The glycolytic metabolite methylglyoxal covalently inactivates the NLRP3 inflammasome

SUMMARY

The NLRP3 inflammasome promotes inflammation in disease, yet the full repertoire of mechanisms regulating its activity is not well delineated. Among established regulatory mechanisms, covalent modification of NLRP3 has emerged as a common route for the pharmacological inactivation of this protein. Here, we show that inhibition of the glycolytic enzyme phosphoglycerate kinase 1 (PGK1) results in the accumulation of methylglyoxal, a reactive metabolite whose increased levels decrease NLRP3 assembly and inflammatory signaling in cells. We find that methylglyoxal inactivates NLRP3 via a non-enzymatic, covalent-crosslinking-based mechanism, promoting inter- and intraprotein MICA (methyl imidazole crosslink between cysteine and arginine) posttranslational linkages within NLRP3. This work establishes NLRP3 as capable of sensing a host of electrophilic chemicals, both exogenous small molecules and endogenous reactive metabolites, and suggests a mechanism by which glycolytic flux can moderate the activation status of a central inflammatory signaling pathway.

In brief

Stanton et al. demonstrate that pharmacological inhibition of the glycolytic enzyme PGK1 hinders assembly of the NLRP3 inflammasome. Methylglyoxal, a reactive metabolite that accumulates in response to inhibition of upstream glycolysis, covalently modifies NLRP3 by crosslinking MICA modifications, preventing functional inflammasome assembly.

Graphical Abstract

INTRODUCTION

Vital to regulating the levels of pro-inflammatory cytokines and pyroptotic cell death, inflammasomes are large cytosolic protein complexes controlled by pattern recognition receptors. One such inflammasome is nucleated by NLRP3 (NOD-, LRR-, and pyrin domain-containing protein 3), which responds to various danger- and pathogen-associated molecular patterns. Activation of the NLRP3 inflammasome requires both priming and activation of the transcription factor nuclear factor κB (NF-κB) to augment transcription of inflammasome components, as well as a secondary activating signal, such as K⁺ efflux, lysosomal signaling, or mitochondrial reactive oxygen species. The presence of both signals promotes the assembly of the active NLRP3 inflammasome, a complex formed by association of NLRP3 with NEK7, ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain) and pro-caspase-1. Upon assembly, pro-caspase-1 activation cleaves pro-interleukin (IL)-1β and pro-IL-18 to active forms. Additionally, caspase-1 cleaves gasdermin D, which forms pores in the plasma membrane resulting in pyroptotic cell death. Given the broad array of signals to which NLRP3 responds, the NLRP3 inflammasome is associated with sterile inflammation in disease.^1–7 For this reason, NLRP3 remains a promising drug target with the potential to mitigate pathology in numerous inflammatory conditions. However, at present, there are no clinically approved NLRP3 inhibitors, though several are being evaluated in the clinic.^8,9

Dapansutrile and ZYIL1, inhibitors targeting the NACHT domain of NLRP3, which reduces ATPase activity, have shown promising activity in phase 2 studies.^9–11 However, among promising approaches to inhibit NLRP3 pharmacologically, covalent inhibitors of NLRP3 have become commonplace in the clinic and in literature. RRx-001, a covalent modifier of C409 that inhibits protein interactions with NEK7, is in phase 3 studies for cancer, but since its identification as an inflammasome inhibitor, it is now also being studied for various neurodegenerative diseases.¹² Likewise, other inhibitors targeting reactive sensor cysteines in NLRP3 have been reported, including oridonin, costunolide, Bay-11–7082, 3,4-methylenedioxy-β-nitrostyrene, parthenolide, and Tubocapsanolide A.^13–17 Recently, we identified new covalent scaffolds showing robust covalent modification of NLRP3 across several domains.¹⁸ NLRP3 is also inhibited through covalent modification by several reactive metabolites, including diroximel fumarate,¹⁸ a fumarate derivative, and itaconate, which modifies C548 of murine NLRP3 to inhibit NLRP3 interactions with NEK7.¹⁹ These observations indicate that NLRP3 serves as an electrophile sensor within the cell, linking the covalent modification of its sensor cysteines to decreased NLRP3 inflammasome activation and downstream inflammatory signaling.

The capacity for NLRP3 to respond to electrophilic small molecules is reminiscent of the established electrophile sensor KEAP1. KEAP1 serves as a negative regulator of the oxidative stress-responsive transcription factor NRF2, promoting its ubiquitination and degradation through interactions with the Cullin-RING E3 ubiquitin ligase protein complex. KEAP1 bears 12 sensor cysteines, which respond to covalent modification by liberating NRF2 to enact a protective transcriptional response.²⁰ In addition to exogenous electrophilic chemicals, KEAP1 covalently senses endogenous reactive electrophilic metabolites induced by perturbations in the citric acid cycle (e.g., fumarate, itaconate) and glycolysis, resulting in NRF2 activation.^21–27

We identified the compound CBR-470-1 as a NRF2 activator.²⁸ This compound inhibits the glycolytic enzyme phosphoglycerate kinase 1 (PGK1), leading to an accumulation of upstream metabolites 1,3-bisphosphoglycerate (1,3-BPG), glyceraldehyde-3-phosphate (GAP), and dihydroxyacetone phosphate (DHAP). In response to PGK1 inhibition, methylglyoxal (MGO), the non-enzymatic dicarbonyl elimination product of GAP and DHAP, accumulates and induces the formation of KEAP1 dimers via MICA (methyl imidazole crosslink between cysteine and arginine) modifications, increasing NRF2-dependent expression of target genes. An analog of this compound, CBR-470-2, was shown to be active in vivo, protecting against oxidative damage induced by pathogenic UV exposure.²⁸

Because NLRP3 and KEAP1 both sense electrophilic chemicals from exogenous and endogenous sources, we hypothesized that NLRP3 also senses and responds to glycolytic metabolites, like MGO. Here, we report that the CBR-470 compound series inhibits NLRP3 inflammasome assembly through MGO-dependent covalent crosslinking and inactivation of NLRP3.

RESULTS

Pharmacologic PGK1 inhibitors block NLRP3 inflammasome assembly and activity

The assembly and activity of the NLRP3 inflammasome is sensitive to reactive metabolites generated by alterations to metabolic pathways. We sought to determine whether PGK1 inhibitors such as CBR-470-1 and CBR-470-2 (Figures S1A and S1B)—both compounds that generate the reactive metabolite MGO²⁸—inhibit NLRP3 inflammasome assembly and activity. Initially, we monitored assembly of the NLRP3 inflammasomes in THP1 cells stably expressing GFP-tagged ASC (ASC-GFP). Treatment of these cells with lipopolysaccharide (LPS), followed by the activating compound nigericin, triggers NLRP3 inflammasome assembly, which is followed by the formation of ASC-GFP specks (Figure S1C). As described previously, ASC-GFP speck formation could be dose-dependently inhibited by pretreatment with the NLRP3 inflammasome inhibitor MCC950 (Figure 1A).¹⁸ Pre-treatment with either CBR-470-1 or CBR-470-2 for 2 h inhibited ASC-GFP speck formation in these cells with IC₅₀ (half maximal inhibitory concentration) of 25 and 3 μM, respectively (Figure 1A). The inhibition afforded by these two PGK1 inhibitors was similar to that observed for MCC950, and minimal increases in inhibition were observed by increasing the pretreatment time beyond 2 h (Figures 1A–1C, S1D, and S1E).

Figure 1. CBR-470 series compounds inhibit the NLRP3 inflammasome.

(A) Relative ASC-GFP specks per cell in LPS-primed THP1-ASC-GFP cells treated for 2 h with indicated concentrations of compound and activated with 10 μM nigericin for 2 h. Error bars show SEM for n = 3 replicates.

(B and C) Relative ASC-GFP specks per cell in LPS-primed THP1-ASC-GFP cells pretreated with CBR-470-1 (B) and CBR-470-2 (C) at 12.5 μM for 0–24 h and activated with 10 μM nigericin for 2 h. Error bars show SEM for n = 3 replicates. ****p < 0.0001 for ordinary one-way analysis of variance (ANOVA).

(D) Relative level of secreted IL-1β from WT THP1 cells primed with LPS, pretreated with MCC950 (10 μM), CBR-470-1, or CBR-470-2, and activated with 5 mM ATP. IL-1β measured by SEAP (secreted embryonic alkaline phosphatase) secretion from HEK-Blue-IL-1β reporter cells. Error bars show SEM for n = 3 replicates.

(E) Relative level of secreted IL-1β from primary murine dendritic cells stimulated with LPS and pretreated with 10 μM CBR-470-1 or CBR-470-2 for 1 h prior to addition of 5 mM ATP, as measured by SEAP secretion from HEK-Blue-IL-1β reporter cells. Error bars show SD for n = 16 replicates. ****p < 0.0001 for ordinary one-way ANOVA with Tukey correction for multiple comparisons between conditions.

(F) Relative viability of LPS-primed WT THP1 following cell death induced by nigericin (10 μM, 2.5 h), pretreated with CBR-470-1 and CBR-470-2 (2 h) in dose response or with 10 μM MCC950 (2 h). Error bars show SEM for n = 6 replicates.

Next, we determined the potential for these compounds to inhibit NLRP3-inflammasome-dependent activities, including caspase-1 activation, IL-1β secretion, and pyroptotic cell death. Pretreatment with CBR-470-1 or CBR-470-2 for 2 h inhibited IL-1β secretion from both THP1 cells and primary dendritic cells, as measured using HEK293-Blue-IL-1β reporter cells (Figures 1D and 1E), and reduced caspase-1 activity (Figure S1F). CBR-470 pretreatment also reduced IL-1β secretion in response to NLRC4-activating stimuli, although inhibition in response to NLRP1-activating stimuli was unable to be determined due to confounding toxicity (Figures S1G and S1H). CBR-470-1 or CBR-470-2 for 2 h also blocked pyroptotic cell death induced by nigericin in LPS-primed THP1 cells (Figure 1F). These results demonstrate that two PGK1 inhibitors block NLRP3 inflammasome assembly and downstream signaling. Since CBR-470-2 showed superior potency compared to CBR-470-1, we focused on CBR-470-2 for further mechanistic experimentation.

CBR-470-2-dependent NLRP3 inhibition is independent of NRF2 activation and NF-κB inhibition

PGK1 inhibitors, including CBR-470-2, activate NRF2 in mammalian cells.²⁸ Since NRF2 activation inhibits the NLRP3 inflammasome,^29,30 we sought to determine whether the inhibition of NLRP3 assembly and activity observed with this compound is attributed to NRF2 activation. A 2 h treatment with CBR-470-2 did not induce expression of the NRF2 target gene NQO1 in THP1 cells (Figure 2A). However, we did observe increased NQO1 expression following 16 h treatment with CBR-470-2. This increase in NQO1 expression was blocked by co-treatment with the NRF2 inhibitor ML385. Similar results were observed for the alternative NRF2 target gene HMOX1 (Figure S1I). These results indicate that NRF2 is not activated in the 2 h pretreatment time course where we observe CBR-470-2-dependent inhibition of NLRP3 inflammasome assembly and activation.

Figure 2. CBR-470-2 inhibition of the NLRP3 inflammasome does not rely on NRF2 activation or NF-κB inhibition.

(A) Relative transcript level for NQO1 from WT THP1 cells pretreated with or without 10 μM ML385 for 30 min and then treated with CBR-470-2 for 2 or 16 h. Error bars show SEM for n = 3 replicates.

(B) Relative ASC-GFP specks per cell in LPS-primed THP1-ASC-GFP cells pretreated with or without 10 μM ML385 for 30 min and then for 2 h with CBR-470-1 and CBR-470-2, followed by activation with 10 μM nigericin for 2 h. Error bars show SEM for n = 3 replicates.

(C) Relative viability of LPS-primed WT THP1 following cell death induced by nigericin (10 μM, 2.5 h), pretreated with or without 10 μM ML385 for 30 min and then with CBR-470-2 (2 h) in dose response or with 10 μM MCC950 (2 h). Error bars show SEM for n = 3 replicates.

(D) Relative transcript levels for NF-κB target genes NLRP3 and IL-1β from WT THP1 cells treated with 1 μg/mL LPS for 3 h and 10 μM CBR-470-2 for 3 or 24 h. Error bars show SEM for n = 3 replicates.

(E) Western blot for NLRP3, ASC, NEK7, pro-IL-1β, IL-1β, pro-CASP1, CASP1, and Tubulin in LPS-primed THP1 cells treated with vehicle or CBR-470-2 (50 μM) for 2 h.

(F) Average intensity of bands of western blot in (E) for each protein. Error bars show SEM for n = 3 replicates.

To further probe the dependence of compound-dependent NLRP3 inhibition on NRF2, we monitored ASC-GFP speck formation and pyroptotic cell death in THP1 cells pretreated with the NRF2 inhibitor ML385 for 30 min followed by CBR-470-2 for 2 h. Pretreatment with ML385 did not influence the potency or efficacy of CBR-470-2-dependent inhibition of NLRP3 inflammasome assembly or activity in these assays (Figures 2B and 2C). This shows that the inhibition of NLRP3 inflammasomes afforded by CBR-470-2 is not mediated through NRF2 activation.

Next, we determined the potential impact of CBR-470-2 on NF-κB-dependent expression of NLRP3 inflammasome components induced by LPS priming. Co-treatment with CBR-470-2 for 3 or 24 h does not influence the expression of NLRP3 in LPS-primed THP1 cells, although it does modestly reduce expression of pro-IL1b (Figure 2D). However, protein levels of NF-κB targets, including NLRP3, ASC, and pro-IL-1β, as well as other NLRP3 inflammasome components, were not significantly decreased following a 2 h treatment with CBR-470-2 in LPS-primed THP1 cells (Figures 2E and 2F). These results indicate that CBR-470-2 does not influence the expression of NF-κB target genes induced by LPS priming.

CBR-470-2 inhibits NLRP3 through the increased production of MGO downstream of PGK1

Pharmacologic PGK1 inhibition with CBR-470-2 leads to the accumulation of the reactive metabolite MGO.²⁸ Since the NLRP3 inflammasome is sensitive to inactivation by reactive metabolites, we predicted that CBR-470-2 inhibits NLRP3 inflammasome assembly and activity through the accumulation of MGO. To test this, we monitored NLRP3 inflammasome assembly (via ASC-GFP speck formation) in THP1 cells co-treated with CBR-470-2 and glutathione (GSH), the latter a treatment that neutralizes reactive metabolites. Co-treatment with GSH abrogated the inhibition of ASC-GFP speck formation afforded by CBR-470-2 (Figure 3A). Identical results were observed in THP1 cells treated with CBR-470-1 (Figure S2A). This supports a model whereby these PGK1 inhibitors inhibit inflammasome assembly through a mechanism involving the accumulation of a reactive metabolite such as MGO.

Figure 3. Inhibition by CBR-470 is mediated by PGK1 inhibition and accumulation of methylglyoxal.

(A) Relative ASC-GFP specks per cell in LPS-primed THP1-ASC-GFP cells pretreated with 0 or 10 mM GSH and treated in dose response with CBR-470-2, followed by activation with 10 μM nigericin for 2 h. Error bars show SEM for n = 3 replicates.

(B) Relative transcript level of PGK1 48 h after knockdown with siRNA targeting PGK1 in THP1-ASC-GFP cells. Error bars show SEM for n = 3 replicates. ****p < 0.0001 for ordinary one-way analysis of variance (ANOVA).

(C) ASC speck formation in LPS-primed THP1-ASC-GFP cells 48 h following transfection with siRNAs targeting PGK1, activated with 10 μM nigericin for 2 h. Error bars show SEM for n = 3 replicates. ****p < 0.0001 for ordinary one-way anlysis of variance (ANOVA).

(D) Relative ASC-GFP specks per cell in LPS-primed THP1-ASC-GFP cells pretreated with or without 10 μM ML385 for 30 min and then for 2 h with MGO in dose response, followed by activation with 10 μM nigericin for 2 h. Error bars show SEM for n = 3 replicates.

(E) Fluorescence (excitation [Ex.] 355, emission [Em.] 400 nM) increase from untreated control for methylglyoxal-level assay for lysates from THP1 cells treated with 50 μM CBR-470-2 for indicated time point or 1 mM MGO for 1 h. Error bars show SEM for n = 6 replicates.

CBR-470-2 generates MGO by inhibiting PGK1 enzymatic activity.²⁸ We predicted that genetic depletion of PGK1 should inhibit NLRP3 inflammasome assembly. We depleted PGK1 in THP1-ASC-GFP cells using three different small interfering RNAs (siRNAs) and monitored NLRP3 inflammasome assembly by ASC-GFP speck formation. We confirmed that siRNA treatment reduced the expression of PGK1 in these cells by RT-qPCR (Figure 3B). Reductions in PGK1 afforded by all three siRNAs reduced ASC-GFP speck formation in THP1 cells (Figure 3C). This shows that genetic depletion of PGK1 recapitulates the reduced NLRP3 inflammasome assembly induced by pharmacologic PGK1 inhibition.

Next, we determined the potential for direct administration of MGO to inhibit NLRP3 inflammasomes. We treated THP1 cells with increasing concentrations of MGO and monitored NLRP3 assembly using our ASC-GFP speck formation assay. Pretreatment with MGO for 2 h inhibited NLRP3 inflammasome assembly with an IC₅₀ of ~0.52 mM (Figure 3D). Similar results were observed for pyroptotic cell death, where pretreatment for 2 h with MGO improved the viability of THP1 cells treated with LPS and nigericin (Figure S2B). MGO activates NRF2 through the covalent targeting of KEAP1. However, we found that pretreatment with the NRF2 inhibitor ML385 did not decrease the inhibition of ASC-GFP speck formation or the protection from pyroptotic cell death afforded by MGO treatment (Figures 3D and S2B). These results suggest that MGO-dependent NRF2 activation does not contribute to the observed reduction in NLRP3 inflammasome assembly and activity induced by this metabolite. Our results are consistent with a model whereby MGO inhibits NLRP3 inflammasome assembly through a direct mechanism involving covalent protein modification.

We next investigated whether treatment with CBR-470-2 increases MGO levels in THP1 cells. We used a fluorescence indicator probe for MGO generation³¹ to monitor the production of MGO in THP1 cells treated with CBR-470-2 (Figure S2C). We observed a rapid increase of MGO in THP1 cells treated with CBR-470-2, which declined ~8 h after treatment (Figure 3E). The level of intracellular MGO generated by compound treatment was nearly identical to that observed in THP1 cells treated with 1 mM MGO—a dose sufficient to fully inhibit NLRP3 inflammasome assembly (Figure 3D). We also observe KEAP1 crosslinking following treatment of THP1 cells with CBR-470-2 for <1 h, supporting this rapid accumulation of MGO within the cell (Figure S2D). This indicates that treatment with CBR-470-2 increases intracellular MGO to levels sufficient to inhibit NLRP3 inflammasome assembly. These results indicate that CBR-470-2 inhibits NLRP3 inflammasome assembly through a mechanism involving PGK1 inhibition and subsequent accumulation of MGO.

MGO induces inter- and intramolecular crosslinks of NLRP3 monomers through MICA modifications

MGO inhibits the NRF2 suppressor KEAP1 by inducing MICA modifications on neighboring protomers. Thus, we sought to determine whether MGO generated by CBR-470-2 treatment similarly induces intermolecular crosslinking of NLRP3. Treatment with MGO for 2 h increased crosslinking of NLRP3-FLAG expressed in HEK293T cells, as measured by immunoblotting, and the accumulation of high-molecular-weight (HMW) NLRP3 (Figure S3A). Treatment with CBR-470-2 also increased HMW NLRP3-FLAG in HEK293T cells. Further, treatment with CBR-470-2 increased endogenous NLRP3 crosslinking in THP1 cells, evident by reductions in soluble NLRP3 and increased HMW NLRP3 in the insoluble fraction (Figure 4A). However, CBR-470-2 did not induce the formation of HMW species of other inflammasome components (e.g., NEK7 or ASC), although we did observe modest increases in HMW CASP1, albeit at a higher MW than a CASP1 dimer (Figure S3B). Given that we do not observe ASC speck formation in cells pretreated with CBR-470, which is upstream of CASP1 recruitment, NLRP3 crosslinks are likely more relevant than CASP1 crosslinks.

Figure 4. CBR-470-2 inhibits NLRP3 through MICA crosslinks.

(A) Western blot of soluble and insoluble NLRP3 (NBD) (MW: 118 kDa) and Tubulin from WT THP1 cells treated with 50 μM CBR-470-2 for 1–16 h.

(B) Western blot for FLAG and Tubulin in HEK293T cells overexpressing the indicated FLAG-tagged NLRP3 domain constructs treated with 50 μM CBR-470-2 for 4 or 8 h (WT: 121 kDa, ΔPYN: 106 kDa, ΔNBD: 61 kDa, ΔLRR: 78 kDa, PYN: 18 kDa, NBD: 63 kDa, LRR: 46 kDa).

(C) Western blot of FLAG from HEK293T cells overexpressing NLRP3-FLAG PYN domain constructs with cysteines mutated and individually reintroduced, treated with 50 μM CBR-470-2 for 6 h.

(D) Schematic depicting cysteine-arginine MICA crosslinks observed within NLRP3 PYN domain.

(E–G) Extracted ion chromatograms from liquid chromatography-tandem mass spectrometry (LC-MS/MS) analyses of gel-isolated and digested HMW PYN (CBR-470-2 and MGO-induced) and monomeric PYN for C38-R07 (E), C08-R43 (F), and C130-R89 (G) crosslinked peptides.

(H) Schematic depicting the communication between glucose metabolism and the NLRP3 inflammasome mediated by MGO crosslinking of NLRP3 and inhibition of inflammasome assembly.

For all images, black arrow = monomer, red arrow = dimer, and blue arrow = trimer.

We sought to identify the specific domains of NLRP3 that were most sensitive to crosslinking induced by treatment with CBR-470-2 or exogenous MGO. NLRP3 contains multiple domains, including a pyrin (PYN) domain, a nucleotide-binding domain (NBD), and a leucine-rich repeat (LRR) domain. We expressed NLRP3-FLAG constructs lacking each of these individual domains (e.g., DPYN, DNBD, DLRR) or each individual domain on their own (e.g., PYN, NBD, LRR) in HEK293T cells and monitored crosslinking induced by treatment with CBR-470-2 or MGO. All these constructs demonstrated increases in HMW bands in cells treated with either CBR-470-2 or MGO (Figures 4B and S3C). However, loss of the PYN domain from the full-length NLRP3-FLAG construct strongly inhibited crosslinking under both conditions. Likewise, when expressed in isolation, the PYN domain showed the most robust increase in HMW bands when compared to other domains individually expressed. Recombinant PYN (residues 3–110) also forms HMW bands corresponding to MWs of PYN dimers and trimers when treated with 1 mM MGO for 1 h (Figures S3D–S3F). Thus, we probed the nature of this modification using the PYN domain construct.

MICA modifications proceed by crosslinking Cys residues to Arg residues. We overexpressed PYN domains lacking all Cys residues within this domain in HEK293T cells and monitored crosslinking induced by treatment with CBR-470-2 by immunoblotting. This showed that loss of all Cys residues prevented CBR-470-2-dependent increases in PYN crosslinking (Figure 4C). Reintroduction of C108 or C130 into the Cys-less PYN domain partially restored CBR-470-2-dependent crosslinking. Similar results were observed in MGO cells overexpressing these PYN Cys mutants, although all constructs showed some modest increase in MGO-dependent crosslinking (Figure S3G). These results identify C108 and C130 of the PYN domain as two sites for covalent modification. Using two alkyne-containing probe compounds, mCMF859 and P207–9175, which inhibit NLRP3 and covalently modify C130 in the PYN domain (Figure S3H), we further confirmed that CBR-470-2 treatment decreases the probe labeling of this Cys residue (Figure S3I). This further supports a model whereby MGO generated by CBR-470-2-dependent PGK1 inhibition leads to the covalent modification of this site.

Finally, to confirm that MICA modifications occur within the NLRP3 PYN domain, we overexpressed PYN in HEK293T cells treated with CBR-470-2 or MGO. We then excised the HMW band corresponding to the PYN dimer and the monomeric PYN domain from SDS-PAGE gels and subjected these isolates to enzymatic digestion. We monitored the formation of MICA modifications by mass spectrometry. Crosslinked peptides with the corresponding MICA mass addition (36 Da) were observed in isolates prepared from CBR-470-2- and MGO-treated samples between C08-R43, C08-R81, C38-R07, C38-R43, C108-R89 (CBR only), C130-R89, and C130-R126 and were not observed in the PYN monomer bands isolated from untreated cells (Figures 4D–4G and S4A–S4F). The formation of crosslinks involving C08 and C38, combined with the minimal increase in HMW bands for PYN constructs containing only these Cys residues (Figure 4C), suggests that these crosslinks are likely intramolecular. This suggests that apart from the intermolecular crosslinks observed by SDS-PAGE (evident by HMW bands), CBR-470-2 treatment also leads to intramolecular crosslinks. This mix of crosslinks suggests that NLRP3 is highly sensitive to glycolytic perturbations, resulting in MGO modification at multiple sites that integrate to inhibit NLRP3 inflammasome assembly and activity.

DISCUSSION

We discovered that inhibition of the glycolytic enzyme PGK1 in a monocyte cell line leads to the accumulation of MGO, a reactive metabolite that covalently crosslinks and inactivates NLRP3, the central pattern recognition receptor of the NLRP3 inflammasome. Inhibition of other glycolytic enzymes has been found to modulate the activity of the NLRP3 inflammasome, as hexokinase inhibition leads to increases in cytosolic mtDNA, which induces inflammasome activation.^32,33 Despite this, regulation of the NLRP3 inflammasome by glycolytic metabolites is poorly understood, as inhibition at different nodes of glycolysis leads to differing effects on inflammasome activity. For example, inhibiting glycolytic commitment using 2-deoxyglucose can either activate or inhibit the NLRP3 inflammasome.^33–35 Nevertheless, these data, along with the work presented here, provide evidence that glycolytic metabolism is intimately linked to NLRP3 inflammasome pathway activity. Certain cell types, like pro-inflammatory macrophages (M2 macrophages), which rely on augmented Warburg-like levels of glycolysis, may sense and respond to modulated glycolytic flux, and the mechanism delineated here may provide feedback to resolve inflammatory activation.

Inhibition of the NLRP3 inflammasome by MGO occurs as the result of covalent modification of NLRP3. MGO accumulates rapidly in response to CBR-470-2 treatment, and the first stage of covalent modification of NLRP3 by MGO is the mono-modification of cysteine(s), resulting in the formation of a hemithioacetal. This mono-modification is likely sufficient to inhibit the assembly of NLRP3, as we observe inhibition prior to observable crosslinking, but likely insufficient to induce its aggregation and insolubility. This hemithioacetal subsequently reacts with nearby arginine(s) to form the crosslinking MICA modification, which leads to the formation of NLRP3 oligomers. These crosslinks occur through a variety of cysteines and arginines across multiple domains of NLRP3. We reported that VLX1570, a covalent inhibitor of NLRP3, crosslinks NLRP3 through its multiple covalent reactive groups. The PYN domain was also the most sensitive to crosslinking by VLX1570, although, like MGO, we observed crosslinking of other NLRP3 domains. In both cases, the crosslinking appears to be non-selective for a specific site but instead forms disordered multimers of various sizes, which we posit inhibits the appropriate oligomerization and assembly of the NLRP3 inflammasome. Given the propensity of the PYN domain, and more generally NLRP3, it is possible that covalent crosslinking and inactivation is an evolved mechanism to sense the presence of an array of reactive metabolites.

Here, we show that MGO acts similarly to the reactive metabolites fumarate and itaconate, capable of both activating NRF2 and inhibiting the NLRP3 inflammasome. This concurrent signaling suggests a coordinated response to promote cellular survival and dampen inflammation. The extent that MGO might modulate additional related cellular responses will be of keen interest in future work. Nevertheless, these data suggest that one reactive electrophilic compound is capable of modulating both pathways, potentially for therapeutic benefit in disease. There are numerous disease states in which NRF2 activation and NLRP3 inhibition are known to be beneficial, including neurodegenerative diseases, metabolic diseases, gastrointestinal diseases, and autoimmune disorders. Using a compound such as CBR-470-2, which beneficially regulates both pathways, may provide an additive or synergistic therapeutic effect to improve disease outcome.

Limitations of the study

The protein targets and metabolic impacts of CBR-470 were previously characterized in IMR32 cells.²⁸ However, THP1 cells were primarily used for this study. While we demonstrate that PGK1 depletion recapitulates the reduction in NLRP3 inflammasome activity induced by treatment with CBR-470-2 or MGO, we cannot explicitly rule out the possibility that this compound inhibits inflammasome activity through a mechanism independent of PGK1 inhibition. Further, in our mass spectrometry, we only searched for crosslinks between cysteines and arginines within the NLRP3-PYN domain, leaving the potential for other CYS-ARG crosslinks within or between other domains of NLRP3 or other proteins that could contribute to the observed inflammasome inhibition.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Michael Bollong (mbollong@scripps.edu).

Materials availability

All cell lines, plasmids, and other stable reagents generated in this study are available from the lead contact with a completed Materials Transfer Agreement.

Data and code availability

• All data reported in this paper will be shared by the lead contact upon request.

• All original mass spectrometry data has been deposited at MassIVE and is publicly available as of the date of publication. DOIs are listed in the key resources table.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
NLRP3 (NBD)	Cell Signaling	RRID:AB_2722591
NLRP3 (PYN)	Adipogen	RRID:AB_2490202
Tubulin	Sigma	RRID:AB_477584
HIS-Tag	Santa Cruz	RRID:AB_627727
FLAG	Sigma	RRID:AB_262044
GFP	Abcam	RRID:AB_2313768
ASC	Cell Signaling	RRID:AB_2798325
CASP1	Cell Signaling	RRID:AB_2069051
NEK7	Cell Signaling	RRID:AB_2150676
IL-1β	GeneTex	RRID:AB_378141
Chemicals, peptides, and recombinant proteins
Ultrapure LPS, E. coli 0111:B4	Invivogen	tlrl-3pelps
MCC950	SelleckChem	S7809
Nigericin Sodium Salt	Cayman Chem	11437-25
Methylglyoxal	Acros Organics	175791000
Lipofectamine™ RNAiMAX Transfection Reagent	Invitrogen	13778075
ML385	Cayman Chemical	21114-5
Hoechst	Invitrogen	H3570
GSH	Sigma Aldrich	458139
ATP	Jena Bioscience	NU-1010-10G
Val-boroPro	Invivogen	tlrl-vbp-10
LFn-Rod	Invivogen	tlrl-rod
Anthrax Protective Antigen	Fisher Scientific	PEP0210
QUANTI-Blue™ Solution	Invivogen	rep-qbs
CellTiter-Glo	Promega	G7570
Poly-d-lysine	Thermo Fisher	A3890401
Optimem	Gibco	31985062
Fugene	Promega	E2311
TNF-α	Life Technologies	A42552
RIPA	Millipore	20188
1.7 Tris((1-benzyl-1H-1,2,3-triazol-4-yl)methyl)amine	Sigma	678937
tBuOH	Thermo Scientific	041857.AK
CuSO4	Sigma	C1297
tris(2-carboxyethyl)phosphine	RPI	T26500
1,8-diaminonaphthalene	Sigma	D21405
Critical commercial assays
Caspase-Glo® 1 Inflammasome Assay	Promega	G9951
RNeasy Mini Kit	Qiagen	74106
High-Capacity cDNA Reverse Transcription Kit	Advanced Biosystems	4368814
Power SYBR Green PCR Master Mix	Advanced Biosystems	4367659
Anti FLAG magnetic beads	Thermo Fisher	Cat#M8823
Deposited data
MICA Modification Mass Spectrometry Data	MassIVE	MSV000095503
Experimental models: cell lines
THP1-ASC-GFP Cells	Invivogen	thp-ascgfp
THP1-Null	Invivogen	thp-null
HEK-Blue™ IL-1β Cells	Invivogen	hkb-il1bv2
RAW-ASC Cells	Invivogen	raw-asc
Oligonucleotides
PGK1 siRNA 1	IDT	rArCrCrUrUrCrCrArUrGrUrCrArArGrArUrUrCrArGrCrUAG,
DNA Duplex		rCrUrArGrCrUrGrArArUrCrUrUrGrArCrArU rGrGrArArGrGrUrUrU
PGK1 siRNA 2	IDT	rCrArUrCrArArArUrUrCrUrGrCrUrUrGrGrArCrArArUrGGA,
DNA Duplex		rUrCrCrArUrUrGrUrCrCrArArGrCrArGrArAr UrUrUrGrArUrGrCrU
PGK1 siRNA 3	IDT	rGrCrUrCrUrCrArGrCrArArUrArUrUrUrArGrUrArCrUrUTC,
DNA Duplex		rGrArArArGrUrArCrUrArArArUrArUrUrG rCrUrGrArGrArGrCrArU
PGK1 Forward	IDT	CAAGCTGGACGTTAAAGGGA
NLRP3 Forward	IDT	GATCTTCGTTGCGATCAACA
Pro-IL-1B Forward	IDT	AGCTCGCCAGTGAAATGATG
NQO1 Forward	IDT	GCCTCCTTCATGGCATAGTT
HMOX1 Forward	IDT	GAGTGTAAGGACCCATCGGA
GAPDH Forward	IDT	AATGAAGGGGTCATTGATGG
RiboPro Forward	IDT	CGTCGCCTCCTACCTGCT
PGK1 Reverse	IDT	CAAGCTGGACGTTAAAGGGA
NLRP3 Reverse	IDT	GGGATTCGAAACACGTGCATTA
Pro-IL-1B Reverse	IDT	GGTGGTCGGAGATTCGTAGC
NQO1 Reverse	IDT	GGACTGCACCAGAGCCAT
HMOX1 Reverse	IDT	GCCAGCAACAAAGTGCAAG
GAPDH Reverse	IDT	AAGGTGAAGGTCGGAGTCAA
RiboPro Reverse	IDT	CCATTCAGCTCACTGATAACCTTG
Recombinant DNA
NLRP3-MYC-FLAG	Origene	RC220952
Software and algorithms
GraphPad Prism 10	Prism	RRID:SCR_002798

STAR★METHODS

EXPERIMENTAL MODEL AND SUBJECT PARTICIPANT DETAILS

Cell lines

WT THP1 cells, THP1-ASC-GFP Reporter cells, and HEK-Blue IL-1β cells were obtained from Invivogen and maintained according to Invivogen protocols. HEK293T cells were obtained from ATCC and maintained according to ATCC protocols. All cells were maintained at 37°C in 5% CO2 in a humidified incubator according to ATCC’s recommendations.

METHOD DETAILS

Methods generally are adapted from Stanton et al.¹⁸ and are here reviewed with any modifications.

Compounds, antibodies, and plasmids

Ultrapure LPS, E. coli 0111:B4 was dissolved in ultrapure water and administered at 1 μg/mL. MCC950 was dissolved in water and administered at 10 μM. Nigericin Sodium Salt dissolved in ethanol and administered at 10 μM. Methylglyoxal (MGO) was freshly diluted in cell culture media prior to administration at the indicated concentrations.

Synthesis of CBR-470-1 and CBR-470-2

CBR-470-1 and CBR-470-2 were synthesized in house as described in Bollong et al.²⁸ and are as follows.

CBR-470-1

A solution of 4-chlorothiophenol (2.56 g, 17.7 mmol) was added dropwise to a stirred solution of N-bromosuccinimide (3.15 g, 17.7 mmol) in CH₂Cl₂ (25 mL). After 30 min of stirring, a solution of 3-sulfolene (2.09 g, 17.7 mmol) was added dropwise to the reaction mixture. After 2 h of stirring, the reaction mixture was quenched with H₂O. The aqueous layer was extracted with CH₂Cl₂ and the combined organic layers were dried over MgSO₄ and concentrated in vacuo. The residue was purified by flash column chromatography on silica gel (EtOAc: n-hexane = 1 : 4) to give 2.04 g (34%) of 2 as white solid. Pyridine (1.20 mL, 14.9 mmol) was added to a stirred solution of 2 (2.04 g, 5.94 mmol) in CH₂Cl₂ (15 mL). After 3 h of stirring at 70°C, the reaction mixture was cooled to room temperature and quenched with saturated aq. NH₄Cl. The aqueous layer was extracted with CH₂Cl₂ and the combined organic layers were dried over MgSO4 and concentrated in vacuo. The residue was purified by flash column chromatography on silica gel (EtOAc: n-hexane = 1 : 3) to afford 513 mg (33%) of 3 as white solid. m-CPBA (14.2 g, 57.5 mmol, 70–75%) was added to a stirred solution of 3 (6.01 g, 23.0 mmol) in CH₂Cl₂ (100 mL). After 3 h of stirring, the reaction mixture was filtered and quenched with saturated aq. NaHCO₃. The aqueous layer was extracted with CH₂Cl₂ and the combined organic layers were dried over MgSO₄ and concentrated in vacuo. The crude mixture was used for next step without further purification. i-butylamine (2.29 mL, 23.0 mmol) was added to a stirred solution of the sulfone in CH₃CN (50 mL). After 2 h of stirring, the reaction mixture was concentrated in vacuo. The residue was purified by preparative HPLC and lyophilized to afford 1.60 g (19%) of CBR-470-1 as white solid.

CBR-470-2

3,4-dichlorobenzenethiol (1.62 mL, 12.7 mmol) was added dropwise to a stirred solution of N-bromosuccinimide (2.26 g, 12.7 mmol in CH₂Cl₂ (20 mL). After 30 min of stirring, a solution of 3-sulfolene (1.50 g, 12.7 mmol) was added dropwise to the reaction mixture. After 2 h of stirring, the reaction mixture was quenched with H₂O. The aqueous layer was extracted with CH₂Cl₂ and the combined organic layers were dried over MgSO₄ and concentrated in vacuo. The residue was purified by flash column chromatography on silica gel (EtOAc: n-hexane = 1 : 5) to give 974 mg (20%) of 4 as white solid. Pyridine (0.260 mL, 3.23 mmol) was added to a stirred solution of 4 (487 mg, 1.29 mmol) in CH₂Cl₂ (10 mL). After 1 h of stirring at 70°C, the reaction mixture was cooled to room temperature and quenched with saturated aq. NH₄Cl. The aqueous layer was extracted with CH₂Cl₂ and the combined organic layers were dried over MgSO4 and concentrated in vacuo. The residue was purified by flash column chromatography on silica gel (EtOAc: n-hexane = 1 : 2) to give 310 mg (81%) of 5 as white solid. m-CPBA (2.18 mg, 8.85 mmol, 70–75%) was added to a stirred solution of 5 (1.05 g, 3.54 mmol) in CH₂Cl₂ (20 mL). After 2 h of stirring, the reaction mixture was filtered and quenched with saturated aq. NaHCO₃. The aqueous layer was extracted with CH₂Cl₂ and the combined organic layers were dried over MgSO₄ and concentrated in vacuo. The crude mixture was used for next step without further purification. Glycine (0.104 mL, 1.05 mmol) and 1N NaOH (2.0 mL) was added to a stirred solution of the sulfone in CH₃CN (20 mL). After 1 h of stirring, the reaction mixture was concentrated in vacuo. The residue was purified by preparative HPLC and lyophilized to give 483 mg (34%) of CBR-470-2 as white solid.

Cell culture and ASC speck assay

siRNA transfections were performed using Lipofectamine RNAiMAX Transfection Reagent following manufacturer’s protocol with 9 μL of RNAiMAX reagent and 3 μL of 10 μM siRNA in 300 μL of Optimem per 2 million cells in a 6 well dish. Cells with knocked down PGK1 from siRNA experiments were tested 2 days following transfection in respective assays or harvested for gene knockdown confirmation. For ASC-GFP Speck assay, cells were plated in black, clear-bottom 384 well plates at 20,000 cells/well in 50μL of media. Compounds in dose response were transferred using an Agilent Bravo outfitted with a pintool head to transfer 100 nL. Following pre-treatment time (1–24 h), 10 μL of media containing Nigericin and Hoechst were added to each well for 2 h (final concentration 10 μM Nigericin, 5 μg/mL Hoechst). Cells were imaged on the Cellinsight CX5 HCS Platform and ASC-Specks were quantified using the SpotDetector function of the Cellinsight High Content Analysis Platform. GSH curve shift assays were performed in a similar manner except 10 μL of GSH (final concentration 10 mM) was added 15 min prior to addition of CBR-470 compounds.

Caspase-1 activity assay

The Caspase-Glo 1 Inflammasome Assay was followed according to manufacturer’s instructions. In summary, THP1 cells primed with LPS (1 μg/mL, 16h) and plated at a density of 400,000 cells/mL. The following day, cells were plated 40 μL in a 384 well dish. Compounds were added in 10 μL of media and incubated for 2 h. 1 μM Nigericin was added to each well for 45 min. Subsequently, caspase 1 reagent with MG132 and ac-YVAD were added to respective wells. Reactions were incubated at room temperature for 10 min and luminescence was measured with an Envision plate reader.

IL-1β secretion inhibition assay

WT THP1, RAW 264.7 cells ectopically expressing ASC, or murine primary dendritic cells were primed and treated with compound at the indicated concentrations. Following pre-treatment of 2 h with inhibitors, cells were treated with ATP pH 7.4 (final concentration 5 mM), Val-boroPro (Final concentration 10μM), or LFn-Rod and protective antigen (440 ng/mL, 220 ng/mL respectively). Cells were allowed to secrete IL-1β overnight. The next day, 10,000 HEK-Blue IL-1β cells were plated in 30 μL of media in black, clear bottom 384-well plates, and 10 μL of IL-1β conditioned media added to each well. Cells were incubated overnight to produce SEAP. The following day, 30 μL of QUANTI-Blue was added to each well and incubated at 37°C for 30 min-24 h (until visibly observable signal) and absorbance at 655 nM was measured.

Pyroptotic cell death assay

WT THP1 cells were primed with LPS, pretreated with compounds, and activated with Nigericin as described in the ASC-Speck Assay. 10 μM MCC950 was used as a control to inhibit pyroptotic cell death, and cells not treated with Nigericin were used for no pyroptotic cell death (maximal viability). Two hours after addition of Nigericin, 30 μL of CellTiter-Glo (diluted 1:6 in water), was added to each well and luminescence was measured with an Envision plate reader.

qPCR

Cells were lysed and RNA purified using a RNeasy Mini Kit following the manufacturer’s protocol. 400 μg of RNA was converted to cDNA using the High-Capacity cDNA Reverse Transcription Kit which was then diluted 1:4 with DNase/RNAse free water. qPCR reactions were prepared using Power SYBR Green PCR Master Mix and gene specific primers. qPCR reactions were performed using the QuantStudio 7 Flex with an initial melting period of 95°C for 10 min and then 40 cycles of 15 s at 95°C, 1 min at 60°C.

Immunoblotting and immunoprecipitation

To express NLRP3 in HEK293T cells, 10⁶ cells were plated on 6-well plates coated with poly-d-lysine and transfected with 2 μg of DNA per well and 7 μL of Fugene in 100 μL of Optimem. After 48 h, the cells were treated with compound. For experiments done with endogenous NLRP3, WT THP1 cells were primed overnight with 1 μg/mL LPS, 1 ng/mL TNF-α and treated with compound before collection. Each well was harvested in RIPA buffer with protease inhibitor and lysed on ice for >15 min. Insoluble material was separated via centrifugation. Concentration of soluble lysates was measured via absorbance with a nanodrop instrument. Lysate samples loaded typically were 50 μg. For samples with FLAG immunoprecipitation, lysates were normalized to 1 mg/mL and 20 μL of Magnetic FLAG Beads were added to each sample and incubated at 4°C overnight. Beads were immobilized using a magnetic Eppendorf holder and washed 2X with lysis buffer and 1X with DPBS. Beads were either resuspended in loading dye for analysis via Western blot, or in 100 μL of DPBS for click chemistry. Click Chemistry Master Mix was comprised of 6 μL of 1.7 Tris((1-benzyl-1H-1,2,3-triazol-4-yl) methyl)amine (Sigma) in 4:1 tBuOH:DMSO solution, 2 μL of 50 mM CuSO4 (Sigma) in water, 2 μL of 5 mM rhodamine-azide in DMSO and 2 μL of 50 mM tris(2-carboxyethyl)phosphine (TCEP) in water. To each sample, 12 μL of click chemistry master mix was added and incubated for 1 h in the dark at room temperature. Beads were immobilized using a magnetic Eppendorf holder and washed 2X with DPBS. Beads were resuspended in loading dye for analysis via Western blot and boiled for 15 min at 95°C. Protein material was separated via SDS–polyacrylamide gel electrophoresis (SDS-PAGE) (12, 15, or 17 well gels, 4–20% Bis-tris BOLT gels, Invitrogen), with 1X MOPS Buffer (Invitrogen) at 200 V for 40 min. Proteins were transferred to polyvinylidene difluoride (PVDF) membrane using a semi-dry transfer at 20 V for 1 h, and the PVDF membrane was blocked with 5% blotting grade milk in TBST (20 mM Tris, 137 mM NaCl, 0.1% Tween 20) for 1 h. Membranes were incubated with respective primary antibodies overnight at 4°C in 5% blotting grade milk or 5% BSA. The next day, membranes were washed 6X over 30 min with TBST, and then incubated in species specific secondary antibody (LI-COR, 1:5000), (HRP, 1:3333) in 5% milk for 1 h. Blots were washed 10X over 1 h with TBST and either imaged using LI-COR fluorescence imager or incubated in HRP substrate (West Dura) and developed with autoradiography film.

Cellular methylglyoxal level assay

WT THP1 cells were treated with CBR-470-2 or MGO for the indicated time frames and concentrations. Cells were harvested via centrifugation, rinsed twice with DPBS, and then lysed in 100 μL of DPBS per 1 million cells via probe sonication. Insoluble material was separated via centrifugation. 30 μL of clarified lysate, 50 μL of DPBS, and 10 μL of 100 μM 1,8-diaminonaphthalene were added to each well in a clear 96 well plate and incubated for 30 min. Fluorescence (Ex: 355 nm, Em: 400 nm) was measured using a SpectraMax iD3 plate reader. Signal was normalized to baseline fluorescence of vehicle treated cells and change in fluorescence compared to baseline was plotted.

Mass spectrometry

The gel band sample was destained, reduced (10 mM DTT), alkylated (55mM iodoacetamide) and digested with trypsin overnight before being analyzed by nano-LC-MS/MS. The nano LC-MS/MS analysis was carried out on a Thermo Scientific Easy nano LC 1200 coupled with Thermo Scientific Q Exactive Plus using Nanospray Flex ion source. Eight μL of digested peptides were analyzed by reverse-phase chromatography using 14cm length × 75mm inner diameter nanoelectrospray capillary column packed in-house with Phenomenex Aqua 3 μm C18 125 Å. Mobile phase A and B were water +0.1% formic acid and 80% acetonitrile +0.1% formic acid. The elution gradient started at 2% B for 5min, ramped up to 25% B for 100mins, 40%B for 20 min, 95%B for 1 min and held at 95% B for 14 min. Data acquisition was performed using Xcalibur (version 4.3). One MS scan of m/z 400–2000 was followed by 10 MS/MS scans on the most abundant ions with application of the dynamic exclusion of 10sec.

Recombinant PYN Domain Expression:

NLRP3 PYN Domain (3–110) was cloned into a PET28b vector and transformed into BL21 (DE3) competent E. coli. Starter cultures were grown overnight in Luria broth, and 250 mL cultures inoculated with 2.5 mL each of starter culture. Cultures were induced with 0.1 mM IPTG when they reached an OD₆₀₀ of 0.6–0.8, and then grown overnight at 18°C. Cultures were collected by centrifugation and lysed via sonication in lysis buffer (20 mM HEPES (pH 7.5), 150 mM NaCl, and 0.5 mM tris(2-carboxyethyl)phosphine (TCEP)). Lysate was clarified by centrifugation and soluble lysate was incubated with Ni-NTA Agarose beads (Qiagen, 30210) for 2 h at 4°C. The beads were added to a column, the unbound fraction flowed through and removed, and the beads washed with 50 mM imidazole. His-tagged NLRP3 PYN was eluted with 200 mM imidazole and purification was confirmed with SDS-PAGE and Coomassie staining.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistics were calculated in PRISM 10 (GraphPad, San Diego, CA). Data are presented as mean ± SEM and were analyzed by the student t-test or one- or two-way ANOVA, as indicated in the accompanying figure legends.

Highlights.

• PGK1 inhibition by CBR-470 compounds inhibits the assembly of the NLRP3 inflammasome

• Accumulation of the glycolytic byproduct methylglyoxal inhibits the inflammasome

• Methylglyoxal inhibits by modifying NLRP3 via covalent crosslinking MICA modifications