Nanoparticles activate the NLR pyrin domain containing 3 (Nlrp3) inflammasome and cause pulmonary inflammation through release of IL-1α and IL-1β

To determine the possible IL-1–stimulatory effect of metallic nanoparticles, the human macrophage cell line THP1, was stimulated with various nanoparticles for 6 h. In contrast to ZnO (15 nm), both TiO₂ (anatase, 20 nm or rutile, 80 nm) and SiO₂ (15 nm) induced caspase-1 cleavage and IL-1β secretion (Fig. 1A). Because the organ primarily exposed to everyday metallic nanoparticles is the skin, we also stimulated primary human keratinocytes with nanoparticles (Fig. 1B). Similar to macrophages, human keratinocytes secreted mature IL-1β in a dose-dependent manner following exposure to nano-TiO₂ and nano-SiO₂ but not carbon nanotubes, another nanoparticle, which is used primarily in electronic and material science.

To examine whether mouse and human cells behaved similarly, mouse bone marrow-derived dendritic cells (BMDC) (Fig. 1C) and bone marrow-derived macrophages (BMDM) (Fig. S1) were treated with nano-TiO₂ and nano-SiO₂. Nano-TiO₂ and nano-SiO₂ were potent stimuli for IL-1β secretion in both murine myeloid cell types, whereas release of inflammasome-independent cytokines such as TNF were minimally affected by nano-TiO₂ stimulation as compared with the TLR9 agonist CpG DNA, which induced TNF secretion without activating inflammasome-dependent IL-1β secretion (Fig. S2).

As particulate danger signals (e.g., asbestos, MSU, and alum) trigger secretion of mature IL-1β via the Nlrp3 inflammasome, we hypothesized that the Nlrp3 inflammasome may also mediate responses to nano-TiO₂ and nano-SiO₂. To test this hypothesis, we treated human and mouse macrophages in which Nlrp3 inflammasome components were knocked down or knocked out, respectively, with nano-TiO₂ or nano-SiO₂. THP1 cells stably infected with shRNA directed against caspase-1 or Nlrp3 (Fig. 1A) or BMDCs from ASC^−/− or Nlrp3^−/− mice (Fig. 1C) did not secrete IL-1β in response to nano-TiO₂ or nano-SiO₂, indicating that the Nlrp3 inflammasome mediated responses to these agonists. In contrast, knockdown or knockout of the related NLR family member, IPAF (Nlrc4) which can also mediate inflammasome assembly, did not impair IL-1β secretion (Fig. 1A, Fig. S1).

The precise mechanism of inflammasome activation remains poorly characterized at present; however, potassium efflux and intracellular ROS have been proposed to be criticial for Nlrp3 inflammasome activation in response to “classic” Nlrp3 agonists. To further characterize the activating capacity of nanoparticles in this context, we examined whether Nlrp3 inflammasome activity was suppressed by the ATP-sensitive potassium channel inhibitor, glibenclamide or the chemical ROS scavenger (2R,4R)-4-aminopyrrolidine-2, 4-dicarboxylate (APDC). As for “classic” Nlrp3 agonists such as MSU, cellular pretreatment with glibenclamide or APDC diminished inflammasome-dependent IL-1β secretion triggered by nanoparticles (Fig. S3A).

Although the above results suggested the current inflammasome activation scheme proposed for other particles, we found an important difference between nano-TiO₂ and known particulate activators such as MSU, alum, hemozoin, and μm-sized SiO₂ (20). Whereas these above-mentioned particulates can activate the Nlrp3 inflammasome in myeloid cells, they do not trigger activation in keratinocytes (Fig. 2 A and B). This is likely to reflect the low capacity of keratinocytes to phagocytose particles. In contrast, small diameter (15 nm) but not large diameter (1.5 μm) SiO₂ could activate the inflammasome in keratinocytes (Fig. 2B), whereas both nano-sized and μm-sized SiO₂ activated the Nlrp3 inflammasome in myeloid cells (Fig. 2A), suggesting that the inflammasome-activating signal was triggered without the need for particle phagocytosis. In support of this, disruption of actin-mediated phagocytosis by cytochalasin D was unable to suppress nanoparticle-induced IL-1β secretion in macrophages (Fig. 2C). Electron microscopy studies further revealed that mainly clustered TiO₂ was located free in the cytoplasm of THP1 cells and not located in phagosomes or lysososmes, as would be expected if they were internalized by phagocytosis (Fig. 2D). To further elucidate other possible mechanisms of nanoparticle uptake, THP1 cells were either treated with methyl-β-cyclodextrin to disrupt lipid raft-mediated endocytosis filipin or nystatin to block the caveolae-dependent endocytic pathway or chlorpromazine hydrochloride to inhibit clathrin-mediated endocytosis but nano-TiO2–mediated inflammasome activation could not be blocked. (Fig. S3 B–D)

Taken together, the above results suggested a proinflammatory activity of nano-TiO₂ and nano-SiO₂ in vitro. We used an established inflammatory peritonitis model (17) to confirm such a role in vivo. In this model, i.p. injection of inflammasome activators drives the influx of neutrophilic granulocytes to the peritoneal cavity. Compared with the solvent control, TiO₂ evoked significant neutrophil recruitment 6 h postinjection (Fig. 3A). The influx of neutrophils was accompanied by significant sercretion of IL-1β and IL-6 into the peritoneal cavity, while cytokine levels for IL-1α and TNF remained unchanged (Fig. S4A). When signaling by both IL-1α and IL-1β was blocked through deficiency in their common receptor, IL-1 receptor (IL-1R), mice exhibited strongly reduced neutrophil influx (Fig. 3B). In strict accordance with impaired neutrophil recruitment, a drastic decrease in IL-6 secretion was detected in IL-1R^−/− mice, suggesting the dependence of IL-6 secretion on IL-1R signaling (Fig. S4B). As the ex vivo findings indicated a critical function for the inflammasome components ASC and Nlrp3 in the response to specific nanoparticles, we also examined TiO₂-dependent neutrophil influx in mice deficient in ASC and Nlrp3. Whereas TiO₂-induced peritonitis appreciably depended on the ASC adapter, surprisingly, Nlrp3 deficiency did not significantly affect TiO₂-induced peritonitis (Fig. 3 C and D).

Due to our observation that IL-1R deficiency more strongly suppressed TiO₂-dependent neutrophil influx compared with ASC or Nlrp3 deficiency, we hypothesized that IL-1α and IL-1β may act in concert to mediate neutrophil influx via their common receptor. Indeed, IL-1α deficiency protected mice from inflammatory cell recruitment (Fig. 3E).

As the effects of IL-1α appeared to dominate over Nlrp3/IL-1β in this in vivo setting, we reexamined the effect of Nlrp3 deficiency following IL-1α depletion. Following injection of a neutralizing anti-IL–1α antibody, Nlrp3 deficiency significantly suppressed neutrophil influx compared with WT, indicating that both IL-1α and IL-1β can mediate TiO₂-induced peritonitis, although the effect of IL-1α predominates (Fig. 3F).

As the peritonitis model suggested a major role for IL-1α in TiO₂-dependent inflammation in vivo, we next investigated whether nanoparticles regulated the secretion of IL-1α by murine dendritic cells (DCs) in vitro. Whereas release of IL-1β by MSU or nano-TiO₂ was completely dependent on ASC and Nlrp3 as expected (Fig. 3G), IL-1α release displayed only a partial dependency on ASC and Nlrp3 (Fig. 3H). The partial ASC dependency of IL-1α release was also apparent in murine keratinocytes (Fig. 3I).

Data from the peritonitis model and the in vitro data provided proof of principle that TiO₂ can induce IL-1–dependent inflammatory responses. However, the two most common routes of nano-TiO₂ exposure to humans are via the skin (e.g., application of nano-TiO₂–containing cosmetics) and the lungs (e.g., dust inhalation). To more closely mimic a physiological route of cutaneous exposure, we treated shaved back skin of mice with either a single application of TiO₂ or PBS dissolved in olive oil or for chronic exposure once daily over a period of 5–10 consecutive days. In agreement with previous reports suggesting that TiO₂ nanoparticles do not penetrate into living skin (22, 23), we found that application of the nanoparticle emulsion or its vehicle control did not lead to clinically visible inflammation. Histologically, 8/11 (72%) mice treated with one single application of TiO₂ displayed acanthosis (epidermal thickening) without any inflammatory infiltrate, whereas the PBS-treated mice displayed epidermal alterations in 2/11 (18%) cases. Acanthosis without inflammatory infiltrate might be due to mechanical irritation and could be caused by the mice scratching or licking away the TiO₂. Upon chronic exposure, the incidence of epidermal thickening did not further increase and still no inflammatory infiltrates were detectable.

We next investigated the effect of pulmonary exposure to nano-TiO₂. A single dose of intranasally administered TiO₂ induced an inflammatory response leading to neutrophil recruitment in the bronchoalveolar lavage fluid (BALF), and IL-1R or caspase-1 deficiency significantly blocked neutrophil influx (Fig. 4A). Similar to the observations in the peritonitis model, neutrophil influx depended significantly on the presence of the adapter protein ASC, but exhibited only a partial dependence on Nlrp3, which did not reach significance (Fig. 4 B and C). As for the peritonitis model, neutrophil recruitment was highly dependent on IL-1α (Fig. 4C), and so the effects of IL-1α in this model may mask the potential contribution of IL-1β/Nlrp3. In accordance with our observation in the peritonitis model, only a minor secretion of IL-1α and TNF into the BALF could be observed, whereas TiO₂ caused a prominent increase of IL-1β and IL-6 (Fig. 4D). Regarding total lung tissue, an induction of both IL-1α and IL-1β could be observed after the administration of TiO₂, reflecting an increased synthesis of the two cytokines.

C57BL/6J, Nlrp3^−/− (17), Asc^−/−, Ipaf^−/− (26), IL-1 α (27), IL-1R^−/−, and caspase-1^−/− (25) mice (on C57BL/6J background) and BALB/c and IL-1R^−/− (28) mice (on BALB/c background) were housed at the animal facility at the University of Lausanne or the Transgenose Institute (Centre National de la Recherche Scientifique, Orleans, France). All animal experiments followed the French government's ethical and animal experiment regulations and the Swiss Federal Veterinary Office guidelines. Six- to 10-wk-old C57BL/6J and BALB/c mice were purchased from The Jackson Laboratory.

Nano-sized particles (TiO₂, anatase, 20 nm; TiO₂, rutile, 80 nm; ZnO, 15 nm; SiO₂, 15 nm, carbon nanotubes) were obtained from Iolitec. Before each use, the particles were sonicated 15 min and vigorously vortexed for 1 min. Nigericin, MSU, cytochalasin D, nystatin, methyl-β-cyclodextrin, and chlorpromazine hydrochloride were purchased from Sigma; phorbol 12-myristate 13-acetate and APDC from Alexis.

Anti-human cleaved IL-1β (2021L) was purchased from Cell Signaling and anti-human caspase-1 (sc-622) was obtained from Santa Cruz; anti-IL-1β p35 is a sheep antibody made in our laboratory. Antibodies against mouse IL-1β and mouse caspase-1 (p20) were generous gifts from Roberto Solari, Glaxo, and Peter Vandenabeele (Ghent University, Belgium).

For cytokine detection, ELISA kits for murine IL-1α (R&D System), IL-1β, and TNF-α (eBioscience) were used according to the manufacturers’ instructions.

Bone marrow-derived dendritic cells and bone marrow-derived macrophages were differentiated from tibial and femoral bone marrow progenitors as described elsewhere (29, 30). Cells were primed with 10 ng/mL ultra-pureLPS (Invivogen) for 3 h. Cell culture medium was replaced by OPTIMEM (Invitrogen) for the duration of the stimulation. Supernatants (SN) were precipitated by methanol-chloroform precipitation and cell extracts (XT) and SN of ≈200,000 cells were loaded for Western blot.

Primary adult murine tail epidermal keratinocytes were isolated from adult mice according to standard procedure (31). Tail skin was digested overnight in 10 mg/mL Dispase II (Roche). After removal of the epidermal sheets and trypsin digestion (0.05% Trypsin/EDTA), the epidermal suspension was filtered through a 100-μm mesh. The cells were cultured in Cnt-57 basal keratinocytes medium (CellnTec) on collagen-coated dishes (Purecol, Nutacon).

Primary human keratinocytes were isolated from neonatal foreskin (32). Briefly, the s.c. fatty tissue was trimmed and the biopsies were placed in 10 mg/mL Dispase II (Roche). The epidermal sheets were removed, chopped, and incubated in 0.05% Trypsin/EDTA. After filtration through a 100-μm mesh, the cells were cultured in Cnt-57 basal keratinocytes medium (CellnTec).

THP1 cells stably expressing shRNA against Nalp3 and Caspase-1 were obtained as previously described (13).

For experiments, THP1 cells were differentiated 3 h with 0.5 μM phorbol 12-myristate 13-acetate (Enzo Lifesciences). Cell extracts and precipitated supernatants were analyzed by Western blot.

Samples were fixed in 2% paraformaldehyde, 2.5% glutaraldehyde, and postfixed in 1% osmium tetroxide, 1.5% potassium ferrocyanide, and stained with 1% uranyl acetate. After dehydration, the samples were embedded in Durcupan. Ultrathin sections were contrasted with lead citrate and then examined with a Philips CM12 electron microscope.

Cholera toxin B staining (Invitrogen) was performed according to the manufacturer´s instructions. Slides were examined with a Leica SP5 tandem upright confocal microscope and the LAS AF lite software.

Peritonitis was induced in 6- to 10-wk-old age- and sex-matched mice by injection of 1–1.5 mg of sonicated and thoroughly vortexed TiO₂ particles in 400 μL sterile PBS. After 5–6 h, the mice were killed by CO₂ exposure and the peritoneal cavity was flushed with 5.5 mL of sterile PBS. The lavage fluid was centrifuged and the cell pellets analyzed for neutrophil influx by flow cytometry (CD11b⁺, Ly-6C⁺, Ly-6G^high, F4/80⁻) using the following monoclonal antibodies: anti-CD11b (M1/70), anti-F4/80 (BM8) (eBioscience), anti–Ly-6C (AL-21), and anti–Ly-6G (1A8) (BD Biosciences). Data were collected using a FACSCanto (BD Biosciences) and analyzed using FLOWJO software (Tree Star).

The supernatants were concentrated using Amicon Ultra centrifugal filters (Millipore) and analyzed using ELISA kits for murine IL-1α (R&D System), IL-1β, IL-6, and TNF (eBioscience) according to the manufacturers’ instructions.

For experiments in which IL-1α was blocked with a neutralizing antibody, mice were injected with anti-IL–1α (clone ALF-161, eBioscience) 80 μg i.p., 30 min before TiO₂ stimulation.

Intact murine back skin was challenged with nano-TiO₂ in PBS/olive oil or the PBS/olive oil carrier control by cutaneous exposure on the back skin. The back skin of the mice was shaved 24 h before exposure. Besides a single application, we applied the topical treatment daily for 5–10 consecutive days. Twenty-four hours after the last exposure, mice were killed by CO₂ exposure, skin biopsies were taken, fixed in formalin, and stained with hematoxylin and eosin (H&E) using standard procedures.

TiO₂ (7.5–30 mg/kg) in saline or the vehicle alone was administered by nasal instillation in a volume of 40 μL under light ketamine-xylasine anesthesia. Six hours postapplication, BAL was performed. Briefly, the trachea was incised and a homemade small plastic canula was inserted to flush both lungs three times with 1 mL PBS, going back and forth twice each time. The recovery of the total lavage exceeded 95%. The samples collected were centrifuged (600 g for 10 min) and cell pellets were resuspended in 0.4 mL PBS and maintained at 4 °C until cell determination. Differential cell counts were performed on cytospin preparations (Cytospin 3, Thermo Shandon) after May–Grünwald staining (Sigma) according to the manufacturer´s instructions. Differential cell counts were made on 200 cells using standard morphological criteria. For cytokine determination, the supernatants of the BALF were analyzed by ELISA.

After BAL, the lungs were perfused with Isoton (Coulder) to flush the vascular content. Lungs were homogenized by a rotor-stator (Ultra Turrax), the extract was centrifuged, and IL-1α and IL-1β levels in the supernatants of the lung homogenates were determined using ELISA.

The significance of the differences between groups was calculated using the unpaired t test and the Bonferroni posttest (GraphPad Prism version 5.0). Differences were considered significant when P < 0.05.