Data are expressed as mean ± SEM [Color figure can be viewed at wileyonlinelibrary.com] the SOD1G93A mouse model do indeed express NLRP3 with elevated levels observed in the pathological state, suggesting a functional role for the NLPR3 pathway in ALS mice.
3.2 | Soluble and aggregated SOD1G93A induce
inflammasome-mediated secretion of mature IL-1β in primary microglia
Next, we aimed to examine the response spine oncology of microglia to ALS protein aggregates, focusing initially on mutant SOD1G93A, as this protein is known to induce microglial activation (Zhao et al., 2010), and is present in the SOD1G93A ALS mouse model where we identified microglial NLPR3 expression (Figure 1c). We first characterized the preparations of soluble and aggregated SOD1G93A by examining the morphology of the SOD1G93A aggregates using transmission electron microscopy (TEM), which confirmed the formation of fibrillar aggre- gates when compared to soluble (natively folded) SOD1G93A protein (Figure 2a,b). Aggregate formation was further confirmed using a ThT spectroscopic assay, where aggregated SOD1G93A protein prepared with DTT and EDTA generated a significantly higher fluorescence sig- nal compared to the soluble form (Figure 2c), indicating fibrillar aggre- gates were present.To assess inflammasome activation with these proteins, primary microglial cells were treated with ultrapure lipopolysaccharide (LPS; 200 ng/ml) for 3 hr to “prime” the inflammasome. Priming was con- firmed by the presence of pro-IL-1β in cell lysates (Figure 2e). Cells were then washed, and fresh media containing either soluble or aggre- gated SOD1G93A (50 –200 μg/ml) was added for 48 hr. Microgliastim- ulated with soluble SOD1G93A released IL-1β into the supernatantin a dose-dependent manner, as measured by ELISA (Figure 2d). Inflammasome-mediated cleavage of IL-1β was confirmed by immuno- blotting of cell supernatants, demonstrating dose-dependent produc- tion of cleaved (i.e., mature) IL-1β (Figure 2e,f). Microglia stimulated with aggregated SOD1G93A also secreted IL-1β, although the magni- tude of IL-1β was higher than that observed for the soluble species (Figure 2g). Inflammasome activation by SOD1G93A aggregates was also confirmed using immunoblotting, demonstrating dose-dependent production of cleaved IL-1β in the supernatant (Figure 2h,i). We con- firmed the requirement for intact SOD1G93A protein in the inflammasome activation response, by denaturing the protein through boiling for 10 min prior to the assay, which completely abolished its capacity to activate inflammasome responses (Figure 2j). Addition of an equal concentration of a irrelevant protein (BSA) to primed microglia was also not able to activate cells (Figure 2j).
Furthermore, addition of SOD1G93A alone (in the absence of priming) was unable to generate IL-1β (Figure2j),demonstrating a requirement for inflammasome priming in the SOD1G93A-mediated inflammasome response. Interestingly, addition of SOD1WT soluble and aggregated protein to microglia was also able to generate IL-1β (Figure 2k). Finally, we also examined if SOD1G93A protein could additionally induce secretion of IL-18, another cytokine target of the inflammasome. Whilst ATP could generate IL-18 release, neither solu- ble nor aggregated SOD1G93A induced IL-18 secretion under the con- ditions tested (Figure 2l).Microglial cells were next treated with SOD1G93A proteins and the release of IL-1β measured at various time points following exposure. IL-1β secretion progressively increased in a time-dependent manner, for both soluble and aggregated forms; however, a minimum 24 hr incubation was required to detect IL-1β secretion, indicating delayed inflammasome activation kinetics (Figure 3a,b). Inflammasome signal- ing requires activation of caspase-1 in order to induce IL-1β cleavage and secretion (Martinon, Petrilli, Mayor, Tardivel, & Tschopp, 2006). Thus, we assessed the caspase-1 dependency of SOD1G93A-mediated IL-1β production by pre-incubating microglia with the caspase-1 inhib- itor, ac-YVAD-fmk (20 μM). Caspase inhibition abolished the release of IL-1β induced by both soluble, and aggregated proteins (Figure 3c, d), confirming caspase dependency for the microglial IL-1β secretion response to SOD1G93A.
FIGURE 4 SOD1G93A activates microglial caspase-1 and generates IL-1β specifically through the NLRP3 inflammasome pathway. (a) Primary mouse microglia pre-treated with the NLRP3 inflammasome inhibitor, MCC950 (100 nM), significantly attenuates IL-1β secretion in response to soluble or aggregated SOD1G93A (100 μg/ml; n = 6 –8, *** p < .001, one-way ANOVA with post hoc Tukey's multiple comparison test).
(b) Western blot and (c,d) densitometry quantification of inflammasome components (caspase-1 and IL-1β) in supernatants of soluble and
aggregated SOD1G93A treated microglia following MCC950 treatment (n = 3, * p < .05, ** p < .01, *** p < .001, one-way ANOVA with post hoc Tukey's multiple comparison test). (e) Microglia cultured from Nlrp3−/− mice are unable to secrete IL-1β in response to soluble and aggregated SOD1G93A protein (100 μg/ml). Wild-type (WT) microglia were included as a control in the same experiment (n = 4, *** p < .001 for Nlrp3−/−
versus WT for soluble and aggregated SOD1G93A, one-way ANOVA with post hoc Tukey’s multiple comparison test). (f) Western blot confirms a lack of caspase-1 activation, mature IL-1β, and NLRP3 formation in Nlrp3−/− microglia. Data are expressed as mean ± SEM [Color figure can be viewed at wileyonlinelibrary.com]
In addition to the maturation and secretion of IL-1β, inflammasome -mediated caspase-1 activation can induce an inflammatory form of cell death known as pyroptosis (Gustin et al., 2015; Patel et al., 2017). To determine whether pyroptosis was also induced by anatomical pathology SOD1G93A proteins upon caspase-1 activation, we measured LDH release in the culture supernatants as a marker for cell rupture. Microglia treated for 48 hr with both soluble and aggregated SOD1G93A demonstrated no increased production of LDH compared to LPS-primed cells (Figure 3e), indicating that cell membrane integrity is not significantly compromised upon SOD1G93A mediated inflammasome activation. By contrast, treat- ment with the known inflammasome-mediated pyroptosis inducer nigericin (Gustin et al., 2015), produced a significant increase in LDH release (Figure 3e), demonstrating that microglia are capable of under-going significant pyroptosis through inflammasome activation, but this is not detectable when caspase-1 is activated in response to SOD1G93A.Another key event in the activation of many inflammasomes is the formation of ASC specks (Sester et al., 2016). We, therefore, performed immunocytochemistry experiments to identify ASC speck formation in response to SOD1G93A protein. We observed consistent ASC speck for- mation in microglia incubated with both soluble and aggregated SOD1G93A, while no ASC-specks were formed when cells were primed with LPS alone (Figure 3f-h).
3.3 | Soluble and aggregated SOD1G93A specifically activate the NLRP3 inflammasome complex
The NLRP3 inflammasome is emerging as a key driver of microglial IL- 1β secretion in response to brain-derived pathogenic proteins, yet sur- prisingly, a prior study reported that SOD1G93A mediated caspase-1 activation in microglia was independent of NLRP3 (Meissner)
FIGURE 5 SOD1G93A activates microglial
inflammasomes through reactive oxygen species (ROS) production and PTX7 receptors. (a) Soluble and aggregated SOD1G93A (100 μg/ml) significantly increases ROS production in primary microglia. Hydrogen peroxide (H2O2) was used as a positive control (n = 6, *** p < .001, one-way ANOVA with post hoc Tukey's multiple comparison test). (band c) Microglia treated with the ROS inhibitor AP-DC (100 μM) prior to soluble or aggregated SOD1G93A administration, had significantly decreased IL-1β production at 48 hr (n = 4, *** p < .001, one-way ANOVA with post hoc Tukey's multiple comparison test). (dande) Microglia pretreated with the P2X7 receptor inhibitor, A438079 (10 μM), showed reduced IL-1β generation in response to soluble and aggregated SOD1G93A protein at 48 hr (n = 3-4, *** p < .001, one-way ANOVA with post hoc Tukey's multiple comparison test). Data are expressed as mean ± SEM [Color figure can be viewed at wileyonlinelibrary.com] Molawi, & Zychlinsky, 2010). We first aimed to confirm these findings by pharmacologically inhibiting NLRP3 activity. MCC950 is a potent and selective small molecule inhibitor of the NLRP3 inflammasome, which inhibits IL-1β release in response to diverse activation signals (Coll et al., 2015; Gordon et al., 2018; Guo, Callaway, & Ting, 2015). Microglia were therefore primed, and then treated with MCC950 (100 nM) prior to the addition of SOD1G93A proteins. Treatment with MCC950 markedly reduced both soluble and aggregated SOD1G93A induced release of IL-1β into the supernatant (Figure 4a)). This was also demonstrated using immunoblotting, where NLRP3 inhibition with MCC950 significantly inhibited inflammasome-mediated cleav- age of caspase-1 and IL-1β, without affecting the pro-forms of these proteins, or NLRP3 expression (Figure 4b-d).
Given that these findings seem to contradict a prior report (Meissner, Molawi, & Zychlinsky, 2010), we next further confirmed the NLRP3-dependency for SOD1G93A-induced inflammasome activa- tion in our assay by utilizing microglia obtained from Nlrp3−/− mice. We first performed a direct comparative assay between wild-type and NLRP3-deficient microglia, and measured the release IL-1β in response to soluble and aggregated SOD1G93A. Confirming the phar- macological inhibition studies, Nlrp3−/− microglia cells were completely unresponsive to SOD1G93A in this IL-1β secretion assay (Figure 4e), and immunoblot for cleaved caspase-1 and IL-1β (Figure 4f). Finally, we also performed dose-response and temporal studies to rule out any potential differences in the kinetics of activa- tion in NLRP3-deficient microglia; however, we failed to detect any measurable IL-1β secretion following SOD1G93A at any of the doses or time points measured (Figure S1). Overall, these results clearly show that these SOD1G93A proteins specifically activate the NLRP3 inflammasome complex to induce caspase-1 activation and IL-1β cleavage and secretion.
3.4 | Mutant SOD1G93A activates the NLRP3 inflammasome through the generation of ROS
The mechanisms driving mutant SOD1G93A toxicity in ALS are incom- pletely understood; however, there is strong evidence suggesting that mutant SOD1 may induce cellular oxidative stress in microglia, leading to neurotoxicity (Apolloni et al., 2013). Since ROS are also known to activate NLRP3 in macrophages (Gross et al., 2016), we thus explored whether SOD1G93A-mediated ROS generation was required to induce NLRP3 inflammasome activation in microglia. We first measured the levels of microglial ROS generation following SOD1G93A incubation, and found that both soluble and aggregated forms of the protein markedly induced cellular ROS (Figure 5a). Next, to confirm whether ROS generation by SOD1G93A was required for inflammasome activa- tion, we utilized a ROS inhibitor, AP-DC, and measured IL-1β secre- tion after incubating cells with SOD1G93A proteins. AP-DC (100 μM) suppressed the secretion of IL-1β in response to soluble and aggre- gated SOD1G93A, indicating that ROS is required for microglial NLRP3 inflammasome signaling (Figure 5b,c). Microglial NLPR3 activation can also occur through activation of the purinergic receptor P2X7, as a consequence of cell-mediated ATP secretion (Gombault, Baron, & Couillin, 2012), and so we also tested the impact of purinergic
FIGURE 6 Inflammasome
components are upregulated in TDP- 43Q331K mice, and TDP43 proteins activate the microglial NLPR3 inflammasome. (a-c) mRNA expression of (a) Nlrp3, (b) Pycard, and (c) Casp1, in the spinal cord of TDP43Q331K mice (TDP-43) relative to age matched wild-type mice (WT) at 16 months of age (n = 6, ** p < .01, *** p < .001, unpaired Student's t test). (b) Primary mouse microglia pre-treated with NLRP3 inflammasome inhibitor, MCC950 (100 nM), significantly attenuates IL- 1β secretion in response to recombinant aggregated TDP43WT, TDP43A315T, and TDP43Q331K incubation for 48 hr (100 μg/ml; n = 4-6, * p < .05, *** p < .001, unpaired Student's t test). Data are expressed as mean ± SEM [Color figure can be viewed at wileyonlinelibrary.com] receptor inhibition using A438079 (10 μM) on SOD1G93A inflammasome activation. P2X7 receptor-inhibited microglia produced significantly less IL-1β following soluble and aggregated SOD1G93A incubation (Figure 5d,e), although not to the extent of ROS inhibition, indicating a partial role for this receptor in NLRP3 inflammasome acti- vation induced by this ALS protein.
3.5 | NLRP3 inflammasome components are upregulated in TDP-43Q331K ALS mice, and TDP-43 proteins can activate microglial NLPR3 inflammasomes
The SOD1G93A mutation represents a small subset of ALS cases, and so to confirm whether the microglial NLRP3 inflammasome pathway could be relevant for more widespread ALS cases, we investigated NLRP3 inflammasome components in a TDP-43 based ALS model. Spinal cord gene expression of the inflammasome components NLRP3, caspase-1, and ASC were significantly increased in TDP- 43Q331K mice, compared to their wild-type littermates (Figure 6a-c). We also examined whether wild-type and various mutant forms of TDP-43 could also activate microglial inflammasomes and generate IL-1β similar to SOD1G93A. Both wild-type and mutant (A315T, Q331K) recombinant forms of TDP43 protein activated primed microglia to generate IL-1β, which was abolished by MCC950 pre- treatment (Figure 6d). These data suggest a common pathway of microglial NLPR3 inflamamasome activation between distinct ALS pathological proteins.
4 | DISCUSSION
Innate immune-mediated neuroinflammation occurs during ALS, and is believed to contribute to propagation of the disease by inducing reactive gliosis and the upregulation of proinflammatory and neuro- toxic molecules (Hensley et al., 2006; Nguyen, Julien, & Rivest, 2001). NLRP3 inflammasome dysregulation is linked to multiple neurodegen- erative diseases (Gordon et al., 2018; Heneka et al., 2013; Song, Pei, Yao, Wu, & Shang, 2017), and drives neuroinflammation through caspase-1 and IL-1β activation. Caspase-1 and IL-1β have previously been identified as key pathogenic drivers of ALS (Friedlander, Brown, Gagliardini, Wang, & Yuan, 1997; Li et al., 2000; Meissner, Molawi, & Zychlinsky,2010), and prior studies have demonstrated inflammasome activation and upregulation of NLRP3 in both human ALS, and SOD1G93A rodent models (Bellezza et al., 2018; Debyeet al., 2018; Gugliandolo, Giacoppo, Bramanti, & Mazzon, 2018; Johann et al., 2015), suggesting that the NLRP3 inflammasome could play a key role in ALS disease pathology. Surprisingly, however, a previous study showed that although SOD1G93A activated caspase-1 and pro- duced mature IL-1β in microglia, this occurred independently of NLRP3 inflammasomes (Meissner, Molawi, & Zychlinsky, 2010). Here, we revisited this, using the specific NLRP3 inhibitor MCC950, and NLRP3 knockout microglia. We identified that microglial SOD1G93A-mediated IL-1β release was indeed through NLRP3-dependent inflammasome pathways, and required ROS.
We demonstrate that Nlrp3, Pycard, Casp1, and Il-1β gene expres- sion are upregulated in SOD1G93A mice, supporting prior findings (Johann et al., 2015). We also provide the first report of increased inflammasome gene expression in TDP-43Q331K mice. Microglial NLRP3 expression is well documented in mouse models and patients suffering from Alzheimer’s disease and Parkinson’s disease (Gordon et al., 2018; Heneka et al., 2013), so it is somewhat surprising that a prior study was unable to detect NLPR3 expression in microglia from SOD1G93A mice. In our study, we demonstrated both microglia and astrocytes express the Nlrp3 locus in end-stage(P155-170) SOD1G93A mice by utilizing a Nlrp3-GFP knock-in mouse, which expresses GFP under the control of endogenous Nlrp3 regulatory ele- ments and negates the need for an anti-NLRP3 antibody. One possi- bility for this discrepancy is that the prior study utilized mice at presymptomatic (P63) and early symptomatic (P98) ages, which may have missed a later-age microglial NLRP3 upregulation in line with reactive gliosis observed in this model.
This study utilized both soluble and aggregated forms of SOD1G93A with both demonstrating delayed activation of NLRP3 inflammasomes. The aggregated form of SOD1G93A, however, pro- duced a greater degree of activation, as demonstrated by higher supernatant concentrations of IL-1β, and by the fact that lower doses of aggregates were required to generate IL-1β, indicating that aggre- gates may have greater neuroinflammatory and neurotoxic potential. This supports previous findings demonstrating that SOD1 aggregate formation is linked to accelerated disease progression in ALS patients (Prudencio, Hart, Borchelt, & Andersen, 2009; Seetharaman et al., 2009). Importantly, microglial inflammasome activation was abolished when SOD1G93A protein was heat-inactivated, indicating that intact SOD1 protein is required for specific NLRP3 activation. SOD1WT pro- tein was also demonstrated to activate the inflammasome in microglia, suggesting that the SOD1 mutation is not a specific requirement for NLRP3 activation. Notably, there is emerging evidence that SOD1WT forms aggregates and may play a role in sporadic and non-SOD1 familial ALS pathology (Forsberg et al., 2019; Pare et al., 2018; Rot- unno & Bosco, 2013), so could represent a potential additional source of pathogenic ALS protein in humans. Our study also confirmed an essential role for microglial NLRP3 inflammasomes in generating IL-1β in response to both soluble and aggregated SOD1G93A, where phar- macological inhibition or genetic deletion of NLRP3 completely inhibited the maturation of IL-1β. This result contrasts with a prior study which indicated that SOD1G93A mediated inflammasome activa- tion was NLRP3-independent (Meissner, Molawi, & Zychlinsky, 2010). Primary microglia treated with TDP-43Q331K, TDP-43A315T, and TDP- 43WT protein aggregates also activated inflammasomes, and extended from the work of Zhao and colleagues by demonstrating this response is NLPR3-dependent (Zhao et al., 2015).TDP-43 spontaneously aggregates once generated, so we were unable to test whether solu- ble TDP-43 forms would similarly activate microglial NLRP3, as we observed with SOD1 wild-type and mutant proteins.
The discrepancy between our study and the study by Meissner, Molawi, and Zychlinsky (2010), could be due to the difference in the cells utilized.Although the Meissner study demonstrated that SOD1G93A could activate inflammasomes in primary mouse microglia (as measured by cleaved caspase-1 generation and IL-1β secretion), the experiment examining NLRP3 involvement was conducted in Nlrp3−/− bone marrow-derived macrophages. Although macrophages and microglia have many similarities in phenotype and function, prior studies have demonstrated distinct differences in inflammasome acti- vation between these cell types (Burm et al., 2015). Our studies using both pharmacological (MCC950) and genetic (Nlrp3−/−) approaches now clearly demonstrate that SOD1 and TDP-43 proteins induce inflammasome activation in primary mouse microglia through NLRP3 signaling. One of the keydownstream events of NLRP3 inflammasome activation is pyroptosis, a process of programmed cell death distinct from apoptosis, mediated through activation of caspase-1 (Harijith, Ebenezer, & Natarajan, 2014; Martin-Sanchez et al., 2016).The present study demonstrates that while microglia are capable of undergoing pyroptosis, SOD1G93A mediated NLRP3 inflammasome activation does not induce detectable pyroptosis (as measured through LDH release) under the conditions tested. This suggests that activation of NLRP3 inflammasomes by SOD1G93A is controlled predominantly by a nonlytic mechanism, similar to an alter- native inflammasome pathway recently described (Gaidt et al., 2016), and recently documented for α-synuclein (Gordon et al., 2018). In sup- port of this, SOD1G93A was also unable to generate IL-18 under the conditions tested, despite robust IL-1β and IL-18 generation from ATP-stimulated microglia.
We also demonstrated that SOD1G93A induces oxidative stress (i.e., ROS production) in primary microglia. This finding is in agreement with previous studies demonstrating that mutant SOD1 proteins cause oxidative stress in other cell types (Pollari, Goldsteins, Bart, Koistinaho, & Giniatullin, 2014). The aggregated form of SOD1G93A also produced greater oxidative stress, which may explain the higher level of IL-1β generation observed. Collectively, these findings suggest that increases in microglial ROS can induce IL-1β production by trig- gering NLRP3 inflammasome assembly.The direct interaction between ROS production and inflammasome activation was also con- firmed, where inhibition of ROS diminished IL-1β production in response to mutant SOD1 protein. This is in agreement with previous studies, where other neurodegenerative protein aggregates including β-amyloid and α-synuclein were also observed to cause oxidative stress, which in turn increased IL-1β production (Codolo etal., 2013; Parajuli et al., 2013). A partial role for cellular ATP release and purinergic P2X7 receptor activation was also revealed, supporting prior microglial activation studies using β-amyloid (Sanzetal., 2009).
In summary, the present study confirms that SOD1G93A ALS mouse microglia express NLPR3 protein, in line with increased gene expression of inflammasome components in vivo. Mutant and wild- type SOD1 directly activate caspase-1 dependent inflammasomes in primary microglia. Pharmacological and genetic approaches confirmed that microglial inflammasome activation by SOD1G93A was entirely driven through NLRP3 inflammasome-dependent pathways, and was reliant on the generation of ROS and ATP. Mutant and wild-type TDP-43 proteins also activated microglial inflammasomes in a NLPR3-dependent manner, with TDP-43Q331K mice also expressing increased inflammasome components during disease progression. Given NLRP3 upregulation and inflammasome activation is observed in both ALS patients andALS mouse models, our data provide a mech- anistic link between protein aggregate formation and NLRP3 inflammasome activation. Furthermore, the confirmed role for caspase-1 activation and IL-1β in ALS pathogenesis (Meissner, Molawi, & Zychlinsky, 2010), suggests that the microglial NLRP3 inflammasome could be a key upstream pathway triggered by neuro- toxic ALS proteins, that drives neuroinflammation.Therapeutic targeting of this innate immune pathway with selective inhibitors, such as MCC950, may thus be a potential approach to slow propaga- tive neuroinflammatory cell death in ALS and ameliorate disease progression.
