Boron-Based Inhibitors of the NLRP3 Inflammasome

Abstract

NLRP3 is a receptor important for host responses to infection, yet is also known to contribute to devastating diseases such as Alzheimer's disease, diabetes, atherosclerosis, and others, making inhibitors for NLRP3 sought after. One of the inhibitors currently in use is 2-aminoethoxy diphenylborinate (2APB). Unfortunately, in addition to inhibiting NLRP3, 2APB also displays non-selective effects on cellular Ca²⁺ homeostasis. Here, we use 2APB as a chemical scaffold to build a series of inhibitors, the NBC series, which inhibit the NLRP3 inflammasome in vitro and in vivo without affecting Ca²⁺ homeostasis. The core chemical insight of this work is that the oxazaborine ring is a critical feature of the NBC series, and the main biological insight the use of NBC inhibitors led to was that NLRP3 inflammasome activation was independent of Ca²⁺. The NBC compounds represent useful tools to dissect NLRP3 function, and may lead to oxazaborine ring-containing therapeutics.

Keywords: NLRP3 inflammasome; boron; inflammation; inhibitor; interleukin-1.

Graphical abstract

Figure 1

Establishing the Importance of Boron in 2APB for NLRP3 Inflammasome Inhibition (A–G) Mouse peritoneal macrophages were primed with bacterial endotoxin (lipopolysaccharide [LPS], 1 μg mL⁻¹, 2 hr) and then stimulated with vehicle (0.5% DMSO) or 2APB (75 μM) before stimulation with ATP (5 mM, 20 min) (A), nigericin (20 μM, 15 min) (B), sphingosine (20 μM, 1 hr) (C), monosodium urate crystals (MSU; 250 μg mL⁻¹, 1 hr) (D), calcium pyrophosphate dehydrate crystals (CPPD; 250 μg mL⁻¹, 1 hr) (E), or aluminum hydroxide (Alum; 250 μg mL⁻¹, 1 hr) (F). The half-maximal inhibitory concentration (IC₅₀) for the effects of 2APB on IL-1β release induced by ATP was established using a 3-parameter logistical sigmoidal model (G). (H) Chemical structures of B-containing compounds 2APB (1) and DPBA (2) and C-containing 2APB analogs (3–12). (I and J) Mouse peritoneal macrophages were primed as before and stimulated with vehicle (0.5% DMSO) or inhibitor (1–5, 75 μM) before stimulation with ATP (5 mM, 20 min) (I). Mouse BMDMs were primed with LPS (1 μg mL⁻¹, 4 hr) and incubated with vehicle (0.5% DMSO) or molecules (NCI1–7, 6–12, 40 μM) for 15 min before ATP stimulation (5 mM, 1 hr) (J). In all cases supernatants were analyzed by ELISA. Data are presented as mean percentage of IL-1β release versus vehicle (DMSO) control + SEM (n = 3–9). *p < 0.05, **p < 0.01, ***p < 0.001, significant difference from 100% IL-1β release (Holm-Sidak corrected one-sample t test).

Figure 2

Identification of an Oxazaborine Ring in an Improved Pharmacophore Primary mouse BMDMs were primed with LPS (1 μg mL⁻¹, 4 hr) and incubated with vehicle (0.5% DMSO) or molecules (BC1–24, Figure S1) at 40 μM for 15 min before ATP stimulation (5 mM, 1 hr). The effects of the molecules on IL-1β release were measured by ELISA and normalized to ATP-induced IL-1β release in the absence of any inhibitor (A). The chemical structures (i) and half-maximal inhibitory concentration curves (IC₅₀, ii) for BC7 (B) and BC23 (C) are also presented using a 3-parameter logistical sigmoidal model. Data are presented as mean percentage of IL-1β release versus vehicle (DMSO) control + SEM of at least 3 experiments. *p < 0.05, ***p < 0.001, significant difference from 100% IL-1β release (Holm-Sidak corrected one-sample t test). ^###p < 0.001, significant improvement from 2APB treatment (Holm-Sidak corrected post hoc comparison).

Figure 3

Refinement of the Structure-Activity Relationship (A) Pathway for oxazaborine syntheses. The method for the synthesis of the oxazaborine compounds are described as types A, B, and C. (i) RCN, Zn(acac)₂/SnCl₄, dichloromethane/toluene, room temperature to 80°C, 3–16 hr; (ii) K₂CO_3(sat), EtOH, room temperature, 24 hr; (iii) DPBA, tetrahydrofuran (THF), 50°C, 16 hr; (iv) Cl₃CCN, NaOAc, EtOH, room temperature, 16 hr (A). (B–D) Table of oxazaborines synthesized with structure type (A, B or C) identified (B). Ad, adamantyl; Cy, cyclohexyl; Pyr, pyrene; Py, pyridinyl; EPPS, 4-(2-hydroxyethyl)piperazine-1-propanesulfonic acid. Also shown in (B) to (D) is the percentage of inhibition of IL-1β release from LPS and nigericin-treated THP-1 cells with 10 μM inhibitor and the calculated cLogP and cLogS values for each compound. (C) Pathway for dioxaborine syntheses. (iii) DPBA, THF, 50°C, 16 hr. ^aNBC9 was isolated as a by-product during NBC5 synthesis. (D) Pathway for diazaborine synthesis. (v) RNH₂, THF, 50°C, 24 hr. (E) Summary of SAR analysis of NBCs. (F) Half-maximal inhibitory concentration curve (IC₅₀) for NBC6 (Fi) is presented using a 3-parameter logistical sigmoidal model (n = 6) (Fii). *p < 0.05, **p < 0.01, ***p < 0.001, significant difference from 100% IL-1β release (Holm-Sidak corrected one-sample t test), n = 4. ns, not significant.

Figure 4

X-Ray Crystallography and Computational Modeling of NBCs (A and B) Crystal (X-ray) and predicted structure (Calc) of NBC6 (15) (A) and NBC11 (16) (B), calculated at the M06-L/6-31G* level of theory. The ring is boat-like in conformation (atom numbering shown in A). (C) Computed structure and Mulliken partial charges on ring atoms of oxazaborine NBC19 (17). (D) Computed structures of NBC27 (18) and NBC30 (19) illustrating the planarity of the diazaborine ring and Mulliken partial charges of NBC27 and NBC30. (E) Steric field arising from topomer CoMFA of 24 oxazaborine compounds, superimposed on structures of (left) NBC19 and (right) NBC20.

Figure 5

NBCs Are Effective NLRP3 Inflammasome Inhibitors (A) The effects of 2APB, BC7, BC23, and NBC6 on ASC speck formation following ATP stimulation were measured. iBMDMs stably expressing ASC protein conjugated to mCherry were primed with LPS (1 μg mL⁻¹, 2 hr), then pre-treated with selected drug (indicated concentration, 15 min) before stimulation with ATP (5 mM, 30–45 min) under live microscopy. Formation of ASC specks (examples indicated by white arrows, Ai [no drug], Aii [plus NBC6]) were quantified (Aiii) and presented as mean percentage of specks counted versus vehicle + SEM (n = 5–6). **p < 0.01, ***p < 0.001, significant difference from 100% speck formation (Holm-Sidak corrected one-sample t test, n = 5–6). Scale bars, 20 μm. (B) Recombinant caspase-1 (10 U mL⁻¹) was incubated with 0.5% DMSO, YVAD (100 μM), or 2APB (75 μM) before addition of the fluorogenic substrate Z-YVAD-AFC. Caspase-1 activity was measured 2 hr later (Bi) (***p < 0.001, significant difference from vehicle control, Holm-Sidak corrected post hoc comparison, n = 4). Hypotonic THP-1 cell lysate assay was also used to measure the effects of 2APB on caspase-1 activity. 2APB (75 μM) was added to the cells just prior to, or following, lysis in hypotonic buffer. The lysate was incubated with Z-YVAD-AFC and caspase-1 activity measured 2 hr later (Bii). YVAD or high K⁺ concentration were included as controls (Bii) (***p < 0.001, significant difference from relevant lysis vehicle control, Holm-Sidak corrected post hoc comparison, n = 4). (C) LPS-primed (1 μg mL⁻¹, 4 hr) mouse primary BMDMs were treated with NBC6 (10 μM) or vehicle (DMSO) 15 min prior to 1 hr treatment with small-molecule NLRP3 activator imiquimod (70 μM) or DMSO control. Imiquimod significantly induced IL-1β release (**p < 0.01) and this was inhibited by NBC6 treatment (^#p < 0.05, Holm-Sidak corrected post hoc comparison, n = 4). (D) Mouse primary BMDMs were primed with Pam3CSK4 (100 μg mL⁻¹, 4 hr) followed by 15 min NBC6 (1 μM), MCC950 (1 μM), or vehicle, then treated with intracellular LPS (2 μg mL⁻¹, transfected with Lipofectamine 3000, 24 hr) or Lipofectamine alone (**p < 0.01, significant induction of IL-1β [Di] or IL-1α [Dii] release versus Lipofectamine-alone control; ^##p < 0.01, significant inhibition of IL-1β release; Holm-Sidak corrected post hoc comparison, n = 4). (E) Mouse primary BMDMs were primed with LPS (1 μg mL⁻¹, 4 hr) followed by 15 min NBC6 (10 and 30 μM), MCC950 (30 μM), YVAD (100 μM), or vehicle, then treated with canonical NLRP3 activator ATP (5 mM, 1 hr), NLRC4 activator (flagellin, 667 ng mL⁻¹, transfected with Lipofectamine 3000), or AIM2 activator (poly(dA:dT), 667 ng mL⁻¹, transfected with Lipofectamine 3000) (*p < 0.05, **p < 0.01, ***p < 0.001, significant inhibition of IL-1β release, Holm-Sidak corrected post hoc comparison, n = 3). (F) Mouse primary bone marrow neutrophils from WT and NLRP3 KO mice (n = 4) were primed with LPS (1 μg mL⁻¹, 2 hr), then NBC6 (10 μM) was added 15 min prior to the addition of nigericin (10 μM), which significantly induced IL-1β release (***p < 0.001), which was inhibited by NBC6 treatment (^###p < 0.001, Holm-Sidak corrected post hoc comparison). Data are presented as the mean + SEM.

Figure 6

NBCs Are Effective against NLRP3 In Vivo (A) HEK293 or HepG2 cells were treated with NBC6 (10 μM), MCC950 (10 μM), or DMSO for 4 hr, 8 hr, and 24 hr. Cell death was measured by lactate dehydrogenase release and expressed as percentage lysis control. No significant effects were observed (two-way repeated-measures ANOVA). (B) LPS-primed (1 μg mL⁻¹, 2 hr) iBMDMs were pre-treated with drugs (BC23, NBC6, 30 μM; MNS, 100 μM; YVAD, 100 μM; 2APB, 75 μM) or vehicle (DMSO) in serum-free media for 15 min and washed 3 times, before inflammasome activation was initiated by adding ATP (5 mM) for 1 hr. IL-1β release was measured by ELISA (*p < 0.05, **p < 0.01, ***p < 0.001, significant inhibition of IL-1β release compared with vehicle-ATP control; ^#p < 0.05, ^###p < 0.001, significant effect of washing compared with no-wash drug-ATP control, Holm-Sidak corrected post hoc comparison, n = 5–6). (C) C57BL/6 and NLRP3 KO mice (n = 6) were injected intraperitoneally with LPS (10 mg kg^-1, 3 hr). Separate groups of WT animals receiving LPS were also given a 50 mg kg¹ dose of MCC950 or NBC13. IL-1β in peritoneal lavage (Ci) and plasma (Cii) was measured by ELISA. IL-1α in plasma was measured by ELISA (Ciii). ***p < 0.001, significant difference from saline vehicle control; ^##p < 0.01, ^###p < 0.001, significant difference from LPS vehicle group (Holm-Sidak corrected post hoc comparison). Data are presented as the mean + SEM.

Figure 7

Ca²⁺-Independent Effects of the NBCs (A) To induce volume-regulated Cl⁻ currents (VRAC), LPS-primed (1 μg mL⁻¹, 2 hr) iBMDMs were superfused with hypotonic solution. Representative current traces are shown, which have been measured in the absence (VRAC) or presence of 30 μM NBC6 (Ai), 30 μM BC7 (Aii), or 30 μM BC23 (Aiii). (B–E) LPS-primed (1 μg mL⁻¹, 2 hr) iBMDMs were kept untreated or were pre-treated for 2 min with 75 μM 2APB or 30 μM NBC6. Subsequently, 100 μM ATP was added to the bath solution. (B–D) Representative Ca²⁺ traces of ATP-stimulated cells in the absence (B, n = 12) or presence of 75 μM 2APB (C, n = 14), or 30 μM NBC6 (D, n = 12). (E) Mean peak Ca²⁺ concentrations determined in cells treated with ATP alone (ATP) or with ATP in the presence of inhibitors. ns, no significant difference; ***p < 0.001, significant difference from Ca²⁺ signals of ATP-stimulated cells determined in the absence of inhibitors (Holm-Sidak corrected one-sample t test). (F and G) LPS-primed iBMDMs were also stimulated with 5 mM ATP. Following development of sustained Ca²⁺ increases, 75 μM 2APB or 30 μM NBC6 was added to the ATP-containing solution. Images show representative examples of Ca²⁺ responses following treatment with ATP and the addition of 2APB (n = 12, F) or the addition of NBC6 (n = 12, G). (H and I) LPS-primed iBMDMs were treated with concentrations of the inhibitors maximal for blocking IL-1β release (i.e., 2APB = 75 μM, NBC6 = 30 μM). Inhibitors were added to the iBMDMs 5 min before or 5 min after the addition of ATP (5 mM, 1 hr) (H) with IL-1β release measured by ELISA (I). ***p < 0.001, significant difference from corresponding vehicle control; ns, no significant effect of ATP administration time. Data are presented as representative traces from calcium imaging experiments (B–D, F, and G), mean + SEM peak Ca²⁺ concentrations versus treatment with ATP alone (E), or mean + SEM IL-1β release as detected by ELISA (I).