Proapoptotic Chemotherapeutic Drugs Induce Noncanonical Processing and Release of IL-1β via Caspase-8 in Dendritic Cells

Innate and adaptive immune responses play important roles in cancer progression and therapy (1, 2). IL-1β is a pleiotropic proinflammatory cytokine that is predominantly expressed in myeloid leukocytes. Depending on context, IL-1β can contribute to beneficial antitumor immune responses or maladaptive responses such as neovascularization, leading to tumor survival. Much attention has been focused on the mechanisms of IL-1β production within particular tissue niches, including tumor microenvironments. Activation of TLRs or receptors for proinflammatory cytokines (including IL-1β per se) induces the NF-κB–dependent expression of proIL-1β (33 kDa) as a cytosolic, biologically inactive precursor protein. The canonical cleavage and processing of proIL-1β into mature IL-1β cytokine (17 kDa) are catalyzed by caspase-1, a pathway regulated by multiprotein inflammasome signaling complexes. Prior to its subclassification within the caspase family of proteases, caspase-1 was originally described as the IL-converting enzyme. The most intensively studied inflammasome comprises an oligomeric complex of procaspase-1 with the NLRP3 and ASC adapter proteins that rapidly assembles in response to diverse stress stimuli such as increased reactive oxygen species (ROS) (3), mitochondrial dysfunction (4), perturbation of intracellular ion homeostasis (5–7), disruption of lysosomal membrane integrity (8), and activation of deubiquitinases (9–11).

On the one hand, NLRP3/caspase-1 inflammasomes play a central role in the regulation of IL-1β production within tumor loci or in response to chemotherapeutic drugs. In tumor-bearing mice treated with oxaliplatin, NLRP3-dependent IL-1β release from dendritic cells (DCs) occurs via paracrine activation of P2X7 receptors (P2X7R) in response to ATP released from dying tumor cells (12). In this model, DC-derived IL-1β contributes to the tumor-specific immune response by polarizing tumor Ag-reactive CD8⁺ T cells into IFN-γ–producing cytotoxic T cells. Two other drugs, gemcitabine and 5-fluorouracil, have been shown to elicit NLRP3-dependent IL-1β production in response to the accumulation of cytosolic cathepsin B in myeloid-derived suppressor cells with consequent promotion of tumor growth and vascularization (13). Other reports have described the ability of proapoptotic chemotherapeutic agents, such as doxorubicin (Dox) and staurosporine (STS) analogs, to directly activate NLRP3- and caspase-1–dependent IL-1β processing in ex vivo LPS-primed macrophages via mechanisms correlated with ribotoxic stress or increased mitochondrial dysfunction (4, 14).

In contrast, recent studies have indicated that caspase-8 can substitute for caspase-1 and can function as an efficient IL-converting enzyme for proteolytic maturation of IL-1β. Dectin-1 engagement by fungal ligands can engage a noncanonical ASC/caspase-8 inflammasome for processing of IL-1β (15). In addition, Maelfait et al. (16) observed that activation of the Toll/IL-1R domain-containing adapter inducing IFN-β (TRIF) signaling pathway by TLR3 or TLR4 induced a caspase-8 signaling pathway that, when combined with cycloheximide-mediated inhibition of protein translation, was sufficient to drive efficient IL-1β processing even in caspase-1 knockout macrophages. A subsequent study found that treatment of LPS-primed macrophages or DCs with second mitochondrial-derived activator of caspases (Smac)–mimetic drugs triggered a robust maturation of IL-1β that was mediated in part by caspase-8 (17). Most recently, the engagement of Fas (CD95) signaling in wild-type (WT) or caspase-1 knockout macrophages was shown to induce IL-1β and IL-18 processing via caspase-8 (18). A Fas-induced caspase-8 cascade was also identified as a major pathway for IL-1β and IL-18 production in Listeria-infected peritoneal exudate cells isolated from B6 mice (19). Caspase-8 is best characterized for its multiple roles in the regulation of cell death via apoptosis or necroptosis (20). These reports of noncanonical IL-1β processing via caspase-8 indicate that proinflammatory cytokine production may be mediated by caspase-8, potentially in signaling complexes such as ripoptosomes, known to control apoptotic or necroptotic death cascades.

In this study, we assessed the relative contributions of caspase-1 and caspase-8 to the processing and release of IL-1β in LPS-primed DCs stimulated with the chemotherapeutic drug, Dox, or the proapoptotic agent, STS. Our observations support a model wherein TLR4 activation induces upregulation of proIL-1β together with the TRIF-dependent assembly of a latent caspase-8 signaling platform. Proapoptotic agents, such as Dox or STS, act to license such caspase-8 signaling complexes for maximal caspase-8–mediated cleavage of proIL-1β, in part by depleting cellular inhibitor of apoptosis proteins (cIAPs). Comparative analyses of DCs isolated from control mice versus Casp1/11^−/− or Casp8^−/−Rip3^−/−− mice indicated that caspase-8 acts as the predominant IL-1β–converting enzyme in response to engagement of TLR4 signaling when the intrinsic apoptotic pathway is triggered by chemotherapeutic agents.

Key reagents and their sources were as follows: Escherichia coli LPS serotype O1101:B4 (List Biological Laboratories), Pam₃CSK₄ (Invivogen), Ac-YVAD-cmk (Bachem), z-IETD-fmk (R&D Systems), zVAD-fmk (Tocris or Abcam), murine rGM-CSF (PeproTech), murine rTNF-α (PeproTech), Dox (LC Laboratories or Sigma-Aldrich), STS (LC Laboratories), UCN-01 (Enzo), and necrostatin-1 (Nec-1; Tocris). Oxaliplatin, cisplatin, TRIzol reagent, nigericin, and ATP were purchased from Sigma-Aldrich. Anti–caspase-1 (p10) rabbit polyclonal (sc-514), anti-actin goat polyclonal (sc-1615), anti–Fas-associated death domain protein (FADD) goat polyclonal (sc-6036), and all HRP-conjugated secondary Abs were from Santa Cruz Biotechnology. The 3ZD anti–IL-1β mAb, which recognizes both 33-kDa proIL-1β and 17-kDa mature IL-1β in Western blot analysis, was provided by the Biological Resources Branch, National Cancer Institute, Frederick Cancer Research and Development Center (Frederick, MD). Other Abs included anti-cIAP1 mouse monoclonal (1E1-1-10) from Enzo, anti–receptor-interacting protein (RIP) mouse monoclonal (38-RIP) from BD Biosciences, and anti-NLRP3 mouse monoclonal (Cryo-2) from AdipoGen. Anti–caspase-7 (9492), anti–caspase-8 (4927), and anti–poly-ADP-ribosyltransferase (PARP; 9542) rabbit polyclonal Abs were from Cell Signaling. Murine IL-1β DuoSet ELISA kit was from R&D Systems, and the murine TNF-α ELISA kit was from BioLegend. Cell Titer-Glo Luminescent Viability Assay kit was from Promega. EnzChek Caspase-3 assay kit and DEVD-cho were from Invitrogen. RT² SYBR Green/ROX quantitative PCR (qPCR) Master Mix (PA-012) and predesigned qPCR primers for murine IL-1β (PPM03109E), murine TNF-α (PPM03113F), and murine GAPDH (PPM02946E) were from SA Biosciences. Transcriptor first-strand cDNA synthesis kit was from Roche.

WT C57BL/6 mice were purchased from Taconic. Mice lacking both caspase-1 and caspase-11 on the C57BL/6 background (Casp1^−/− Casp11^−/−) have been previously described (21–23). Asc^−/−, Trif^−/−, and Nlrp3^−/− Nlrc4^−/− double-knockout mouse strains (C57BL/6 background) were obtained from E. Pearlman (Case Western Reserve University). Femoral and tibial bones for bone marrow–derived DC (BMDC) cultures were also isolated from Casp8^−/−Rip3^−/−, Casp8^+/+Rip3^−/−, and litter-matched control Casp8^+/+Rip3^+/+ mice, which have been described previously (24). All experiments and procedures involving mice were approved by the Institutional Animal Care and Use Committees of Case Western Reserve University.

BMDC from 9- to 12-wk-old mice were isolated by minor modification of previously described protocols (25). Mice were euthanized by CO₂ inhalation. Femura and tibiae were removed and briefly sterilized in 70% ethanol, and PBS was used to wash out the marrow cavity plugs. The bone marrow cells were resuspended in DMEM (Sigma-Aldrich); supplemented with 10% bovine calf serum (HyClone Laboratories), 100 U/ml penicillin, 100 μg/ml streptomycin (Invitrogen), 2 mM l-glutamine (Lonza), and 15 ng/ml GM-CSF; plated onto 150-mm dishes; and cultured in the presence of 10% CO₂. On day 3 postisolation, 80% of the nonadherent population was removed and centrifuged at 300 × g for 5 min at room temperature, and fresh medium was applied. Five days postisolation, the resulting loosely adherent BMDC were collected; resuspended to a cell density of ∼1 × 10⁶/ml in the above differentiation medium; plated into 6-well (2 ml/well), 12-well (1 ml/well), 24-well (0.5 ml/well), or 96-well (0.1 ml/well) plates as needed for particular experiments; and used between days 7 and 10 postisolation.

For experimental tests, plated BMDC were centrifuged at 300 × g for 5–10 min, and the differentiation medium was removed and replaced with low-serum DMEM (0.1% bovine calf serum plus penicillin, streptomycin, and l-glutamine). The cells were equilibrated for 15 min at 37°C in 10% CO₂ prior to addition of test reagents. BMDC were routinely primed with 1 μg/ml LPS (unless indicated otherwise) for 4 h to activate TLR4 signaling prior to treatment with indicated concentrations of Dox, STS, or other chemotherapeutic drugs for additional periods ranging from 2 to 18 h (with the LPS being present throughout). In some experiments, the cells were primed with 200 ng/ml Pam₃CSK₄ to induce TLR2, rather than TLR4, signaling cascades. Stimulation of NLRP3/caspase-1 inflammasome signaling by either P2X7R activation or nigericin treatment was routinely used as a positive control by supplementing the medium of LPS-primed BMDC with either 5 mM ATP or 10 μM nigericin for the final 30 min of test incubations. Where indicated, the BMDC cultures were treated with various pharmacological inhibitors (50–100 μM Ac-YVAD-cmk, 100 μM z-VAD-fmk, 100 μM z-IETD-fmk, 100 μM DEVD-cho, 50 μM Nec-1) either before or after the LPS priming steps.

BMDC were seeded in 6- or 12-well plates and treated, as described above, prior to processing of both extracellular medium and whole-cell lysates. Extracellular medium samples were concentrated by trichloroacetic acid precipitation/acetone washing, whereas cell lysates were prepared by detergent-based extractions, as described previously (21), prior to standard processing by SDS-PAGE, transfer to polyvinylidene diflouride membranes, and Western blot analysis. Primary Abs were used at the following concentrations: 5 μg/ml for IL-1β, 1 μg/ml for caspase-1, 1 μg/ml for actin, 0.1 μg/ml for caspase-7, 1.2 μg/ml for caspase-8, 0.05 μg/ml for PARP, 2 μg/ml for cIAP, 1 μg/ml for NLRP3, 0.4 μg/ml for RIP, and 1 μg/ml for FADD. HRP-conjugated secondary Abs were used at a final concentration of 0.13 μg/ml. Where indicated, Western blots were probed with anti–caspase-1 and anti–IL-1β Abs simultaneously. Chemiluminescent images of the developed blots were detected, stored, and quantified using a FluorChemE processor (Cell Biosciences).

BMDC were seeded in 24-well plates. Extracellular media samples were removed and centrifuged at 10,000 × g for 15 s to pellet floating BMDC. The supernatants were then assayed for murine IL-1β or murine TNF-α by standard ELISA (R&D Systems for IL-1β or BioLegend for TNF-α), according to the manufacturer’s protocol. All test conditions for ELISA-based experiments were performed in duplicate.

BMDC were seeded in 96-well plates. Viability of BMDC was measured by quantifying intracellular ATP content using the Cell Titer-Glo Luminescent Viability Assay (Promega). Luminescence was quantified in relative light units using a BioTek Synergy HT plate reader and normalized to the values measured in control, untreated BMDC.

BMDC seeded in 12-well plates were primed for 4 h with LPS and then incubated with various chemotherapeutic agents ± 100 μM DEVD-cho for an additional 2–18 h prior to preparation of cell lysates for quantification of accumulated caspase-3/7 activity by EnzChek Fluorescence Caspase-3 assay kits (Invitrogen).

Casp8^+/+Rip3^+/+, Casp8^−/−Rip3^−/−, and Casp8^+/+Rip3^−/− BMDC were treated with 1 μg/ml LPS for 1 or 4 h prior to extraction of total RNA by TRIzol reagent. Transcriptor First Strand cDNA Synthesis kit was used to generate first-strand cDNA from the purified RNA. IL-1β, TNF-α, and GAPDH transcripts were quantified using a StepOne-Plus Real-Time PCR System (Applied Biosystems) with reactions performed in 25 μl containing RT² SYBR Green/ROX qPCR Master Mix (12.5 μl), 1:100 dilutions of reverse-transcriptase product, and 1 μM PCR primer pair stock that were run in triplicate. Amplification cycle conditions were 95°C for 10 min, followed by 40 cycles of the following: 95°C, 15 s; 55°C, 30–40 s; and 72°C, 30 s. Expression of IL-1β or TNF-α was calculated using the ΔΔC_t method using StepOne software version 2.1 with values normalized to GAPDH expression.

BMDC (4–5 × 10⁶ in 60-mm culture dishes) were stimulated for 2 h with or without LPS (1 μg/ml) or TNF-α (50 ng/ml) in the absence or presence of 50 μM zVAD in DMEM containing 10% bovine calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine. Other dishes were primed with LPS (1 μg/ml) for 4 h and then stimulated for an additional 3–8 h with 10 μM Dox in the absence or presence of 50 μM zVAD. Incubations were terminated by removal of supernatant medium for sedimentation and isolation of detached cells. Detached cells were centrifuged at 400 × g for 5 min and washed with 1 ml ice-cold PBS. Whole-cell detergent lysates were prepared by addition of 85 μl radioimmunoprecipitation assay lysis buffer (0.5% Na deoxycholate, 0.1% SDS, 1% Igepal CA630 in PBS [pH 7.4] plus protease inhibitor mixture) to the adherent cells on the dish and incubated on ice for 20 min. Lysed adherent cells were scraped with a rubber policeman, pooled with detached cells, and extracted for an additional 10 min on ice. The whole-cell lysates were separated into detergent-soluble and detergent-insoluble fractions by centrifugation at 15,000 × g for 15 min at 4°C. SDS sample buffer (20 μl) was added to the detergent-soluble fractions, whereas 58 μl radioimmunoprecipitation assay lysis buffer (supplemented with 5 mM MgCl₂) was added to the insoluble lysate pellet. The insoluble lysate pellet was subjected to 5× freeze-thaw cycles (dry ice/ethanol and 37°C heating blocks) and vortexing after each freeze-thaw cycle. Each sample was then DNase treated (2 U/sample) and incubated on ice for 10 min prior to addition of SDS sample buffer (12 μl) and extraction at 100°C for 5 min.

All experiments were repeated two to eight times with separate BMDC preparations. Figures illustrating Western blot results are from representative experiments. As indicated, figures illustrating IL-1β ELISA, cell viability, or caspase activity results represent either the mean (±SE) of data from several identical experiments or the average (±range) of duplicate samples from single representative experiments. Experiments with three or more repeats were analyzed by one-way ANOVA with Bonferroni posttest comparison using Prism 3.0 software. The absolute magnitudes of maximal IL-1β release as quantified by ELISA could vary by ∼2-fold between different BMDC preps. Thus, for some experimental series, IL-1β release measured in different conditions was normalized to the maximal release for the particular BMDC preparation; the normalized values from several identical experiments were then used to generate the mean (±SE) for evaluation by ANOVA.

Fig. 1A illustrates the ability of several proapoptotic or chemotherapeutic drugs (12-h stimulation) to induce IL-1β processing and release in LPS-primed BMDC. Robust accumulation of extracellular IL-1β (10-20 ng/ml over 12 h) was elicited in response to the following: 1) Dox, an anthracycline inhibitor of topoisomerase II; 2) STS, a broad-specificity kinase inhibitor commonly used to induce intrinsic apoptosis (26, 27); and 3) UCN-01, a 7-OH STS analog and experimental therapeutic for hematopoietic and solid tumors (28, 29). The IL-1β release was comparable in magnitude to that elicited by ATP activation of P2X7R for 30 min. In contrast, oxaliplatin and cisplatin, two widely-used platinum-based chemotherapeutic agents, did not stimulate significant IL-1β production. The ability of Dox, STS, or UCN-01, but not the platinum reagents, to trigger IL-1β release correlated with the relative efficacies of the drugs to induce apoptosis of LPS-primed BMDC, as indicated by intracellular accumulation of active caspase-3/7 (Fig. 1B). These results are consistent with the studies of Sauter et al. (14) and Shimada et al. (4), who reported the abilities of Dox and STS, respectively, to stimulate a NLRP3- and caspase-1–dependent IL-1β–processing response in LPS-primed murine macrophages. It is important to note that LPS was present throughout the duration of drug stimulation to facilitate sustained TLR4 signaling. We characterized the kinetics (Fig. 1C) and concentration–response relationship (Fig. 1D) for Dox-stimulated IL-1β release. After a lag phase of 4 h, 10 μM Dox induced a progressive increase in IL-1β production over the subsequent 8 h with a plateau at times ≥12 h. Increased IL-1β accumulation (at 12 h) occurred over a narrow range of Dox with a threshold at 1 μM and maximal effect at 10 μM; concentrations >10 μM resulted in an attenuated response due likely to a more rapid loss of BMDC viability. Although 1 μg/ml LPS was routinely used for TLR4 activation, equivalent Dox-induced IL-1β release was observed in BMDC primed with 10 or 100 ng/ml LPS, whereas 1 ng/ml LPS was subthreshold for supporting the response to Dox (Supplemental Fig. 1A). Western blot analysis (Fig. 1E) of BMDC stimulated with 10 μM Dox revealed the following: 1) the time-dependent accumulation of extracellular IL-1β assayed by ELISA reflected the release of mature (17-kDa) IL-1β rather than unprocessed (33-kDa) proIL-1β; 2) the release of mature IL-1β was preceded by the processing and release of the p10 subunit of active caspase-1; and 3) the proteolytic maturation of IL-1β was temporally correlated with the proteolytic processing and release of the p20 subunit of caspase-7, another defined substrate for caspase-1 in myeloid lineage leukocytes.

Given the strong temporal association of the accumulation of mature IL-1β and active caspase-1 p10 subunit in the Dox-treated BMDC (Fig. 1E), we hypothesized that the IL-1β processing reflected Dox-induced activation of the NLRP3/caspase-1 inflammasome pathway. This was tested by directly comparing (in the same experiments) the kinetics and magnitudes of Dox-induced IL-1β release in LPS-primed WT versus Casp1/11^−/− BMDC. To our surprise, we observed that the rate and magnitude of mature IL-1β production were only modestly delayed (by ∼2 h) and decreased (by ∼20%) in the Casp1/11^−/− BMDC as monitored by either Western blot (Fig. 2A) or ELISA (Fig. 2B). Efficient ablation of caspase-1 signaling in the knockout cells was confirmed by the following: 1) the absence of p45 procaspase-1 and the p10 subunit of active caspase-1 in the cell lysates and extracellular medium; and 2) suppression of P2X7R-stimulated IL-1β maturation and release. As with WT cells, 1 ng/ml LPS was subthreshold for supporting Dox-induced IL-1β release in the Casp1/11^−/− BMDC, and equivalent secretion was observed when 10 μM Dox was combined with 10, 100, or 1000 ng/ml LPS (Supplemental Fig. 1B). The observation that WT, Asc^−/−, and Nlrp3^−/− Nlrc4^−/− BMDC (all LPS primed) exhibited similar magnitudes of IL-1β release in response to 12-h stimulation with 10 μM Dox further indicated the presence of an alternative inflammasome-independent pathway for IL-1β processing (Fig. 2C). As with the Casp1/11^−/− cells (Fig. 2B, inset), the Asc^−/− or Nlrp3^−/− Nlrc4^−/− BMDC exhibited a reduced IL-1β release at 8 h of Dox stimulation, but the differences did not reach statistical significance at p < 0.05. In contrast, ATP-stimulated IL-1β production was completely suppressed in the same Asc^−/− and Nlrp3^−/− Nlrc4^−/− BMDC preparations.

Production of mature IL-1β in response to Dox was also observed in WT LPS-primed BMDC treated with YVAD, an inhibitor of caspase-1 activity (Fig. 2D, 2E). YVAD only modestly reduced (by ∼15%) the IL-1β release stimulated by maximally effective (10 μM) Dox while producing a ∼50% suppression of the response to submaximal (3 μM) Dox and complete inhibition of P2X7R-induced IL-1β release. Inclusion of YVAD during Dox treatment of Casp1/11^−/− BMDC did not reduce IL-1β maturation. Using a murine bone marrow–derived macrophage (BMDM) experimental model, Sauter et al. (14) reported that the ability of shorter-term Dox exposure (8 h) to stimulate IL-1β processing was almost completely suppressed in macrophages isolated from Asc^−/−, Nlrp3^−/−, or Casp1^−/− mice. We found that more prolonged (12-h) Dox treatment also induced caspase-1–independent IL-1β release in murine BMDM (Supplemental Fig. 2A). We similarly observed that LPS plus Dox elicited IL-1β release in the RAW264.1 murine macrophage cell line (Supplemental Fig. 2B) that lacks expression of functional ASC (30, 31) and cannot process IL-1β in response to canonical inflammasome agonists such as ATP. These data suggest that intracellular signals generated during the initial phases of Dox-induced apoptosis in DCs and macrophages elicit modest assembly of NLRP3/ASC/caspase-1 inflammasomes, but that this canonical pathway is superseded by a caspase-1–independent pathway for IL-1β processing driven by signals that develop with sustained apoptotic stress.

As indicated in Fig. 1B, Dox treatment induces peak accumulation of apoptotic executioner caspase-3/7 activity within 4–8 h, which precedes the major phase of IL-1β processing and release. We tested whether active caspase-3/7 comprises a necessary upstream signal for the caspase-1–independent IL-1β processing by including the caspase-3/7 inhibitor DEVD during LPS plus Dox stimulation. DEVD had no inhibitory effect on LPS plus Dox–stimulated IL-1β release in WT or Casp1/11^−/− BMDC (Fig. 2F), but completely suppressed the increased caspase-3/7 activity stimulated by Dox in parallel BMDC samples (Fig. 2G). Thus, Dox-induced accumulation of active apoptotic executioner caspases can be dissociated from its stimulatory action on noncanonical IL-1β processing and release.

We hypothesized that caspase-8 was the alternative IL-1β–processing enzyme activated in Dox-treated BMDC given its demonstrated ability to act as an IL-1β convertase in other models of macrophage innate immune response (15–18). However, testing this by utilizing the pan-caspase inhibitor, zVAD, or the caspase-8 selective inhibitor, IETD, in LPS-primed BMDC is complicated by the risk of unleashing RIP1/RIP3-dependent necroptotic death downstream of TLR4 activation (32, 33). Consistent with the induction of necroptosis under such conditions, we observed a time-dependent accumulation of extracellular proIL-1β in BMDC within 4 h after initiation of the cotreatment with LPS and zVAD (Fig. 3A). The ability of zVAD to completely inhibit the caspase-1–mediated IL-1β–processing response to P2X7R activation verified the efficacy of zVAD as a pan-caspase inhibitor in these experiments. The LPS plus zVAD–induced release of proIL-1β was completely suppressed by the RIP1 inhibitor, Nec-1, consistent with a role for necroptosis in the release of extracellular proIL-1β. Similarly, combined LPS plus IETD treatment of WT BMDC, either with or without Dox stimulation (12 h), resulted in the accumulation of extracellular proIL-1β (Fig. 3B). In contrast, IETD completely suppressed the accumulation of extracellular mature IL-1β triggered by LPS plus Dox. The induction of necroptosis in BMDC cotreated with LPS plus zVAD or LPS plus IETD (versus LPS alone) was also verified by observing decreased cell viability (as indicated by total intracellular ATP content) and the ability of Nec-1 to prevent this loss of viability (Fig. 3C).

We also assessed the roles of caspase-8 in the LPS plus zVAD–induced death response in WT, Casp8^−/−Rip3^−/−, and Casp8^+/+Rip3^−/− BMDC. Casp8^−/− mice undergo midgestational death due to unrestrained necroptosis, but this phenotype is reversed in mice that additionally lack RIP3 (Casp8^−/−Rip3^−/−) (24, 34). Consistent with this phenotype, the decrease in BMDC viability induced by combined LPS plus zVAD was absent in Casp8^−/−Rip3^−/− or Casp8^+/+Rip3^−/− BMDC (Fig. 3D). The ability of Dox treatment alone to elicit similar decreases in BMDC viability, independent of caspase-8 and/or RIP3, was consistent with initiation of the intrinsic apoptotic cascade with likely engagement of caspase-9 as the initiator caspase. Kinetic analysis revealed that Dox alone triggered a rapid decrease (50% within 4 h) in the viability of WT or Casp1/11^−/− BMDC (Fig. 3E) that was suppressed in LPS-primed cells, consistent with a protective effect of NF-κB–dependent antiapoptotic gene expression. However, sustained (>8-h) cotreatment with LPS plus Dox resulted in decreased viability, indicating engagement of an alternative regulated cell death pathway. The similar kinetics and magnitude of this response to LPS plus Dox in WT and Casp1/11^−/− BMDC argued against involvement of pyroptosis. In contrast, the rescued viability in Casp8^−/−Rip3^−/− cells, but not in WT or Casp8^+/+Rip3^−/− cells, indicated that TLR4 activation coupled with Dox treatment mediated caspase-8–dependent apoptosis (Fig. 3D).

We also used the Casp8^−/−Rip3^−/− and Casp8^+/+Rip3^−/− BMDC to assess the role of caspase-8 as the alternative IL-1β–processing enzyme stimulated by LPS plus Dox. We first determined how canonical NLRP3/caspase-1 inflammasome signaling pathways are operative in the knockout BMDC. LPS triggered rapid (within 1 h) and quantitatively equivalent accumulation of IL-1β mRNA (and TNF-α mRNA; data not shown) in WT (Casp8^+/+Rip3^+/+), Casp8^−/−Rip3^−/−, and Casp8^+/+Rip3^−/− BMDC (Fig. 4A). LPS priming for 8 h induced similar accumulation of proIL-1β protein in the three cell types (Fig. 4B). Inclusion of nigericin (Fig. 4B) or ATP (Fig. 4C) as canonical NLRP3 stimuli during the final 30 min of the 8-h LPS treatment periods also resulted in robust release of mature IL-1β in each BMDC genotype, albeit at modestly lower levels in the Casp8^−/−Rip3^−/− cells. When the total LPS treatment time was extended to 16 h (Fig. 4D) or 18 h (Fig. 4H), we observed that the proIL-1β protein levels were markedly reduced in Casp8^−/−Rip3^−/− BMDC relative to those in WT or RIP3-ko cells at these time points, or in Casp8^−/−Rip3^−/− cells at 8 h post-LPS. In contrast, LPS stimulation for 16 h induced equivalent TNF-α secretion in the three BMDC genotypes (Fig. 4G). Thus, the combined absence of caspase-8 and RIP3 apparently results in signals (or loss of sustaining signals) that result in gradual attenuation of ongoing TLR4-driven transcription and translation of proIL-1β. Consistent with this model, inclusion of nigericin or ATP during the final 30 min of 16-h LPS treatment periods resulted in markedly reduced levels of mature IL-1β production in Casp8^−/−Rip3^−/− BMDC, as assayed by Western blot (Fig. 4D) and ELISA (Fig. 4F). Notably, the ability of nigericin to elicit production and release of caspase-1 p10 subunit by Casp8^−/−Rip3^−/− BMDC was suppressed when these cells were treated with LPS for 16 h (Fig. 4D), but not for 8 h (Fig. 4B); the former effect was correlated with the reduced ability to sustain elevated levels of TLR4-dependent NLRP3 protein expression (Fig. 4H).

These analyses of canonical NLRP3 inflammasome signaling provide an important context for comparison and interpretation of the noncanonical IL-1β release responses to LPS plus Dox in WT, Casp8^−/−Rip3^−/−, and Casp8^+/+Rip3^−/− BMDC. IL-1β release was completely suppressed in Casp8^−/−Rip3^−/− BMDC compared with WT BMDC treated with LPS and Dox, as assayed by ELISA (Fig. 4E) and Western blot analysis (Fig. 4D). The LPS plus Dox–induced IL-1β release was attenuated by ∼50–60% in Casp8^+/+Rip3^−/− cells, suggesting a possible modulatory role for RIP3 in the response to Dox. It is important to note that the Dox treatment in these experiments was initiated after 4 h of LPS priming and maintained during the next 12 h of LPS stimulation. Despite the gradual decrease in TLR4-dependent proIL-1β protein expression in Casp8^−/−Rip3^−/− BMDC, the cells contained high levels of this cytokine precursor during the initial several hours of Dox treatment. Thus, the complete inhibition of Dox-induced mature IL-1β release in these cells most likely reflects the absence of caspase-8 as a major noncanonical IL-1β–converting enzyme rather than reduced levels of the proIL-1β substrate per se.

Consistent with a role for caspase-8 as an IL-1β–converting enzyme, procapase-8 processing and accumulation of mature extracellular caspase-8 were detected in WT and Casp1/11^−/− BMDC treated with LPS plus Dox for >4 h (Fig. 4J). Interestingly, the clearance of procaspase-8 induced by Dox occurred more rapidly in Casp1/11^−/− BMDC compared with WT BMDC. With recognition of the caveats in using pharmacological inhibitors of caspase-8 to assess its role in IL-1β processing in LPS-primed BMDC, we observed that the ability of LPS plus Dox to stimulate extracellular accumulation of ELISA-measurable IL-1β was greatly suppressed by IETD in both WT and Casp1/11^−/− BMDC (Fig. 4I). The RIP1 inhibitor Nec-1 also suppressed these IL-1β release responses. These results implicate caspase-8 signaling platforms, possibly in RIP1-containing ripoptosomes, as key mediators of LPS plus Dox–stimulated noncanonical IL-1β processing.

Maelfait et al. (16) demonstrated that the ability of TLR4 and TLR3 activation (combined with cycloheximide-mediated suppression of protein turnover) to induce caspase-8 maturation of IL-1β was compromised in macrophages from TRIF-knockout mice. Kaiser and Offermann (35) found that TRIF possesses a RIP homotypic interaction motif domain that facilitates association with RIP1 (and RIP3). Thus, activation of TLR4 in macrophages and DCs results in NF-κB–dependent upregulation of proIL-1β, via both the MyD88 and TRIF adapters, as well as the assembly of latent RIP1/FADD/caspase-8 ripoptosome complexes, orchestrated by the TRIF adapter (Fig. 5A). Because TLR4 activation drives the NF-κB–mediated upregulation of NLRP3 expression (Fig. 4H) (36), LPS-treated cells will also be primed for induction of canonical caspase-1–mediated IL-1β maturation in the presence of appropriate signal 2 stress stimuli. In contrast, whereas TLR2 activation can also trigger the NF-κB–dependent upregulation of proIL-1β and NLRP3 that facilitates utilization of caspase-1 as an IL-1β–converting enzyme, this pathway will not engage TRIF-dependent signaling cascades, such as those leading to possible assembly of RIP1/FADD/caspase-8 ripoptosomes (Fig. 5A). To test the differential contributions of MyD88 versus TRIF signaling to LPS plus Dox–induced IL-1β processing, WT and Trif^−/− BMDC were primed either with the TLR4 agonist, LPS, or the lipopeptide TLR2 agonist, Pam₃CSK₄, for 4 h before stimulation with Dox for 12 h. LPS plus Dox–induced IL-1β maturation and release as assayed by Western blot (Fig. 5B) and ELISA (Fig. 5C, 5D) were significantly reduced in Trif^−/− BMDC. Prolonged LPS treatment resulted in lower levels of proIL-1β in the Trif^−/− cells (Fig 5B), but this partial reduction in proIL-1β substrate contrasted with the near-complete suppression of IL-β maturation in response to Dox. Inclusion of YVAD did not further suppress the residual LPS plus Dox–triggered IL-1β release observed in Trif^−/− BMDC (Fig. 5D). Although Pam₃CSK₄ and LPS induced equivalent accumulation of proIL-1β in WT cells (Fig. 5B), combined stimulation by Pam₃CSK₄ plus Dox resulted in ∼4-fold less production of mature IL-1β compared with LPS plus Dox treatment (Fig 5C). Regardless, the magnitude of Pam₃CSK₄ plus Dox–induced IL-1β release was similar in Trif^−/− and WT BMDC. These data support a role for TRIF in mediating a TLR4-dependent activation of caspase-8 and consequent IL-1β maturation when combined with Dox.

We next explored the mechanism by which Dox may license the activation of caspase-8 within the presumed signaling platforms assembled in response to the TLR4–TRIF cascade. Tenev et al. (37) reported that genotoxic stressors such as etoposide (a topoisomerase II inhibitor like Dox) facilitate ripoptosome assembly and caspase-8–mediated cell death in nonmyeloid tumor cells by inducing the degradation of cIAPs. Furthermore, Vince et al. (17) showed that Smac mimetics, which induce degradation of cIAPs, activated IL-1β processing in LPS-primed BMDC via a caspase-8 pathway that required coordinate downregulation/inhibition of all three major inhibitor of apoptosis proteins (IAPs), including cIAP1, cIAP2, and XIAP. We assessed the levels of cIAP1 in WT and Casp1/11^−/− BMDC treated (or not) with LPS only, Dox only, or LPS plus Dox. The amount of cIAP1 was greatly reduced in LPS plus Dox–treated BMDC and attenuated in the LPS-only or Dox-only treatment conditions compared with untreated cells (Fig. 5E). As a positive control, we showed that Dox treatment reduced cIAP1 levels in Jurkat leukemic T cells.

We initially assessed whether LPS plus Dox stimulated the formation of RIP1/FADD/caspase-8 ripoptosome complexes in BMDC by adapting coimmunoprecipitation protocols developed for the characterization of such platforms in tumor cells or nonmyeloid cell types challenged with proapoptotic stimuli, TLR3 agonist, or TNF-α (37–39). WT BMDC were treated with various combinations of LPS, Dox, and zVAD, or TNFα ± zVAD, prior to generation of whole lysates by standard detergent extraction, removal of detergent-insoluble material by centrifugation, incubation of the detergent-soluble lysates with anti-FADD, collection of anti-FADD immune complexes, and Western blot analysis of the immune complexes for caspase-8, FADD, and RIP1. ZVAD is commonly used in such analyses to stabilize caspase-8–containing signaling complexes (37–39). No caspase-8 or RIP1 was observed in FADD immunoprecipitates prepared from cells treated with any combination of those stimuli. However, we consistently found that the detergent-soluble faction (i.e., the fraction subjected to subsequent immunoprecipitation) of lysates from BMDC treated for 2 h with LPS plus zVAD or TNF-α plus zVAD contained markedly lower levels of procaspase-8 and FADD (Fig. 5F). To our surprise, analysis of the detergent-insoluble lysate pellets from cells stimulated with LPS plus zVAD or TNF-α plus zVAD revealed robust enrichment of 57-kDa procaspase-8 and 45-kDa partial cleavage fragment of caspase-8, as well as lower amounts of 18-kDa processed caspase-8 subunit; FADD was also enriched in the same detergent-insoluble fractions. This redistribution of caspase-8 and FADD was not observed in lysates from cells treated with only zVAD (data not shown) or in BMDC that were primed with LPS for 4 h, followed by treatment with Dox for an additional 8 h (Fig. 5F) or 3 h (data not shown). However, inclusion of zVAD during the Dox treatment did facilitate caspase-8 and FADD partitioning into the insoluble lysate pellets (Fig. 5F). Analysis of RIP1 recruitment to these pellets was limited by the finding that total RIP1 levels in BMDC (but not BMDM or J774.1 murine macrophages) were below the detection limit of the Ab used (Fig. 6E). These data indicate that TLR4 activation in BMDC induces the recruitment of caspase-8 and FADD to signaling pltaforms that may underlie noncanonical IL-1β processing when combined with proapoptotic stimuli that downregulate IAPs.

STS has been studied extensively with regard to its ability to induce a very rapid apoptotic response in most cell types via incompletely understood mechanisms that include suppression of multiple prosurvival kinases (40, 41), disruption of mitochondrial integrity (4), and allosteric modulation of caspase-9 independently of Apaf-dependent apoptosome assembly (42). Shimada et al. (4) demonstrated that STS also stimulates a rapidly developing NLRP3- and caspase-1–dependent IL-1β processing response in LPS-primed murine macrophages via a mechanism linked to disruption of mitochondrial integrity. Fig. 6A shows that STS triggers a similarly rapid (near maximal within 4 h) increase in IL-1β processing and release in LPS-primed WT BMDC. Notably, STS also induced a robust accumulation of mature extracellular IL-1β in Casp1/11^−/− BMDC. The STS-stimulated processing and release of IL-1β in both WT and Casp1/11^−/− cells were temporally correlated with production and release of the p18 subunit of active caspase-8 (Fig. 6A). As with LPS plus Dox treatment, LPS plus STS–induced clearance of procaspase-8 occurred more rapidly in Casp1/11^−/− BMDC compared with WT BMDC even though mature 18-kDa caspase-8 accumulation appeared to be equivalent. ELISA measurements indicated that STS-induced IL-1β release (at 4 h) was nonsignificantly attenuated by ∼25% in either Casp1/11^−/− BMDC (Fig. 6B) or WT BMDC treated with YVAD (data not shown). STS-induced IL-1β release from either WT (Fig. 6C) or Casp1/11^−/− (data not shown) BMDC was markedly suppressed by IETD and partially (∼50%) decreased by Nec-1 (Fig. 6C). A 75% reduction in STS-triggered IL-1β secretion was also observed in Casp8^−/−Rip3^−/− BMDC (Fig. 6D). Similarly to treatment with 12 h of Dox, a 4-h exposure to STS suppressed the expression of cIAP1 in WT BMDC (Fig. 6E, upper panel). In contrast, 18 h of oxaliplatin induced no change in cIAP1 expression relative to untreated BMDC. Additionally, a 4-h treatment with STS was sufficient to produce marked depletion of cIAP1 levels in other murine myeloid cell types, including BMDC, BMDM, and the J774.1 murine macrophage line (Fig. 6E, lower panel). Rapid STS-triggered depletion of cIAP1 in these cells was correlated with apoptotic induction, as indicated by depletion of intact 116-kDa PARP and accumulation of 89-kDa PARP cleavage product. STS treatment also decreased RIP1 protein levels in these detergent-solubilized cell lysate samples. Despite the complexity of STS action, these data indicate that this apoptosis inducer also triggers caspase-8–dependent IL-1β processing and release in LPS-primed BMDC.

This study describes a novel and robust mechanism of caspase-8–dependent processing of IL-1β induced by the chemotherapeutic drug, Dox, in LPS-primed BMDC. This release did not require caspase-1 based on the fact that mature IL-1β was produced by Casp1/11^−/− BMDC as well as by WT BMDC treated with YVAD (Fig. 2). The proapoptotic agent, STS, triggered caspase-1–independent IL-1β release in LPS-primed BMDC even more rapidly than Dox (Fig. 6). Dox- and STS-induced IL-1β production in LPS-primed WT BMDC was attenuated in the presence of the caspase-8 inhibitor, IETD, or the RIP1 inhibitor, Nec-1, whereas Dox- and STS-induced IL-1β release was greatly suppressed in LPS-primed Casp8^−/−Rip3^−/− BMDC, implicating a caspase-8 signaling platform for IL-1β processing in TLR4-activated DCs (Figs. 4, 6). The marked attenuation of LPS plus Dox–induced IL-1β release from Trif^−/− BMDC relative to WT cells supported an important role for TRIF in coupling TLR4 stimulation to caspase-8 activation and noncanonical IL-1β processing (Fig. 5). In the presence of a pan-caspase inhibitor, TLR4 activation induced recruitment of caspase-8 and FADD to a detergent-insoluble compartment (Fig. 5) and also triggered RIP1-dependent necroptotic death (Fig. 3). The ability of both Dox and STS to trigger IL-1β release in LPS-primed BMDC was correlated with the degradation of cIAP1 (Figs. 5, 6), which, together with cIAP2 and XIAP, is known to restrain ripoptosome activity by destabilizing such complexes. Taken together, these results support a model wherein TLR4 drives TRIF-dependent assembly of caspase-8 ripoptosomes that are licensed as IL-1β–converting enzyme platforms in the presence of proapoptotic stimuli, most likely via modulated expression of IAPs and/or other factors (e.g., cellular FLICE inhibitory protein [cFLIP]) that control ripoptosome assembly, stability, and activity. However, a rapidly expanding literature is identifying roles for caspase-8 as a component of multiple signaling complexes or pathways linked to innate immune response (15, 18, 19, 43, 44). Thus, ripoptosomes may comprise only one of several caspase-8–based pathways that participate in noncanonical IL-1β processing.

Consistent with the reported role for the NLRP3 inflammasome in LPS plus Dox–induced IL-1β production by bone marrow–derived macrophages (14), we observed reduced IL-1β release at the early (≤8-h) stages of LPS plus Dox stimulation in BMDC lacking caspase-1 (Fig. 2A, 2B), ASC (Fig. 2C), or NLRP3 (Fig. 2C). However, when Dox stimulation was sustained, the relative contribution of this NLRP3/caspase-1 cascade to IL-1β accumulation was superseded by an alternative pathway that was attenuated in the absence of caspase-8 or TRIF (Figs. 2, 4, 5). Robust proteolytic maturation of caspase-1 per se was invariably observed in WT BMDC treated with LPS plus Dox (Figs. 1, 2, 4) or LPS plus STS (Fig. 6), despite the fact that caspase-1 was dispensable for maximal IL-1β processing. We initially expected that the activation of caspase-1 and caspase-8 in response to Dox or STS comprised independent IL-1β–processing pathways operating in parallel. However, our observations and recent studies by others suggest a linked, but asymmetric relationship in the ability of proapoptotic stimuli to activate the caspase-1 versus caspase-8 pathways coupled to IL-1β maturation and release. Although the caspase-8 pathway was strongly activated by LPS plus Dox in BMDC that lack caspase-1, caspase-11, ASC, or NLRP3 (Fig. 2), activation of caspase-1– and caspase-1–mediated IL-1β processing was greatly reduced in Casp8^−/−Rip3^−/− BMDC (Fig. 4) but only modestly attenuated in cells lacking only RIP3 (Fig. 4D, 4E). This was mediated in part by the reduced ability of the Casp8^−/−Rip3^−/− cells to sustain high expression of proIL-1β and NLRP3 protein during prolonged TLR4 activation (Fig. 4D, 4H). Notably, the ability of ATP or nigericin, as rapidly acting signal 2 stimuli for NLRP3 inflammasome assembly, to trigger caspase-1 processing (Fig. 4B) and IL-1β release (Fig. 4B, 4C) was normal in Casp8^−/−Rip3^−/− BMDC during the earlier (<8-h) stages of TLR4 activation, but was markedly reduced when the Casp8^−/−Rip3^−/− cells were primed with LPS for longer durations (>12 h) before being challenged with nigericin or ATP (Fig. 4D, 4F). These data indicate that caspase-8 in DCs and macrophages can act directly as an alternative IL-1β–converting enzyme and indirectly as a modulator of the expression of proIL-1β and NLRP3. Although the underlying mechanism for this latter role of caspase-8 remains to be defined, the dependence on prolonged TLR4 activation suggests feedback regulation by autocrine or paracrine pathways that fine-tune ongoing transcription, translation, and turnover of proIL-1β and NLRP3, and possibly other proinflammatory gene products.

Other investigators have recently described complex roles for caspase-8 in the direct and indirect regulation of IL-1β production (17, 45). Kang et al. (45) observed that accumulation of active RIP3 and RIP1 in BMDC isolated from mice with a DC-restricted deletion of Casp8 was sufficient to license rapid assembly of NLRP3 inflammasomes in response to TLR4 activation in the absence of a signal 2 stimulus, such as ATP or nigericin. This ability of caspase-8 deletion to potentiate TLR4 induction of NLRP3 inflammasomes was independent of significant RIP1/RIP3-induced cell death. Kang et al. (45) additionally noted that caspase-8 and FADD were coimmunoprecipitated with NLRP3 in lysates from LPS-stimulated WT BMDC. We found that caspase-8 and FADD were recruited to a detergent-insoluble compartment in DCs treated with LPS and zVAD (Fig. 5F). Vince et al. (17) characterized additional interactions between caspase-8, RIP3, and NLRP3 signaling in their model of Smac-mimetic–induced IL-1β accumulation by murine BMDM and BMDC. Activation of RIP3 was correlated with production of ROS that stimulated IL-1β processing via both caspase-1 and caspase-8. Those investigators also reported recruitment of caspase-8 and RIP1 to detergent-insoluble fraction in lysates of the Smac-mimetic–stimulated BMDM. Another study found that caspase-8 interacted with ASC and colocalized with AIM2/ASC specks during cell death triggered by Francisella tularensis infection of caspase-1–deficient BMDM (44). Finally, Sagulenko et al. (43) have described a direct interaction between the death effector domain of procaspase-8 and the pyrin doman of ASC that can occur in the context of NLRP3- or AIM2-inflammasome activation in macrophages.

These various findings underscore complex interactions between caspase-8 and proteins/pathways associated with canonical caspase-1 inflammasome assembly and activity. Our data suggest that proapoptotic agents, in conjunction with TLR4 activation, elicit a temporally defined hierarchy of reactions that use various combinations of TRIF, FADD, RIP1/3, and conventional inflammasome proteins for the assembly of caspase-8 and caspase-1 platforms that catalyze IL-1β processing. Although both Dox and STS induced proteolytic processing of both caspase-1 and caspase-8 (Figs. 4J, 6A), STS stimulated these responses much more rapidly with significant activation of caspase-1 activation preceding that of caspase-8. This more rapid engagement of the caspase-1 pathway may underlie the lower efficacy of IETD or caspase-8 deletion in attenuating STS-induced IL-1β processing (Fig. 6C, 6D) relative to Dox-induced IL-1β production (Fig. 4E, 4I). The relative rate at which a particular “signal 2” stimulus triggers activation of the NLRP3/caspase-1 pathway most likely determines whether the more slowly developing caspase-8 pathway will, or will not, comprise a major route for IL-1β processing. ATP acting directly via P2X7 nonselective cation channel receptors, and nigericin, acting as a direct K⁺ ionophore, trigger dissipation of the normal trans-plasma membrane K⁺ gradient within minutes, and the resulting decrease in cytosolic [K⁺] elicits exceptionally rapid (complete within 30 min) and efficient assembly of NLRP3 inflammasomes (Figs. 1E, 2D, 4B, 4D, 4J) (7). Thus, very robust processing of proIL-1β occurs prior to any substantial activation of caspase-8. STS also triggers K⁺ efflux-dependent NLRP3 activation (4), but the K⁺ efflux most likely occurs as a secondary and more slowly induced consequence of mitochondrial dysfunction and reduced ATP generation. As a result, STS-induced activation of caspase-1 occurs only modestly faster than induction of the caspase-8 pathway (Fig. 6A). Finally, because changes in mitochondrial function (and apoptotic signaling) are even more slowly induced by Dox, the activation of both the caspase-1 and caspase-8 pathways develops over similar time periods (Fig. 4J).

We observed a strong correlation between the relative efficacies of proapoptotic agents to induce IL-1β production and their abilities to stimulate apoptotic executioner caspase-3/7 activity in DCs (Fig. 1A, 1B). Perturbation of mitochondrial function comprises one possible link by which proapoptotic agents can engage both the caspase-1 and caspase-8 pathways for IL-1β production. STS elicits exceptionally rapid changes in mitochondrial function and apoptotic progression in macrophages (4) and most other cell types by incompletely understood mechanisms. Given its ability to inhibit topoisomerase II, Dox can induce conventional DNA damage-dependent apoptosis in tumor cells and other rapidly dividing cells (46). In quiescent or slowly proliferating cells (e.g., cardiomyocytes), Dox also triggers apoptosis via activation of iron-dependent redox-cycling reactions in mitochondria that result in accumulation of ROS (46). Accumulation of ROS together with release of mitochondrial DNA has been linked to assembly of the NLRP3 inflammasome pathway for IL-1β production (3). However, another hallmark of the intrinsic apoptotic pathway is the release of Smac from dysfunctional mitochondria. Binding of Smac to IAPs either directly inhibits their ability to bind to caspases, as with XIAP, or induces their autoubiquitination and proteasome-mediated clearance as with the cIAP1/cIAP2 E3 ubiquitin ligases (47). The cIAPs per se target proapoptotic proteins and components of the ripoptosome, including RIP1 and RIP3, for ubiquitin-mediated degradation (Fig. 5A). Genotoxic stressors, such as etoposide, or the degradation of cIAPs using Smac mimetics, promote ripoptosome assembly (17, 37). We found that Dox and STS, but not oxaliplatin, induced marked downregulation of cIAP1 levels in BMDC (Figs. 5E, 6E); this most likely comprises one mechanism by which these agents trigger caspase-8 activation when combined with TLR4 stimulation. A limitation of our study is that we assayed only changes in cIAP1 levels. Vince et al. (17) found that the ability of Smac mimetics to maximally induce IL-1β production by LPS-primed BMDM or BMDC required coordinated targeting of cIAP2 and XIAP, in addition to cIAP1. Thus, it will be important to assess the effects of Dox and other proapoptotic agents on the expression/activity of cIAP2 and XIAP in future studies of noncanonical IL-1β production. In addition to IAPs, other E3 ubiquitin ligases may contribute to caspase-8 regulation in our model. Jin et al. (48) identified a critical role for cullin-3 in mediating the polyubiquitination, aggregation, and full activation of caspase-8 during TRAIL-mediated apoptosis of cancer cells. Notably, those investigators also observed the recruitment of caspase-8 and FADD to a detergent-insoluble compartment similar to our findings in LPS plus zVAD–stimulated BMDC (Fig. 5F).

The role of cIAPs in inflammasome assembly and caspase-1 signaling is more complex. Labbé et al. (49) found that cIAP1 and cIAP2 were required for efficient caspase-1 activation and IL-1β processing in response to multiple stimuli for NLRP3 and NLRC4 inflammasome assembly. This reflected a direct interaction of the cIAPs with caspase-1 that results in an activating and nondegradative K63-linked ubiquitination of caspase-1. In contrast, Vince et al. (17) showed that the loss of cIAPs upon stimulation with Smac mimetics promoted RIP3-dependent caspase-1 and caspase-8 activation for IL-1β processing. These nominally conflicting observations are particularly germane given the E3 ubiquitin ligase activity of IAPs and recent findings that the ubiquitination status of NLRP3 per se (9–11) and caspase-1 (49) strongly modulate the assembly and activity of caspase-1 inflammasomes. Our observations that the caspase-8 pathway progressively predominates over the caspase-1 pathway during Dox and STS treatment may reflect the opposing consequences of IAP suppression on these signaling cascades. Progressive loss of the cIAPs may act to bias the signaling network by reversing the stimulatory effect of K63 ubiquitination on caspase-1 while enhancing ripoptosome assembly/activity. Further investigation is required to assess the differential roles of IAPs and other E3 ubiquitin ligases in the engagement of the canonical and noncanonical pathways for IL-1β processing by various cell death and inflammatory stimuli.

The ability of LPS plus Dox to stimulate IL-1β maturation and release was greatly reduced in Trif^−/− BMDC (Fig. 5B–D). The absence of TRIF could reduce the contribution from the TLR4→TRIF→IFN regulatory factor 3/7→IFN-β→caspase-11 signaling cascade that amplifies NLRP3/caspase-1 inflammasome activation (Fig. 5A) (50). However, another major role for TRIF is recruitment of RIP1 via RIP homotypic interaction motif domain interactions with consequent induction of RIP1/FADD/caspase-8 complexes. The assembly, activity, and stability of caspase-8–containing ripoptosomes are additionally modulated by association with isoforms of cFLIP, a noncatalytic paralogue of caspase-8 (20). Maximal ripoptosome-associated caspase-8 activity (sufficient to drive apoptosis) is restrained via association of caspase-8 with long isoform cFLIP. However, the low-level ripoptosome-associated caspase-8 activity of the long isoform cFLIP-containing heterodimeric complex is sufficient to either cleave RIP1 and/or RIP3 directly or eliminate another process that impacts function and thereby suppresses the assembly of the necrosome complexes necessary for necroptosis. When the activity of this heterodimeric complex is completely inhibited, either by pharmacological inhibition of caspase-8 or when the complex contains short isoform cFLIP, the suppression of RIP1 and RIP3 is relieved; RIP1 then phosphorylates RIP3, which regulates an incompletely understood necroptotic signaling cascade (34, 38). Inhibition of RIP1 kinase activity by Nec-1 blocks this cascade. This latter necroptotic pathway was operative in our BMDC experimental model, as indicated by morphology as well as the loss of DC viability in response to LPS stimulation in combination with either pan-caspase inhibition or selective caspase-8 inhibition (Fig. 3A–C). The inhibitory effects of Nec-1 on LPS plus Dox or STS-induced IL-1β release illustrated in Figs. 4I and 6C further suggested that this IL-1β production involves a RIP1-regulated pathway for engagement of caspase-8. The accumulation of extracellular IL-1β was temporally correlated with caspase-8 activation in WT and Casp1/11^−/− BMDC stimulated with Dox (Fig. 4J) or STS (Fig. 6A) and markedly suppressed in Casp8^−/−Rip3^−/− BMDC (Figs. 4D, 4E, 6D).

There are several limitations in our experimental results and their implications regarding the possible involvement of RIP1/FADD/caspase-8 ripoptosomes as the major signaling platform for noncanonical IL-1β processing. The genetic support for caspase-8 as the alternative IL-1β–converting enzyme used cells deficient in both caspase-8 and RIP3. Experiments with cells lacking only caspase-8 would eliminate potentially confounding effects of additional RIP3 ablation on the Dox-stimulated responses. However, unlike Casp8^−/−Rip3^−/− BMDC, cells deficient in only RIP3 were characterized by only modestly attenuated IL-1β processing and release in response to LPS plus Dox (Fig. 4D, 4E) or LPS plus STS (Fig. 6D). This suggests that the absence of caspase-8 rather than RIP3 underlies the very strong suppression of IL-1β production in Casp8^−/−Rip3^−/− cells. Our results also suggest stimulus-specific roles for RIP1 because Nec-1 almost completely suppressed Dox-stimulated IL-β production (Fig. 4I), but only partly attenuated the response to STS (Fig. 6C). However, an additional caveat is that Nec-1 inhibits IDO, a known immunomodulatory enzyme (51). Finally, the isolation of RIP1, FADD, and caspase-8 in coprecipitating immune complexes would provide stronger biochemical support for ripoptosomes as an underlying signaling platform for noncanonical IL-1β processing. Studies of ripoptosome assembly have predominantly used cancer cells or established cell lines that may express higher levels of the interacting signaling proteins (35–37). Moreover, the accumulation of RIP1/FADD/caspase-8 ripotosomes as soluble protein complexes within the cytosol may be an intermediate state superseded by the recruitment or aggregation of such complexes into higher-order macromolecular ensembles. This possibility is supported by our observations (Fig. 5F) and those of Vince et al. (17) demonstrating the recruitment of caspase-8 into a detergent-insoluble compartment of BMDC treated with LPS plus zVAD or BMDM treated with LPS plus Smac mimetic, respectively. Additional studies are necessary to address how RIP1, RIP3, and cFLIP isoforms modulate the ability of various proapoptotic stimuli to induce specific caspase-8 signaling complexes that mediate processing of IL-1β.

The differential viabilities of Casp8^−/−Rip3^−/−, Casp8^+/+Rip3^−/−, and control Casp8^+/+Rip3^+/+ BMDC treated with Dox only, LPS only, or LPS plus Dox indicated that Dox may engage either a conventional caspase-9–mediated intrinsic apoptotic program in the absence of TLR4 costimulation or a caspase-8–mediated apoptosis in the context of TLR4 activation (Fig. 3D). The viability of Casp8^−/−Rip3^−/− BMDC treated with LPS plus Dox, but not Dox only, was sustained, thereby implicating caspase-8–dependent apoptosis as the mode of cell death in this instance. Pyroptosis was eliminated as a major mode of cell death or IL-1β release under LPS plus Dox treatment conditions because similar time courses of reduced viability (Fig. 3E) and extracellular IL-1β accumulation (Fig. 2B) were observed in WT and caspase-1–deficient BMDC. Dox-induced accumulation of active executioner caspase-3/7 activity could also be dissociated from the parallel IL-1β–processing response to this proapoptotic agent (Fig. 2F, 2G). Notably, mature IL-1β rather than proIL-1β was the predominant form in the extracellular medium of cells treated with LPS plus Dox (Fig. 1E). This indicated that the production of mature IL-1β and its release from the DCs were tightly coordinated with the decrease in cell viability orchestrated by apoptotic signaling. In contrast, induction of necroptosis by LPS plus zVAD or LPS plus IETD resulted in a massive release of proIL-1β that was suppressed by Nec-1 (Fig. 3A).

It will be important to determine whether Dox and related anthracyclines can engage this caspase-8–mediated pathway for IL-1β production under in vivo conditions wherein these agents exert their antitumor chemotherapeutic and immunogenic actions. In this regard, two questions are particularly germane. First, might DCs or macrophages within tumor loci be exposed to the micromolar Dox concentrations that trigger this signaling cascade in cell culture conditions? Both the tumor cells and immune cells within tumor sites can be exposed to high local concentrations of chemotherapy drugs depending on the method and site of drug delivery. There is increasing development of new modes of Dox therapy that involve tumor-directed delivery via encapsulation within liposomes or conjugation with nanoparticles and Ab complexes (52, 53). Such particulate-based therapies can result in phagocytosis of Dox-containing liposomes or conjugates by tumor-resident DCs and macrophages. The second question concerns possible in vivo mediators within the tumor microenvironment that might act as TLR4 agonists to facilitate the TRIF-dependent signals that are also required for caspase-8 activation. Several studies have identified high-mobility group protein B1 released from dying tumor cells or other host cells as a relevant danger-associated molecular pattern agonist for TLR4 that supports activation of IL-1β processing and release (12, 54–57). Moreover, advanced stages of cancer progression or cytotoxic cancer therapies can result in compromised barrier function of the gut epithelia and increased circulating levels of pathogen-associated molecular patterns derived from commensal bacteria (58, 59). This study further validates the importance of exploring the direct effects of proapoptotic chemotherapeutic drugs on IL-1β production by tumor-resident immune cells in addition to previously described models wherein other danger-associated molecular patterns, such as ATP, released from dying chemotherapy-treated tumor cells initiate canonical inflammasome signaling (60). Moreover, the overall consequences of the myeloid-driven IL-1β production response to various cancer chemotherapy drugs may vary with the stage of tumor progression (61). For example, IL-1β production elicited in the early stages of cancer therapy may be advantageous to the tumor-bearing host in initiating an IL-1β–dependent immunogenic antitumor response (12), but, at more advanced stages, may contribute to tumor growth (13). Overall, better understanding of the relative contributions of various IL-1β–processing pathways elicited by cancer chemotherapeutics may aid in the development of refined therapeutic approaches that limit tumor progression.

We thank Eric Pearlman and Mausita Karmakar for providing Asc^−/−, Trif^−/−, and Nlrp3^−/− Nlrc4^−/− double-knockout mouse strains.

The authors have no financial conflicts of interest.