This article is mentioned in:

Introduction

Cervical cancer is one of the most prevalent cancer worldwide with an incidence of 500,000 new cases and mortality of 270,000 women recorded in 2018 (1). Patients with high levels of estrogen and the estrogen receptor alpha (ERα) are at increased risk of developing this type of tumor (2). Recently, our group described the tumor-promoting effect of 17β-estradiol (E2) through the induction of changes in cervical cancer cells metabolism (3). It remains to define how E2 produces these modifications.

Metabolism is one of the most complex processes in biological systems. The effect of the metabolism in cancer cells was first described by Otto Warburg in 1924 (4). Several hypotheses have been proposed to explain how tumor cells activate glycolysis under aerobic conditions. Asgari et al (5) described seven subsystems that form the basis of the Warburg effect, including glutamine metabolism, nucleotides, glycolysis, oxidative phosphorylation system (OXPHOS), pentose phosphate pathway, tricarboxylic acid (TCA) cycle and pyruvate metabolism. Even though Warburg postulated impaired mitochondrial respiration in tumor cells (4), the role of mitochondria in the malignant transformation is still not well understood.

Mitochondria are organelles with vital roles in cellular energy production. Functionally, they are best known for their ability to generate the majority of ATP and free radicals from NADH and FADH using molecular oxygen (O2) via electron transfer coupled with the OXPHOS (5). Also, they are involved in the modulation of various cellular processes, such as the intracellular calcium homeostasis, fatty acid oxidation, urea cycle, biosynthesis of amino acids, lipids, hemes and purines, and other central metabolic pathways. A close link between attenuated mitochondrial bioenergetics and enhanced glycolysis dependency is present in human tumor cells (6).

Estrogen has been shown to influence mitochondrial structure, biogenesis and activity (7). The majority of the biological effects of E2 are mediated via two ERs, namely ERα and ERβ. Both ERs are localized in the mitochondria and they play an important role in the regulation of the organelle structure and function (7,8).

Under various stress conditions, the major function of E2 is to maintain OXPHOS in mitochondria, explaining the neuroprotective and cardioprotective effects of estrogens (9,10). The stressed mitochondria produce excessive amounts of reactive oxygen species (ROS), which can damage in the lipid bilayers, mutate DNA and alter the activity of specific enzymes critical for the maintenance of oxidative function (11). Estrogens act as free-radical scavengers, and modify the cytosolic and mitochondrial influx of calcium ions (Ca2+) potentially providing protection to the cell from toxic Ca2+ influx (12). Collectively, these effects indicate that the hormone protects from mitochondrial membrane potential collapse, avoids the loss of the inner mitochondrial membrane integrity and inhibits the release of pro-apoptotic factors.

In tumor cells, there is a change in the metabolic profile to support proliferation and the increased biosynthetic demands. These cells have an alteration of the mitochondrial activity and a high rate of glycolysis, followed by increased lactic acid production under aerobic conditions (13). In this work, we tested the association between the activity of E2 and its ability to moderate the Warburg effect and the mitochondrial function in cell lines derived from cervical cancer with the aim to define the role of this hormone in the process of cervical carcinogenesis.

Materials and methods

Reagents

E2, phorbol myristate acetate (PMA; cat. no. PI585-1 mg), ionomycin calcium salt (cat. no. IO634), cisplatin (cat. no. 479306), sulfuric acid (cat. no. 7664-93-9), phosphotungstic acid hydrate (cat. no. 12501-23-4), 2-thio-barbituric acid (TBA; cat. no. 504-17-6) and metformin were purchased from Calbiochem (cat. no. D150959; Merck KGaA). 1-[4-(6-bromobenzo[1,3](8)dioxol-5yl)-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]quinolin-8-yl]-ethanone (G-1; cat. no. 10008933) was obtained from Cayman Chemical Company. DMEM, charcoal stripped fetal bovine serum (FBS) and antibiotics were purchased from Gibco (Thermo Fisher Scientific, Inc.). mitoCapture™ (cat. no. K250) was acquired through BioVision, Inc., and 1-butanol and Baker Analyzed™ A.C.S. were from Avantor, Inc.

Cell lines

Cervical carcinoma cell lines were obtained from the American Type Culture Collection, including HeLa [positive to human papillomavirus (HPV) 18], SiHa (positive to HPV 16) and C33A (negative to HPV). Non-tumorigenic HaCaT was cultured as mock control; cells maintained normal properties of differentiation and were kindly provided by Dr. Petra Boukamp (German Cancer Research Center; DKFZ). Cell lines were tested and authenticated by the Multiplexion's Multiplex Cell Line Authentication tests in September 2018 (www.multiplexion.de).

Cell culture

Cells were cultured in DMEM containing 1% antibiotic [penicillin G (10,000 U/ml) and streptomycin (10,000 µg/ml)], amphotericin B (250 µg/ml) and 10% of charcoal-stripped FBS at 37°C in a humidified atmosphere with 5% CO2. Cells were grown until reaching 70-80% confluence. Three to four passages were achieved before the experiments were performed.

Mitochondrial permeability

Cell lines were cultured on 8-well slides (cat. no. 177402; Sigma-Aldrich; Merck KGaA) at 5×103/well for HeLa and SiHa or at 10×103/well for C33A and HaCaT. Cells were stimulated with E2 (10 nM), as previously determined (3) and incubated for 24 h. Staining was performed using the MitoCapture™ staining according to the manufacturer's instructions and examined under a fluorescence microscope coupled to a digital camera (magnification, ×10 and ×40). In healthy cells, dye aggregates and stains the mitochondria red; in abnormal cells, the mitochondrial membrane potential collapses and the dye remains in the cytoplasm in its green fluorescent monomeric form. As in recent years evidence suggested that low concentrations of chemotherapeutic agents, like cisplatin, facilitate malignancy of carcinoma cells (14,15), cisplatin (0.5 µg/ml) was used as control of cellular and mitochondrial damages and were compared with a basal control (no treatment). The evaluation of the mean optical density was determined using Image-pro Plus 6.0 (Media Cybernetics, Inc.) based on the red staining; data were normalized to the fluorescence of vehicle-treated cells. Each sample was tested by triplicate in three independent experiments.

Calcium pathway

The effect of E2 (10 nM) on intracellular Ca2+ concentration was measured using the Fluo-4 Calcium kit (cat. no. F36206; Molecular Probes; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The supernatant of the cells was obtained 48 h after applying E2. Ionomycin (750 ng/ml) and PMA (10 ng/ml) stimuli were applied for 5 min at 37°C before the analysis of the samples in the spectrophotometer. Excitation and emission wavelengths were 494 and 516 nm, respectively. Readings were recorded every minute over 16 min. G1 (1 µM) was used as a positive control alone or in combination with PMA (10 ng/ml) and ionomycin (750 ng/ml) and were compared with a basal control.

Nitric oxide (NO) concentration

Cells (5×103) were seeded in 96-well plates with DMEM containing 10% charcoal-stripped FBS. Cells were stimulated with E2 (10 nM) and/or metformin (10 mM) and were compared with a basal control. After 48 h, nitrite concentrations were measured using the Measure-iT™ High-Sensitivity Nitrite assay (Invitrogen; Thermo Fisher Scientific, Inc.). The amount of nitrite is extrapolated to NO concentrations following the manufacturer's instructions. Reagent (Measure-iT™ nitrite quantitation reagent; 2.0 ml of 100X in 0.62 M HCl; 1:100; 100 µl) was added to the wells and samples (10 µl) were added. After 10 min at room temperature, 5 µl developer solution was added and the fluorescence was measured using a spectrophotometer (excitation and emission wavelength, 360 and 460 nm, respectively).

Lipid peroxidation

The levels of lipid peroxides were calculated via determining the malondialdehyde (MDA) concentrations. Cells were stimulated with E2 (10 nM) and/or metformin (10 mM) and were compared with a basal control. After 48 h, 300 µl cell supernatant was mixed with 2 ml sulfuric acid (N/12), 0.3 ml 10% phosphotungstic acid and 1 ml 0.6% TBA. The mixture was placed in a boiling water bath at 95°C for 1 h, followed by cooling at room temperature. Then, 1.3 ml of n-butane was added, mixed vigorously and centrifuged at 1,400 × g for 15 min. The absorption of the organic phase was recorded at 534 nm.

Concentration of iron and ferritin

Cell supernatants were used to determine iron concentrations and to evaluated ferritin levels. Cells were stimulated with E2 (10 nM) and incubated for 48 h and were compared with a basal control. The VITROS Fe slide method was performed using the VITROS Fe slides and the VITROS Calibrator 4 kit (Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The difference in the density of the reflectance after 1-5 min is proportional to the concentration of iron in the samples. The determination of the ferritin levels was determined using the same samples with the IMMULITE® kit (cat. no. L2KFE2; Siemens Healthineers) following the manufacturer's instructions.

Concentration of glucose

Cells were stimulated with E2 (10 nM) and/or metformin (10 mM) and were compared with a basal control. The supernatant was obtained after 48 h. Glucose concentrations were measured in the BS-120 Clinical Chemistry analyzer (Shenzhen Mindray Bio-Medical Electronics Co., Ltd.) using the glucose oxidase method (COD11803; BioSystems S.A.) following the manufacturer's instructions. Glucose in the samples generated a colored complex that was quantified spectrophotometrically at 500 nm.

Concentration of lactic acid

Cells were stimulated with E2 (10 nM) and/or metformin (10 mM) and were compared with a basal control. The supernatant was obtained after 48 h. The VITROS LAC Slide method was performed using the VITROS LAC slides and the VITROS Calibrator kit 1 (Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. Lactate in samples is oxidized to pyruvate and hydrogen peroxide (H2O2). The generated H2O2 oxidizes 4-aminoantipy-rine, which forms a complex with 1,7-dihydroxynaphthalene. Absorbance at 540 nm was determined spectrophotometrically.

RNA extraction

Cells were stimulated with E2 (10 nM) and incubated for 48 h and were compared with a basal control. Total RNA was extracted from SiHa and HeLa using the RNeasy Plus Mini kit (cat. no. 74136; Qiagen, Inc.), according to the manufacturer's instructions. RNA was quantified measuring the absorbance at 260/280 nm.

RNA sequencing

Transcriptome sequencing was performed using RNAseq technology. The samples (SiHa and HeLa cells) stimulated with E2 were sent to the Genome Institute (Novogene Corporation, Inc.), where the Illumina HiSeq 4000 system was used; the sequencing strategy included a 250-300 bp insert cDNA library, HiSeq platform, paired-end 150 and 50 bp single-end. The RNA quality was measured and selection criteria were: Amount ≥1 µg; volume ≥20 µl; concentration ≥50 ng/µl; RNA integrity number value ≥7; 28S/18S >1; no degradation or pollution. Subsequently, fragmentation of mRNA and synthesis of cDNA of the first and second chain was performed. For the mRNA isolation and fragmentation, 200 ng total RNA was purified using oligo-dT beads and poly (A)-containing mRNA were fragmented into small pieces with Fragment Buffer. For the cDNA synthesis, first-strand cDNA was generated using the First Strand Master Mix and Super Script II (Invitrogen; Thermo Fisher Scientific, Inc.) for 25°C for 10 min, 42°C for 50 min and 70°C for 15 min. Then Second Strand Master Mix was added to synthesize the second-strand cDNA (16°C for 1 h).

cDNA fragments were purified, resolved for the final repair and the adenine single nucleotide addition was performed. cDNA fragments were ligated using adapters and fragments with a suitable size were selected for PCR amplification. The library construct was prepared for the quantification and qualification of libraries. For the end repair, the purified fragmented cDNA was combined with End Repair Mix and incubated at 30°C for 30 min. The end-repaired DNA was purified with Ampure XP beads (Thermo Fisher Scientific, Inc.) and incubated with A-Tailing Mix at 37°C for 30 min. For the adapter ligation, the adenylated 3'-DNAs were combined and RNA Index Adapter and Ligation Mix was added at 30°C for 10 min. The end-repaired DNA was purified with Ampure XP beads. Several rounds of PCR amplification with the PCR Primer Cocktail and PCR Master Mix were performed to enrich the cDNA fragments and products were purified with the Ampure XP beads.

For the validation of the library, quantitation was performed in two ways: (i) Determination of the mean molecule length using the Agilent 2100 bioanalyzer (Agilent Technologies, Inc.); (ii) reverse transcription-quantitative PCR (TaqMan Probe; Illumina, Inc.). Library sequencing was performed as follows: (i) Amplification within the flow cell on the Bot instrument for cluster generation (HiSeq® 4000 PE Cluster Kit; Illumina, Inc.); and (ii) loading of the clustered flow cell onto the HiSeq 4000 Sequencer for paired-end sequencing (Illumina; Inc.) with recommended read lengths of 100 bp.

Downstream analysis was performed using a combination of programs, including FLEXBAR version 2.5 for filtration (16). Alignments of readings with index hg38 were parsed employing Kallisto version 0.44.0 (17). Differential expression analysis between groups (two biological replicates per group) was performed using the DESeq2 R package version 1.24.0 (18). The resulting P-values were adjusted applying the Benjamini-Hochberg approach to control the false discovery rate (FDR). Genes with an adjusted P<0.05 were considered as differentially expressed. An absolute fold change of 1 was considered in the significant differential expression. Gene Ontology (19) and KEGG enrichment (20,21) were implemented by the Cluster Profiler R package (https://www.R-project.org). A corrected P<0.05 was considered significantly enriched by differential expressed genes (DEGs).

Cell proliferation assay

Cell proliferation of HeLa, SiHa, C33A and HaCaT cells was assessed using the xCELLigence platform (Roche Applied Science). Cells were cultured at 0.25×104/well in 96-well E-plates (cat. no. 058416; ACEA Biosciences, Inc.) and were introduced into the xCELLigence reading station. After 4 h, cells were stimulated with 10 nM E2, 10 mM metformin, 1 µg/ml cisplatin alone or in combination (E2+metformin, E2+cisplatin and E2+metformin+cisplatin), a vehicle-stimulated (DMEM+ethanol) cells and were compared with basal control. The proliferation rate was determined in real time every 30 min over 72 h. Cisplatin stimulation was used as a cell death control. Three independent experiments were performed with three replicates in each case.

Statistical analysis

Statistical analyses were made using SPSS 20 (IBM Corp.) and GraphPad Prism 6 (GraphPad Software, Inc.). Experiments were performed in triplicate and tested in three independent trials. One-way ANOVA with Bonferroni correction was used to determine significance following Levene's test for homogeneity of variances. Data are presented as the mean ± standard error of the mean. P<0.05 was considered to indicate a statistically significant difference.

Results

E2 decreases the mitochondrial membrane permeability in cervical cancer cell lines

The effect of E2 on the mitochondrial permeability of cervical cancer and HaCaT cell lines was evaluated using a selective dye. In HeLa, SiHa and C33A, E2 significantly decreased the mitochondrial permeability compared with the basal control (P<0.0001; Fig. 1). The green color indicated that the dye accumulated in the cytoplasm. In HaCaT cells, the opposite effect was observed with increased mitochondrial permeability following E2 treatment as suggested by the red color (P<0.0001).

Figure 1 E2 affects mitochondrial permeability of cervical cancer cells. Mitochondrial staining was performed in HeLa, SiHa and C33A cervical cancer cells and HaCaT following treatment with E2 or metformin; red, healthy cells and green, abnormal cells; magnification, x40. The quantification refers to the mean optical density of red stain. Experiments were performed in triplicate and with three repeats. *P<0.05. E2, 17β-estradiol.

E2 alters the calcium pathways in SiHa and HeLa

E2 had no significant effect on the calcium concentrations in cervical cancer cells or HaCaT compared with the basal control (P>0.05; Fig. 2). The use of ionomycin plus PMA significantly increased the ion concentration compared with the basal control (HeLa, P=0.003; SiHa, P=0.001; C33A, P=0.001). Exposure to E2, ionomycin and PMA significantly decreased the calcium concentration in SiHa and HeLa compared with the associated control (P<0.05), suggesting that E2 altered the calcium signaling pathway. In C33A and HaCaT, E2 did not significantly modify the cellular response to stimulation with ionomycin plus PMA (P>0.05). G1, a selective agonist of the ER in the membrane (GPR30) associated with non-genomic effects, such as calcium mobilization, was used as positive control. E2 and G1 exhibited similar behaviors in all cell lines.

Figure 2 E2 affects the calcium regulation in cervical cancer cells. Ca2+ levels were determined in HeLa, SiHa and C33A cervical cancer cells and HaCaT following various treatments. Experiments were performed in triplicate and with three repeats. *P<0.05. E2, 17β-estradiol; PMA, 1,1,3,3-tetramethoxypropane; G1, 1-[4-(6-bromobenzo[1,3]dioxol-5yl)-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]quinolin-8-yl]-ethanone.

E2 has an antioxidant effect in cervical cancer cell lines

To evaluate the participation of E2 in other mechanisms involved in the mitochondrial activity, we analyzed whether E2 modulated oxidative stress. First, we determined the effect of E2 on the concentration of NO in cervical cancer cells, as well as in HaCaT. E2 significantly decreased NO levels in HeLa, SiHa and C33A compared with the basal control (P=0.0002, P=0.0014 and P=0.0003; respectively; Fig. 3). This effect was significantly reversed when treating cells with metformin and E2 (P<0.05). In HaCaT cells, no differences were observed following treatment with E2, metformin or a combination of these compared with the control.

Figure 3 E2 affects nitric oxide production in cervical cancer cells. Nitric oxide levels were determined in HeLa, SiHa and C33A cervical cancer cells and HaCaT following various treatments. Experiments were performed in triplicate and with three repeats. *P<0.05. E2, 17β-estradiol.

Furthermore, we analyzed the concentration of MDA, which is a final product of the lipid peroxidation and is used as a direct indicator of cell damage (22). E2 induced a significant decrease in lipid peroxidation in HeLa (P=0.0004) and SiHa (P=0.0005; Fig. 4) compared with the basal control. In C33A, the effect was the opposite and lipid peroxidation was significantly increased following E2 treatment (P=0.0002). Treatment with metformin significantly alleviated E2 effects in HeLa and C33A cells (P<0.05), but no significant effects were observed in SiHa (P>0.05). In HaCaT cells, no significant differences were observed following treatment with E2, metformin or a combination of these compared with the control.

Figure 4 E2 affects lipid peroxidation in cervical cancer cells. Lipid peroxide levels were determined in HeLa, SiHa and C33A cervical cancer cells and HaCaT following various treatments. Experiments were performed in triplicate and with three repeats. *P<0.05. E2, 17β-estradiol.

E2 does not modify iron metabolism in cervical cancer cells

The role of iron in lipid peroxidation has been associated with a form of cell death other than apoptosis or necrosis. Ferritin binds to iron and this complex is vital in controlling ROS generation (22). E2 stimulus did not significantly modify intracellular iron or ferritin levels in cervical cancer cells compared with the basal control (P>0.05; Fig. 5). In HaCaT, intracellular iron was significantly decreased following E2 treatment compared with the basal control (P=0.007).

Figure 5 E2 affects the iron metabolism in cervical cancer cells. Intracellular iron and ferritin levels were determined in HeLa, SiHa and C33A cervical cancer cells and HaCaT following E2 treatment. Experiments were performed in triplicate and with three repeats. *P<0.05. E2, 17β-estradiol.

E2 favors the Warburg effect

The mitochondrial bioenergetics dysfunction favors the activation of the glycolytic pathway in different types of tumors (13). In HeLa, SiHa and C33A supernatants, there was a significant decrease in glucose levels following E2 stimulus (P=0.0001, P=0.002 and P=0.0003; respectively; Fig. 6), suggesting that the cells increased the uptake of glucose. A similar significant decrease was observed using metformin treatment (P<0.01). The combination of both treatments showed glucose levels similar to those observed in the basal control (P>0.05). It was hypothesized that the hypoglycemic effect of metformin was associated with E2. Furthermore, E2 significantly increased lactic acid levels in HeLa and SiHa compared with the basal control (P=0.0001 and P=0.004; respectively). The effects were more pronounced with metformin or a combination treatment with E2 (P<0.001).

Figure 6 E2 affects the Warburg effect in cervical cancer cells. Glucose and lactic acid ferritin levels were determined in HeLa, SiHa and C33A cervical cancer cells and HaCaT following E2 treatment. Experiments were performed in triplicate and with three repeats. *P<0.05. E2, 17β-estradiol.

E2 modulates DEGs in SiHa and HeLa

Considering that E2 modified the metabolism of the cervical cancer cells, we analyzed the effects of E2 on the expression of molecules associated with metabolic pathways using the RNA sequencing technology. The total number of DEGs in SiHa following E2 treatment was ~3-fold higher than that in HeLa cells (Fig. 7). E2 stimulus had an impact on various metabolic pathways, including the upregulation of genes involved in the synthesis of proteins associated with the mitochondrial electron transport chain, which may affect OXPHOS and the energy metabolism (Table I). Additionally, upregulation of genes associated with glycolysis was observed (Table I and II), supporting the theory of the role played by E2 in the regulation of the Warburg effect associated with cervical cancer (Fig. 8). Defects in the metabolism of lipids, amino acids, proteoglycans, nucleotides, inositol phosphate, and choline have been reported in different types of cancer (23-25). E2 stimulation modulated various metabolic pathways (Table I). Table II summarized the DEGs shared by HeLa and SiHa.

Figure 7 E2 affects expression of genes belonging to metabolic pathways in HeLa and SiHa. All experiments were performed in duplicate four times. E2, 17β-estradiol; DEG, differentially expressed gene.

Figure 8 Schematic representation of the involvement of E2 in metabolic pathways in HeLa and SiHa cells; with upregulated genes in purple, downregulated genes in brown, metabolic intermediates and enzymes in red, metabolic pathways in light green and hormone receptors in blue. E2, 17β-estradiol; ER, estrogen receptor; ROS, reactive oxygen species.

Table I Differential expression of genes associated with the metabolism in HeLa and SiHa cells following 17β-estradiol stimulation.

Table I

Differential expression of genes associated with the metabolism in HeLa and SiHa cells following 17β-estradiol stimulation.

A, HeLa
Pathway	N	Downregulated genes	Upregulated genes
Oxidative phosphorylation	24	NDUFV1, COX6B2, TCIRG1	COX6C, NDUFAB1, NDUFA4, NDUFS5, COX5B, COX7C, COX7B, ATP5F1A, ATP5F1B, ATP5F1E, ATP5MC2, APT5MC3, ATP5MG, ATP5PB, ATP5PO, COX6A1, COX6B1, COX8A, UQCRB, UQCRH, UQCRQ
Glycolysis/gluconeogenesis	6	PFKP	GAPDH, LDHA, LDHB, PGK1, TPI1
Glycan biosynthesis and metabolism	18	NDST1, SGSH, LSS, MGAT5B, PIGQ, FUK, MGAT1, GPAA1, GAA, MAN2A2, MOGS, PIGQ, B4GALT2	EXT1, PIGK, GALNT1, RPN2, TUSC3
Lipid metabolism	14	AGPS, FASN, ACADVL, LSS, ACSL4, DGKD, SPHK1, MGAT1	PTDSS1, PTGES3, CDS2, HMGCS1, PLA2G4A, PPT1, CPS1
Metabolism of cofactors and vitamins	8	RDH10, NADK, NAPRT	MTHFD2, NAMPT, EPRS, RFK, PANK3
Amino acid metabolism	12	DNMT1, ANPEP, OPLAH	MDH1, CPS1, LDHA, HMGCS1, ODC1, TPI1, LDHB, GAPDH, CPS1
Nucleotide metabolism	10	POLD1, POLE, CAD, AMPD2, IMPDH1	PAICS, DUT, GMPS, HPRT1, PAPSS2
Choline metabolism	4	DGKD, PIP5K1C	PLA2G4A, PPT1
Inositol phosphate metabolism	4	PLCD3, PIP5K1C, INPPL1	TPI1
B, SiHa
Pathway	N	Downregulated genes	Upregulated genes
Oxidative phosphorylation	45	ATP5F1D, ATP6AP1, ATP6V0C, ATP6V0E1, ATP6V0E2, CYC1, NDUFA1, NDUFA11, NDUFS6, NDUFS7	ATP6V1E1, ATP5MC3, NDUFA4, NDUFA5, NDUFA8, NDUFAB1, NDUFB3, NDUFB4, NDUFS5, NDUFV2, ATP6V1B2, ATP6V1C1, COX15, ATP5F1C, ATP5F1E, ATP5MC2, ATP5ME, ATP5MG, ATP5PB, ATP5PD, ATP5PF, ATP5PO, COX5A, COX5B, COX6A1, COX6C, COX7B, COX7C, ATP5F1A, ATP5F1B, SDHB, SDHC, UQCRB, UQCRC2, UQCRH
Glycolysis/gluconeogenesis	13	ALDH1A3, ALDH3B1, ALDOA, PFKL, PFKP	DLAT, ALDH7A1, PGK1, TPI1, PKM, LDHA, LDHB, ADH5
Glycan biosynthesis and metabolism	40	ALG12, ALG14, B3GALT6, B3GAT3, B4GALNT1, B4GALT7, CHPF, GAA, GALNT2, GPAA1, HYAL2, IDUA, LSS, MAN1B1, MAN2A2, MGAT4B, MGAT5B, MOGS, NAGLU, NDST1, SGSH, ST3GAL1, ST3GAL2, ST6GALNAC4, ST6GALNAC6, XYLT1, XYLT2	B4GALT1, FKTN, C1GALT1C1, ALG8, EXT1, EXTL2, CSGALNACT2, PIGF, PIGK, PIGN, DPM1, STT3A, TUSC3
Lipid metabolism	59	LSS, ACADVL, ADO, AGPAT2, AGPS, B4GALNT1, CAD, CBR1, CDIPT, CKB, DGKQ, DGKZ, DHCR7, FASN, GPAT4, HSD17B1, MCAT, MGAT1, MGLL, PAFAH1B3 PGP, PGS1, PISD, PLPP1, PLPP2, PNPLA2, PTDSS2, PTGES2, SGMS2, SMPD4, SPHK1, SPHK2, ST3GAL1, ST3GAL2, TKFC, ST6GALNAC4, ST6GALNAC6	LTA4H, ACSL3, ACADSB, PTGES3, PAFAH1B1, ELOVL5, HACD3, HADH, HADHB, CRLS1, ETNK1, KDSR, SPTLC2, CYP51A1, CPS1, B4GALT1, ALDH7A1, ACAT1, HADHA, PPT1, SMS, PLA2G4A
Metabolism of cofactors and vitamins	21	ALPL, ALPP, COASY, FPGS, NADK, NAPRT, PANK4, PDXK, PDXP, PNPO, SHMT1, SHMT2	ACP1, DHFR, NNT, MTHFD1, PANK1, EPRS, PNP, PSAT1, COX15
Amino acid metabolism	56	ALDH6A1, ALDOA, AMD1, AMDHD1, ASS1, BCAT2, SRM, BCKDHA, CAD, CBS, CHDH, DNMT3A, FAHD1, GAMT, GLUL, GPT2, MPST, MRI1, NAT8L, OAT, OGDH, OPLAH, PGP, PHGDH, PYCR1, PYCR3, SHMT1, SHMT2	GOT2, PSAT1, ASNS, AHCYL2, RIMKLB, P4HA1, SAT1, ODC1, MAT2B, GCLM, MAT2A, KYNU, HIBADH, MCCC2, ACAA2, ATIC, IDH1, PGK1, TPI1, PKM, LDHA, LDHB, GCSH, DBT, HIBCH, HADHA, ACADSB, CPS1
Nucleotide metabolism	39	IMPDH1, ADK, AMPD2, PFAS, GUK1, POLD1, POLD2 TK1, POLR1A, POLR2E, POLR2F, POLR2L, POLR3H, NME3, NME4, POLD4, DCTPP1, TYMP	GMPS, HPRT1, ADSL, ADSS, PAICS, HPRT1, IMPDH2, GART, POLA1, POLD3, POLE3, POLR2B, POLR2G, PRIM2, ME1, NME7, NT5C3A, CMPK1, UMPS, PKM, NME1
Choline metabolism	8	CKB, DGKQ, DGKZ, PLPP1, PLPP2, PIP5K1C	PLA2G4A, PPPI
Inositol phosphate metabolism	4	PLCD3, PIP5K1C, INPPL1	TPI1

Table II Differentially expressed genes shared by HeLa and SiHa associated with the metabolism following stimulation with 17β-estradiol.

Table II

Differentially expressed genes shared by HeLa and SiHa associated with the metabolism following stimulation with 17β-estradiol.

Pathway	N	Downregulated genes	Upregulated genes
Oxidative phosphorylation	19	TCIRG1	COX6C, NDUFAB1, NDUFA4, NDUFS5, COX5B, COX7C, COX7B, ATP5F1A, ATP5F1B, ATP5F1E, ATP5MC2, ATP5MG, ATP5PB, ATP5PO, COX6A1, UQCRB, UQCRH, MDH1
Glycolysis/gluconeogenesis	5	PFKP	LDHA, LDHB, PGK1, TPI1
Glycan biosynthesis and metabolism	10	GAA, NDST1, SGSH, PIGQ, MGAT5B, GPAA1, MGAT1	EXT1, PIGK, TUSC3
Lipid metabolism	10	MGAT1, AGPS, FASN, ACADVL, CAD, SPHK1	CPS1, PTGES3, PLA2G4A, PPT1
Metabolism of cofactors and vitamins	3	NADK, NAPRT	EPRS
Amino acid metabolism	9	ANPEP, OPLAH, CAD	MDH1, LDHA, LDHB, ODC1, PGK1, TPI1, CPS1
Nucleotide metabolism	9	FUK, POLD1, AMPD2, IMPDH1, PFAS	PAICS, GMPS, HPRT1, NME1
Choline metabolism	3	PIP5K1C	PLA2G4A, PPT1
Inositol phosphate metabolism	4	PLCD3, PIP5K1C, INPPL1	TPI1

E2 regulation of the differential expression of transcripts with functional roles in the processes of energy production was demonstrated by evaluating the impact on OXPHOS, glycolysis/gluconeogenesis, lipid processing, vitamins and cofactors, as well as on the metabolism of amino acids, nucleotides, phosphatidylinositol and choline (Table I). E2 regulated the expression of genes involved in mitochondrial complexes as follows: (i) Complex I, NDUFAB1, NDUFA4, and NDUFS5 upregulated; (ii) complex III, UQCRB and UQCRH upregulated; (iii) complex IV, COX6C, COX5B, COX7C, COX7B and COX6A1 upregulated; (iv) complex V, ATP5F1A, ATP5F1E, ATP5F1B, ATP5MC2, ATP5MG, ATP5PB and ATP5PO upregulated, and TCIRG1 downregulated.

Absolute changes in DEGs ranged between −0.4985 and 0.2496. A total of 5 genes associated with glycolysis were modified in both HeLa and SiHa by E2 stimulus, including LDHA, LDHB, PGK1, TPI1 and PFKP (absolute changes, −0.1893 to 0.2598). Other gene alterations modulated by E2 include: (i) Metabolism and synthesis of glycans (n=10), with GAA, NDST1, SGSH, PIGQ, MGAT5B, GPAA1 and MGAT1 downregulated and EXT1, PIGK and TUSC3 upregulated; (ii) lipid metabolism (n=10), with MGAT1, AGPS, FASN, ACADVL, CAD and SPHK1 upregulated and CPS1, PTGES3, PLA2G4A and PPT1 downregulated; (iii) metabolism of amino acids and nucleotides (n=9); (iv) metabolism of inositol phosphate (n=4); and (v) metabolism of choline, vitamins and cofactors (n=3; Table II and Fig. 8).

A total of 12 signaling pathways were affected by E2, including hypoxia-inducible factor (HIF)-1, Ras, mitogen-activated protein kinase (MAPK), vascular endothelial growth factor (VEGF), phospholipase D, 5'-AMP-activated protein kinase (AMPK), cGMP-dependent protein kinase (PKG), calcium, appeline, phosphatidylinositol, PI3K-Akt and pathways mediated by sphingolipid (Table III). It should be noted that of these, genes that affected HIF-1, Ras, MAPK, VEGF and phospholipase D pathways that were upregulated included PGK1, LDHA and PLA2G4A (absolute changes, 0.1606, 0.2167 and 0.2494, respectively). Decreased genes following E2 stimulus included PFKP (−0.1893), FASN (−0.5459), CAD (−0.3228), SPHK1 (−0.3164), PLCD3 (−0.5829), INPPL1 (−0.7263) and LAMA5 (−0.7294), corresponding to the AMPK, cGMP-PKG, calcium, sphingolipid-mediated, appeline, phosphatidylinositol and PI3K-Akt signaling pathways (Table III).

Table III Differentially expressed genes affecting signaling pathways associated with metabolism shared by HeLa and SiHa after stimulation with 17β-estradiol.

Table III

Differentially expressed genes affecting signaling pathways associated with metabolism shared by HeLa and SiHa after stimulation with 17β-estradiol.

A, Downregulated genes
Signaling pathway	Gene	P-value
VEGF	SPHK1	1.6706x10-2
Phospholipase D	SPHK1	1.6706x10-2
	PIP5K1C	7.1061x10-3
AMPK	PFKP	1.0293x10-2
	FASN	1.2382x10-34
cGMP-PKG	CAD	1.5215x10-4
Calcium	SPHK1	1.6706x10-2
	PLCD3	8.8258x10-7
Sphingolipid	SPHK1	1.6706x10-2
Apelin	SPHK1	1.6706x10-2
Phosphatidylinositol	PLCD3	8.8258x10-7
	PIP5K1C	7.1062x10-3
	INPPL1	1.7603x10-11
PI3K-Akt	LAMA5	1.1881x10-85
B, Upregulated genes
Signaling pathway	Gene	P-value
HIF-1	PGK1,	4.7152x10-4
	LDHA	6.4214x10-7
Ras	PLA2G4A	9.7518x10-3
MAPK	PLA2G4A	9.7518x10-7
VEGF	PLA2G4A	9.7518x10-7
Phospholipase D	PLA2G4A	9.7518x10-7

Fig. 8 outlines important DEGs regulated by E2 and those directly involved in mitochondrial function, metabolism and associated signaling pathways. The impact of E2 in the pathophysiology of cervical cancer was suggested to be associated with the dysregulation of the respiratory mitochondrial electron chain, increasing OXPHOS, and activating glycolysis and the pentose phosphate pathway (Warburg effect). E2 was also suggested to inhibit the degradation of amino acids and provide metabolites for the synthesis of nucleotides, nucleic acids and coenzymes. Furthermore, it was hypothesized that E2 stimulates the synthesis of fatty acids associated with steroids, leukotrienes and prostaglandins metabolites, while it reduces the β-oxidation of lipids. It was suggested that E2 reduces oxidized glutathione, which regulates the oxidative state of tumor cells.

E2 does not modify metformin antiproliferative effect

Metformin is proposed as an antitumor drug (26). The most relevant mechanism of its activity is the regulation of cellular metabolism (27). The potential role of E2 in metformin-induced cell death in cervical cancer was assessed in this study. In cervical cancer cells treated with metformin a significant decrease in proliferation was observed compared with the basal control at incubation times >20 h (P=0.0001; Fig. 9). In HeLa, the effect of metformin was similar to that induced by cisplatin, thus metformin was suggested to be a potential anti-carcinogen. E2 treatment did not markedly alleviate effects exerted by metformin.

Figure 9 Effects of E2 and metformin on proliferation of cervical cancer cells. Proliferation of HeLa, SiHa, C33A and HaCaT cells is presented for 24 h intervals over 72 h following various treatments. Cisplatin was used as a cell death control. Experiments were performed in triplicate and with three repeats. E2, 17β-estradiol.

Discussion

Chronic and excessive exposure to estrogen is associated with an increased cancer risk incidence and the classical mechanism to explain this association is that estrogen affects the cell proliferation (28). Previously, effects of E2 on the mitochondrial metabolism were shown by increased mitochondrial activity based on MTT assays, which are known to alter the complex IV levels and therefore inferred a modification in mitochondrial metabolism (3). However, in this study we approached other aspects associated with this alteration in cervical cancer, such as the influence of the hormone in mitochondrial permeability, calcium mobilization, NO levels, the Warburg effect and modifications of pathways important in tumorigenesis.

The presence of E2 decreased the mitochondrial membrane permeability, which was in response to several factors, such as calcium concentration, oxidative stress and ATP depletion. It has been shown that estrogen affects the mitochondrial function through the modulation of Ca2+ levels. Excessive Ca2+ loading coupled with oxidative stress triggers a collapse of the mitochondrial membrane potential (Δψm) accompanied by edema of the matrix and initiation of apoptosis due to the release of cytochrome c (10). E2 protects mitochondria by preventing Δψm collapse and this may explain the described ability of estrogen to prevent the release of apoptotic factors (8). Intracellular calcium signals are critical for mitochondrial-nuclear crosstalk and the maintenance of cellular homeostasis (12). In the central nervous system, E2 exhibits neuroprotective characteristics preventing the influx of cytosolic and mitochondrial Ca2+ during excitotoxic stimulation (9,10). Additionally, the repression of Ca2+ mobilization by E2 may serve to protect cells from death during oxidative stress. In previous studies, it was observed that the use of E2 inhibits the effect of Ca2+ levels on pro-oxidant substances, such as 3-nitropropionic acid and H2O2 (9,11). In the present study, E2 inhibited the response induced by ionomycin and PMA with regards to the mobilization of calcium and it was hypothesized that E2 caused alterations in the calcium signaling pathway. Furthermore, the results suggested that E2 may protect cervical cancer cells.

NO is a well-known intracellular messenger in a range of physiological processes and a regulator of cell death and survival. It modulates apoptotic pathways and long-term exposure to µmolar NO can induce cell death by apoptosis (13,28). When NO is combined with superoxide radicals, it generates compounds, such as peroxynitrite, that lead to irreversible protein nitration, enzymatic inactivation, DNA damage, disruption of the mitochondrial integrity and cell apoptosis either by direct contact or by inducing lipid peroxidation (29). It has been reported that E2 has an antioxidant mechanism by increasing a blockage of the sensitivity of mitochondria to calcium, enhancing antioxidant systems and decreasing the expression of NO synthetase (10–12,30–32); this may explain why no lipid peroxidation was observed in HeLa and SiHa. However other mechanisms may further be involved. The protective effect of E2 on the survival of the cervical cancer cells was demonstrated in our study by reporting a decrease in NO levels following E2 stimulus.

Excessive iron availability in mitochondria exacerbates oxidative stress and promotes membrane lipid peroxidation, affecting the integrity of the inner mitochondrial membrane and the mitochondrial function (33). Several studies showed the participation of estrogen in iron metabolism. In general, the hormone appears to protect the cells from a distinct nonapoptotic cell death (ferroptosis) because it lowers the levels of iron and ferritin, and it regulates the expression of other proteins that participate in iron metabolism (22,34–36). This effect was not observed in the present study, as we did not find any significant impact of E2 on iron and ferritin levels in cervical cancer cells. It is known that for rapid proliferation, cells enhance metabolic activity allowing for increased DNA biosynthesis; this may describe why HaCaT cells to have high iron levels and require less ferritin. However, in SiHa, oxidative stress may be controlled through maintaining lower free iron levels, as intracellular iron may be harmful due to its ability to produce ROS (37–40).

The metabolic networks that give rise to metabolic phenotypes in most tumors are unknown. In this study, an essential role of E2 favoring the Warburg effect was demonstrated. It increased the glucose consumption and lactic acid levels in cervical cancer cell lines. Glucose utilization provides tumor cells with a constant energy supply, as well as precursors for de novo macromolecular biosynthesis that are essential for cell growth and proliferation (13). Estrogen substantially increased the glycolytic activities and glucose utilization in the uterus of ovariectomized rats in concert with the uterine growth (41). The hormone increases the glucose uptake in muscle and adipose tissues through the regulation of the expression of the glucose transporter 4 and decreases gluconeogenesis in the liver (36,42).

Although the Warburg effect predominantly focuses on glycolysis and suggesting attenuation in the mitochondrial metabolism, more recent works suggested that certain TCA cycle intermediates are elevated and important in the context of cancer (43–45). Thus, while glycolysis is drastically upregulated in most cancer cells, mitochondrial respiration continues to operate normally at rates proportional to oxygen supply.

To gain a better understanding of estrogen effects in the metabolism of cervical cancer, we performed a detailed analysis of the metabolic gene expression modulated by E2 in SiHa and HeLa cells, which are representative of ~99% of cervical tumors (46). E2 upregulated various genes associated with the protein components of the electron respiratory chain complexes. Consistent with these observations, previous work suggested that female liver mitochondria have greater capabilities of OXPHOS than the mitochondria from males (47). Mitochondrial (mt) DNA contains sequences similar to the estrogen response element (ERE) and glucocorticoid responsive elements (48). Chen et al (49) observed that recombinant human ERα and ERβ bind to mtEREs and this binding is enhanced by E2. Studies in which MTT is used, reported an increase in the conversion of MTT to formazan in the presence of E2 (50,51). Furthermore, nuclear genes that participate in the transcription of proteins of the respiratory chain of electrons are suggested to be modulated by estrogen. The expression of the genes involved in mitochondrial ATP synthase subunit β and F complex is upregulated by E2 in prostatic dysplasia (52), and the subunit c isoform of the F0 complex and mitochondrial ribosomal protein 3 are other nuclear estrogen-regulated genes identified in ERα-positive breast cancer cells (53).

In neuroblastoma, E2 has a protective effect against ATP depletion, Δψm decline and the generation of ROS induced by an inhibitor of the succinate dehydrogenase complex (54). There is potential that E2 mediated these effects via stimulating the synthesis of protein components of the electron respiratory chain complexes, and contributed to the preservation and regulation of the mitochondrial structure and functions. The presence of ERs has been associated with the preservation of ATP levels via enhancing OXPHOS and reducing ATPase activity, and thus, increasing mitochondrial respiration efficiency in neurons (55).

In association with these effects, E2 enhances the antioxidant systems in the brain by inducing the expression of glutathione, thioredoxin and superoxide dismutase proteins (30,32). With these observations and the findings reported in our bioenergetics analysis at the gene level, we hypothesized that in cervical cancer, E2 protects tumor cells from stress-induced death during malignant transformation partially by modulating mitochondrial respiratory chain structure and function.

Mitochondrial alterations as part of the adaptation of tumor cells to their microenvironment favor the Warburg effect (43). In this study, E2 modulated numerous genes associated with an increase in the expression of glycolytic proteins. The only gene downregulated by the hormone shared by HeLa and SiHa in this metabolic pathway was PFKP. The conversion of fructose-6-phosphate to fructose-1,6-bisphosphate catalyzed by phosphofructokinase-1 is the most important control step in mammalian glycolysis (56,57). Downregulation of PFKP potentiates cancer cell survival under metabolic stress through the maintenance of pentose phosphate pathway flow (57). Dynamic regulation of glucose flux between aerobic glycolysis and the pentose phosphate pathway provides the tumor cell with the metabolites needed to proliferate. NADPH production is an important strategy against oxidative stress and oxidative stress, not ATP depletion, is the primary trigger of cell death during metabolic stress (57). Downregulation of PFKP by E2 constituted a transcendent mechanism of evasion of cell death in difficult situations.

Our bioinformatics analysis suggested other metabolic pathways modulated by E2. Abnormal choline metabolism has emerged as a hallmarks of cancer, and increases in choline kinase-α expression and activity, a higher rate of choline transport and increased phosphatidylcholine-specific phospholipase C and D activities have been well documented within the pathophysiology of several cancers (57). The aberrant choline phospholipid metabolism of breast cancer cells and its correlation with the malignant progression has been reported (58). In our study, E2 was suggest to upregulate the expression of enzymes associated with the metabolism of choline and other lipids, which can affect the cholesterol metabolism and the synthesis of steroids, prostaglandins and leukotrienes. Physiologically, it has been reported that estrogen, in general, inhibits de novo lipogenesis in the liver (59), but increase cholesterol biosynthesis for cell proliferation (60).

In one study, a metabolomics approach for evaluating the estrogenic effects in uterus established that the TCA cycle, lipid metabolism, choline and amino acid metabolisms are disturbed (61). The reported alterations dependent on the ERs include: (i) Phenylalanine, tyrosine and tryptophan biosynthesis; (ii) alanine, aspartate, glutamate, glycerophospholipids and pyruvate metabolism; (iii) the citrate cycle; and (iv) glycolysis or gluconeogenesis. These observations are in accordance with our results and led to the hypothesis that E2 may exert physiological functions in an erroneous context within the pathophysiology of cervical cancer.

Another remarkable finding of our work was that E2 modulated signaling pathways that are essential for metabolism and tumorigenesis. The role of HIF-1 (62) and Ras-MAPK in the metabolism of cancer has been well documented (63,64). AMPK is a pivotal intracellular energy sensor in normal and tumor cells. Studies have been reported that high glucose levels reduce AMPK phosphorylation and activate the MAPK signaling pathway in cancer cells (64,65). Inactivation of AMPK accelerates tumorigenesis (66). There are no conclusive results about the effects of estrogen on AMPK expression in different tumors; some evidence suggests AMPK activation by ERα (67), while other evidence considers that the hormone inhibits AMPK by repressing the expression of its activating proteins (68). Interestingly, our study showed that E2 was suggested to decrease the expression of genes that indirectly repress AMPK.

Further observation of this work included that E2 modulated the differential expression of several genes associated with metabolic pathways. The number of DEGs in SiHa was 3 times greater than in HeLa; this may be explained due to these cells being derived from different epithelial cells (squamous cell carcinoma and adenocarcinoma) and that they are infected by two different virus genotypes (16 and 18, respectively) that may modify DEGs involved in signaling pathways, and therefore, affected pathogenicity mechanisms and treatment responses. Furthermore, our results indicated that, although from the same anatomical site, SiHa and HeLa displayed marked differences following treatment with E2. Regional differences in HPV genotypes may affect DEGs and the phenotype of the cervical cells. The difference in HPV genotype leads to different treatment responses with different clinical characteristics, pathogenicity mechanisms and histopathological subtypes (69). These results are essential for the design of new therapeutic models against cervical cancer.

Metformin has gained considerable attention as a potential anticancer agent. One of its main mechanisms of action is the suppression of the mitochondrial respiration by inhibiting the respiratory complex I (70). In human breast adipose stromal cells, it activates the liver kinase B1-AMPK signaling pathway and inhibits aromatase expression (71). Furthermore, the negative modulation of metformin on ERs, in particular ERα, has been demonstrated in numerous studies (72–74). In this work, the effect of metformin on E2 metabolic actions was evaluated. There are specific molecular features that control the uptake of biguanides into mitochondria (70), so we suggested that modifications in the mitochondrial permeability by E2 may be another factor that conditioned the entry of metformin in the organelle. In addition, although there are no specific transporters for biguanides, uptake is tissue-specific due to varying levels of distinct transporter expression (70). This evidence supported the different effects of metformin in the metabolic regulation of various cell lines, even when derived from the same type of tumor.

Measuring the metabolic responses of cervical cancer cell lines to E2 stimulus, allowed us to identify a distinct bioenergetics phenotype in cervical cancer that was the result of alterations in OXPHOS and the glycolytic energy metabolism. We concluded that E2 induced changes in the energy metabolism that allowed cell survival in high-stress conditions. The antioxidant and cytoprotective power of E2 in the context of cancer was demonstrated in this work.

Funding

This work was supported by CB 2017-2018 CONACYT (grant no. A1-S-51207) and the Health Research Fund from IMSS (grant no. FIS/IMSS/PROT/PRIO/15/046).

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors' contributions

ARL and PCOL performed the experimental work, searched for scientific literature, contributed to figures and the writing of the manuscript. LFJS, ARdA and JFMV participated in contributed to the data analysis and reviewed the manuscript. AAL, YMOG, CABG, RSM and SLdA performed experiments. APS conceived and designed the study and wrote the manuscript. All authors read and approved the final manuscript.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Acknowledgments

Not applicable.

References