Intrinsic adaptations in OXPHOS power output and reduced tumorigenicity characterize doxorubicin resistant ovarian cancer cells

Abstract

Although the development of chemoresistance is multifactorial, active chemotherapeutic efflux driven by upregulations in ATP binding cassette (ABC) transporters are commonplace. Chemotherapeutic efflux pumps, like ABCB1, couple drug efflux to ATP hydrolysis and thus potentially elevate cellular demand for ATP resynthesis. Elevations in both mitochondrial content and cellular respiration are common phenotypes accompanying many models of cancer cell chemoresistance, including those dependent on ABCB1. The present study set out to characterize potential mitochondrial remodeling commensurate with ABCB1-dependent chemoresistance, as well as investigate the impact of ABCB1 activity on mitochondrial respiratory kinetics. To do this, comprehensive bioenergetic phenotyping was performed across ABCB1-dependent chemoresistant cell models and compared to chemosensitive controls. In doxorubicin (DOX) resistant ovarian cancer cells, the combination of both increased mitochondrial content and enhanced respiratory complex I (CI) boosted intrinsic oxidative phosphorylation (OXPHOS) power output. With respect to ABCB1, acute ABCB1 inhibition partially normalized intact basal mitochondrial respiration between chemosensitive and chemoresistant cells, suggesting that active ABCB1 contributes to mitochondrial remodeling in favor of enhanced OXPHOS. Interestingly, while enhanced OXPHOS power output supported ABCB1 drug efflux when DOX was present, in the absence of chemotherapeutic stress, enhanced OXPHOS power output was associated with reduced tumorigenicity.

1. Introduction

Cancer treatment is often thwarted by the development of drug resistance. Emerging evidence indicates that the acquisition of drug resistance typically occurs in parallel with augmented mitochondrial metabolism [1–6]. Specifically, relative to chemosensitive controls, drug resistant cells typically have larger mitochondrial networks and higher respiration [1–3]. Based on these mitochondrial signatures, there is great interest in developing therapies to eliminate drug resistant clones by targeting cancer cell mitochondrial bioenergetics [7]. Although a handful of mitochondrial inhibitors have shown anti-cancer efficacy in pre-clinical models [1,8–15], the therapeutic window for mitochondrial inhibition is likely constrained by unintended systemic toxicity across the body’s preeminent oxidative organs (e.g., heart, brain, muscle). Thus, of paramount importance is the identification of actionable phenotypes unique to cancerous mitochondria.

Using chemoresistant model systems of epithelial origin, Giddings et al., recently linked elevated mitochondrial respiration in chemoresistant cells to loss of DnaJ homolog subfamily C member 15 (DNAJC15) -- a negative regulator of respiratory CI [1]. Importantly, despite no evidence of systemic toxicity, forced inhibition of CI with a peptide mimetic of DNAJC15 restored in vivo chemosensitivity [1], presumably as a consequence of lowered cancer cell mitochondrial respiration. In this model, elevated mitochondrial respiration in chemoresistant cells was demonstrated to be necessary to support the ATP needs of ABCB1 (also called P-glycoprotein) -- an ABC transporter responsible for cytotoxic drug efflux. Drug efflux pumps, like ABCB1, are common signatures of cancer drug resistance [16]. By using the free energy of ATP hydrolysis, these pumps expel chemotherapeutics at the expense of cellular ATP. Thus, ABCB1 upregulation presumably elevates cellular energetic demand. In response to this demand, intrinsic adaptions would be hypothesized to take place across the cancerous mitochondrial network to boost OXPHOS power output. Due to the limited trackability of whole cell respirometry, elucidating intrinsic adaptations in OXPHOS power output necessitates biochemical methodologies capable of independently interrogating the OXPHOS system [17]. That said, direct interrogation of the OXPHOS system in cancer mitochondria is relatively understudied.

The purpose of the present study was twofold: 1) test the impact of ABCB1 activity on cancer cell basal respiration, and 2) characterize how the OXPHOS system is intrinsically remodeled to support epithelial cancer cell drug resistance. To do this, we leveraged an in-house phenotyping platform that integrates comprehensive bioenergetic flux measurements with the underlying mitochondrial proteome [4,18,19]. Consistent with prior work [1], DOX resistant ABCB1 overexpressing cells presented with elevations in both basal and maximal respiration. Across distinct cancer cell models, acute inhibition of ABCB1, with zosuquidar [20], lowered basal respiration only in ABCB1-expressing chemoresistant cells. Although zosuquidar did lower respiration in an ABCB1-dependent manner, basal respiration remained elevated in chemoresistant cells. Interestingly, upon correcting for group differences in mitochondrial content and total cellular protein, differences in cellular respiratory kinetics were eliminated. Thus, in addition to enhanced ATP resynthesis demand driven by ABCB1, higher basal respiration in DOX resistant cells is attributable to both larger cell size and enhanced mitochondrial volume. With respect to intrinsic bioenergetic remodeling, in-depth bioenergetic analysis of chemoresistant mitochondria revealed adaptations in favor of increased respiratory CI function. Such changes boosted intrinsic OXPHOS power output, consistent with prior findings detailing an absolute requirement for mitochondrial ATP to support ABCB1-dependent DOX efflux [1]. Interestingly, although both ABCB1 upregulation and mitochondrial remodeling afforded cancer cell survival in the presence of DOX, in the absence of DOX, 2-dimensional proliferation, anchorage independent growth, as well as in vivo tumorigenicity were all lower in DOX resistant cells. Thus, in the absence of a chemotherapeutic stressor, elevated OXPHOS, and/or increased reliance on CI functionality may constrain tumorigenicity.

2. Materials and Methods

2.1. Materials

All chemicals for both mitochondrial and proteomic analysis were purchased from either Millipore Sigma or Thermo Fisher Scientific. Both doxorubicin hydrochloride and vincristine sulfate were purchased from Tocris. Anchorage independent growth assays were performed using assay kits supplied by Cell Biolabs Inc. Fetal bovine serum (FBS) was purchased from Peak Serum (Wellington, CO).

2.2. Cell culture

NCI/ADR-Res and OVCAR-8 cells were previously described [21]. Human VCR resistant cell line, HL-60/VCR, was developed by Safa and colleagues from HL-60/Vinc cells, originally developed by Melvin S. Center, Division of Biology, Kansas State University, Manhattan, KS, by step-wise selection to increasing concentrations of VCR[22]. HL-60 cells were obtained from the ATCC, Manassas, VA. All cells were cultured in RPMI-1640 medium, supplemented with 10% FBS, 100 units/ml penicillin, and 100 μg/ml streptomycin. NCI/ADR-Res cultures media was supplemented with 1μM DOX dissolved in water. HL60/VCR culture media was supplemented with 1.0 μg/ml (1.2 μM) vincristine sulfate dissolved in water. Prior to experimentation, DOX was removed from NCI/ADR-Res culture media for 24hr preceding cell harvest. For xenograft experiments, DOX was removed from the NCI/ADR-Res culture media for 7 days prior to experimentation. Cell lines were not tested or authenticated over and above documentation provided by ATCC, which included antigen expression, DNA profile, short tandem repeat profiling, and cytogenetic analysis.

2.3. Cell proliferation assays

Following 5 days exposure to either vehicle or DOX, cell viability was determined by fluorescence measurement as previously described[23]. Briefly, cells were seeded in Corning black wall, 96-well plates, treated with indicated agents (vehicle or DOX) in a final volume of 0.2 ml RPMI-1640 complete medium, and incubated at 37° C, 5% CO₂, for the times indicated. Viability was determined using propidium iodide (PI) as follows. Cells (positive controls corresponding to 100% cell death) were permeabilized by addition of 10μl of 1 mg/ml digitonin and incubated at 37° C, 5% CO₂ for 20 min, followed by addition of PI dissolved in PBS, at a final concentration of 5μM. The plate was incubated for 20 minutes, and viability was calculated as the mean fluorescence (minus non-permeabilized control) at 530 nm excitation and 620 nm emission, using a fluorescent plate reader. Viability was also determined using trypan blue (0.4%) exclusion and cell counting on a Countess II apparatus using disposable hemocytometers. Cell size was determined in single-cell suspensions, after trypsinization, using a Countess II apparatus.

2.4. Anchorage independent growth

Assays were performed in Corning black wall, 96-well plates assay components supplied by Cell Biolabs Inc. A solution of 1.3% agar was mixed with 2x RPMI-1640 complete media and 50μL was added to each well. Agar was allowed to solidify for 30 minutes at 4°C. After 30 minutes, cell suspensions were prepared according to the following formula: 1.2% agar, 2x RPMI-160 complete media, and cells suspended in 1x RPMI-1640 complete media (1:1:1). Cell concentrations were 0.2 × 10⁶ cells/mL to achieve a seeding density of 5,000 cells when 75μL of the 1:1:1 mixture was added to each well. Following addition of the cells, agar was allowed to solidify for 15 minutes at 4°C. Following agar solidification, 100μL of RPMI-1640 complete media, supplemented with either vehicle or DOX was added to each well. Plates were either analyzed immediately (i.e., baseline plate) or cultured for a period of 8 days to allow colony formation. At the end of the 8 day period, assay media was decanted and 50μL of agar solubilization solution was added to each well. Plates were incubated for 1 hour at 37°C. Individual wells were mixed 5–10 times with a multi-channel pipette to ensure complete agar solubilization. To each well, 25μL of concentrated lysis buffer was added and again individual wells were mixed using a multi-channel pipette. Plates were incubated for 15 minutes at room temperature. From each well, 10μL of cell lysate was added to individual wells of a 96 well plate and 90μL of CyQuant working solution was added to each well. The plate was incubated for 10 minutes at room temperature to allow fluorescence to develop. Plates were analyzed in a fluorescent plate reader using excitation/emission parameters of 485/520. Data were expressed relative to fluorescence intensity of the baseline plate analyzed at the time of cell seeding.

2.5. Cellular respirometry (intact and permeabilized)

High-resolution O₂ consumption measurements were conducted using the Oroboros Oxygraph-2K (Oroboros Instruments, Innsbruck, Austria) in intact and digitonin permeabilized cells. For each intact cell experiment, cells were centrifuged at 300 × g for 7 min at room-temperature, washed in PBS, centrifuged once more and then suspended in assay media at a cell concentration of ~1 × 10⁶ viable cells/ml. Assay media was RPMI 1640, without bicarbonate, supplemented with 20 mM HEPES (pH 7.4), 10% FBS, 100 units/ml penicillin, and 100 μg/ml streptomycin. All experiments were carried out at 37°C in a 1-mL reaction volume. Following the assessment of basal respiration, the following sequential additions were made (oligomycin (0.02μM), FCCP (0.5, 1, 2, 3μM), antimycin A (Ant A; 0.5μM). For experiments designed to assess the impact of ABCB1, zosuquidar (5μM) was added prior to oligomycin to intact cells. For permeabilized cell experiments, cells were centrifuged at 300 × g for 7 min at room-temperature, washed in respiration buffer, centrifuged once more and then suspended in respiration buffer at a cell concentration of ~1 × 10⁶ viable cells/mL. Respiration buffer consisted of potassium-MES (105mM; pH 7.2), KCl (30mM), KH₂PO₄ (10mM), MgCl₂ (5mM), EGTA (1mM), and BSA (2.5 g/L). After recording basal respiration, cells were permeabilized with digitonin (20 μg/mL), energized with various carbon substrates (pyruvate, malate, glutamate, succinate, octanoyl-l-carnitine; P,M,G,S,O; 5 mM, 2 mM, 5 mM, 5 mM, 0.2 mM) and flux was stimulated across a physiological ATP free energy demand using the creatine kinase (CK) clamp as previously described. For complete details regarding the calculation of ΔG_ATP at each titration point see [19]. For experiments designed to assess CI/CII respiration, malonate (20mM) was added to inhibit succinate supported respiration [24]. Additional respiratory stimulation was carried out via respiratory uncoupler (FCCP) titration (0.5–3.0μM). Cytochrome C (10μM) was added to check the integrity of the outer mitochondrial membrane. Note, the absence of an increase in respiration, relative to the pre cytochrome C rate, was used as a quality control assessment for outer membrane integrity. Non-mitochondrial respiration was controlled for by adding Ant A (0.5μM). Data were normalized to viable cell count and expressed as pmol/s/million cells and then corrected for the mitochondrial enrichment factor (MEF) calculated for each group (see proteomics methodology below regarding MEF). All additions were made directly to the O2K chamber during the period of each assay. Typical assay length was 20–40 minutes.

2.6. Mitochondrial ATP synthesis, P/O determination, and OXPHOS power output

Parallel respiration and fluorometric ATP synthesis experiments were carried out in order to generate an ATP/O (P/O) ratio as previously described [18,25]. Fluorometry experiments were carried out using a QuantaMaster Spectrofluorometer (QM-400; Horiba Scientific) at 37°C in a 200μL reaction volume with continuous stirring. All experiments were conducted using the hexokinase (HK) clamp to maintain the desired ADP concentration (0.15mM). Respiration buffer was supplemented with HK (1U/mL), glucose (5mM), glucose-6-phosphate dehydrogenase (2U/mL), and NADP⁺ (4mM), as well as P1,P5-di(adenosine-5’)pentaphosphate (Ap5A; 0.2mM). Ap5A was included to inhibit adenylate kinase, preventing the generation of ATP from ADP alone [25]. Respiration was assessed in parallel as described above. For both respiration and ATP synthesis assays, to begin, digitonin permeabilized cells were added, followed by ADP/AP5A, respiratory substrates (P/MG then O/S), and oligomycin (0.02μM). Respiration experiments also included an FCCP titration (0.5, 1, 2μM), as well as Ant A (0.5μM). The rate of change in NAD(P)H fluorescence (Ex:Em, 350:450) was equated to the rate of ATP production (JATP), as previously described [25]. The P/O ratio was then calculated as JATP/JO₂, divided by 2.

As previously described[4], empirically derived P/O ratios were then used to convert oxygen consumption rates recorded during the CK clamp assay (using identical substrates) to ATP production rate [(oxygen consumption rate) × (P/O × 2)]. A fixed extra-mitochondrial force was assumed to be applied via the CK clamp at each PCr titration. These forces corresponded to −54.16, −58.93, −60.64, and −61.49 kJ/mol. These forces were used along with ATP production rate to quantitate OXPHOS power output in Watts (J/s) according to the following formula [(pmol ATP/s/million cells) × (5.416 or 5.893 or 6.064 or 6.149 * 10⁻⁸ J/pmol ATP) = μWatts/million cells]. All data were corrected for group specific MEF, as described above.

2.7. Nucleotide analysis in intact cells

OVCAR-8 and NCI/ADR-Res cells were seeded in 10cm dishes and allowed to reach ~70% confluency. The 24hr prior to cell harvest, DOX was removed from the NCI/ADR-Res cultures. At the time of harvest, plates were placed on ice. Growth media was aspirated, and cells were washed twice with ice-cold PBS. Cells were harvested in ice-cold acetonitrile/methanol/water (40%/40%/20%) using a cell scraper. Following sonication, samples were incubated for 20 minutes at −20°C. Samples were thawed and centrifuged at 12,000 × g for 10 minutes at 4°C. Supernatant was transferred to fresh tubes and stored at −80°C until analysis. A sample aliquot was dried by centrifuge vacuum concentrator (Eppendorf Vacufuge Plus), reconstituted in water, and then analyzed by ultra-performance liquid chromatography (Waters Acquity H-class Premier system), as previously described [26]. ATP, ADP, and AMP were measured simultaneously in a single run per sample with quantification by UV absorbance. To quantitate total protein, cell pellets were dissolved in 0.2M NaOH and total protein was determined. Energy charge was calculated according to the following formula [(ATP) + ½(ADP)]/[ATP + ADP + AMP] [27].

2.8. In vivo experiments

All experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of ECU. OVCAR-8 and NCI/ADR-Res cells were cultured for a period of seven days. All NCI/ADR-Res cultures were devoid of DOX. Single cell suspensions of both cell lines were prepared in sterile PBS. Under sterile conditions, cells (1 × 10⁶ cells in 0.2mL of PBS) were injected into the subcutaneous space of the right flank of male NSG mice (Jackson Laboratories Strain ID: 005557). Mice were 10 weeks of age at the state of cell injection. Tumors were allowed to develop for 45 days. Mice were sacrificed and tumors were excised to be weighted and analyzed.

2.9. Mitochondrial lysis and sample prep for label-free proteomics

Cell pellets were lysed in urea lysis buffer (8M urea in 40mM Tris, 30mM NaCl, 1mM CaCl₂, 1 × cOmplete ULTRA mini EDTA-free protease inhibitor tablet; pH=8.0), as described previously [18,28]. The samples were subjected to two freeze-thaw cycles, and sonicated with a probe sonicator in three 5s bursts (Q Sonica #CL-188; amplitude of 30). Samples were centrifuged at 10,000 × g for 10min at 4°C. Protein concentration was determined by BCA. Equal amounts of protein were reduced with 5mM DTT at 37°C for 30min, and then alkylated with 15mM iodoacetamide for 30min in the dark. Unreacted iodoacetamide was quenched with DTT (15mm). Reduction and alkylation reaction were carried out at room temperature. Initial digestion was performed with Lys C (1:100 w:w) for 4hr at 32°C. Following dilution to 1.5M urea with 40mM Tris (pH=8.0), 30mM NaCl, 1mM CaCl₂, samples were digested overnight with sequencing grade trypsin (50:1 w/w) at 32°C. Samples were acidified to 0.5% TFA and then centrifuged at 4,000 × g for 10min at 4°C. Supernatant containing soluble peptides was desalted, as described previously [28] and then eluate was frozen and subjected to speedvac vacuum concentration.

2.10. nLC-MS/MS for label-free proteomics

Final peptides were resuspended in 0.1% formic acid, quantified (ThermoFisher Cat# 23275), and then diluted to a final concentration of 0.25μg/μL. Samples were subjected to nanoLC-MS/MS analysis using an UltiMate 3000 RSLCnano system (ThermoFisher) coupled to a Q Exactive Plus Hybrid Quadrupole-Orbitrap mass spectrometer (ThermoFisher) via a nanoelectrospray ionization source. For each injection, 4μL (1μg) of sample was first trapped on an Acclaim PepMap 100 20mm × 0.075mm trapping column (ThermoFisher Cat# 164535; 5μL/min at 98/2 v/v water/acetonitrile with 0.1% formic acid). Analytical separation was performed over a 95min gradient (flow rate of 250nL/min) of 4–25% acetonitrile using a 2μm EASY-Spray PepMap RSLC C18 75μm × 250mm column (ThermoFisher Cat# ES802A) with a column temperature of 45°C. MS1 was performed at 70,000 resolution, with an AGC target of 3×10⁶ ions and a maximum injection time (IT) of 100ms. MS2 spectra were collected by data-dependent acquisition (DDA) of the top 15 most abundant precursor ions with a charge greater than 1 per MS1 scan, with dynamic exclusion enabled for 20s. Precursor ions isolation window was 1.5m/z and normalized collision energy was 27. MS2 scans were performed at 17,500 resolution, maximum IT of 50ms, and AGC target of 1×10⁵ ions.

2.11. Data analysis for label-free proteomics

As described previously [18,28], with some modification, Proteome Discoverer 2.2 (PDv2.2) was used for raw data analysis, with default search parameters including oxidation (15.995 Da on M) as a variable modification and carbamidomethyl (57.021 Da on C) as a fixed modification. Data were searched against the Uniprot Homo Sapiens reference proteome (Proteome ID: UP000005640), as well as the Human Mito Carta 3.0 database [29]. PSMs were filtered to a 1% FDR and grouped to unique peptides while maintaining a 1% FDR at the peptide level. Peptides were grouped to proteins using the rules of strict parsimony and proteins were filtered to 1% FDR. Peptide quantification was done using the MS1 precursor intensity. Imputation was performed via low abundance resampling. Using only high confidence master proteins, mitochondrial enrichment factor (MEF) was determined as previously described [18] by comparing mitochondrial protein abundance (i.e., proteins identified to be mitochondrial by cross-reference with the MitoCarta 3.0 database) to total protein abundance.

2.12. Statistical evaluation

All proteomics samples were normalized to total protein abundance, and the protein tab in the PDv2.2 results was exported as a tab delimited .txt. file and analyzed. Protein abundance was converted to the Log₂ space. For pairwise comparisons, tissue mean, standard deviation, p-value (p; two-tailed Student’s t-test, assuming equal variance), and adjusted p-value (Benjamini Hochberg FDR correction) were calculated [30]. Mitochondrial functional assay results are expressed as the mean ± SEM (error bars). Data were normalized to viable cell counts and then corrected for the group mean MEF, with the final values expressed as pmol/s/million cells/MEF [18]. Throughout the paper, differences between groups were assessed by t-test, or one-way ANOVA, followed by Tukey’s test where appropriate using GraphPad Prism 8 software (Version 8.4.2). Other statistical tests used are described in the figure legends. Statistical significance in the figures is indicated as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Unless otherwise stated, figures were generated using GraphPad Prism 8 software (Version 8.4.2) or Biorender.

3. Results

3.1. Verification of doxorubicin resistance and heightened mitochondrial metabolism in NCI/ADR-Res.

The present study made use of ovarian cancer cells either sensitive (OVCAR-8) or resistant (NCI/ADR-Res) to DOX. To verify DOX resistance, cell lines were continuously cultured in DOX for a period of 5 days. In OVCAR-8, DOX exposure dose-dependently lowered viable cell counts (Fig. 1A), with higher concentrations leading to cytotoxicity (Fig. 1B). In contrast, both proliferation and cell viability were largely unaffected by DOX in NCI-ADR-Res (Fig. 1A–B), consistent with their known drug resistant phenotype [1,21].

Fig. 1. Increased basal and maximal respiration characterize NCI/ADR-Res.

(A) Viable cell counts in OVCAR-8 and NCI/ADR-Res cultured in 0, 0.5, 2.0, and 10μM DOX for five days. (B) Percent viability for each group at the indicated DOX dosages. (C) Oxygen consumption in intact OVCAR-8 and NCI/ADR-Res cells under basal conditions, as well as following the addition of oligomycin (Oligo; 0.02μM), FCCP titration, and antimycin A (Ant). Figures generated using GraphPad Prism 8 software (Version 9.1.0). Data are Mean ± SEM, (A, B) N=3/group, (C) N=11/group. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Like other drug resistance models, NCI/ADR-Res are biochemically characterized by heightened mitochondrial oxidative metabolism[1]. To confirm this phenotype, both OVCAR-8 and NCI/ADR-Res were suspended in complete growth media (HEPES buffered RPMI, 10% FBS) and respiratory flux was measured using high resolution respirometry. Measurements were made under basal condition, as well as in response to the following additions: a) Oligomycin to inhibit ATP synthase; b) Uncoupler (FCCP) titration to determine maximal respiratory capacity; and c) Antimycin A (an inhibitor of Complex III) to control for non-mitochondrial O₂ consumption. Relative to their DOX-sensitive counterparts, NCI/ADR-Res cells presented with higher basal, as well as maximal respiration (Fig. 1C). While the addition of oligomycin inhibited basal respiration in OVCAR-8, NCI/ADR-Res were refractory to oligomycin (Fig. 1C; “Oligo”).

3.2. ATP-dependent drug efflux contributes to higher basal respiration in NCI/ADR-Res.

DOX resistance, in NCI/ADR-Res, is attributable to ABCB1-dependent drug-efflux [1]. Because drug efflux has an absolute requirement for ATP hydrolysis, the increased ATP demand commensurate with ABCB1 activity may directly contribute to NCI/ADR-Res’ high mitochondrial respiration. However, whether ABCB1’s ATPase function is sufficient to stimulate whole cell respiration remains unclear. To specifically evaluate the bioenergetic cost of ATP-dependent drug efflux, respirometry experiments were repeated in the absence and presence of either extracellular DOX (an ABCB1 substrate) or zosuquidar [20] (an ABCB1 inhibitor). As a function of ABCB1 expression, the following two hypotheses were formulated: 1) DOX administration was expected to further activate ABCB1-dependent ATPase activity, quantifiable by an increase in basal respiration; 2) Akin to the effects observed in response to ATP synthase inhibition with oligomycin, acute inhibition of ABCB1 was anticipated to blunt respiration. Surprisingly, DOX had no effect on basal respiration in either cell line (Fig. 2A). However, acute inhibition of ABCB1 with zosuquidar lowered respiration exclusively in DOX resistant cells (Fig. 2B). Although 2.5μM zosuquidar completely restored oligomycin sensitivity in NCI/ADR-Res, complete inhibition of ABCB1 activity was not sufficient to normalize basal respiration between groups (Fig. 2B). Consistent with this, relative to vehicle control, zosuquidar administration, to intact NCI/ADR-Res, lowered basal respiration by only ~5–10% (Fig. 2C). To further quantitate the contribution of ABCB1 flux to basal respiration, zosuqiduar exposure experiments were repeated in a different ABCB1-dependent drug resistant cancer cell line[3]. Like NCI/ADR-Res, relative to wildtype controls, HL60 cells resistant to vincristine (HL60/VCR) also upregulate ABCB1 (Fig. 2D). Like that seen in NCI/ADR-Res, acute ABCB1 inhibition lowered respiration by ~ 5–10% in vincristine resistant cells (Fig. 2E). Taken together, although the ATPase activity of ABCB1 does contribute to elevated basal mitochondrial metabolism, ABCB1 activity alone is not sufficient to fully explain higher basal mitochondrial respiration in chemoresistance.

Fig. 2. Impact of ABCB1 activity and mitochondrial content on intact cellular respiration.

(A) Intact basal respiration in the absence and presence of 3μM DOX. Data expressed as a fold change from the pre-DOX condition. (B) Intact basal respiration in the absence and presence of zosuquidar (2.5μM), followed by oligomycin (0.02μM). (C) Intact basal respiration in the presence of vehicle or 2.5μM zosuquidar, followed by oligomycin (0.02 μM). Data expressed as a percentage of basal. (D) ABCB1 expression in HL60wt versus HL60/VCR cells. (E) Intact basal respiration in the presence of vehicle or 2.5μM zosuquidar, followed by oligomycin (0.02 μM). Data expressed as a percentage of basal. (F) Volcano plot depicting changes in proteome abundance between NCI/ADR-Res and OVCAR-8 cells. Proteins corresponding to the MitoCarta 3.0 database are shown in red. (G) Calculated mitochondrial enrichment factor (MEF). (H) Cell size quantification performed in single cell suspensions. (I) Intact basal respiration normalized to total protein and corrected for group mean MEF. Figures generated using GraphPad Prism 8 software (Version 9.1.0). Data are Mean ± SEM, (A) N=7/group, (B) N=3/group, (C) N=4/group, (D) N=5/group, (E) N=3/group, (F) N=4/group, (G) N=4/group, (H) N=9/group, (I) N=17/group. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

3.3. Changes in basal respiration largely reflect differences in both cell size and mitochondrial volume.

Given that ABCB1 activity was insufficient to fully explain basal respiratory kinetics, we hypothesized that higher basal mitochondrial flux in DOX resistant cells could also be influenced by differences in cellular mitochondrial content. To test this, we performed whole proteome screens on representative cell aliquots from both OVCAR-8 and NCI/ADR-Res cells (all proteomics data available in Supplemental Table 1). As validation of our proteomics approach, relative to OVCAR-8, the most over-expressed protein in NCI/ADR-Res was ABCB1 -- the ATP-dependent transporter responsible for DOX resistance (Fig. 2F). With respect to the mitochondrial proteome, MitoCarta 3.0 positive proteins were skewed to the right side of the volcano plot (Fig. 2F), a phenomenon indicative of increased mitochondrial content. In support of this, when we compared the abundance of all MitoCarta 3.0 positive proteins to total protein abundance, the resulting mitochondrial enrichment factors (MEF) were indeed higher in NCI/ADR-Res (Fig. 2G). Thus, the percentage of total protein attributable to the mitochondrial network was higher in NCI/ADR-Res, consistent with mitochondrial network expansion. Interestingly, cell size, assessed in single cell suspensions, was also slightly larger in NCI/ADR-Res (Fig. 2H). Based on these data we hypothesized that elevations in NCI/ADR-Res basal respiration may reflect both their larger cell size and a more expansive mitochondrial network. To test this, basal respiration was normalized to cellular protein scaled to MEF. Upon normalization, group differences in basal respiratory kinetics were eliminated (Fig. 2I). Thus, higher basal respiration, scaled to cell number, in NCI/ADR-Res is largely driven by both increases in cell size and mitochondrial network expansion.

3.4. Intrinsic alterations in NCI/ADR-Res OXPHOS kinetics link to CI upregulation.

A novel feature of our mitochondrial phenotyping approach lies in its ability to integrate sophisticated flux measurements with the underlying mitochondrial proteome. In this way, both mitochondrial composition (i.e., proteome) and function (i.e., flux) can be interpreted per mitochondrion (i.e., upon normalization to a fixed amount of mitochondrial protein). This is achieved by normalizing mitochondrial protein expression data to total MitoCarta 3.0 abundance and then scaling all flux measurements to group specific MEF. From first principles, the OXPHOS system consists of a series of multi-subunit protein complexes, embedded in the inner mitochondrial membrane, that either pump or conduct protons (Fig. 3A). Interestingly, despite clear heterogeneity in OXPHOS subunit expression (Fig. 3B), the summed abundances of respiratory CI and CIII were increased in NCI/ADR-Res (Fig. 2C). The summed abundance of CII was downregulated in NCI/ADR-Res (Fig. 3C). Despite differences in respiratory complex expression, maximal respiratory capacity, in substrate replete digitonin permeabilized cells, was identical between groups (Fig. 3D). In a subset of cell preparations, maximal respiratory capacity was assessed in parallel using both intact and permeabilized cells. This allowed us to compare respiratory capacity across the disparate methodologies. Interestingly, unlike that observed in NCI/ADR-Res, OVCAR-8 respiratory capacity was consistently higher in permeabilized, relative to intact cells (Fig. 3E), suggesting the existence of negative respiratory regulators in OVCAR-8 cytosol.

Fig. 3. Functional upregulations in CI characterize NCI/ADR-Res.

(A) Cartoon depiction of the OXPHOS system. (B) Heatmap displaying OXPHOS protein abundance, separated by OXPHOS complex. Individual protein abundance was scaled to mean. Asterisk (*) denotes significant differences using an adjusted P value cutoff of 0.1. (C) Mean abundance of each respiratory complex. (D) Assessment of respiratory capacity in digitonin permeabilized cells in response to carbon substrate energization and FCCP titration. (E) Relationship between respiratory capacity assessed in either permeabilized or intact cells. (F) Assessment of OXPHOS kinetics using the CK clamp technique. (G) OXPHOS kinetics at minimal ATP free energy in response to the sequential additions of substrates and inhibitors designed to quantify maximal CI versus CII supported respiratory flux. (H) Ratio of CI to CII respiration. Respiratory assays are expressed in pmol/s/million cells/MEF. Figures generated using either Biorender or GraphPad Prism 8 software (Version 9.1.0). Data are Mean ± SEM, (B-C) N=4/group, (D) N=9/group, (E) N=14, (F) N=10/group, (G-H) N=8/group. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Given our interest in investigating intrinsic OXPHOS remodeling commensurate with DOX resistance, a series of assays were designed to evaluate group differences in OXPHOS kinetic efficiency. To do this, we made use of the creatine-kinase clamp system [19,31,32]. This system leverages the enzymatic activity of creatine-kinase (CK) to couple the interconversion of ATP and ADP to that of phosphocreatine (PCr) and free creatine (Cr). In so doing, this technique allows the extramitochondrial ATP free energy (i.e. ΔG_ATP) to be clamped at an empirically defined value and then further titrated with sequential PCr additions. Thus, stimulation of mitochondrial respiration can proceed across a physiological ΔG_ATP span (i.e. −56 to −64 kJ/mol) [33]. Using this approach, following digitonin-permeabilization, the CK clamp was established at −54.16 kJ/mol, an ATP/ADP ratio that is expected to induce near maximal OXPHOS flux. Mitochondria were then energized with CI-specific substrates (pyruvate/malate/glutamate; P/M/G) followed by the additions of both octanoyl-carnitine and succinate (O/S), as well as PCr titration, oligomycin, and FCCP titration. Despite similar OXPHOS conductance in the presence of P/M/G/O/S, maximal OXPHOS flux supported exclusively by P/M/G was higher in NCI/ADR-Res (Fig. 3F), suggesting potential CI upregulation. To further probe complex-specific differences in OXPHOS kinetics, CK clamp experiments were repeated in permeabilized cells energized with substrates specific to either CI or CII. Although differences in respiration with NAD-linked Pyr/M, G, or O did not reach significance (Fig. 3G), the ratio of CI/CII respiration was higher in NCI/ADR-Res (Fig. 3H). With respect to the OXPHOS proteome, although expression differences were evident across CI, CII, and CIII (Fig. 3B–C), only alterations in CI appeared functionally relevant. Thus, in addition to having more mitochondria per cell (Fig. 2G), NCI/ADR-Res mitochondria appear intrinsically remodeled to favor CI-supported OXPHOS.

3.5. ATP energy charge in unchanged in OVCAR-8 and NCI/ADR-Res cells.

Given the observed differences in mitochondrial bioenergetics between OVCAR-8 and NCI/ADR-Res, we hypothesized that such differences may be associated with changes in cellular ATP energy charge. To evaluate this, we assessed the concentrations of ATP, ADP, and AMP in intact cells under basal conditions. In intact cells, concentrations of ATP, ADP, and AMP were unchanged across groups (Fig. 4A–C). In addition, calculated energy charge was also not different across groups (Fig. 4D). Thus, despite intrinsic differences in mitochondrial bioenergetics, both OVCAR-8 and NCI-ADR-Res cells maintain an identical energy charge under basal conditions.

Fig. 4. Analysis of Energy charge in intact cells.

Quantification of ATP (A), ADP (B), AMP (C). (D) Calculated energy charge. Figures generated using GraphPad Prism 8 software (Version 9.1.0). Data are Mean ± SEM, N=3/group.

3.6. Increases in OXPHOS power output negatively correlate with tumorigenicity.

Given that differences in OXPHOS respiratory kinetics were apparent between OVCAR-8 and NCI/ADR-Res, we sought to determine if these changes translated to an increase in ATP output. To do this requires empirical determination of the ATP to O₂ stoichiometry (i.e., P/O ratio) [17]. Thus, we designed two parallel assays that pair measurements of oxygen consumption to the spectrofluorometric detection of ATP synthesis (JATP). In either assay, both respiration and JATP were stimulated using an ADP concentration (0.15mM) mimicking that elicited by a ΔG_ATP of −54.16 kJ/mol. All assays included the adenylate kinase (AK) inhibitor Ap5A to control for AK-dependent JATP [25]. Energization was first initiated with the addition of complex I substrates (P/M/G), followed by the addition of substrates to saturate the carbon pool (O/S). Oligomycin (an inhibitor of ATP synthase) was added to control for non ETS-dependent JATP. Consistent with earlier findings, NCI/ADR-Res supported respiration was higher with CI substrates (Fig. 5A; ‘P/M/G’). However, contrary to that seen with either FCCP alone (Fig. 3D) or the CK clamp (Fig. 3F), ADP stimulated respiration was also higher in NCI/ADR-Res upon saturation of the carbon pool and these differences persisted in response to FCCP titration (Fig. 5A). Parallel determination of ATP synthesis (Fig. 5B) revealed similar fold differences in JATP (Fig. 5C), such that the P/O ratio was similar across groups (Fig. 5D). Armed with group specific P/O, we used these data to convert respiratory flux at each ΔG_ATP to ATP production rate [JO₂ × (P/O*2) = ATP production rate]. Assuming extra-mitochondrial force applied via the CK clamp is fixed for a given ΔG_ATP, ATP production rate can be used to quantitate OXPHOS power output in Watts (J∙s⁻¹) [4]. Relative to OVCAR-8, maximal OXPHOS power output was higher in NCI/ADR-Res (Fig. 5E). Taken together, these data inform a model whereby in the absence of any changes in total respiratory capacity, intrinsic adaptions across the NCI/ADR-Res mitochondrial network boost OXPHOS power output, presumably to support ABCB1-dependent DOX efflux.

Fig. 5. Intrinsic elevations in OXPHOS power output characterize NCI/ADR-Res.

(A) Assessment of ADP-supported respiration in digitonin permeabilized cells. (B) Representative image of a JATP synthesis assay in digitonin permeabilized cells. (C) Quantified JATP synthesis and JO2 in digitonin permeabilized cells energized with either P/M/G or P/M/G/O/S. (D) P/O ratio. (E) Assessment of OXPHOS power output across a ΔG_ATP span. Figures generated using GraphPad Prism 8 software (Version 9.1.0). Data are Mean ± SEM, (A) N=5–6/group, (C-D) N=5/group, (E) N=10/group, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Although enhancements in OXPHOS power output likely underscore NCI/ADR-Res survival when DOX is present, in the absence of DOX pressure, we hypothesized that such alterations may differentially impinge on cell growth and tumorigenicity. To address this, cell proliferation was assessed across a five-day period in the absence of DOX. Relative to NCI/ADR-Res, cellular proliferation was faster in OVCAR-8 (Fig. 6A). Next, to assess how increases in OXPHOS power correlate to tumorigenicity, we conducted an anchorage independent growth assay in soft agar. Experiments were performed in the presence of vehicle alone, as well as upon exposure to DOX. In the presence of DOX, despite no impact on NCI/ADR-Res tumorigenicity, colony formation was eliminated in OVCAR-8 (Fig. 6B). Interestingly, relative to OVCAR-8, tumorigenicity in the absence of DOX was reduced by nearly 50% in NCI/ADR-Res (Fig. 6B). To confirm our results in vitro, we performed xenograft experiments in immunocompromised mice (Fig. 6C). To do this, both OVCAR-8 and NCI/ADR-Res cells were cultured in the absence of DOX for a period of 7 days. After DOX removal for 7 days, both basal and maximal respiration remained higher, when normalized to cell count, in NCI/ADR-Res (Fig. 6D). After the DOX washout period, viable cells (1×10⁶ cells, suspended in sterile PBS) were injected subcutaneously into the right flank of immunocompromised male mice. Tumors were allowed to develop for 45 days, at which time mice were sacrificed and tumor burden was assessed. Consistent with our in vitro results, tumor burden was decreased in mice injected with NIC/ADR-Res cells (Fig. 6E–F). Collectively, these data suggest the intriguing possibility that intrinsic bioenergetic adaptations in favor of enhanced OXPHOS power and/or an increased reliance on respiratory CI confers a fitness disadvantage that manifests exclusively in the absence of chemotherapeutic stress, resulting in impaired tumorigenicity.

Fig. 6. Reduced tumorigenicity in NCI/ADR-Res.

(A) Cell proliferation assessed in the absence of DOX for five days. Data expressed as fold change in viable cells seeded at day 0. (B) Anchorage independent growth in OVCAR-8 and NCI/ADR-Res cells exposed to vehicle or DOX (3μM). Data expressed as fold change from day 0. (C) Cartoon schematic of xenograft experiments performed in immunocompromised mice. (D) Oxygen consumption in intact OVCAR-8 and NCI/ADR-Res cultured in the absence of DOX for 7 days. (E) Representative images of tumors resected from mice 45 days post cell injection. (F) Quantified tumor burden. (G) Expression of ABCB1 in tumor lysates. Figures generated using GraphPad Prism 8 software (Version 9.1.0), as well as Biorender. Data are Mean ± SEM, (A) N=3/group, (B, D, F) N=6/group, (E) N=3/group. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

4. Discussion

Measurements of intact cellular respiration typically begin in a black box, as changes in oxidative metabolism could be attributable to modifications in total cellular mitochondrial content, bioenergetic efficiency, or cellular energetic demands [17]. To begin to fill this gap in knowledge, we leveraged a comprehensive diagnostic biochemical workflow to investigate the bioenergetic mechanisms underlying the enhanced oxidative metabolism reported in ABCB1 overexpressing, doxorubicin resistant ovarian cancer cells. In agreement with prior reports [1], DOX resistant cells presented with elevations in both basal and maximal respiration. In addition to alterations in cellular bioenergetics, paired mitochondrial phenotyping and proteomics analyses revealed functional upregulations in respiratory CI in DOX resistant cells. Our findings demonstrating specific upregulations in CI agree with a recent report that linked elevated respiration in chemoresistant cells to loss of the CI negative regulator DNAJC15 [1]. With respect to the impact of CI upregulation on NCI/ADR-Res bioenergetics, mechanistic studies in digitonin permeabilized cells associated CI upregulation to an increase in OXPHOS power output. Power is a critical diagnostic of mitochondrial bioenergetics because it encompasses thermodynamic, kinetic, and stochiometric descriptions that are reflective of the organelle’s ability to generate an intracellular ΔG_ATP charge (i.e., displace the ATP/ADP ratio from equilibrium). Interestingly, OXPHOS power output was higher in NCI/ADR-Res only at the lowest ΔG_ATP tested. Chronically less negative ΔG_ATP would be expected to maintain high mitochondrial oxidative metabolism and thus could explain the consistency by which chemoresistant models present with high respiration. However, we did not observe any differences, under basal conditions, in cellular energy charge, suggesting that differences in basal OXPHOS flux are likely driven by kinetic differences in ATP turnover. Indeed, acute inhibition of ABCB1 activity lowered respiration only in ABCB1 expressing cells.

Cell proliferation has been shown to result in a reduction in the intracellular ATP/ADP ratio [34,35], and by extension a less negative ΔG_ATP. Although the consequential increase in basal respiratory activity is often attributed to elevated ATP resynthesis, OXPHOS has been shown to be dispensable for tumor outgrowth [36]. Moreover, recent data suggests that the cancer cell’s demand for biomass outpaces its need for mitochondrial ATP resynthesis such that ATP availability is not rate limiting to proliferation [37]. Here, we show that in the absence of chemotherapeutic stress, increases in OXPHOS power output negatively correlate with proliferation and tumorigenic growth. While our results cannot be used to infer a causal relationship between OXPHOS power and tumorigenic growth potential, it is tempting to speculate that constrained OXPHOS power output may be favorable for uncontrolled cell proliferation, but unfavorable for cells reliant on ATP-dependent drug efflux as a mechanism of chemoresistance. In this mechanism, cells with intrinsically elevated OXPHOS power output are potentially selected for in the presence of DOX, as these adaptations may oppose growth pressures, while at the same time supplying the ATP needs for ABCB1-dependent DOX efflux.

With respect to ABCB1, although the ATPase activity of ABCB1 was directly linked to elevated basal mitochondrial metabolism, ABCB1 activity alone was not sufficient to fully explain higher basal mitochondrial respiration in chemoresistance. Basal respiration in intact cells is generally accepted to reflect cellular energy demand for ATP resynthesis. However, whether basal respiration in proliferating cells is indeed responsive to cellular demand for ATP resynthesis is not clear. In addition to ATP resynthesis, many cellular processes, including proton leak, are driven by the mitochondrial proton current. Prior work has demonstrated that ATP, per se, is not rate limiting to proliferation [37], thus is it possible that mitochondria inside proliferating cells are optimized for a specific flux. In this way, basal respiration may be buffered by shifting proton current through various ATP and non-ATP dependent resistors to achieve an optimized flux. In the present study, although increased ABCB1 activity was directly linked to elevations in basal mitochondrial respiration, acute ABCB1 inhibition failed to completely normalize differences in respiration across distinct chemoresistant models. Thus, future work is required to elucidate the mechanistic drivers responsible for differences in basal mitochondrial metabolism in the setting of drug resistance. Mechanistic understanding as to how cancer cells sustain and regulate basal mitochondrial metabolism may have important therapeutic implications.

5. Conclusions

Modifications to OXPHOS have been implicated in the development of drug resistance to a variety of chemotherapeutics, leading to an increased focus on the implementation of mitochondrial-targeted anti-cancer therapies to combat disease progression [38]. While these strategies have demonstrated clinical relevance as a potentially effective approach to tackle chemoresistance, the development of novel, cancer-specific therapeutics is potentially hindered by a limited understanding of the functional changes these various metabolic adaptations impart on cancer mitochondrial bioenergetics. While the present study suffers from the limitation of using cancer cell lines alone, our use of comprehensive bioenergetic phenotyping to unveil novel features of cancerous OXPHOS control may serve as a blueprint for future discovery.

Supplementary Material

Supp Table 1

Supplemental Table 1. Proteomics Data. (A) Exported results from Proteome Discoverer 2.2 using the whole mouse proteome database. (B) Analyzed results, inclusive of MEF calculation. (C) MitoCarta 3.0 database. (D) Exported results from Proteome Discoverer 2.2 using the MitoCarta 3.0 database only. (E) Analyzed results for MitoCarta 3.0 positive proteins. (F) Mean abundance of the OXPHOS complexes.

Supp Table 2

Supplemental Table 2. Source Data. Individual data points for all figures.

Highlights.

• Basal respiration is elevated in ABCB1-dependent DOX resistant cancer cells

• ABCB1 dependent drug efflux accounts for ~7% of basal mitochondrial respiration

• Mitochondria from DOX resistant cells upregulate CI and OXPHOS power

• In the absence of chemotherapy, enhanced OXPHOS correlates to impaired tumorigenicity

Abbreviations

OXPHOS

oxidative phosphorylation

CI

complex I

DOX

doxorubicin

ATP

adenosine triphosphate

VCR

vincristine

P

pyruvate

M

malate

G

glutamate

O

octanoyl-carnitine

S

succinate

Ant

antimycin A

Rot

rotenone

Oligo

oligomycin

ZOQ

zosuquidar

JATP

rate of ATP synthesis

JO₂

rate of oxygen consumption

ΔG_ATP

free energy of ATP hydrolysis

MEF

mitochondrial enrichment factor

Data availability

All raw data for proteomics experiments is available online using accession number “PXD031629” for Proteome Xchange [39] and accession number “JPST001475” for jPOST Repository [40]. Proteomics data also available in Supplemental Table 1. All other data are available in the source data files provided with this paper (Supplemental Table 2).