RETRACTED ARTICLE: Curcumin synergizes with 5-fluorouracil by impairing AMPK/ULK1-dependent autophagy, AKT activity and enhancing apoptosis in colon cancer cells with tumor growth inhibition in xenograft mice

Colon cancer is one of the most common malignancies in human worldwide [1]. 5-Fluorouracil (5-Fu), a fluoropyrimidine analog, is chemotherapeutic agent widely used for the treatment of this cancer type [2]. While the non-specific cytotoxicity narrows its clinical therapeutic index with small differences between therapeutic and toxic doses, therapeutic resistance of 5-Fu is often occurred and results in poor outcome for the patients [3]. Although the combinational use of 5-Fu with other agents such as oxaliplatin, irinotecan or bevacizumabhas has significantly improved the prognosis and clinical benefits [4, 5], there remains a critical need for better understanding of molecular basis that accounts for the chemotherapeutic resistance, and hereby to uncover novel therapeutic strategies for extending survival while decreasing resistance and increasing therapeutic window in colon cancer patients.

Cancer cells trigger multiple signaling to escape from the cytotoxicity of chemotherapeutics. Autophagy, as a route of programmed cell death, has been increasingly studied in cancer therapy [6], and is thought to contribute to autophagic cell death via lysosomes-related cell degradation [7]. On the other hand, autophagy could promote tumor progression by providing metabolic fuel for cell survival when encountered environmental stressors such as nutrient starvation, hypoxia or treatment with chemotherapeutic agents [8‐10]. Such a “double-edged sword” role of autophagy in cancer is dependent on tumor cell types and their specific microenvironment. Nevertheless, evidence has been accumulated in support of the notion that autophagy could be an important cellular mechanism towards chemoresistance in various malignancies [11, 12]. Modulation of autophagy could be therefore a promising new strategy to overcome chemoresistance in cancer therapy.

Curcumin is well-known to be the main active component responsible for the majority of the medicinal properties of turmeric. In addition to otherwise described, curcumin has increasingly attracted scientific and clinical interests due to its wide spectrum of pharmacological activities upon multiple biological targets in preventing tumor initiation, progression, and dissemination in a number of human cancers [13, 14]. Moreover, the neglectable toxicity makes curcumin a very suitable adjuvant in disease, including cancer treatment [15]. Indeed, curcumin could act as cancer-specific chemosensitizer, presumably via induction of autophagic signaling pathways [16]. To the context of gaining insights into novel therapeutic strategies, however, the precise therapeutic effect of curcumin, especially under circumstance of chemo-related resistance, on autophagy in determining tumorous cells’ fate remains unclear. Thus, this study was designed to investigate the differential modulations of the treatments either with 5-Fu alone or 5-Fu combined with curcumin on cellular autophagic responses and viabilities in the human colon cancer cells HCT116 and HT29, and then to further explore if such autophagic responses could be attributed to curcumin-mediated changes on Akt/mTOR/ULK1 and AMPK-ULK1 signal transductions and hereby potentiate 5-Fu’s cytotoxicity in vitro and in vivo.

Serious toxicity/side-effect and therapeutic resistance are considered to be two major obstacles for a successful cancer chemotherapy translated from bench to bed [17]. This may be due to the scarce knowledge of cancer cell signaling network and bypass mechanisms. As an effort among others that deserves attention, a novel combinational strategy with aims to reverse chemoresistance has recently been proposed, in which an adjuvant drug is sequentially added to target resistance caused by major chemotherapy [18]. The present study demonstrates a unique combination of curcumin and 5-Fu, which though has been previously documented [19], to assess if and how curcumin could enhance the anti-cancer efficacy of 5-Fu in human colon cancer cells.

It is well-known that curcumin is efficient and safe for the prevention and treatment of varied pathological conditions, including cancer [20], despite its clinical benefits of curcumin are still limited due to its poor pharmacokinetic properties and bioavailability in vivo [21]. Given the fact that curcumin induces apoptosis via pleiotropic mechanisms in various cancer types [20, 22], several studies demonstrated the high interest of using curcumin as potent chemosensitizer to improve the therapeutic effects of cisplatin, mitomycin C, γ-Radiation, and other chemotherapeutics [23‐25]. Particularly, co-treatment of curcumin and 5-Fu, both at a single dose-level, promoted chemosensitivity in colon carcinoma cells [26]. In the present study, with initial observations of 5-Fu-resistance in HCT116 and HT29 cells (Fig. 1a, Additional file 1: Figure S1A), we investigated the differential sensitizing effects of curcumin on the cytotoxicity of 5-Fu by taking three different combinations, i.e., pre-treatment of curcumin followed by 5-Fu (pre-Cur), co-treatment of two agents (5-Fu + Cur) and post-treatment of curcumin following 5-Fu (pre-5-Fu). The cellular viability data showed that the pre-Cur was the most effective regimen compared to others (Fig. 1b, Additional file 1: Figure S1B), and that curcumin pretreatment at 20μM could significantly sensitize the anti-tumor activity of 5-Fu at a lower dosage (20μM), as compared with that of 5-Fu alone (Fig. 1b-d, Additional file 1: Figure S1B–d). Similarly, in the HCT116-derived xenograft mouse model, we further found that pretreatment of curcumin followed by a lower dosage of 5-Fu (30 mg/kg) exhibited not only a significant reduction of the tumor size but also an improvement in food-intake status, which was absent in mono-treatment groups (Fig. 6a and c). Nevertheless, our findings, from both in vitro and in vivo assays, provide valuable insights into the novel benefits of curcumin on increasing chemosensitization and decreasing undesirable toxicity, whatever being contained in a regular diet or as an adjuvant medicine, for long-term use in patients with colon cancer prior to or during 5-Fu-based chemotherapy.

Autophagy is an evolutionarily conserved catabolic process with essential functions in cellular homeostasis and cell survival under both physiological and pathological conditions [27]. In addition to its housekeeping roles in removing damaged DNA, dysfunctional proteins and defective organelles, autophagy also involves in tumorigenesis and cancer cell metabolism [28, 29]. Here, we proved that the dysregulation of autophagy also contributes to reduced cytotoxicity of 5-Fu. To understand synergistic effects of curcumin with 5-Fu, we compared the effects of 5-Fu alone with that of pre-Cur on cells’ autophagic process by immunofluorescence Cyto-ID Green staining and Western blotting of the autophagic proteins p62, beclin-1 and LC3II/LC3I. We showed that 20 μM 5-Fu activated autophagy after 24 h (Fig. 2, Additional file 3: Figure S3), which was reserved by pre-Cur combinational treatment (Fig. 3, Additional file 4: Figure S4-A). In contrast to our findings, Yao et al. reported that 140 mM 5-Fu reduced autophagy in SNUC5 colon cancer cells after more than 6 months treatment [30]. While another recent study, in concert with that of us, demonstrated that autophagy inhibitor 3-MA could potent 25 μM 5-Fu’s cytotoxicity in HT29 colon cancer cells after 48 h treatment [31]. The discrepancies between seemingly different impacts of 5-Fu on autophagy would be explained by the differences on its using dosages, times, as well as the different context with different colorectal cancer cell types [32, 33]. On the other hand, from the pathophysiological view of autophagy, which is generally regarded as a cellular adaptation mechanism to counteract cellular stress, for example in chemotherapy, that would trigger pro-survival signals escaping from apoptosis or cell death [34, 35], the 5-Fu-triggered autophagy activation in our experiments may be a survival response of the colon cancer cells to the cytotoxic stimulus of 5-Fu. In addition to 5-Fu, curcumin has also been believed to be an autophagy regulator associated with its anti-cancer activity, for instance as an autophagy inducer in human gastric cancer cells [36], human melanoma cells [37], osteosarcoma MG63 cells [38] and HCT116 colon cancer cell line [39], and as a blocker in malignant mesothelioma cells [40]. The molecular changes underlying curcumin-mediated autophagic responses were also documented for cutaneous T-cell lymphoma to be relevant to the degradation of beclin-1, which is a component of class III phosphatidylinositol 3-kinase (IIIPI3K) and has an up-regulating effect on autophagosome [41], and thereby the accumulation of microtubule-associated protein-1 light chain 3 (LC3I), which promotes the death of the cancer cells [42]. To our knowledge, there is no report addressing the critical role of the combinational use of curcumin with 5-Fu in autophagic regulation that promotes chemosensitization in the colon cancer cells. Collectively, 5-Fu-induced activation of autophagy might suggest a cellular mechanism responsible for, at least in part, the low susceptibility of HCT116 and HT29 colon cancer cells to 5-Fu mono-treatment, and the cellular autophagic turnover we observed following the pre-Cur combinational treatment might thus further reveal an autophagy inhibitory mechanism underlying, at least partially, the increased cytotoxicity of 5-Fu when curcumin actioned as an autophagy inhibitor.

Given the fact that curcumin’s pleiotropic activity on cancer prevention [2‐4], several earlier studies have further illustrated this interest of combinational using of curcumin with 5-Fu, by showing its sensitizing effect against 5-Fu resistance in varied cancer cell types such as human gastric cancer cells through inhibition of the NFκB survival-signaling pathway [43], or particularly in different sub-types of colon cancer cell lines, via miRNA-induced suppression of epithelial-to-mesenchymal transition [26], or the modulation of EGFR and IGF-1R. [19] With attempt to explore molecular interactions underlying the autophagic responses following 5-Fu mono-treatment and the pre-Cur combinational treatment, we investigated the changes within the core autophagy machinery including the autophagic trigger ULK1 and its upstream effectors Akt, mTOR and AMPK. It is generally believed that AMPK activation could dampen mTOR expression and thereby trigger autophagy via phosphorylation of ULK1. Accumulated evidence has indicated that under different nutritional statuses AMPK is coordinated closely with mTOR in regulating autophagy through direct phosphorylation of ULK1, mainly at site Ser317 and/or Ser777 within the Ser/Thr-rich domain [44, 45]. For instance, under basal condition, activated Akt/mTOR signaling is able to inhibit autophagy by disrupting the AMPK-ULK1 interaction, whereas during nutrient deficiency AMPK could promote autophagy by direct activation of ULK1, presumably at site Ser317 and /or Ser777 [45]. Although these studies has proposed a direct connection between nutrient-sensing kinases and autophagic activation, the further challenge still remains as to if these are other signaling pathways, such as feedback signals, that could even more criticaly regulate and control autophagy, as autophagy itself is such a “double-edged sword” process and requires to be precisely regulated [46]. Our data indicate that mono-5-Fu-activated autophagy in HCT116 and HT29 cells (Fig. 2, Additional file 3: Figure S3) appeared to be caused by the blockage of phosphorylation of Akt/mTOR and thereby activation of P-ULK1(Ser317), though no obvious changes found for AMPK signaling (Fig. 4a), whereas the pre-Cur combinational treatment appeared to down-regulate not only the P-Akt and P-mTOR expressions but also the P-AMPK and P-ULK1(S317) levels (Fig. 4b, Additional file 4: Figure S4-B). Although the question to these results still remains as to why under the circumstance of 5-Fu mono-treatment, ULK1(Ser317)-mediated activation of autophagy was apparently responsible for the inhibition of Akt/mTOR signaling pathway but not activation of AMPK, our data imply a functional substrate (ULK1Ser317)-competitive interaction between Akt/mTOR/ULK1 and AMPK/ULK1 pathways in favor of the latter that eventually leads to the autophagy reversal from being activated in response to 5-Fu alone to being inhibited in response to the pre-Cur treatment (Figs. 2 and 3, Additional file 3: Figure S3 and Additional file 4: Figure S4). Moreover, PI3K signaling takes vital role in tumor initiation and progression, and the signaling pathway is also genetically altered in numerous cancer types, including the tumor of colon, which was detected with high frequency of PIK3CA activating mutation as well as relative lower frequency of PTEN inactivating mutation [47]. PIK3CA and PTEN mutations both direct PI3K tumorigenesis largely through mediating Akt activity. And we selected two PIK3CA activating mutation cells HCT116 and HT29 to study the anti-tumor efficiency of different treatments either with 5-Fu alone or 5-Fu plus curcumin, so it is of interest to compare their impacts on Akt. As mentioned above, curcumin potentiated the inhibition of Akt by 5-Fu in a dose-dependent manner in HCT116 cells (Fig. 4b). In contrast, the treatment of 5-Fu alone only showed mild and seemingly transient Akt inhibition, which is absent at high dosages of 5-Fu (Fig. 4a), implying that curcumin not only functions as an inhibitor of autophagy to synergize the cytotoxicity of 5-Fu in colon cancer cells, but also works as a cytotoxicity enhancer via directly inhibition of Akt activity. Furthermore, curcumin-mediated AMPK-dependent autophagic turnover towards sensitization of HCT116 cells to 5-Fu was oppositely confirmed by addition of A-769662, a selective agonist of AMPK, to the pre-Cur treatment, in which the A-769662-activated AMPK led to an autophagic reverse as evidenced by the increased levels of both beclin-1 and the ratio of LC3II/LC3I and an decreased expression of p62 (Fig. 5b), and by the elevated intensity of Cyto-ID Green Staining (Fig. 5a) as compared with those found in the pre-Cur treatment, and consequently resulting in a reduced apoptosis and inactivation of caspase 3 (Fig. 5c and d, Additional file 5: Figure S5). More importantly, all these cellular and molecular findings could be well translated into in vivo anti-tumor outcomes as shown by the measures of tumor-size and food-intake for the HCT116-delivered xenograft mice following the 25-day combination treatment of curcumin (40 mg/kg) and 5-Fu (30 mg/kg) (Fig. 6a and c). Correspondingly, an AMPK/ULK1-regulated autophagic mechanism underlying the tumor-suppression efficacies of the combinational therapeutics in the tumor-bearing mice was also evident (Fig. 6b). Taking together, our in vitro and in vivo studies provide novel information regarding to 5-Fu’s lower therapeutic index and the underlying cellular and molecular mechanisms, as well as potential use of curcumin as an autophagy inhibitor to synergize with 5-Fu’s anti-tumor effects.

In summary, we show that autophagy was activated by mono-treatment of 5-Fu in vitro in the human colon carcinoma HCT116 and HT29 cell lines and in vivo in tumor tissues of the xenograft mice. This autophagy activation was found to predispose insensitivity of the tumor cells to 5-Fu. Pre-treatment with curcumin followed by 5-Fu (combination treatment), however, caused autophagy turnover both in vitro and in vivo, which was found to contribute to increased susceptibility of the colon cancer cells/xenograft to the anti-proliferative/anti-growth activities of 5-Fu.

Our results suggest, as illustrated in Fig. 7, that additive using of curcumin could amplify 5-Fu anti-tumor effects through suppression of Akt signaling and autophagic activity via damping AMPK/ULK1 signaling.