Chemical Modulation of the 1-(Piperidin-4-yl)-1,3-dihydro-2H-benzo[d]imidazole-2-one Scaffold as a Novel NLRP3 Inhibitor

In the search for new chemical scaffolds able to afford NLRP3 inflammasome inhibitors, we used a pharmacophore-hybridization strategy by combining the structure of the acrylic acid derivative INF39 with the 1-(piperidin-4-yl)1,3-dihydro-2H-benzo[d]imidazole-2-one substructure present in HS203873, a recently identified NLRP3 binder. A series of differently modulated benzo[d]imidazole-2-one derivatives were designed and synthesised. The obtained compounds were screened in vitro to test their ability to inhibit NLRP3-dependent pyroptosis and IL-1β release in PMA-differentiated THP-1 cells stimulated with LPS/ATP. The selected compounds were evaluated for their ability to reduce the ATPase activity of human recombinant NLRP3 using a newly developed assay. From this screening, compounds 9, 13 and 18, able to concentration-dependently inhibit IL-1β release in LPS/ATP-stimulated human macrophages, emerged as the most promising NLRP3 inhibitors of the series. Computational simulations were applied for building the first complete model of the NLRP3 inactive state and for identifying possible binding sites available to the tested compounds. The analyses led us to suggest a mechanism of protein–ligand binding that might explain the activity of the compounds.


Introduction
The nucleotide-binding oligomerization domain leucine rich repeat and pyrin domain containing protein 3 (NLRP3) inflammasome is a cytosolic pattern recognition receptor (PRR) that plays a fundamental role in the response to exogenous and endogenous stimuli. The NLRP3 inflammasome belongs to the NOD-like receptor (NLR) family of PRR; it is a multiprotein complex constituted by the NLRP3 protein, the apoptosis-associated specklike protein containing a caspase-recruiting domain (ASC) and procaspase-1 [1-3]. The NLRP3 protein can be seen as the central core of the NLRP3 inflammasome. It is constituted by three domains: the C-terminal leucine-rich repeat domain (LRR), the nucleotide-binding and oligomerization domain (NACHT) and the N-terminal pyrin domain (PYD). Each of these domains play a different role. The role of the LRR domain is still to be fully clarified: LRR appears to be dispensable for NLRP3 activation [4], however, by binding to NIMA-related kinase 7 (NEK7), it facilitates the ATP-dependent activation of NLRP3 [2,5]; the NACHT domain is mainly responsible for oligomerization and possesses ATP-binding and hydrolysis abilities [6]; finally, the PYD domain interacts with ASC through homotypic (PYD-PYD) interactions [7]. NLRP3 expression is induced following the stimulation by danger-associated molecular patterns (DAMPs) or pathogen-associated molecular patterns (PAMPs) of Toll-like receptors (TLRs), such as TLR4 receptors, or cytokine receptors, such as TNF receptor. This process, known as the priming step, is not mandatory in any cell types [8]. NLRP3 is mainly expressed in monocytes, macrophages, lymphocytes, neutrophils and dendritic cells; however, it is also expressed in microglia, astrocytes and epithelial cells [9,10]. NLRP3 usually needs a second signal, generally referred to as an activation signal, to be fully activated. Different signals can lead to inflammasome activation [7,[11][12][13]. Nigericin, ATP, molecular particulates and crystals induce, through different mechanisms, a K + efflux that is considered a common trigger for NLRP3 activation [14]. However, altered Ca 2+ homeostasis and other ion fluxes (e.g., Na + and Cl − ) have also been implicated in NLRP3 activation [15,16]. Several studies demonstrated that reactive oxygen species (ROS) production, lysosomal destabilization and post-translational modifications are other cellular triggers able to promote the activation and assembly of NLRP3 inflammasome [17][18][19][20][21]. Once activated, NLRP3 undergoes a conformational change and then oligomerizes to give rise to a functional inflammasome, which can trigger the auto-proteolytic cleavage of pro-caspase-1 into the active caspase-1 [22]. Finally, this cysteine protease converts the pro-inflammatory cytokines pro-interleukin (IL)-1β and pro-IL-18 into their active counterparts IL-1β and IL-18. The activated caspase-1 is also able to cleave the protein gasdermin-D (GSDMD). The cytotoxic GSDMD-N terminal fragments can assemble in large circular structures that, by localizing into the cellular membrane, cause the formation of pores that leads to the membrane rupture and the pyroptotic cell death [23]. This process, known as pyroptosis, contributes to the release of IL-1β and other intracellular material into the extracellular space, thus exacerbating the inflammatory process.
The aberrant activation of NLRP3 inflammasome is involved in the onset and progression of a wide range of human diseases [24], among which auto-inflammatory [25], inflammatory and autoimmune [26][27][28][29], neurodegenerative [30,31], cardiovascular [32][33][34] and metabolic diseases [35][36][37] are the most studied. It is not surprising that a concerted research effort toward the discovery of small molecules able to block NLRP3 activation is being performed both by industry and academia. These efforts have enabled the discovery of interesting compounds, some of which are now in clinical development [38][39][40].
Recently, our research group developed covalent NLRP3 inhibitors based on the acrylic acid scaffold (e.g., INF4E, INF39, INF58; Figure 1). These compounds bear an electrophilic moiety with a properly tuned reactivity toward nucleophiles and are able to block NLRP3 activation, NLRP3-dependent pyroptosis and the consequent release of IL-1β [41][42][43]. One compound, namely INF39, proved able to prevent DNBS-induced colitis in vivo after oral administration in mice at both 50 and 25 mg/kg [43,44]. INF39 inhibited ATP hydrolysis by isolated recombinant NLRP3 and prevented conformational changes necessary for NLRP3 activation in NLRP3 expressing HEK293 cells [43]. A recent report demonstrated that INF39 is also able to inhibit the NEK7-NLRP3 interaction attenuating NLRP3 assembly in macrophages [45].
A benzo [d]imidazole-2-one derivative, HS203873 (Figure 1), was recently identified using a fluorescent-linked enzyme chemoproteomic strategy (FLECS). HS203873 was able to bind to NLRP3 and its ability to inhibit NLRP3 activation and IL-1β release in differentiated THP-1 cells was confirmed [46]. To discover new molecular scaffolds endowed with NLRP3 inhibition potential, we applied a pharmacophore-hybridization strategy based on merging two molecular probes: the acrylate INF39 and HS203873. The latter molecules have been conjugated to obtain compound 1, bearing an electrophilic acrylamide substructure and is still able to inhibit pyroptotic cell death and IL-1β release in human macrophages. To investigate the possibility of generating compounds devoid of electrophilic properties, and thus to obtain non-covalent NLRP3 inhibitors, derivative 2 was synthesised. Since compound 2 showed reduced activity, we further modulated this scaffold to design more active non-covalent compounds able to block NLRP3 inflammasome activity.
In this work, we describe the synthesis and the preliminary in vitro pharmacological screening of a new series of benzo [d]imidazole-2-one derivatives ( Figure 2) as potential inhibitors of NLRP3-dependent IL-1β release and pyroptotic cell death in differentiated THP-1 cells, we report data on the ability of the selected compounds to attenuate the ATPase activity of immobilised NLRP3 protein, and we present a possible model of interaction with NLRP3 protein.
The second series of molecules (derivatives 6-8), bearing a spacer between the amide group and the piperidine ring, was synthesised using the synthetic approaches depicted in Scheme 2. The acid 29 was converted into its O-hydroxysuccinimide ester 29a using dicyclohexylcarbodiimide (DCC) and N-hydroxysuccinimide (NHS) and then reacted with the aminoesters 30-31 to afford the intermediates 32 and 33, which were hydrolysed in basic medium (NaOH 2.5 M) to give the acids 34 and 35. These acids were coupled with 26 under the HBTU/HOBt/DIPEA-mediated conditions described above to give the desired derivatives 6 and 8. The synthesis of 7 was attained with a slightly different procedure. The piperidine derivative 26 was first coupled with Boc-protected β-alanine using HBTU/HOBt/DIPEA as activating agents. The tert-butyl protected intermediate 36 was cleaved in 10% TFA in CH 2 Cl 2 to afford the free amine 37, which was subsequently coupled with hydroxysuccinimide ester 29a, under the previously described conditions to afford 7 in 36% overall yield.
The third series of molecules, formally derived by the opening of the piperidine ring, was synthesised according to Scheme 3. The aminoalkylbenzimidazol-2-one derivatives 44 and 45 were obtained through a nucleophilic substitution of the fluorine atom on 1fluoro-2-nitrobenzene using the Boc-protected alkylamines 38 and 39 to afford nitrobenzene derivatives 40 and 41. Catalytic hydrogenation over Pd/C 10% allowed the reduction of the nitro group to afford aniline derivatives 42 and 43, which were cyclised using CDI to afford benzimidazolone derivatives 44 and 45 in good overall yields. Deprotection with TFA 10% in CH 2 Cl 2 afforded the free amines 46 and 47, which were reacted with the activated ester 29a in DIPEA/DMF to afford the final compounds 9 and 10. To obtain the N-methyl substituted derivative 11, a different approach was used. The benzimidazol-2one was converted into the intermediate 48 by protection with di-tert-butyldicarbonate and subsequent alkylation with excess 1,3-dibromopropane. Nucleophilic substitution of the bromine with aqueous methylamine afforded 49, which, in its turn, was coupled with 29 using CDI to afford the protected intermediate 50. The usual deprotection with TFA 10% in CH 2 Cl 2 gave the desired 11 in modest yield.
Compounds 12 and 13, bearing a cyanoguanidine residue in place of the imidazole-2-one ring, were obtained in high yield according to the pathway described in Scheme 4. The acid 29 was converted into the corresponding acyl chloride using SOCl 2 /DMF at room temperature and immediately reacted with commercially available tert-butyl piperidin-4ylcarbamate to give 51 in nearly quantitative yield. The amino group was deprotected and reacted with diphenyl cyanocarbonimidate to afford the O-phenylisourea derivative 52.

Biological Evaluation
Different techniques could be used to study NLRP3 inhibitors [47]; the use of differentiated THP-1 cells treated with NLRP3 activating stimuli is suitable for the screening of large series of compounds [48]. All the synthesised compounds ( Figure 2) were evaluated for their NLRP3 inhibitory activity by measuring their ability to prevent NLRP3-dependent pyroptosis in differentiated THP-1 cells. THP-1 cells were differentiated into macrophages by treatment with phorbol-myristate acetate (PMA; 50 nM; 24 h) and then treated with lipopolysaccharide (LPS; 10 µg/mL; 4 h) to promote NLRP3 expression. The cells were then incubated with test compounds at the fixed concentration of 10 µM for 1 h. Test compounds were dissolved in serum free medium containing 0.1% DMSO as the vehicle. NLRP3 was activated by the addition of ATP (5 mM), and the pyroptotic cell death was evaluated after 1.5 h by measuring the lactate dehydrogenase (LDH) released in the cell supernatants. The obtained results, expressed as pyroptosis% decrease with respect to vehicle-treated cells, are collected in Table 1. In the same experiments, the release of IL-1β was measured via an ELISA assay and the results, expressed as % inhibition of IL-1β release versus vehicle-treated cells, are reported in Table 1. Finally, the cytotoxicity of the synthesised compounds was evaluated by MTT assay after treatment of THP-1 cells with increasing concentration of test compounds (0.1-100 µM) for 72 h. The results, expressed as TC 50 , are reported in Table 1. Analysis of the obtained results showed that by merging the structure of INF39 with the 1-(piperidin-4-yl)-2,3-dihydro-1,3-benzoimidazol-2-one moiety present in HS-203873, to obtain 1, a compound able to prevent pyroptosis (24.9 ± 6.3%) and IL-1β release (19.4 ± 0.4%) to a similar extent at 10 µM was generated. When the electrophilic substructure was eliminated (compound 2), the anti-pyroptotic activity dropped down to 14.9 ± 5.8% inhibition (p = 0.198); however, the effect of compound 2 appeared to be concentration-dependent as the pyroptosis inhibition increased to 29.1 ± 4.8% (p< 0.05) at 50 µM. The structural modulation of compound 2 was then carried out in order to understand whether it was possible to restore/increase the activity without the re-introduction of a Michael acceptor substructure. Table 1. Inhibitory effect of synthesised compounds on pyroptotic cell death, IL-1β release in differentiated THP-1 cells and cytotoxicity in THP-1 cells.
The last series of compounds encompasses derivatives 16-23 ( Figure 2). These compounds were deprived of the piperidine ring and the benzoimidazol-2-one structure was directly linked to the 2-chlorophenyl substructure through a propanoyl or an ethyl linker. In this series of compounds, the benzimidazol-2-one ring was substituted with an acidic group or with an ethyl ester, used as the carboxylic acid prodrug, at the terminal nitrogen. The biological results, reported in Table 1, showed that the 1-(3-(2-chlorophenyl)propanoyl)-1,3-dihydro-2H-benzo[d]imidazol-2-one scaffold could be the minimal requirement for NLRP3 inhibition (compound 16: 37.7 ± 7.6% pyroptosis reduction; 14.9 ± 8.8% IL-1β inhibition). When this scaffold was functionalised with an acetic acid residue (18) or a tetrazol-5-yl-methyl residue (19), two active compounds were obtained (Table 1). Interestingly, the conversion of 18 into the corresponding ethyl ester 17 maintained the activity as well as the removal of the carbonyl group in compound 23. When the carboxylic group was further spaced apart from the benzoimidazol-2-one ring, the activity was suppressed (derivative 21). It is difficult to rationalize the behaviour of compounds 20 and 22.
All the synthesised compounds did not show relevant toxicity in THP-1 cells (Table 1), with the partial exception of compound 1 (TC 50 32.9 ± 19 µM), possibly because of the electrophilic character of this derivative.
From the preliminary screening, compounds 6, 9, 13 and 18 were selected for further biological evaluation. The anti-pyroptotic effect of the selected compounds was determined in human macrophages using increasing concentrations (0.1-50 µM) of test compounds ( Figure 3). In these conditions, 6, 9, 13 and 18 showed a conc.-dependent inhibition of NLRP3-dependent pyroptosis with a maximal inhibition ranging between 40 and 60%. Interestingly, compounds 9, 13 and 18 maintained a significant ability to inhibit pyroptosis down to 1 µM (approx. 20-30% inhibition).

Inhibition of NLRP3 ATPase Activity
Selected compounds were further assessed for their ability to directly inhibit the ATP hydrolysis activity of the NLRP3 enzyme in vitro. The in vitro assays of NLRP3 ATPase activity were completed using 100 µM compound concentrations in order to provide the maximal inhibitory potential for the pre-incubation conditions used in the assay. We choose a rather high concentration for the inhibitors, expecting a low general effect if they were acting in a competitive manner toward substrate ATP. The collected results, expressed as % activity decrease with respect to vehicle-treatment, are presented in Figure 4. Compound 1, a conjugation of the acrylate INF39 and HS203873, displayed inhibitory potential toward the enzymatic activity of NLRP3 with a 28.4 ± 2.6% reduction in ATP hydrolysis relative to vehicle control. Likewise, derivative 2 was also able to suppress the ATPase activity by 28.6 ± 5.5%. Compounds 6 (INF148), 9 (INF120), 13 (INF156) and 18 (INF172) demonstrated similar inhibitory potential when used in the in vitro ATPase assay. In this case, the compounds were associated with 22.5 ± 1.4%, 18.3 ± 3.2%, 26.5 ± 3.1% and 27.2 ± 6.3% inhibition, respectively.  All compounds were added at concentrations of 100 µM and the ATP concentration in all assays was 800 µM. Mean ± SEM; Significantly different from vehicle control using unpaired Student's t-test with Welsch's correction. * p ≤ 0.05, ** p ≤ < 0.005, *** p ≤ 0.0005; n = 4 assays per condition.
Shortening the carbon-chain linker appeared to reduce (compound 4) or abolish (compound 3) the inhibitory potential when compared to 1 and 2. When the piperidine ring in compound 2 was replaced by a three-methylene chain (compound 10) no effect on ATPase inhibition was detected, while the use of a two-methylene chain (compound 9, INF120) restored the ATPase inhibition. Among the three cyanoguanidine-containing compounds 12, 13 (INF156) and 15, only compounds 13 and 15 showed a significant inhibition of ATPase activity. Finally, among the compounds belonging to series D (Figure 2), the ethyl ester derivative 17 was inactive while the corresponding acid 18 (INF172) was able to reduce ATPase activity. In this series of derivatives, both the lengthening of the chain bearing the COOH group (compound 21) or the replacement of the COOH with a tetrazol-5-yl (compound 19) reduced the inhibitory potential. This observation indicates that the presence of an acidic function in a correct spatial orientation might be important for the inhibition of the ATPase activity in this series of NLRP3 inhibitors.
The inhibitory potentials of selected compounds were also assessed at 1 mM (data not shown). The effective inhibition was found to be similar at both concentrations ( Figure 5). Specifically, no significant differences were identified between the two concentrations for compounds 6 (INF148), 9 (INF120), 13 (INF156) and 18 (INF172). A non-competitive inhibition vs. ATP, together with a low apparent Ki might be at the basis of this behaviour. Moreover, we completed a comparison of the impact of the different experimental methods (i.e., IL-1β maturation, ATPase inhibition and pyroptosis). Although not all compounds were examined by all experimental methods, those compounds demonstrating inhibitory potential in the ATPase assay were also associated with effective attenuation of IL-1β release ( Figure 5). However, compounds that suppressed pyroptosis were generally not well aligned with inhibitory effects on either IL-1β secretion or enzymatic ATPase activity. Figure 5. Effect of the experimental method on NLRP3 inflammasome attenuation. The impact of selected compounds on inflammasome outputs (i.e., pyroptosis, ATPase activity, IL-1β secretion) was assessed by two-way ANOVA with Tukey's multiple comparison test showing 95% CI. No difference between the observed effects suggests the INF compound administration was similar in outcome. Only those compounds which had been analysed by all three methods were included in the statistical analysis.

Molecular Modelling
We next investigated the potential binding mode of the model compounds (6, 9, 13 and 18) with NLRP3. The NLRP3 protein in complex with ADP and Mg 2+ ion was modelled on PDB entry 6NPY and submitted to extended (1.150 µs) plain molecular dynamics (MD). The Root Mean Square Deviation (RMSD) of the backbone atoms was calculated for checking the structural convergence of the protein ( Figure S1). As the structure was obtained by homology modelling (see Methods), the average RMSD along simulation time was quite high. According to the RMSD plot, the 750-1150 ns time frame showed a less dispersed profile, suggesting the achievement of a more stable conformation. To better investigate the structural convergence of the trajectory, the covariance matrix was calculated and diagonalised. Then, eigenvectors were sorted according to their eigenvalues. The resulting principal component analysis (PCA) plot highlighted a large movement along the first eigenvector during the first part of the trajectory, while the last part of the trajectory showed a large displacement along the second eigenvector ( Figure S2). The order of magnitude of the principal modes was quite high, reflecting NLRP3 intrinsic flexibility. In agreement with the RMSD analysis, the last part of the simulation (750-1150 ns) explored a less dispersed scattering in the essential space, and finally concentrated in an ellipsoid restricted area during the last 150 ns. Finally, the RMSD matrix was calculated, for assessing the difference among the conformations explored during the MD simulation. Consistently with the previous analyses, the matrix showed a fragmented pattern of minor conformational clusters until approximately 750 ns, while from 750-1150 ns, macro clusters were detected ( Figure S3). In accordance with the results shown by the RMSD plot, PCA and RMSD matrices, we retained the 750-1150 ns time frame of the MD simulation to perform the clustering with the Gromos method and a 0.4 nm cut-off. The cluster analysis highlighted five major clusters in the 750-1150 ns timeframe; the representative structure for each of these was extracted. Figure 6 shows the distribution of the five clusters along simulation time, highlighting that the first, second and third cluster well represent the trajectory. The medoid structure of each of these three clusters was extracted and used as target structure for docking studies. From here on, these medoids will be referred to as Med1, Med2 and Med3, i.e., the representative structure of the first, second and third most populated clusters, respectively. The quality of the NLRP3 model was assessed by calculating the Ramachandran plot for the initial structure and for the medoids extracted from the trajectory and used for subsequent docking studies ( Figures S4-S7). With respect to the initial model, MD structures showed a decreased number of outliers and an increased percentage of residues in allowed/favoured regions (Table S1). We thus analysed the three medoids looking for pockets close to the ADP binding site in Med1, Med2 and Med3. Pockets were calculated with FLAPsite to assess the extent of the cavity ( Figure S8). The volume of the ADP pocket greatly varies along the MD simulation, as the ADP molecule is bound between two subdomains that are known to undergo drastic reorientation upon protein activation [2]. In our MD simulation, the ADP pocket merged ( Figure S8, panel a) or split ( Figure S8, panels b and c) to other minor pockets, considerably changing its shape and extent. We extensively investigated the docking pose of the mentioned compounds in this variable and very large ADP pocket, which we separated into several sub-pockets identified by corresponding centroids having a radius of 8-10 Å (Table S2). Since docking studies in these pockets showed no consensus poses and low scores, we investigated all the pocketomes in NLRP3 to check whether other pockets in communication with the ADP site could better and more consistently accommodate the compounds. Pockets coupled to the ADP site along the MD trajectory (750-1150 ns timeframe) were detected with Pocketron (see Methods section), and among these, the ones showing less than 90% persistency were discarded. The retained cavities with 90% or more persistency, namely p1, p4, p8, p16, p25, p30 and p31, were further investigated in docking studies ( Figure 7). Figure 7. (a) Dynamic pocket network. Pockets with 90% or higher persistency are represented as spheres, with the sphere radius being proportional to pocket volume. Each pocket represents a node in the network, and dark blue edges represent the cross talk between pockets, with their thickness being proportional to the degree of correlation. Protein and ADP molecules are reported as cartoons and sticks, respectively. (b) Close-up of the NACHT domain. Pockets with 90% or higher persistency are labelled according to their pocket ID. ADP pocket (p21) is highlighted in yellow, p16 pocket is coloured in bright green. (c) Communication between the ADP pocket and p16. Persistency (% of frames) of residues belonging to the ADP pocket (p21, yellow circles) compared to those belonging to p16 (green squares).
The compounds were docked in these pockets, searching for consensus among the five top-ranked poses. The best results were achieved in pocket p16, which showed the highest dynamic correlation with the ADP pocket. The cross talk between the ADP pocket and surrounding pockets was monitored using the Pocketron tool implemented in the BiKi software suite as detailed in the Methods section. Overall, residues Met408, Thr439, Lys570, Glu520 and Ile521 are frequently exchanged between the ADP pocket and p16, highlighting a hydrophobic communication path between these two sites ( Figure 7c). Consequently, the binding of ligands at the p16 site could affect this cross talk and have an impact on the ADP binding site. Docking of 9 (INF120) and other model compounds 6, 13 and 18 in p16 showed a consensus in the overall binding poses (Figures S9-S11). A hydrophobic sub-pocket lined by residues Val404, Leu405, Met408 and Ile653 hosts the 2-chlorophenyl moiety of ligands. The amide linker is usually engaged in variable polar contacts with surrounding residues such as Thr439 and the backbone of Lys437 and Asn656. The environment interacting with the benzimidazole-2-one moiety is mainly composed by the basic side chains of residues Arg578, His663 and Lys437, which open the door to the modulation with hydrogen bond acceptor substituents as done in 18 (INF172). Additional hydrophobic interactions of the benzimidazol-2-one involve Leu681.
To better investigate how 9 binds to p16, induced fit docking studies were carried out, as described in the Methods section. The best outcome was related to docking in Med3, in which the LRR domain appears closer to the NACHT domain than in Med1 and Med2.
Compound 9 docking pose in Med3 ( Figure 8) was chosen for 100 ns MD in NLRP3, for checking whether the pose was stable inside the p16 pocket. The RMSD plot ( Figure S12) shows a slight rearrangement in the binding pose, which, however, was conserved throughout the MD simulation.

Chemistry
All the reactions were monitored by Thin Layer chromatography (TLC) on Merck 60 F254 (0.25 mm) plates, which were visualised by UV inspection (254 nm) and/or by spraying KMnO 4 (0.5 g in 100 mL 0.1 N NaOH). Na 2 SO 4 was used as drying agent for the organic phases. Flash chromatography (FC) purifications were performed using silica gel Merck with 60 mesh particles. Unless otherwise specified, all reagents were used as received without further purification. Dichloromethane was dried over P 2 O 5 and freshly distilled under nitrogen prior to use. DMF was stored over 3 Å molecular sieves. Anhydrous THF was freshly distilled under nitrogen from Na/benzophenone ketyl. 1 H and 13 C-NMR spectra were registered on JEOL ECZR600 spectrometer, at 600 and 151 MHz. Coupling constants (J) are given in Hertz (Hz) and chemical shifts (δ) are given in ppm, calibrated to solvent signal as internal standard. Following abbreviations are used to describe multiplicities: s= singlet, d = doublet, t = triplet, q = quadruplet, m = multiplet and br= broad signal. The following abbreviations are used to identify exact proton: ArH = Aromatic proton, BzImH= benzimidazolone ring, Pip = piperidine. ESImass spectra were recorded on a Waters Micromass Quattro Micro equipped with an ESI source. Melting points were measured with a capillary apparatus (Büchi 540). The purity of the final compounds was determined by RP-HPLC. Analyses were performed with a HP1100 chromatograph system (Agilent Technologies, Palo Alto, CA, USA) equipped with a quaternary pump (G1311A), a membrane degasser (G1379A) and a diode-array detector (DAD) (G1315B) integrated into the HP1100 system. Data analyses were processed using a HP ChemStation system (Agilent Technologies). The analytical column was a LiChrospher ® 100 C18-e (250 × 4.6 mm, 5 µm) (Merck KGaA, 64271 Darmstadt, Germany) eluted with CH 3 CN 0.1% TFA/H 2 O 0.1% TFA in a ratio that depended on the characteristics of the compound. All compounds were dissolved in the mobile phase at a concentration of about 0.01 mg/mL and injected through a 20 µL loop. HPLC retention times (t R ) were obtained at flow rates of either 1.0 or 1.2 mL/min and the column effluent was monitored using the DAD. The DAD acquired the UV spectra in the range from 190 to 800 nm, and the HPLC chromatogram was recorded at 226, 254, 580 and 660 nm (with 800 nm as the reference wavelength). The purity of the test samples was evaluated as the percentage ratio between the areas of the main peak and of possible impurities at the three wavelengths, and also using a DAD purity analysis of the chromatographic peak. The purity of all the target compounds was found to be ≥ 95%. (24) The reaction was conducted in nitrogen atmosphere. Sodium hydride (60%, 2.13 g, 21.9 eq) was added to a stirred solution of tert-butyl diethylphosphonoacetate (6.17 mL, 26.3 eq) in DMF (40 mL) at 0 • C. The reaction mixture was stirred 2 h at room temperature. p-chlorobenzilbromide (2.84 mL, 21.9 eq) was added dropwise at 0 • C, and the solution was stirred 2 h at room temperature. The reaction mixture was cooled to 0 • and water was added (20 mL). The solvent was reduced under reduced pressure. The residue was dissolved in diethyl ether and washed with water (2 × 10 mL), brine (15 mL), dried (Na 2 SO 4 ) and concentrated under reduced pressure to give 24 (8.20 g, 99.7%) as a white solid.

General Procedure for the Preparation of Compounds 32 and 33
To a stirred solution of 29a (1 mmol) in DMF (10 mL), the appropriate amine (1 mmol) and the appropriate base (2 mmol) were added at room temperature. When the reaction was complete (TLC), the mixture was concentrated under reduced pressure. Next, 10% Na 2 CO 3 solution was added (20 mL), and the product was extracted with EtOAc (3 × 25 mL). The combined organic phases were washed with brine, dried (Na 2 SO 4 ), filtered and concentrated under reduced pressure.

General Procedure for the Preparation of Compounds 36 and 6-8
To a stirred solution of the appropriate carboxylic acid (1 mmol) in DMF (2 mL), DI-PEA (2 mmol), HOBt (0.15 mmol), HBTU (1.5 mmol) and the appropriate amine (1.1 mmol) were added at room temperature and the mixture was stirred overnight. The solvent was evaporated under reduced pressure and a solution of NaHCO 3 10% (20 mL) was added. The mixture was extracted with EtOAc (3 × 20 mL). The combined organic phases were washed with brine (15 mL), dried (Na 2 SO 4 ) and concentrated under reduced pressure.   (7) (8)

General Procedure for the Preparation of Compounds 40 and 41
p-fluoronitrobenezene (1 mmol) was dissolved in a stirred solution of N-Boc alchildi amine (1.1 mmol) and a suspension of fine-powdered K 2 CO 3 (2 mmol) in dry DMF under nitrogen gas. The reaction mixture was kept at 70 • C overnight. The mixture was concentrated under reduced pressure. Brine (20 mL) was added and the product was extracted with EtOAc. The combined organic phases were washed with brine (15 mL), dried (Na 2 SO 4 ) and concentrated under reduced pressure.

General Procedure for the Preparation of Compounds 44 and 45
The appropriate p-diaminobenzene (1 mmol) derivatives were added to a stirred solution of CDI (1.1 mmol) in THF. The reaction was kept at room temperature overnight. The reaction mixture was concentrated under reduced pressure. The residue was dissolved in DCM and washed with water (2 × 15 mL), with brine (15 mL), dried (Na 2 SO 4 ) and concentrated under reduced pressure. (44) (46) Compound 44 (0.292 g, 1.05 mmol) was dissolved in a stirred solution of TFA/DCM (10%, 5.5 mL) at room temperature. After 4 h, the reaction mixture was concentrated under reduced pressure to give 46 (0.191

1-(3-(2-chlorophenyl)propanoyl)-1,3-dihydro-2H-benzo[d]imidazol-2-one
3.5. Molecular Modelling 3.5.1. MD Setup Uniprot sequence Q96P20 was modelled using the SWISS-MODEL webserver, using the Cryo-EM structure of human NLRP3 (PDB ID 6npy) as a template and retaining ADP [50]. The PYD domain, already missing in the Cryo-EM structure was not modelled, given the absence of structural indication about its position with respect to the other domains. The NEK7 protein, originally present in 6npy, was removed to reduce the structural dimension and allow more profitable MD simulation. NEK7 is fundamental for NLRP3 assembly and activation but, apparently, not for ATPase activity. Despite it being unnecessary, we thus kept the NEK7 while looking for a putative binding site for the tested compounds. For assessing the quality of the model, Ramachandran plots were generated using MolProbity website [51]. As mentioned, the ADP pose was retained as in the Cryo-EM structure 6npy. The Mg 2+ ion, known to be essential for NLRP3 activity was added and coordinated by the beta-phosphate group of ADP, as reported for other AAA+ proteins [52]. Parameters for ADP and Mg 2+ were retrieved from Meagher et al. [53] and Allnér et al. [54], respectively. Ionizable residues were assigned the default tautomeric state at pH = 7.4, while all histidines were treated as N ε -H (HIE), the only exception being His522, directly involved in a polar interaction with ADP and treated as N d -H (HID). The system was embedded in a cubic 130 × 130 × 130 Å box, filled with TIP3P water molecules and 11 Na + atoms to neutralize the protein charge. Periodic boundary conditions were set, the bond length between hydrogens and heavy atoms was kept at equilibrium distance with the LINCS algorithm. The Particle Mesh Ewald (PME) method was used for computing long-ranged interactions, and a cutoff value of 11 Å was set for both Van der Waals and electrostatic interactions. The protein converged to Fmax after 6284 steepest descent and 277 Polak-Ribière conjugate gradient minimization cycles. To gradually heat up the system from 0 to 300 K, six 1 ns thermalization steps were run with 1 fs time step in NVT ensemble using backbone restraints when heating from 0 to 200K. Just before the long MD production (1.150 µs), the system underwent a short equilibration (2 ns) in the NPT ensemble. All analyses on the trajectory were carried out using Gromacs 4.6.1, unless otherwise specified.
Essential dynamics. Principal modes of the protein were extracted by first fitting the trajectory to a reference structure (last equilibration step) to remove roto-translations. Then, the covariance matrix was calculated and diagonalized and eigenvectors were sorted out in descending order according to their eigenvalue. Principal component analysis (PCA) plot was obtained by plotting the trajectory in the orthogonal space described by the first vs. the second eigenvector.

Clusters
The 750-1150 ns timeframe of the trajectory was clustered on backbone atoms, using the GROMOS method, a cutoff = 0.4 nm and a 100 ps stride. Three medoids deriving from the cluster analyses well represented the equilibrated part of the trajectory and were further analysed in docking experiments (Figure 4).

Static Pocket Analysis
Pockets were calculated with FLAPSite [55] with standard parameters for each of the three extracted medoids, to better evaluate the immediate surroundings of the ADP site.

Dynamic Pocket Analysis
Pockets were calculated with the Pocketron, the pocket tracker tool implemented in BiKi Life Science v. 1.3.5 software suite (http://www.bikitech.com/) [56], along the equilibrated part of the trajectory (750-1150 ns, stride = 100, with a 3 Å probe radius). This tool is aimed at identifying pockets and monitoring the crosstalk between them, i.e., the exchange of atoms between adjacent pockets [57]. The persistency of the pockets was quantified, and only pockets with >90% or more persistency were extracted and evaluated in docking studies.

Induced Fit Docking
First, the target structures were prepared with the Protein Preparation Wizard in Maestro and protonated at a pH of 7.4 using PROPKA. The alpha carbon of Thr304 was chosen to define the target docking grid. INF ligands 6, 9, 13 and 18 were docked with Glide SP [60].

Ligand Parametrization
The top-ranked pose of 9 in Med3 was chosen for assessing the stability of ligands in pocket p16. Ligand 9 was parametrized using the ab initio RESP charge fitting method included in the BiKi Life Science v. 1.3.5 software suite. Ligand-NLRP3 binary complex was then submitted to the same minimization/equilibration protocol reported above, and a 100 ns MD production in the NVT ensemble was carried out for checking out the stability of the docking pose.

Conclusions
In our continuous effort to advance the discovery of new NLRP3 inhibitors, we employed a pharmacophore-hybridization strategy to identify new structural templates endowed with the ability to dampen NLRP3 function. The merging of the pharmacophore of INF39 and HS-203873 allowed the identification of compound 1 as a prototypical template. With the aim of obtaining new non-electrophilic derivatives able to target NLRP3, the chemical modulation of 1 was performed. A preliminary screening of the synthesised compounds in an established cellular model of NLRP3 activation led to the selection of a subset of active molecules. Compounds 6, 9, 13 and 18, proved able to moderately reduce NLRP3-dependent pyroptosis and IL-1β release in human macrophages. Interestingly, the selected compounds proved able to decrease the ATPase activity of immobilised NLRP3. A homology model of the inactive state of NLRP3 was built and molecular simulations and docking studies allowed for the identification of putative binding pockets for the selected compounds. A full pharmacological and mechanistic characterization of the selected compounds was out of the scope of this work, this representing an obvious limitation. However, thanks to the preliminary studies reported herein and to the model built with the use of the active scaffold, we are now working toward improved NLRP3 inhibitors.