Autophagy Regulates IL-23 Secretion and Innate T Cell Responses through Effects on IL-1 Secretion Free

Autophagy is a highly conserved homeostatic mechanism that orchestrates the sequestration and degradation of damaged or noxious cytosolic constituents, including organelles, and regulates the catabolism of macromolecules during nutrient deprivation. More recently, autophagy has been shown to regulate inflammation and immune responses to infection. In particular, autophagy can mediate the intracellular killing of some bacteria by macrophages (1) and is involved in the presentation of Ags via MHC class I and class II molecules (2). Autophagy is regulated by cytokines; in macrophages, it is induced by IFN-γ and TNF-α (3, 4) and inhibited by IL-4, IL-13, and IL-10 (5‑8). Numerous studies have also demonstrated a role for autophagy in regulating the secretion of cytokines (reviewed in Ref. 9), including members of the IL-1 family.

Inhibition of autophagy allows the processing and secretion of the proinflammatory cytokine IL-1β in response to TLR3 and TLR4 agonists, and this process may be dependent on TIR domain-containing adaptor protein inducing IFN-β (TRIF), reactive oxygen species (ROS), and mitochondrial DNA (10–13). In addition, inhibition of autophagy enhances the secretion of IL-1α and IL-18 in response to LPS (10). Conversely, autophagosomes have been shown to target both pro–IL-1β and the inflammasome components ASC and NLRP3 within cells, thus directly regulating the production and availability of IL-1β (10, 14). Moreover, both IL-1β and IL-1α induce autophagy in macrophages (15), suggesting that these cytokines might regulate their own secretion. These data, together with studies showing that autophagy alters disease outcomes in animal models of sepsis (10, 11) and colitis (12), suggest that autophagy is a potent regulator of inflammatory responses. Supporting this, polymorphisms in human ATG16L1, a protein involved in autophagosome formation, have been associated with an increased risk of Crohn’s disease (16, 17).

Both IL-1β and IL-1α, as well as IL-18, can drive IL-17 secretion by Th17 and γδ T cells, but only in the presence of IL-23 (18–20). This pathway can drive inflammatory pathologies in multiple autoimmune diseases, including multiple sclerosis and rheumatoid arthritis (21–24), and is important in immunity to infection (18, 25, 26). In this study, we demonstrate that inhibition of autophagy promotes the secretion of IL-1α, IL-1β, and IL-23 by macrophages and dendritic cells (DC), which in turn influences the innate secretion of IL-17, IFN-γ, and IL-22 by γδ T cells. Moreover, this is dependent on IL-1 and NF-κB signaling, and both IL-1α and IL-1β augment IL-23 production. These data demonstrate a potentially pivotal role for autophagy in regulating inflammatory cytokines and innate T cell-driven immune responses.

Wortmannin, LY294002, and rapamycin were from Sigma-Aldrich (Poole, U.K.), and 3-methyladenine (3-MA) was from Sigma-Aldrich or Merck Millipore (Billerica, MA). Hydrocinnamoyl-l-valyl pyrrolidine (IL-1R antagonist [IL-1Ra]) and BAY 11-7082 (NF-κB inhibitor) were from Merck Millipore. LPS from Escherichia coli serotype R515 was from Enzo Life Sciences (Exeter, U.K.), and polyinosinic:polycytidylic acid [poly(I:C)], R837, and CpG were from InvivoGen (San Diego, CA). PP242 was from Chemdea (Ridgewood, NJ). Recombinant human (rh)IL-1β was from Immunotools (Friesoythe, Germany). Granzyme B-cleaved and full-length murine IL-1α were gifts from Prof. S. Martin (Trinity College Dublin, Ireland).

C57BL/6 mice were obtained from Harlan Olac (Bicester, U.K.) and were used at 8–16 wk of age. Animals were maintained according to the regulations of the European Union and the Irish Department of Health. Bone marrow-derived DC (BMDC) were generated as previously described (27) and grown in RPMI 1640 medium (Biosera, East Sussex, U.K.) with 10% FCS, l-glutamine (2 mM), and penicillin and streptomycin (50 U/ml and 50 μg/ml, respectively) (complete medium). The medium was supplemented with GM-CSF (40 ng/ml). Cells were plated out on day 10 and stimulated on day 11. BMDC from TRIF^+/+ and TRIF^−/− mice were provided by Prof. K. Fitzgerald (University of Massachusetts Medical Center, Worcester, MA).

CD3⁺ T cells were purified from spleens of C57BL/6 mice using magnetic beads for negative selection (Pan T Cell Isolation Kit II, mouse; Miltenyi Biotec), according to the manufacturer’s instructions. In some experiments, cells were cultured with IL-1β (10 ng/ml), IL-23 (10 ng/ml), LPS (10 or 100 ng/ml), 3-MA (5–10 mM), IL-1Ra (5 μg/ml), or combinations thereof. Alternatively, CD3⁺ T cells were cultured with supernatants (50 μl/200 μl CD3⁺ cells) obtained from DCs cultured for 24 h with 3-MA (5–10 mM) in the presence or absence of LPS (10 ng/ml). Similarly, CD3⁺ T cells were cultured with supernatants obtained from immortalized bone marrow-derived macrophages (iBMM) transfected with control (scrambled), beclin I, or Atg7 small interfering RNA (siRNA) stimulated for 18 h with LPS (10–100 ng/ml). After 72 h, IL-17A, IL-17F, and IFN-γ concentrations in supernatants were determined by ELISA.

For characterization of γδ T cells, lymph node T cells depleted of γδ T cells (2 × 10⁵) or γδ T cells (2 × 10⁴) enriched from lymph nodes using a TCRγ/δ⁺ T cell MACS Isolation kit, according to manufacturer’s instructions (Miltenyi Biotec), were cultured for 72 h in 200 μl complete medium. Supernatants (50 μl) or recombinant cytokines were added to each well, as described above. Cytokine (IL-17A, IL-22, and IFN-γ) secretion was measured by ELISA.

iBMM from wild-type C57BL/6, generated using J2 transforming retroviruses, were a gift from Prof. D. Golenbock (University of Massachusetts Medical Center). iBMM stably expressing EGFP-LC3 (GFP-LC3) were generated as previously described (10) and expanded under constant selection with 10 μg/ml puromycin. All iBMM were grown in complete medium.

M. tuberculosis strain H37Rv was grown in Middlebrook 7H9 broth with 0.2% glycerol and albumin–dextrose catalase supplement. Mycobacteria were grown to log phase before use, declumped by vortexing with glass beads for 60 s, and resuspended in RPMI 1640 medium with 10% FCS before use.

iBMM or BMDC were transfected by nucleoporation with an Amaxa Nucleofector Device (Lonza, Wokingham, U.K.) as described previously (5). iBMM were harvested after 2–3 d in culture and resuspended in 100 μl of the appropriate electroporation buffer (Lonza AG) with 10 nM siGENOME SMARTpool siRNA against beclin 1, Atg7, or siGENOME nontargeting control siRNA (Thermo Scientific, Lafayette, CO). Transfected cells were incubated overnight in complete medium before stimulation with LPS for 24 h. Targeting efficiency was analyzed by Western blot for beclin 1 and Atg7.

Cytokine secretion in cell culture supernatants was measured by ELISA, according to the manufacturer’s standard protocols (R&D Systems, Abingdon, U.K.; eBioscience, San Diego, CA; and BD Biosciences, San Diego, CA). Absorbance was read on a Multiscan FC plate reader and analyzed with SkanIt for Multiscan FC software (Thermo Scientific).

BMDC were grown and stimulated in 96 well U-bottom tissue culture plates. After treatment with LPS and 3-MA, cells were incubated with propidium iodide (1 μg/ml) for 5 min. A CyAN (DakoCytomation, Dublin, Ireland) flow cytometer was used to analyze fluorescence, and the data were analyzed using FlowJo software (Tree Star, Ashland, OR).

Mouse BMDC were stimulated for 4 h with 3-MA (2.5–10 mM) in the presence or absence of LPS (10 ng/ml). RNA was purified with the TRIzol (Invitrogen)/chloroform method; this process was followed by transcription into cDNA with a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Real-time PCR for the detection of IL-23p19 (il23a) (Mm00518984_m1) and IL-1β (il1b) (Mm00434228_m1) mRNA was performed with predesigned TaqMan gene expression assays (Applied Biosystems) using a 7500 Fast Real-Time PCR machine (Applied Biosystems). Expression was normalized to 18s rRNA.

Immortalized GFP-LC3⁺ iBMM were grown and stimulated on nitric acid-treated glass coverslips (19 mm diameter) in 12-well plates. Cells were fixed with 2% paraformaldehyde for 30 min at room temperature and permeabilized for 10 min with Triton X-100 (0.1% in PBS). All subsequent steps were conducted at room temperature. The cells were blocked with 5% goat serum in PBS with 1% BSA (blocking buffer) for 1 h and stained with primary Ab (1/200 in blocking buffer) for 1 h. After washing with PBS (five times), the cells were stained with Alexa Fluor 568-conjugated goat anti-rabbit IgG (1/500 in blocking buffer; Life Technologies, Grand Island, NY) for 1 h. In some experiments, cells were stained with tetramethylrhodamine isothiocyanate–phalloidin (American Peptide, Sunnyvale, CA) at 1 μg/ml to stain polymerized actin. The coverslips were mounted onto glass slides with fluorescent Mounting medium (DakoCytomation) and analyzed on an Olympus FV1000 laser scanning confocal microscope.

BMDC and immortalized iBMM were washed twice with PBS and lysed, according to a previously described protocol (28). Samples were loaded and separated on 10, 12, or 15% SDS-polyacrylamide gels and transferred to a nitrocellulose membrane. The membrane was blocked for 1 h with 5% milk in PBS/Tween 20 (0.1%). After washing with PBS/Tween 20, the membrane was probed with primary Ab (1/500 in 3% BSA) overnight at 4°C, washed three times with PBS/Tween, and probed with HRP-conjugated secondary Ab (1/2000 in PBS) for 1 h at room temperature. The membrane was washed three times with PBS/Tween 20 and developed with freshly prepared luminol-based detection solution. Densitometric analysis of blots was conducted using ImageJ software (National Institutes of Health, Bethesda, MD).

Female BALB/c mice were injected i.p. with 200 μl PBS, LPS (10 μg/mouse in 200 μl PBS), or LPS with rapamycin (1.5 mg/kg in 200 μl PBS). After 4 h, mice were bled, and ELISA analysis was performed on plasma samples.

Unless otherwise stated, data sets were analyzed for statistical significance (p < 0.05) using one-way ANOVA, followed by a Bonferroni post hoc test. Statistical analysis was conducted using GraphPad Prism software.

3-MA is a PI3K inhibitor that blocks autophagic degradation of proteins (29). Our previous work has demonstrated that treatment of murine macrophages and DC with 3-MA stimulates the secretion of IL-1β, IL-1α, and IL-18 in responses to specific TLR ligands (10). To determine whether IL-23 secretion was also regulated by autophagy, we treated human PMA-differentiated THP-1 monocyte/macrophages with 3-MA and LPS overnight and measured cytokines in the supernatants. At concentrations of 2.5 mM and above, 3-MA stimulated the secretion of both IL-23 and IL-1β in response to LPS (Fig. 1A, 1B). IL-1α secretion by THP-1 cells was also stimulated under these conditions (Supplemental Fig. 1A). In contrast, LPS-induced TNF-α secretion was enhanced by 3-MA at 2.5 mM but inhibited by 10 mM (Fig. 1C). In primary murine BMDC, 3-MA increased LPS-induced secretion of IL-23 and IL-1β (Fig. 1D, 1E), as well as IL-1α (Supplemental Fig. 1B). Levels of IL-6 secreted by these cells were not significantly affected (Fig. 1F). This effect was also observed in primary human monocyte-derived DC in which 3-MA augmented LPS-induced IL-23 secretion but inhibited TNF-α secretion (Supplemental Fig. 1C, 1D). In addition, 3-MA enhanced the secretion of IL-23 by BMDC in response to heat-killed E. coli (Fig. 1G). Secretion of IL-1α and IL-1β was also increased in response to E. coli with 3-MA (Supplemental Fig. 1E, 1F), whereas IL-6 secretion was either unaffected or inhibited (Supplemental Fig. 1G). Similarly, stimulation of THP-1 cells with 3-MA enhanced IL-23 secretion in response to live virulent M. tuberculosis H37Rv (Fig. 1H), whereas TNF-α secretion was either unaffected or inhibited (Supplemental Fig. 1H). To confirm the autophagy-inhibiting activity of 3-MA, we treated BMDC with the autophagy inducer rapamycin or with LPS for 2 h in the presence or absence of 3-MA and bafilomycin (to inhibit autophagic flux (30)) and measured levels of LC3 I (cytosolic) and LC3 II (lipidated autophagosomal membrane-bound) by Western blotting. Both rapamycin and LPS increased the amount of LC3 II present, indicating an increase in autophagosome biogenesis, whereas 3-MA inhibited this effect (Fig. 1I).

Treatment with other broad-range PI3K inhibitors has also been shown to inhibit autophagy (30). In our model, LPS-induced IL-23 secretion by BMDC was increased in the presence of the PI3K inhibitors wortmannin and LY294002 (Fig. 1J, 1K), and both IL-1β and IL-23 secretion were enhanced by LY294002 in THP-1 cells (Supplemental Fig. 1I, 1J). These data demonstrate that autophagy-inhibiting drugs augment IL-23 secretion by macrophages and DC in response to microbial stimuli.

IL-1β secretion in the absence of autophagy is dependent on the TLR adaptor molecule TRIF and thus only occurs in response to TLR3 and TLR4 agonists, which signal through this pathway (10, 12). In this study, we also observed enhanced secretion of IL-23 by 3-MA–treated BMDC in response to TLR3 and TLR4 ligands but not ligands of other TLRs. Both IL-23 and IL-1β secretion were increased in cells treated with 3-MA in combination with the TLR3 ligand poly(I:C) or the TLR4 ligands LPS and monophosphoryl lipid A (which preferentially signals through TRIF) but not the TLR2 ligand PAM₃CysK₄, the TLR7 agonist R837 (imiquimod), the TLR7/8 agonist R848, or the TLR9 ligand CpG (Fig. 2A–D). Both IL-1β and IL-23 secretion were augmented by 3-MA in response to zymosan, a ligand for TLR2, but which also signals through binding to dectin-1 (Fig. 2C, 2D). To determine whether the effect of 3-MA on IL-23 secretion was dependent on signaling through TRIF, which has been previously demonstrated for IL-1β secretion (10, 12), we treated BMDC from wild-type and TRIF^−/− mice with LPS and 3-MA. Both IL-1β and IL-23 secretion was inhibited in the TRIF^−/− cells (Fig. 2E, 2F).

Knockdown of autophagy-specific genes, using siRNA targeting beclin 1 and Atg7, enhances LPS-induced secretion of IL-1β by macrophages (10, 12). In this study, we found that siRNA depletion of beclin 1 in murine iBMM enhanced secretion of IL-1α and IL-1β but did not affect secretion of IL-12/23 p40 (Fig. 3A‑D). Moreover, knockdown of either beclin 1 or Atg7 enhanced secretion of IL-23, but not TNF-α, in response to LPS (Fig. 3E–H), replicating the effects produced by treatment with 3-MA. Inhibition of autophagy with 3-MA has been demonstrated to drive IL-1β secretion by human PBMC in response to M. tuberculosis (31). Similarly, we show in this study that knockdown of beclin 1 in BMDC, even though only partial, significantly enhanced IL-23 but not TNF-α secretion in response to irradiated M. tuberculosis strain H37Rv (Fig. 3I, 3J). These data demonstrate that in microbe-activated macrophages and DC, the specific inhibition of autophagy enhances IL-23 secretion.

The effects of autophagy inhibition on LPS-driven IL-1β secretion have previously been shown to occur at the transcriptional level (31, 32). In this study, we examined cytokine mRNA expression in murine BMDC in response to LPS and 3-MA. After 4-h treatment with LPS, expression of IL-1β mRNA was increased, but this was unaffected by 3-MA (Fig. 4A). However, expression of IL-23 mRNA was enhanced by LPS, and this was further augmented by the addition of 3-MA (Fig. 4B). Expression of IL-10 mRNA was increased by LPS but unaffected by 3-MA (Fig. 4C). Similarly, in iBMM transfected with beclin 1 siRNA, expression of IL-23p19 mRNA in response to LPS was significantly increased after 4 h compared with cells transfected with scrambled siRNA (Fig. 4D). Thus, inhibition of autophagy not only leads to inflammasome activation but also to the upregulation of IL-23 at the transcriptional level.

In human PBMC exposed to wheat gliadin, IL-23 secretion is driven by IL-1 signaling (33). To determine whether IL-23 secretion was dependent on IL-1 signaling in our model, we treated BMDC and PMA-differentiated THP-1 cells with LPS and 3-MA in the presence or absence of hydrocinnamoyl-l-valyl pyrrolidine, an IL-1Ra. In BMDC, IL-1Ra inhibited the secretion of both IL-23 and IL-1β in response to LPS and 3-MA but had no effect on IL-12p40 or IL-12/23p70 secretion (Fig. 5A–D), whereas in THP-1 cells, IL-1Ra inhibited the secretion of IL-23, but not IL-1β, in response to LPS and 3-MA (Fig. 5E, 5F). In THP-1 cells, a neutralizing Ab against IL-1β inhibited IL-23 secretion at all concentrations tested, whereas a neutralizing Ab against IL-1α had a small inhibitory effect only at the lowest concentration tested (Fig. 6A). Similarly, secretion of IL-23 by THP-1 cells stimulated with heat-killed E. coli and 3-MA was inhibited with neutralizing Abs against IL-1β but not IL-1α (Fig. 6B). To further assess the ability of IL-1 to induce IL-23 secretion, THP-1 cells were treated with rhIL-1α or rhIL-1β. At 10 ng/ml, rhIL-1α induced IL-23 secretion, both in the presence and absence of LPS (Fig. 6C). Similarly, rhIL-1β stimulated IL-23 secretion by THP-1 cells, regardless of exposure to LPS (Fig. 6D).

Because the IL-1R1 is known to signal through the NF-κB pathway, we assessed the effects of a NF-κB inhibitor, BAY 11-7082, on LPS/IL-1β–induced IL-23 secretion. In PMA-differentiated THP-1 cells, the inhibitor abrogated the effects of LPS and rhIL-1β, but not LPS alone, on IL-23 secretion (Fig. 6E). In addition, LPS and rhIL-1β enhanced the secretion of TNF-α by THP-1 cells, and this was again inhibited by the NF-κB inhibitor (Fig. 6F). These data suggest that activation of NF-κB via the IL-1R1 stimulates the production of IL-23 as well as TNF-α.

Previous reports have suggested that ROS promote activation of the inflammasome and secretion of IL-1β in autophagy-deficient cells (10, 12). Given the role of IL-1β in promoting IL-23 secretion, we assessed the potential role of ROS on IL-23 secretion by BMDC. The ROS scavenger N-acetyl-l-cysteine completely abrogated the secretion of IL-23 in response to LPS and 3-MA in BMDC at all concentrations tested (Supplemental Fig. 2A). In addition, at higher concentrations, N-acetyl-l-cysteine inhibited the secretion of IL-1β and IL-1α (Supplemental Fig. 2B, 2C). IL-6 secretion was affected only at the highest concentration (50 mM) (Supplemental Fig. 2D). Taken together, these data demonstrate that IL-23 secretion in autophagy-deficient cells is directly regulated by IL-1 signaling and is dependent on the generation of ROS.

Our previous work demonstrated that in LPS-stimulated iBMM, autophagosomes sequester pro–IL-1β for degradation (10). To assess whether the same was true for IL-23, we stained LPS-treated GFP-LC3–expressing iBMM with Ab against IL-1β or IL-23 and examined the cells by confocal microscopy. As previously shown, after overnight treatment with LPS, IL-1β colocalized with GFP-LC3 (Fig. 7A). However, no IL-23 could be detected in GFP-LC3⁺ autophagosomes (Fig. 7B). This would suggest that unlike IL-1β, intracellular IL-23 is not sequestered by autophagosomes.

To determine the effect of autophagy induction on IL-23 production, we treated BMDC with LPS and PP242, an autophagy-inducing mammalian target of rapamycin (mTOR) inhibitor. At all concentrations tested, PP242 inhibited LPS-induced IL-23 production (Fig. 7C) Moreover, when another mTOR inhibitor, rapamycin, was administered i.p. in a murine model of endotoxin-induced sepsis, it abrogated the stimulatory effects of LPS on serum IL-23 (Fig. 7D), as we have previously demonstrated for IL-1β (10). These data would suggest that although blocking autophagy allows IL-23 production through the enhancement of IL-1 secretion, the induction of autophagy has the opposite effect and downregulates IL-23 secretion.

Both IL-1β and IL-1α can induce autophagy (15), suggesting that these cytokines may limit their own secretion. In agreement with this, we found that both IL-1β and IL-1α increased autophagosome formation in GFP-LC3–expressing iBMM (Supplemental Fig. 3A, 3B). Interestingly, granzyme B-processed IL-1α, which has recently been shown to have greater proinflammatory effects than the full-length form (34), was a more potent autophagosome inducer (Supplemental Fig. 3B). In addition, IL-23 increased autophagosome formation in GFP-LC3–expressing iBMM (Supplemental Fig. 3C). These data would indicate that autophagy represents a negative feedback mechanism for the control of IL-1β secretion and the subsequent production of IL-23.

Given that IL-1 and IL-23 synergistically stimulate TCR-independent IL-17 production by Th17 cells (20) and that the inhibition of autophagy in macrophages and DC enhances the secretion of IL-1β, IL-1α, and IL-23, we sought to determine whether this could, in turn, stimulate innate Th17-like cytokine responses in vitro. Supernatants from BMDC treated with LPS and 3-MA (to induce IL-1β and IL-23 secretion; Supplemental Fig. 4A, 4B) stimulated IL-17A and IL-17F secretion by MACS-purified splenic CD3⁺ T cells (Fig. 8A, 8B). These supernatants also enhanced T cell secretion of IFN-γ (Fig. 8C). Moreover, LPS and 3-MA had no direct effect on T cells (Fig. 8D–F), suggesting that this effect was due to soluble factors secreted by the 3-MA–treated BMDC. Supernatants from iBMM transfected with siRNA against beclin 1 similarly enhanced secretion of IL-17A from splenic T cells (Supplemental Fig. 4C‑H), demonstrating that this effect is due to impaired autophagy. Because γδ T cells are a key source of innate cytokines secreted in response to IL-1β and IL-23 in the absence of TCR stimulation (20), we isolated γδ⁺ and γδ⁻ T cells from lymph nodes and assessed their capacity to produce IL-17, IL-22, and IFN-γ when exposed to supernatants from autophagy-deficient DC. When cultured with supernatants from BMDC treated with LPS and 3-MA or with recombinant IL-1β and IL-23, γδ T cells were responsible for the majority of IL-17, IL-22, and IFN-γ secretion (Fig. 8G–L). Thus, our data demonstrate that, through the regulation of IL-1β and IL-23 secretion, autophagy can influence innate cytokine production through the indirect stimulation of γδ T cells.

Numerous cytokines have been shown to regulate autophagy in macrophages and DC (9). In particular, IFN-γ, TNF-α, IL-1α, and IL-1β increase autophagosome formation in macrophages (3, 4, 15), whereas IL-4, IL-13, and IL-10 inhibit autophagy (5–8). More recent studies have also demonstrated that autophagy can directly influence the secretion of cytokines, including IL-1β. Inhibition of autophagy enhances IL-1β production and secretion, because of the intracellular accumulation of endogenous inflammasome-activating factors, such as mitochondrial DNA and ROS (10–13, 31, 32). Conversely, autophagosomes themselves can directly target pro–IL-1β and the inflammasome components ASC and NLRP3 (10, 14). In addition, autophagy can regulate the secretion of IL-18 and IL-1α, although the mechanisms behind the latter may be different to those characterized for IL-1β (10, 12). In the current study, we have demonstrated that the inhibition of autophagy can also enhance IL-23 secretion in response to LPS and whole bacteria. A previous study has reported that wortmannin, but not LY294002, enhances IL-23 secretion by human DC in response to TLR ligation via a type I PI3K-independent mechanism (35). In this study, we observed a similar effect, although in our system, LY294002 did enhance IL-23 secretion, as did both 3-MA and siRNA knockdown of the autophagy proteins beclin 1 and Atg7. This effect was specific to IL-23; IL-12p70 secretion was inhibited by 3-MA, whereas IL-12/23p40 levels were unaffected by autophagy inhibition.

Our data suggest that autophagy has a largely indirect effect on IL-23 through its regulation of IL-1β secretion. Previous studies have demonstrated that increased IL-1β secretion in autophagy-deficient cells is dependent on TRIF and ROS (10, 12). The present study suggests that increased IL-23 secretion in autophagy-deficient cells is similarly dependent on both ROS and TRIF. Interestingly, zymosan, which is not known to signal through TRIF but signals through TLR2/MyD88 and the β-glucans receptor Dectin-1 receptor (36), also acts synergistically with 3-MA to induce IL-23. However, Dectin-1 also signals through a Syk kinase-dependent pathway that occurs independently of TLR2 (37), and a previous study has demonstrated that β-glucans can induce IL-23 via an IL-1–dependent mechanism (33). Our previous study has demonstrated that pro–IL-1β colocalizes with GFP-LC3 in LPS-treated cells, and induction of autophagy, both in vitro and in vivo, has been shown to inhibit IL-1β secretion (10). In the current study, we have demonstrated that although IL-23 is not sequestered by autophagosomes, induction of autophagy in vitro and in vivo does inhibit IL-23 production, perhaps as a result of effects on IL-1β secretion.

Although neutralizing IL-1α had little or no effect on IL-23 secretion, IL-1α did itself enhance IL-23 secretion. This suggests that IL-1R signaling is important but perhaps that IL-1α release in response to 3-MA does not have consequences as significant as those of IL-1β. Interestingly, a recent study has suggested that IL-1α secretion may also depend on IL-1β (38), further implicating IL-1β as a key driver of inflammation. In agreement with our findings, IL-1β alone induced IL-23 secretion by human PBMC, and secretion of IL-23 in response to gliadin and β-glucans was also IL-1 dependent (33). Our data suggest that IL-1 drives IL-23 (and TNF-α) secretion through NF-κB signaling. Indeed, the IL-23 p19 promoter contains a κB site, which is required for the induction of p19 transcription (39), and IL-1R1 signaling activates NF-κB (40). Taken together, the data presented in this article, coupled with previous studies, suggests that in the absence of autophagy, TRIF- and ROS-dependent inflammasome activation, and IL-1β secretion drives IL-23 production and secretion in an NF-κB–dependent manner. It is notable that in some cases the effects of 3-MA on IL-23 are nonlinear, whereas those on IL-1β are linear at the dose range tested. This could be due to conflicting effects of 3-MA on autophagy and other PI3K-dependent processes, although further studies would be required to test this. Induction of autophagy, in contrast, leads to the sequestration and degradation of inflammasome components, pro–IL-1β, and other inflammatory stimuli, leading to reduced IL-1β secretion and thus reduced IL-23 production.

Intriguingly, IL-1β, IL-1α, and IL-23 can all induce autophagy, suggesting a potentially important negative feedback mechanism for the control of excessive inflammation. IL-23 is important in driving intestinal inflammation in various mouse models of inflammatory bowel disease (41–43). In humans, genome-wide association studies have identified the IL-23R as an important correlate in Crohn’s disease (44). Moreover, polymorphisms in ATG16L1, which encodes a protein required for autophagy, have been associated with susceptibility to Crohn’s disease (16, 45). In a murine model of dextran sodium sulfate-induced colitis, loss of the autophagy protein Atg16L1 results in more severe disease, with increased weight loss, enhanced colonic inflammation, and decreased survival (12). The same study also demonstrated greatly enhanced serum IL-1β levels in Atg16L1-deficient animals, although the effects on IL-23 were not examined. Thus, our data indicate that autophagy and IL-23 might be closely linked through the autophagic control of IL-1β secretion in this model.

γδ T cells are a rapid and potent innate source of cytokines, including IL-17, IL-22, and IFN-γ. Although IL-17 and IFN-γ are thought to be key modulators of inflammation in disease models, such as dextran sodium sulfate-induced colitis, IL-22 acts on nonimmune cells, particularly in the skin, gut, liver, lungs, and kidneys, to induce production of antimicrobial peptides and to protect and maintain epithelial barriers. IL-1β drives—and IL-23 amplifies—such Th17-like cytokine responses, particularly from γδ T cells (19, 20, 26, 46, 47). Given that autophagy is a potent regulator of the inflammasome and IL-1 secretion, our data provide evidence that through the regulation of IL-1, autophagy is able to influence not only IL-23 secretion but also innate cytokine responses, characterized by robust secretion of IL-17, IL-22, and IFN-γ by γδ T cells independently of TCR stimulation.

Autophagy may thus represent a potent target for anti-inflammatory therapies that could impact on a number of autoimmune diseases, including multiple sclerosis and inflammatory bowel disease. Conversely, directed inhibition of autophagy might be useful to drive Th17-like cytokine responses against specific Ags. This could, in turn, increase the efficacy of some vaccines that require these cytokines, as has been suggested for subunit vaccines against Streptococcus pneumonia and M. tuberculosis (48, 49).

In summary, our data uncovers a novel role for autophagy in the regulation of IL-23 secretion by macrophages and dendritic cells and, as a result, IL-17, IFN-γ, and IL-22 production by γδ T cells. This is dependent on IL-1 secretion, which is regulated by autophagy through at least two separate mechanisms. These findings further highlight the potential of autophagy as a therapeutic target for inflammatory conditions mediated by IL-1.

We thank Gavin McManus and Barry Moran for technical assistance.

K.H.G.M. is a cofounder and shareholder in Opsona Therapeutics Ltd., involved in the development of anti-inflammatory therapeutics, and Trimod Therapeutics, involved in the development of cancer vaccines and therapeutics. The other authors have no financial conflicts of interest.