Identification of an Annonaceous Acetogenin Mimetic, AA005, as an AMPK Activator and Autophagy Inducer in Colon Cancer Cells

Annonaceous acetogenins, a large family of naturally occurring polyketides isolated from various species of the plant genus Annonaceae, have been found to exhibit significant cytotoxicity against a variety of cancer cells. Previous studies showed that these compounds could act on the mitochondria complex-I and block the corresponding electron transport chain and terminate ATP production. However, more details of the mechanisms of action remain ambiguous. In this study we tested the effects of a set of mimetics of annonaceous acetogenin on some cancer cell lines, and report that among them AA005 exhibits the most potent antitumor activity. AA005 depletes ATP, activates AMP-activated protein kinase (AMPK) and inhibits mTOR complex 1 (mTORC1) signal pathway, leading to growth inhibition and autophagy of colon cancer cells. AMPK inhibitors compound C and inosine repress, while AMPK activator AICAR enhances, AA005-caused proliferation suppression and subsequent autophagy of colon cancer cells. AA005 enhances the ATP depletion and AMPK activation caused by 2-deoxyglucose, an inhibitor of mitochondrial respiration and glycolysis. AA005 also inhibits chemotherapeutic agent cisplatin-triggered up-regulation of mTOR and synergizes with this drug in suppression of proliferation and induction of apoptosis of colon cancer cells. These data indicate that AA005 is a new metabolic inhibitor which exhibits therapeutic potentials in colon cancer.


Introduction
In tumor cells there is an increased glycolytic pathway enzymes and glucose transporters even in the presence of a high O 2 concentration, ultimately leads to an elevated ATP production rate [1]. Clinical evidence has linked cell metabolism with cancer outcomes, and reprogramming energy metabolism has been approved to be an emerging hallmark of cancer [2]. These observations have raised interest in targeting energy metabolism for cancer therapy for both hypoxic (glycolytic) and oxidative tumors [1], albeit also concerns that these therapies would have unacceptable effects on normal cells. Intriguingly, some of the first cancer therapies targeting specific metabolic needs of cancer cells remain effective in the clinic today, re-inspiring efforts to target metabolic dependencies of cancer cells as a selective anticancer strategy [3].
AMP-activated protein kinase (AMPK) which exists in cells as a heterotrimeric complex composed of a catalytic kinase subunit (a) and two regulatory subunits (b and c), is a sensor of energy status that maintains cellular energy homeostasis. It is activated by a fall in ATP (concomitant with a rise in ADP and AMP) or stimuli that increase the cellular AMP/ATP ratio, resulting in the activation of catabolic pathways and the inhibition of anabolic pathways [4,5]. Essential to activation of AMPK is its phosphorylation at Thr-172 by an upstream AMPK kinases (AMPKKs) [6] and tumor suppressor LKB1 [7] which is a serine/threonine kinase associated with gastrointestinal polyposis and cancer [8] and lung cancer [9]. AMPK phosphorylates two rate-limiting enzymes in fatty acid and cholesterol synthesis: acetyl-CoA carboxylase (ACC) and HMG-CoA reductase, as well as other downstream targets, culminating in the inhibition of anabolic pathways and the activation of catabolic pathways [5]. AMPK activation directly limits translational initiation and protein synthesis [10], through inhibition of translation elongation factor 2 (EF2) [11], and indirectly through TSC2, leading to suppression of mammalian target of rapamycin (mTOR) [12] which can phosphorylate and activate p70 S6 kinase and 4E-binding protein (4EBP) [13].
AMPK activation by AMP analog 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) accumulates cyclin dependent kinase inhibitors p21 and p27 and down-regulates cyclin D1 in human hepatocellular carcinoma cells, leading to cell cycle arrest at G1 phase [14]. Population studies provide clues that the use of metformin which is an AMPK activator, may be associated with reduced incidence and improved prognosis of certain cancers [15,16]. In breast cancer, metformin exerts inhibitory effects via inhibition of mTOR-dependent translation initiation [17,18]. Metformin inhibits proliferation, decreases cell viability and blocks cell cycle in G1 phase in prostate cancer cells, and in vivo treatment with metformin leads to a significant reduction of tumor growth in mice bearing xenografts of prostate cancer cells [19]. Metformin and AICAR induce apoptosis and suppresses the tumor growth of colon cancer line HCT116 p53(2/2) xenografts, and trigger autophagy of HCT116 p53(+/+) cells [20]. AICAR and protein folding inhibitor 17-AAG, especially when combined, show efficacy against aneuploid human cancer cell lines [21]. These results indicate that AMPK could be a rational drug target and lead compounds should be identified or designed for the development of therapeutic avenues for cancers.
Annonaceous acetogenins represent a class of naturally occurring polyketides isolated from various species of the plant genus Annonaceae [22]. These compounds exhibit diverse bioactivities, including promising cytotoxicites and antiparasitic activities [23]. Though studies show that annonaceous acetogenins can disrupt mitochondrial function through blocking mitochondria complex I and ubiquinone-linked NADH oxidase [23], and bind the third matrix-side loop of ND1 subunit in mitochondrial NADHubiquinone oxidoreductase [23,24], the mechanisms of action of these compounds in fighting cancer remain largely unknown. The chemistry group of our team had designed and synthesized a series of annonaceous acetogenin mimetics [25][26][27][28]. In this work we tested the biological activity of some new analogs and investigated the mechanisms underlying annonaceous acetogenins' cytotoxicities, and reported that a mimetic AA005 which showed potent and selective inhibitory activities against a variety of cancer cells, was able to activate AMPK and induce cell cycle arrest followed by autophagy, demonstrating its therapeutic potentials.

Cell Culture and Transfection
The lung cancer cell line A549, breast cancer cell line MCF-7, cervical cancer cell line HeLa, colon cancer cell lines LOVO, SW480, HCT116 and HT29, and human embryonic kidney HEK-293T cells were obtained from the American Tissue Culture Collection (ATCC). Human embryonic lung fibroblast MRC-5, gastric cancer cell line SGC7901, and hepatic cancer cell line BEL7402 were purchased from the Cell Resource Center, Chinese Academy of Medical Sciences (Beijing). The normal human bronchial epithelial cells (HBEpiC) were purchased from ScienCell (ScienCell Research Laboratories, San Diego, California). The 293T, A549, BEL7402, HeLa, MCF-7, SGC7901 and the human normal bronchial epithelial cells BEAS-2B [29] cells were cultured in Dulbecco modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS; Gibco/BRL, Grand Island, NY), 100 U/ml penicillin and 100 mg/ml streptomycin. LOVO, SW480, HCT116 and HT29 cells were cultured in DMEM/F12 supplemented with 10% FBS, 100 U/ml penicillin and 100 mg/ml streptomycin. MRC-5 cells were cultured in MEM/EBSS medium supplemented with non-essential amino acids, 10% FBS, 100 U/ml penicillin and 100 mg/ml streptomycin. HBEpiC cells were cultured in a serumfree bronchial epithelial cell medium (ScienCell Research Laboratories) containing bronchial epithelial cell growth supplement (ScienCell Research Laboratories). The pQCXIP-GFP-LC3 and pQCXIP-GFP plasmids [30] were transfected into LOVO cells using the Lipofectamine 2000 (Qiagen) according to the recommended protocol by the manufacturer.

Cell Viability, Cell Proliferation and Clonogenic Assays
Cancer cells (5610 3 ) were seeded in each well of 96-well tissue culture plates (Coaster, Charlotte, NC) and treated with annonaceous acetogenin mimetics for 48 h at 37uC in a 5% CO 2 atmosphere. MTT assay was performed as described [31], and the IC 50 values were calculated using the CalcuSyn software (version 2.0, Biosoft, Cambridge, UK). Cell viability was estimated by trypan blue dye exclusion. The potential synergistic, additive or antagonistic effect between AA005 and 2-DG or cisplatin was carefully assessed using the Calcusyn Software (Biosoft, Cambridge, UK) as described [32]. The dose-effect curves of single or combined drug treatment were analyzed by the median-effect method [33], where the combination indexes (CI) less than, equal to, and greater than 1 indicate synergistic, additive, and antagonistic effects, respectively.
For foci formation, HT29, LOVO or HCT116 treated with AA005 were seeded in triplicate into 35 mm plates (200 cells per plate). After 8 days of culturing, cells were stained with Giemsa and clones containing more than 50 cells were counted [29].

Analysis of Cell Cycle and Apoptosis
To detect the cell cycle distribution, colon cancer cells were synchronized to G1/S boundary by a double thymidine block [31] and then exposed to AA005 at indicated concentrations for 24 h. Cells were harvested, fixed with 70% cold ethanol in 4uC overnight. The cells were centrifuged and washed with PBS, followed by incubation with RNase and propidium iodide (PI) (Sigma-Aldrich). Cell cycle distribution was analyzed by flow cytometry (BD FACS Vantage Diva, USA). Cell apoptosis was measured using PE Annexin V/PI Apoptosis Detection kit (BD Biosciences, San Jose, CA) according to manufacturer's instruction.

Analysis of Cells with GFP-LC3 Vesicles
Cells were transfected with pQCX-IP-GFP-LC3 and pQCX-IP-GFP-expressing plasmids for 24 h, and then treated with various concentrations of AA005 for another 24 h. Cells were fixed with 4% paraformaldehyde/PBS for 15 min at room temperature and analyzed using confocal microscopy at 636 magnification. The percentage of GFP-positive vesicles cells was assessed [30].

Measurement of Mitochondrial Transmembrane
Potential, ATP Content and NAD + /NADH Ratio Cells were treated with different concentrations of annonaceous acetogenin mimetics for 24 h. The mitochondrial transmembrane potential was examined as described [34], ATP content was

Confocal Microscopy Analyses
Cells were grown on coverslips and fixed in 4% paraformaldehyde. After a brief washing in PBS supplemented with 100 mM glycine, slides were blocked with 5% bovine serum albumin (BSA; Sigma) and 0.3% Triton X-100 in PBS for 30 min at room temperature, and stained for fluorescence microscopy as described [31]. Cells upon AA005-flu were examined by use of a Zeiss LSM 510 META microscope equipped with a 636 oil-immersion objective. Image processing and analysis were done with Zeiss LSM 510 software version 3.2, ImageJ Version 1.42 (National Institutes of Health), and Adobe Photoshop Version 7.0 (Adobe Systems).

Statistical Analysis
All experiments were repeated at least three times and the data are presented as the mean6SD unless noted otherwise. Differences between data groups were evaluated for significance using Student t-test of unpaired data or one-way analysis of variance and Bonferroni post-test. P values,0.05 were considered statistically significant.

Structure activity Relationship (SAR) Analysis of Annonaceous Acetogenin Mimetics
A serial annonaceous acetogenin mimetics had been synthesized by replacement of both tetrahydrofuran (THF) rings of natural bullatacin with a simple diethylene glycol ether unit by the chemistry group of our team [25][26][27][28]. Nine novel analogs were synthesized in this work, and we found that the mimetics AA005 [25] and AA093, a compound with the introduction of a 10hydroxy group onto AA005, represented the two compounds which had the lowest IC50 values in this setting ( Table 1). The observation that compound AA090 and AA091 lacking the right lactone unit and left 11-carbon tail exhibited a 17 to .170 fold lower cytotoxicity in cancer cells suggested that these groups are essential for the cytotoxicity of AA005. AA101 with an additional lactone unit embedded in the left hydrocarbon chain part exhibited a 23-142 fold lower cytotoxicity in cancer cells, further confirming the importance of the long hydrophobic tail and the right terminal lactone in the mimicry. Adding a middle ether unit to the mimetics (compounds AA102-105) slightly enhanced their anti-proliferative activity as compared to AA101, suggesting that a diethylene glycol ether unit is essential for the anti-proliferative activity. The diverse biological activity of these mimetics indicates that the structural analogs may not be functional analogs.

Inhibitory Effects of AA005 on Cancer Cells
Because AA005 was the most potent cytotoxic agent among these mimetics, we further tested its effects on 11 human cancer cell lines and 4 noncancerous cell lines (HBEpiC, MRC5, HLF and 293T), and found that AA005 showed diverse effects on cancer cells in that it had potent inhibitory effect on colon (HCT116, HT29, LOVO and SW480), gastric (SGC7901), hepatic (BEL7402), lung (A549) and breast (MCF7) cancer lines, and weak effect on cervical (HeLa) cancer cells ( Figure 1A). AA005 exhibited inhibitory effects on HCT116 ( Figure 1B), HT29 ( Figure 1C) and LOVO ( Figure 1D) cells in a dose-and timedependent fashion. Interestingly, AA005 showed an even weaker activity against noncancerous (HBEpiC, MRC5, HLF, BEAS-2B and 293T) cells ( Figure 1A and Table 1). These results indicate that the relative selective inhibitory effects of AA005 on cancer cells warrant further investigation.

AA005 Suppresses Cell Proliferation and Colony Forming Activity of Colon Cancer Cells
We further analyzed the effects of AA005 on colon cancer cells. By using the trypan blue exclusion analyses, we showed that treatment with AA005 at 50 to 200 nM for 24 to 48 h markedly inhibited proliferation of HT29, LOVO and HCT116, but not HBEpiC or BEAS-2B cells (Figure 2, A and B). Foci formation assay showed AA005's potent inhibitory effects on colony forming activity of colon cancer cells ( Figure 2C). We analyzed the effects of AA005 on cell cycle and found that AA005 caused a substantial increase in the percentage of colon cancer cells in G1 phase in a dose-dependent fashion ( Figure 2D).

AA005 Targets Mitochondria, Depletes ATP and Activates AMPK in Colon Cancer Cells
Fluorescein-labeled AA005 (AA005-flu, Figure 3A) was successfully accomplished by a biological activity assessmentaided protocol after examining a number of potential derivative positions in parallel [36]. AA005-flu was found to exhibit similar cell selectivity to its parental molecule, and accumulate in the mitochondria of hepatic cancer but not normal cells [36]. By using immunofluoresence confocal microscopy analysis, we demonstrated that AA005-flu could co-localize with mitochondria in HT29, HCT116 and LOVO cells ( Figure 3B). However, AA005-flu signal was very weak in mitochondria of HBEpiC or 293T cells ( Figure 3B). While AA005-flu inhibited proliferation of LOVO, HT29 and HCT116 cells in a dose-dependent fashion, its cytotoxic effect on HBEpiC was weak ( Figure 3C). Furthermore, we reported that AA005 could increase Rhodamine 123-negative fractions in HT29 and LOVO but not HBEpiC cells ( Figure 3D). These results suggest that AA005 might target mitochondrial molecules and thus perturb energetic pathway of cancer cells.
Mitochondria targeting agents (mitocans) are prone to disrupting oxidation phosphorylation pathway and reducing NAD + / NADH ratio, resulting in inhibition of ATP production and activation of AMPK that is reflected by phosphorylation and inactivation of Acetyl-CoA Carboxylase (ACC) (Ser79) which is an indicator for AMPK activity [37]. We tested the effect of AA005 on cellular NAD + /NADH ratio and ATP content, and found that in HT29 and LOVO cells, treatment with AA005 at 50 to 200 nM for 24 h reduced NAD + /NADH ratio ( Figure 3E) and depleted ATP in a dose-dependent manner ( Figure 3F), while interestingly,   AA005 could not significantly decrease NAD + /NADH ratio and ATP content in HBEpiC cells ( Figure 3E and F). Furthermore, AA005 markedly up-regulated p-AMPK and p-ACC in a doseand time-dependent fashion ( Figure 3G). However, these phenomena were not observed in HBEpiC cells ( Figure 3G).
We showed that AA005 as well as the mitocans AICAR and rotenone caused ATP depletion in HT29 and LOVO cells, while AMPK inhibitors compound C and inosine attenuated this effect ( Figure 4A). AA005, AICAR and rotenone up-regulated p-AMPK and p-ACC, while compound C and inosine reduced AA005triggered up-regulation of the two molecules ( Figure 4B). In addition, while AA005 inhibited proliferation of HT29 and LOVO cells, compound C and inosine reduced this effect ( Figure 4C). These results suggest that AMPK activation is required for AA005-induced cytotoxicity to colon cancer cells.

AMPK Activation Mediates AA005-induced mTOR Complex 1 (mTORC1) Inhibition in vitro
Activated AMPK phosphorylates TSC2 at Thr-1227 and Ser-1345 and increases the activity of TSC1-TSC2 complex to inhibit mTOR [12]. We tested the effects of AA005 on mTORC1, and reported that AA005 decreased p-mTOR and p-S6K (p85 and p70 S6K) in LOVO and HT29 but not HBEpiC cells ( Figure 3G). AICAR and rotenone also down-regulated p-mTOR in HT29 and LOVO cells ( Figure 4B). However, AA005-induced p-mTOR down-regulation could be partially repressed by inosine and compound C ( Figure 4B). Compound C and inosine also partially attenuated cyclin D1 down-regulation caused by AA005 ( Figure 4B). AMPK/mTOR Signaling is Involved in AA005 Induced Autophagy in vitro AMPK activation can induce autophagy via inhibition of mTOR (49). We tested whether AA005 could induce autophagy in colon cancer cells or not by detecting the changes of cytosolic form (LC3-I) and lipidated form (LC3-II) of the autophagy marker LC3. Interestingly, we found that AA005 induced accumulation of LC3-II in LOVO cells in a time-and dose-dependent fashion ( Figure 5A). The pQCXIP-GFP-LC3 plasmid was transfected into LOVO cells which were then treated with AA005 at 100 nM for 24 h, followed by confocal microscopy assessment. We showed that while control cells displayed a diffuse staining, LOVO cells upon AA005 or mTOR inhibitor rapamycin (100 nM) exhibited a speckled fluorescent staining pattern, indicating the redistribution of LC3 to autophagosomes ( Figure 5B). Interestingly, inosine attenuated AA005-caused formation of LC3 autophagosomes ( Figure 5B) and accumulation of LC3-II ( Figure 5C). These results indicate that AA005 induces autophagy of colon cancer cells via AMPK/mTOR signaling pathway. However, treatment with AA005 at 50-200 nM for 24 h or 100 nM for 12-48 h did not result in marked apoptosis of LOVO cells, reflected by Annexin V/PI staining and flow cytometry assessment ( Figure 5D) or Western blot analysis of cleavage of PARP, a substrate of activated casp-3 ( Figure 5E).

AA005 Synergizes with 2-Deoxyglucose and Cisplatin in Inhibiting Colon Cancer Cell Proliferation by Modification of AMPK and mTOR
2-Deoxyglucose (2-DG) is a synthetic glucose analogue capable of inhibiting glycolysis and ATP production [38]. We tested the combined effects of AA005 and 2-DG in LOVO cells, and found that combined use of the two agents exerted synergism in inhibition of cell proliferation ( Figure 6A and B) and suppression of ATP generation ( Figure 6C). AA005 in combination with 2-DG led to enhanced up-regulation of p-AMPK and p-ACC, and down-regulation of p-mTOR and CDK4 ( Figure 6D).
Chemotherapeutic agent cisplatin can up-regulate mTOR survival pathway which confers drug resistance to cancer cells [39]. We showed that while AA005 and cisplatin caused synergistic inhibitory effect on LOVO cell proliferation ( Figure 6E and F), AA005 decreased cisplatin-triggered upregulation of mTOR and S6K ( Figure 6G). Combined use of AA005 and cisplatin led to enhanced apoptotic effect on LOVO cells, reflected by Annexin V/PI staining and flow cytometry assessment ( Figure 6H) or Western blot analysis of cleavage of PARP ( Figure 6G). However, combination of these two agents did not cause synergy in depletion of ATP content ( Figure 6I), cell cycle arrest ( Figure 6J), or autophagy (reflected by the expression LC3-II, Figure 6G) in LOVO cells.

Discussion
Annonaceous acetogenins represent a series of C-35/C-37 natural products with hydroxylated THF and the c-lactone rings structures [25,26,28]. Many members of this family display cytotoxic activity by perturbation of the terminal electron transfer step in complex I of mitochondria [40]. AA005 is a mimetic of annonaceous acetogenin in which both the THF rings are replaced by an ethylene glycol ether unit. AA005 retains the essential functionalities of the natural acetogenins and shows more powerful biological activity [41]. In this study, we report that AA005 is able to activate AMPK and inhibit mTOR, therefore arrests cell cycle at G1 phase and induces autophagy of colon cancer cells. AA005 synergies with glycolysis inhibitor 2-DG for marked ATP generation blockade, and antagonizes cisplatininduced up-regulation of p-mTOR, leading to enhanced proliferation inhibition and apoptosis induction in colon cancer cells. These results indicate that AA005 bears therapeutic potentials at least in this kind of malignant neoplasm.
Targeting cancer cell metabolism, especially glycolysis inhibition, has emerged as a new promising strategy to fight cancer [42][43][44]. Mitochondria not only manage energy generation via citric acid cycle (tricarboxylic acid cycle, TCA cycle), but also play a key role in apoptosis regulation through release of cytochrome C. Although there has much debate on whether mitochondria have defects in oxidative phosphorylation pathway, it could be a potential target for cancer therapy [42,45]. Some mitocans such as metaformin and vitamin E analogues which inhibit complex I and II show effective and selective anti-cancer activity. These agents induce cancer cell death through generation of superoxide and mitochondrial destabilization [46]. ATP could also be depleted to some extent upon treatment of mitocans [42,45]. Mitochondria complex I has been shown to be targeted by annonaceous acetogenin [40]. We report that AA005 can co-localize with mitochondria in colon cancer but not HBEpiC or 293T cells ( Figure 3A), suggesting that AA005 may reduce NAD + /NADH ratio and ATP production in cancer cells. This possibility is confirmed in LOVO, HT29 and HCT116 cells ( Figure 3C). However, why mitochondria in cancer cells are more prone to be targeted by AA005 remains an open question. Inhibition of mitochondria by metaformin and rotenone may lead to activation of AMPK [42]. We show that AA005 can also suppress mitochondria and activate AMPK (Figure 3). At this point, inosine and compound C antagonize AA005-induced AMPK activation and cytotoxicity (Figure 4), indicating a direct link between ATP depletion through mitochondria inhibition and AMPK activation, which ultimately results in mTOR inhibition. AA005 also synergizes with 2-DG in ATP depletion and AMPK activation, indicating that AA005 bears therapeutic potentials for colon cancer.
The oncoprotein mTOR suppresses while activation of AMPK promotes autophagy [47]. We show that AA005 down-regulates p-mTOR as well as its downstream effector S6K (Figure 3), while compound C and inosine partially reverse this effect ( Figure 4). Inosine also represses AA005-induced autophagy of LOVO cells reflected by decrease LC3-II accumulation ( Figure 5C). While cisplatin up-regulates mTOR, AA005 attenuates this phenomenon ( Figure 6G) and synergizes with this agent in inducing apoptosis of LOVO cells ( Figure 6H). These results indicate that AA005 can inhibit mTOR via activation of AMPK, and further demonstrate the benefits of this annonaceous acetogenin mimetic for colon cancer cells.