Introduction

While medical technology to extend human lifespan continues to develop, cancer mortality is still high, and the social burden is also increasing [1]. In particular, colorectal cancer has a high diagnosis rate, and although it has a high survival rate in the early stage, its mortality rate is high among all cancers [2]. When colorectal cancer is detected early, the cure rate is high, but it is often detected late because certain symptoms or pain do not appear in the early stages. With continuing research on conquering cancer, first-generation chemical anticancer drugs, second-generation targeted anticancer drugs, and third-generation immune anticancer drugs have been developed, and recently, fourth-generation anticancer drugs are being developed. Existing anticancer drugs have disadvantages such as side effects and resistance. Natural products are compounds produced by living organisms in nature that can be used as food, and it is expected that safe natural products can be used for anticancer therapy to compensate for the shortcomings of existing anticancer drugs.

Turmeric (Curcuma longa L.) is a perennial herbaceous plant belonging to the Zingiberaceae family that is cultivated in India, Indonesia, China, and other Asian countries. C. longa is frequently used as a medicinal herb and a food supplement owing to its health benefits. C. longa is known for many biological activities, among which anticancer activity has been reported, including against glioma cancer [3], cervical cancer [3], prostate cancer [3, 4], oral cancer [4], breast cancer [5, 6], colon cancer [7], and liver cancer [8] in vitro, and liver cancer [9], colon cancer [10], and breast cancer [11] in vivo.

In addition to studies on extracts, many anticancer studies have been conducted on major compounds contained in C. longa. As a representative compound of C. longa, curcumin has been reported to be highly cytotoxic to cancer cells. DNA damage [12], c-jun N-terminal kinase dependent apoptosis [13], and downregulation of E2F4 expression and apoptosis [14] were investigated in the human colon cancer cell line (HCT116) treated with curcumin. Curcumin is easily converted to demethoxycurcumin by demethoxylation from its benzene ring, resulting in a more stable structure [15]. Demethoxycurcumin has also been subjected to cytotoxic studies of cancer cell lines including fibrosarcoma [16], breast cancer [17], prostate cancer [18], lung cancer [19,20,21], bladder cancer [22], cervical cancer [23], oral squamous cell carcinoma [24], osteosarcoma [25], and brain glioblastoma [26]. However, there are no reports on colorectal cancer. We tried to confirm the growth-inhibiting activity of C. longa extract in colorectal cancer cells, as well as whether demethoxycurcumin exhibited the same activity. To support the interpretation of their metabolic mechanisms, metabolites of colorectal cancer cells treated with extract of C. longa and demethoxycurcumin were analyzed using nuclear magnetic resonance (NMR) spectroscopy. Metabolites are final products generated by metabolic processes, and because they reflect reactions that occur in the body, insight into the mechanism of cancer cells can be additionally obtained by applying metabolomics.

Materials and methods

Plant materials

The Curcuma longa used in this study was cultivated in the Jindo region of the Republic of Korea. A voucher specimen (MPS004295) was deposited at the Herbarium of the Department of Herbal Crop Research, National Institute of Horticultural and Herbal Science, Rural Development Administration, Eumseong, Republic of Korea. Dried rhizome of C. longa was ground and extracted twice by reflux extraction with 50% aqueous fermented ethanol at 80 °C for 4 h. The extract was filtered through a filter paper (Whatman, Maidstone, UK) and vacuum-concentrated under reduced pressure. The concentrated extract was lyophilized under reduced pressure (100 mTorr) (EYELA, Tokyo, Japan). Demethoxycurcumin was isolated from the C. longa using medium-pressure liquid chromatography (MPLC), NMR techniques and comparison with literature sample (Sigma, CAS # 22608-11-3).

Cell culture

The HCT116 human colorectal cancer cell line was purchased from the Korean Cell Line Bank (Seoul, Korea). HCT116 cells were cultured in DMEM containing 10% FBS with 100 U/mL penicillin and 100 U/mL streptomycin, and cultured at 37 °C in an atmosphere containing 5% CO2.

Cell viability assay

HCT116 cells (2 × 104 cells/well) were seeded in 96-well plates, and cell viability was detected using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT). HCT116 cells were cultured with various concentrations of C. longa extract (0, 50, 100, 200, and 300 μg/mL) or demethoxycurcumin (0, 1.5, 3.125, 6.25, 12.5, 25, and 50 μM). Following treatment for 24 h, 10 µL MTT solution (0.5 mg/mL) was added to each well, and cells were incubated at 37 °C for 2 h. The culture medium was subsequently removed and 100 µL dimethyl sulfoxide was added to each well at room temperature while being shaken for 20 min. The optical density of HCT116 cells was measured at 570 nm wavelength on a multiwell plate reader.

Western blot analysis

HCT116 cells (5 × 105 cells/well) were seeded in 6-well plates, and the expression of PARP, caspase-3, caspase-9, Bcl-2, Bax, and p53 protein was detected via Western blot analysis. Following treatment for 24 h with demethoxycurcumin (0, 20, 40, and 60 µM) or doxorubicin (1 µM), HCT116 cells were washed twice with ice-cold PBS and lysed for 30 min on ice in cell-lysis buffer (1 × RIPA; Thermo Fisher Scientific, USA). Protein concentration was determined using a BCA protein assay kit. Protein samples (10–30 µg/well) were resolved using 10% SDS-PAGE. Separated proteins were subsequently transferred onto polyvinylidene difluoride (PVDF) membranes at 70 V for 1.5 h. The membranes were blocked with 5% nonfat milk powder in TBS Tween 20 (TBST) buffer at room temperature for 1 h with shaking. The PVDF was incubated with anti-PARP (1:1000; Cell Signaling Technology, USA), anti-cleaved caspase-3 (1:1000; Cell Signaling Technology, USA), anti-cleaved caspase-9 (1:1000; Cell Signaling Technology, USA), anti-Bcl-2 (1:500; Santa Cruz Biotechnology, USA), anti-Bax (1:500, Santa Cruz Biotechnology, USA), anti-p53 (1:1000, Santa Cruz Biotechnology, USA) and anti-β-actin (1:5000, Santa Cruz Biotechnology, USA) at room temperature for 3 h. Following 3 washes with TBST twice for 30 min, the membranes were incubated with anti-immunoglobulin secondary antibody (1:2000–1:5000, Santa Cruz Biotechnology, USA) at room temperature for 1 h with shaking. The bands were detected using an Enhanced Chemiluminescence Prime Western blotting kit (Thermo Fisher Scientific, USA). Statistical significance was tested using Student’s t-test. P < 0.05 was considered statistically significant.

Sample preparation for NMR analysis

For the metabolic analysis, HCT116 cells were cultivated in the same conditions as the cell viability assay. In order to obtain statistical significance, 10 samples of the control group (CON), 11 samples of the doxorubicin-treated group (DOX), and 10 samples of the C. longa-treated group (CL) were cultivated and analyzed. In addition, 10 samples of CON, 11 samples of DOX, 10 samples of the demethoxycurcumin 20 μM group, 10 samples of the demethoxycurcumin 40 μM group, and 10 samples of the demethoxycurcumin 60 μM group were cultivated and analyzed. Harvested HCT116 cells were washed 3 times with PBS, then an HCT116 cell pellet (approximately 1 × 107 cells) was suspended in 400 μL of pre-chilled methanol, and 325 μL of cold water and 400 μL of cold chloroform were added step-by-step. After centrifugation (4 °C, 4000 rpm, 10 min), aqueous phase and organic phase were separated. The aqueous phase of each sample was collected in a glass vial and the methanol in the sample was eliminated using SpeedVac (EYELA, Tokyo, Japan) for 6 h, and all solvent was lyophilized overnight. Samples were re-dissolved using 560 μL of deuterated 0.2 M sodium phosphate buffer containing 0.2 mM of 3-(trimethylsilyl) propionic-2,2,3,3-d4 acid sodium salt (TSP-d4) and transferred to 5 mm NMR tubes.

NMR data acquisition and processing

One-dimensional (1D) 1H-NMR spectroscopy was conducted using a 700 MHz Bruker NMR spectrometer (Bruker Biospin, Germany) equipped with a cryoprobe at the KBSI Ochang Center, Republic of Korea. Pulse sequence for 1D experiment was a 1H nuclear Overhauser effect spectroscopy (NOESY). The parameters were set as 2.00 s of relaxation delay, 50.0 ms of mixing time, 12.2 μs of 90º pulse width, and 128 transients. This resulted in a total experiment time of 7 min 48 s. Two-dimensional (2D) NMR spectroscopy experiments were performed using a 900 MHz Bruker NMR spectrometer equipped with a cryoprobe at the KBSI Ochang Center. 2D 1H-1H correlation spectroscopy (COSY) and 1H-13C heteronuclear single quantum correlation (HSQC) were performed. All acquired data were manually phased and baseline corrected using Topspin v3.6.5 (Bruker Biospin, Germany). Metabolic identification and quantification were performed using Chenomx NMR Suite 8.4 Professional (Chenomx Inc., Edmonton, Canada). Biomarker and correlation analyses of quantified metabolites were conducted by MetaboAnalyst 5.0 (https://www.metaboanalyst.ca). For multivariate statistical analysis, NMR spectra were binned to 0.001 ppm size and normalized to total area using Chenomx NMR Suite 8.4 Professional. The binning results were aligned by the icoshift algorithm of MATLAB (MathWorks, USA). Principal component analysis (PCA) and orthogonal partial least square discriminant analysis (OPLS-DA) were performed using SIMCA 15.0.2 (Umetrics, Sweden).

Results and discussion

The cytotoxic effects of C. longa extract and demethoxycurcumin, a single compound in the plant, on the HCT116 cell line were examined by MTT assay. The results show that HCT116 cell growth was dose-dependently inhibited by C. longa extract treatment for 24 h (Fig. 1A) and 48 h (Fig. 1B). The cell viability was less than 40% after treatment with 300 µg/mL of C. longa extract for 24 h, and less than 20% after treatment with 300 µg/mL of extract for 48 h. C. longa extract effectively inhibited the survival of HCT116 cells. Demethoxycurcumin treatment also inhibited the growth of HCT116 cells in a dose-dependent manner (Fig. 1C). C. longa extract majorly contains curcumin, and there have been many reports of curcumin inhibiting the growth of HCT116 cells [12,13,14]. As a result of confirming the cytotoxicity of the single compound contained in C. longa, the IC50 of curcumin was 12 µM (data was not shown), and the IC50 of demethoxycurcumin was 38.5 µM. The reason why C. longa extract effectively inhibited the growth of the HCT116 cell line is expected to be due to the combined action of the activity of curcumin, which is high in turmeric extract, and the activity of demethoxycurcumin. We attempted to determine how C. longa extract and demethoxycurcumin affect metabolic changes associated with HCT116 cell growth inhibition.

Fig. 1
figure 1

Effects of C. longa extract or demethoxycurcumin on cell viability of HCT116 cells. A Cell viability of HCT116 cells treated with C. longa extract (0, 50, 100, 200, and 300 µg/mL) for 24 h and B 48 h, C Cell viability of HCT116 cells treated with demethoxycurcumin (0, 1.5, 3.125, 6.25, 12.5, 25, and 50 µM) for 24 h

Full size image

To confirm the metabolic perturbation by C. longa treatment in HCT116 cells, NMR-based metabolic analyses were performed. To obtain the 1D 1H-NMR spectra, a 700 MHz NMR spectrometer was used. To identify metabolites in the 1D spectra, 2D COSY and HSQC-DEPT experiments were performed using a 900 MHz NMR spectrometer. All spectra were analyzed for identification and quantification of metabolites using Chenomx NMR suite software, the open Human Metabolome Database (HMDB), and measured 2D NMR spectra (Additional file 1: Figures S1 and S2). A total of 45 primary metabolites were analyzed in the extract of HCT116 cells (Additional file 1: Table S1). Figure 2 shows a representative NMR spectrum of the HCT116 extract and annotation of identified metabolites.

Fig. 2
figure 2

Representative one-dimensional NMR spectra of the HCT116 cell extract with annotation of major metabolites. 4-HP trans-4-hydroxy-L-proline, Ala alanine, Asn asparagine, Asp aspartate, Glu glutamate, Gly glycine, GPC sn-glycero-3-phosphocholine, GSH glutathione, His histidine, Ile isoleucine, Leu leucine, mI myo-inositol, NAA N-acetylaspartate, PC O-phosphocholine, Phe phenylalanine, Pro proline, Thr threonine, Tyr tyrosine, Val valine

Full size image

Quantified metabolites were analyzed by biomarker discovery analysis using receiver operator characteristic (ROC) curves with sensitivity and specificity. The area under the ROC curve (AUC) for each metabolite was calculated when comparing the C. longa extract treated group with the control group and the demethoxycurcumin 60 μM treated group with the control group (Table 1). In the comparison of the C. longa extract treated group with the control group, the AUC value of alanine, arginine, aspartate, betaine, creatine, glutamate, glutathione, glycine, lactate, phenylalanine, threonine, uracil, and valine was 1.00, which makes it an excellent prediction biomarker. In general, AUC values of 0.9–1.0, 0.8–0.9, and 0.7–0.8 are considered excellent, good, and fair predictive biomarkers, respectively [27]. In the comparison of the demethoxycurcumin 60 μM treated group with the control group, the AUC value of AMP, formate, glycine, lactate, NAD + , succinate, UDP-N-acetylglucosamine, UMP, uracil, uridine, myo-inositol, and glycerophosphocholine was also 1.00. The metabolites that showed the same change trend of concentration in the two comparisons with AUC over 0.7 were filtered. The result showed that lactate and uracil were increased after treatment with C. longa extract and demethoxycurcumin 60 μM, and aspartate, ATP, betaine, dimethylamine, formate, glycerophosphocholine, myo-inositol, and uridine were decreased after both treatments.

Table 1 Potential biomarkers of metabolites for filtering significantly changed by C. longa extract and demethoxycurcumin 60 µM treatment (AUC ≥ 0.7)
Full size table

NMR spectra were binned for multivariate statistical analysis. On the PCA score plot of the C. longa extract treatment comparison, three groups were clearly separated (Fig. 3A). OPLS-DA was performed to confirm the difference between the control group and the C. longa extract treated group. The OPLS-DA score plot shows a clear separation between the two groups (Fig. 3B), and the S-line plot shows that metabolites contributed to this separation (Fig. 3C). Demethoxycurcumin treated groups were similarly classified on PCA and OPLS-DA score plots (Fig. 3D, E). The group with low-dose demethoxycurcumin treatment showed less difference from the control group, but the groups treated with 40 and 60 μM were similar and distributed far from the control group. As shown in the biomarker analysis, increased lactate and decreased uridine, myo-inositol, and dimethylamine were confirmed on the S-line plot of the comparison between control and demethoxycurcumin treated groups (Fig. 3F).

Fig. 3
figure 3

Multivariate statistical analyses of NMR data. A PCA score plot of control group, doxorubicin treatment group and C. longa extract treatment group (R2X = 0.875, Q2 = 0.722), B OPLS-DA score plot of control group and C. longa extract treatment group (R2X = 0.531, R2Y = 0.995, Q2 = 0.983), C OPLS-DA S-line plot of control group and C. longa extract treatment group, D PCA score plot of control group, doxorubicin treatment group and demethoxycurcumin treatment (20, 40, 60 µM) groups (R2X = 0.241, Q2 = 0.15), E OPLS-DA score plot of control group and demethoxycurcumin 60 µM treatment group (R2X = 0.383, R2Y = 0.997, Q2 = 0.928), F OPLS-DA S-line plot of control group and demethoxycurcumin 60 µM treatment group

Full size image

By comprehensively interpreting the results of multivariate statistical analysis and biomarker analysis, it was possible to identify significantly changed metabolites. Glutathione was upregulated and glutamate, which is involved in the biosynthesis of glutathione, was reduced by exposure to C. longa extract. Therefore, glutathione and glutamate had a negative correlation. It was reported that the α, β-unsaturated carbonyl group, which is contained in curcumin and curcumin derivatives, causes oxidative stress [28]. It was also reported that oxidative phosphorylation is reduced to lower the generation of reactive oxygen species as a mechanism to protect DNA and protein from oxidative stress [29]. This can increase lactate by increasing energy production using glycolysis. Pyruvate is converted to lactate by lactate dehydrogenase, with oxidation of NADH to NAD + . In the results of our study, a significant increase in lactate and an increase in NAD + were confirmed (Fig. 4A). In addition, alanine, which is converted from pyruvate and glutamate by aminotransferase, was significantly increased and glutamate was decreased (Fig. 4B). An increase in energy production using glycolysis may result from mitochondrial damage. In a previous report, boiled extract of C. longa induced apoptosis by reducing the membrane potential through mitochondrial damage in glioma (A172), prostate cancer (PC3), and uterine cancer (HeLa) cell lines [3]. C. longa extract, which is a mixture of various compounds, can inhibit cancer cells through various mechanisms. The root of C. longa contains approximately 2% curcumin [30], and much research has been done on the activity of curcumin. The root of C. longa also contains curcumin derivatives, among which demethoxycurcumin accounts for roughly one-third of the curcumin content [31], and the activity of demethoxycurcumin on cancer cells has rarely been studied.

Fig. 4
figure 4

Metabolic pathways involving significantly altered metabolites. A Pyruvate to lactate reduction by lactate dehydrogenase, B Cahill cycle by aminotransferase

Full size image

To confirm the inhibitory effect of demethoxycurcumin on cancer cells, we investigated the apoptosis-related proteins in HCT116 cells following demethoxycurcumin treatment by Western blot analysis. The expression levels of apoptosis-related proteins were evaluated by assessing pro-PARP, PARP, Bax, Bcl-2, and p53 protein levels in HCT116 cells incubated with normal medium (negative control), doxorubicin (positive control), or demethoxycurcumin (20, 40, and 60 μM) (Fig. 5). The β-actin content was confirmed to be constant, indicating that changes in these proteins cannot be explained by unequal loading of proteins. Demethoxycurcumin treatment at 40 and 60 μM significantly increased PARP, caspse-3, and caspase-9 cleavage. The increase in cleaved PARP was associated with a decrease in pro-PARP. Bax was upregulated following demethoxycurcumin treatment at 40 and 60 μM, whereas Bcl-2 was suppressed. Bax is a pro-apoptotic protein belonging to the Bcl-2 family that moves to the mitochondrial outer membrane and oligomerizes to release cytochrome c into the cytoplasm [7]. Bax was increased by demethoxycurcumin treatment, whereas Bcl-2 expression was decreased. Cytochrome c released from the mitochondria served as an apoptosis signal and increased PARP, caspase-3, and caspase-9 cleavage according to demethoxycurcumin treatment. The expression of p53 was also increased with demethoxycurcumin treatment, and a greater increase with doxorubicin treatment was confirmed. Doxorubicin directly causes DNA damage in cancer cells and activates p53, which can cause apoptosis through activation of Bax. In this result, the doxorubicin treated group showed a significant increase in p53 compared to the demethoxycurcumin treated groups.

Fig. 5
figure 5

Effects of demethoxycurcumin on apoptosis-related protein expression in HCT116 cells by western blot analysis. Data are presented as mean ± SD of three independent experiments (*p < 0.05, **p < 0.01, ***p < 0.001 versus control group)

Full size image

In the case of doxorubicin treatment, the PCA score plot of the metabolic data shows a significantly different distribution pattern from those of the C. longa extract and demethoxycurcumin treatment groups. This pattern can also be confirmed in the heatmap (Fig. 6). In particular, the amino acids tyrosine, methionine, isoleucine, leucine, arginine, phenylalanine, and valine showed a pattern of decreasing with doxorubicin treatment. This pattern was the same even with a low concentration of demethoxycurcumin, and the expression of apoptosis-related proteins in this group was similar to that of the control group. Therefore, it is considered that the decrease of these amino acids is not directly related to apoptosis. On the other hand, the decrease in 4-hydroxyproline, choline, NAD + , and glycerophosphocholine in the doxorubicin and demethoxycurcumin (40 and 60 µM) groups was confirmed on the heatmap. Among them, choline and glycerophoshophcholine are part of the CDP-choline pathway, which is activated in cancer, and it was reported that glycerophosphocholine levels are characteristically high in cancer [32]. It was also reported that cancer cells increase the uptake of choline and the biosynthesis of phosphatidylcholine, and, through their degradation, produce lipid mediators, such as free fatty acids used as fuel for β-oxidation, or phosphatidic acids for migration of cancer cells for escaping from immune cells. Phosphatidylcholine is the major component of the cellular membrane; therefore, its biosynthesis is closely related to the proliferation of cancer cells [33]. In our results, the contents of choline and glycerophosphocholine formed by the degradation of phosphatidylcholine were the highest in the control group, and were decreased with doxorubicin and demethoxycurcumin treatment. Phosphocholine, which is also part of the CDP-choline pathway, was decreased after treatment with demethoxycurcumin. It is considered that the CDP-choline pathway, which is activated for the survival of cancer cells, is blocked by treatment with demethoxycurcumin (Fig. 7).

Fig. 6
figure 6

Pearson correlation heatmap of metabolites in the HCT116 cell extracts. A Comparison of control group, doxorubicin treatment group and C. longa extract treatment group, B comparison of control group, doxorubicin treatment group and demethoxycurcumin treatment (20, 40, 60 µM) groups

Full size image
Fig. 7
figure 7

CDP-choline pathway actively occurring for the survival of cancer cells

Full size image