Brazilin is a natural product inhibitor of the NLRP3 inflammasome

Summary Excessive or aberrant NLRP3 inflammasome activation has been implicated in the progression and initiation of many inflammatory conditions; however, currently no NLRP3 inflammasome inhibitors have been approved for therapeutic use in the clinic. Here we have identified that the natural product brazilin effectively inhibits both priming and activation of the NLRP3 inflammasome in cultured murine macrophages, a human iPSC microglial cell line and in a mouse model of acute peritoneal inflammation. Through computational modeling, we predict that brazilin can adopt a favorable binding pose within a site of the NLRP3 protein which is essential for its conformational activation. Our results not only encourage further evaluation of brazilin as a therapeutic agent for NLRP3-related inflammatory diseases, but also introduce this small-molecule as a promising scaffold structure for the development of derivative NLRP3 inhibitor compounds.


INTRODUCTION
Inflammasomes are cytosolic, multimeric protein complexes which assemble within cells in response to danger signals evoked by tissue injury and/or infection.Inflammasomes serve a dual purpose of facilitating the maturation and release of pro-inflammatory cytokines interleukin (IL)-1b and IL-18 and initiating a lytic form of cell death known as pyroptosis.Several inflammasomes have been identified, each named after a unique pattern-recognition receptor (PRR) protein which recognizes pathological stimuli and initiates inflammasome assembly.Such PRRs include absent-in-melanoma 2 (AIM2), pyrin and members of the nucleotide-binding oligomerization domain (NOD), leucine-rich repeat (LRR)-containing protein family ('NLR'); NLRP1, NLRP3, NLRP6 and NLRC4. 1 The NLRP3 inflammasome is the most widely characterized of all inflammasomes and has been the focus of our current research.
Canonical NLRP3 inflammasome activation is governed by a dual signaling mechanism.The first 'priming' signal involves extracellular tolllike receptor (TLR) or cytokine receptor activation, which initiates nuclear factor-kB (NF-kB)-dependent transcription of the NLRP3 protein and several pro-inflammatory cytokines including pro-IL-1b. 2,3A second 'activation' stimulus is then required to trigger inflammasome assembly.A range of damage-or pathogen-associated molecular patterns (DAMPs/PAMPs, respectively) have been found to activate the NLRP3 inflammasome, including crystalline/particulate substances (e.g., uric acid crystals or silica), extracellular ATP 2,4 and signals associated with viral, 5 fungal 6 and bacterial 7 infection.These activating stimuli evoke various cellular stress responses, including K + efflux, calcium influx, reactive oxygen species production, mitochondrial DNA release, lysosomal destabilisation 8,9 and a disruption of endosomal trafficking, 10 which ultimately trigger the oligomerization of NLRP3 proteins.This provides a platform for the polymerization of apoptosis-associated speck-like protein containing a CARD (ASC), creating filament amalgamations known as ASC specks, which are considered a hallmark of inflammasome activation.Inflammasome assembly is complete with the recruitment of pro-caspase-1 enzymes, which undergo proximity-induced autocatalytic activation to form mature caspase-1 proteases. 8,9Caspase-1 then catalyzes the maturation of IL-1b and IL-18 cytokines and the cleavage of gasdermin-D (GSDMD) proteins.GSDMD N-terminal fragments insert into the cell membrane and create pores, facilitating the extracellular release of mature cytokines. 11,12][15][16]

OPEN ACCESS
Appropriate inflammasome activation and production of pro-inflammatory IL-1b is an essential part of a protective host immune response, however, dysregulation of such a response can potentiate or even initiate disease pathologies.8][19] Inhibition of the NLRP3 inflammasome is a promising therapeutic strategy to attenuate inflammation-associated pathogenesis of such life altering and often under-treated conditions.MCC950, otherwise known as CRID3 or CP-456,773, is one of the first known and well-studied small-molecule inhibitors of the NLRP3 inflammasome. 20,21MCC950 potently and specifically inhibits NLRP3 inflammasome formation by binding to a site within the NLRP3 NACHT domain.][24] However, concerns surrounding liver toxicity have prevented any clinical application of MCC950. 18As such, the development of new, safe NLRP3 inflammasome inhibitors remains an active and evolving area for drug discovery.There is now a growing library of NLRP3 inhibitor compounds which differ in their mechanisms of action and potential clinical applicability, with some currently in or having completed early-phase clinical trials. 25ioactive compounds from plants present an important avenue for drug discovery, with such natural products potentially carrying a lower risk of severe adverse effects and lower developmental costs relative to synthetic compounds.Brazilin (Figure 1A) is a small-molecule isoflavonoid (MW 286.28) which occurs naturally as the (S) stereoisomer (reference to brazilin here refers to this isomer).This compound yields a red colored brazilein product upon oxidation (Figure 1B).Brazilin is highly abundant in the heartwood of several tree species, including brazilwood (Caesalpinia echinate), sappan (Caesalpinia sappan) and Mexican logwood (Haematoxylum brasiletto). 26Heartwood extracts from these trees have been widely used as colorants in foods, fabrics and cosmetics and have long been used as traditional remedies for a range of conditions, including dermatological problems, diabetes, fever and cancer. 27,28Preclinical research has described several pharmacological activities of brazilin, including an ability to attenuate the aggregation and cytotoxicity of amyloid beta protein, 29 limit cancer cell proliferation 30 and reduce inflammation.Brazilin has been shown to reduce the production of pro-inflammatory mediators by cultured macrophages in response to bacterial lipopolysaccharide [31][32][33] and to exert an anti-inflammatory effect in rodent models of renal injury, significantly decreasing tumor necrosis factor (TNF) a and IL-1b expression in the tissue 34 and increasing anti-inflammatory cytokine production. 35However, to the best of our knowledge, the specific effect of brazilin on inflammasome responses has not yet been investigated.
Here we report that brazilin is an inhibitor of the NLRP3 inflammasome.Molecular modeling and simulation proposes that brazilin can adopt a favorable pose in the inhibitor binding site of the NLRP3 NACHT domain; the same site to which the selective NLRP3 inhibitor MCC950 binds to hold NLRP3 in an inactive conformational state.We show that brazilin can significantly inhibit the NLRP3 inflammasome response in primary murine bone marrow-derived macrophages (BMDMs) and induced pluripotent stem cell (iPSC)-derived human microglia.We also report that brazilin treatment can significantly attenuate the NLRP3 inflammasome response in vivo, using a mouse model of acute peritoneal inflammation, highlighting the potential effectiveness of using brazilin for the development of new therapeutics targeting NLRP3dependent disease.

Molecular modeling predicts a direct interaction of brazilin with NLRP3
Firstly, we evaluated the potential for brazilin to bind favorably to the NLRP3 protein using computational modeling.Interestingly, due to its fused ring structure, the conformer of brazilin forms a rigid, twisted L-shaped structure (Figure 1A).For comparison, the binding modes of the oxidized form, brazilein (Figure 1B), were similarly characterized.24]36 To evaluate its potential to bind to NLRP3, brazilin was docked in both the cofactor and the inhibitor binding sites, with subsequent refinement by molecular dynamics simulation.
Brazilin was predicted via docking to bind to the inhibitor binding site in the NACHT domain of NLRP3 in two distinct binding modes, denoted I and II, occupying opposite ends of the site (Figure 1D).After MD simulation of the docked pose I, with ADP bound to the cofactor site, the OH groups of brazilin form hydrogen bonds with residues Asp662, Ile370 and Glu369 of NLRP3 (Figure 1E).MD simulation of pose II of brazilin optimized hydrogen bonds with Thr439 as well as Ala228 and Arg578; the latter two amino acids are identified as key amino acids for binding of MCC950 and its analogs in the NACHT domain X-ray structure of NLRP3 (Figure 1F, and Video S1).These poses are predicted to be similar in affinity, based on computed binding free energies using these trajectories, with values of À31.6 and À31.7 kcal/mol for poses I and II, respectively (Table 1).These values are rather lower in affinity than the predicted free energy of nanomolar ligand NP3-146 (Table 1); however, the similarity in ligand efficiency of NP3-146 and brazilin (Table 1) does suggest a favorable interaction of brazilin with the inhibitor site of NLRP3.
For comparison, brazilin was modeled into the cofactor binding site using the same docking/MD protocol.The ligand made a range of interactions in the site, including a hydrogen bond between an OH group of brazilin and the backbone N atom of Thr169 (Figure 1G and 1H); and several hydrophobic residue contacts (Table S1).However, the computed binding free energy for brazilin in this site was rather less than for the inhibitor site, with a value of À20.2 kcal/mol (Table 1).
Finally, the oxidized form of brazilin, namely brazilein, was docked into the two binding sites on the NLRP3 NACHT domain.The two MDrefined poses of brazilein in the inhibitor site exhibited a significant decrease in binding affinity relative to brazilin (Table 1).For pose I, flipping of brazilein occurred during the simulation and no hydrogen bonding was found (Table 1).For pose II of brazilein, a hydrogen bond formed between its OH group and Arg578 during molecular dynamic simulation (Figure 1I), but the additional hydrogen bonds formed with Ala228 and Thr439 by brazilin were lacking (Figure 1F).As for brazilin, the predicted binding energy of brazilein in the cofactor site was rather lower than that its interaction with the inhibitor site (Table 1), and 2 kcal/mol lower than for brazilin in the cofactor site (Table 1, and Figure 1G).Therefore, our computational modeling suggests that the twisted L-shape of brazilin (Figure 1A) is capable of interacting well with the NACHT inhibitor site of NLRP3, in terms of shape complementarity and forming a hydrogen bonding network; however, the more planar curved shape of brazilein (Figure 1B) is predicted to be less compatible with either the cofactor or inhibitor sites.Therefore, we hypothesized that brazilin was a strong candidate as an inhibitor of the NLRP3 inflammasome.

Brazilin inhibits activation of the NLRP3 but not AIM2 or NLRC4 inflammasomes in murine BMDMs
We next evaluated the ability of brazilin to inhibit the activity of NLRP3 experimentally using primary murine macrophages.LPS-primed (1 mg mL À1 , 4 h) BMDMs were treated with brazilin (0.01-30 mM, 15 min) prior to the addition of nigericin (10 mM, 2 h) to activate NLRP3.Brazilin caused a dose-dependent inhibition of nigericin-induced IL-1b release, with an IC 50 of 1.98 mM (Figure 2A (i) ) and cell death also decreased in a dose-dependent manner (Figure 2A (ii) ).Brazilin itself was determined as non-toxic to the cells, as treatment of either naive cells or LPS-primed cells with brazilin in the absence of nigericin stimulation did not induce cell death at any concentration (Figure 2A (ii) ).
To deduce whether the inhibitory activity of brazilin was specific to the NLRP3 inflammasome, we also tested the effects of brazilin on NLRC4 and AIM2 inflammasome activation.Treatment of LPS-primed (1 mg mL À1 , 4 h) BMDMs with brazilin (0.01-30 mM, 15 min) did not alter IL-1b release or cell death induced by subsequent transfection with poly(dA:dT) (AIM2 activator) (Figure 2B) or flagellin (NLRC4 activator) (Figure 2C).Collectively, these data indicate that brazilin can inhibit the activation of the NLRP3 inflammasome, but not AIM2 or NLRC4 inflammasome activation.

Brazilin inhibits the activation step of the canonical NLRP3 inflammasome
As discussed previously, the NLRP3 inflammasome can be activated by a range of DAMPs and PAMPs.Despite having different upstream signaling pathways, most known stimuli ultimately cause K + efflux as a necessary signal for NLRP3 activation.This includes nigericin (K + / H + ionophore), ATP (P2X purinoceptor 7 agonist), silica (particulate material) and LLOMe (lysosome disrupting agent) 9.However, the NLRP3 inflammasome can also be activated independently of K + efflux using the imidazoquinoline compound imquimod. 37We sought to investigate whether brazilin could directly inhibit NLRP3 inflammasome activation induced by these different stimuli, to address whether the inhibitory effect was dependent upon an interference with K + efflux and/or was confined to disrupting a particular activation pathway.
We first confirmed a significant inhibition of nigericin-induced NLRP3 inflammasome activation by brazilin (10 mM) (Figure 3A (i-ii) ), before testing brazilin at this concentration against other NLRP3 activating stimuli.Brazilin was also found to significantly inhibit other K + efflux-dependent pathways of activation, namely activation by ATP (Figure 3A (i-ii) ), silica (Figure 3B (i-ii) ) and LLOMe (Figure 3C (i-ii) ).K + free buffer was also used to directly evoke cellular K + efflux and NLRP3 inflammasome activation, in the absence of additional upstream signals, and under these conditions brazilin also significantly inhibited inflammasome activation (Figure 3D (i-ii) ).Brazilin also inhibited K + efflux-independent NLRP3 inflammasome activation in response to imiquimod (Figure 3E (i-ii) ).Brazilin treatment in between LPS priming and nigericin or ATP stimulation prevented the cleavage of IL-1b and caspase-1 into their mature forms, and the cleavage of gasdermin D to produce its active N-terminal fragment (Figure 3F).Brazilin treatment did not affect pro-IL-1b expression but caused a slight reduction in NLRP3 levels when added after LPS priming.
We next tested whether the inhibition of NLRP3 caused by brazilin was reversible.When brazilin was removed after 15 min treatment of LPS-primed (1 mg mL À1 , 4 h) BMDMs, and not replaced before the addition of nigericin, IL-1b release and pyroptosis were no longer inhibited.However, when the cells remained exposed to brazilin for the duration of stimulation with nigericin (1 h), either when brazilin was not removed or brazilin was removed after 15 min treatment and then replaced before the addition of nigericin, inflammasome inhibition was maintained (Figure 3G (i-ii) ).These data indicated that the inhibitory effect of brazilin on NLRP3 inflammasome activation was reversible.
Collectively, these results confirm an ability of brazilin to directly inhibit the activation step of the canonical NLRP3 inflammasome pathway, whether activation is dependent upon or independent of K + efflux, and that the mechanism of inhibition is reversible.

Brazilin can also inhibit the priming step of canonical NLRP3 inflammasome activation
As brazilin was effective at inhibiting the activation step of the NLRP3 inflammasome, we then investigated the role of brazilin on the priming step.To assess the potential effect of brazilin on the priming stage of the canonical NLRP3 inflammasome pathway, primary murine BMDMs were treated for 15 min with either DMSO (vehicle) or brazilin (10 mM) prior to the addition of LPS (1 mg mL À1 , 4 h).Brazilin pre-treatment significantly reduced expression of pro-IL-1b, NLRP3 (Figure 4A) and IL-6 (Figure 4B i ) in response to LPS priming.These data suggest that brazilin was also inhibiting priming of the canonical NLRP3 inflammasome.However, brazilin-mediated inhibition of priming was found to occur independently of its influence on the activation step of the pathway, as pro-IL-1b levels were only reduced when brazilin was added prior to but not after LPS priming (Figure 3F).In contrast, brazilin pre-treatment significantly enhanced LPS-induced TNFa release from the cells (Figure 4B ii ).Cell death was not induced by either LPS stimulation or drug treatment (Figure 4B iii ).
These data suggested that brazilin was affecting NF-kB response genes in a manner comparable to the effects of enhanced Nuclear factor erythroid 2-related factor 2 (NRF2) signaling.NRF2 is a transcription factor which is inactive when bound to Kelch-like ECH-associated protein 1 (KEAP1) in the cytoplasm.Some electrophilic agents and reactive oxygen species can induce KEAP1 degradation, allowing NRF2 to translocate into the nucleus where it can induce the expression of several target genes, but also suppress the transcription of NF-kB secondary response genes, including IL6 and IL1B. 38,39NRF2 accumulation has a much lesser effect on NF-kB primary response genes, such as TNFA. 39,40Since brazilin suppressed IL-6 and IL-1b but not the production of TNFa protein, we investigated whether brazilin could induce NRF2 signaling.BMDMs were pre-treated with DMSO, brazilin (10 mM) or dimethyl fumarate (DMF) (30 mM), a known inducer of NRF2 signaling, 41 for 15 min before LPS priming (1 mg mL À1 , 6 h).As expected, DMF alone induced NRF2 accumulation (Figure 4C i ) and this effect was enhanced by the addition of LPS (Figure 4C i-ii ).DMF pre-treatment also significantly reduced pro-IL-1b (Figure 4C i,iii ) and IL-6 levels (Figures 4D and S1A  These results show that brazilin can inhibit LPS-induced priming of the NLRP3 inflammasome, which may be a consequence of brazilin enhancing NRF2 transcription factor signaling.

Brazilin significantly reduces NLRP3 inflammasome activation in a human iPSC-microglial cell line
Dysregulated NLRP3 inflammasome activation has been implicated in several central nervous system (CNS) disorders, including stroke, Alzheimer's disease and Parkinson's disease. 19To inform upon the potential clinical translation of brazilin for such conditions, we next sought to examine the effect of brazilin on inflammasome activity in human microglia; the predominant CNS-resident immune cells.We generated iPSC-derived microglial-like cells according to previously established methods 42,43 (see also STAR methods: 'cultured cells').Firstly, we verified that our iPSC-microglia expressed the key markers IBA1, TREM2 and P2Y12 using immunocytochemistry and fluorescence microscopy (Figure S2A i-iii ).We also confirmed the microglia were functional and could respond to inflammatory stimuli by demonstrating their ability to phagocytose pHrodo-labeled E.coli bioparticles (Figure S2B) and their release of various cytokines (TNFa, IL-6, IL-10) and chemokines (CXCL10) following stimulation with LPS (Figure S2C).We also confirmed that our iPSC-microglia can mount an inflammasome response to LPS and nigericin stimulation by transfecting the cells with lentiviral human ASC-GFP and visualizing ASC speck formation (Figure S2D) (see also STAR methods: 'human iPSC-microglia characterization').
To explore whether brazilin could influence the NLRP3 inflammasome response, human iPSC-microglia were primed with LPS (100 ng mL À1 , 16 h) followed by 30-min treatment with brazilin (10-100 mM), MCC950 (0.1 mM) or DMSO control, before NLRP3 inflammasome activation with nigericin (10 mM, 2 h).Brazilin (R10 mM) treatment significantly reduced both the amount of IL-1b released (Figure 5A; Figure S3 raw data) and the number of ASC specks formed (Figures 5B and 5C) in iPSC-microglia, indicating a significant inhibition of NLRP3 inflammasome activation.Although brazilin appeared to inhibit IL-1b release in a dose-dependent manner, this was not the case with ASC speck formation, as brazilin concentrations >30 mM did not cause any further reductions in speck numbers (Figure 5B).

Brazilin significantly inhibits the NLRP3 inflammasome response to acute inflammation in vivo
To investigate whether brazilin treatment could inhibit the NLRP3 inflammasome response in vivo, mice were treated with a single i.p. dose of LPS, immediately followed by a single i.p. dose of brazilin (50 mg/kg), MCC950 (20 mg/kg) or vehicle (1% DMSO in PBS).After 2 h, mice were stably anesthetized and either PBS control or ATP was administered i.p. for 15 min under terminal anesthesia, the latter used to activate the NLRP3 inflammasome.Peritoneal lavage was then performed and the fluid used for cytokine analyses (Figure 6A).As expected, vehicle pretreated mice that received both LPS and ATP had elevated IL-1b levels in the peritoneum, compared to vehicle pre-treated mice that received LPS only (Figure 6B), indicating an activated NLRP3 inflammasome response.However, mice pre-treated with either brazilin or MCC950 had significantly lower levels of peritoneal IL-1b in response to LPS and ATP administration, compared to the vehicle-treated group (Figure 6B).This indicated an inhibition of the inflammasome response by both MCC950 and brazilin in vivo.Neither brazilin nor MCC950 significantly altered peritoneal levels of TNFa (Figure 6C) or IL-6 (Figure S4) evoked by LPS and ATP.

DISCUSSION
The NLRP3 inflammasome is a cytosolic protein complex that drives pro-inflammatory cytokine production and is an executioner of lytic cell death.Inflammasome responses can serve as a rapid cellular defense mechanism to limit tissue damage and/or spread of infections, however, when dysregulated can become a facilitator or even instigator of disease progression.Molecular inhibition of the NLRP3 inflammasome has translational relevance for numerous conditions involving acute or chronic inflammation.However, no NLRP3 inflammasome inhibitors have yet been approved for therapeutic use in the clinic, with toxicity concerns prohibiting the progression of some compounds through clinical trials.][46] Here we have identified that the natural product brazilin can significantly inhibit the NLRP3 inflammasome response in cultured mouse macrophages and human iPSC-derived microglia.Our experimental data suggest that brazilin exerts a dual inhibitory effect upon the canonical NLRP3 inflammasome, whereby it can inhibit both the priming and activation steps of the pathway.By pre-treating murine BMDMs with brazilin, the expression of pro-IL-1b, IL-6 and NLRP3 was significantly reduced in response to priming with bacterial LPS.These findings are in line with previous studies that report brazilin to suppress the pro-inflammatory response of cultured macrophages to LPS. [31][32][33] However, our data also indicate that brazilin inhibits LPS-induced expression of the NLRP3 protein.
Brazilin has been proposed to limit the cellular production of pro-inflammatory mediators through a general suppression of NF-kB signaling, by reducing NF-kB nuclear translocation 34 and/or the binding of NF-kB to DNA. 31 However, we found that brazilin treatment did not reduce the secretion of TNFa from macrophages, indicating that there was not global inhibition of NF-kB signaling in brazilin-treated cells.Elevated NRF2 signaling has previously been reported to underpin cardioprotective effects of brazilin against ischemia-reperfusion injury. 47NRF2 is a transcription factor which is known to inhibit the expression of NF-kB secondary response genes, including IL1B and IL6, 38 without inhibiting the expression of NF-kB primary response genes such as TNFA. 41Additionally, brazilin has been shown to induce the expression of heme oxygenase-1 (HO-1) in cultured cells 32,48 which may also be due to brazilin-mediated NRF2 activation, given that H O -1 is a target gene of NRF2. 39,49Therefore, we investigated the accumulation of NRF2 in BMDMs following LPS stimulation and indeed found that this was enhanced by brazilin treatment.Therefore, we propose that in the context of inflammation, brazilin may inhibit the expression of specific LPS-induced gene subsets by increasing NRF2 signaling, which consequently contributes to an inhibition of NLRP3 inflammasome priming.However, the ability of brazilin to reduce NLRP3 expression after LPS priming may be independent of NRF2 signaling, as the NRF2 inducer DMF does not affect LPS-induced NLRP3 production. 41razilin can also inhibit NLRP3 activation, via a mechanism that is reversible and independent of its influence on inflammasome priming.Treatment of LPS-primed BMDMs with brazilin was sufficient to inhibit processes downstream of NLRP3 activation, namely the maturation of caspase-1, gasdermin-D and IL-1b proteins, without altering the upstream production of pro-IL-1b.Brazilin also significantly reduced the formation of ASC specks and the release of IL-1b from human iPSC-microglia in response to nigericin.3][24] Our computational modeling predicts that brazilin may also directly interact with this inhibitor binding site.Furthermore, our biological assays showed that brazilin was able to inhibit NLRP3 activation evoked by diverse signaling pathways, dependent or independent of potassium ion efflux, and brazilin had no inhibitory effect upon the activation of the AIM2 or NLRC4 inflammasome.Collectively, these findings support our hypothesis that brazilin may interact directly and selectively with the NLRP3 sensor protein to inhibit inflammasome activation.Through direct binding, brazilin may mediate degradation of the NLRP3 protein, which may explain the inhibition of LPS-induced NLRP3 expression and/or may contribute to the slight reduction in NLRP3 levels and inhibition of the inflammasome activation step that we report.However, further experiments are needed to address these hypotheses.
Brazilin has a lower potency to inhibit the NLRP3 inflammasome in BMDMs (IC 50 1.98 mM) compared to MCC950 (IC 50 7.5 nM), 20 which may be partially due to an oxidation of brazilin to form the red-pigmented brazilein product, which our computational modeling predicts to have a much lower binding affinity for NLRP3 than brazilin.The brazilin molecule could serve as a scaffold for the development of structural analogues with improved stability and potency to inhibit the NLRP3 inflammasome.Nevertheless, we have also shown that brazilin treatment can significantly attenuate peritoneal IL-1b release in response to LPS and ATP administration in mice, confirming that brazilin in its current form is sufficiently stable and active to inhibit an acute NLRP3 inflammasome response in vivo.Brazilin is the most abundant constituent of heartwood from the Haematoxylum brasiletto and Caesalpinia sappan tree species.While there are no commercially licensed products containing purified brazilin, heartwood extracts from these trees have long been used as coloring agents in foods and beverages and as medicinal remedies. 27,28This would suggest that brazilin is likely safe for human consumption; however, thorough preclinical and clinical toxicology investigations are still required.][33] Brazilin also had no obvious adverse effects on the appearance, behavior or survival of the mice in our in vivo study, although their exposure to the treatment was short (2 h) and no formal postmortem analyses of organs were performed.We found that brazilin may enhance TNFa release from LPS-primed BMDMs, which may be a concern as TNFa is itself known to drive a range of inflammatory diseases. 50However, brazilin treatment did not enhance peritoneal TNFa levels beyond that associated with the inflammation itself in our mouse model.It is also worth noting that brazilin had no inhibitory effects on the AIM2 inflammasome response to DNA or the NLRC4 inflammasome response to bacterial flagellin in cultured BMDMs, which may be an important attribute when considering such a compound as a potential immunomodulatory therapeutic.However, further experiments are required to confirm this in vivo.
In conclusion, we have identified that the natural product brazilin can exert a dual inhibitory effect over both the priming and activation steps of the canonical NLRP3 inflammasome pathway.Our results indicate that brazilin can enhance NRF2 signaling, which may contribute to the inhibition of inflammasome priming, while our computational modeling suggests that brazilin may also bind directly to the NLRP3 protein to inhibit inflammasome activation.We have confirmed that brazilin can significantly inhibit the NLRP3 inflammasome response in cultured murine macrophages, human microglial cells, and in a mouse model of acute inflammation.Our results encourage further evaluation of brazilin as a promising therapeutic agent for NLRP3-related inflammatory diseases.Moreover, this small-molecule may serve as a scaffold for the development of new selective NLRP3 inflammasome inhibitors.

Limitations of the study
Taken together, our data strongly suggest that Brazilin binds directly to the NLRP3 protein to inhibit downstream inflammasome assembly and IL-1b maturation.However, this hypothesis is mainly derived from computational modeling data.Future work will confirm this direct interaction of Brazilin with the NLRP3 protein using biological binding assays, such as those recently described by Teske et al. 51 and humidity, ad libitum food and water and environmental enrichment).All experiments were performed in accordance with the UK Home Office regulations (PPL: P4035628) and reported according to the ARRIVE guidelines for experiments involving animals. 53Animal work was approved by The University of Manchester Animal Welfare and Ethical Review Board.

METHOD DETAILS Molecular modeling and simulation
A complete NACHT domain structure of NLRP3 was prepared for computational docking from its X-ray structure (PDB code 7ALV) 22 using SWISS-MODEL server (https://swissmodel.expasy.org).Docking was performed using the OpenEye software suite. 54The OMEGA classic tool 55 was used to create 3D structures of brazilin and brazilein using a maximum of 500 conformations for each compound.The NLRP3 cofactor site and inhibitor site were prepared for docking using the make_receptor module.The FRED module [56][57][58] was used to dock compounds using the Chemgauss4 scoring function.The best 50 poses for each compound were visualized using Vida 4.4.0 and MOE 2022.02. 59his docking protocol was previously validated and reported 46 by redocking of ADP in the cofactor binding site and NP3-146 (an analogue of MCC950) in the inhibitor site which reproduced the crystal structure pose well in both cases.
Nevertheless, molecular dynamics (MD) simulations [60][61][62] and experimentally determined structures 23,24,36,63,64 of NLRP3 point to the flexibility of the protein.Consequently, we refined our docked poses using MD simulation, following an approach recently applied by us in the analysis of tubulin inhibitors. 65Therefore, MD simulations of NLRP3-ligand complexes were performed for selected docked poses.These calculations employed the AMBER 19 package. 52Atomic partial charges of ligands were assigned via the AM1-BCC method using the antechamber module of AMBER.The gaff2 66 and ff14SB 67 force fields were used to describe ligands and receptor, respectively.Force field parameters for ADP were assigned from the AMBER database (http://amber.manchester.ac.uk/) 68 and gaff2. 66The systems were solvated in an octahedral TIP3P water box 69 that extends at least 10 A ˚from the protein-ligand surface.The system was neutralized by addition of 2-5 chloride counterions to the solvated system.This led to $15,664 water molecules for each simulation system.The generated topology files were edited with the parmed 70 module of AMBER 19 to repartition the mass of heavy atoms into the bonded hydrogen atoms.The new topology file was designed to use hydrogen mass repartitioning 71 in which the time step of the simulation was 4 fs.The nonbonded cut-off of 9.0 A ˚was used, along with the particle mesh Ewald 72 method for long range electrostatic interactions.
MD simulations were performed using the pmemd.cudamodule of AMBER 19.Before simulation, water molecules were relaxed by energy minimizing the solvated system.The system was heated to 300 K in two steps under NVT conditions (constant Number, Volume and Time) over 700 ps using the Langevin thermostat with a coupling constant value of g = 2.0. 73Then the system was equilibrated at 300 K and 1 atm with a Monte Carlo barostat 74 for 10 ns with a coupling constant value of 25 and compressibility of 44.6.The production MD simulation was conducted for 40 ns in an NPT ensemble (constant Number, Pressure and Time), during which configurations were sampled every 10 ps.
For ligand-protein systems, the MM/GBSA method 75 was applied to the final 10 ns of the trajectory to estimate binding free energies.These calculations were performed using the MMPBSA.py 76tool of AMBER 19.The internal and external dielectric constants were set to 1.0 and 80.0, respectively.The ionic strength was set to 0.15 mM.MM/GBSA calculations were performed using 100 snapshots/compound.The electrostatic contribution to binding free energy DG el was a sum of electrostatic protein-ligand and solvation components; the non-electrostatic contribution DG nonel was a sum of protein-ligand van der Waals and non-electrostatic solvation terms.Ligand efficiency (LE) was calculated by dividing the DG tot by the molecular weight.
For the fluorescent ASC speck assay, cells were fixed in 4% PFA as described above, after which a Wheat Germ Agglutinin (WGA) cell mask (Invitrogen, W32466; 1:200 in PBS) and Hoescht nuclear stain (Invitrogen, 62249; 1:4000 in PBS) were added.Cells were then washed twice prior to confocal imaging.
For phagocytosis experiments, iPSC-microglia (11-14 DIV) were pre-treated with 3uM Cytochalasin D (Sigma, C8273) or DMSO control for 30 min then incubated with 0.1 mg/mL pHrodo-labelled E.coli Bioparticles (Sartorius, 4616; Go ¨ttingen, Germany).Cells were imaged using a Incucyte cell imaging system every hour for 24 h (10X, Phase and green fluorescence).Using the IncuCyte ZOOM software, image masks for the fluorescent signal were generated and quantified.
For cytokine release experiments, iPSC-microglial cells (11-14DIV) were stimulated with 100 ng/mL of LPS.Supernatants were harvested after 24 h and analyzed with a custom Human Luminex Discovery Assay kit (R&D Systems, LXSAHM; Abingdon, UK) to examine secretion of TNF-⍺, IL-6, IL-10, and CXCL10 following stimulation.The Luminex assay was run according to the kit protocol.In brief, supernatant was diluted to within the linear range of the assay if required.Analyte-specific antibody-coated magnetic microparticles were then incubated subsequently with the sample, a biotin-antibody cocktail and streptavidin-PE, with 3 washes on a magnetic plate washer in between.The microparticles were then resuspended and read on a Luminex MAGPIX instrument, and analyte concentration was calculated from a standard curve.
At the end of these experiments, supernatants were collected and processed using cell death and/or ELISA analyses, while lysis buffer (50 mM Tris-HCl, 150 mM NaCl, Triton 1%; pH 7.5) was added to the cells.Alternatively, cells were lysed in-well by adding protease inhibitor cocktail (Merck Millipore, 539131) (1% v/v) and Triton-X-100 (1% v/v) (Sigma, X-100) directly into the culture medium, to assess total protein content in combined cell lysate and supernatant via Western blot (as indicated in the results).All samples were stored at À20 C between use.

Human iPSC-microglia
To examine inflammasome activation, iPSC-microglial cells were transduced with a lentivirus expressing hASC (NM-013258.5)fused to GFP at 8-day-in vitro (DIV).After 72 h, cells were treated with or without LPS (100 ng mL À1 ) for 16 h.Microglial cells were then treated with or without MCC950 (0.1 mM), brazilin (10, 30, 60 or 100 mM) or DMSO control as indicated for 30 min prior to stimulation with nigericin (10 mM) for 2 h.Images were captured using SX5 incucyte imaging system (20X phase and green fluorescence) and speck analysis was performed by blinded manual counting of specks in two images for each biological replicate in each treatment condition.Media supernatant was collected from the cells at the end of the experiment and IL-1b protein was analyzed using an IL-1b Human Luminex Discovery Assay kit (R&D Systems), as detailed above.

Figure 1 .
Figure 1.Predicted NLRP3 binding mode of brazilin from docking and simulation (A and B) 2D and 3D structures of (A) (S)-brazilin in which the gray highlighted OH group represents the hydroxyl group that oxidizes to give (B) (S)-brazilein.(C) Superposition of full-length NLRP3 cryo-EM structure (gray, PDB code 7PZC) 23 with X-ray structure of NACHT domain (blue, PDB code 7ALV) 22 ; inset shows in more detail the two binding sites in NACHT domain.Walker A (WA) motif is colored pink.(D) Superposition of X-ray pose of NP3-146 (gray), and top-ranked docked pose I (brown) and pose II (orange) of brazilin in inhibitor binding pocket (gray surface) of NLRP3-NACHT structure.Detailed interatomic interactions of (E) pose I (brown) and (F) pose II of brazilin (orange) with inhibitor binding site using NACHT-NLRP3 structure after 40 ns MD simulation.(G) Superposition of X-ray pose of ADP (gray), post-MD structure of brazilin (cyan) and brazilein (dark cyan) in cofactor binding pocket (gray surface) of NLPR3-NACHT structure.(H) Detailed interatomic interactions of brazilin (cyan) with cofactor binding site using NACHT-NLRP3 structure after 40 ns MD simulation.(I) Detailed interatomic interactions of pose II of brazilein (green) in the inhibitor binding site using the NACHT-NLRP3 structure after 40 ns MD simulation.Dashed lines represent hydrogen bonds and lost interactions indicated by dashed line with red cross.See also Video S1 showing the MD simulation of interactions between S-brazilin and the inhibitor binding site of NACHT domain.
raw data) in response to LPS priming compared to DMSO treated controls, but did not affect LPS-induced NLRP3 expression (Figure4Civ ) or TNFa release (Figures 4E, and S1B raw data).Treatment with brazilin alone did not induce significant NRF2 accumulation (Figure 4C i ).However, as observed with DMF treatment, brazilin strongly enhanced LPS-induced NRF2 accumulation (Figure 4C i-ii ) whilst significantly inhibiting LPSinduced pro-IL-1b (Figure 4C i,iii ) and IL-6 production (Figures 4D and S1A raw data), without affecting TNFa release (Figures 4E and S1B raw data).Brazilin also significantly reduced NLRP3 expression in LPS-primed cells (Figure 4C iv ), as we observed previously (Figure 4A i ).

Figure 4 .
Figure 4. Pre-treatment with brazilin reduces LPS-induced priming of the canonical NLRP3 inflammasome in association with enhanced NRF2 expression (A and B) BMDMs were treated with vehicle (DMSO) or brazilin (10 mM) for 15 min.LPS (1 mg mL À1 , 4 h) was then added to the wells to induce priming.(A) Concentrated protein content from combined supernatant and cell lysates were probed by Western blotting for pro-IL-1b and NLRP3 proteins.The blot shown is representative of 3 biological repeats.(B) In a separate experiment, supernatants alone were analyzed by ELISA for (B i ) IL-6 (N = 3 or 5), (B ii ) TNFa (N = 5) and (B iii ) LDH release (N = 5).(C-E) BMDMs were treated with vehicle (DMSO), brazilin (10 mM) or DMF (30 mM) for 15 min.LPS (1 mg mL À1 , 6 h) was then added to the wells to induce priming.(C) Cell lysates were probed by Western blotting for NRF2, pro-IL-1b and NLRP3 proteins.(C i ) The blot shown is representative of 4 biological repeats.For LPS+ samples only, expressions of (C ii ) NRF2 (N = 3), C( iii ) pro-IL-1b (N = 4) and C( iv ) NLRP3 (N = 4) proteins were quantified by densitometry (expressed relative to DMSO treatment/normalized to b-actin).(D and E) Supernatants from the same cells were analyzed by ELISA for (D) IL-6 (N = 4) and (E) TNFa (N = 4) release.Data show cytokine levels relative to DMSO + LPS treatment.See also Figure S1 for cytokine release data without normalization.All data are presented here as mean G SEM, each data point ('N') representing a biological repeat.Each dataset (B, Cii-iv, D and E) was generated from two experimental repeats.Statistical analyses following normality testing: B( i-iii ) unpaired t-Tests; C(ii-iv) and (D-E) one-sample t-tests, comparing means to a hypothetical mean of 1.0.*p < 0.05, **p < 0.01, ***p < 0.001.BMDMs, bone marrow-derived macrophages; DMSO, dimethyl sulfoxide; DMF, dimethyl fumarate; NRF2; nuclear factor erythroid 2-related factor 2.

Figure 5 .
Figure 5. Brazilin significantly inhibits NLRP3 inflammasome activation by nigericin in human iPSC-microglia Human iPSC-derived microglial cells expressing lentiviral hASC-GFP were primed with LPS (100 ng mL À1 ) or vehicle (16 h) before treatment with MCC950 (0.1 mM), brazilin (10, 30, 60 or 100 mM) or DMSO control (30 min).Nigericin was then added to activate the NLRP3 inflammasome (10 mM, 2 h).(A) Supernatant IL-1b was quantified using an IL-1b Human Luminex Discovery Assay kit, expressed relative to DMSO control cells treated with LPS and ATP.See also Figure S3 for IL-1b release data without normalization.N = 3 biological repeats/group.(B) ASC specks were quantified via blinded manual speck counts from two images (field of view; F.O.V.) obtained from four biological replicates of each treatment condition (N = 8).(C) Representative images used for ASC speck counting.ASC specks are indicated by arrow heads and a zoomed in region is placed in the top right corner of each image, indicated by a white box.Images were captured using SX5 incucyte imaging system (20X phase, green GFP fluorescence).Scale bars represent 200 mm.Statistical analyses following normality testing: (A) one-sample t-tests, comparing means to a hypothetical mean of 1.0.(B) Kruskal-Wallis test with Dunn's multiple comparisons.All data are expressed as mean G SEM.Each dataset (A and B) was generated from two experimental repeats.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 6 .
Figure 6.Pre-treatment with brazilin can significantly attenuate the NLRP3 inflammasome response in vivo (A) Schematic of the treatment protocol and sample collection from male C57 mice.Intraperitoneal (i.p.) administration of either brazilin (50 mg/kg, 2h) or MCC950 (20 mg/kg, 2h), alongside LPS (i.p., 1 mg, 2h), (B) significantly attenuated subsequent IL-1b release in the peritoneum in response to ATP administration (i.p., 100 mM in PBS (500 mL/mouse), 15 min), compared to vehicle (1% DMSO in PBS) treated mice.Neither brazilin nor MCC950 treatments had any effect on peritoneal (C) TNFa release in response to LPS and ATP administration.See also Figure S4 for peritoneal IL-6 data.Data are presented as mean G SEM. N = 5 mice per treatment group.Statistical analyses following normality testing: (B) one-way ANOVA with Tukey's post-hoc comparisons, (C) Kruskal-Wallis test with Dunn's post hoc comparisons were used to assess the effect of drug treatment between groups treated with both LPS and ATP.*p < 0.05, **p < 0.01 and ***p < 0.001.DMSO, dimethyl sulfoxide; LPS, lipopolysaccharide; ATP, adenosine triphosphate; i.p., intraperitoneal.

Table 1 .
Computed binding free energies of brazilin and brazilein in the inhibitor and cofactor binding sites of the NLRP3 protein