Inflachromene inhibits autophagy through modulation of Beclin 1 activity

Autophagy is an essential life-sustaining process that mediates the clearance of aggregated proteins and damaged organelles for maintenance of cellular homeostasis. Autophagy is tightly regulated to promote cellular survival and is constitutively active at a basal level in cells. Beclin 1 accurately regulates autophagy by combining with the PI3K–VPS34 complex and positive and negative cofactors (Kang et al., 2011; Levine et al., 2015). Modulation of autophagy holds great potential for the treatment of a wide range of diseases (Rubinsztein et al., 2012; Galluzzi et al., 2017). For example, induction of autophagy has been associated with beneficial effects in neurodegenerative diseases and infectious diseases. Conversely, the decreased autophagy seen upon inactivation of the gene encoding Beclin 1 was found to increase protein aggregation and neurodegeneration upon mutant Huntingtin protein expression (Shibata et al., 2006; Jia et al., 2007). Beclin 1 transfer or overexpression can reduce aggregation of amyloid β and α-synuclein and improve neuronal function (Pickford et al., 2008; Spencer et al., 2009). However, excessive autophagy is cytotoxic and aggravates some diseases, such as ischemia and reperfusion injuries of the heart and brain (Gustafsson and Gottlieb, 2009; Shi et al., 2012) as well as X-liked myopathy with excessive autophagy (XMEA) (Kalimo et al., 1988). The development of an autophagy inhibitor could benefit certain diseases depending on the autophagy condition. Beclin 1 is a key target molecule for inhibiting autophagy as a part of a lipid kinase complex that mediates the initial stage of autophagosome formation.

HMGB1 is a nuclear protein that contributes to DNA repair, replication and transcriptional regulation (Lee et al., 2014b). HMGB1 is translocated to the cytoplasm upon proinflammatory or oxidative stress and is eventually secreted into the extracellular space after post-translational modifications such as acetylation and phosphorylation (Bonaldi et al., 2003; Youn and Shin, 2006). Extracellular HMGB1 functions as a danger-associated molecular pattern (DAMP) molecule capable of inducing inflammation and chemotaxis, and HMGB1 can augment inflammation after direct binding lipopolysaccharide (LPS) and lipoteichoic acid (Youn et al., 2008; Kwak et al., 2015). Cytoplasmic HMGB1 induces autophagy by directly binding to Beclin 1 after displacing Bcl-2 from Beclin 1 (Tang et al., 2010; Zhu et al., 2015).

Inflachromene (ICM) is a benzopyran-embedded tetracyclic compound that has been reported to inhibit LPS-induced secretion of HMGB1 by modulating post-translational modifications in microglial cells, resulting in amelioration of experimental autoimmune encephalomyelitis (EAE) and sepsis (Lee et al., 2014a; Cho et al., 2017). In this study, we hypothesized that inhibition of HMGB1 transport to the cytoplasm would decrease autophagy. We studied the effect of ICM on autophagy and demonstrated that ICM inhibited autophagy by inhibiting HMGB1 migration to the cytoplasm. In parallel, autophagy was also inhibited through ICM-mediated Beclin 1 degradation because of the activation of RNF216, which increased Beclin 1 K48 ubiquitylation. Our study demonstrated the novel inhibitory effects of ICM on autophagy. ICM could be a therapeutic candidate for treatment of HMGB1-related proinflammatory diseases and diseases related to excessive autophagy.

ICM is a small molecule that binds to HMGB1 and HMGB2 and exerts anti-inflammatory effects with the potential to prevent EAE and sepsis (Lee et al., 2014a; Cho et al., 2017) (Fig. S1A). HMGB1 plays an important role in autophagy induction by interacting with Beclin 1 (Tang et al., 2010; Zhu et al., 2015). To determine whether ICM could inhibit autophagy, we tested the effect of ICM on autophagy under the starvation conditions produced by Earle's balanced salt solution (EBSS) culture. ICM inhibited autophagy under starvation conditions to a similar extent to that caused by the autophagy inhibitors 3-MA and spautin-1. Spautin-1 is a recently described synthetic chemical that inhibits autophagy by inhibiting VPS34 (also known as PIK3C3 in mammals) (Liu et al., 2011) (Fig. S1B). ICM dose-dependently inhibited LC3-II conversion (denoting the change from the LC3-I form to the LC3-phosphatidylethanolamine conjugate that binds to autophagosomal membranes), under starvation condition or H₂O₂ treatment in HEK293T cells (Fig. 1A,B). Rapamycin is an mTOR inhibitor that induces autophagy, while NH₄Cl and chloroquine (CQ) are lysosome-neutralizing agents that inhibit autophagy by impairing autophagosome–lysosome fusion. ICM inhibited rapamycin-induced, NH₄Cl-treated and CQ-treated autophagy with or without starvation (Fig. 1C–E; Fig. S1C).

We next performed an mRFP–GFP–LC3 puncta formation assay to evaluate autophagy flux. The low pH of lysosomes quenches the fluorescence signal of GFP but not that of RFP, thus enabling mRFP–LC3 to be detected in autolysosomes (Mizushima et al., 2010). HEK293T cells were transfected with mRFP–GFP–LC3 and pre-treated with ICM, and then further incubated under starvation conditions. As shown in Fig. 1F, ICM treatment reduced the number of autophagosome puncta (double positive for RFP and GFP) compared with either the control or NH₄Cl treatment under starvation conditions. These results suggest that ICM functions as an early-stage inhibitor of autophagy.

To investigate the reason why ICM inhibited the early stage of autophagy flux, we next studied the effect of ICM on the nucleocytoplasmic translocation of HMGB1 after starvation, since nucleocytoplasmic translocation of HMGB1 is related to autophagy (Tang et al., 2010; Livesey et al., 2012; Yang et al., 2015; Zhu et al., 2015). By using confocal microscopy and nuclear/cytoplasmic analysis, we determined that ICM inhibited the nucleocytoplasmic translocation of HMGB1 under starvation conditions (Fig. 2A,B), consistent with a previous report (Lee et al., 2014a). HEK293T cells were transfected with a mixture of HMGB1 and Beclin 1 plasmids, and then treated with ICM. The interaction between HMGB1 and Beclin 1 was increased under starvation conditions but was decreased by ICM treatment (Fig. 2C). HEK293T cells were transfected with HMGB1 C106S and HMGB1 C23S/C45S mutant plasmids. HMGB1 C106S is known to be located in the cytoplasm (Yang et al., 2012), and HMGB1 C23 and C45 residues are important for Beclin 1 binding (Tang et al., 2010). As expected, HMGB1 C106S showed a high binding affinity for Beclin 1 under starvation conditions and binding was inhibited by ICM, while double mutant HMGB1 C23S/C45S showed almost no binding to Beclin 1. Regardless of the form of HMGB1, the interaction between Beclin 1 and HMGB1 was inhibited by ICM (Fig. 2D). These results suggest that the inhibition of HMGB1 nucleocytoplasmic translocation by ICM decreased Beclin 1 interaction and autophagy. ICM can also bind to HMGB2 and inhibit its secretion (Lee et al., 2014a). Hence, we measured the effect of ICM on HMGB2-mediated autophagy flux since HMGB2 is related to autophagy formation (An et al., 2015), and we observed that ICM also inhibited HMGB2-mediated autophagy flux (Fig. S2A,B).

Interestingly, the level of Beclin 1 protein was significantly decreased by ICM treatment (Fig. 3A). We hypothesized that ICM had another role in autophagy inhibition. Therefore, we studied Beclin 1 protein degradation upon ICM treatment. Downregulation of the Beclin 1 protein level is a critical factor in autophagy termination (Xu et al., 2014; Liu et al., 2016). ICM dramatically decreased the endogenous Beclin 1 protein level in a dose- and time-dependent manner whereas the levels of ATG5–ATG12 conjugate, ATG7 and ATG16L1 were not changed (Fig. 3B,C), suggesting that ICM specifically influenced the degradation of Beclin 1 protein. When HEK293T cells were cultured under starvation conditions with ICM and the proteasomal inhibitor MG132, Beclin 1 protein levels recovered (Fig. 3D). These data demonstrate that ICM specifically induced Beclin 1 degradation via the proteasome system.

Next, we investigated the effect of ICM on the degradation of Beclin 1. HEK293T cells were transiently transfected with a Flag-tagged Beclin 1 plasmid and cultured under starvation conditions, and ICM was added with or without MG132. The Beclin 1 ubiquitylation level was increased by ICM treatment and was profoundly increased by MG132 treatment. In addition, K48-specific ubiquitylation of Beclin 1 was increased (Fig. 4A). When HEK293T cells were transiently transfected with Flag-tagged Beclin 1 and treated with ICM in the presence or absence of EBSS, ICM increased Beclin 1 ubiquitylation regardless of starvation status (Fig. 4B). Similar results were observed when HEK293T cells were co-transfected with Beclin 1 and HA-tagged ubiquitin (Ub) plasmids in the presence or absence of ICM. MG132 was used as a positive control (Fig. 4C). HEK293T cells were transfected with HA-tagged wild-type (WT) Ub, Ub K48 (Ub where all lysine residues are mutated except K48) or Ub K48R (Ub where only K48 is mutated into Arg) plasmids for 48 h and treated with 25 μM ICM for 6 h. ICM increased K48-linked ubiquitylation of Beclin 1, and the ubiquitylation was dramatically decreased when the Ub K48R mutant was transfected (Fig. 4D). These results show that ICM induced degradation of Beclin 1 by promoting K48-linked ubiquitylation.

Proteasome-dependent protein degradation requires an E3 Ub ligase to conjugate Ub to a target protein (Bedford et al., 2011). We hypothesized that the ICM treatment could influence the interaction between an E3 Ub ligase and Beclin 1. Several Beclin 1 E3 ligases have been identified (Shi and Kehrl, 2010; Platta et al., 2012; Xia et al., 2013; Xu et al., 2014); we focused on the RNF216 E3 ligase because RNF216 mediates K48-linked ubiquitylation of Beclin 1. We tested whether ICM influences the interaction between RNF216 and Beclin 1. HEK293T cells were co-transfected with V5-tagged Beclin 1 and Flag-tagged RNF216 plasmids with or without MG132. The Beclin 1 protein level was decreased in a dose-dependent manner by RNF216 plasmid transfection and recovered upon treatment of MG132 (Fig. 5A). Moreover, the level of Beclin 1 ubiquitylation was dose-dependently increased by RNF216 plasmid transfection in HEK293T cells (Fig. 5B). ICM treatment led to an increased ubiquitylation level of Beclin 1 even in the RNF216-transfected cells (Fig. 5C). When we added ICM to cells that were co-transfected with Beclin 1 and RNF216 plasmids, the interaction between RNF216 and Beclin 1 was significantly increased by ICM, although the Beclin 1 level was decreased in the input because of the RNF216 overexpression (Fig. 5D). We further investigated the interaction between RNF216 and Beclin 1 using a proximity ligation assay (PLA). The interaction between Beclin 1 and RNF216 was significantly increased by ICM treatment (Fig. 5E), and these interactions occurred in the cytoplasm (Fig. 5F) (Xu et al., 2014). This suggests that ICM promotes the interaction between Beclin 1 and RNF216, leading to Beclin 1 ubiquitylation.

In summary, ICM suppresses early stage autophagy by inhibition of the translocation of HMGB1 and promotion of Beclin 1 degradation. Therefore, ICM could act as a treatment in for diseases related to excessive autophagy as well as inflammatory disease (Fig. 6).

Autophagy is a catabolic process required for nutrient homeostasis, the degradation of damaged organelles and aggregated intracellular proteins, and energy salvage. Sustained autophagy is essential for intracellular homeostasis. Optimal autophagy is pro-survival but excessive autophagy is considered cytotoxic and may result in autophagic cell death. Excessive autophagy aggravates ischemia and reperfusion injury, such as ischemic heart disease (Gustafsson and Gottlieb, 2009) and cerebral ischemia (Shi et al., 2012), vein graft neointima formation (Chang et al., 2016) and XMEA (Kalimo et al., 1988). Therefore, small molecules that inhibit excessive autophagy could be potentially beneficial therapeutics.

In this study, we demonstrated that autophagy is inhibited by ICM, a factor originally known to bind to HMGB1 and inhibit its extracellular secretion, resulting in reduced inflammation. HMGB1 induces autophagy by binding to Beclin 1 (Tang et al., 2010); thus, nucleocytoplasmic translocation of HMGB1 and binding to Beclin 1 is an important process that is required to initiate autophagy. ICM can inhibit LPS-stimulated phosphorylation and acetylation of HMGB1 (Lee et al., 2014a), modifications which are important for the translocation and secretion of HMGB1 (Bonaldi et al., 2003; Youn and Shin, 2006).

Another novel finding from our study was that ICM enhanced the interaction between Beclin 1 and RNF216 E3 ligase necessary for Beclin 1 ubiquitylation, which resulted in Beclin 1 degradation. ICM treatment caused a reduction of the Beclin 1 protein level but not that of ATG5–ATG12 conjugate, ATG7 and ATG16L1, demonstrating that ICM specifically targeted Beclin 1 degradation. Autophagy regulation is complex and Beclin 1 ubiquitylation is an important factor in this process. TRAF6 and A20 (also known as TNFAIP3), respectively, regulate K63-linked ubiquitylation and deubiquitylation of Beclin 1 after TLR4-induced autophagy (Shi and Kehrl, 2010), Nedd4 induces K11-linked polyubiquitylation (Shi and Kehrl, 2010), WASH1 (Wiskott Aldrich Syndrome protein and Scar homologue 1, also known as WASHC1) mediates suppression of ubiquitylation (Xia et al., 2013), and RNF216 mediates K48-linked ubiquitylation of Beclin 1 (Xu et al., 2014). Thus, further investigation is needed to elucidate the molecular mechanism through which ICM mediates binding of Beclin 1 to RNF216 and other Ub-related enzymes. Given that Beclin 1 is transcriptionally upregulated by the p65 subunit of NF-κB (also known as RelA) (Copetti et al., 2009), ICM could inhibit the Beclin 1 protein level since ICM inhibits NF-κB activity (Lee et al., 2014a).

Autophagy inhibitors could enhance the therapeutic efficacy of cancer treatment by altering sensitivity to chemotherapy or radiotherapy. For example, spautin-1, a small-molecule inhibitor of autophagy that promotes the degradation of VPS34 complexes by inhibiting ubiquitin-specific peptidases USP10 and USP13 (Liu et al., 2011), could be used as a possible lead anticancer compound for chronic myeloid leukemia (Shao et al., 2014). Preliminary data from our laboratory show that ICM induced apoptosis and inhibited mammosphere formation in MCF-7 cells (unpublished results, Y.H.K.). In summary, the ICM molecule, originally reported to inhibit HMGB1 secretion during sterile inflammation, could be used as a therapeutic factor for modulation of excessive autophagy or in autophagy-related diseases.

Human embryonic kidney (HEK) 293T (ATCC^® CRL-1573™) cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Life Technologies, Camarillo, CA), 100 U/ml penicillin, 100 µg/ml streptomycin and 2 mM L-glutamine at 37°C under 5% CO₂. Earle's balanced salt solution (EBSS) medium (Wellgene, Gyeongsan-si, Gyeongsangbuk-do, Korea) was used for starvation. The autophagy inhibitor chloroquine (Sigma-Aldrich, St Louis, MO), proteasome inhibitor MG132 (Sigma-Aldrich), 3-methyladenine (3-MA; Sigma-Aldrich), mTOR inhibitor rapamycin (TOCRIS, Bristol, UK) and VPS34 inhibitor spautin-1 (TOCRIS) (Liu et al., 2011) were used. The HMGB1 and HMGB2 secretion inhibitor ICM (AOBIOUS, Gloucester, MA) was used.

Flag- and V5-tagged Beclin 1 plasmids for mammalian cell expression were kindly provided by Prof. Eun-Kyeong Jo (ChungNam Univ., Daejoen, Korea). Flag-tagged RNF216 was purchased by GenScript (#OHu08423). mRFP–GFP–LC3 plasmid was kindly provided by Prof. Myung-Shik Lee (Yonsei University, Seoul, Korea). HA-Ub K48 (#17605) and HA-Ub K48R (#17604) were purchased from Addgene (Lim et al., 2005). Myc-tagged HMGB1 WT, and the HMGB1 C23S/C45S and C106S mutants were described previously (Youn and Shin, 2006). cDNA for HMGB2 was subcloned into a pCMV-Myc plasmid.

HEK293T cells were transiently transfected with various combinations of plasmids via Lipofectamine 2000 (Invitrogen, Camarillo, CA) following the manufacturer's instructions. For immunoblotting, cells were washed with PBS and lysed with 1× RIPA buffer (GenDEPOT, Barker, TX) containing 150 mM NaCl, 1% Triton X-100, 1% deoxycholic acid sodium salt, 0.1% SDS, 50 mM Tris-HCl pH 7.5, 2 mM EDTA, and a protease inhibitor cocktail. Whole-cell lysates (WCLs) were centrifuged at 20,000 g for 10 min at 4°C. A protein sample buffer (100 mM Tris-HCl pH 6.8, 2% SDS, 25% glycerol, 0.1% Bromophenol Blue and 5% β-mercaptoethanol) was added to the WCLs followed by heating at 94°C for 5 min. In all experiments, 30 μg of protein was separated by SDS-PAGE and transferred onto PVDF membranes (Millipore, Billerica, MA). The membranes were blocked with Tris-buffered saline (TBS) supplemented with 0.1% Tween 20 and 5% skimmed milk for 1 h. Anti-HMGB1 (1:3000; ab18256, Abcam, Cambridge, MA); anti-Beclin 1 (1:1000; NB500-249), anti-ATG5 (1:1000; NB110-53818), anti-ATG7 (1:1000; NBP2-24682), anti-ATG16L1 (1:1000; NB110-82384) (all Novus, Littleton, CO); anti-HA (1:3000; sc-805), anti-Sp1 (1:1000; sc-59) (both Santa Cruz Biotechnology, CA); anti-GAPDH (1:3000; YF-MA10022, Abfrontier, Seoul, Korea); anti-actin (1:5000; #4967, Cell Signaling Technology, Danvers, MA); anti-Ub (1:1000; MAB1510), anti-Ub K48 (1:1000; 05-1307) (both Millipore); anti-c-Myc (1:3000; Invitrogen, 13-2500); anti-LC3B (1:2000; L7543), and anti-Flag (1:4000; F3165, F7425) (both Sigma) antibodies were used. Membranes were washed three times and probed with a horseradish peroxidase (HRP)-conjugated secondary antibody for 1 h. Enhanced chemiluminescent (ECL) substrate (GenDEPOT) was used for visualization.

Immunoprecipitation was performed with 20 μl of Dynabead protein G (Invitrogen). The beads were washed three times with PBST buffer using DynaMag™-2 (Invitrogen) and incubated with 1 µg of the specific antibody for 1 h at room temperature. The beads were washed and incubated overnight with 200 μg of protein at 4°C followed by washing. Proteins were separated using SDS-PAGE.

To observe the effect of ICM on autophagy flux, HEK293T cells were transfected with a mRFP–GFP–LC3 plasmid and treated with 25 μM ICM for 4 h followed by starvation stress with EBSS for 2 h. Autophagy flux was measured by using confocal microscopy to count the puncta that were both GFP+ and mRFP+ (yellow dot, autophagosome) and those that were GFP− and mRFP+ (red dot, autolysosome) puncta.

To observe the subcellular localization of HMGB1 in HEK293T cells under starvation conditions, HEK293T cells were transfected with GFP-tagged HMGB1 and cultured on a cover glass in a six-well plate. After 48 h, cells were treated with 50 μM ICM for 6 h, 10 mM NH₄Cl for 2 h and then incubated in EBSS medium for 2 h. Cells were fixed with 4% paraformaldehyde-PHEM buffer for 30 min at room temperature and washed with cold PBS. After mounting with 4′,6′-diamidino-2-phenylindole (DAPI, Vectashield, Burlingame, CA), cell fluorescence was measured with a confocal FV1000 microscope (Olympus, Shinjuku-ku, Tokyo, Japan).

To measure the interaction between Beclin 1 and RNF216, HEK293T cells were transfected with plasmids encoding Flag-tagged Beclin 1 and RNF216 in LabTek chamber slides (Nunc, Rochester, NY). Cells were treated with 25 μM ICM for 4 h and fixed as described above. Cells were sequentially washed with 1% Triton X-100, 0.02% Tween 20, and 0.02% Tween 20 with 1% BSA in PBS. PLA was performed by using a Duolink-in situ kit (Sigma) following the manufacturer's instructions. Briefly, cells were treated with anti-Beclin 1 and anti-Flag antibodies overnight at 4°C followed by washing. Cells were incubated with a PLA probe mixture and ligation solution. For signal amplification, cells were incubated with a polymerase solution. After mounting with DAPI, fluorescence was visualized using a confocal microscope.

To measure nucleocytoplasmic translocation of HMGB1 and measure the interaction between Beclin 1 and RNF216 upon ICM treatment in HEK293T cells, nuclear and cytosolic parts were fractionated by using a nuclear–cytosolic fractionation kit (BioVision, Milpitas, CA). Briefly, cells were harvested and 200 μl of cytosol extraction buffer (CEB)-A was added. Samples were then incubated with 11 μl of ice-cold CEB-B, and the supernatant fraction (cytoplasmic fraction) was harvested after centrifugation. Pellets containing the nuclei were washed and resuspended in 100 μl of ice-cold nuclear extraction buffer followed by centrifugation.

Analyses were performed using one-way ANOVA or Student's t-test using GraphPad Prism5 software. Data represent the mean±s.d. for at least three individual experiments. Differences were considered statistically significant when P<0.05.

We thank Dr Seung Bum Park for providing ICM.