Covalent targeting as a common mechanism for inhibiting NLRP3 inflammasome assembly

The NLRP3 inflammasome is a cytosolic protein complex important for the regulation and secretion of inflammatory cytokines including IL-1β and IL-18. Aberrant overactivation of NLRP3 is implicated in numerous inflammatory disorders. However, the activation and regulation of NLRP3 inflammasome signaling remains poorly understood, limiting our ability to develop pharmacologic approaches to target this important inflammatory complex. Here, we developed and implemented a high-throughput screen to identify compounds that inhibit inflammasome assembly and activity. From this screen we identify and profile inflammasome inhibition of 20 new covalent compounds across 9 different chemical scaffolds, as well as many known inflammasome covalent inhibitors. Intriguingly, our results indicate that NLRP3 possesses numerous reactive cysteines on multiple domains whose covalent targeting blocks activation of this inflammatory complex. Specifically, focusing on compound VLX1570, which possesses multiple electrophilic moieties, we demonstrate that this compound allows covalent, intermolecular crosslinking of NLRP3 cysteines to inhibit inflammasome assembly. Our results, along with the recent identification of numerous covalent molecules that inhibit NLRP3 inflammasome activation, suggests that NLRP3 serves as a cellular electrophile sensor important for coordinating inflammatory signaling in response to redox stress. Further, our results support the potential for covalent cysteine modification of NLRP3 for regulating inflammasome activation and activity.

inflammasome inhibitors. Here, we employed a high-throughput screen (HTS) of diverse chemical libraries to identify >450 new inhibitors of NLRP3 inflammasome assembly across numerous chemical scaffolds.
Intriguingly, the vast majority of these compounds appear to be covalent modifiers of NLRP3 cysteines, suggesting that NLRP3 is highly sensitive to electrophilic modification. We demonstrate that these covalent compounds, including known NLRP3 inhibitors, modify NLRP3 on multiple cysteines in various domains. These results indicate that NLRP3 is likely a highly sensitive electrophile sensor within the cell, revealing insights into the regulation of this complex during conditions of stress and further motivating the development of new covalent ligands to inhibit NLRP3 inflammasome activity in the context of disease.

HTS Identification of inhibitors of NLRP3 inflammasome assembly and activity:
To identify novel inhibitors of NLRP3 inflammasome activation, we developed a high-content, imaging-based screening assay of inflammasome assembly using THP1-ASC-GFP cells (Invivogen). These cells express GFPtagged ASC under control of a NFkB-regulated promoter. In the absence of stimuli, ASC-GFP is diffuse throughout the cell; however, upon administration of an activating stimulus such as ATP or nigericin, ASC-GFP forms discrete puncta, referred to as ASC specks, which reflects inflammasome assembly ( Fig. 1a,b, Fig. S1a).
We monitored ASC speck formation in THP1-ASC-GFP cells in 384-well format through induction of specks in cells pretreated with LPS (priming signal), followed by treatment with nigericin (activation signal) (Fig. 1a). We then quantified specks per cell by defining the number of GFP puncta in individual cells. Treatment with nigericin significantly increased ASC specks in THP1-ASC-GFP cells primed with LPS (Fig. 1c). Pre-treatment with known inflammasome inhibitors such as MCC950 or oridonin robustly inhibited speck formation (Z' = 0.8 for MCC950), confirming the sufficiency of this assay for HTS.
We then used this imaging-based assay to screen the Calibr ReFRAME library, comprised of ~14k compounds that have undergone extensive preclinical or clinical testing, for inhibitors of NLRP3 inflammasome assembly [40]. Initially, we performed this screen by pre-treating cells for 24 h with compound prior to nigericin administration. This identified 111 hits that inhibited inflammasome assembly (Fig. S1b), which included 17 inhibitors of HSP90, a known mechanism of NLRP3 inflammasome inhibition [41,42]. Intriguingly, we also identified 15 compounds with known cysteine reactive moieties, suggesting these compounds inhibit inflammasome assembly through a covalent mechanism. To better define this mechanism, we rescreened the ReFRAME library in THP1-ASC-GFP cells pre-treated with compounds for 1 h -a timepoint that would minimize potential off-target inhibitory mechanisms and increase identification of direct inhibitors of inflammasome assembly. We also expanded this screen to include a library of covalent compounds and a library of nucleoside mimetics. This more focused screen identified 487 compounds that inhibited inflammasome assembly (Fig. 1d).
While the libraries screened were chemically diverse and contained many non-covalent scaffolds, an overwhelming number of top hits contained electrophilic moieties suggesting they likely act by a covalent mechanism ( Table 1). From this screen, we selected 20 of the most potent compounds from 9 potentially covalent, structurally diverse scaffolds for further characterization (Fig. 1e, Fig. S2a). We showed that these compounds inhibit inflammasome assembly with IC50 values ranging from 30 nM to 2 µM ( Table 1).
Given that inflammasome activation leads to the secretion of the pro-inflammatory cytokine IL-1b and pyroptotic cell death, we established secondary screening assays to evaluate if the identified inhibitors of inflammasome assembly also decreased IL-1b secretion from WT THP-1 cells co-treated with LPS and ATP (HEK-Blue IL-1b reporter assay) and reduced pyroptotic cell death of WT THP-1 cells treated with LPS and nigericin (measured by CellTiter-Glo). We confirmed that pre-treatment with the established NLRP3 inhibitor MCC950 inhibited IL-1b secretion and prevented inflammasome-dependent cell death of WT THP-1 cells (Fig.   S2b,c). Similarly, all prioritized NLRP3 inhibitors identified in our HTS decreased IL-1β secretion and improved viability of stimulated WT THP-1 cells ( Table 1). These results confirm that, apart from inflammasome assembly, NLRP3 inhibitors block downstream activities of this complex implicated in inflammatory disease.

VLX1570 promotes NLRP3 crosslinking:
Prioritized inhibitors identified in our HTS all contain potential sites of covalent modification that indicate these compounds likely act through a covalent mechanism. To initially test if these compounds act covalently, we monitored ASC speck formation in stimulated THP1-ASC-GFP cells pretreated with glutathione (GSH) prior to the addition of covalent inhibitor. We found that IC50 for speck formation increased ~10-fold for compounds administered in the presence of GSH (Fig. 2a), consistent with a covalent targeting mechanism. We also evaluated the toxicity of these compounds by measuring changes in viability of WT THP-1 cells 24 h after treatment. We observed some toxicity for select compounds at higher doses (Table 1); however, this toxicity did not preclude us from utilizing these compounds for monitoring NLRP3 inhibition at shorter timepoints.
The most potent compound identified in our screen was VLX1570 (Fig. 1e, Table 1, Fig. 2b) -a compound that was developed as a covalent inhibitor of deubiquitinating enzymes (DUBs) including USP14 and UCHL5 with an IC50 of DUB inhibition of ~10 µM in vitro [43,44]. Thus, we sought to better define the activity of this compound. Initially, we monitored the in vivo activity of VLX1570 to inhibit inflammasome-associated proinflammatory signaling. Towards that aim, we IP administered VLX1570 1 h prior to IP administration of LPS.
We then collected peritoneal fluid and monitored levels of pro-inflammatory cytokines 3 hr after LPS administration. Interestingly, we found that VLX1570 reduced levels of both IL-1b and IL-6 in these animals ( Fig.   2c). However, VLX1570 did not influence TNF-α -a cytokine that is not regulated by inflammasome activity. We did observe reductions in both pro-IL-1b and NLRP3 in peritoneal cells by immunoblotting (Fig. S3a). This suggests decreased immune infiltration in the peritoneum. These results indicate that VLX1570 inhibits inflammasome-dependent pro-inflammatory signaling in LPS-treated mice.
VLX1570 contains three potential sites for covalent modification (Fig. 1e). Intriguingly, immunoblotting shows that treatment with VLX1570 increases high molecular weight (HMW) populations of NLRP3 in LPStreated THP1 cells (Fig. 2d, Fig. S3b). This appears specific to NLRP3, as other inflammasome components NEK7, ASC, or CASP1 did not show similar effects (Fig. S3c). We observed increased levels of HMW NLRP3 in insoluble fractions from VLX1570-treated THP1 cells, suggesting that this modification reduces protein solubility (Fig. S3d). VLX1570-dependent increases in HMW NLRP3 was also observed in HEK293T cells overexpressing NLRP3-MYC-FLAG (Fig. S3e,f). Interestingly, NLRP3-GFP efficiently co-eluted with HMW . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. ; https://doi.org/10.1101/2023.06.01.543248 doi: bioRxiv preprint NLRP3-MYC-FLAG in FLAG IPs from VLX1570-treated HEK293T cells co-overexpressing these proteins, indicating that these HMW species contain multiple crosslinked NLRP3 proteins (Fig. 2e). However, we did not observe USP14 or UCHL5 in NLRP3 immunopurifications from THP1 cells treated with VLX1570, indicating that these HMW NLRP3 species do not contain known targets of VLX1570 (Fig. S3g). We showed that recombinant NLRP3 also forms VLX1570-dependent HMW species indicating these HMW species are comprised of intermolecularly cross-linked NLRP3 molecules and the formation of which is not dependent on alternative cellular functions or additional proteins (Fig. S3h).
NLRP3 contains 45 cysteine residues localized throughout its pyrin (PYN), nucleotide binding (NBD), and leucine-rich repeat (LRR) domains (Fig. 2f). To identify the specific domain responsible for VLX1570-dependent induction of HMW species, we overexpressed these individual domains and monitored formation of HMW species by immunoblotting. Surprisingly, we found that each overexpressed domain showed increased formation of HMW species, indicating that VLX1570 can induce crosslinking across multiple different cysteine residues ( Fig. 2g). This suggests that VLX1570 specifically crosslinks proteins containing multiple electrophile sensing cysteine residues. Thus, to identify specific cysteines modified by VLX1570, we specifically focused on the NLRP3 pyrin domain, which contains only 4 cysteines (Fig. S3k). Importantly, mutation of all cysteines within the pyrin domain to serine completely blocked VLX1570-dependent induction of HMW species (Fig. 2h). Next, we re-introduced these cysteines back into the cysteine-less pyrin domain to identify those specifically modified by VLX1570. Interestingly, while we found that all cysteines showed some modification, Cys 108 and Cys 130 were shown to be the two cysteine residues that most strongly induced formation of HMW species (Fig. 2h).
This indicates that these two cysteines may be more sensitive to electrophilic modification. Recombinant pyrin domain, only retaining cysteine 108, forms HMW pyrin species confirming C108 as a site of modification for VLX1570 (Fig. S3i,j). This further highlights the ability of this compound to modify multiple cysteines on NLRP3.
We propose that VLX1570 may target highly reactive cysteines such as those found on electrophile sensors.
Consistent with this, we found that VLX1570 also induced crosslinking of KEAP1-FLAG -another electrophile sensor containing multiple reactive cysteine residues -in HEK293T cells (Fig. S3l). Collectively, these results suggest that VLX1570 induces crosslinking of NLRP3 through reactive cysteines on multiple different domains to block inflammasome assembly and activation.

A 2-Sulfonylpyrimidine-Containing Scaffold Covalently Modifies NLRP3
To further characterize the capacity for electrophile sensing by reactive cysteines on NLRP3, we selected two novel covalent scaffolds, a 2-sulfonylpyrimidine containing scaffold (e.g., D053-0025) and a chloroacetamide containing scaffold (e.g., K784-5117), for further study. These compounds inhibit ASC-speck formation with IC50 values of 84 and 330 nM, respectively (Fig. 1e). We next performed structure activity profiling of each scaffold to identify specific locations within their structures to incorporate click-compatible alkyne moieties (Fig. 3a, Fig.   S4a-c). These efforts established that the 2-sulfonylpyrimidine scaffold was a more tractable scaffold for the optimization of inhibitory potency, resulting in the identification of D053-0367 (IC50 = 70 nM) (Fig. 3b,c) which acted covalently in the GSH shift assay (Fig. 3d) and inhibited pyroptotic cell death (Fig. 3e), as well as an alkyne derivatized probe compound (P207-9174, IC50 = 70 nM). We confirmed that this probe efficiently inhibited ASC speck formation in THP1-ASC-GFP cells to extents similar to that observed for the parent compound (Fig.   3f).

Covalent inflammasome inhibitors modify cysteines of NLRP3 on multiple domains.
The 2-sulfonylpyrmidine scaffold can modify biological thiols covalently, acting through an SNAr based mechanism, resulting in the elimination of the sulfonyl group (Fig. S5a). The nature of this modification was confirmed via mass spectrometry of D053-0367 incubated with GSH (Fig. S5b). We then probed the ability of this compound to covalently modify NLRP3. We treated THP-1 cells with probe compound and then monitored NLRP3 labeling by appending a biotin to the probe using click chemistry followed by streptavidin isolation and immunoblotting for NLRP3. P207-9174 showed dosable modification of NLRP3 using this assay (Fig. S5c). Dose dependent modification was also observed for overexpressed FLAG-NLRP3 in HEK293T cells (Fig. S5d).
Further, we did not observe labeling of other inflammasome subunits such as NEK7 by this probe (Fig. S5e).
Importantly, probe-dependent labeling of NLRP3 was inhibited by co-treatment with excess parent compound, demonstrating that this labeling reflected compound modification (Fig. 4a). Collectively, these results confirm that the identified 2-sulfonylpyrimidine scaffold covalently modifies NLRP3.
We next sought to identify the specific NLRP3 domains modified by the 2-sulfonylpyrimidine probe. We overexpressed FLAG-tagged constructs of individual NLRP3 domains in HEK293T cells and monitored compound modification in FLAG immunopurifications followed by click chemistry-dependent appendage of a fluorescent rhodamine to the probe. Intriguingly, we found that the probe labeled all individually expressed domains of NLRP3, indicating that this compound targets multiple cysteines throughout the protein sequence ( Fig. 4b). To identify specific cysteines modified by this probe, we again focused on the NLRP3 pyrin domain, which contains only 4 cysteines (Fig. S3k). Mutation of all cysteines within the pyrin domain to serine completely blocked probe-dependent labeling (Fig. 4c) When these cysteines were re-introduced back into the cysteineless pyrin domain to identify those specifically modified by our probe, we found that while all cysteines showed some modification, Cys 8 and Cys 130 were shown to be the two cysteine residues most prominently labeled by our probe (Fig. 4c). This indicates that these two cysteines may be more sensitive to electrophilic modification.
This further highlights the ability of these compounds to modify multiple cysteines on NLRP3.
Numerous established covalent NLRP3 inflammasome inhibitors have been previously suggested to work by targeting specific cysteines on the protein structure. Thus, the ability for our probe to broadly modify multiple cysteines on different domains provides an opportunity to evaluate the specific or broad nature of cysteine targeting by these established compounds and other covalent compounds identified in our screen using a competition assay. Surprisingly, we found that the vast majority of covalent compounds, including the established NLRP3 inhibitors oridonin and RRx-001, competed with our probe in labeling of endogenous NLRP3 in THP1 cells (Fig. 4d). In contrast, no competition was observed for non-covalent NLRP3 inhibitors such as MCC950. These results indicate that covalent compounds, including the established inhibitor oridonin and many identified through our screens, appear to modify multiple cysteine residues throughout the NLRP3 sequence, indicating that the protein is highly sensitive to regulation through covalent modification.
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. ; https://doi.org/10.1101/2023.06.01.543248 doi: bioRxiv preprint

Concluding Remarks:
Our data demonstrates that NLRP3 inflammasome assembly can be inhibited by a variety of previously unpublished covalent chemical scaffolds at multiple cysteines. Identification of structurally diverse covalent inhibitors of NLRP3 indicates that NLRP3 may be a good covalent drug target for the treatment of inflammatory diseases through modification of cysteines resulting in inhibitory mechanisms not previously shown. This may be advantageous compared to an ATPase inhibitor, as covalent drugs often display increased specificity, potency, and duration of engagement [45]. Identified here, the deubiquitinase inhibitor VLX1570 inhibits the NLRP3 inflammasome through induction of high molecular weight NLRP3 aggregates via covalent cross-linking.
This demonstrates a novel, alternative mechanism for covalent NLRP3 inflammasome inhibition, independent of ATPase activity or NEK7 interactions. We also show that VLX1570 is active in vivo and can inhibit NLRP3 inflammasome-dependent inflammation induced by LPS challenge.
In conjunction with previous publications showing covalent modification and inhibition of NLRP3, this data indicates that numerous cysteines in NLRP3, including those in the pyrin domain which have not been previously shown to be modified, may be targeted by covalent molecules of varying chemical scaffolds. However, our data demonstrates the identified covalent inhibitor, VLX1570, the 2-sulfonylpyrimidine scaffold, and previously published inhibitors, are likely not modifying a single cysteine, but an array of NLRP3 cysteines which contribute to the inhibition. Additionally, this modification of multiple cysteines by each compound makes it difficult to increase potency of covalent compounds through structure activity relationship analysis as there is not a single binding site to optimize compound binding. Despite this, the high sensitivity of NLRP3 to covalent modification indicates that NLRP3 serves as an electrophile sensor within the cell to regulate inflammatory signaling, providing justification for further covalent NLRP3 inhibitor development for the treatment of inflammatory disease.
With further characterization of covalent inhibition at additional sensor cysteines on NLRP3, more potent and specific NLRP3 inhibitors may be developed in the future.

Conflict of interest statement:
The authors declare no competing interest.

Compounds, Antibodies, and Plasmids
The ReFRAME Library was accessed through CALIBR at Scripps Research. The covalent compound and the nucleoside mimetic libraries were purchased from ChemDiv. For follow-up assays, compounds identified from

IL-1β Secretion Inhibition Assay:
WT THP-1 cells were primed and treated with compound as described in ASC-Speck assay. Following pretreatment of 1 hr with covalent inhibitors, cells were treated with ATP in 10 µL water, pH 7.4 (Final concentration 5 mM). Cells were allowed to secrete IL-1β overnight. The following 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 was added to each well. Cells were incubated overnight to produce SEAP. The following day, 30 µL of QUANTI-Blue (Invivogen) was added to each well and incubated at 37 °C for 30 min-24 hr and absorbance at 655 nM was measured. IL-1β secretion inhibition by compounds was normalized to the maximal inhibition of MCC950.

Pyroptotic Cell Death Assay:
WT THP-1 cells were primed with LPS, pretreated with covalent inhibitor compound, and activated with Nigericin as previously 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 as a control for no pyroptotic cell death (maximal viability).
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. ; https://doi.org/10.1101/2023.06.01.543248 doi: bioRxiv preprint Two hours after addition of Nigericin, 30 µL of CellTiter Glo (Promega, diluted 1:6), was added to each well and luminescence was measured with an Envision plate reader. Compound cytotoxicity was measured using a similar protocol where compound was added to LPS-primed WT THP-1 cells and viability was measured 24 hr following compound treatment.

Immunoblotting and Immunoprecipitation:
For expression of NLRP3 in HEK293T cells, 10 6 cells were plated on poly-d-lysine coated 6-well plates and  Table 2. The next day, membranes were washed 6X over 30 min with TBST, and then incubated in secondary antibody (LI-COR, 1:5000), (HRP, 1:3333) in 5% milk for 1 hr. Blots were washed 10X over 1 hr with TBST and either imaged using LI-COR fluorescence imager or incubated in HRP substrate (West Dura) and developed with autoradiography film.

NLRP3 Target Confirmation and Competition:
10 mL of WT THP-1 cells at 1x10 6 cells/mL were primed overnight with 1 ng/mL LPS and either pre-treated with compound before probe treatment, or treated only with probe, and then collected. Cells were lysed in 1 mL of DPBS using a tip sonicator. Insoluble material was separated via centrifugation. Samples were diluted to 2 tubes . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. ; https://doi.org/10.1101/2023.06.01.543248 doi: bioRxiv preprint of 0.5 mL of 2 mg/mL of lysate and treated each with 55 µL of Biotin Click master mix comprised of 30 µl of 1.7 Tris((1-benzyl-1H-1,2,3-triazol-4-yl)methyl)amine (Sigma) in 4:1 tBuOH: DMSO solution, 10 µl of 50 mM CuSO4 (Sigma) in water, 10 µl of 5 mM biotin-PEG3-azide in DMSO and 10 µl of 50 mM tris(2carboxyethyl)phosphine (TCEP) in water, and incubated for 1 hr in the dark. Reaction was stopped with addition of 500µL of ice cold methanol and precipitated proteins were pelleted via centrifugation and rewashed with methanol. Protein pellets were resuspended by sonication in 12 mL of 0.1% SDS in DPBS to which 220 µL of pre-equilibrated streptavidin beads (Thermo Scientific, PI20357) were added. Samples were incubated overnight at room temperature and then was washed twice with 0.1% SDS in DPBS, twice with DPBS, and then twice with water, 10 mL each. Beads were resuspended in 200 µL of loading dye with DTT and boiled at 95 °C for 15 min to elute. Immunoblotting was performed as described above.

In vivo LPS Challenge:
To evaluate in vivo efficacy of VLX1570, 5 9-week-old C57BL/6J male mice per group (4 groups total) were  was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. ;     was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. ;  LD50 for THP-1 cells primed overnight with 1 ng/mL LPS and treated with compound for 24 hr. was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. ;    . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 1, 2023. ; https://doi.org/10.1101/2023.06.01.543248 doi: bioRxiv preprint