PtdIns4P on dispersed trans-Golgi network mediates NLRP3 inflammasome activation

Abstract

The NLRP3 inflammasome, which has been linked to human inflammatory diseases, is activated by diverse stimuli. How these stimuli activate NLRP3 is unknown. Here we show that different NLRP3 stimuli lead to disassembly of the trans-Golgi network (TGN). NLRP3 is recruited to the dispersed TGN (dTGN) through ionic bonding between its conserved polybasic region and negatively charged phosphatidylinositol-4-phosphate (PtdIns4P) on the dTGN. The dTGN then serves as a scaffold for NLRP3 aggregation into multiple puncta, leading to polymerization of the adaptor protein ASC, thereby activating the downstream signalling cascade. Disruption of the interaction between NLRP3 and PtdIns4P on the dTGN blocked NLRP3 aggregation and downstream signalling. These results indicate that recruitment of NLRP3 to dTGN is an early and common cellular event that leads to NLRP3 aggregation and activation in response to diverse stimuli.

Extended Data Figure 1.. NLRP3 forms multiple puncta to activate the inflammasome pathway.

a, Endogenous NLRP3, ASC and caspase-1 are not detectable in HEK293T. Extracts from HEK293T cell lines stably expressing the indicated proteins were examined by immunoblotting. *, a nonspecific band. b, Reconstitution of NLRP3 inflammasome pathway in HEK293T. Cells stably expressing murine NLRP3, ASC and caspase-1 were treated with nigericin (Nig) (10 μM) for 60 minutes followed by immunoblotting. c, Highly purified NLRP3 showed signal-dependent activity in the in vitro assay. NLRP3-GFP was purified through fractionation and immunoprecipitation as detailed in Methods, before the purity and activity were examined by silver staining (left panel) and the in vitro assay (right panel) respectively. d, The in vitro assay detects activation of endogenous NLRP3. RAW 264.7 was treated with LPS (50 ng/mL) for 3 hours and stimulated with nigericin (10 μM) for 60 minutes before cell extracts were collected for the in vitro NLRP3 activity assay. e, Human NLRP3 also formed various puncta in response to nigericin. HeLa cells stably expressing human Flag-NLRP3 were treated with nigericin (10 μM) for 80 minutes before immunostaining with a Flag antibody. f, NLRP3 formed multiple puncta in the absence of ASC but a large speck in the presence of ASC. Cells stably expressing the indicated proteins were treated as in (e) before imaging. g, NLRP3 puncta possessed high activity. Left panel: nigericin (10 μM, 60 minutes)-induced NLRP3 puncta remained in the cells after saponin treatment. Nu, Nucleus. Right panel: cell extracts with or without saponin treatment were examined by the in vitro assay. The p10 level in Lane 4 is approximately 6.5 fold of that in Lane 6 based on quantification in imageJ (normalization by NLRP3 bands intensity). Only Activator cell extracts were used for tubulin immunoblot. h, Constitutively active mutants of NLRP3 formed puncta without stimulation. Cells stably expressing the indicated proteins were imaged in the absence of stimulation.

Extended Data Figure 2.. NLRP3 aggregates on stimulus-triggered dTGN.

a, Nigericin treatment induced giant vesicles to form in the perinuclear region. HeLa NLRP3-GFP cells treated with or without nigericin (10 μM) for 80 min were examined under phase contrast microscope. Mag: magnification. Percentage of cells with giant vesicle formation was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable. b, Ultrastructural analysis of nigericin-induced dispersed TGN (dTGN) vesicles. HeLa cells treated as in (a) were examined by transmission electron microscopy. Nu, nucleus. Blue arrowheads indicate Golgi stacks under resting conditions, while red arrowheads indicate nigericin-induced dTGN vesicles. Representative images from two independent experiments (more than 30 cells were examined for each condition in each experiment) are shown. c, Nigericin triggered the formation of dTGN, on which NLRP3 aggregated. HeLa cells stably expressing the indicated protein were stimulated as in (a) before immunostained for TGN38 and GOLGA4, two TGN markers. d, e, Cis- and medial-Golgi remained intact after nigericin treatment. Cells treated as in (a) were immunostained for GM130 (cis-Golgi maker) or giantin (cis-/medial-Golgi marker). f, NLRP3 aggregates mostly on dispersed TGN38-positive EEA1 but also on some EEA1-positive vesicles. HeLa cells stably expressing NLRP3-GFP and EEA1-HA were treated as in (a) before immunostained for TGN38 and HA (EEA1-HA). g, ATP stimulation led to NLRP3 aggregation on dTGN. HeLa cells stably expressing NLRP3-GFP and P2X7-HA were treated −/+ ATP (5 mM) for 80 min before imaging. P2X7 is a purinergic receptor essential for ATP-mediated NLRP3 inflammasome activation. h, Gramicidin stimulation led to NLRP3 aggregation on dTGN. HeLa cells were treated −/+ gramicidin (5 μM) for 80 min before imaging. i, DNA stimulation does not cause TGN dispersion or AIM2 recruitment to TGN. HeLa cells stably expressing AIM2-Flag were mock-transfected, transfected with poly(dA:dT) (1.5 μg/mL) for 3 hours or incubated with nigericin (10 μM) for 80 min before immunostained with antibodies against Flag (AIM2-Flag) or TGN38.

Extended Data Figure 3.. Endogenous NLRP3 is recruited to dTGN in primary macrophages.

a, Dramatic TGN disassembly occurred at early time points in WT BMDMs. Cells were primed with LPS (50 ng/mL) for 3 hours, followed by nigericin (10 μM) or ATP (5 mM) stimulation for the indicated time and immunostained for TGN38. To quantify the level of TGN disassembly, the numbers of TGN structures not connected with each other for each cell were quantified from 100 randomly selected cells and grouped as shown in the right panels. b, c, Nigericin-induced caspase-1 and IL-1β cleavage didn’t occur until 15 min (for nigericin) or 10 min (for ATP) post stimulation in WT BMDMs. Cells were treated as in (a) before lysates were collected for immunoblotting. d, Endogenous NLRP3 aggregation on dTGN could be detected as early as 10 min post nigericin treatment in ASC-deficient BMDMs. Cells were primed with LPS (50 ng/mL) for 3 hours, followed by nigericin (10 μM) treatment for 0, 10 or 30 minutes before imaging. Percentage of cells with NLRP3 puncta on dTGN was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable.

Extended Data Figure 4.. NLRP3 activity is strongly associated with dTGN but not mitochondria.

a, NLRP3 did not translocate to mitochondria upon stimulation in HeLa cells. HeLa NLRP3-GFP cells were stimulated with nigericin (10 μM) or gramicidin (5 μM) for 80 min before immunostained for TOM20 (mitochondrial marker). b, Neither nigericin- nor ATP-induced NLRP3 puncta were colocalized with mitochondria in ASC-deficient BMDMs. Cells were primed with LPS (50 ng/mL) for 3 hours, followed by nigericin (10 μM) or ATP (5 mM) treatment for 60 min before immunostained for endogenous NLRP3 and TOM20. c, NLRP3 activity in P100 (light membrane) fraction was strongly associated with dTGN but not mitochondria. P100 fraction collected from Fig. 1c was subjected to sucrose gradient ultracentrifugation, before fractions were collected and examined by the in vitro NLRP3 activity assay (top panel). TOM20 (mitochondrial marker) and GM130 (cis-Golgi marker) were not detectable on immunoblots even after prolonged exposure. d, dTGN-localized NLRP3 puncta can initiate aggregation of ASC^PYD. HeLa cells stably expressing the indicated proteins were incubated −/+ nigericin (10 μM) for 80 min before imaging. Mag, magnification. ASC^PYD: aa 1–90 of murine ASC. Right panel (from top to bottom): reconstituted Z-stack image of a representative nigericin-treated cell; nigericin-treated cell co-immunostained for TGN38 (pseudo-colored in magenta); percentage of cells with ASC^PYD filaments originating from dTGN-localized NLRP3 puncta was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable.

Extended Data Figure 5.. The polybasic region mediates the dTGN recruitment and activation of NLRP3.

a, NLRP3 is not required for dTGN formation in response to nigericin in HeLa. HeLa cells (without NLRP3 expression) were treated −/+ nigericin (10 μM) for 80 min and immunostained for TGN38. Cell borders are outlined with dashed lines. b, ATP induced dTGN formation in a manner dependent on P2X7 but not NLRP3. HeLa cells with or without P2X7-HA stable expression were incubated −/+ ATP (5 mM) for 80 min before immunostained for TGN38. c, Constitutively active mutants of NLRP3 can bypass the step of TGN recruitment by spontaneously forming aggregates in the cytosol. Cells stably expressing the indicated proteins were treated −/+ nigericin (10 μM) for 80 min before immunostained for TGN38. d, Immunoblots for cells used in Fig. 3c. e, The KKKK motif is critical for nigericin-induced NLRP3 activation. Cells stably expressing the indicated proteins were treated −/+ nigericin (10 μM) for 60 min before examined with the in vitro NLRP3 activation assay. f, Mutations in the KKKK motif do not compromise the ability of constitutively active NLRP3 to activate caspase-1. Cells stably expressing the indicated proteins were examined by fluorescence microscopy (upper panel) and the in vitro NLRP3 activity assay (lower panel) without any stimulation. LP, L351P; 4KA, K127/128/129/130A.

Extended Data Figure 6.. The polybasic region of NLRP3 functions through its positive charge.

a, Mutations of the KKKK motif to arginine do not affect nigericin-induced NLRP3 puncta formation. HeLa cells stably expressing the indicated proteins were treated −/+ nigericin (10 μM) for 80 min before imaging. The percentage of cells with NLRP3 puncta was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable. n.s., not significant (alpha = 0.01). 2KR, K127/128R; 3KR, K127/128/129R; 4KR, K127/128/129/130R. b, Mutations of the KKKK motif to arginine do not impair nigericin-induced NLRP3 activation. Cells stably expressing the indicated proteins were treated −/+ nigericin (10 μM) for 60 min before cell extracts were examined by the in vitro NLRP3 activity assay. c, The positive charge of the KKKK motif is important for gramicidin- and ATP-induced NLRP3 puncta formation. HeLa cells stably expressing the indicated proteins were treated −/+ gramicidin (5 μM) (left panel) or ATP (5 mM) (right panel) for 80 min, before the percentage of cells with NLRP3 puncta was analyzed as in (a). d, The positive charge of the KKKK motif is essential for NLRP3 to polymerize ASC. Cells stably expressing the indicated proteins were treated −/+ nigericin (10 μM) or gramicidin (5 μM) for 80 min before immunostained for both NLRP3 and ASC. The percentage of cells with NLRP3-ASC speck was analyzed as in (a). e, Immunoblots for cells used in Fig. 3d. f, g, The second positively-charged region of NLRP3 is also important for its aggregation on dTGN and activation in a manner dependent on the number of remaining positively-charged residues. Cells stably expressing the indicated proteins were treated −/+ nigericin (10 μM) before imaging (f) or examined by the in vitro NLRP3 activity assay (g).

Extended Data Figure 7.. NLRP3 is recruited to dTGN via binding to PI4P.

a, The polybasic region of NLRP3 interacted with phospholipids in vitro through its positive charge. Left panel: Coomassie Blue staining of purified proteins of interest, with Flag-GFP as a control. Right panel: PIP Strip membranes blotted with various lipids were incubated sequentially with proteins of interest, Flag antibody and HRP secondary antibody before exposure. Phospholipids with positive binding on the second PIP Strip are highlighted in red. b, Catalytically inactive Sac1 did not impair the recruitment of NLRP3 to dTGN. COS-7 cells stably expressing Flag-NLRP3 were transiently transfected with TGN38-FRB and mRFP-FKBP12-Sac1(C389S), incubated with rapamycin (1 μM) and nigericin (10 μM) for 80 min before imaging. Note that nigericin-induced dTGN in COS-7 cells sometimes look like a single cluster because of the relatively small size of COS-7 cells and the high intensity of fluorescence signal. Under phase contrast microscope these dTGN were separated vesicles adjacent to each other (data not shown). c, Only PI4P phosphatase blocked NLRP3 recruitment to dTGN. Similar to (b), except that indicated phosphatases were used. The target phospholipids are labeled after “x”. d, Immunoblots of HeLa NLRP3-GFP cells stably expressing TGN38-Sac1 (WT or C389S). CS, C389S. e, TGN-targeted Sac1 didn’t affect general cell morphology or nigericin-induced dTGN formation. Cells were treated −/+ nigericin (10 μM) for 80 min before imaged under phase contrast microscope. The percentage of cells with dTGN formation was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable. n.s., not significant (alpha = 0.01). f, TGN-targeted Sac1 didn’t affect AIM2 aggregation. Cells stably expressing the indicated proteins were transfected with or without poly(dA:dT) (1.5 μg/mL) for 3 hours before imaging. The percentage of cells with AIM2 aggregates was analyzed as in (e). g, NLRP3 puncta were restricted to PI4P-enriched microdomains. Cells stably expressing the indicated proteins were treated −/+ nigericin (10 μM) for 80 min before imaging. Inset: higher magnification of dTGN. Colocalization analysis was performed by calculating Pearson’s correlation coefficient (threshold regression: Costes) using Coloc 2 plugin of ImageJ. Data are presented as mean ± SD.

Extended Data Figure 8.. Binding to PI4P on dTGN is essential for NLRP3 inflammasome activation.

a, b, The KKKK motif of NLRP3 can be functionally replaced with a PI4P-binding domain of OSBP. (a) Cells stably expressing the indicated proteins were treated −/+ nigericin (10 μM) for 80 min before imaging. The percentage of cells with NLRP3 on dTGN was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable. 2RE, R107/108E. Note that NLRP3(ΔKKKK OSBP^PH) constitutively localized on intact TGN without stimulation was not counted in the quantification. (b) Replacement of the KKKK motif with OSBP^PH domain allowed NLRP3 to be constitutively localized on TGN. Cells were treated as in (a) before immunostained for TGN38. c, Mutations of OSBP PH domain that abrogates its PI4P binding also abolish its ability to functionally rescue NLRP3(ΔKKKK). Cells were treated as in (a) before extracts were examined by the in vitro NLRP3 activity assay. d, Immunoblots for primary NLRP3-deficient BMDMs rescued with Flag-NLRP3 (WT or mutants), which were used for experiment in Fig. 5b and 6d. The cells were infected with lentivirus encoding the indicated proteins for 6 days before immunoblotting. e, Recruitment of NLRP3 to non-PI4P-enriched regions of TGN is not sufficient to support its activation. Cells stably expressing the indicated proteins were treated as in (a) before examined by fluorescence microscopy (left panel; images for NLRP3(ΔKKKK OSBP^PH) are shown in (a)) and the in vitro assay (right panel). f, Extracellular KCl at 30 mM was sufficient to completely block nigericin-induced K⁺ efflux. Cells were treated −/+ nigericin (10 μM) for 80 min in the presence of increasing concentrations of KCl, before cell extracts were collected for measurement of intracellular K⁺ concentration (shown as percentage change compared to untreated sample, mean ± SD). Representative results from two independent experiments (each contains two samples for each condition) are shown. g, Incubation in K⁺-free medium induced spontaneous K⁺ efflux. Cells were incubated in Hanks buffer containing 5 mM K⁺ for the first two conditions, or Hanks buffer without K⁺ (replaced by Na⁺) for the third condition (K⁺-free). The second condition also contained nigericin (10 μM). After 80 min, the cell extracts were collected for intracellular K⁺ measurement with methods similar to (f). h, K⁺ efflux alone is not sufficient to activate either WT NLRP3 or NLRP3(ΔKKKK OSBP^PH). Cells stably expressing the indicated proteins were treated as in (g) before examined by the in vitro NLRP3 activity assay. i, Incubation in K⁺-free medium induced spontaneous K⁺ efflux in primary WT BMDMs. Cells were primed with LPS (50 ng/mL) for 3 hours, before treatment as in (g) with the indicated time lengths. Intracellular K⁺ concentrations were then measured and analyzed as in (f). j, K⁺ efflux alone is not sufficient to activate endogenous NLRP3 in primary WT BMDMs. Cells treated as in (i) were used for immunoblotting. lys, lysate; sup, supernatant.

Extended Data Figure 9.. K⁺ efflux-independent stimuli also induced TGN dispersion and PI4P-dependent NLRP3 recruitment.

a, b, K⁺ efflux-independent stimuli also induced NLRP3 aggregation on dTGN through the KKKK motif. HeLa cells stably expressing the indicated proteins were treated −/+ imiquimod or CL097 (45 μg/mL) for 80 min before imaging. High magnification images are shown in the inset. Arrows indicate representative plasma-membrane-localized NLRP3 puncta, which were separated from TGN38-positive compartments due to the partial separation of PI4P and TGN38. The percentage of cells with NLRP3 puncta on dTGN was quantified from 100 cells (n = 3, mean ± SD; two-sided t test). N.D., not detectable. c, Neither imiquimod- nor CL097-induced NLRP3 puncta were colocalized with mitochondria in ASC-deficient BMDMs. Cells were primed with LPS (50 ng/mL) for 3 hours and incubated −/+ imiquimod or CL097 (45 μg/mL) for 60 min, before immunostained for endogenous NLRP3 and TOM20 (mitochondrial marker). d, High extracellular KCl had no significant effect on imiquimod- or CL097-induced NLRP3 aggregation on dTGN. HeLa NLRP3-GFP cells were treated −/+ imiquimod or CL097 (45 μg/mL) in the presence of KCl (0 or 30 mM) for 80 min before imaging. Results were analyzed as in (b). n.s., not significant (alpha = 0.01).

Extended Data Figure 10.. Model: NLRP3 aggregation on dTGN through PI4P binding is a common cellular signal essential for the inflammasome activation by diverse stimuli.

Central panel: under basal conditions, NLRP3 is diffused in the cytosol while TGN (shown in red on the Golgi stack) remains as a single cluster attached to medial- and cis-Golgi (colored in purple). Left panel: when cells are stimulated with K⁺ efflux-dependent stimuli such as nigericin and ATP, TGN is disassembled into multiple dispersed structures (dTGN), whereas cis- and medial- Golgi stacks still remain intact. These stimuli also trigger K⁺ efflux, which helps recruit NLRP3 to dTGN via ionic bonding between the negatively-charged PI4P on dTGN membranes and the positively-charged polybasic region of NLRP3. Right panel: when cells are stimulated with K⁺ efflux-independent stimuli such as imiquimod and CL097, TGN is also disassembled, although in a more dramatic way. PI4P is partially separated from other TGN compartments and some can be enriched in the plasma membrane. NLRP3 is then recruited to these PI4P-containing microdomains through its polybasic region. For both types of stimulation, dTGN serves as a scaffold for NLRP3 to aggregate in the form of multiple puncta, which then interact with ASC to activate the downstream signaling cascade. This model shows that aggregation of NLRP3 on dTGN through PI4P binding is a common cellular signal essential for its activation by diverse stimuli.

Figure 1.. NLRP3 forms multiple puncta after stimulation.

a, b, In vitro assay for examining NLRP3 activity. 293 NLRP3 cells were stimulated with either nigericin (Nig) (10 μM) or gramicidin (Gra) (5 μM) for 60 minutes, before cell extracts were collected and mixed with PFO-permeabilized 293 ASC + casp1 cells. After incubation, the reaction mix was analyzed by immunoblotting. c, NLRP3 activity resided in light membrane (P100) and to a less extent heavy membrane (P5), but not in cytosol (S100). The fractions served as “Activator” as in (a) in the NLRP3 activity assay. d, HeLa cells stably expressing NLRP3-GFP were stimulated −/+ nigericin (10 μM) for 80 minutes before imaging.

Figure 2.. NLRP3 aggregates on dispersed TGN.

a, b, Nigericin induced NLRP3 aggregation on dispersed TGN (dTGN) in HeLa cells stably expressing NLRP3-GFP. Percentage of cells with NLRP3 puncta on dTGN was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable. c, d, Nigericin (10 μM) and ATP (5 mM) both induced endogenous NLRP3 aggregation on dTGN in primary ASC-deficient BMDMs. The cells were primed with LPS (50 ng/mL) for 3 hours. The data was collected and analyzed as in (b). e, Immunoblotting of indicated organelle markers in P5 (heavy membrane), P100 (light membrane) and S100 (cytosol) fractions. f, The P5 as shown in e was further fractionated by sucrose gradient ultracentrifugation followed by NLRP3 activity assay (top panel) or immunoblotting of each fraction (remaining panels).

Figure 3.. NLRP3 is recruited to dTGN via its polybasic region.

a, HeLa cells stably expressing the N-terminally truncated NLRP3 proteins were stimulated −/+ nigericin. b, The polybasic region of NLRP3 is highly conserved in all identified NLRP3 orthologs (aligned using Clustal Omega). c, Cells stably expressing the indicated proteins were treated as in (a), before percentage of cells with NLRP3 puncta was quantified from 100 cells (n = 3, mean ± SD). N.D., not detectable. d, HeLa cells stably expressing NLRP3 WT, 4KA or 4KR and other indicated proteins were treated −/+ nigericin or gramicidin before immunoblotting. 4KA, K127/128/129/130A; 4KR, K127/128/129/130R.

Figure 4.. NLRP3 is recruited to dTGN via binding to PI4P.

a, Inducible recruitment system of phosphatases: the addition of rapamycin (Rapa) (1 μM) induced the heterodimerization of FRB and FKBP12, thus recruiting the phosphatase to TGN to hydrolyze its target phospholipid. b, Inducible translocation of PI4P phosphatase Sac1 to TGN inhibited nigericin-induced NLRP3 puncta formation. Arrows indicate cells with Sac1 expression. c, HeLa cells stably expressing the indicated proteins were treated −/+ nigericin or gramicidin. The percentage of cells with NLRP3 puncta was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable. n.s., not significant (alpha = 0.01). CS, C389S. d, Lysates from the cells treated as in (c) were analyzed by the in vitro NLRP3 activity assay.

Figure 5.. PI4P binding by NLRP3 is essential for the inflammasome activation

a, HeLa cells stably expressing the indicated proteins were treated −/+ nigericin before examined by the in vitro NLRP3 activity assay. 2RE, R107/108E. b, Primary NLRP3-deficient BMDMs reconstituted with the indicated proteins were primed with LPS before nigericin stimulation. Cell lysates (lys) and supernatants (sup) were analyzed by immunoblotting to detect cleaved caspase-1 and IL-1b. c, Cells were treated −/+ nigericin in the presence of increasing concentrations of KCl, before the percentage of cells with dTGN formation was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable. n.s., not significant (alpha = 0.01). d, Cells were treated as in (c), before the percentage of cells with NLRP3 on TGN (either intact or forming dTGN) was analyzed as in (c). e, Cells treated as in (c) and the cell lysates were analyzed by the in vitro NLRP3 activity assay.

Figure 6.. Binding to PI4P on dTGN is essential for K⁺ efflux-independent NLRP3 activation.

a, HeLa cells stably expressing the indicated proteins were treated −/+ imiquimod or CL097 (45 μg/mL) for 80 minutes before cell lysates were analyzed by the in vitro NLRP3 activity assay. b, c, Primary ASC-deficient BMDMs were primed with LPS and incubated −/+ imiquimod or CL097 (45 μg/mL) for 60 minutes. The percentage of cells with NLRP3 puncta on dTGN was quantified from 100 cells (n = 3, mean ± SD, two-sided t test). N.D., not detectable. d, Primary NLRP3-deficient BMDMs reconstituted with the indicated proteins were primed with LPS and treated −/+ imiquimod. Cell lysates (lys) and supernatants (sup) were analyzed by immunoblotting with the indicated antibodies to assess the inflammasome activation. e, HeLa cells stably expressing the indicated NLRP3 proteins were treated −/+ imiquimod in the presence of increasing concentrations of KCl and the cell lysates were analyzed by the in vitro NLRP3 activation assay.