Glucose modulation induces reactive oxygen species and increases P-glycoprotein-mediated multidrug resistance to chemotherapeutics

Abstract

Background and Purpose

Cancer cells develop resistance to stress induced by chemotherapy. In tumours, a considerable glucose gradient exists, resulting in stress. Notably, hypoxia-inducible factor-1 (HIF-1) is a redox-sensitive transcription factor that regulates P-glycoprotein (Pgp), a crucial drug-efflux transporter involved in multidrug resistance (MDR). Here, we investigated how glucose levels regulate Pgp-mediated drug transport and resistance.

Experimental Approach

Human tumour cells (KB31, KBV1, A549 and DMS-53) were incubated under glucose starvation to hyperglycaemic conditions. Flow cytometry assessed reactive oxygen species (ROS) generation and Pgp activity. HIF-1α, NF-κB and Pgp expression were assessed by reverse transcriptase-PCR and Western blotting. Fluorescence microscopy examined p65 distribution and a luciferase-reporter assay assessed HIF-1 promoter-binding activity. The effect of glucose-induced stress on Pgp-mediated drug resistance was examined after incubating cells with the chemotherapeutic and Pgp substrate, doxorubicin (DOX), and performing MTT assays validated by viable cell counts.

Key Results

Changes in glucose levels markedly enhanced cellular ROS and conferred Pgp-mediated drug resistance. Low and high glucose levels increased (i) ROS generation via NADPH oxidase 4 and mitochondrial membrane destabilization; (ii) HIF-1 activity; (iii) nuclear translocation of the NF-κB p65 subunit; and (iv) HIF-1α mRNA and protein levels. Increased HIF-1α could also be due to decreased prolyl hydroxylase protein under these conditions. The HIF-1α target, Pgp, was up-regulated at low and high glucose levels, which led to lower cellular accumulation of Pgp substrate, rhodamine123, and greater resistance to DOX.

Conclusions and Implications

As tumour cells become glucose-deprived or exposed to high glucose levels, this increases stress, leading to a more aggressive MDR phenotype via up-regulation of Pgp.

Tables of Links

TARGETS
Transportersa	Enzymesb
ABCG2	HDAC1
GLUT1	Other protein targets
MRP1 (ABCC1)	VEGFA
P-glycoprotein (ABCB1; MDR1)

LIGANDS
ATP	H2O2
D-glucose	Pyruvate
Doxorubicin (DOX)	Rhodamine123
Glutamine	Vinblastine
Glutathione

These Tables list key protein targets and ligands in this article which are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Pawson et al., 2014) and are permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 (^a,bAlexander et al., 2013a,b).

Introduction

The intracellular glucose concentration markedly depends on glucose uptake, cellular metabolism and the concentration of extracellular glucose (Prochazkova et al., 2011). Malignant cells display enhanced glycolysis as a result of their accelerated metabolism, high glucose requirements and increased glucose uptake (Chen and Russo, 2012). As a consequence of the increased reliance on aerobic glycolysis, as well as vascularization and elevated proliferation, there exists a considerable glucose-gradient within tumours (Chaplain, 1996; Pelicano et al., 2006; Annibaldi and Widmann, 2010). This microenvironment exposes a significant portion of cancer cells to extremely low or high glucose concentrations (Casciari et al., 1988; Walenta et al., 2002; Yeom et al., 2012). Moreover, studies have associated increased reactive oxygen species (ROS) with conditions of excess and limited glucose (Yu et al., 2011; Graham et al., 2012).

Previous studies have demonstrated that intracellular ROS are primarily formed via NADPH oxidases (NOX) or as a by-product of the electron-transport chain (Murphy, 2009; Block and Gorin, 2012). While ROS can mediate cytotoxicity, there is also evidence to support their role in signal transduction (Behrend et al., 2003). Furthermore, ROS can modulate transcription factor activity, including hypoxia-inducible factor-1 (HIF-1) and NF-κB (Denko, 2008; Hagen, 2012). These transcription factors have been linked to increased expression of the ATP-dependent, drug-efflux ‘pump’, P-glycoprotein (Pgp) (Milane et al., 2011). This trans-membrane protein is the most consistently overexpressed ATP-binding cassette (ABC) transporter involved in multidrug resistance (MDR), which represents a major concern to successful chemotherapeutic treatment (Longley and Johnston, 2005; Milane et al., 2011).

This study aimed to establish the relationship between stress induced by the tumour cell microenvironment and modulation of the drug-resistant phenotype, using a cell culture model. Herein, we report that variation in glucose levels confers Pgp-mediated drug resistance to the chemotherapeutic and Pgp substrate, doxorubicin (DOX). Low or high glucose levels, relative to standard culture conditions, induced oxidative stress, nuclear translocation of the NF-κB p65 subunit and increased HIF-1α-expression. This increase in HIF-1α was implicated in the increased Pgp expression, which leads to greater DOX resistance. These studies highlight how glucose-induced stress increases MDR in a tumour cell culture model.

Methods

Cell culture

The human DMS-53 small cell lung carcinoma cell line and the A549 non-small cell lung carcinoma cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA). The human KB31 and KBV1 cervical carcinoma cell lines were from Dr M. Kavallaris (Children's Cancer Institute Australia, NSW). Medium for growing KBV1 cells was supplemented with VBL (0.8 μg·mL⁻¹) for maintenance of a partial MDR phenotype, allowing Pgp induction upon stimulation. All media were supplemented with penicillin (100 U·mL⁻¹), streptomycin (100 mg·mL⁻¹), glutamine (2 mM), non-essential amino acids (100 mM) and pyruvate (100 mM; all from Life Technologies). Cells were maintained in DMEM ( [d-glucose] = 25 mM; Life Technologies). At the time of treatment, media was replaced with glucose-free DMEM (-d-glucose, 0 mM) or this medium supplemented with d-glucose (final [glucose] = 12.5 or 50 mM).

For all experiments, to study glucose-induced stress, cells were exposed to either: (i) glucose-deprivation (0 mM); (ii) low glucose (12.5 mM); (iii) the glucose concentration used in DMEM (from the manufacturer) to routinely culture cells (25 mM; herein referred to as ‘normal glucose’); or (iv) high glucose (50 mM) for 30 min/37°C. Preliminary studies demonstrated that because the cells were routinely cultured under standard growth medium using glucose at 25 mM, they had become accustomed to these conditions. Indeed, the higher and lower concentrations of glucose implemented were essential in order to observe the glucose-induced stress response. Hence, this culture model simulates the hypo- and hyperglycaemic conditions found in vivo. Studies were also performed in an attempt to adjust the cells to lower glucose concentrations (<20 mM) by prolonged incubations at these levels. However, these experiments were unsuccessful and led to a significant decrease in cellular viability (not shown).

Protein extraction and Western blotting

Whole cell, membrane protein extractions and Western blotting were performed using standard procedures (Zhong et al., 2008; Saletta et al., 2010). Nuclear/cytosolic extractions were achieved via the NE-PER nuclear cytoplasmic kit (ThermoFisher, VIC, Australia). GAPDH and histone deacetylase-1 (HDAC1) were used as controls for cytoplasmic and nuclear fractions respectively (Kovacevic et al., 2013). H₂O₂ was used as it increases nuclear p65 localization (Takada et al., 2003).

The role of ROS in this pathway was assessed by the utilization of the ROS scavenger and GSH-precursor, NAC (5 mM; De Flora et al., 1995)], or the NOX inhibitor, apocynin (50 μM; Petronio et al., 2013). NAC or apocynin were added before (30 min/37°C) and during the glucose treatment. H₂O₂ and the mitochondrial electron-transport chain inhibitor, AM (10 μM), were used as positive controls for the effects of ROS.

Membranes were probed using mouse anti-Pgp (Cat.#:P7965, 1:5000, Sigma-Aldrich), rabbit anti-GLUT1 (Cat.#:SAB4502803, 1:1000, Sigma-Aldrich); mouse anti-HIF-1α (Cat.#:610959, 1:400, BD Biosciences, San Jose, CA, USA); rabbit anti-prolyl hydroxylase 2 (PHD2; Cat.#:ab4561, 1:1000, Abcam, Cambridge, UK), rabbit anti-GAPDH (Cat.#:ab9485; 1:1000, Abcam); mouse anti-HDAC (Cat.#:5356, 1:1000, Cell Signaling, Danvers, MA, USA); or rabbit anti-p65 XP® (Cat.#:8242, 1:1000, Cell Signaling). Incubations were performed overnight/4°C followed by appropriate secondary antibody [HRP-conjugated goat anti-mouse (Cat.#:A4416, 1:10,000, Sigma-Aldrich) and goat anti-rabbit (Cat.#:A6154, 1:10,000; from Sigma-Aldrich) for 1 h/room temperature. Membranes were developed with enhanced chemiluminescence reagent (Amersham Pharmacia Biotech, Amersham, UK) and visualized using a ChemDoc system (BioRAD, Hercules, CA, USA). Densitometry was performed using ImageLab Software (BioRAD). β-actin (Cat.#:A5441; 1:500, Sigma-Aldrich) was used as a protein-loading control.

Cellular ROS assay

ROS were measured using the H₂DCFDA assay and detected using flow cytometry (FACSCanto™, BD Biosciences) at 488 nm excitation/530 nm emission. H₂DCFDA is de-acetylated by intracellular esterases to 2′,7′-dichlorodihydrofluorescein (H₂DCF) (Karlsson et al., 2010). The H₂DCF directly detects hydroxyl radicals and indirectly detects superoxide and H₂O₂ as they become converted into hydroxyl radicals, which then oxidizes non-fluorescent H₂DCF to fluorescent DCF (Karlsson et al., 2010). Hence, the DCF assay estimates total amount of intracellular ROS generated. Cells were incubated for 30 min/37°C with the Pgp inhibitor, Ela (0.2 μM), before the addition of the H₂DCFDA probe (25 μM; Life Technologies) to this medium for a further 30 min/37°C incubation. The medium was then replaced with glucose treatments (0–50 mM), and also Ela (0.2 μM) to prevent Pgp-dependent efflux of DCF (Prochazkova et al., 2011). H₂O₂ (50 μM) was used as a positive control for DCF fluorescence, as shown previously (Keston and Brandt, 1965). Data were collected for 10 000 cells per sample and analysed using FlowJo software, version 7.5.5 (Tree Star, Inc., Ashland, OR, USA).

Transient NOX4-silencing

Cells were incubated for 6 h/37°C with an siRNA-Lipofectamine mixture (50 nM; NOX4-siRNA Cat.#:30141, Santa Cruz, CA, USA, and 1:200 Lipofectamine-2000, Life Technologies). This mixture was then replaced with media for a further 72 h/37°C before glucose treatments. The effectiveness of NOX4 silencing was assessed using Western blotting. As a control, scrambled-siRNA (Scr-siRNA, Life Technologies) was used at the same concentration as NOX4-siRNA.

Mitochondrial ROS determination

Mitochondrial ROS were determined using the MitoSOX Red mitochondrial superoxide indicator (5 μM; Life Technologies). The incubation procedure was the same as that used for the H₂DCFDA probe and included the Pgp inhibitor, Ela (0.2 μM), to prevent probe efflux. Fluorescence intensity was measured using flow cytometry (FACSCanto, BD Biosciences) at 488 nm excitation/595 nm emission. Data were collected from 10 000 cells per sample and analysis performed using FlowJo software.

Mitochondrial membrane potential

Mitochondrial membrane potential was evaluated using the JC-1 probe (5,5′,6,6′–tetrachlor-1,1′,3,3′-tetraetylbenzimidazolylcarbocyanine iodide; Life Technologies). Cells were incubated for 30 min/37°C with glucose treatments in the presence or absence Ela (0.2 μM) to prevent Pgp-dependent probe efflux (Chaoui et al., 2006). The JC-1 probe (10 μM) was then added to this medium for a further 15 min/37°C. The cells were then imaged using a confocal microscope (LSM 510 Meta; Zeiss, Oberkochen, Germany) and Axiovision software (Zeiss). Quantification of the red/green fluorescence was performed using ImageJ (NIH, Bethesda, MD, USA).

Pgp expression

Pgp expression was assessed by incubating cells with a FITC-conjugated Pgp antibody (Cat.#:557002, 1:5, BD Biosciences) for 30 min/37°C and detected using flow cytometry (FACSCanto, BD Biosciences) at 488 nm excitation/530 nm emission. Data were collected from 10 000 cells per sample. Data analysis was performed using FlowJo software.

Rh123 accumulation assay

Pgp functionality was assessed by measuring intracellular accumulation of the fluorescent Pgp substrate, Rh123 (Kawabata et al., 1997). Cells were treated with glucose in the presence or absence NAC (5 mM). NAC was added before (30 min/37°C), and during the 24 h/37°C glucose treatment. After glucose treatments, cells were incubated with Rh123 (10 μM) for 30 min/37°C and analysed by flow cytometry (FACSCanto, BD Biosciences) at 510 nm excitation/595 nm emission. Data were collected from 10 000 cells per sample.

Immunofluorescence

Cells were fixed with paraformaldehyde (4%; 15 min/20°C) and permeabilized with digitonin (200 μM; 10 min/20°C). After blocking with 5% BSA, immunofluorescence was performed using a 16 h/4°C incubation with anti-NF-κB p65 antibody (Cat.#:8242, 1:100, Cell Signaling), followed by an incubation of 16 h/4°C with an Alexa Fluor-conjugated secondary antibody (Cat.#:11012, 1:1000, Life Technologies). Images were captured with a confocal microscope (LSM 510 Meta; Zeiss) using green (495 nm excitation/516 nm emission) and red (577 nm excitation/592 nm emission) fluorescence.

RNA isolation and reverse transcriptase-PCR (RT-PCR)

Total RNA was isolated using TRIzol® (Life Technologies). RT-PCR was performed (Whitnall et al., 2006) using the primers in Table 1. β-actin was used as an RNA-loading control. RT-PCR was shown to be semi-quantitative and in the log-phase of amplification.

Table 1.

List of primer sets, expected product size and PCR programmes

Gene	Accession number	Species		Forward (5′ – 3′)	Product size	Program (temp/cycle)
MDR1	NM_000927.4	Human	F	TCACCAAGCGGCTCCGATACAT	1065	64/24
			R	CCCGGCTGTTGTCTCCATAGGC
GLUT1	NM_006516.2	Human	F	CTTTGTGGCCTTCTTTGAAGT	168	58/24
			R	CCACACAGTTGCTCCACAT
HIF-1α	NM_001530.3	Human	F	CTCGGCGAAGCAAAGAGT	578	56/38
			R	CAAGCACGTCATGGGTGG
β-actin	NM_001101.3	Human	F	CCCGCCGCCAGCTCACCATGG	397	56/24
			R	AAGGTCTCAAACATGATCTGGGTC

HIF-1 promoter luciferase assay

Cells were transfected with a HIF-1 promoter luciferase construct containing the mammalian HIF-1 transcriptional regulatory-element sequence (5′-TACGTGCT-3′) (Wang and Semenza, 1993). Cells were also transfected with a constitutively expressing Renilla luciferase construct and a non-inducible Firefly luciferase construct, which acted as positive and negative controls, respectively, to validate transfection (see ‘Positive’ and ‘Negative’). Luciferase assays were carried out using the Qiagen Luciferase Assay System (SAB Biosciences, VIC, Australia). Cells were transfected (24 h/37°C) before treatment. Protein extraction was performed using Luciferase Cell Culture Lysis Buffer (Promega, Madison, WI, USA). Luminescence was measured using a FLUOstar Omega Luminometer (BMG Labtech, VIC, Australia).

Proliferation assay

Cellular proliferation was assessed after drug treatments, in the presence or absence of a non-toxic concentration of Ela, using phase contrast microscopy and also MTT assays, which were validated by viable cell counts (Richardson et al., 1995).

Data analysis

Results are expressed as mean ± SD (n ≥ 3 experiments). Statistical analysis was performed using Student's t-test, with results being significant when P < 0.05. Concentration–response curves were fitted in Prism 6.0 (Graphpad Software, San Diego, CA, USA) to obtain IC₅₀ values.

Chemicals

DOX was obtained from Pfizer (New York, NY, USA). 3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT), superoxide dismutase–polyethylene glycol (PEG-SOD), apocynin, antimycin A (AM), H₂O₂, vinblastine (VBL), N-acetyl-L-cysteine (NAC) and rhodamine123 (Rh123) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Elacridar (Ela; PSC833) was provided by GlaxoSmithKline (London, UK). MitoSOX™ and 2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA) were from Life Technologies (Carlsbad, CA, USA).

Results

Multidrug-resistant cells are susceptible to glucose-induced ROS

Considering the link between glucose levels and cellular stress (Graham et al., 2012), we assessed whether the well-characterized Pgp-expressing KBV1 cell line (Ludwig et al., 2006) and their very low Pgp-expressing parental counterparts, namely KB31 cells (Figure 1A), were susceptible to glucose-induced stress.

Figure 1.

Changes in glucose levels lead to NOX4-mediated ROS generation, mitochondrial superoxide generation and mitochondrial membrane hyperpolarization (A) Western blot showing Pgp protein expression in KB31 and KBV1 cells. (B) Incubating KB31 and KBV1 cells for 30 min/37°C with low (0 and 12.5 mM) or high (50 mM) glucose compared with normal glucose (25 mM) increases DCF fluorescence. (C) Western blot showing NOX4 protein expression after silencing with NOX4-siRNA compared with Scr-siRNA-treated KB31 and KBV1 cells. (D) Incubation for 30 min/37°C with apocynin (50 μM), a NOX inhibitor, and transient silencing of NOX4, using NOX4-siRNA, decreases DCF fluorescence. (E) Mitochondrial superoxide production measured by MitoSOX Red (5 μM) increases after the glucose treatments in (B). Upon incubation of cells with PEG-SOD (1000 U·mL⁻¹) and the glucose treatments, MitoSOX was markedly decreased compared with the respective glucose treatments alone. (F) Increased red JC-1 fluorescence after a 30 min/37°C incubation with low (0 mM) or high (50 mM) glucose compared with normal glucose (25 mM). JC1 fluorescence was quantified and expressed as a ratio (red /green fluorescence. Data in (A, C) are typical blots from five experiments. Results in (B, D, E) are mean ± SD (n = 5). *P < 0.05, **P < 0.01, ***P < 0.001, versus control (25 mM glucose), ^#P < 0.05, ^###P < 0.001, ^†P < 0.05 versus respective glucose treatment alone in (D) Scr-siRNA cells, or (E) without PEG-SOD in KB31 or KBV1 cells. Data in (B, D, E) were normalized to the control (glucose; 25 mM) for both cell types. In (B, D, E), mean fluorescence intensity is presented as arbitrary units (a.u.). Immunofluorescence photographs in (F) are representative of three experiments and the quantified fluorescence intensity is presented as mean ± SD (n = 3). Scale bar: 50 μm.

Initially, we examined glucose-induced stress by assessing cellular ROS production, as measured by 2′,7′-dichlorofluorescein (DCF) fluorescence (Jansson et al., 2010). For both cell types, an incubation for 30 min/37°C in media devoid of glucose (0 mM) resulted in significantly (P < 0.001–0.01) increased ROS levels as measured by DCF fluorescence relative to the normal glucose concentration (25 mM) (Figure 1B). At low glucose (12.5 mM), a slight, but significant (P < 0.05) increase in DCF fluorescence was observed with KB31 cells, but not KBV1 cells, relative to normal glucose (25 mM; Figure 1B). The high (50 mM) glucose concentration significantly (P < 0.01) elevated ROS generation compared with normal glucose (25 mM) in both cell lines (Figure 1B). The positive control, H₂O₂ (50 μM), significantly (P < 0.001) increased DCF fluorescence under normal glucose conditions in both cell types. Collectively, Pgp-expressing KBV1 and non-Pgp-expressing KB31 cells showed increased ROS generation in response to altered glucose levels. Studies then assessed the intracellular source of ROS production.

Mitochondrial NOX4 contributes to glucose-induced ROS production

The most abundant NOX4 is a major enzymatic generator of cellular H₂O₂ (Takac et al., 2011), and was evaluated as a source of the redox-stress in Figure 1B in KB31 and KBV1 cells. To specifically assess the role of NOX4, the NOX4-isoform was silenced using siRNA and resulted in a significant (P < 0.001) 88 ± 4% decrease in NOX4 protein expression in both cell types relative to the respective Scr control (Figure 1C). Studies then assessed the effect of modulating glucose on ROS generation using these cells treated with Scr- and NOX4-siRNA, and also with the Scr-siRNA-treated cells incubated with the NOX inhibitor, apocynin (50 μM; Li et al., 2010). In these experiments, cellular ROS generation was measured using the DCF assay (Figure 1D).

Upon modulation of glucose, KB31 (−Pgp) and KBV1 (+Pgp) cells incubated with Scr-siRNA responded in a similar way to their non-transfected counterparts (Figure 1B), with increased ROS being detected at low and high glucose levels relative to normal glucose (Figure 1D). The addition of the NOX inhibitor, apocynin, to Scr-siRNA-treated cells, resulted in a significant (P < 0.05) decrease in ROS generation following incubation with low glucose (0 mM in KB31 and KBV1 cells) and high glucose (50 mM in KB31 cells) relative to the respective Scr-siRNA glucose treatment (Figure 1D). Similarly, NOX4-siRNA significantly (P < 0.05) reduced ROS production at low (0 mM) and high (50 mM) glucose levels in both cell types, compared with the respective Scr-siRNA glucose treatment (Figure 1D). These results using apocynin or NOX4-siRNA indicated that NOX4 contributed to cellular ROS generation under glucose-induced stress conditions.

Modulation of media glucose concentration elevates mitochondrial electron-transport chain ROS production

Considering that inactivation of NOX activity or expression using apocynin or NOX4-siRNA, respectively, did not completely abolish ROS generation upon glucose modulation (Figure 1D), it was hypothesized that an additional, more limited source of redox-stress could also be involved. In fact, mitochondria produce a significant amount of ROS via the electron-transport chain (Nickel et al., 2014), and this could be responsible. To assess this, the mitochondrial superoxide indicator, MitoSOX Red (Mukhopadhyay et al., 2007), was used to measure mitochondrial superoxide, rather than H₂O₂ generated by NOX4 (Takac et al., 2011). Notably, the low glucose (0 and 12.5 mM in KB31 and KBV1 cells) and high glucose levels (50 mM in KBV1 cells) significantly (P < 0.05) increased MitoSOX fluorescence compared with normal glucose (25 mM; Figure 1E). In these studies, the established electron-transport chain complex III inhibitor, AM (10 μM), was used as a positive control for induction of mitochondrial superoxide (Mattiazzi et al., 2004), and significantly (P < 0.01) increased MitoSOX fluorescence compared with normal glucose (25 mM) in both cell types (Figure 1E).

As an additional control, the ROS scavenger, PEG-SOD (1000 U·mL⁻¹), a membrane-permeable antioxidant (Webb et al., 1998), was used to inhibit ROS generation (Figure 1E). Incubation of KB31 (−Pgp) and KBV1 (+Pgp) cells with PEG-SOD for 30 min/37°C prior to, and during the 30 min/37°C glucose treatment, markedly and significantly (P < 0.001) reduced MitoSOX fluorescence for all glucose levels compared with their respective glucose treatment without PEG-SOD (Figure 1E). Together, these results in Figure 1D and E indicate that alterations in glucose levels increase ROS predominantly via NOX4-mediated H₂O₂ generation, but also through mitochondrial superoxide production.

Alteration in media glucose levels hyperpolarizes mitochondrial membranes

Hyperpolarization and depolarization of the mitochondrial membrane leads to ROS production that is triggered by an electron ‘leak’ from the mitochondrial respiratory chain (Wei and Dirksen, 2012), and may explain the results in Figure 1E. Using the well-characterized JC-1 probe, we examined the mitochondrial membrane potential as an indicator of mitochondrial dysfunction (Perelman et al., 2012). Low mitochondrial inner-membrane potential leads to a green fluorescence with the JC-1 probe, while hyperpolarization increases JC1 red fluorescence (Perelman et al., 2012). Significantly, experiments were performed in the presence or absence of the Pgp inhibitor, Ela (0.2 μM), to prevent Pgp-mediated efflux of JC-1 which is a Pgp substrate (Chaoui et al., 2006).

Incubation of KB31 (−Pgp) cells with low (0 mM) or high (50 mM) glucose for 30 min/37°C resulted in a marked and significant (P < 0.01) increase in the ratio of red to green fluorescence compared with normal (25 mM) glucose-treated cells (Figure 1F). Similarly, significant (P < 0.01) alterations in fluorescence after glucose modulation were also observed for KB31 (−Pgp) cells in the presence of the Pgp inhibitor, Ela. In contrast, very few KBV1 (+Pgp) cells were fluorescent under the glucose treatments in the absence of Ela, with the fluorescence ratio not being calculated (Figure 1F). This occurred probably because the JC-1 probe was effectively effluxed out of these cells via Pgp (Chaoui et al., 2006). However, in the presence of Ela, incubation of KBV1 (+Pgp) cells with low (0 mM) or high (50 mM) glucose resulted in a marked (P < 0.01) increase in red fluorescence compared with normal (25 mM) glucose-treated cells (Figure 1F). These results show that alterations in glucose availability cause hyperpolarization of the mitochondrial inner membrane.

Alteration in media glucose levels regulates Pgp function and expression

Considering cellular ROS generation after glucose modulation (Figure 1B,D and E), and that ROS signalling alters cellular phenotype (Dworakowski et al., 2006), we assessed if glucose modulation could promote MDR. In these studies, we examined intrinsic and glucose-induced Pgp expression (Figure 2A, Bi), in addition to plasma membrane Pgp function, in KB31 (−Pgp) and KBV1 (+Pgp) cells (Figure 2Bii, Ci,ii). Furthermore, to assess if the response was cell type-specific, A549 and DMS-53 cells, which endogenously express Pgp, were also examined (Figure 2A,B and Ciii,iv).

Figure 2.

Changes in glucose levels increase Pgp-expression and function. KB31, KBV1, A549 and DMS-53 cells were examined for: (A) Pgp protein expression assessed by Western blotting following a change from the normal glucose level (25 mM) to low (0 or 12.5 mM) or high (50 mM) glucose after a 24 h/37°C incubation; (B)(i) Cell plasma membrane Pgp measured by flow cytometry under conditions of normal glucose (25 mM); (ii) Pgp function analysis by flow cytometric measurement of the intracellular accumulation of Rh123 in normal glucose (25 mM) over a 30 min/37°C incubation. (C) Decreased intracellular Rh123-accumulation following a 24 h/37°C incubation with low or high glucose. The Pgp inhibitor, Ela (0.2 μM), or antioxidant NAC (5 mM), increased cellular accumulation of Rh123 in all treatment conditions relative to its respective glucose treatment alone in KBV1, A549 and DMS-53 cells. The results in (A) are typical of three experiments, densitometry is mean ± SD (n = 3). *P < 0.05, relative to normal glucose (25 mM). For (B), mean fluorescence intensity is presented as arbitrary units (a.u.). Results in (C) are mean ± SD (n = 5). *P < 0.05, **P < 0.01, ***P < 0.001, versus normal glucose (25 mM), ^#P < 0.05, ^###P < 0.001, ^†P < 0.05, ^††P < 0.01, ^†††P < 0.001 versus respective glucose treatment alone.

All cell types were subjected to a shift in glucose levels during a 24 h/37°C incubation, after which protein levels were assessed by Western blotting utilizing an isolated membrane fraction. No Pgp was detected in KB31 cells under all conditions (Figure 2A). This finding was irrespective of the very high protein-loading and exposure time of KB31 cells (i.e., 500 μg and 300 s exposure time; Figure 2A). In contrast, modulation of glucose concentrations using KBV1, DMS-53 or A549 cells, resulted in significantly (P < 0.05) increased Pgp expression (Figure 2A). Notably, as evident from the lower protein loading and shorter exposure time (i.e., 30 μg and 40 s; Figure 2A), Pgp expression in KBV1 cells was far greater than in A549 or DMS-53 cells (i.e., protein load: 150 μg and exposure time: 120 s; Figure 2A).

To measure membrane Pgp levels, flow cytometry utilizing a FITC-labelled Pgp antibody was implemented (Figure 2Bi). This more sensitive method detected Pgp in all cell lines under normal glucose levels (25 mM). The Pgp levels were greatest in KBV1 cells followed by DMS-53, A549 and KB31 cells (Figure 2Bi). Notably, KB31 cells express only very low Pgp levels. Moreover, Pgp functionality in these cells under normal glucose (25 mM; Figure 2Bii) was found to correspond to total plasma membrane Pgp expression (Figure 2Bi), as measured by the accumulation of the fluorescent Pgp substrate, Rh123, in all cell types via flow cytometry (Figure 2Bii). In this case, low cellular Rh123 accumulation indicates high Pgp activity and efflux of Rh123 out of cells.

Interestingly, varying glucose to low (0 mM) or high (50 mM) concentrations for 24 h/37°C, only slightly, but significantly (P < 0.05), decreased Rh123 accumulation relative to normal glucose (25 mM) in KB31 cells (Figure 2Ci). The other three cell types that express significant Pgp levels (i.e., KBV1, A549 and DMS53; Figure 2A and Bi, showed a far more pronounced and significant (P < 0.001–0.05) decrease in Rh123 accumulation upon changing glucose concentrations from normal levels (25 mM; Figure 2Cii-iv). This latter observation is consistent with increased Pgp-mediated efflux of Rh123 under glucose-induced stress. Indeed, Rh123 accumulation was Pgp-dependent, as the potent Pgp inhibitor, Ela (0.2 μM) (Akhtar et al., 2011), significantly (P < 0.001) increased Rh123 accumulation in KBV1, A549 and DMS-53 cells (Figure 2Cii-iv). The greater effect of Ela in KBV1, A549 and DMS-53 cells relative to KB31 cells, was probably due to their much higher Pgp levels (Figure 2A and Bi).

In these studies in Figure 2C, ROS were found to be important for increasing Pgp function, as incubation with NAC significantly (P < 0.001–0.05) reversed the glucose-induced increase in Pgp activity (at 0 and 50 mM glucose) relative to the respective glucose treatment only, as measured by Rh123 accumulation (Figure 2Ci-iv). This was particularly evident in cells expressing significant Pgp levels (i.e., KBV1, A549 and DMS53; Figure 2Cii-iv). Hence, these findings using NAC suggest ROS could be responsible for the increased Pgp expression at low and high glucose levels. Collectively, these observations in Figure 2 demonstrate that changes in glucose concentration lead to a higher Pgp-expressing phenotype. Therefore, to elucidate how glucose regulates Pgp expression, further mechanistic studies were performed utilizing KBV1 cells to take advantage of their marked Pgp expression/function.

Glucose variation-induced stress increases NF-κB nuclear translocation

During stress, HIF-1α is transcriptionally regulated by the redox-sensitive transcription factor, NF-κB, which binds at a distinct element in the promoter of the HIF-1α gene (van Uden et al., 2008). Therefore, we investigated the impact of glucose levels on NF-κB activity. Considering that NF-κB transcriptional activity requires translocation of its p65 subunit from the cytosol to the nucleus (Ghosh and Hayden, 2008), we initially investigated p65 expression in nuclear and cytosolic extracts of KBV1 (+Pgp) cells by Western blotting (Figure 3A). Expression of HDAC1 and GAPDH were only expressed in the nuclear and cytosolic fractions respectively.

Figure 3.

Changes in glucose levels induce translocation of the activated NF-κB p65 subunit into the nucleus. Analysis of KBV1 cells for (A) p65 expression in: (i) whole cells; (ii) the cytosolic fraction; and (iii) nuclear fraction, demonstrates a decrease in cytosolic p65 protein levels with a simultaneous increase in nuclear p65 protein accumulation after a 24 h/37°C incubation with low (0 and 12.5 mM) or high (50 mM) glucose. Fraction purity was confirmed with HDAC1 (nuclear marker) and GAPDH (cytosolic marker). (B) Immunofluorescence detection of p65 co-localization with the nuclear marker, DAPI, in KBV1 cells following a 4 h/37°C incubation with low, normal and high glucose levels. Co-localization of p65 with the nuclear marker, DAPI, was quantified as fluorescence intensity per cell. The results in (Ai–iii) are typical of three experiments, while densitometry is mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001 relative to normal glucose (25 mM). Immunofluorescence photographs in (Bi) are representative of three experiments and the quantified fluorescence intensity in (Bii) is mean ± SD (n = 3). Scale bar: 10 μm in each panel.

Incubating cells for 4 h/37°C with low (0 or 12.5 mM) or high (50 mM) glucose levels or with H₂O₂ (50 μM) did not significantly (P > 0.05) increase total p65 expression in KBV1 whole cell lysates relative to normal (25 mM) glucose-treated cells (Figure 3Ai). However, varying glucose concentrations (0, 12.5 and 50.0 mM) relative to normal levels (25 mM), significantly (P < 0.001–0.05) decreased p65 expression in the cytosol relative to normal (25 mM) glucose (Figure 3Aii).

In contrast to the cytosolic fraction, a significant (P < 0.001) increase in p65 was detected in the nuclear fraction at low (0 mM) and high (50 mM) glucose treatments relative to the normal (25 mM) glucose-treated cells (Figure 3Aiii). Hence, the studies in Figure 3A demonstrate that following low (0 mM) or high (50 mM) glucose treatments, marked p65 translocation from the cytosol to nucleus occurred.

The effect of glucose treatments on nuclear p65 was also assessed by immunofluorescence microscopy (Figure 3B). Relative to normal glucose (25 mM), incubation of KBV1 cells with low glucose (0, 12.5 mM) or high glucose (50 mM) led to nuclear p65 accumulation, as demonstrated by merging p65 (red) with DAPI (blue), forming a purple co-localization pattern (Figure 3Bi; see boxes as relevant examples of purple co-localization in the merged image). Quantification of p65 nuclear intensity demonstrated that H₂O₂ or low and high glucose relative to normal glucose levels, significantly (P < 0.001) increased p65 nuclear localization in KBV1 cells (Figure 3Bii). These findings demonstrate variations in glucose caused nuclear translocation of the NF-κB p65 subunit.

Glucose variation-induced stress regulates HIF-1α and MDR1 expression in MDR tumour cells

Importantly, it is known that HIF-1α is up-regulated by NF-κB (van Uden et al., 2008) and can increase Pgp expression (Comerford et al., 2002). Considering this, we examined the molecular events following glucose variation-induced stress in KBV1 (+Pgp) cells by examining the mRNA expression of HIF-1α, and its target gene, MDR1 (encoding Pgp). The glucose transporter gene, GLUT1, was used as a control as it displays a well-characterized homeostatic response to glucose, with low cellular glucose levels up-regulating its mRNA and high cellular glucose leading to its down-regulation (Mandarino et al., 1994; Hayashi et al., 2004; Thorens and Mueckler, 2010). In these studies, there was significant (P < 0.001–0.01) up-regulation of HIF-1α- and MDR1-mRNA expression following the low glucose (0 mM) and high glucose (50 mM) treatments for 24 h compared with normal glucose levels (25 mM; Figure 4A). GLUT1-mRNA significantly (P < 0.001) increased with glucose starvation (0 mM) relative to normal glucose (25 mM), but significantly (P < 0.05) decreased relative to the normal control (25 mM) at high (50 mM) glucose levels (Figure 4A). The positive control, H₂O₂, significantly (P < 0.001) up-regulated HIF-1α-, MDR1- and GLUT1-mRNA expression (Figure 4A). Hence, HIF-1α- and MDR1-mRNA demonstrate similar responses to glucose modulation and H₂O₂-mediated redox stress.

Figure 4.

Changes in glucose levels regulate the mRNA and protein levels of HIF-1α, MDR1 (Pgp) and GLUT1, and the protein levels of PHD2. Analysis of KBV1 cells show: (A) increased HIF-1α- and Pgp- (MDR1) mRNA expression via RT-PCR after a 24 h/37°C incubation with low (0 mM), normal (25 mM), or high (50 mM) glucose levels. GLUT1 mRNA expression was inversely proportional to glucose levels. (B) Increased Pgp and HIF-1α protein expression measured by Western blot after a 24 h/37°C incubation with low (0 and 12.5 mM), normal (25 mM) and high (50 mM) glucose. GLUT1 protein levels were inversely proportional to glucose levels. PHD2-protein expression was decreased after varying glucose levels from normal glucose (25 mM). The antioxidant, NAC (5 mM) or the NOX inhibitor, apocynin (50 μM) prevented the glucose variation-induced regulation of HIF-1α, Pgp, GLUT1 and PHD2 expression. H₂O₂ and AM (10 μM) were used as positive controls for the effects of ROS. (C) HIF-1α promoter-binding activity under varying glucose levels (see ‘Test’). The results in (A, B) are typical of three experiments with the densitometry being mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001 relative to normal glucose (25 mM). The results in (C) are presented as arbitrary units (a.u.) and are mean ± SD (n = 3). **P < 0.01, ***P < 0.001, relative to the respective control (i.e., normal glucose at 25 mM).

Glucose variation-induced stress regulates HIF-1α- and Pgp protein expression in MDR tumour cells

The translation of HIF-1α- and MDR1-mRNA was confirmed, as their protein levels were significantly (P < 0.01) increased in KBV1 cells following low (0 mM) and high (50 mM) glucose treatments for 24 h relative to normal glucose levels (25 mM; Figure 4B). Notably, HIF-1α appeared as two closely associated bands (Figure 4B), which are due to post-transcriptional modification (Zagzag et al., 2005). In contrast, GLUT1 protein remained inversely proportional to glucose availability (Figure 4B), as reported (Mandarino et al., 1994; Hayashi et al., 2004). The positive controls, H₂O₂ and AM, significantly (P < 0.01) increased HIF1α- and Pgp expression relative to normal glucose (25 mM), while it significantly (P < 0.01) decreased GLUT1 relative to normal glucose (25 mM; Figure 4B).

Considering the results mentioned earlier, PHD2 protein expression was then examined as it is responsible for oxygen-dependent degradation of HIF-1α (Bruick and McKnight, 2001), and may be regulated by altered glucose. Under the glucose levels used, PHD2 expression was inversely correlated to HIF-1α- and Pgp levels (Figure 4B). In fact, at lower (0 and 12.5 mM) and higher (50 mM) glucose, PHD2 expression was significantly (P < 0.01–0.05) decreased relative to normal (25 mM) glucose. The antioxidant, NAC (De Flora et al., 1995), and the NOX inhibitor, apocynin (Petronio et al., 2013), prevented the glucose variation-induced regulation of HIF-1α, Pgp, GLUT1 and PHD2 expression (Figure 4B).

Hence, at the lowest and highest glucose concentrations, greater HIF-1α expression was observed, which corresponded to the increased nuclear p65 (Figure 3Aiii and Bi,ii) and decreased HIF-1α degradation by PHD2 at these glucose levels relative to normal glucose (Figure 4B). This process can be prevented by an antioxidant or the inhibition of NOX.

Alterations in glucose increases HIF-1 promoter binding activity

To monitor HIF-1-regulated signal transduction pathways during glucose modulation, KBV1 cells were transfected with a HIF-1 promoter luciferase construct. Cells were also transfected with a constitutively expressing Renilla luciferase construct and a non-inducible Firefly luciferase construct which acted as positive and negative controls, respectively, to validate transfection (see ‘Positive’ and ‘Negative’; Figure 4C). The HIF-1 promoter-binding activity (see ‘Test’; Figure 4C) significantly (P < 0.001–0.01) increased following incubation of cells for 24 h with low (0 and 12.5 mM) or high (50 mM) glucose compared with normal glucose (25 mM; Figure 4C). This increase in HIF-1 promoter-binding activity was consistent with elevated HIF-1α protein levels at low and high glucose (Figure 4B) via p65 transcription factor translocation (Figure 3Aiii and Bi,ii).

Alterations in glucose modulates Pgp-induced DOX resistance

Considering that altering glucose levels changed Pgp transport activity (Figure 2C), we next examined if glucose modulation also results in Pgp-induced resistance to the chemotherapeutic agent and Pgp substrate, DOX (Shen et al., 2008) (Figure 5). Initial studies examined KB31 cells which express the lowest Pgp levels among the cell types (Figure 2A and Bi), and displayed slight, but detectable alterations in Pgp transport activity upon modulating glucose (Figure 2Ci). A 72 h incubation of KB31 cells with low (0 mM) or high glucose (50 mM), compared with normal (25 mM) glucose levels significantly (P < 0.01) increased the DOX IC₅₀ (i.e., concentration required for 50% inhibition of cellular proliferation; Figure 5A). This observation indicated that glucose variation increases DOX resistance, as higher DOX concentrations were required to inhibit proliferation (i.e., higher IC₅₀; Figure 5A). Hence, these results demonstrate that incubation of KB31 cells under glucose starvation conditions, results in increased Pgp activity, which leads to enhanced resistance to DOX and a higher IC₅₀ value. The role of Pgp in this increased resistance was confirmed by the Pgp inhibitor, Ela, which significantly (P < 0.01–0.05) reversed the IC₅₀ to levels similar to normal glucose (25 mM).

Figure 5.

Glucose-induced stress increases cellular resistance to the chemotherapeutic and Pgp-substrate, DOX. Cellular proliferation of: (A) KB31, (B) KBV1, (C) A549 and (D) DMS-53 cells in culture using medium supplemented with low (0 and 12.5 mM), normal (25 mM), or high glucose (50 mM) concentrations for 24 h/37°C before and during the incubation with DOX (24 h/37°C). Studies were performed in the presence or absence of the Pgp inhibitor Ela (0.2 μM). Results in (A–D) are mean ± SD (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001, versus normal glucose (25 mM), ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 versus respective glucose treatment concentration.

Similarly, in KBV1 cells that possess the highest Pgp expression and activity following low (0 mM) and high (50 mM) glucose treatment (Figure 2), there was a far more marked and significant (P < 0.001) increase in the DOX IC₅₀ value compared with normal glucose (25 mM; Figure 5B) relative to KB31 cells (Figure 5A). Moreover, the effect of Pgp inhibition with Ela was significant (P < 0.001) for all glucose levels, reducing the DOX IC₅₀ to similar levels as non-Pgp-expressing KB31 cells (Figure 5A and B). Notably, Ela also reversed resistance of KBV1 cells to DOX observed at normal glucose levels (25 mM; Figure 5B), as this cell type expresses high Pgp levels and activity under these conditions (Figure 2A, Bi and Bii). The A549 and DMS-53 cell types, which express appreciable Pgp (Figure 2A and Bi), showed similar results as KBV1 cells (Figure 5C and D).

Collectively, these data indicate that glucose levels regulate the sensitivity of cancer cells to DOX via Pgp induction. This effect was most pronounced in KBV1 cells that highly express Pgp, but could also be detected in KB31 cells that express much lower Pgp levels.

Discussion

Tumour cells exist in a stressful micro-environment, where crucial nutrients, such as glucose and oxygen, are heterogeneous (Pelicano et al., 2006; Annibaldi and Widmann, 2010). As a consequence, cancer cells are constantly adapting their metabolism to the tumour micro-environment. This investigation assessed how glucose variation-induced stress affected drug resistance in a cell culture model. To achieve this, glucose levels were altered to either a lower or higher glucose level to induce stress. Collectively, the present studies show that cancer cells develop a resistant phenotype upon ROS generation because of altered glucose availability.

The generation of ROS plays an important role in regulating genes responsible for the metabolic differences between normal and cancer cells (Denko, 2008; Hagen, 2012). Interestingly, ROS have been demonstrated to regulate downstream target genes of NF-κB and HIF-1α, including vascular endothelial growth factor A (VEGFA; Liu et al., 2006), which promotes angiogenesis (Byrne et al., 2005; Pages and Pouyssegur, 2005) and GLUT1, which activates glucose transport (Hayashi et al., 2004; Thorens and Mueckler, 2010). These intracellular ROS are primarily generated via NOX and are also formed as a by-product of the electron transport chain (Murphy, 2009; Block and Gorin, 2012). This study demonstrates that altered glucose levels increase ROS (Figure 1B, D and E) and that decreasing NOX activity or NOX4 protein levels reduces oxidative species generation following glucose modulation (Figure 1D). Additionally, the superoxide indicator, MitoSOX, demonstrated some mitochondrial superoxide generation as a consequence of altered glucose levels (Figure 1E), with the mitochondrial membrane becoming hyper-polarized (Figure 1F). Indeed, inhibition of NOX has been shown to inhibit glucose withdrawal-induced ROS (Graham et al., 2012). Also, mitochondrial ROS are linked to mitochondrial membrane potential (Griffiths, 2000), with hyperpolarization promoting ROS production through mitochondrial complex I (Yu et al., 2006). Together, ROS primarily from NOX4, and to a lesser extent, the mitochondrial complex, explain glucose variation-induced ROS generation (Figure 1B and 6 ).

Figure 6.

Schematic illustration of the glucose-induced Pgp resistance phenotype observed in this study. Alterations in glucose concentrations induces: (i) production of mitochondrial ROS via primarily the NOX4 enzyme, but also the electron-transport chain; (ii) translocation of NF-κB's active p65 subunit into the nucleus, where it is able to induce transcription of the master transcription factor; HIF-1α; (iii) decreased expression of PHD2, which prevents degradation of HIF-1α via the proteosome; (iv) increased expression of HIF-1, that binds to the hypoxia-response element (HRE) in the MDR1 gene promoter and leads to transcription of Pgp mRNA (MDR1) and its subsequent translation; and (v) increased Pgp on the plasma membrane, which acts as a ‘drug-pump’ to increase efflux of the cytotoxic Pgp substrate, DOX, from the cell. This response induces resistance to this chemotherapeutic.

While ROS can mediate cytotoxicity, there is also evidence to support their role in signal transduction (Behrend et al., 2003). In the current investigation, we observed that alterations in glucose levels resulted in increased Pgp expression (Figure 2A) and function (Figure 2Ci-iv), which was more apparent in cells with higher basal Pgp. As a consequence of glucose modulation, the involvement of ROS-mediated signalling (Behrend et al., 2003), was substantiated by significantly reduced Pgp activity via the antioxidant, NAC (Figure 2C). In agreement with our studies, up-regulation of Pgp expression was found when the antioxidant, glutathione, was depleted in rat brain endothelial cells (Hong et al., 2006).

Although previous studies showed that glucose deprivation abolishes Pgp activity through ATP restriction (Sauna and Ambudkar, 2007), we demonstrated that Pgp activity remains increased under such conditions, as measured by decreased Rh123 accumulation and increased Pgp expression (Figure 2B and C). Interestingly, it has been shown that while complete glucose starvation reduces ATP levels to 50%, this is not sufficient to reduce Pgp activity (Trach et al., 2012). This may be a result of the cell utilizing alternative metabolic sources for ATP production, such as fatty acids and proteins (Hsu and Sabatini, 2008). This may explain the sustained Pgp activity in our studies during glucose starvation.

The oxidant-inducible transcription factor, NF-κB (Piva et al., 2006), requires translocation of its p65 subunit to the nucleus (Oeckinghaus and Ghosh, 2009) to modulate HIF-1α expression during oxidative stress (Belaiba et al., 2007). In this investigation, glucose-induced stress led to translocation of the p65 subunit of NF-κB to the nucleus (Figure 3Aiii and Bi,ii). Consistent with p65 nuclear localization, increased HIF-1α expression was observed following alterations in glucose levels (Figure 4A and B), with a HIF-reporter assay confirming activation of HIF-regulated signal transduction (Figure 4C). Further, upon glucose-induced stress, reduced PHD2 expression also occurred (Figure 4B), which will decrease HIF-1α degradation (Bruick and McKnight, 2001). Hence, under these conditions, increased HIF-1α can induce Pgp transcription (Marxsen et al., 2004). Importantly, the central role of ROS in this pathway was confirmed by the utilization of the antioxidant, NAC or the NOX inhibitor, apocynin, which suppressed glucose-induced HIF-1α induction, while preventing the decrease in PHD2 (Figure 4B). This inhibition of ROS generation also resulted in decreased Pgp expression, indicating that glucose-induced ROS production was responsible for the observed increase in Pgp expression (Figure 6).

The present studies also demonstrated that cancer cells develop a resistant phenotype to the chemotherapeutic substrate, DOX, upon altered glucose availability. The increase in glucose-induced resistance was found to be due to Pgp, as co-incubation with the Pgp inhibitor, Ela, reversed DOX resistance (Figure 5). While the focus of this paper was Pgp, glucose variation-induced ROS may also regulate the expression of other ABC transporters including ABCG2 and MRP1, which are also HIF-1 responsive (Chen et al., 2009; Mo and Zhang, 2012). However, unlike the Pgp inhibitor, Ela, studies using the ABCG2 inhibitor, KO143 (Allen et al., 2002), or the MRP1 inhibitor, MK571 (van der Kolk et al., 1998), did not alter the cytotoxicity of DOX in the KBV1 cells which hyperexpressed Pgp in our studies (data not shown). This observation suggests that the glucose-induced resistance observed was due to Pgp expression.

Hence, our findings highlight that MDR might not only arise from repeated treatment of patients with anti-cancer drugs that can select for cells with high Pgp expression, but can also occur as a consequence of the stressful tumour micro-environment. As such, enhanced MDR because of glucose-induced stress could play a vital role in the response of tumours to chemotherapeutics.

Conclusions

Herein, for the first time, we show that glucose-induced stress acts to increase HIF-1α activity, resulting in increased Pgp expression and function in a culture model. Consequently, the resulting MDR phenotype increased DOX resistance. These studies highlight the potential importance of the tumour micro-environment in MDR and are important to consider for cancer treatment and future drug development.

Glossary

Abbreviations

ABC

ATP-binding cassette

AM

antimycin A

DOX

doxorubicin

Ela

Elacridar

DCF

2′,7′-dichlorofluorescein

H₂DCF

2′,7′-dichlorodihydrofluorescein

H₂DCFDA

2′,7′-dichlorodihydrofluorescein diacetate

HIF-1

hypoxia-inducible factor-1

MDR

multidrug resistance

NOX

NADPH oxidase

PEG-SOD

superoxide dismutase–polyethylene glycol

Pgp

P-glycoprotein

PHD2

prolyl hydroxylase 2

Rh123

rhodamine123

ROS

reactive oxygen species

VBL

vinblastine

Author contributions

P. J. J. and D. R. R. contributed equally to the work as co-corresponding and senior authors; P. J. J., N. A. S. and D. R. R. planned the research; P. J. J. and D. R. R.-directed experiments; N. A. S. performed experiments; all of the authors analysed data; N. A. S, D. R. R. and P. J. J. wrote the paper.

Conflicts of interest

None.