Metabolomic analysis reveals the mechanism of aluminum cytotoxicity in HT-29 cells

Chemicals

Aluminum chloride was obtained from Sigma-Aldrich, St. Louis, MO, USA. TRIzol was bought from Ambion, Life Technologies (Grand Island, NY, USA). The iTaq™ universal SYBR^® Green one-step kit was purchased from Bio-Rad (Berkeley, CA, USA). All of the materials used for cell experimentation were bought from Shanghai Chemical Reagent Company (Shanghai, China). MTT and dimethyl sulfoxide (DMSO) used for detecting cell viability were purchased from Gibco, Life Technologies (Grand Island, NY, USA). Milli-Q AdvantageA10 used to provide ultrapure water was obtained from Millipore (Boston, MA, USA).

Cell culture

HT-29, the human colon cancer cell line, acquired from the Shanghai cell bank of the Chinese Academy of Sciences, was cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin at 37 °C in a 5% CO₂ atmosphere. The cell culture was passaged every other day by trypsinizing the cells and diluting them at 1:3/1:2 in fresh medium to start a new cycle. The cells were then used in the following analysis when they reached the logarithmic growth phase.

Metabolic activity detection

MTT assay for cell viability was performed to determine the half-maximal inhibitory concentration (IC50) of Al ion (Nzengue et al., 2008; Xu et al., 2010). Briefly, the cells in the logarithmic growth phase were added to a 96-well plate at a density of 103–104 cells per well. Once each well was coated with a monolayer, fresh medium supplemented with sterile phosphate-buffered saline (PBS; filtered through a 0.22-μm-pore size Millipore filter) containing Al ions at a final concentration of 0, 2, 4, 6, 8, or 10 mM was added to separate wells. After 24-h incubation, 20 μL of five mg/mL MTT was added to each well, and the cells were incubated for another 4 h. Subsequently, the medium in each well was replaced with 150-μL DMSO, and the cell viability was determined by measuring the absorbance of well contents at 490 nm.

Cell collection and cell extraction

The Al ion concentration of four mM was selected for further analyses based on the results of our MTT assay and of a previous study (Pineton de Chambrun et al., 2014). HT-29 cells were seeded at a density of 106 cells per well in medium containing four mM Al ion and were incubated for 24 h. Cells incubated in medium devoid of Al ion were considered as control. After incubation, the cells were washed twice with PBS and frozen in liquid nitrogen for quenching metabolism. The procedure of extraction reported in a previous study was used with slight modifications (Van den Eede et al., 2015). Briefly, one mL of trichloromethane–water–methanol mixture at a ratio of 1:1:4 (v/v/v) was added to the wells for 5 min. The cells were then detached from the wells using sterile rubber scrapers and transferred to precooled bottles (−20 °C). Each well was rinsed with another one mL of extraction solution, and the residue was also transferred to the bottle. The samples were vortexed and subjected to ultrasonication in an ice-water bath at 300 W for 3 min (6-s/4-s on/off pulses). The samples were then transferred into 1.5-mL Eppendorf tubes and subjected to supersonic extraction for 3 min, followed by centrifugation at 15,000 rpm for 10 min (4 °C). The supernatant (200 μL) was used for metabolomic analysis by LC-MS.

Metabolomic analysis

Agilent 1290 Infinity Ultra High-Performance Liquid Chromatography (UHPLC) system combined with Agilent 6538 UHD and Accurate-Mass Q-TOF/MS were used for data processing in the positive and negative ionization modes. The parameters of UHPLC and MS were further optimized based on a previous study (Van den Eede et al., 2015) to meet the needs of the current conditions. A C₁₈ column (Waters ACQUITY UPLC@HSS T3; 100 × 2.1 mm, 2.1 μm) was used to separate the samples. Solvent A was 0.1% (v/v) formic acid in water, and solvent B was 0.1% (v/v) formic acid in acetonitrile. Each time, a three μL sample was injected into the column maintained at 40 °C. The liquid flow rate was 0.4 mL/min, with the following gradient conditions: 0 min, 5% B; 2 min, 5% B; 13 min, 95% B; 15 min, 95% B.

The capillary voltages used for the negative- and positive-ion modes were 3,500 V and 4,000 V, respectively. The reference masses 112.985587 m/z and 1033.988109 m/z were used in the negative mode and 121.0509 m/z and 922.0098 m/z were used in the positive mode. A quality control sample was injected every eight samples to ensure consistent quality.

Data treatment

Agilent Mass Hunter Qualitative software was used for data conversion. Peak alignment and integration were conducted using XCMS (https://metlin.scripps.edu/) (Smith et al., 2006). A 3D matrix with retention time (RT), peak intensity, and m/z were obtained. The following XCMS parameters were used: profmethod = bin, method = matched Filter, step = 0.1, full width at half maximum (peak width) = 8, bandwidth = 10, snthresh = 5, ppm = 20, mzdiff = 0.01, and minfrac = 0.8. Mass range was 50–1100 m/z, mass tolerance was 20 ppm, RT range was 0.5–15.0 min, and RT width threshold was 0.2 min. The following formula was used to integrate the area of the peaks: unimodal area/area of all peaks in a single sample × 10,000. The SIMCA-P+ 14.0 software was used for principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA). Two principle components for the positive mode and two for the negative mode were used in PCA. Five principal components for the positive mode and four for the negative mode were used in PLS-DA. Variables with variable importance in projection (VIP) values of >1 were considered as differential variables. Seven-round cross-validation with 200 response permutation tests was used to prevent model overfitting and determine the model quality.

Metabolite screening

Metabolites were screened using univariate and multivariate analyses, the screening criteria (VIP > 1, P < 0.05, fold change > 1.5 or < 0.67), and the mass tolerance threshold (20 mDa). The HMDB and METLIN databases were used for metabolite identification, and the MBRole database was used for pathway enrichment.

Verification of key enzymes by RT-qPCR

The relative expressions of nine related genes in the altered metabolomics pathways were further validated using quantitative reverse transcription polymerase chain reaction (RT-qPCR). The corresponding encoded proteins were found to be succinate dehydrogenase (sdhA), citrate synthase (CS), isocitrate dehydrogenase (IDH), lactic dehydrogenase (LDH), pyruvate dehydrogenase (PDH), pyruvate kinase (PK), glutamic oxaloacetic transaminase (GOT), glutathione peroxidase (GPx), and glutathione reductase (GR). The genes of the first three, middle three, and last three proteins are related to the TCA cycle, glycolysis, and glutathione (GSH) metabolism, respectively. Total RNA was extracted using the traditional TRIzol method. TAKARA RR047A kits were used to obtain cDNA by reverse transcription. Subsequently, 20-mL reaction mixture was prepared by combining cDNA, 2× Bio-Rad iTaq™ Universal SYBR^® Green Supermix, and specific primer pairs (Table 1) and subjected to RT-qPCR. Each reaction had three parallels, and the β-actin gene was used as the internal reference gene.

The relative amounts of mRNA were calculated using the following equation: $$\text{Relative\hspace{0.17em}expression\hspace{0.17em}}={\text{2}}^{\wedge }-\left[\left({\text{CT}}_{\text{AT}}-{\text{CT}}_{\text{AR}}\right)-\left({\text{CT}}_{\text{CT}}-{\text{CT}}_{\text{CR}}\right)\right],$$ where AT is the target gene in the Al-treated group, AR is the reference gene in the Al-treated group, CT is the target gene in the control/untreated group, and CR is the reference gene in the control group.

MTT assay for cell viability

The HT-29 cell viability decreased with the increase in Al concentration from 0 to 10 mM (Fig. 1; Dataset S1). Cell viability decreased to approximately 50% at four mM of Al ion concentration. This concentration was used to study Al cytotoxicity in the subsequent experiment, consistent with that in a previous study (Pineton de Chambrun et al., 2014).

Metabolomic characterization of cells

The metabolites in the Al-treated group were compared with those in the control group using typical base peak intensity chromatograms. The results are presented in Fig. S1 for both the positive- and negative-ion modes. In total, 995 and 2,272 discrepant metabolites in the negative and positive modes, respectively, were selected by UPLC/MS for further analysis.

First, unweighted analysis was used to evaluate the positive and negative data. The relationships of the metabolites between the Al-treated and control groups were identified by comparative analysis. The R2X, a representative parameter of the PCA model, was 0.472 and 0.548 for the positive and negative modes, respectively. The PCA score plots in Fig. 2 show that six replicates of the Al-treated group were separated from the samples in the control group. This result indicates that Al exposure caused some alteration in the metabolic profile of HT-29 cells.

Extraction of special metabolites

The Al-treated cells and untreated cells were further analyzed by PLS-DA. The more closely the three indexes (R²X, R²Y, and Q²) approached 1, the more credible and predictable are the metabolomics data. As shown in Fig. 3, in the above case, the R²X values were 0.725 and 0.745, the R²Y values were 1 and 0.997, and the Q² values were 0.474 and 0.823 for the positive- and negative-ion modes, respectively. Partial least squares-discriminant analysis was cross-validated, and the results are shown in Fig. S2. All of the metabolites were judged by their VIP values, which reflected the proportion of their influence on the model.

The following screening criteria were rigorously applied to screen the metabolites as potential biomarkers: fold change > 1.5 or < 0.667, VIP > 1, and P < 0.05 (relative peak areas of the metabolite contents). The METLIN and HMDB databases were used to screen and characterize the candidates, which were then annotated as 39 and 42 metabolites in the positive- and negative-ion modes, respectively (Tables 2 and 3; Dataset S2). The metabolites GSH, glutamic acid, four phosphatidylcholines (PC (14:1), PC (14:0), PC (15:0), and PC (14:1)), phosphatidylethanolamine (PE (22:4)), pantothenate, creatine, and choline, were consistently downregulated after Al exposure.

Pathway enrichment analysis

The MBRole-ID conversion was used to identify metabolites and enrich the pathways in KEGG. One pathway in the positive-ion mode, namely arginine and proline metabolism, and 17 pathways in the negative-ion mode, including pyrimidine metabolism, the TCA cycle, pyruvate metabolism, and GSH metabolism, were ultimately identified (all P < 0.05, Fig. 4). Moreover, some metabolites, such as pyruvate, citrate, succinate, leucine, tryptophan, phenylalanine, glutamate, were significantly reduced after Al exposure. The results of RT-qPCR shown in Fig. 5 (Dataset S3) presented the relative gene expression levels in the Al-treated group compared with those in the control group. Specifically, the expression of GR, PK, and LDH genes was significantly upregulated, whereas that of the other four genes (GPx, PDH, GOT, and IDH) was significantly downregulated (P < 0.05).

Effects of Al cytotoxicity on energy metabolism

The TCA cycle is a common pathway in all aerobic organisms used to produce energy through the oxidation of acetyl-CoA derived from fats, carbohydrates, and proteins into ATP and CO₂. It also provides precursors for the biosynthesis of certain amino acids and NADH. Pyruvate dehydrogenase catalyzes the decarboxylation of pyruvate, the end product of glycolysis, into acetyl-CoA. The RT-qPCR results showed that the expression of PDH in Al-treated cells was decreased to 0.65-fold of that in the untreated cells, which consequently decreased the formation of acetyl-CoA, resulting in the shortage of substrate for the TCA cycle. Consequently, the levels of intermediate compounds in the TCA cycle, such as citrate and succinate, decreased after Al exposure. Furthermore, decreased citrate and succinate levels in turn reduced NADH production, which is essential for the delivery of electrons to oxygen through the electron transport chain to produce ATP. Thus, an important mechanism of Al toxicity appears to be the perturbation of mitochondrial ATP production. These results indicate that Al exposure has a deleterious effect on the energy output and causes various cellular abnormalities. Another reason that explains the decrease in some components of the TCA cycle is the oxidation of the TCA cycle enzymes due to Al exposure. The TCA cycle enzymes are very sensitive to oxidative stress (Tretter & Adam-Vizi, 2005). Indeed, the relative mRNA expressions of IDH and GOT in Al-treated cells were found to be significantly decreased to 0.42- and 0.74-fold of those in the untreated cells, respectively (Fig. 5). Kil et al. (2006) reported that heavy metal exposure decreased IDH levels in human embryonic kidney 293 cells, impairing the conversion of citrate into isocitrate. Murakami & Yoshino (2004) found that Al inhibited IDH expression, thereby decreasing GSH regeneration in mitochondria. Benderdour et al. (2004) reported decreased IDH expression as a marker of oxidative stress in cardiac hypertrophy development.

Cell growth requires glycolysis for ATP generation to maintain the energy supply and for the accumulation of glycolytic intermediates to meet the needs of rapid cell proliferation and the need for rapid synthesis of nucleotides, lipids, and proteins (Ye et al., 2012). In the oxidative environment, cells can use glycolysis for ATP production, without the participation of mitochondria. The function of this metabolic route is crucial to maintain the energy balance of organisms in various conditions (Dang & Semenza, 1999; Kim et al., 2006). Mailloux & Appanna (2007) reported that glycolysis increases with the increase in oxidative stress in Al-treated cells. Two enzymes, PK and LDH, which are critical for ATP production in the anaerobic environment, exhibited dramatically increased activity upon Al exposure in HT-29 cells in our study. The study by Mailloux & Appanna (2007) reported similar results for Al-treated HepG2 cells. Lactic dehydrogenase is also required for the recycling of NAD⁺, which is essential for activating the subsequent step of glycolysis (Dang & Semenza, 1999).

Effects of Al cytotoxicity on oxidative stress and amino acid metabolism

The main mechanism of Al toxicity is the induction of oxidative stress by producing excessive reactive oxygen species (ROS) (Kumar, Bal & Gill, 2009). Excess ROS levels cause imbalances and disturbances in signaling processes, such as MAPK signaling pathways, leading to growth inhibition and damage to human cells (Sharma & Dietz, 2009). Reactive oxygen species are also responsible for exacerbating DNA damage (Santos et al., 2010). Glutathione can protect cellular components against oxidative injury by directly neutralizing ROS or functioning as a cofactor against free radicals in cells (Carrola et al., 2016). In addition, being a thiol compound, GSH can bind to free metal ions with high affinity and reduce metal availability, thus indirectly reducing ROS production and alleviating oxidative stress (Sharma & Dietz, 2009). In our study, GSH level was lower in Al-treated cells than in untreated cells, suggesting that Al exposure triggered oxidative stress. Our findings also revealed decreased levels of glutamate, a GSH precursor, in the Al-treated cells, supporting the hypothesis of deregulating GSH synthesis. Glutathione peroxidase catalyzes the oxidation of GSH to GSSG, whereas GR catalyzes the reduction of GSSG to GSH. Our RT-qPCR results showed that GPx expression was reduced by 0.36-fold, whereas GR expression was increased by 1.74-fold in Al-treated cells compared with untreated cells (Fig. 5). This result indicates that the decrease in GPx expression and increase in GR expression may be a mechanism to alleviate the Al exposure-induced oxidative stress in Al-treated cells.

Upon exposure to toxins, such as toxic metals, several species downregulate or upregulate the synthesis of diverse metabolites and cause the conversion of specific amino acids. For example, glycine, alanine, serine, tryptophan, cysteine, and threonine can be converted to pyruvate through the TCA cycle, whereas aspartate and asparagine are converted to oxaloacetate (Owen, Kalhan & Hanson, 2002). Our results showed decreased intracellular levels of the ketogenic amino acids leucine, tryptophan, and phenylalanine and the glycogenic amino acid glutamate in Al-treated HT-29 cells, consistent with the results of a previous study on human epidermal keratinocytes (Carrola et al., 2016). One previous study reported that the metabolism of glutamine, the most abundant naturally occurring amino acid in the body, is upregulated during the glucose shift to provide substrate for increased lipogenesis and nucleic acid biosynthesis crucial for cell proliferation (Dakubo & Safarina, 2010). After the conversion during the glucose shift, it may be transformed into α-ketoglutarate and enter the TCA cycle. Thus, it can be speculated that decrease of amino acid level may be an indirect cause for energy and lipids metabolism alterations induced by Al exposure.

Effects of Al cytotoxicity on lipid metabolism and cellular apoptosis

Phospholipids, such as PCs, PEs, monoglyceride, and sphingomyelin (SM), are indispensable components of cellular membranes, signaling pathways, and cellular processes such as cell–cell interactions, cell proliferation, cell differentiation, and cellular apoptosis (Hla & Dannenberg, 2012). Membrane lipids containing polyunsaturated fatty acids are highly sensitive to free radicals and are susceptible to lipid peroxidation. Our results revealed a decrease in choline and phosphocholine levels in Al-treated cells, suggesting that cell membrane disruption caused by Al exposure and the corresponding oxidative stress. In particular, the levels of PCs, LysoPCs, and PEs—crucial cellular membrane constituents—in Al-treated cells decreased to 0.28–0.64-fold of those in untreated cells (Tables 2 and 3), suggesting the degradation of membrane phospholipids and cellular apoptosis. The level of SM (d18:1/14:0) in Al-treated cells decreased to 0.43-fold of that in untreated cells, also indicating the occurrence of cellular apoptosis. A decrease in the levels of creatine, which plays a key role in energy transport across the mitochondrial membrane. Notably, a decrease in choline levels can disturb transmethylation and may also result in a decline in creatine levels. Lower carnitine level was also observed after Al treatment. A decrease in the carnitine level has been reported to cause accumulation of endotoxins and disruption of several enzyme reactions (Coulter, 1991; Russell, 2007). Carnitine deficiency is implicated in various conditions such as diabetes mellitus, Alzheimer’s disease, and heart failure (Ha et al., 2012; Li et al., 2015). After CoA biosynthesis from the precursor pantothenate, CoA is combined with a long-chain fatty acid to produce acetyl-CoA that enters the TCA cycle. Thus, disruption in the CoA biosynthesis pathway is closely related to abnormalities in energy metabolism and fatty acid biosynthesis, as well as imbalances in various cellular processes. Consequently, the availability of pantothenate plays a key role in this pathway, as indicated by a previous finding that the rate of the CoA production step is limited by the phosphorylation of pantothenate. Previous studies demonstrated that the lack of pantothenate has a severe impact on energy production and fatty acids synthesis (Rock et al., 2000).

Other effects of Al cytotoxicity

The level of edetic acid, a common metal chelator, in Al-treated cells decreased to 0.66-fold of that in the untreated cells (Table 3). A reduction in edetic acid levels would affect the trace elements levels. Consistent with this result, Al exposure reduces the Fe, Ca, and Mg levels in mice tissues, as well as Cu, Zn, Fe, and Ca levels in humans serum (Yu et al., 2017; Metwally & Mazhar, 2007). Taken together, our results indicate that a decrease in GSH levels and disturbance in trace element levels may be the main mechanisms of Al-induced oxidative stress in HT-29 cells.

The level of 8-Oxo-dGDP, generated by the oxidation and phosphorylation of dGTP, in Al-treated cells decreased to 0.43-fold of that in the untreated cells (Table 2), suggesting that Al exposure led to nucleotide damage directly (Hayakawa et al., 1995). Moreover, Al exposure would reduce the amount of metabolites that can metabolize toxic substances. For example, 4-hydroxymidazolam level reduced to 0.65-fold after Al treatment. It is an active metabolite of midazolam, can further produce to glucuronide, metabolizing several exogenous and endogenous toxic substances and thus exerting positive and protective effects (Seo et al., 2010).