Background
Colorectal cancer (CRC) is the third most commonly diagnosed cancer worldwide and ranks second for malignancy-related mortality [
1]. Despite recent progress in treatment, the prognosis of CRC remains poor because of the high metastasis and post-intervention recurrence rates [
2]. Chemotherapeutic drugs are the main treatment in CRC patients with progression and metastasis, and 5-fluorouracil (5-FU) is a widely used single-agent or key component of systemic chemotherapy for CRC treatment [
3,
4]. However, drug resistance can emerge after 5-FU treatment. Although 5-FU pro-drugs and several 5-FU-based combination chemotherapy regimens show beneficial effects, patients with advanced CRC to therapy continue to show poor prognosis due to the development of drug resistance [
5]. Therefore, elucidation of the underlying 5-FU resistance mechanisms is critical.
Metabolic reprogramming is one of the important hallmarks of cancer cells [
6,
7]. Metabolic reprogramming is characterized by reduced function of the mitochondrial oxidative phosphorylation (OXPHOS) system with energy compensation by glycolysis or the pentose phosphate pathway (PPP) in the presence of abundant oxygen in cancer cells. The primary site of energy production shifts from the mitochondria toward the cytosol [
8,
9]. These metabolic transitions ensure a supply of energy and provide building blocks, serving as a key contributor to tumor progression and chemotherapy resistance [
10,
11]. Strategies that target metabolic abnormalities will become an effective treatment alternative to enhance drug susceptibility, directly or indirectly [
12].
The transcription factor hypoxia inducible factor 1α (HIF-1α) is a key regulator that increases glycolysis and drives tumor development under anoxic conditions [
13,
14]. HIF-1α transcriptionally up-regulates glycolytic enzymes and membrane transporters to increase glucose flux and enhance glycolysis [
14,
15]. Under normoxia, HIF-1α has also been demonstrated as a major mediator in tumor progression and recurrence [
16]. It is worthy to note that the influence of HIF-1α expression on glucose metabolism and resistance of chemotherapy is yet to be fully explored in 5-FU-resistant CRC.
HIF-1α signaling is the classical response to the state of oxygen deficiency [
11,
15]. Furthermore, defining the mechanisms of HIF-1α stabilization and activation in normoxia is important to clarify the effect of the complex oxygen environment on the biological behaviors of tumors in vivo. However, the underlying mechanisms are still elusive. Reactive oxygen species (ROS) regulates HIF-1α levels, and HIF-1α also provides a negative feedback to ROS levels [
17‐
19]. The mechanisms of crosstalk between ROS and HIF-1α in regulating glycolysis and 5-FU resistance in CRC have never been reported. Additionally, several studies reported that the Wnt/β-catenin pathway is associated with glycolysis in CRC [
20], and up-regulated Wnt/β-catenin promotes resistance to chemotherapy in multiple cancers [
21]. HIF-1α is a target of β-catenin, and nuclear β-catenin cooperates with HIF-1α to regulate its transcriptional activity [
22]. Based on these facts, it is worth demonstrating whether β-catenin assists HIF-1α to modulate phenotypes in 5-FU resistance in CRC.
Here, we show that the metabolic reprogramming that facilitates 5-FU resistance in CRC arises from HIF-1α upregulation in non-classical ways, by ROS and the Wnt/β-catenin signaling pathway, independently from external oxygen concentrations. Our results provide new insights toward novel therapeutic strategies for suppressing the expression of HIF-1α, which restored the sensitivity of CRC to 5-fluorouracil. Moreover, we provide preliminary data into the potential value of HIF-1α as a biomarker for 5-FU resistance and assessing poor prognosis in CRC patients with 5-FU treatment.
Materials and methods
Cell lines
The following colorectal adenocarcinoma cell lines with STR profiling, were obtained from KeyGEN BioTECH, Jiangsu: HCT8 (male), HCT15 (male), HCT116 (male), LoVo (lymph node metastasis, male), SW480 (male), SW1116 (male), HT29 (female), Caco-2 (male), DLD-1 (male), and T84 (lung metastasis, male). The DiFi cell line was a generous gift of Dr. Li Luo. Cells were cultured in Roswell Park Memorial Institute (RPMI)-1640 medium supplemented with 10% fetal bovine serum (FBS), penicillin (100 mg/ml) and streptomycin (100 mg/ml), and cells were subcultured after dissociation with 0.25% trypsin-EDTA once they had reached 80% confluence. In general, all cultures were maintained in a 37 °C 5% CO2 incubator, or incubated in a hypoxia chamber at 94% N2, 5% CO2, and 1% O2 for certain experiments.
Mice
Male athymic BALB/c nude mice (4 weeks old) and male NOD/scid mice (4 weeks old) were purchased from SPF (Beijing) Biotechnology Co., Ltd. and housed in the animal center of Qianfoshan Hospital Affiliated to Shandong University. All procedures were approved by the Institutional Animal Care and Use Committee of Qianfoshan Hospital Affiliated to Shandong University. All animal studies complied with the relevant ethical regulations for animal testing and research.
Samples from CRC patients and the public GEO database
Fresh and neutral-buffered formalin-fixed tumor samples from 42 patients with CRC were collected from the Department of Gastrointestinal Surgery, Qianfoshan Hospital Affiliated to Shandong University and the Department of Colorectal Surgery, Qilu Hospital of Shandong University (Table S
1). At the time of tumor sample collection, 5 mL peripheral venous blood from the cubital vein and 5 mL blood from the tumor reflux veins were also obtained from these patients. An informed consent form was signed by each patient. The Ethics Committee of Qianfoshan Hospital Affiliated to Shandong University granted approval for this study. Datasets involved (GSE104645, GSE69657) were downloaded from the public Gene Expression Omnibus (GEO) database (
https://www.ncbi.nlm.nih.gov/gds/).
Stable acquired 5-FU resistance cell model generation
HCT8, HCT15, HCT116, LoVo, T84, and DiFi cell lines were continuously treated with a gradually increasing concentration of 5-FU rising from 10− 8 M to 10− 4 M initially. We assessed 5-FU resistance at each dose by calculating the IC50 using CCK8 assays. We defined 5-FU resistance as a resistance index (RI, IC50 of the WT cells/IC50 of the 5-FU-R cells) > 10 after 5-FU treatment of approximately 8 months duration. We succeeded in generating stable cells lines of HCT8 (RI = 396.63), HCT15 (RI = 110.84), and LoVo (RI = 347.60) with acquired 5-FU resistance.
Knockdown of HIF1A and CTNNB1
Small interfering RNA (siRNA) duplexes and negative control (NC) targeting HIF1A and CTNNB1 were transfected into cells by overnight incubation using Lipofectamine RNA iMAX reagent according to the manufacturer’s protocol. For lentivirus constructs, the short hairpin RNA (shRNA) for HIF1A was cloned into a hU6-MCS-Ubiquitin-EGFP-IRES-puromycin lentiviral (GV248) vector. After lentiviral-mediated transfection and puromycin selection, transfection efficiency was verified by RT-qPCR and WB. The sequences of siRNA and shRNA targeting HIF1A or CTNNB1 are listed in an additional file.
CCK8 cytotoxicity assay
For cell cytotoxicity assays, 5000 cells per well were seeded into 96-well plates 24 h before treatment. The cells were then treated with the indicated agents for 48-72 h, after which 10% v/v of water-soluble WST-8 dye was added for 0.5-2 h. The absorbance of the formazan was measured at 450 nm using a microplate reader after mixing gently on an orbital shaker for 1 min. All experiments were performed with at least three replicates.
Cell morphology visualization
Cells were visualized using the CytoPainter phalloidin-iFluor 488 reagent. Fixed and permeabilized cells were stained with phalloidin-iFluor 488 in 1% BSA in PBS for 1 h to label the F-actin. DAPI was used to stain the nuclei. Slides were covered with mounting medium containing DAPI and photographed using an Axio Scope A1 microscope (Zeiss, Germany) at 400× magnification.
5-ethynyl-2′-deoxyuridine (EdU) cell proliferation assay
Cells at 5000 per well were plated in 96-well plates for 24 h, and then treated with 10− 5 M 5-FU for 48 h. After incubation with 10 μM EdU for a further 2 h, the cells were fixed in 4% paraformaldehyde (PFA) and stained with Click-iT reaction solution at 100 μL per well. Hoechst 33342 was used to stain the nuclei. Images were obtained using an ImageXpress Micro High Content Screening System (Molecular Devices, USA) at 200× magnification, and the ratio of EdU/Hoechst 33342-double-positive cells was quantified.
Flow cytometry analysis of the cell cycle and apoptosis
Cells were seeded into 6-well plates and treated with or without 10− 5 M 5-FU for 48 h, after which they were harvested using trypsin without EDTA. For cell apoptosis analysis, washed cells were immediately stained with PE-conjugated annexin V and 7-AAD and analyzed using a BD FACSAria II instrument (BD Biosciences, USA) and FlowJo software. For cell cycle analysis, cells were fixed in 70% ethanol overnight at − 20 degrees and then resuspended in 1 ml of PBS containing propidium iodide and RNase. Cell cycle stage detection was performed on a FACSAria II machine (BD Biosciences, USA) and data analyzed with ModFit LT software.
Flow cytometry of Mito-tracker
Cells were cultured in 6-well plates until 80% confluence was achieved. To analyze mitochondria organization, cells were incubated with 1 mL of the working solution of Mito-Tracker for 1 h at 37 °C and kept on ice protected from light. Texas Red fluorescence was visualized using a BD FACSAria II instrument (BD Biosciences, USA) and flow cytometry data were analyzed using FlowJo software.
Transmission electron microscopy for mitochondrial morphology analysis
Cells were dissociated using trypsin-EDTA solution and collected by centrifugation. They were then fixed in electron microscope fixation liquid for 2 h at room temperature and stored in the dark at 4 °C. Next, fixed samples were sequentially treated with agarose pre-embedding, post-fixation, dehydration, resin penetration, EMBed 821 embedding, polymerization, ultrathin sectioning, and staining. The cuprum grids were observed under an Hitachi HT7800 transmission electron microscope (Hitachi, Japan) and images captured.
2-NBDG uptake assay
1 × 105 cells were plated per well in 6-well plates, incubated for 24 h at 37 °C and then treated with the indicated agents for the appropriate time or transfected with HIF1A siRNA or control siRNA. Cells were incubated with sugar-free RPMI-1640 medium with 100 μM 2-NBDG for 2 h after starving them in glucose in sugar-free RPMI-1640 medium overnight. After 2-NBDG incubation, washed cells were digested and collected in 1 ml of PBS. The mean fluorescence intensity (MFI) of samples was measured by BD FACSAria II (BD Biosciences, USA) flow cytometry using the FITC channel.
Lactate release assay
To measure lactate levels secreted into the culture supernatants, 2 × 104 cells per well were seeded into 24-well plates. The medium was refreshed with 1 mL complete RPMI-1640 medium per well for 24 h after treating with the indicated agents. The next day, culture supernatants were harvested and the lactate concentration was measured by colorimetric assays according to the manufacturer’s protocol of the Lactate Assay Kit. Additionally, protein quantitation was measured by the BCA method and lactate release counts were normalized to total protein concentrations.
1H-nuclear magnetic resonance (1H-NMR) platform for intracellular lactate
Cultured cells were washed with ice-cold PBS and removed from the plate by gentle scraping. 2 × 107 cells per sample were collected in triplicate. After removal of the supernatant, 1 mL ice-cold methanol was added and cells were fragmented by ultrasound in an ice bath. Finally, the supernatant was collected after centrifugation, dried and resolubilized in 450 μL D2O containing 30 μM 3-(tetramethysilane) propionic acid-2,2,3,3-d4 (TMSP) for spectral referencing. All 1H-NMR spectra were recorded at 298 K using a Bruker Avance III 600 MHz spectrometer (Bruker Biospin, Germany), and a standard Bruker noesygppr1d pulse sequence for water suppression was used. The raw 1H-NMR spectra were imported into MestReNova 9.0.1 software (Mestrelab Research, Spain) for Fourier transformation, phase and baseline correction.
Ultra-high pressure liquid chromatography-coupled tandem mass spectrometry (UHPLC-MS/MS) platform for metabolites of central carbon metabolism
Cells (1 × 107 cells per sample) were suspended in 80% methanol, spiked with 20 μL of internal standard solution (2 μg/mL succinic acid-13C4, 4 μg/mL Fructose-1,6-bisphosphate-13C6, 10 μg/mL glucose-13C6), and processed by 5 cycles of ultra-sonication in an ice-water bath. After 30 min at − 20 °C, the cell mixture was centrifuged at 4 °C and 15,000 g for 15 min. The cell-free supernatant was evaporated to dryness and reconstituted in 40 μL of 50% acetonitrile prior to UHPLC-MS/MS analysis. The quality control (QC) sample was obtained by isometrically pooling all the prepared samples. All glycometabolism standards were prepared separately and mixed to a 25 μg/mL standard solution, including glucose, citric acid, cis-aconitic acid, isocitric acid, α-ketoglutaric acid, succinic acid, fumaric acid, malic acid, lactic acid, pyruvic acid, Glucose-6-phosphate (G6P), fructose-6-phosphate (F6P), fructose-1,6-bisphosphate(FBP), ribulose-5-phosphate (Ru5P), xylulose-5-phosphate (Xu5P), sedoheptulose-7-phosphate (S7P), 3-phosphoglyceric acid (3-PGA), glyceraldehyde-3-phosphate (GAP), phosphoenolpyruvic acid (PEP), ribose-5-phosphate (R5P), erythrose-4-phosphate(E4P), and dihydroxyacetone phosphate (DHAP).
The UHPLC-MS/MS analysis was performed on an Agilent 1290 Infinity II UHPLC system coupled to a 6470A Triple Quadrupole mass spectrometer (Santa Clara, USA). Samples were injected into a Waters BEH Amide column (100 mm × 2.1 mm, 1.7 μm) at a flow rate of 0.25 mL/min. The mobile phase consisted of (A) water in 15 mM ammonium acetate at pH 8.5 and (B) 90% acetonitrile. The chromatographic separation was conducted by a gradient elution program as follows: 0-2 min, 90% B; 14 min, 75% B; 15 min, 65% B; 15.2-16.9 min, 50% B; 17-20 min, 90% B. The eluted analytes were ionized by an electro-spray ionization source in positive mode (ESI−). The temperatures of the ESI− source drying gas and sheath gas were 300 °C and 350 °C. The flow rates of ESI− source drying gas and sheath gas were 5 and 11 L/minute, respectively. The pressure of the nebulizer was 40 psi, and capillary voltage was 4000 V. Dynamic multiple reaction monitoring (dMRM) was used to acquire data in optimized MRM transition. MassHunter software (version B.08.00, Agilent) was used to control instruments and acquire data.
The raw data were processed on the MassHunter Workstation (version B.08.00, Agilent) using the default parameters with manual inspection asssitance to ensure the qualitative and quantitative accuracies of each compound. The concentrations (C, μg/mL) of metabolites in prepared samples (for determination) were quantified automatically, and finally the output for quantitative calculation of each tube sample in Excel was established with the following formula: Content (μg/sample) = C × V, where C is the concentration quantified in the prepared sample (μg/mL), V is the volume of reconstituted solvent (μL).
Flow sorting on the basis of GLUT1 or MCT4 expression
5 × 106 cells were harvested and washed in pre-cooled PBS. Cells were incubated with anti-GLUT1 (1:50, Abcam) or anti-MCT4 (1:200, Abcam) primary antibodies for 1 h on ice. After washing 3x with cold PBS, cells were incubated with Alexa-conjugated-488 secondary antibody (1:500) for 30 min on ice. At the end of the incubation, cells were sorted on a BD FACSAria III instrument (BD Biosciences, USA) for high and low expression of GLUT1 or MCT4. Sorted cells were then cultured and CCK8 cytotoxicity assays performed.
Measurement of oxygen consumption rates (OCR) and extracellular acidification rates (ECAR)
1 × 104 cells were seeded into XFe96 cell culture plates in complete RPMI-1640 medium and incubated at 37 °C overnight in 5% CO2. To equilibrate temperature and pH of the detection system, cells were washed with assay RPMI-1640 medium and incubated at 37 °C for 1 h in a CO2-free incubator before assessment. OCR and ECAR were detected with an Agilent Seahorse XFe96 extracellular flux analyzer (Agilent Technologies, USA). To examine mitochondrial respiratory activity, cells were treated with oligomycin (1.5 μM), FCCP (1 μM) and rotenone/antimycin A (0.5 μM) using a Seahorse XF Cell Mito Stress Test Kit. For assessment of glycolytic activity, cells were treated with glucose (100 mM), oligomycin (10 μM) and 2-deoxy-D-glucose (2-DG, 500 mM) using a Seahorse XF Glycolytic Rate Assay Kit in sequence. All experiments were done in seven replicas each time and data expressed as means with SEM.
Fluorometric measurement of ATP production
4 × 105 cells per well were seeded into 6-well plates until 70-90% confluence was achieved. Cells were lysed and cell lysates used for detecting ATP and total protein concentrations. ATP concentrations were measured using a luminometer and determined based on the standard curve. The results were normalized to protein content (BCA assay).
Flow cytometry for intracellular ROS levels
When cells in T25 flasks had reached 60% confluence, they were probed using 10 μM 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) incubated at 37 °C for 20 min. Fluorescence signals in the FITC channel were acquired on a BD FACSAria II (BD Biosciences, USA) instrument for measurement of intracellular ROS. Data were analyzed using FlowJo software.
Colorimetric measurement of enzyme activities
Lactate dehydrogenase (LDH), catalase (CAT), glutathione peroxidase (GPx), and superoxide dismutase (SOD) activities were measured by colorimetry, following the manufacturer’s protocols. Enzyme activities were normalized to total protein levels (BCA assay).
Protein extraction and Western blotting
To extract total proteins, cells were lysed in RIPA buffer containing a protease inhibitor and a phosphatase inhibitor. In addition, nuclear, cytosolic, and membrane proteins were extracted using a Nuclear Protein Extraction Kit or a Membrane Protein Extraction Kit according to the manufacturer’s instructions. Protein solutions were boiled in 5 × loading buffer at 98 °C for 10 min and then resolved by SDS-PAGE. All the primary antibodies used in the experiment were: anti-CDK2 (1:1000, Proteintech), anti-Cyclin D1 (1:1000, Proteintech), anti-p21 (1:2000, Proteintech), anti-Bcl-XL (1:1000, Proteintech), anti-Cleaved caspase 3 (1:500, Affinity), anti-Cleaved caspase 9 (1:1000, Affinity), anti-NDUFB8 (1:2000, Abcam), anti-SDHB (1:100000, Abcam), anti-UQCRC1 (1:1000, Proteintech), anti-COX4 (1:2000, Abcam), anti-ATP5F1 (1:500, Proteintech), anti-GLUT1 (1:5000, Abcam), anti-GLUT2 (1:5000, Abcam), anti-GLUT3 (1:1000, Abcam), anti-GLUT4 (1:500, Abcam), anti-MCT1 (1:1000, Abcam), anti-MCT2 (1:500, Proteintech), anti-MCT4 (1:200, Santa), anti-LDHA (1:1000, Proteintech), anti-HK2 (1:1000, Abcam), anti-GAPDH (1:10000, Proteintech), anti-ENO1 (1:1000, Abcam), anti-PKM2 (1:500, CST), anti-G6PD (1:1000, Abcam), anti-HIF-1α (1:500, Abcam), anti-Catalase (1:2000, Proteintech), anti-GPx1 (1:2000, Abcam), anti-SOD1 (1:50000, Abcam), anti-SOD2 (1:1000, Abcam), anti-PI3K (1:1000, Affinity), anti-Phospho-PI3K (1:1000, Affinity), anti-AKT (1:500, Abcam), anti-Phospho-AKT (1:1000, Abcam), anti-β-catenin (1:5000, Proteintech), anti-Axin2 (1:1000, Abcam), anti-Dvl1 (1:1000, Proteintech), anti-TCF1 (1:1000, Proteintech), anti-TCF4 (1:1000, Proteintech), anti-β-Actin (1:5000, Proteintech), anti-Lamin B1 (1:2000, Proteintech). Detailed antibody information was listed in additional file. Primary antibodies were detected by goat anti-rabbit/mouse IgG (H + L), horse radish peroxidase (HRP)-conjugated secondary antibodies, and reacted with a chemiluminescent HRP substrate. Chemiluminescence images were captured using an Amersham Imager 680 (GE Healthcare, USA).
Co-immunoprecipitation (co-IP) analysis
Pierce Co-IP Kits (Thermo Scientific) were used for the isolation of natural protein complexes. Our approach is briefly summarized in the following steps. Primary antibodies against β-catenin (0.5 μg/μL, Proteintech) or HIF-1α (0.165 μg/μL, Abcam) were first covalently immobilized to AminoLink Plus coupling resins for 2 h at room temperature in a tube rotator, with IgG serving as the negative control. Cells were solubilized in lysis buffer (Tris 0.025 M, NaCl 0.15 M, EDTA 0.001 M, NP-40 1%, and glycerin 5%), and 500 μL of the cell lysates incubated with pre-prepared anti-β-catenin or anti-HIF-1α primary antibody-conjugated resins overnight at 4 °C in a tube rotator. After incubation, the natural protein complexes were eluted and analyzed by Western blotting with anti-β-catenin (1:5000, Proteintech), anti-HIF-1α (1:500, Abcam), anti-TCF1 (1:1000, Proteintech), and anti-TCF4 (1:1000, Proteintech) primary antibodies as described previously.
Total RNA extraction and real-time quantitative PCR (RT-qPCR)
Total RNA was extracted with TRIzol reagent (TaKaRa) according to the manufacturer’s instructions. Extracted total RNA was measured on a NanoDrop spectrophotometer (NanoDrop Technologies, USA) and reverse transcribed to cDNA using ReverTra Ace qPCR RT Kit (TOYOBO). The cDNA was detected using a mixed system containing 1 μL cDNA, 0.8 μL gene-specific primers, and 5 μL SYBR Green Realtime PCR Master Mix (TOYOBO) using a LightCycler 480 II instrument (Roche, Switzerland). Relative mRNA quantification was performed by the ΔΔCt method and the housekeeping gene ACTB was used as an internal reference. Detailed information of primer sequences is listed in the additional files.
Immunocytofluorescence
1 × 105 cells were seeded on cell climbing slices coated with poly-D-lysine in 24 well plates. Prepared cells were fixed in 4% PFA for 20 min, permeabilized with 0.5% Triton X-100 for 5 min, and blocked in 10% goat serum for 1 h at room temperature. The following primary antibodies were incubated overnight at 4 °C: anti-LDHA (1:250, Proteintech), anti-β-catenin (1:200, Proteintech), or anti-HIF-1α (1:200, Abcam; 1:100 Proteintech). Next day, cells were incubated with secondary antibodies for 1.5 h at room temperature: CoraLite594-conjugated goat anti-rabbit IgG(H + L) (1:250, Proteintech), CoraLite488-conjugated goat anti-mouse IgG(H + L) (1:250, Proteintech). Cell climbing slices were mounted on slides with mounting medium containing DAPI and photographed by a TCS SP8 confocal laser scanning microscopy (Leica, Italy) or an Axio Scope A1 microscope (Zeiss, Germany) at 400× magnification.
Subcutaneous cell-derived xenograft (CDX) nude mouse model
Male athymic BALB/c nude mice were raised under 12 h light/12 h dark cycles in an SPF environment. We subcutaneously implanted 5 × 106 WT or 5-FU-R cells with or without stable knockdown of shHIF1A into the right forelimb underarm of 4 week-old nude mice. One week after subcutaneous injection, the mice were intraperitoneally injected with 5-FU (25 mg/kg, three times a week), or saline as a control. In the experiments on HIF-1α pharmacological inhibition, one week after subcutaneous injection, both WT or 5-FU-R cell-bearing nude mice were then randomly divided into 4 groups: control, 5-FU, IDF-11774, and 5-FU + IDF-11774. 5-FU was intraperitoneally injected at the dose of 25 mg/kg three times a week, IDF-11774 was intraperitoneally injected at the dose of 30 mg/kg twice a week, and saline as a control. Subcutaneous tumor-bearing mice were continuously monitored, and tumor volumes (V = L × W2/2, where L is the length and W is the width) were assessed. Mice were sacrificed under anesthesia when the tumor reached a diameter of about 1.5 cm.
Patient derived xenograft (PDX) nude mouse model
Tumor tissues derived from a rectal cancer patient (male, Dukes’ C) received neoadjuvant chemotherapy containing 5-FU and subsequent surgery. Fresh well-trimmed tumor tissues were cut into pieces about 3 mm in size and immediately placed in RPMI-1640 medium supplemented with 50% FBS and penicillin/streptomycin. We subcutaneously implanted tumor pieces into the right forelimb underarm of 4 week-old NOD/scid mice, and xenografts were harvested when they reached a diameter of about 1.5 cm. They were then cut up and implanted again into fresh NOD/scid mice. When these xenografts had been growing for about 20 days, mice were divided into four groups: control (saline), 5-FU (25 mg/kg, three times per week), IDF-11774 (30 mg/kg, twice per week), and 5-FU with IDF-11774. Mice were sacrificed under anesthesia when the PDXs reached a diameter of about 1.5 cm.
Immunohistochemistry (IHC)
CDX/PDX tumors and CRC tissues of patients were fixed in 4% PFA, embedded in paraffin, and sectioned. Tumor sections were dewaxed with dewaxing agent and dehydrated in graded alcohol concentrations. After antigen retrieval with heated citrate buffer and blocking with 10% goat serum, the following primary antibodies were incubated overnight at 4 °C: anti-Ki-67 (Proteintech, 1:5000), anti-HIF-1α (1:200, Abcam), anti-GLUT1 (1:500, Abcam), anti-HK2 (1:100, Abcam), anti-PKM2 (1:200, CST), anti-LDHA (1:200, Proteintech), anti-MCT4 (1:50, Santa), anti-Phospho-AKT (1:100, Abcam), anti-β-catenin (1:1000, Proteintech), and anti-TCF1 (1:200, Proteintech). Next day, the sections were incubated with HRP-conjugated anti-rabbit/mouse IgG and detected by 3,3′-diaminobenzidine (DAB) staining. The nuclei were stained with hematoxylin. The stained sections were photographed using an Axio Scope A1 microscope (Zeiss, Germany) at 200× magnification.
In vivo glucose uptake assay
WT or 5-FU-R cell-bearing nude mice were injected with 10 nmol IRDye 800CW 2-DG Optical Probe (Li-Cor Biosciences) via the tail vein, and then imaged using an IVIS Kinetic (Caliper Life Sciences, USA) small animal imaging system with a cooled Hamamatsu ORCA-R2 camera (Hamamatsu, Japan) 24 h later. Probe signals were displayed as pseudo-colored bioluminescent images and merged with grey-scale white light images of the mice. Circular ROIs were drawn over the areas and quantified, and the results are reported as total radiant efficiency.
Lactate-magnetic resonance spectroscopy (MRS)
Tumor lactate concentrations of WT or 5-FU-R cell-bearing nude mice were quantified with MRS. The data were acquired on a Siemens Skyra 3.0 T MRI scanner (Siemens Skyra, Germany) using an 8-channel small animal special coil (mouse coil) for reception of the signal. T2-weighted anatomical images were acquired using a standard 3D sequence with an isotropic voxel size of 1 mm. Subsequently, the lactate spectrum was acquired from a single voxel snug-fit to the tumor with TR/TE of 1700 ms/135 ms. The acquisition duration was set to 853 ms with a bandwidth of 1200 Hz.
Enrichment and detection of circulating tumor cells (CTCs)
Tumor-reflux venous blood was collected from patients with CRC during surgery, and peripheral venous blood was collected at the same time. Three mL normal saline (0.9%) and 200 μL PFA (8%) were added to 5 mL EDTA-anticoagulated whole blood samples, which were then transferred to a filter with an 8 μm diameter aperture membrane. CTCs were captured on the membrane by a CTCBIOPSY device (YZYBIO Company, China). Given that captured cells on the membrane included CTCs but also normal blood cells, candidate CTCs were distinguished by Wright’s staining. After destaining, detection of HIF-1α expression was performed by immunofluorescence analysis.
Statistical analysis
All data are shown as means ± standard errors of the mean (SEM). A two-tailed Student’s t-test was used to compare variables of two groups, and one-way or two-way ANOVA were performed for multi-group comparisons. Correlation analysis used Pearson or Spearman correlation analysis. Patient survival data related to HIF-1α expression was evaluated by the Kaplan-Meier method with survival analysis using Log-rank (Mantel-Cox) testing. Significance of differences is marked ns = not significant, * p < 0.05, ** p < 0.01, and *** p < 0.001, and all p values < 0.05 were considered statistically significant. Statistical details are included in the respective figure legends.
Discussion
For more than 60 years, 5-FU has been the first-line drug for both single-drug and multi-drug chemotherapy and the most effective systemic agent for managing advanced and metastatic CRC [
3,
4]. However, a considerable fraction of patients still acquire 5-FU resistance [
5]. Therefore, identifying strategies on how to improve the efficacy of 5-FU and reverse 5-FU resistance are important challenges in clinical practice. Several mechanisms of 5-FU resistance have been reported, including changes in the rate of drug influx or efflux [
52], intra-tumor heterogeneity [
52], epigenetic factors [
53], tumor microenvironment [
54], and 5-FU metabolic enzymes [
55]. Here, we present a mechanism of 5-FU resistance in CRC by which cancer cells remodel glucose metabolism, which renders 5-FU ineffective. We present strong evidence for the use of inhibitors or gene knock-down targeting HIF-1α, a “master regulator” of glucose metabolism [
14], in diminishing 5-FU resistance in CRC.
Metabolic reprogramming is a hallmark in cancer cells [
6,
7]. We found that the intracellular glucose metabolic pools changed significantly in 5-FU-resistant CRC, with abnormal glucose and lactate transport and utilization. Our results also showed that metabolic reprogramming events in the process of 5-FU resistance are an exacerbation from OXPHOS to glycolysis and PPP.
One of the most important characteristics of metabolic reprogramming in tumors is increased dependency on glycolysis for energy generation [
8]. Although the energy conversion efficiency of glycolysis is not as high as OXPHOS, increased glucose uptake and a higher glycolytic flux can compensate [
8]. In addition to providing energy, high glycolytic flux also provides a variety of raw materials for biosynthesis [
10]. Increased aerobic glycolysis contributes to tumor progression by giving cancer cells growth advantage and drug resistance phenotypes [
11]. Drug resistance of CRC cells to vincristine and oxaliplatin is overcome by knocking out polypyrimidine tract binding protein 1 (PTBP1) [
56], a regulator of glycolysis. HK2 catalyzes the first rate-limiting step in glucose metabolism, and 2-DG, an inhibitor of HK2, reverses drug resistance in several in vitro models [
57]. Silencing PKM2 increased docetaxel accumulation and promoted anti-tumor activity in lung cancer cells [
58]. Targeting MCT1 greatly enhanced the sensitivity of human osteosarcoma cells to chemotherapy [
59]. These data implied that inhibition of glycolysis by key enzymes targeting the glycolytic pathway may be a potential broad-spectrum therapeutic approach for reversing drug resistance. Our results showed that knock down or pharmacological inhibition to down-regulate HIF-1α expression inhibited glycolysis and ultimately reversed 5-FU resistance, and this may be a much more effective approach than other agents because HIF-1α regulates the entire glycolytic pathway as a critical up-stream regulator.
Activation of PPP is implicated in the development of various tumors, and PPP is strongly activated in chemotherapy-resistant cells and tumor tissues [
11]. We also demonstrated a higher PPP flux both oxidative and non-oxidative pathways in 5-FU-resistant CRC cells. Flux through the oxidative PPP has a stronger ability to generate NADPH [
41]. Given that NADPH has been mainly associated with ROS scavenging, high levels of ROS in 5-FU-R CRC cells may drive the activation of oxidative PPP. In additional, PPP is a necessary pathway for the biosynthesis to meet metabolic demands of glycolysis-dependent cells [
41]. The higher metabolic flux into the non-oxidative PPP increases nucleotide biosynthesis and causes an increase intracellular nucleotide pools. As 5-FU is a pyrimidine analog drug, it will likely be diluted by increased intracellular nucleotide concentrations. Future research should explore a method to reverse resistance in CRC to 5-FU and other fluoropyrimidine by inhibiting non-oxidative PPP related nucleotide biosynthesis.
Previous studies showed that inhibition of energy substrates relieves drug resistance [
60]. Substantial glucose influx and high expression of GLUT1 ensure the supply of metabolic substrates for a high rate of glycolysis and PPP [
7,
8]. This is consistent with our observations, in which activated glycolysis was accompanied by increased glucose uptake and up-regulated GLUT1 expression in 5-FU-R cells in vitro and in vivo. Furthermore, increased level of GLUT1 was related to 5-FU resistance in 5-FU-R cells. Thus, to limit the use of glucose as an energy source might have advantages as therapeutic strategies in 5-FU resistance in CRC. We also observed significantly increased lactate efflux and up-regulation of MCT1/MCT4 in 5-FU-R cells, and high expression of MCT4 was related to 5-FU resistance. Hence, targeting MCTs not only limited glycolytic flux but also reversed 5-FU resistance, which profoundly improved the cell-killing effect of 5-FU.
Although a high glycolysis rate confers 5-FU-R cells with growth and resistance advantages, it also renders cancer cells susceptible to glucose deprivation. The glucose metabolic reprogramming in vitro and in nude mice reflects the glucose consumption capacity of 5-FU-R cells but this capacity does not necessarily reflect a real condition of glucose usage in vivo. In low glucose conditions, we found that glucose deprivation was severely detrimental to 5-FU-R cells, while lactate rescued cancer cells from extremely low glucose condition. Thus, disrupting lactate balance and use in 5-FU-resistant cells or tumors could expose their vulnerability to glucose deprivation and ultimately reverse 5-FU resistance. Co-blockade of GLUT1 and MCTs for tumor cell killing and resistance reversal should be pursued in future studies.
Increased stability of HIF-1α and enhanced transcription of its downstream genes have been widely reported in a variety of tumors during malignant progression [
61]. Consistently, HIF-1α which was considered to be a major orchestrator of cellular adaptation to oxygen environment, whereas the precise oxygen-dependent prolyl hydroxylases (PHDs)-mediated regulation [
62]. In addition, some non-classical regulation of HIF-1α, independent of external oxygen concentrations, has been reported [
63]. The oxygen environment is complicated and fluctuant in tumor tissue in vivo. We observed high expressions of HIF-1α in 5-FU-R cells in both normoxia and hypoxia conditions. In this study, we identified novel mechanisms of HIF-1α regulation in 5-FU-resistant CRC that are independent of the classical PHD-mediated oxygen-dependent pathway. The activated PI3K/Akt pathway plays an important role in human cancer [
42], and activated PI3K/Akt signaling up-regulates
HIF1A transcription and translation [
43,
64]. Moreover, PI3K/Akt signal stabilizes and trans-activates HIF-1α regardless of oxygen levels [
43]. We observed that ROS accumulation leads to HIF-1α up-regulation via activated PI3K/Akt signaling pathway in 5-FU-R cells. HIF-1α mRNA and protein were regulated by PI3K/Akt signaling in 5-FU-resistant CRC, in response not to oxygen conditions but to the overload of ROS.
The Wnt/β-catenin signaling pathway plays an important role in CRC initiation and progression [
47]. The correlation of HIF-1α and β-catenin is commonly detected in cancer and indicates malignant phenotypes [
48]. We demonstrated that β-catenin is involved in the stabilization and translocation of HIF-1α from the cytoplasm to the nucleus in 5-FU resistance. The β-catenin transcriptional co-factor TCF1, but not TCF4, interacts with the HIF-1α/β-catenin complex. In-depth research on the mechanisms of the HIF-1α/β-catenin complex and co-factor TCFs can help us to identify more potential targets for molecular inhibitors. We found that high levels of HIF-1α in 5-FU resistance in CRC were attributed to increased stability, prolonged half-life, cellular translocation, and up-regulated transcription. ROS/PI3K/Akt signaling and Wnt/β-catenin signaling are responsible at least in part for the comprehensive changes of HIF-1α and the metabolic reprogramming in our 5-FU resistance models.
Data from CRC patients with preoperative fluorouracil analog-based chemotherapy revealed that increased HIF-1α expression in primary tumors or CTCs may suggest chemotherapy resistance, and high HIF-1α expression was associated with a significantly decreased disease-free survival in patients who had been previously treated with fluorouracil analog. Detection of CTCs in liquid biopsies is a promising strategy for diagnosing cancer, monitoring relapse and metastasis, and evaluating cancer prognosis and therapy [
36,
37]. Studies showed that the presence of CTCs in CRC patients is a strong predictor of poor prognosis [
37]. Our preliminary studies also suggested that CTCs have great potential to serve as a prognostic biomarker of CRC for evaluating the outcome of 5-FU-based chemotherapy, especially CTCs in the reflux veins of tumor, which carry more accurate information about the primary tumor. Analyses with larger samples in multi-center, prospective randomized and controlled trials are needed to draw more definitive conclusions.