Activation of inflammasomes requires intracellular redistribution of the apoptotic speck-like protein containing a caspase recruitment domain (ASC)

Abstract

Activation of caspase-1 is essential for the maturation and release of interleukin (IL)-1β and IL-18, and occurs in multi-protein complexes, referred to as inflammasomes. The apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) is the essential adaptor protein for recruiting pro-caspase-1 into inflammasomes, and consistently gene ablation of ASC abolishes caspase-1 activation and secretion of IL-1β and IL-18. However, distribution of endogenous ASC has not yet been examined in detail. In the present study we demonstrated that ASC localized primarily to the nucleus in resting human monocytes macrophages. Upon pathogen infection ASC rapidly redistributed to the cytosol, followed by assembly of perinuclear aggregates, containing several inflammasome components, including caspase-1 and Nod-like receptors (NLRs). Prevention of ASC cytosolic redistribution completely abolished pathogen induced inflammasome activity, which affirmed that cytosolic localization of ASC is essential for inflammasome function. Thus, our study characterized a novel mechanism of inflammasome regulation in host defense.

Inflammatory reactions in response to pathogen infection are highly coordinated processes. Leukocytes are recruited to sites of infection where they become activated and enlist tissue cells to aid with pathogen clearance. This response is orchestrated by a complex array of soluble mediators, including locally released cytokines and chemokines, such as macrophage-released interleukin (IL)-1β³ and IL-18. Generation of mature IL-1β and IL-18 is regulated at multiple steps, including transcription, posttranslational processing and receptor binding (1). Processing into the bioactive secreted 17 or 18 kDa forms, is dependent on the proteolytic activity of caspase-1, which itself becomes activated in molecular platforms, referred to as inflammasomes (2). Inflammasomes emerged recently as multi protein complexes that link recognition of damage-associated molecular patterns (DAMPs)³ by members of the Nod-like receptor (NLR)³ family of cytosolic pattern recognition receptors (PRRs) to the activation of caspase-1 and processing and release of the pro-inflammatory cytokines IL-1β and IL-18 (3-8). Core inflammasome proteins therefore include, NLRs, the pyrin domain (PYD)³ and caspase recruitment domain (CARD)³-containing adaptor protein apoptotic speck-like protein containing a CARD (ASC)³, and caspase-1 (2, 9). The inflammasome-initiating event is recognition of intracellular DAMPs derived from pathogens (PAMPs)³ or host (danger or stress signals) by a cytosolic NLR. Thus, inflammasomes are assumed to form in the cytosol of phagocytic cells, such as monocytes and macrophages. NLRs undergo ATP-dependent oligomerization in response to DAMP recognition, and recruit ASC by PYD-PYD interaction (10, 11). Subsequently, caspase-1 is recruited through the CARD of ASC, which is essential for its activation (12, 13). Consistently, macrophages deficient in ASC are impaired in their ability to activate caspase-1 in response to infection and tissue damage, emphasizing its central role for the activation of inflammatory caspases (14, 15).

NLRP3 (Cryopyrin) is activated by Gram positive bacteria and multiple DAMPs including MDP, bacterial and viral RNA, toxins that cause a decrease in intracellular potassium levels, genomic DNA, irritants, such as trinitro-chlorobenzene and dinitro-1-fluorobenzene and reactive oxygen species, as well as uric acid and calcium pyrophosphate crystals (16–22). NLRP3 also functions in concert with P2X7 receptors recognizing extracellular ATP (17, 20). Due to the central role of IL-1β and IL-18 in regulating inflammation and immunity, dysregulation of pathways leading to caspase-1 activation and the resulting uncontrolled secretion of these pro-inflammatory mediators is directly linked to human inflammatory disorders. Variants of NLRP1 are associated with autoimmune diseases that cluster with vitiligo (23). NLRP3-containing inflammasomes were recently linked to contact hypersensitivity, sunburn, essential hypertension, gout and pseudogout, and elevated expression of NLRP3 is detected in synovial fluids of RA patients (19, 24–27). Furthermore, hereditary mutations in NLRP3 rendering the protein constitutively active, are directly linked to cryopyrin-associated periodic syndromes (CAPS) (28, 29). Mutant NLRP3 proteins efficiently form complexes with ASC to mediate caspase-1 activation independent of an activating ligand. This finding demonstrates the potential benefits of controlling the accessibility of ASC for cytosolic NLRs (30). Several mechanisms have been proposed by which activation of inflammasomes might be regulated, including single PYD or CARD-containing proteins and pyrin (30). However, regulation of inflammasomes still remains largely elusive.

Here we describe a novel regulatory mechanism for inflammasome assembly. We propose that ASC is sequestered in the nucleus in resting monocytes and macrophages, and only becomes available to bridge NLRs and caspase-1 in the cytosol, once macrophages are activated in response to pathogen infection and cellular stress. Such a mechanism might provide an additional safety checkpoint to limit spontaneous activation of caspase-1 in resting cells.

Experimental Procedures

Cell culture

THP-1, U-937, HL-60, and HEK293 cells were obtained from the American Type Culture Collection and maintained as recommended by ATCC. HEK293 cells were transiently transfected using Polyfect (Qiagen) according to the manufacturer’s instructions. THP-1 cells were used at low passage numbers and were regularly tested for mycoplasma infection (MycoAlert, Lonza). THP-1 cells were transiently transfected using the Nucleofector II (Amaxa) with solution V and protocol V001 using two ASC-specific siRNAs targeting the 3′UTR and nucleotides 474–492 of the ASC ORF (31). THP-1 cells were seeded into collagen I (5 μg/cm²)-coated glass cover slips and differentiated into adherent macrophages by overnight culture in medium supplemented with 15 ng/ml of phorbol 12-myristate 13-acetate (PMA) and further cultured for 2 days. Where indicated, cells were treated with 2 μg/ml of E. coli total RNA (Ambion), or 2×10⁵ cfu/ml of heat killed Legionella pneumophila (HKL) or Staphylococcus aureus (HKSA) (InvivoGen).

Isolation of primary monocytes from human peripheral blood

Human monocytes were isolated by Ficoll-Hypaque centrifugation (Sigma) from heparanized blood. Mononuclear cells were removed, washed and resuspended in serum-free DMEM and isolated by adherence to plastic dishes. Adherent monocytes were washed, incubated in complete RPMI 1640 media overnight in glass chamber slide, and were left untreated or treated with 2 μg/ml of E. coli total RNA (Ambion). Macrophages were obtained by culture of adherent monocytes on collagen type I (5 μg/cm²)-coated glass cover slips for 7 days in medium containing 20% FBS. Alternatively, monocytes were isolated from PBMCs by countercurrent centrifugal elutriation in the presence of 10 μg/ml polymyxin B sulfate using a JE-6B rotor (Beckman Coulter).

Expression plasmids

The nuclear localization sequence (NLS) from the SV40 large T antigen (DPKKKRKV) was utilized to generate a 3xNLS-ASC fusion protein with RFP epitope tags in pcDNA3 (Invitrogen) and pMSCV-puro (Clontech) expression vectors (32). 3 silent point mutations were introduced in the siRNA recognition sequence (Quick-change, Stratagene) to prevent siRNA-mediated degradation of ASC and NLS-ASC. Authenticity of all plasmids was confirmed by sequencing. All other plasmids have been previously described (33).

Retroviral infections

293GP2 cells (Clontech) were transiently transfected with RFP-tagged ASC, NLS-ASC, empty pMSCV-puro, or GFP-ASC in modified pMSCV-puro expression vectors (Clontech) or pRNATin-H1.4 (Genscript), in combination with a VSV-G-encoding expression plasmid to produce a pseudo-typed recombinant retrovirus. 5× 10⁵ THP-1 cells were infected by spinoculation in the presence of 6 μg/ml polybrene (Sigma) at 32°C. Stable cells were selected 72 hours post infection with 3 μg/ml puromycin or 400 μg/ml hygromycin for 14 days. Expression of ASC and NLS-ASC was verified in pooled cell populations by immunoblot, and knock-down of ASC was verified by immunoblot in cells sorted by FACS for cGFP expression.

Generation of shRNA plasmids

shRNA expression constructs were generated by inserting double stranded oligonucleotides into the MluI and XhoI sites of pRNATin-H1.4 (Genscript) downstream of the RNA polymerase III H1 promoter. Stable knock down of ASC in THP-1 cells was achieved by infection with a VSV-G-pseudotyped retrovirus encoding an shRNA targeting ASC (targeting sequence gctcttcagtttcacacca) or firefly luciferase (targeting sequence gatttcgagtcgtcttaat) as control by spinoculation on three consecutive days. Cells were selected in hygromycin (400 μg/ml) for 14 days and sorted by FACS for cGFP expression, which is encoded in the vector backbone independent from the shRNA.

Immunofluorescence microscopy

Adherent cells on collagen I-coated cover slips (5 μg/cm²) were fixed in 3.7% paraformaldehyde, incubated in 50 mM glycine for 5 minutes and permeabilized and blocked with 0.5% saponin, 1.5% BSA, 1.5% normal goat serum for 30 minutes. ASC was immunostained with affinity purified monoclonal anti-ASC antibodies (MBL, 1 μg/ml) directed to the PYD, affinity purified polyclonal antibodies directed to the CARD from Chemicon, Calbiochem, and Apotech (1:200), custom raised affinity purified polyclonal antibodies directed to the linker (CS3; 1: 2,500), and monoclonal anti-myc antibodies (Santa Cruz Biotechnology, 1:350). Caspase-1 was immunostained with a monoclonal anti-caspase-1 antibody from BD Biosciences (1:500) and a polyclonal antibody from Santa Cruz Biotechnology (1:50); NLRP3 was immunostained with monoclonal anti-NLRP3 antibodies from Abcam (1:50) and Santa Cruz Biotechnology (1:50); and IL-1β with a polyclonal antibody from Santa Cruz Biotechnology (1:50). Secondary Alexa Fluor 488, 546, and 647-conjugated antibodies, ToPro-3, DAPI, and phalloidin were from Molecular Probes. Cells were washed with PBS containing 0.5% saponin, and cover slips were mounted using Fluoromount-G (Southern Biotech). Suspension cells were fixed and stained as above and adhered to poly-L-lysine-coated slides using a cytocentrifuge (StatSpin). Images were acquired by confocal laser scanning microscopy on a Zeiss LSM 510 or LSM 510 Meta, and epifluorescence microscopy on a Nikon TE2000E2 with a 100x oil immersion objective and image deconvolution.

Subcellular fractionation

10⁶ cells were resuspended in hypotonic lysis buffer (10 mM Tris-HCL pH 7.4, 10 mM NaCl, 3 mM MgCl₂, 1 mM EDTA, and 1mM EGTA, supplemented with protease and phosphatase inhibitors), incubated on ice, adjusted to 250 mM sucrose, and lysed in a Dounce homogenizer. Samples were initially centrifuged at 4°C at 1,000 ×g for 3 minutes to remove any intact cells and then centrifuged at 4°C at 2,000 ×g for 10 minutes to pellet the nuclei. The cytosolic supernatant was removed, and the nuclear pellet was then washed three times in hyptonic lysis buffer with the addition of 250 mM sucrose and 0.1% NP-40 and incubated for 20 minutes on ice. Both fractions were adjusted to 50 mM Tris-HCl pH 7.4, 20 mM NaCl, 3 mM MgCl₂, 250 mM sucrose, 0.5% deoxycholate, 0.1% SDS, 0.2% NP-40, and protease and phosphatase inhibitors, and fully solubilized by brief sonication. 50 μg of protein lysates were separated by SDS-PAGE, transferred to a PVDF membrane, and probed with anti-ASC antibodies (CS3) and HRP-conjugated secondary antibodies (Amersham Pharmacia) in conjunction with an ECL detection system (Pierce). Membranes were stripped and re-probed with anti-GAPDH (Sigma) and anti-Lamin A (Santa Cruz Biotechnology) antibodies as control for cytosolic and nuclear fractions, respectively.

Measurement of secreted IL-1β

IL-1β secretion was quantified in culture supernatants using a commercial ELISA (BD Biosciences) at least in triplicates. Briefly, primary macrophages or THP-1 cells were treated with E. coli total RNA (2 μg/ml), 2×10⁵ cfu/ml of heat killed Legionella pneumophila (HKLP) or 2×10⁵ cfu/ml of heat killed Staphylococcus aureus (HKSA) (InvivoGen) in 12-well dishes for 16 hours and the cleared culture supernatant was analyzed by ELISA. HEK293 cells were grown in collagen I-treated 12-well culture dishes and co-transfected with plasmids encoding mouse pro-IL-1β, pro-caspase-1, NLRP3^R260W, ASC, or NLS-ASC. 24 hours post-transfection, culture media were replaced, collected 36 hours post transfection, clarified, and IL-1β activated by reconstituted inflammasomes was analyzed by mouse IL-1β ELISA according to the manufacturer’s protocol (13, 33, 34).

Statistical methods

A standard two-tailed t-test was used for statistical analysis; P values of 0.05 or less were considered significant.

Results

Nuclear localization of ASC in resting monocytes

ASC has previously been shown to localize diffusely throughout the cell (Fig. 1A, upper panel), as well as to form characteristic aggregates (Fig. 1A, middle panel) (35). In addition, THP-1 cells stably expressing GFP-ASC, predominantly exhibited nuclear distribution of GFP-ASC, as determined by nuclear counter staining (Fig. 1A, lower panel). However, most studies to date have relied upon over-expression of ASC. To clarify the localization of ASC and its implications on the activity of inflammasomes, we studied the subcellular localization of endogenous ASC in monocytes and macrophages, where it is essential for inflammasome function (14, 15). First we examined ASC subcellular localization by immunofluorescence microscopy in primary human peripheral blood monocytes (Fig. 1B, 1^st panel) and a panel of human monocytic cell lines, including HL-60 (Fig. 1B, 2^nd panel), U-937 (Fig. 1B, 3^rd panel) and THP-1 cells (Fig. 1B, 4^th panel). Without exception, all cells showed nuclear localization of ASC, emphasizing that endogenous ASC localized to the nucleus in monocytes. Immunostaining was performed with different anti-ASC antibodies, including commercially available mouse monoclonal (U-937) or rabbit polyclonal (HL-60) antibodies or a custom raised rabbit polyclonal anti-ASC antibody (THP-1 and primary monocytes). To further demonstrate specificity of nuclear ASC staining, we performed an ASC peptide competition assay. The custom raised polyclonal anti-ASC antibody was incubated with a 1000-fold molar excess of the peptide used initially for immunization, before being employed in immunostaining of THP-1 cells (Fig. 1B, 5^th panel). Peptide competition completely abrogated nuclear ASC staining in THP-1 cells. These ASC antibodies are highly specificity for ASC without cross-reactivity by western blot of THP-1 cell lysates (Fig. 1C) and HEK293 cells that were transfected with myc-tagged ASC or control plasmid (Fig. 1D).

Figure 1. Nuclear localization of ASC.

(A) HEK293 cells were transiently transfected with myc-ASC, fixed, and immunostained with anti-myc and Alexa Fluor 488-conjugated antibodies, while GFP-ASC expressing THP-1 cells were fixed and cytospun onto glass slides. ASC either localized diffusively throughout the cell (upper panel) or into a perinuclear aggregate (middle panel), while stable expression of GFP-ASC in THP-1 cells showed nuclear localization (lower panel). (B) Subcellular localization of endogenous ASC was examined in primary human monocytes and several monocytic cell lines. Cells were fixed and ASC was immunostained with monoclonal antibodies from MBL (monocytes) (1^st panel), and polyclonal antibodies from Chemicon, Calbiochem, and custom raised (HL-60, U-937, THP-1) (2^nd, 3^rd, and 4^th panel). THP-1 cells were also immunostained in the presence of a 1000x molar excess of the ASC-specific peptide that is recognized by the CS3 anti-ASC antibody (5^th panel). Secondary Alexa Fluor 488-conjugated antibodies were used in combination with ToPro-3 and Alexa Fluor 546-conjugated phalloidin to counter stain nuclei (DNA) and the actin cytoskeleton to outline cells, respectively. Images were acquired by laser scanning confocal microscopy, and the panels from left to right are showing ASC (green), nucleus (blue), actin (red), and a merged image. (C) THP-1 lysates (75 μg) were analyzed by immunoblot using the anti-ASC antibodies used in immunofluorescence microscopy. An asterisk denotes a splice form of ASC, which is not recognized by our custom-raised polyclonal anti-ASC antibody (CS3). (D) HEK293T cells were either mock transfected or transfected with a myc-tagged ASC expression construct. Cleared cell lysates were analyzed by immunoblot using the anti-ASC antibodies used in immunofluorescence microscopy and anti-myc antibodies (Santa Cruz Biotech) as control. Blots were also stripped and re-probed for GAPDH to demonstrate equal loading.

Stimulation-dependent translocation of ASC from the nucleus to the cytosol in monocytes

Inflammasomes form in response to pathogen and PAMP recognition, and we therefore examined the localization of ASC in primary monocytes and THP-1 monocytes after treatment with E. coli total RNA for 30 minutes, which specifically activates an NLRP3-dependent inflammasome response (20, 36). Contrary to the nuclear localization of ASC observed in resting monocytes (Fig. 1B), immunofluorescence staining of E. coli total RNA activated cells revealed a cytosolic distribution of ASC in primary as well as THP-1 monocytes (Fig. 2A). To confirm this observation, proteins from resting and E. coli total RNA-treated THP-1 cells were fractionated into nuclear and cytosolic fractions and immunoblotted with ASC-specific antibodies. To control fractionation efficiency, membranes were stripped and re-probed with antibodies specific for the cytosolic GAPDH and the nuclear Lamin A proteins, respectively (Fig. 2B). Although the control proteins clearly indicated that the fractionation procedure was highly efficient, ASC was also seen in the cytosol in resting cells, which was not visible by immunofluorescence microscopy and was likely caused by nuclear leakage of the small ASC protein during aqueous fractionation, which is a well-acknowledged problem for small proteins (37, 38). ASC is only 22 kDa in size, and therefore retaining more than 50% of ASC inside the nucleus during fractionation further confirms its predominantly nuclear localization in resting monocytes, and that ASC redistributes from the nucleus to the cytosol following exposure to E. coli total RNA. Therefore, we conclude that PAMPs are capable of inducing translocation of ASC from the nucleus to the cytosol in monocytes.

Figure 2. Cytosolic redistribution of ASC in response to inflammatory stimulation of monocytes.

(A) Subcellular localization of ASC was analyzed by immunofluorescence in primary monocytes and THP-1 monocytes following treatment with E. coli total RNA (2 μg/ml) for 30 minutes, using the monoclonal anti-ASC antibody described in Fig. 1. Images were acquired by laser scanning confocal microscopy. Panels from left to right are showing ASC (green), nucleus (blue), actin (red), and a merged image. (B) Subcellular localization of ASC was determined by subcellular fractionation of control and E. coli total RNA (2 μg/ml)-activated THP-1 cells. 10⁶ cells were fractionated by differential centrifugation into nuclear (N) and cytosolic (C) fractions, and protein lysates (50 μg) were analyzed by immunoblotting with anti-ASC and HRP-conjugated secondary antibodies. Fractionation efficiency was controlled by re-probing membranes with anti-GAPDH (cytosolic) and anti-Lamin A (nuclear) antibodies. *denotes two cross reactive proteins in the cytosolic fraction following E. coli total RNA treatment.

Formation of cytosolic ASC aggregates in response to inflammatory activation of macrophages

Monocytes infiltrate sites of infection and differentiate into inflammatory macrophages to aid pathogen clearance and homeostasis. Therefore, we next examined distribution of ASC in primary macrophages, which were either resting or activated with E. coli total RNA for 6 hours. Similar to monocytes, resting macrophages displayed primarily nuclear ASC (Fig. 3A, upper panel), while the majority of E. coli total RNA-activated macrophages showed the characteristic ASC-containing perinuclear aggregates (Fig. 3A, lower panel), reminiscent of the structures that were observed upon ASC overexpression (Fig. 1A, middle panel). To prove that THP-1 cells can be used as a model for inflammasome studies, ASC-containing aggregates were also assessed in PMA-differentiated THP-1 macrophages by immunostaining before and after activation with E. coli total RNA for 6 hours. As observed for primary macrophages, differentiated resting THP-1 macrophages showed nuclear ASC (Fig. 3B, upper panel) and activation with E. coli total RNA for 6 hours caused ASC to form cytosolic aggregates (Fig. 3B, middle panel). To demonstrate specificity for the staining of ASC-containing aggregates, we also performed a peptide competition assay with the custom rabbit polyclonal anti-ASC antibody, as shown in figure 1B. Peptide competition completely abrogated any ASC aggregate staining (Fig. 3B, lower panel), which confirmed the specificity of the ASC antibody for ASC aggregate staining. Since ASC is the only known adaptor protein for inflammasome mediated caspase-1 activation, ASC-containing aggregate formation may be crucial for caspase-1 activation and consequently, macrophages from ASC^−/− mice fail to activate caspase-1 in response to Gram-positive and Gram-negative bacteria (14). Gram-positive but not Gram-negative pathogens are able to induce inflammasomes in an ASC and NLRP3-dependent manner (14, 17, 20, 39). To demonstrate that the observed response is not only due to a PAMP, THP-1 derived macrophages were infected with heat-killed Gram-positive Staphylococcus aureus (HKSA) for 4 hours, followed by immunofluorescence staining of ASC. Concomitant with NLRP3-dependent pathogen recognition and subsequent inflammasome formation and activation, Gram-positive pathogens also caused ASC aggregate formation (Fig. 3C, upper panel). Additionally, Gram-negative Legionella pneumophilia (HKLP) for which no specific NLR has been defined yet, also caused ASC-containing aggregates (Fig. 3C, lower panel). This data suggest that inflammasomes potentially activated by other NLRs are able to trigger ASC aggregate formation, which is therefore not limited to NLRP3.

Figure 3. Aggregate formation of endogenous ASC in response to inflammatory stimulation in macrophages.

Subcellular localization of ASC was analyzed by immunofluorescence in primary macrophages (A) and PMA-differentiated THP-1 macrophages (B,C), using mono- and polyclonal anti-ASC and Alexa Fluor 488-conjugated secondary antibodies in combination with ToPro-3 nuclear stain and Alexa Fluor 546 conjugated phalloidin to visualize actin. Images were acquired by laser scanning confocal microscopy. All panels show ASC (green), nucleus (blue), actin (red), and a merged image from left to right. (A) Primary macrophages, either untreated (upper panel) or E. coli total RNA treated (2 μg/ml) (lower panel), were immunostained using the CS3 polyclonal anti-ASC antibody. (B) Untreated (upper panel) and E. coli total RNA (2 μg/ml) treated THP-1 macrophages (middle and lower panels) were immunostained using the polyclonal anti-ASC antibody in the absence (middle panel) or in the presence (lower panel) of a 1000x molar excess of the ASC-specific peptide (note the loss of specific ASC aggregate staining in the panel with peptide competition). (C) THP-1 macrophages were activated with 2×10⁵ cfu/ml heat killed Staphylococcus aureus (HKSA) (upper panel) and Legionella pneumophilia (HKLP) (lower panel), and immunostained with the CS3 polyclonal antibody.

ASC aggregation is an early response of macrophage activation

Maturation and release of IL-1β by macrophages occurs rapidly in response to infections. We therefore wanted to investigate a time course of ASC nuclear to cytosolic redistribution and formation of aggregates in response to stimulation with E. coli RNA. THP-1 cells were kept untreated or treated with E. coli total RNA for 0.5, 1, 6 and 24 hours, and distribution of ASC was analyzed in these cells by immunofluorescence (Fig. 4). Within 30 minutes of E. coli total RNA treatment the majority of cells displayed ASC redistribution from the nucleus to the cytosol, with some cells showing diffuse nuclear and cytosolic distribution. After 1 hour, ASC was cytosolic, with aggregate formation visible in some cells. Longer exposure to E. coli total RNA induced ASC-containing aggregates in the majority of cells and further increased the size of these aggregates, which are still present after 24+ hours of activation, without any signs of toxicity. Significantly, duration of ASC-containing aggregates required the persistent presence of E. coli total RNA (or LPS or heat killed pathogens, data not shown), since wash-out of E. coli RNA after 6 hours of exposure caused a dissociation of aggregates with some nuclear accumulation of ASC visible after 12 hours (Fig. 4, bottom panel). This observation suggested that ASC redistribution and aggregation is reversible, which could indicate a function in host response to infections and other stress situations.

Figure 4. Aggregation of ASC occurs within one hour of inflammatory stimulation of macrophages and depends on the continued presence of the stimulus.

Subcellular localization of ASC in THP-1 macrophages was analyzed by immunofluorescence following stimulation of cells with E. coli total RNA (2 μg/ml) for the indicated times. Cells were fixed at 0 minutes, 30 minutes, 1 hour, 6 hours, and 24 hours post-stimulation. In addition, after 6 hours of stimulation, cells were washed extensively and cultured for additional 12 hours in the absence of E. coli total RNA. Fixed cells were stained with the polyclonal CS3 anti-ASC, and Alexa Fluor 488-conjugated secondary antibodies. DNA and the actin cytoskeleton were visualized as above. The panels from left to right are showing ASC (green), nucleus (blue), actin (red), and a merged image.

Caspase-1 and NLRP3 co-localize with ASC to aggregates

Inflammasomes are multi protein complexes consisting of at least an NLR protein, ASC, and caspase-1 (2). Over-expression studies suggested that many ASC-interacting proteins co-localize to aggregates of over-expressed ASC. We find that, in response to pathogen infection, macrophages display a 4 to 6 μm ring shaped perinuclear structure, which contains endogenous ASC (Fig. 5A). Upon compiling of 20 images along the z-axis, a 3 dimensional image revealed that this structure is spherical with a hollow center (Fig. 5B). To characterize localization of additional inflammasome proteins, immunostaining for caspase-1 and NLRP3 was performed. In resting THP-1 macrophages caspase-1 localized to the nucleus and to small vesicular structures in the cytosol (Fig. 5C, upper panel), while E. coli total RNA-activation caused partial localization of caspase-1 to perinuclear aggregates (Fig. 5C, lower panel). Co-staining of caspase-1 with ASC showed that both proteins displayed a similar pattern inside the nucleus in resting THP-1 macrophages (Fig. 5D, upper panel), and that both proteins co-localize to aggregates in E. coli total RNA-treated cells (Fig. 5D, lower panel). NLRP3^R260W is a constitutively active CAPS-linked mutant, causing ligand-independent recruitment of ASC and caspase-1 (40, 41). Co-expression of ASC, pro-caspase-1 and NLRP3^R260W in HEK293 cells, caused formation of ASC-containing aggregates and co-localization of all three inflammasome proteins in these aggregates (Fig. 5E). We did not obtain specific staining for endogenous NLRP3 in resting cells, which is in agreement with reports that NLRP3 is inducibly expressed in response to pro-inflammatory agents (42). Nevertheless, in E. coli total RNA-treated cells, NLRP3 was present in ASC-containing aggregates in some cells, as well as in punctate structures throughout the cytosol (Fig. 5F). Overall, these results demonstrate that macrophages recruit NLRs, ASC and caspase-1 into perinuclear aggregates in response to pathogen infection, providing the intriguing possibility that these aggregates may represent inflammasomes.

Figure 5. Caspase-1 and NLRP3 co-localize with ASC in aggregates.

All images show PMA-differentiated THP-1 macrophages, except (E), which are HEK293T cells. (A) Cells were treated with E. coli total RNA (2 μg/ml) and immunostained with polyclonal anti-ASC and Alexa Fluor 546-conjugated antibodies (red), monoclonal anti-β-tubulin and Alexa Fluor 488-conjugated antibodies (green) and the nuclear stain DAPI (blue). The insert in the lower right half highlights the ring-shaped ASC-containing aggregate. The scale bar measures 10 μm and 1 μm, respectively. (B) E. coli total RNA (2 μg/ml)-treated cells were immunostained with polyclonal anti-ASC and Alexa Fluor 546-conjugated antibodies (red) and DAPI (blue). 20 optical sections at 0.6 μm were captured, deconvoluted and assembled into a 3D structure, showing xy, yz, and xz views of the aggregate. The scale bar is 10 μm. (C, D) Untreated (upper panel) or E. coli total RNA (2 μg/ml)-treated cells (lower panel) were fixed and immunostained with (C) polyclonal anti-caspase-1 and monoclonal β-tubulin and Alexa Fluor 546 and 488-conjugated secondary antibodies, respectively. (D) Polyclonal anti-ASC and monoclonal anti-caspase-1 antibodies and Alexa Fluor 488- and 546-conjugated secondary antibodies, respectively. DNA was visualized with DAPI, and panels from left to right show (C) caspase-1 (red), nucleus (blue), β-tubulin (green), and a merged image and (D) ASC (green), caspase-1 (red), nucleus (blue), and a merged image. (E) Flag-tagged ASC, HA-tagged caspase-1 and myc-tagged NLRP3^R260W were transiently transfected into HEK293 cells and immunostained with rabbit anti-ASC, mouse anti-NLRP3, and rat anti-HA antibodies and Alexa Fluor 546-, 647-, and 488-conjugated secondary antibodies, respectively, to determine co-localization of all three proteins in transfected cells. Panels from left to right show ASC (red), NLRP3^R260W (blue), caspases-1 (green), phase, and a merged image. (F) E. coli total RNA (2 μg/ml)-treated THP-1 macrophages were immunostained with rabbit anti-ASC and goat anti-NLRP3 antibodies and Alexa Fluor 488- and 546-conjugated secondary antibodies, respectively. Panels from left to right show ASC (red) and NLRP3 (green) and a merged image.

ASC aggregate-dependent maturation of IL-1β

To test the hypothesis that these aggregates may represent inflammasomes, we determined IL-1β secretion in cell culture supernatants by ELISA under conditions that cause ASC-containing aggregate formation in E. coli total RNA treated macrophages (Fig. 6A). Secreted IL-1β was significantly elevated in macrophages after E. coli total RNA treatment over untreated control. To directly link elevated levels of released IL-1β to inflammasomes, we generated stable THP-1 cells with an shRNA-mediated knock-down of ASC. shRNA ASC or control THP-1 cells were treated with E. coli total RNA, HKLP or HKSA that cause ASC aggregate formation and secreted IL-1β levels were determined by ELISA in culture supernatants. While control cells show a significant increase in secreted IL-1β, ASC knock-down cells are completely impaired in the release of IL-1β into culture supernatants (Fig. 6B). The efficiency of ASC knock-down was assessed by immunoblot in cleared lysates of control and ASC knock-down cells (Fig. 6B). Furthermore, wild type cells (GFP negative) form ASC-containing aggregates, as determined by immunofluorescence analysis, whereas ASC knock-down THP-1 cells (GFP-positive) do not show ASC-containing aggregates following E. coli total RNA treatment, further demonstrating the link between ASC-containing aggregates and IL-1β release (Fig. 6C).

Figure 6. ASC aggregate formation is linked to the maturation of IL-1β.

(A) Normalized culture supernatants from resting and E. coli RNA (2 μg/ml)-activated primary human macrophages were analyzed by ELISA for released IL-1β. Results represent an average of two independent experiments, and are presented as fold release compared to untreated macrophages. (B) Culture supernatants from THP-1 cells stably transfected with either an shRNA targeting ASC (black bars) or a control shRNA targeting luciferase (gray bars) and left either untreated, or treated with E. coli RNA (2 μg/ml), HKLP (2×10⁵ cfu/ml) or HKSA (2x10⁵ cfu/ml) were analyzed as above for released IL-1β. Cells were previously FACS sorted for GFP expression, which is encoded from the pRNATin-H1.4 (Genscript) shRNA vector backbone. Results represent an average of three independent experiments, and are presented as pg/ml of released IL-1β. The insert shows an immunoblot of control shRNA and ASC shRNA expressing THP-1 cells for ASC and GAPDH as a loading control. (C) ASC-containing aggregates were analyzed following treatment with E. coli total RNA (2 μg/ml) for 6 hours by immunofluorescence in a mixed population of shRNA ASC transfected cells before FACS sorting for GFP expression of the ASC shRNA. ASC was immunostained with the polyclonal CS3 antibody, and images were acquired by laser scanning confocal microscopy. Panel shows from left to right ASC (red), ASC shRNA (green), nucleus (blue), and a merged image. Note that the cell on the left side (white arrow) encodes the ASC shRNA, as indicated by the GFP positive signal and therefore does not form ASC-containing aggregates, while the cell on the right is GFP negative, thus does not encode the ASC shRNA and therefore shows aggregated ASC.

Redistribution of ASC to the cytosol is essential for inflammasome activity

The significance of cytosolic redistribution of ASC is further emphasized by the cytosolic distribution of the inflammasome target, pro-IL-1β, in response to E. coli total RNA treatment of THP-1 cells. Consistent with previous reports that show upregulated pro-IL-1β expression in response to infection (43), no IL-1β staining was observed in resting cells (Fig. 7A, upper panel), while cytosolic distribution is present in activated cells (Fig. 7A, lower panel). To further test the hypothesis that ASC cytosolic redistribution is essential for inflammasome formation and activity, we altered its intracellular localization. A constitutive nuclear ASC was generated by in-frame fusion of ASC with three tandem repeats of the strong consensus nuclear localization sequence (NLS)³ derived from the SV40 large T antigen, which has frequently been employed to target proteins to the nucleus (32, 44). Three copies of the NLS were used to ensure efficient nuclear retention of ASC even after inflammatory stimulation. We chose this approach over mutational analysis because both death domain folds of ASC require correct folding to retain full adaptor function and even point mutations could disrupt the overall structure and function of ASC, which may lead to a wrong interpretation of results, while NH₂ terminal fusion of ASC generally does not alter its function (data not shown). To confirm the nuclear localization of this ASC fusion protein (NLS-ASC), HEK293 cells were transfected and localization was compared to wildtype ASC. As shown in Fig. 1A, overexpression of wildtype ASC results in either nuclear/cytosolic localization, or preferentially the formation of perinuclear aggregates (Fig. 1A and 7B, upper panel). In contrast, transfection of NLS-ASC showed exclusively nuclear localization, suggesting that the fusion of a strong NLS to ASC was sufficient to retain it inside the nucleus (Fig. 7B, lower panel). To determine the functional significance of nuclear versus cytosolic distribution of ASC, we utilized an inflammasome reconstitution assay in HEK293 cells, which are deficient in endogenous inflammasomes, but are able to produces mature IL-1β upon reconstitution of inflammasomes (13, 33, 34). We reconstituted inflammasomes by transient transfection of pro-caspase-1, pro-IL-1β, NLRP3^R260W, and either ASC or NLS-ASC, and determined the amount of secreted IL-1β in culture supernatants. While inflammasomes containing wild type ASC caused processing and release of IL-1β, inflammasomes restored with NLS-ASC were not capable of secreting IL-1β into culture supernatants above baseline level (Fig. 7C). Total cell lysates were used to verify expression of all transfected inflammasome proteins by immunoblotting (Fig. 7C). To directly assess inflammasome activity in macrophages, we infected THP-1 cells with recombinant retroviruses encoding ASC or NLS-ASC, which is not targeted by the siRNA, because we introduced three silent point mutations in the siRNA recognition sequence. We used an RFP fusion to conveniently distinguish between endogenous and ectopic ASC by the increased molecular weight, and chose stable pooled cell populations with expression levels comparable to endogenous ASC, as analyzed by immunoblotting (Fig. 7D, right upper panel). Endogenous ASC was knocked down by two pooled siRNAs, and was confirmed by immunoblotting (Fig. 7D, right lower panel). Next, IL-1β release was measured in response to E. coli total RNA treatment. While activation of control cells strongly promoted release of IL-1β, knockdown of endogenous ASC from wild type cells efficiently reduced IL-1β release, which was restored in RFP-ASC expressing cells. In contrast, expression of NLS-ASC in ASC knockdown cells failed to restore secretion of IL-1β above the level observed in stimulated ASC knockdown cells (Fig. 7D, left panel). These data confirmed that nuclear export of ASC is essential for inflammasome activity, since nuclear retention of ASC completely impaired inflammasome-mediated IL-1β release.

Figure 7. ASC localization to the cytosol is required for inflammasome formation and efficient IL-1β secretion.

(A) Untreated and E. coli RNA-treated THP-1 macrophages were immunostained with anti-IL-1β and Alexa Fluor 488-conjugated secondary antibodies. Nuclei and actin were visualized as above. Panels show from left to right IL-1β (green), nucleus (blue), actin (red), and a merged image. (B) Myc-tagged ASC (upper panel) and myc-tagged NLS-ASC (lower panel) were transiently transfected into HEK293 cells and immunostained with monoclonal anti-myc and Alexa Fluor 546-conjugated secondary antibodies. Nuclei and actin were visualized as above. Panels show from left to right ASC (red), actin (green), nucleus (blue), and a merged image. Images in (A, B) were acquired by laser scanning confocal microscopy. (C) Inflammasomes consisting of a constitutively active NLRP3^R260W, pro-IL-1β, pro-caspase-1, and either ASC or NLS-ASC, were transiently reconstituted in HEK293 cells, as indicated, and inflammasome activity was assayed by analyzing secreted IL-1β by ELISA. Results represent an average of at least three independent experiments +/− SD. Cleared and normalized cellular lysates were analyzed by immunoblot for expression of all transfected inflammasome components, as indicated. (D) THP-1 cells were stably infected with a VSV-G pseudo-typed recombinant retrovirus encoding either red fluorescent protein (RFP)-fused ASC or RFP-fused NLS-ASC with expression levels comparable to endogenous ASC. Mock-infected, RFP-ASC, and RFP-NLS-ASC expressing cells were transiently transfected with two pooled ASC-specific siRNAs to knock-down expression of endogenous ASC, but not RFP-ASC or RFP-NLS-ASC. Cells were seeded into fresh wells 24 hours post nucleofection, and where indicated treated with LPS and E. coli RNA for 12 hours. Secreted IL-1β was determined in culture supernatants by ELISA and results are presented as pg/ml of 10⁶ cells, and represent an average of three independent experiments +/− SD. The insert shows stable expression of RFP-ASC and RFP-NLS-ASC compared to endogenous ASC in THP-1 cells (upper panel) and the siRNA-mediated reduction of endogenous ASC (lower panel) by immunoblot. An asterisk denotes expression of ASC and NLS-ASC containing silent point mutations in the sequence recognized by the siRNA preventing its degradation.

Discussion

In this report we studied the intracellular distribution of the central inflammasome adapter ASC and provide evidence that inducible redistribution of ASC from the nucleus to the cytosol is essential for inflammasome function. While inflammasomes are biochemically characterized, no detailed studies have been undertaken to visualize endogenous inflammasomes. Over-expression studies result predominantly in the formation of aggregates, referred to as specks (35, 45). However, ASC-containing aggregates were believed to represent artificial results driven by enforced over-expression causing self-aggregation of ASC. We observed nuclear localization of ASC in monocytes and macrophages with several anti-ASC antibodies recognizing different epitopes, thus excluding epitope masking as an explanation for our results.

Nuclear localization of ASC, however, cannot explain its essential function as an adaptor protein for inflammasome-mediated caspase-1 activation, which occurs in the cytosol of cells (46). We find cytosolic accumulation of ASC in response to purified PAMPs as well as heat-killed Gram negative and Gram positive bacteria, suggesting that this response is part of the host defense to clear pathogens. Redistribution of ASC from the nucleus to the cytosol can be observed as early as 30 minutes post-stimulation, while longer stimulation caused the aggregation of ASC into structures that resembled the specks observed by enforced over-expression of ASC, including the perinuclear localization and the spherical morphology with a hollow center. Additionally, ASC-containing aggregates appear to grow in size over time. To our knowledge, we provide the first evidence of host defense-mediated assembly of ASC-containing aggregates in macrophages. We propose that inducible redistribution of ASC from the nucleus to the cytosol is required for its interaction with activated NLRs to form caspase-1-activating inflammasomes. ASC-containing aggregates not only assembled in response to the NLRP3 ligands bacterial RNA, and S.aureus, but also in response to Gram negative bacteria and other PAMPs (data not shown), suggesting that other NLRs also aggregate with ASC, consistent with the requirement of ASC for inflammasome-mediated activation of caspase-1 in response to Gram negative and Gram positive pathogens (14).

Caspase-1 activation can induce pyroptosis, which is a form of programmed cell death associated with antimicrobial responses in inflammation, which results in osmotic lysis of cells (47). A caspase-1 and GFP-ASC containing protein complex termed pyroptosome has recently been proposed to facilitate pyroptosis through self-oligomerization of ASC in response to potassium depletion (45). However, other reports did not find a link between ASC and pyroptosis (48). NLRP3 and ASC have also been implicated in caspase-1 and IL-1β-independent pyronecrosis (49). Thus, the role of ASC and caspase-1 containing complexes in pyroptosis and pyronecrosis is still unclear. We propose that ASC may participate in different multi-protein complexes that facilitate either effective IL-1β secretion (inflammasomes) or cell death (pyroptosomes, pyronecrosomes). ASC has also been shown to localize to mitochondria in response to genotoxic insult, which is essential to facilitate p53-mediated Bax translocation to mitochondria and subsequently apoptosis in fibroblasts, hence participating in yet another signaling complex (50). This complex could be specific to DNA damage in fibroblasts.

Although we do not understand yet what the determining factor for such different outcomes is in macrophages, we noticed significant differences in the appearance and functional properties of these multi-protein complexes. Endogenous ASC aggregates in response to NLR activation are 4–6 μm compared to the smaller (~1 μm) and more compact aggregates (specks) caused by over expression of ASC and potassium depletion (45), or that form in response to cytotoxic treatment of macrophages (data not shown). In spite of aggregate formation, cells do not show any sign of cell death for at least 48 hours, which is in sharp contrast to the rapid cell death in response to the formation of GFP-ASC-containing pyroptosomes (45). ASC-containing aggregates are reversible, which is contrary to structures causing cell death, because removal of the activating ligand by washing in culture medium resulted in dissociation of ASC-containing aggregates and diffuse cytoplasmic and even nuclear re-distribution of ASC. Furthermore, these complexes contain caspase-1 and NLRs, though caspase-1 co-localization with ASC was more frequently detectable than co-localization with NLRP3. Ultrastructural studies performed by electron microscopy and molecular approaches investigating the release of IL-1β from activated monocytes also did not find any signs of cell death during IL-1β release, and suggested a secretory mechanism that does not involve apoptosis, cell death, or surface membrane lysis (51, 52).

Since conditions that induced ASC aggregates also caused maturation of IL-1β and IL-18, these ASC-containing aggregates could potentially represent inflammasomes. Inflammasomes have been biochemically characterized as large, inducible protein aggregates, consisting of an NLR, ASC, and pro-caspase-1 (2). Other proteins can be associated with some inflammasomes. Thus, ASC-containing aggregates might provide a scaffold for inflammasome proteins to interact and to promote caspase-1 activation. Activation of NLRs causes their unfolding and NTP-mediated oligomerization, resulting in pentameric and heptameric aggregates in vitro (11). Thus, it is feasible that ASC aggregates might denote higher order oligomeric structures of NLRs, as indicated by their larger size. Alternatively, larger aggregates might assemble in vivo, because incorporation of accessory proteins is lacking by in vitro reconstitution. Recombinant NLRP1 pentamers or heptamers are about 20 nm in diameter (11), while most endogenous ASC aggregates are 4–6 μm. Adherence of cells is not required for ASC-containing aggregates, as we observed it also in activated monocytes (data not shown).

Fusion of proteins with nuclear localization sequences has been widely employed to prevent redistribution of proteins with nuclear and cytosolic localization. We employed this approach to sequester ASC inside the nucleus of macrophages to prevent ASC export. Nuclear ASC completely abrogated endogenous as well as reconstituted inflammasome-mediated maturation of IL-1β, directly demonstrating the requirement for ASC redistribution for inflammasome activity. We also noted co-localization of ASC with pro-caspase-1 in the nucleus of resting macrophages, suggesting that both proteins might pre-assemble in the nucleus of cells even before oligomerization with activated NLRs in the cytosol. In addition, both also localized to ASC-containing aggregates in activated macrophages. Different localizations have been reported for caspases-1, including cytosolic, vesicular, nuclear, as well as the inner and outer face of plasma membranes, which likely reflect different activation states of cells (51, 53–55). Moreover, caspase-1 is responsible for its own secretion alongside its cytokine substrates (2, 56). Nuclear caspase-1 has been suggested to mediate apoptosis in tumor cells, but we did not observe apoptosis in macrophages, in spite of its nuclear localization (53). This is supported by our observation that resting macrophages showed nuclear caspase-1, while a cytosolic to nuclear translocation occurs in neuroblastomas undergoing apoptosis (55).

Whether ASC is only physically separated inside the nucleus to prevent its spontaneous recruitment to NLRs and subsequent secretion of inflammatory cytokines, or whether ASC performs inflammasome-independent functions inside the nucleus, warrants further investigations. The related protein FADD, which links death receptors to caspase-8 activation in the cytosol, has been recently shown to localize to the nucleus in non-apoptotic cells, and its export is required for caspases-8 activation (44, 57). Nuclear and cytosolic FADD is involved in genome surveillance and death receptor-mediated apoptosis, respectively. Nuclear FADD further regulates mitosis and proliferation of T-cells (58). Likewise cytosolic TRADD, another death domain fold-containing adapter protein, participates in death receptor-mediated apoptosis, while nuclear localized TRADD mediates PML and p53-mediated apoptosis and interacts with STAT1-α to modulate IFN-γ signaling in macrophages (59, 60).

Recent evidence suggested a crosstalk between Toll like receptors (TLRs)³ and the NLR system during host defense. Pathogen recognition by TLRs is required for up-regulation of inflammasome components and substrates, while subsequently the NLR system is essential for processing of the cytokine precursors. Our data indicate another layer of complexity in the regulation of IL-1β and IL-18 maturation, where pathogen recognition initially triggers nuclear to cytosolic redistribution of the NLR adaptor ASC, which then links NLRs to caspase-1 activation in inflammasomes (Fig. 8). While the molecular mechanism of ASC cytosolic redistribution is currently elusive, it potentially could be targeted for prevention of inflammasome-mediated inflammatory cytokine secretion in patients with periodic fever syndromes and other inflammatory conditions.

Figure 8. A model for inflammasome formation and activation.

Cross talk between the TLR and NLR system has been proposed, where initial DAMP recognition by TLRs triggers transcriptional up-regulation of pro-IL-1β and other inflammasome components. Subsequently, DAMP recognition by NLRs is required for maturation of pro-IL-1β and pro-IL-18. Our data indicate that DAMP recognition also causes redistribution of ASC from the nucleus to the cytosol by a yet elusive mechanism, which then can be recruited to activated NLRs to assemble inflammasomes. The perinuclear aggregates in activated macrophages contain the core inflammasome proteins and might represent inflammasomes.