An alcoholic extract of Thuja orientalis L. leaves inhibits autophagy by specifically targeting pro-autophagy PIK3C3/VPS34 complex

Autophagy is a lysosome-dependent degradation program to maintain cellular homeostasis in response to a variety of stressful conditions, such as long-lived or non-functional subcellular organelles, protein aggregates, nutrient limitation, and virus/bacteria infection. Accordingly, dysregulation of autophagy is closely associated with many human pathophysiological conditions, such as neurodegenerative diseases, aging, and cancer, and autophagy is highlighted as an important therapeutic target for these human diseases. In autophagy process, PIK3C3/VPS34 complex plays important roles in autophagosome biogenesis. Accumulating evidences that inhibition of PIK3C3/VPS34 complex successfully blocks autophagy make the complex as an attractive target for the development of autophagy-specific inhibitors. However, considering that various forms of PIK3C3/VPS34 complex exist and they are involved in many different cellular functions, the targeting of the pro-autophagy PIK3C3/VPS34 complex is required to specifically inhibit autophagy. To identify autophagy inhibitors targeting the pro-autophagy complex, we have performed the screening of a customized natural product library consisting of 35 herbal extracts which are widely used in the oriental medicine as anti-inflammation and/or anti-tumor reagents. We discovered that an alcoholic extract of Thuja orientalis L. leaves inhibits pro-autophagy complex formation by disrupting the interaction between autophagy-specific factor, ATG14L, and the complex core unit Vps34-Beclin 1 in vitro. Also, it inhibits the nutrient starvation induced autophagy and diminished pro-autophagy PIK3C3/VPS34 complex containing either ATG14L or UVRAG in several cell lines. Our results strongly suggest that Thuja orientalis L. leave extract functions as an autophagy-specific inhibitor not decreasing the complex activity nor the protein level, but preventing protein–protein interaction between autophagy-specific factor (ATG14L and UVRAG) and PIK3C3/VPS34 complex core unit, Vps34-Beclin 1, thereby specifically depleting the pro-autophagy complex to inhibit autophagy.


Results
A screening of natural herbal medicinal extracts disrupting pro-autophagy PIK3C3/VPS34 complex by inhibiting ATG14L binding to the complex core unit, Vps34-Beclin 1. Although the underlying mechanism is still largely unknown, accumulating reports have shown that the medicinal effects of various natural herbal extracts used in oriental medicine are dependent on autophagy 30 . It makes the natural herbal medicinal extracts as an attractive repertoire for developing autophagy-specific modulators. Especially, considering the complexity of protein-protein binding interface, the natural extract may be a good starting resource to obtain PPI inhibitor to prevent ATG14L from binding to PIK3C3/VPS34 complex core unit, because a single small molecule may not good enough to interfere with PPI. For the inhibitor screening, we have selected 35 natural herbal medicinal extracts (E1-E35, Table 1), which are mostly used as anti-inflammatory, anti-tumor, or anti-bacterial/viral agents in oriental medicine, to make up the screening library. Considering the significance of autophagy in immune responses, these natural herbal extracts are expected to have a role in autophagy. In the inhibitor screening, the purified ATG14L protein was added into the immobilized PIK3C3/VPS34 complex core unit, Vps34-Beclin 1, to reconstitute pro-autophagy complex, Vps34-Beclin 1-ATG14L, in which natural herbal extract was added to examine the complex formation by ELISA against ATG14L and Beclin 1, respectively (Fig. 1a). When the extract specifically interferes with PPI between ATG14L and Beclin 1, ELISA signal for ATG14L will be decreased, but the Beclin 1 signal will not change. Therefore, the ratio of Beclin 1/ATG14L ELISA value will be increased. In the initial screening shown in Fig. 1b, ten natural extracts (E4, E8, E10, E12, E19, E21, E22, E29, E33, and E34) showed neither Beclin 1 nor ATG14L ELISA signals, suggesting that they might affect the complex core unit Vps34-Beclin 1 on the plate at 0.2 mg/ml concentration. Except them, thirteen extracts caused the decrease in ATG14L signal more than 80% (Fig. 1b), of which the nine extracts have changed Beclin 1 signal less than 50% decrease so that their Beclin 1/ATG14L ELISA signal ratio was greater than 3 (Fig. 1c). For these nine candidates from the first round screening, we have repeated the same experiment with them at fourfold diluted condition (0.05 mg/ml) to obtain more potent inhibitors. As shown in Fig. 2a, E5 (Forsythia koreana Nakai), E6 (Artemisia scoparia Waldst. & Kitam.), E14 (Morus bombycis Koidz.), E17 (Ampelopsis japonica Makino), and E18 (Thuja orientalis L.) still potently inhibited ATG14L binding more than 70%, and they have almost no effect on the complex core unit (less than 10% decrease in Beclin 1). To evaluate the kinetics for the inhibition of ATG14L binding to the complex core unit, Vps34-Beclin 1, the dose-dependent inhibition of ATG14L binding was tested and their inhibition curve was plotted to obtain the maximum inhibition (%) and IC50. As shown in Fig. 2b. All five natural herbal extracts inhibit ATG14L binding at maximum around 80% and their IC50 was in the range from 1 μg/ml (E5) to 13 μg/ml (E18).
An alcoholic extract of T. orientalis L. inhibits the starvation-induced autophagy in cells. Next www.nature.com/scientificreports/ vation is long-believed to induce autophagy to maintain cellular energy homeostasis 2 . The cells were starved with glucose, a primary energy source in cell culture medium, in which each extract was added. For overnight starvation, almost of all cells were dead in the presence of E5, E14, or E17 at the concentration of 0.2 mg/ml, suggesting their high toxicity in this experimental condition. In contrast, cell viability was little affected by either E6 or E18. As shown in Fig. 3a, glucose starvation induced autophagy as evidenced by the immunoblot showing the increase in LC3-II upon the starvation. In this condition, E6 has no effect on the starvation-induced autophagy, but E18 dramatically decreased the starvation-induced LC3-II, demonstrating that E18, an alcoholic extract of Thuja orientalis L. leaves (hereafter referred to as T. orientalis), functions as an inhibitor of autophagy in response to the starvation. Notably, under T. orientalis, both LC3-II in the absence or presence of a lysosome blocker, NH 4 Cl, was much lower than the control. NH 4 Cl suppresses autophagy flux to block LC3-II degradation in autolysosome, thereby accumulating LC3-II, but it was significantly suppressed by T. orientalis. Similarly, the inhibition of the basal autophagy flux by another lysosomal inhibitor, Chloroquine (CQ), was also significantly decreased by T. orientalis (Fig. 3b). These results collectively suggest that T. orientalis functions as an autophagy inhibitor in a manner different to the lysosome blockers and its inhibition may occur before lysosomal degradation, probably, at the initiation step regulated by PIK3C3/VPS34 complex as it was discovered as an inhibitor of PIK3C3/VPS34 complex formation in vitro. Therefore, T. orientalis is expected to decrease autophagy flux, but not block the flux. Inhibition of T. orientalis on the starvation-induced autophagy was increased in a dosedependent manner (Fig. 3c). Also, the inhibition of the starvation-induced autophagy was observed in several other cell lines, such as HEK293, NIH3T3, and human cancer cell line, HCT116 (Fig. 3d), suggesting that it may function as a general autophagy inhibitor at least upon the nutrient starvation. Next, we have tried to identify an active single component in T. orientalis required for the autophagy inhibition. In search of the literature analyzing the components of T. orientalis 31,32 , we have found that it was made up of more than 20 components, of which top five components more than 5% in the extract total were chosen 31 : α-pinene (~ 29%), Δ-3-carene (~ 20%), α-cedrol (~ 10%), β-caryophyllene (~ 8%), and α-humulene (~ 6%). We have treated these single components at the maximum concentration at which the cells were viable at least more than 50% in overnight treatment and examined their inhibition potency against the starvation-induced autophagy (Fig. 4a). Unfortunately, under the overnight starvation, although there were variations in cell viability, none of any single component was as effective as T. orientalis in the inhibition of the starvation-induced autophagy. In addition, top three components (pinene, carene, and cedrol) were mixed together to test their efficacy of autophagy inhibition, this mixture barely inhibited the starvation-induced autophagy (Fig. 4b). It suggests that inhibition effect of T. orientalis on the starvation-induced autophagy may rely on either the other component present in the smaller amount in the extract or the more complex combination of the extract components.
T. orientalis specifically decreases pro-autophagy PIK3C3/VPS34 complex containing ATG14L (or UVRAG) in cells. Finally, to examine inhibition mechanism of T. orientalis on autophagy, we have analyzed pro-autophagy PIK3C3/VPS34 complex in cells because it was originally identified as an inhibitor target- Table 1. List of 35 natural herbal medicinal extracts. All listed natural herbal extracts were prepared by a standard extraction protocol using either ethanol extraction or steam distillation. They are lyophilized and then dissolved in DMSO at the concentration of 20 mg/ml, which was diluted with the binding assay buffer or cell culture medium in the experiment as indicated. www.nature.com/scientificreports/ ing the formation of pro-autophagy complex, Vps34-Beclin 1-ATG14L, in vitro. To this end, the plasmids encoding HA-Vps34, Flag-Beclin 1, and HA-ATG14L, respectively, were transfected to HEK293 cells and Flag-Beclin 1 was immunoprecipitated. Co-immunoprecipitations (Co-IPs) of HA-Vps34 and HA-ATG14L were tested for their binding to Beclin 1 by their immunoblots. As shown in Fig. 5a, both HA-Vps34 and HA-ATG14L were co-immunoprecipitated by Flag-Beclin1 in control, indicating the formation of pro-autophagy Vps34-Beclin 1-ATG14L complex in this experimental setting. However, in the presence of T. orientalis, Co-IP of ATG14L was significantly decreased in the Beclin 1 immune-complex, whereas Co-IP of Vps34 was not affected. To further confirm that T. orientalis specifically inhibits Beclin 1-ATG14L binding, we carried out two different endogenous Vps34 and ATG14L Co-IP assay. First, we have transfected Flag-Beclin1 into HEK293 cells and then Co-IPs of the endogenous Vps34 and ATG14L in Flag-Beclin 1 immune-complex was examined by immunoblots (Fig. 5b). Consistent with the transfection result in Fig. 5a, almost the same amount of the endogenous Vps34 was coimmunoprecipitated regardless of T. orientalis, but Co-IP of ATG14L was significantly decreased in T. orientalis treatment condition. Second, we have conducted Co-IP assay targeting the endogenous Beclin 1 in both MEFs and HEK293. Consistently, we obtained the same Co-IP results for the endogenous Vps34 and ATG14L in the endogenous Beclin 1 immune-complex from MEFs (Fig. 5c), in which only Co-IP of the endogenous ATG14L was decreased by T. orientalis. We did not obtain a reliable Co-IP result for ATG14L in the endogenous Beclin 1 2 mg/ml concentration (Each extract was originally dissolved in DMSO at 20 mg/ml and is diluted to 1:100 in assay buffer) and ELISA assays for Beclin 1 and ATG14L, respectively, were performed (b) and the resulting ELISA signal ratio between Beclin 1 and ATG14L was represented (c). The relative amount of ATG14L or Beclin 1 bound to the immobilized Vps34 shown in (b) is shown as % of control under 1% DMSO. A red line in (b) and (c) represents an arbitrary cut-off for the inhibitor efficiency: more than 80% ATG14L binding (b) and more than 3 in the ELISA signal ratio between Beclin 1 and ATG14L (c). * denotes the candidates that passed the cutoff. www.nature.com/scientificreports/ immune-complex from HEK293, but we have obtained very clear Co-IP of the endogenous UVRAG in this cell (Fig. 5d). Like as ATG14L, Co-IP of the endogenous UVRAG was dramatically decreased upon T. orientalis, but Co-IP of the endogenous Vps34 was not. Considering that ATG14L and UVRAG compete with Beclin 1 binding and their binding regions on Beclin 1 are closely overlapped, it is expected that T. orientalis can disrupt both ATG14L and UVRAG containing PIK3C3/VPS34 complex. These results collectively support that T. orientalis specifically target pro-autophagy PIK3C3/VPS34 complex containing either ATG14L or UVRAG by disrupting these autophagy-specific subunit binding to Beclin 1, but not the interaction of the complex core unit, Vps34-Beclin 1, as proposed in in vitro complex formation screening.

Discussion
In this study, we have identified an alcoholic extract of T. orientalis L. leaves as an autophagy-specific inhibitor by targeting pro-autophagy PIK3C3/VPS34 complex containing either ATG14L or UVRAG. We have screened a customized natural product library consisting of 35 herbal medicinal extracts with anti-inflammation, anti-tumor, or anti-bacterial/viral activities, which are closely related to pathophysiological roles of autophagy. An in vitro pro-autophagy Vps34-Beclin 1-ATG14L complex formation assay was carried out in the presence of each natural herbal extract and then, the selected hit extracts were tested their autophagy inhibition activities in cells. Among 35 extracts, T. orientalis successfully inhibited the starvation-induced autophagy and specifically disrupt proautophagy PIK3C3/VPS34 complex formation in both in vitro and cells. It specifically targets Beclin 1-ATG14 (or UVRAG) binding, but not the complex core unit, Vps34-Beclin 1, binding. These results indicate that T. orientalis could be an autophagy inhibitor specifically disrupting the pro-autophagy PIK3C3/VPS34 complex by preventing autophagy-specific factors, ATG14L or UVRAG, from the binding to the complex core unit, Vps34-Beclin 1. It is important because the complex core unit Vps34-Beclin 1 is a backbone of the complex, in which Beclin 1 recruits many different subunits as a scaffold subunit to form various PIK3C3/VPS34 complexes in a wide spectrum of cell physiology, such as vesicle trafficking, endocytosis, and amino acid signaling 23 . Therefore, a targeting a proautophagy PIK3C3/VPS34 complex by T. orientalis may result in the regulation of one specific function of various PIK3C3/VPS34 complexes, autophagy, but not the other functions. In this sense, it is different to other PIK3C3/ VPS34 inhibitors, such as 3-MA and Spautin-1, both of which target the complex core unit. 3-MA is a catalytic inhibitor of Vps34 11 and Spautin-1 targets Beclin 1 degradation 21 , thereby they may affect all PIK3C3/VPS34 complex family members. It should be noted that T. orientalis disrupts both ATG14L-and UVRAG-containing pro-autophagy PIK3C3/VPS34 complex. ATG14L and UVRAG participate in the complex by binding to Beclin 1 in a mutually exclusive manner, suggesting that their binding sites on Beclin 1 are overlapped. Therefore, although T. orientalis was initially identified to disrupt Beclin 1-ATG14L binding by an in vitro Vps34-Beclin 1-ATG14L complex formation assay, it is expected that it also affects the UVRAG-containing complex. Indeed,   www.nature.com/scientificreports/ known to play an important role in the maturation of autophagosome as well as endosomal trafficking, whereas ATG14L-containing one is believed to mainly function in the initiation of autophagosome formation 33,34 . Therefore, the fact that T. orientalis also disrupt Vps34-Beclin 1-UVRAG complex may result in its undesirable effects in endolysosomal pathway. Notably, the treatment of T. orientalis appears to slightly decrease the protein level of ATG14L (Fig. 5a,b), therefore, it is possible that the reduction of Vps34-Beclin 1-ATG14L complex in this condition may result from the decrease in ATG14L protein level, but not the inhibition of Beclin 1-ATG14L binding. However, it is also possible that free ATG14L that do not participate in the complex by T. orientalis is unstable to be degraded, decreasing the protein level. One more caution to use T. orientalis as an autophagy inhibitor is that it may not work in PIK3C3/VPS34 independent (or bypass) autophagy. Also, if there is any mutation in Beclin1-ATG14L (or UVRAG) binding interface or surrounding regions, this protein-protein interaction inhibitor may be no longer functional in the pro-autophagy PIK3C3/VPS34 complex containing them, thereby no effect on autophagy. Unfortunately, we could not assign a single active component in the extract of T. orientalis for autophagy inhibition. It may cause some trouble in developing it as a drug, but the extract of T. orientalis has www.nature.com/scientificreports/ been long and extensively used in the oriental medicine, therefore, it will provide an important clue to develop an autophagy-specific inhibitor as a therapeutic reagent for human diseases. To express pro-autophagy PIK3C3/VPS34 complex in cells, the mammalian expression plasmids encoding Vps34, Beclin 1, and ATG14L, respectively, with appropriate N-terminal tags were transfected into HEK293 cells by polyethylenimine (PEI, SIGMA) method as previously reported 35 . After 2 day incubation, the transfected HEK293 cells were washed with ice-cold PBS and lysed in Mild Lysis Buffer (MLB, 10 mM Tris, pH7.5, 2 mM EDTA, 100 mM NaCl, 1% NP-40, 50 mM NaF, 1 mM Na3VO4, and protease inhibitor cocktail) to harvest the total cell lysates for protein purification, immunoblots and immunoprecipitation. Also, the total cell lysates for endogenous protein analysis was also obtained with MLB as above.

Methods
Preparation of alcoholic extracts of T. orientalis L. leaves. The commercially available air-dried T.
orientalis L. leaves were purchased from the Korean traditional medicine market (Seoul Yangnyeong market).
A detailed preparation procedure is demonstrated in elsewhere 32 . Briefly, the dried leaves of T. orientalis L. have been extracted in 70% ethanol with stirring for 24 h at room temperature. The resulting alcoholic extract was filtrated, concentrated in a rotary vacuum evaporator under the heating at 50 °C, and finally lyophilized to the powder. The powder was dissolved in DMSO and kept on − 20 °C.
Immunoblotting and immunoprecipitation. Protein samples were resolved on SDS/PAGE and then, electrophoretically transferred onto a PVDF membrane. The membranes were blocked with TBST buffer containing 5% skim milk at room temperature for 1 h. The membranes were incubated with a primary antibody against the indicated proteins, which is diluted in TBST buffer containing 1% BSA to 1:1000 at 4 °C overnight, followed by the reaction with HRP-conjugated secondary antibodies at room temperature for additional 1 h. Antibody detection was carried out using ECL detection kit (PIERCE). For immunoprecipitation, 0.5-1 mg of total cell lysates reacts with the antibody-coupled bead complex, which was prepared by incubation of a specific target antibody with protein A (or G)-coupled Agarose in TBST containing 5% BSA at 4 °C overnight. The immunoprecipitation reaction was carried out at 4 °C for 3-5 h with a gentle rocking. Unbound proteins were removed by an extensive washing with MLB and the immune-complex was analyzed by immunoblots against the antibody recognizing the protein of interest. www.nature.com/scientificreports/