cAbl kinase regulates inflammasome activation and pyroptosis via ASC phosphorylation

Abstract

Inflammasome activation is regulated in part by the post-translational modification of inflammasome proteins. Tyrosine phosphorylation is one possible modification. Having previously shown that the protein tyrosine kinase (PTK) inhibitor AG126 greatly inhibits inflammasome activation, we sought to uncover the target kinase. To do this we screened a commercial tyrosine kinase library for inhibition of inflammasome-dependent IL-18/IL-1β release and pyroptosis. THP-1 cells (human monocyte cell line) were incubated with PTK inhibitors (0.1, 1 and 10 μM) before stimulation with LPS followed by ATP. The PTK inhibitors DCC-2036 (Rebastinib) and GZD824, specific for Bcr-Abl kinase, showed the most severe reduction of IL-18 and LDH release at all concentrations used. The suggested kinase target, cAbl kinase, was then deleted in THP-1 cells by CRISPR/Cas9 editing and then tested for its role in inflammasome function and potential to phosphorylate the inflammasome adaptor ASC. The cABL KO not only significantly inhibited inflammasome function but also decreased release of phosphorylated ASC after LPS/ATP stimulation. One predicted target of cAbl kinase is tyrosine 146 in ASC. Complementation of ASC KO THP-1 cells with mutated Y146A ASC significantly abrogated inflammasome activation and ASC oligomerization as compared to wild type ASC complementation. Thus, these findings support cAbl kinase as a positive regulator of inflammasome activity and pyroptosis, likely via phosphorylation of ASC.

Introduction

The inflammasome is an innate immune system multi-protein complex that senses DAMPs (danger-associated molecular patterns) and PAMPs (pathogen-associated molecular patterns) and induces pyroptosis to release active IL-1β and IL-18 [1–3]. While there are multiple inflammasomes, they have in common the inclusion of apoptosis-associated speck-like protein containing a CARD (ASC) [1, 4]. ASC links the PYD-containing NOD-like receptor sensors (NLRs) to the CARD domain of caspase-1, 4 and 5 to form an inflammasome [1]. Via polymerization of ASC, the assemblage causes caspase-1 dimerization that induces its auto-activation [5, 6]. This activation is a fast-acting two-step post-translational process such that in monocytes, pre-loaded with inflammasome proteins, endogenous caspase-1/4/5 substrates are cleaved within minutes of stimulation [7, 8]. Indeed, mature IL-18, cleaved caspase-1 and GSDMD-N are detectable 30 minutes after stimulation [6, 8, 9].

The capacity for rapid inflammasome activation suggests that post-translational modifications such as phosphorylation might be required [10–12]. For example, it has been shown that NLRC4 phosphorylation at serine 533 activates its corresponding inflammasome [11]. The NLRP3 inflammasome activation is also complex. JNK1 phosphorylation of NLRP3 at serine 198 [13] and PKA phosphorylation at serine 295 [14] are components of this activation. The NLRP3 inflammasome can also be induced by ERK [9]. But at the same time, phosphorylation events can be inhibitory as shown for NLRP3 phosphorylation at tyrosine 918 by Lyn kinase [15] and for the pyrin inflammasome with phosphorylation of pyrin at serines 205 and 241 [16].

In the context of ASC, regulation by phosphorylation is well recognized but just as complex. For example, inflammasome assembly is promoted by Syk and JNK phosphorylation of ASC at tyrosine 144 in mouse (equivalent to human tyrosine 146) [17, 18]. Sequestration of ASC availability by the kinase IKKα has also been reported [19], but at the same time there is evidence that some ASC phosphorylation events at tyrosine 60 and 137 are inhibitory [20]. Thus, ASC phosphorylation presents challenges as a complexly regulated part of an already complexly regulated inflammasome.

Although the protein tyrosine kinase inhibitor, AG126, potently suppresses the inflammasome, it has not been clearly linked to specific tyrosine kinase targets [21]. The activation of the inflammasome by sodium orthovanadate, a nonspecific inhibitor of tyrosine phosphatases, further supports a regulatory role for tyrosine kinases but does not identify the specific kinase [22]. We therefore chose to use an unbiased approach by screening a commercial library of PTK inhibitors with better-characterized target specificity. Here, we identify cAbl kinase as a tyrosine kinase involved in the phosphorylation of Y146, and we expand on the significance of this tyrosine phosphorylation event and its role in pyroptosis.

Materials and Methods

Cell culture, bacterial strains, infections and transfections

In all experiments we used THP-1 cells (ATCC, Lot 385653, Manassas, VA) and associated Cas9 and CRISPR/Cas9 derivatives. Cells were incubated in RPMI 1640 media + GlutaMax (Gibco) supplemented with 10% heat-inactivated FBS (Atlas Biological, Fort Collins, CO) in the presence or absence of antibiotics. LPS and nigericin were purchased from InvivoGen (San Diego, CA) and ATP from Sigma-Aldrich. Francisella novicida U112 was grown on Chocolate II agar (BD Biosciences, Sparks, MD). Burkholderia cenocepacia (K56–2) and Salmonella enterica serovar Typhimurium (S. Typhimurium) were grown overnight in LB media. OD at 600 nm was used to calculate multiplicity of infection (MOI). F. novicida infection was used at MOI 50–100, B. cenocepacia at MOI 5–10 and S. Typhimurium at MOI 2.5. In general, cells were incubated with bacteria for 2.5 h and then 50 ug/ml of gentamicin (Gibco Life Technologies) was added and cells were incubated for 6 h or overnight. Samples were spun first at 300g to separate cells from bacteria, and then cell culture media was centrifuged at 16,000 g to clear bacteria. For transfection experiments we used either 1 ug poly(dA:dT) complexed with LyoVec (InvivoGen), 50 ng recombinant flagellin from S. Typhimurium (RecFLA-ST, InvivoGen) together with Profect solution (Targeting Systems), or 50 ng TcdB (Abcam). Transfection was initially performed in FBS-free media for 1 h, then cells were incubated overnight in 10% FBS.

Protein Tyrosine Kinase library screening

For the screening of the PTK library (Selleckchem, Houston, TX) we used THP-1 cells and a 96-well plate format. All inhibitors were dissolved in DMSO and used in screening at 3 concentrations: 0.1 μM, 1 μM and 10 μM. Cells were pretreated with inhibitors for 30 min and then stimulated with 1 μg/ml of LPS (InvivoGen, San Diego, CA) for 30 min followed by 5 mM of ATP (Sigma-Aldrich) for 30 minutes. In some experiments, to further evaluate selected inhibitors, cells were infected with 100 MOI of F. novicida or 10 MOI of B. cenocepacia for 6 h. Cell culture media was collected and centrifuged for 5 minutes at 300 g. Cell death was determined by LDH release (Roche Applied Science) and inflammasome-dependent IL-18 release by ELISA, as we described earlier [8]. DMSO pre-treated cells served as a control and the AG126 PTK inhibitor (InvivoGen, San Diego, CA) served as a positive control for inflammasome inhibition [21]. To evaluate the effect of PTK inhibitors, a heat map was generated with fold difference changes related to DMSO control (100% or 1).

Generation of CRISPR/Cas9 stable THP-1 cells

The pFUCas9-mCherry plasmid was used for Cas9 expression and the doxycycline hyclate-inducible pFgH1t_UTG plasmid expressing the EGFP marker (both gifts from Dr. Seth Masters, Walter and Eliza Hall Institute of Medical Research, Australia) [23] was used for CRISPR constructs. Small guide RNAs were designed using CHOPCHOP v2 [24] and were as follows: sgABL (plus strand) TGACTCCAAAACCCCTCCGG, sgABL (minus strand) TCAGTGATGATATAGAACGG, sgPYCARD (ASC) ACCGGGCTGCGCTTATCGCG. Sense and antisense guide RNA oligonucleotides were brought to 100 μM and annealed with NEB3 buffer (New England Biolabs Inc) using program: 95°C – 4 min, 70°C – 7 min, 65°C – 5 min, 60°C – 5 min, 55°C – 5 min, followed by a decrease to 20°C at a rate of 0.1°C/s. The pFgH1t_UTG plasmid was digested with the BsmBI restriction enzyme (New England Biolabs Inc), ran on a 1% agarose gel, cut out, eluted and used for ligation of annealed sgRNA oligonucleotides. Final plasmids were verified by Sanger sequencing. Lentivirus was produced in the HEK293FT packaging cell line (Invitrogen Life Sciences), transfected in the presence of Lipofectamine 2000 (Invitrogen Life Sciences) either with the pCas9-mCherry or pFgH1t_sgRNA_UTG plasmids and complemented with two helper plasmids, pCMVΔR8.2 and pMD.G, as we described earlier [25]. Lentiviral particles were collected at 48 and 72 h post-transfection, combined and centrifuged at 500g for 10 min to remove debris. The supernatant was filtered through a 0.45μm filter (Millex, Merck Millipore Ltd) and further concentrated with Amicon Ultra-15 centrifugal filters (Ultracell - 100 kDa cut off, Merck Millipore Ltd) at 3200g for 30 min, resulting in a 1 × 10⁷ TU/ml titer. To make stable cell lines, THP-1 cells were first spinoculated with Cas9-Cherry lentiparticles (800g for 90 min) using LentiBoost solutions (Sirion Biotech). Cells recovered for several days and were sorted with FACS ARIA III (BD Biosciences). After establishing a stable THP-Cas9-mCherry cell line, cells were spinoculated with FgH1t_sgRNA_UTG lentiparticles and after recovery double positive cells (mCherry/EGFP) were sorted with FACS ARIA III. Finally, sorted double positive cells were induced with 1 μg/ml of doxycycline hyclate (Sigma) for 72 h [23, 26]. Knockout proteins were verified by immunoblot and cells were also tested for the absence of mycoplasma contamination [27].

Knock-in (KI) Wild-type and Y146A mutant of ASC-a isoform into ASC KO THP-1 cells

To check the impact of Y146A ASC on inflammasome activation, the wild type and Y146A mutant of ASC-a isoform (22 kDa) were transduced into ASC KO THP-1 cells. The ASC-a isoform was cloned into the pLenti vector [28] using the EcoRI and XhoI restriction sites. PCR-based site-directed mutagenesis was used to replace tyrosine 146 to neutral alanine. Both plasmids were verified by Sanger sequencing and used in lentivirus production as we reported previously [22]. Because the CRISPR/Cas9 ASC KO THP-1 cells already express mCherry and EGFP markers, WT and Y146A ASC were made without a fluorescent tag. Therefore, KI insertion was done using several rounds of spinoculations of lentiviral particles in the presence of LentiBoost solutions A and B (Sirion Biotech). ASC expression levels were confirmed by immunoblot after every spinoculation. As usual, KI cells were confirmed by PCR to be mycoplasma negative. In addition, to be able to visualize ASC oligomerization in vivo, ASC and Y146A ASC mutant were inserted in pLenti-YFP plasmid using EcoRI/XhoI restriction sites generating YFP-ASC and YFP-Y146A-ASC fusion proteins [25].

ELISA, LDH assay, ASC crosslinking and immunoblots

Cell culture media was collected, centrifuged at 300g for 10 min and used for detection of LDH (Roche Applied Science), IL-18 (MBL antibodies D044–3 and D045–6, Woburn, MA) and IL-1β (clone 2805, R&D Systems used as coating antibody and rabbit polyclonal developed in our lab used as detection antibody) by ELISA. Cells were washed with PBS and lysed in either TN1 buffer (50mM Tris-Cl pH8.0, 125mM NaCl, 10mM EDTA, 10mM sodium fluoride, 10mM sodium pyrophosphate and 1% Triton X-100), CHAPS buffer (20mM HEPES-KOH pH7.2, 5mM MgCl_2, 0.5mM EGTA and 0.1% CHAPS) or a modified HEPES buffer (20 mM HEPES pH7.5, 150mM NaCl, 1mM NaF, 1% NP40 and 0.25% sodium deoxycholate). All lysis buffers were supplemented with 1 mM PMSF, 1:100 protease inhibitor cocktail and 100 μM neutrophil elastase inhibitor methoxy-succinyl-ala-ala-pro-val-chloromethyl ketone (Sigma Aldrich, St. Louis, MO). CHAPS buffer cell lysis was done as previously described [25]. TN1 and modified HEPES buffer cell lysis were performed similarly, but instead of syringe strokes, cells were lysed on ice for 30 min. ASC chemical crosslinking of cell culture media and cell extract was performed with 2 mM disuccinimidyl suberate (Pierce) as we described earlier in detail [29]. Protein concentration was determined with Bio-Rad DC Lowry protein assay (BioRad). After SDS-PAGE gel separation, samples were transferred to a polyvinylidene fluoride (PVDF) membrane, probed with the antibody of interest, and developed either by ECL (Pierce, Thermo Scientific) or by using LI-COR Biosciences System in conjunction with Odyssey® CLx Imaging System (LI-COR Biosciences, Lincoln, NE, USA). For blocking, we used either 10% or 5% milk in TBST solution and washing steps were done as previously described [29]. Rabbit polyclonal Abs against IL-1β and ASC were developed in our laboratory as described [30]. cAbl kinase was detected with polyclonal antibody #2862 from Cell Signaling, and ASC phosphorylation was detected with monoclonal 4G10 (Millipore Sigma) general anti-phosphotyrosine antibody and polyclonal Y144 (ECM Biosciences) antibody raised specifically against phosphorylated tyrosine at position 144/146 in ASC. Secondary antibodies were either donkey anti-rabbit polyclonal antibody or sheep anti-mouse monoclonal antibody (GE Healthcare), both conjugated to horseradish peroxidase (for ECL) as described previously [29]. For utilization of the Odyssey® CLx Imaging System, IRDye 800CW donkey anti-rabbit and IRDye 680RD goat anti-mouse were used (LI-COR). To re-blot, PVDF membranes were incubated in Re-Blot Plus Mild Solution (Millipore Sigma) for 12 minutes, then washed in TBST wash solution, and proceeded with blocking followed by antibody incubation as described earlier. All densitometry was done with ImageJ-win64 software. Negative control lanes were plotted and background area was subtracted from dimers and oligomers prior to comparison between 4G10 or Y-144 signal and ASC signal.

Statistical analysis

All experiments were performed independently three or more times (as indicated in Figure legends) and expressed as mean values ± SEM. Comparison of groups for statistical difference was done using Student’s t test and p value ≤ 0.05 was considered to be significant. The graphs were performed using GraphPad Prism 8.0 (GraphPad Software, San Diego, CA, USA).

Results

Abl kinase inhibitors efficiently prevent inflammasome activation

Although PTK inhibition blocks inflammasome activation [21], the specific kinases targeted by individual tyrophostins remain poorly characterized. In our efforts to identify specific PTK kinases that regulate inflammasome activity we screened a PTK inhibitor library for the ability to inhibit IL-18 processing and pyroptosis. THP-1 cells were pretreated or not with 3 concentrations of inhibitors using a short activation model (LPS (30 min)+ATP (30 min) (Supplementary Figure 1). This model studies inflammasome activation independent of new protein synthesis [9].

After several rounds of screening to eliminate inhibitors with weak or no effect, we selected for inhibitors that grouped around a specific PTK target and also showed matching ability to inhibit both cell death and inflammasome activation (Fig.1 A). To broaden the inflammasome targets beyond the LPS+ATP model, we incubated cells with selected inhibitors prior to incubation with Burkholderia cenocepacia and Francisella tularensis, bacteria known to efficiently activate the pyrin, NLRP3, NLRC4 and AIM2 inflammasomes [31–37] (Fig. 1B). Compared to our positive control prototype PTK inhibitor, AG126, the most consistent and efficient inhibition was observed with DCC2036 (Rebastinib) and GZD824. These inhibitors almost completely blocked LDH and IL-18 release following LPS+ATP, Burkholderia and Francisella challenge and both are recognized inhibitors of Bcr-Abl kinase (Fig.1A, B).

Figure 1. Heatmap of selected PTK inhibitors identified in high throughput screening.

Effect of inhibitors was measured by LDH (pyroptosis) and IL-18 release (inflammasome activation) following LPS/ATP or bacterial stimulation of THP-1 cells and expressed as fold differences relative to the LPS/ATP DMSO inhibitor control (defined as 1.0). Cells were pretreated with three different concentrations of PTK inhibitors 30 min prior to LPS stimulation for 30 min followed by ATP treatment for another 30 min (A), or infected with 100 MOI of F. novicida and 10 MOI of B. cenocepacia for 6 h (B). AG126 was used as a positive control reference inhibitor for LDH and IL-18 inhibition. Detailed readouts for all inhibitors from the library are shown in Supplementary Figure 1.

Inflammasome activation is decreased in ABL KO THP-1 cells

To further investigate the effect of cAbl kinase on inflammasome activation, we created two individual ABL KO THP-1 cell lines using CRISPR/Cas9 technology. As shown on Fig.2A, both ABL KO cell lines lost the cAbl kinase protein after doxycycline treatment. Confirmed ABL KO THP-1 cells were stimulated with LPS for 3h followed by ATP for 30 min (Fig.2B) or LPS alone overnight (Fig.2C), and IL-1β release was measured. To verify post-translational inflammasome activation, we measured LDH (Fig,2D) and IL-18 release after LPS stimulation for 30 min followed by ATP (Fig.2E) or nigericin (Fig.2F) stimulations for another 30 min. All models showed a significant decrease of inflammasome activation (p<0.001) for the ABL KO cell lines. To further validate whether this effect is common for all inflammasome platforms or observed for NLRP3 only, cells were infected with F. novicida, B. cenocepacia or S. Typhimurium (Supplementary Figure 2) or transfected with TcdB, recombinant flagellin or poly(dA:dT) (Supplementary Figure 3). There was no difference detected in LDH and IL-1β release between control and ABL KO cells after bacterial infection (Supplementary Figure 2). Transfection with pyrin, NLRC4 and AIM2 inflammasome ligands also did not show a difference between control and ABL KO cells. However, ASC KO and Y146A KI cells showed a significant decrease in IL-1β release after transfection. In addition, IL-1β release was significantly decreased in pyrin KO cells following pyrin inflammasome ligand TcdB transfection but not after flagellin or poly(dA:dT) transfections (Supplementary Figure 3).

Figure 2. Inflammasome activation is decreased in ABL KO THP-1 cells.

Two stable CRISPR/Cas9 KO THP-1 strains immunoblotted for cAbl kinase and actin proteins are shown (A). The effect of the cAbl kinase deletion on inflammasome activation after LPS/ATP stimulation (3h+30min) is shown in (B) and after LPS stimulation for 16h in (C). IL-1β release was used as a measure of inflammasome activation. The effect of cAbl depletion on LDH release after rapid inflammasome activation is shown in (D). IL-18 release was measured by ELISA after rapid inflammasome activation by LPS for 30 min followed by ATP for 30 min (E) or nigericin (F). n=7–10 independent experiments. *** p<0.001, ** p<0.01, * p<0.05 as compared to Cas9 control cells.

Y146 of ASC is critical for inflammasome activation

Inflammasome activation, at least its first step – priming, happens rapidly before new protein synthesis [9]. Since ASC is the universal inflammasome adaptor protein for most if not all inflammasome constructs, ASC is a logical candidate for post-translational modification and a potential target for tyrosine kinase phosphorylation. ASC contains 6 tyrosine residues, and Y146 has the highest predictive score to be phosphorylated (Supplementary Figure 4). To further evaluate the role of tyrosine 146 in inflammasome priming/activation, we created ASC KO THP-1 using CRISPR/Cas9 and then complemented the ASC KO cell line with either a wild type or Y146A mutant knock-in (KI) using lentiviral delivery (Fig.3A). To eliminate the possibility of differential activation of our CRISPR cells, all cells stimulated with LPS showed equal induction of the inflammasome substrate proIL-1β (Fig.3A). As expected, ASC KO THP-1 cells showed neither IL-1β release after LPS+ATP (3h+30 min) activation (Fig.3B) nor LPS overnight stimulation (Fig.3C), and no IL-18 release after LPS stimulation for 30 min followed by either ATP (Fig.3E) or nigericin (Fig.3F) stimuli, confirming ASC’s critical function in this model of inflammasome activation. Cell death was also significantly decreased in ASC KO and Y146A KI cells following 30 min LPS stimulation complemented with ATP or nigericin stimuli (Fig.3D). Re-introduction of WT ASC into ASC KO THP-1 cells restored inflammasome function in both short term and overnight schemes of activation (Fig.3B, C, E, F). In contrast, the knock-in of the Y146A ASC mutant failed to restore inflammasome function (Fig.3B, C, E, F).

Figure 3. Tyrosine 146 of ASC CARD is critical for IL-1β processing.

THP-1 cells were engineered using CRISPR/Cas9 technology to create ASC KO THP-1 cells. These KO cells were then transduced with a lentivirus knock-in (KI) of either wild type ASC isoform “a” (ASC KI) or the Y146A mutant of ASC “a” (ASC Y146A). The relative expression of ASC in these modified THP-1 cells is shown by ASC immunoblot and the expression of proIL-1β after priming with LPS by immunoblot for IL-1β (A). IL-1β release from THP-1 cells stimulated with LPS+ATP for 3h+30 min is shown in (B) and after LPS overnight in (C) was detected by ELISA. LDH and IL-18 release from cells stimulated with LPS for 30 min followed by ATP or nigericin for another 30 min shown in (D, E and F), respectively. n=5–7 independent experiments. *** p<0.001, ** p<0.01, * p<0.05 as compared to Cas9 control cells.

Release of phosphorylated ASC is decreased in ABL KO cells

To further explore the role of cAbl kinase on mechanisms of inflammasome activation, ASC oligomerization and phosphorylation were measured in wild-type, Cas9, ASC KO and ABL KO THP-1 cells. Phosphorylated ASC dimers and oligomers predominated in the cell culture media compared to the cell-associated fraction after LPS+ATP stimulation (Fig.4A, B). Although ASC oligomerization and release were not different between THP-1 and ABL KO cells, phosphorylation of ASC dimers and oligomers was reduced in ABL KO cells as shown by 4G10 anti-phosphotyrosine antibody using ECL (Fig.4A) or fluorescent secondary antibodies to detect overlapping signals (Fig.4B).

Figure 4. Phosphotyrosine signal is reduced in released ASC dimers and trimers from ABL KO THP-1 cells.

ASC oligomerization and phosphorylation were induced by LPS+ATP stimulation (30 min+30 min) of THP-1 and ABL KO cells. Oligomers from cell culture media were stabilized by DSS crosslinking and detected by polyclonal ASC antibody while phosphorylation was detected with anti-phosphotyrosine 4G10 antibody using ECL (A) or Li-COR (B) detection systems. After 4G10 antibody ECL detection, the membrane was stripped and re-blotted with polyclonal ASC antibody. With LI-COR detection system the monoclonal 4G10 and polyclonal ASC antibodies were probed together using two secondary antibodies, one anti-mouse conjugated with red tag (for 4G10) and the other anti-rabbit conjugated with green tag (for ASC). ASC monomer (22 kDa), and dimer (44 kDa) are shown with arrows and oligomers (66 kDa and up) denoted by line. Densitometry data are expressed as a ratio of intensity of P-tyrosine bands to the corresponding ASC bands. Two immunoblots are presented out of 4 experiments performed.

Since the 4G10 antibody is a nonspecific anti-phosphotyrosine antibody, the ASC specific Y144 antibody (which detects the ASC Y144 in mice and Y146 in human) was also used. Although similar levels of oligomerized ASC were detected in media upon LPS+ATP stimulation in both wild-type and ABL KO THP-1 cells, the specific Y144 antibody showed a decrease in phosphorylation of ASC dimers (Fig.5A) and trimers (Fig.5B) in the ABL KO cells. The lack of the Y144 antibody signal in ASC KO cells suggests that the observed phosphorylation signal is ASC specific. Interestingly, neither the 4G10 nor the Y144 antibody detected the ASC monomer.

Figure 5. ASC phosphorylation is reduced in released ASC dimers and trimers from ABL KO THP-1 cells.

ASC oligomerization and phosphorylation were compared between THP-1 cells, ABL KO THP-1 cells and ASC KO THP-1 cells by comparing the monoclonal anti-Y144 ASC antibody to the polyclonal ASC antibody after LPS+ATP stimulation (30 min+30 min) (A). Tyrosine phosphorylation of released ASC in the modified THP-1 cells after LPS+ATP (30 min+30 min) was imaged with Y144 antibody (B). After Y146 imaging, blots were stripped and re-blotted with the polyclonal ASC antibody. Arrows show ASC monomer (22 kDa), and dimer (44 kDa) and line shows oligomers (66 kDa and up). Densitometry data are expressed as a ratio of intensity of P-tyrosine bands to the corresponding ASC bands.

Wild type but not Y146A mutant knock-in ASC restores release and phosphorylation of ASC.

To further validate that Y146 of ASC is critical in inflammasome activation and also that this is a site of ASC phosphorylation, ASC phosphorylation was analyzed in the media of Cas9 cells, ASC KO, ASC KI and Y146A ASC KI cells stimulated with LPS for 30 min followed by ATP for 30 min. Both the monoclonal general anti-phosphotyrosine (Fig.6A) and polyclonal Y144 specific for tyrosine 146 of ASC (Fig.6B) showed restoration of phosphorylated ASC dimer and trimer from wild-type ASC KI cells but not from Y146A KI cells. Blotting these membranes for total ASC showed equal ASC release and oligomerization in both Cas9 and ASC KI cells. However, total ASC release and oligomerization was greatly reduced in the case of Y146A KI cells. To shed light on the absence of Y146A ASC oligomerization in the media, we lysed cell pellets and separated soluble versus insoluble fractions. As shown in Fig. 6C, Y146A ASC monomer was present in the soluble cell extract, similar to Cas9 and ASC KI THP-1 cells, however the Y146A mutant did not form oligomers and was not present in the insoluble fraction (Fig. 6D). Furthermore, YFP-fused wild type and Y146A mutant ASC presented distinct patterns of ASC oligomerization: well-structured round specks with YFP-ASC (Fig.6E) and filamentous structures with YFP-Y146A-ASC (Fig.6F).

Figure 6. ASC release, oligomerization and phosphorylation is absent in ASC KO cells and restored by ASC wild type but not Y146 mutant knock in.

Phosphorylated ASC in media was detected by 4G10 (A) and Y144 (B) antibodies after LPS+ATP stimulation of THP-1 cells for 30min+30min. After imaging, membranes were stripped and re-blotted with polyclonal ASC antibody. Arrows show ASC monomer (22 kDa), and dimer (44 kDa) and line shows oligomers (66 kDa and up). Cell extracts were also blotted for ASC to show relative levels of soluble ASC and actin in THP-1, WT ASC KI and Y146A ASC KI (C). The insoluble cell pellets of these cell extracts were analyzed for oligomerization as described (D). Images of YFP-ASC oligomerized specks (E) are compared to YFP-Y146A-ASC filaments (F). Scale bar is 125 μm. Experiments were repeated 4 times.

Discussion

Central to most if not all active inflammasome structures is the adaptor protein ASC. Inflammasome function depends upon the remarkable ability of ASC to form polymers that provide the skeleton for this process. ASC contains an amino terminal pyrin domain (PYD) that can form polymers via PYD/PYD domain interactions. These PYD-based polymers leave ASCs carboxy-terminal caspase-recruitment domains (CARD) exposed [38]. Thus, the analogous CARD domains of caspase-1/4/5 can pair up by binding to ASC dimers and oligomers through their analogous CARDs. This caspase dimerization along the ASC backbone triggers the proximity-based, self-activation of caspase-1/4/5 [5, 6, 38]. Thus, the events that trigger ASC polymerization and its binding to the CARDs of caspase-1/4/5 are likely targets for inflammasome regulation. However, the exact mechanisms that regulate ASC polymerization and the binding of caspase-1 to ASC are incompletely characterized.

The present work seeks to uncover post-translational regulatory events that may control the ability of ASC to self-aggregate and provide a platform for caspase-1 activation. Since tyrosine kinases are essential for inflammasome activation [18], we chose to use an unbiased screen of a tyrosine kinase library to identify specific kinase targets. In doing so we identified two distinct inhibitors of Bcr-Abl kinase (DCC-2036 (Rebastinib) and GZD824) that suppressed inflammasome activation as evidenced by their prevention of both pyroptotic cell death and the release of IL-1β and IL-18. Using this lead, we confirmed that cAbl kinase plays a role in this process as evidenced by the fact that cAbl kinase deletion by CRISPR/Cas9 in THP-1 cells suppresses caspase-1 activation. Finally, having noted that the CARD domain of human ASC contains a highly conserved tyrosine domain at Y146, which is predicted to undergo tyrosine phosphorylation, we turned our attention to ASC. Using anti-phosphotyrosine antibodies we confirm that ASC is phosphorylated at a tyrosine residue during inflammasome activation. More importantly, mutating Y146 to a neutral alanine completely disrupts caspase-1 activation events. Thus, this data suggests that a phosphorylation event in the CARD of ASC may be a critical step to promote the final activation event of the inflammasome, the binding of caspase-1’s CARD to the phosphorylated CARD of ASC.

There remain many questions unanswered by this work. Although the data strongly supports that cAbl kinase is involved in the phosphorylation of ASC, the ABL KO did not completely suppress inflammasome activation events after LPS+ATP stimulation and did not suppress inflammasome activation following cell infection with bacteria or transfection with pyrin, NLRC4 or AIM2 inflammasome ligands. Perhaps other kinases may be required to completely phosphorylate ASC at Y146. For example, Pyk2 phosphorylates ASC at Y146 [18] as well as does Syk [17]. In fact, our kinase inhibitor screen (Supplementary Figure 1) did show partial inflammasome inhibition for a number of kinases including Syk, BTK, IGFR and EGFR. One explanation for these findings is that many inhibitors are promiscuous with overlapping target specificity. However, as shown for the knockout of Syk [39] and in the current work by a knockout of cAbl kinase, each target only partially suppresses the inflammasome. Perhaps the redundancy of the signaling networks provides a resiliency to the host defense system such that pathogens need to simultaneously subvert multiple components of the network in order to be completely successful [40]. Alternatively, the redundancy may provide a mechanism to tightly modulate the intensity of the immune response as needed.

Questions also still remain about the specific role that ASC phosphorylation plays in triggering the inflammasome. In this regard, it is noteworthy that our studies only detected ASC phosphorylation in polymeric forms of ASC. Thus, one might hypothesize that ASC polymerization itself requires phosphorylation of ASC as we detected no phosphotyrosine in monomeric ASC. However, we have noted (data not shown) that highly concentrated recombinant ASC spontaneously forms polymers in a cell-free, kinase-free system, suggesting that ASC polymerization is kinase independent. Thus, if polymer formation is not controlled by phosphorylation, it is logical to predict that phosphorylation of the ASC CARD at Y146 might promote capture of the caspase-1 CARD to the polymerizing ASC. If true, Y146 phosphorylation may represent the final step in the proximity-based auto-activation of caspase-1. Further work in this area may provide new insights.

In summary, ASC phosphorylation is critical to inflammasome function and at least partly controlled by tyrosine kinases, which we have now shown includes cAbl kinase. Thus, inhibitors directed at cAbl kinase may provide a novel pathway for the development of therapeutics to modify pyroptosis and cytokine release in inflammatory disorders.

Key Points.

• Tyrosine kinase inhibitors that target cAbl kinase inhibit inflammasome activation.

• cAbl kinase deletion suppresses the ability to activate the NLRP3 inflammasome.

• Tyrosine 146 of ASC is integral to speck formation and inflammasome function.