Pancreatic ductal adenocarcinoma (PDAC) is a very aggressive cancer, with a low percentage of affected patients eligible for surgical resection and highly refractory to conventional therapies [1
]. Therefore, more effective drugs are urged to improve current treatment regimens. Apart from cell growth, DNA repair, invasiveness, and angiogenesis, PDAC cells are hallmarked by mutations in genes involved in metabolism [1
]. New therapeutic strategies targeting metabolism are emerging as promising approaches to overcome chemoresistance [4
]. However, inter- and intra-tumor heterogeneity, often result in different metabolic phenotypes also as a consequence of multiple interactions with the tumor microenvironment [5
]. This poses therapeutic limitations and highlights the importance of preliminary metabolic characterizations of the tumor lineages, preparatory to the administration of effective drugs. We recently demonstrated that two pancreatic cancer cell lines, characterized by a different metabolic profile, produce a dissimilar response to glucose deprivation/galactose substitution, an approach that is able to rewire energy metabolism [6
]. Moreover, our group had already demonstrated the efficacy of dichloroacetate (DCA), an inhibitor of pyruvate dehydrogenase kinase (PDK), to kill cultured cells derived from human oral carcinomas, an effect inversely correlated with the mitochondrial respiratory capacity of the tumor cells [7
]. Several in vivo and in vitro studies describe the ability of DCA to increase mitochondrial oxidative phosphorylation (OxPhos), reverting the Warburg effect and selectively targeting tumor cells [8
]. As well, an extensive body of literature demonstrates the efficacy of DCA to enhance chemo-sensitivity in several cancer types [10
]. DCA treatment has been purposed both for in vitro and in vivo studies also in pancreatic cancer [8
]. Nevertheless, further investigations are necessary to better define the efficacy of the drug in this cancer type, to clarify potential additional mechanisms leading to cell death, and to explore possible further ways to limit the side effects encountered. In this study, we analyzed the effects of DCA on the two PDAC cell lines, PANC-1 and BXPC-3, chosen among others for their similar growth conditions and well-characterized geno-/phenotype [6
]. A broad metabolite and transcriptome profiling of PDAC cell lines identified three tumor subtypes, with PANC-1 and BXPC-3 belonging to a lipogenic cluster hallmarked by a distinct reliance on glucose oxidation and mitochondria-related metabolism [17
]. Testing DCA on 2D and 3D cultures of the PDAC cell lines, we demonstrated that the drug negatively affects vital parameters by reducing the mitochondrial respiratory activity and, most notably, the cancer stem cell compartment. Moreover, we showed that DCA is also able to mitigate in vivo tumor growth in a model of PDAC xenografted mice.
2. Materials and Methods
2.1. Cell Culture
PANC-1 and BXPC-3 cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured at 37 °C in a 5% CO2 humidified atmosphere in complete RPMI medium supplemented with 10% fetal bovine serum, penicillin–streptomycin (100 U/mL), and 2 mM of glutamine, and the glucose concentration was typically 10 mM or 1 mM when indicated. Dichloroacetate (DCA) was purchased from Sigma-Aldrich (St. Louis, MO, USA). For each in vitro experiment, cells were treated with DCA 4 mM and 10 mM at indicated times.
2.2. Cell Growth Curves
Cell growth curves were performed as previously described [18
2.3. Real-Time Cell Proliferation Monitoring by xCELLigence System
The xCELLigence experiments were performed using the RTCA (real-time cell analyzer) instrument, according to the manufacturers’ instructions (ACEA Biosciences, San Diego, CA, USA). The optimal seeding number was previously determined by cell titration and growth experiments (data not shown). The 2500 cells/well were then seeded and their proliferation was automatically monitored every 30 min and 24 h after seeding, cells were treated with DCA. The cell index was monitored up to 90 hours from seeding. Data were analyzed using xCELLigence software (Version 2.0, Acea biosciences, San Diego, CA, USA) and expressed as a mean ± SD of the cell index normalized to the last cell index recorded before the time of DCA addition.
2.4. Apoptosis Assay
After incubation with DCA, cells were stained with Annexin-V-FITC and PI (BD Biosciences). Live, apoptotic, and necrotic cells were detected using flow citometry (Navios, Beckman Coulter, Brea, CA, USA). Three independent experiments were carried out. A total of 104 events for each sample were acquired.
2.5. Migration Assay
The effects of DCA on PANC-1 and BXPC-3 migration capacities were assessed using a scratch wound assay. Briefly, cells were seeded into six-well culture plates and cultured to complete confluence. Subsequently, three parallel, linear wounds were produced in each dish with a 200 μL plastic pipette tip. The cells were then treated with DCA and the wound healing ability, monitored at different time points, was quantified after 48 h. Three representative images of scratched areas from each dish were photographed to estimate the migration of cells. The cell migration rate was calculated using the following formula: [1 − (48 h scratch width/0 h scratch width)] × 100%.
2.6. Lactate Measurements
A lactate colorimetric assay kit (Abcam, Cambridge, MA, USA) was used following the manufacturer’s protocol and the lactate concentration detected (intracellular or released) was normalized to cell number.
2.7. Metabolic Flux Analysis and Mitochondrial Respiratory Complex Enzymatic Activity
The oxygen consumption rate (OCR) and extra-cellular acidification rate (ECAR) were measured in adherent PANC-1 and BXPC-3 cells with an XF96 extracellular flux analyzer (Seahorse Bioscience, Billerica, MA, USA) as previously described [19
]. Briefly, for the OCR analysis, after measuring basal respiration, oligomycin (1 μM), FCCP (1 μM), and rotenone + antimycin A (1 μM + 1 μM) were injected into each well sequentially to assess, respectively, the coupling of the respiratory chain, and the maximal and non-mitochondrial oxygen consumption. For the ECAR analysis, glycolytic flux (basal glycolysis, glycolytic capacity, and glycolytic reserve) was analyzed by the sequential addition of 10 mM glucose, 1 μM oligomycin, and 100 mM 2-deoxyglucose. The OCR and ECAR values were normalized to protein content in each well, determined using BCA assay (Thermo Scientific, Waltham, MA, USA).
2.8. Mitochondrial DNA Quantification
The measurement of mtDNA copy number, relative to nuclear DNA copy number, was determined as previously described [6
2.9. Live Cell Imaging of mtΔΨ and ROS
Cells cultured at low density on fibronectin-coated 35-mm glass-bottom dishes (Eppendorf, Amburgo, Germany) were incubated for 20 minutes at 37 °C with 2 μM of TMRE, 10 μM of DCF (Molecular Probes, Eugene, OR, USA) to monitor mtΔΨ and ROS, respectively. Stained cells were washed with PBS and examined using a Leica TCS SP8 confocal laser scanning microscope. Acquisition, storage, and data analysis were performed with a dedicated instrumental software from Leica (LAS-X, Wetzlar, Germany).
2.10. Western Blotting Analysis
Aliquots, containing 40 μg of proteins from each lysate cell, were subjected to SDS polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane using a Trans Blot Turbo Transfer System. Membranes (Bio-Rad Laboratories, Hercules, CA, USA) were probed with the following primary antibodies: pyruvate dehydrogenase E1-alpha (PDH) and pPDHSer293 (1:500, Abcam, Cambridge, UK), LC3B (1:1000 Cell Signaling Technology), TOM20 (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA, USA), DRP1 (1:1000, BD Bioscences), OPA-1 (1:1000, BD Bioscences), MFN1 (1:1000, Santa Cruz), MFN2 (1:1000, Abnova, Tapei, Taiwan), and CASPASE 3 (1:1000, Cell Signaling Technology, Danvers, MA, USA). After incubation with a correspondingly suited horseradish peroxidase-conjugated secondary antibody (1:2500; Cell Signaling Technology), signals were developed using the enhanced chemiluminescence kit (ClarityTM Western ECL Substrate, Bio-Rad) and the ChemiDoc imaging system XRS + (BioRad), and then analyzed using Image Lab software (version 4.1, Bio-Rad, Hercules, CA, USA). The intensity of LC3B-II (corresponding to the cleaved fraction), TOM20, DRP1, OPA-1, MFN1, and MFN2 bands were normalized to the β-actin signal while PDH phosphorylation was normalized to total proteins.
2.11. Flow Cytometric Detection of Surface Markers
Surface marker expressions CD44, CD24, and EPCAM, was evaluated by citofluorimetric analysis in PANC-1 and BXPC-3 treated with DCA for 24 h. In brief, after trypsinization, cells were incubated in the dark at room temperature for 15 min with CD44-APC, EPCAM-FITC, and CD24-PE directly conjugated monoclonal antibodies (BDB). Cytofluorimetric analysis was performed by Navios (Beckman Coulter). The emitted fluorescent signal of 10,000 events for each sample was acquired and analyzed using the Kaluza Analysis software (version 1.3, Beckman Coulter, Brea, CA, USA).
2.12. Reverse Transcription and Real-Time PCR Analysis
One microgram of total RNA, isolated using Trizol reagent (Life Technologies, Paisley, UK), according to the manufacturer’s instruction and quantified on a Nanodrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), was used in a reverse transcription (RT) reaction using the Transcriptor first strand cDNA synthesis kit (Roche Diagnostic, Penzberg, Germany) according to the manufacturer’s instructions. Quantitative real-time polymerase chain reaction (PCR) was performed in duplicate, using the QuantiTect Primer Assay (Qiagen, Basel, Switzerland) detecting Lin28 mRNA. Quantification of the mRNA levels was performed on a LightCycler® 480 real-time PCR instrument. The relative amounts of Lin28 were normalized with GAPDH expression by Light Cycler® 480 Software version 1.5 (ROCHE) using the 2ΔΔCt method.
2.13. 3D Culture
PANC-1 and BXPC-3 cells were detached with trypsin-EDTA and counted. Then, 1000 cells/well were seeded into ultra-low attachment 96-well round-bottomed plates and cultured in RPMI. To evaluate the DCA effect of preformed spheroids, 3D cultures were maintained for 7 days, obtaining spheroids. Then, medium was replaced with fresh medium and spheroids were treated with DCA 4 mM and 10 mM for 72 h. To evaluate the effect of DCA to spheroid formation, DCA was added to cell suspension when seeded into ultra-low attachment plates and the culture was maintained for 7 days. Spheroids were photographed on an inverted optical microscope (Axio Vert A1, Zeiss, Oberkochen, Germany) and their diameter was measured using the ZEISS ZEN imaging software. Spheroid viability was assessed using an MTS assay. A solution of cellTiter 96® Aqueous MTS Reagent Powder (Promega, Madison, WI, USA) and PMS (Sigma Aldrich, Saint Louis, MO, USA) was added to each well of 3D spheroids culture. After 2 h of incubation at 37 °C the absorbance at 490 nm was measured and the percentage of viability in each well was calculated using the untreated spheroids as 100%.
2.14. Animal Studies
The animal testing was executed in an AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care International, Frederick, MD USA) accredited experimental facility under the approval number ANM14_002/468862. A total number of 5 × 106 BxPC-3-luc cancer cells were cultured, resuspended in 0.1 mL of PBS/matrigel mixture (1:1) and then s.c. injected into the right flank of 5–6 weeks old Nu/Nu nude mice. When tumor size reached an average volume of 100 mm3, BxPC-3-luc tumor-bearing nude mice were randomly assigned into 2 groups (6 mice/group). Group 1 (normal saline, i.p, qw), group 2 (DCA, mg/kg, i.p, qw). Animals had free access to water. DCA was dissolved to generate a final concentration of 100 mg/kg/day (s.c: sub-cutaneous; i.p: intraperitoneal; qw: once a week).
2.15. Statistical Analysis
Experimental data are expressed as the mean ± standard error mean (SEM) or mean ± standard deviation (SD). Data were compared using the unpaired Student’s t-test or one-way Anova, followed by the Bonferroni test. A p value < 0.05 was accepted as statistically significant.
Aberrant metabolic processes generally occur in cancer cells, and therefore targeting metabolism represents an emerging strategy to treat tumors, including pancreatic cancer [31
]. Tumor heterogeneity may result in malignant cells with distinct metabolic phenotype, and consequently, different sensitivity to metabolic drugs as in the case of gemcitabine toward the majority of patients with pancreatic cancer develop resistance [17
In the present study, we tested the efficacy of the metabolic drug DCA in 2D and 3D cultures from two different well-characterized pancreatic cancer cell lines (i.e., PANC-1 and BXPC-3) as well as in a xenograft model of pancreatic cancer. Both cell lines, grown in monolayer, showed a marked sensitivity to DCA at the highest concentration tested (i.e., 10 mM), which halted cell proliferation and severely inhibited their migration capacity. Testing the drug at a lower concentration (i.e., 4 mM DCA) highlighted a higher sensitivity of the PANC-1 cell line. This observation is of interest because PANC-1 is reported as an aggressive and chemo-resistant cell line [21
Analysis of the viability parameters in both the DCA-treated PDAC cell lines unveiled a limited percentage of apoptotic/necrotic cells, thereby suggesting a cytostatic rather than a cytotoxic effect exerted by the drug, confirming previous reports [36
A major effect of DCA is generally attributed to its ability to induce a metabolic switch from glycolysis to mitochondrial glucose oxidation. This is achieved by inhibition of the PDH kinase PDK, thereby shifting PDH toward its more active unphosphorylated state [38
]. Consequently, pyruvate is converted into acetyl-CoA which enters the tricarboxylic acid cycle and fuels the mitochondrial oxidative phosphorylation.
However, in this study, we found that despite a substantial DCA-induced dephosphorylation of PDH no activation of the mitochondrial respiratory activity was observed in both of the drug-treated PDAC cell lines. On the contrary, DCA caused a dose-dependent reduction of mitochondrial OxPhos which was coupled to inhibition of the glycolytic capacity in BXPC-3.
This result was also somewhat surprising considering that in a previous study with oral cancer cell lines we demonstrated that PE15 cells, characterized by a sustained OxPhos, were resistant to the DCA treatment, while HSC2/3 cells, exhibiting a glycolytic profile, appeared more drug-sensitive with a marked effect also on mitochondrial morpho-functional parameters [7
]. Moreover, in another study with PANC-1 and BXPC-3 cell lines, we demonstrated a differential sensitivity to glucose deprivation/galactose substitution, a condition that also fosters the oxidative metabolism, with the more glycolytic BXPC-3 cells being more vulnerable [6
]. This led us to hypothesize that the dissimilar sensitivity of different cell lines to drugs or conditions promoting a pro-oxidative metabolic shift was dependent on their basal metabolic profile, with those relying more on glycolysis and/or with low respiratory capacity being more vulnerable.
The depressing effect of DCA on mitochondrial respiration did not apparently result in changes of the mitochondrial morphology, although a significant reduction of the fission promoting-factor Drp1 was observed in DCA-treated BXPC-3 cells. Likely the basal fragmented phenotype of the mitochondrial network in BXPC-3 hidden to appreciate further mitochondrial fragmentation. However, the mtDNA copy number/cell was significantly increased in both the PDAC cell lines, likely due to a compensatory mechanism as a consequence of mitochondrial dysfunction leading to activation of mitophagy, as demonstrated by the increased cleaved form of LC3B-II. The observed enhanced production of ROS in DCA-treated PDAC cell lines might elicit the organelle quality control to remove damaged mitochondria. Unbalance of ROS homeostasis is commonly related to dysfunction of the mitochondrial respiratory chain, although the relationship is often not clear (i.e., cause, effect, vicious cycle). Generally, considered a pro-survival mechanism protecting cells under stress conditions (oncogenic function) [40
], more recently it has been demonstrated that dysregulation of mitophagy contributes to drug resistance (tumor suppressive role) [41
]. Anyway, induction and inhibition of mitophagy in cancer progression is still controversial.
All together the aforementioned observations do not enable us to rationalize the cytostatic effect of DCA as simply linked to metabolic rewiring of the PDAC cells. It must be taken into account that DCA can target other cellular pathways in addition to PDK. Indeed, it has been reported that DCA affects the CoA biosynthetic pathway [42
], activates the AMPK signaling pathway [43
], antagonizes with acetate [44
], and disturbs the tyrosine catabolism [45
]. Moreover, comparison of the metabolite profiles in cells treated with DCA or more selective novel inhibitors of PDK resulted in different outcomes [46
]. This led us to investigate additional potential off-target effects of DCA to explain its efficacy to hit tumor cells.
Cancer stem cells (CSCs) represent a fraction of the whole tumor mass emerging as responsible for cancer therapy refractory as well as metastasis dissemination and tumor relapse [47
], thereby attracting growing interest as targets for the development of new anticancer therapies [48
]. To the best of our knowledge, there is no report about the effect of DCA on pancreatic cancer stem cells. To dissect this intriguing aspect, first, we evaluated the effect of DCA treatment on Lin 28 expression revealing a significant dose-dependent downregulation detectable in both cell lines. Of note, Lin28 expression is strictly linked to metabolism since it is able to regulate cancer cell progression via PDK1 and to induce an energetic switch [49
]. Lin28 is involved in the formation of CSCs [50
] and its aberrant expression is associated with many human neoplastic diseases, including pancreatic cancer [51
]. FACS analysis of the surface antigens expression CD44, CD24, and EPCAM typically featuring pancreatic CSCs [53
] unveiled that DCA treatment reduced the percentage of the triple positive fraction in PANC-1. In contrast, we did not appreciate any DCA-induced modulation in BXPC-3 which were mainly constituted by triple positive cells. It should be taken into account that although more than 90% of the BXPC-3 was positive for stemness markers, their expression level was relatively low. Conversely, PANC-1 expressed higher levels of stemness markers, even though only in less than 30% of the cell population, suggesting a younger CSCs phenotype characterizing this cell subset. Consistent with this observation is the notion that rather than the absolute expression of the CSC markers it is their ratio to “qualify” the stemness propensity of cancer cells [28
]. In keeping that the CD44/CD24 expression ratio in PANC-1 is much higher than in BXPC-3 this would indicate that, although less populated, the
CSC compartment in PANC-1 is qualitatively more stem cell like. Conversely, the low expression level of the CSC markers
broadly spread in BXPC-3 cell population would phenotype them as early progenitors. This difference in the two pancreatic cancer cell lines might
account for their distinct metabolic phenotypes and sensitivity to
chemotherapeutic drugs as well as to DCA.
Three-dimensional (3D) cell culture technology has become a focus of research in tumor cell biology. Compared to 2D, 3D cultures from cell lines involve an enrichment in CSCs [30
] and by mimicking the metabolic and proliferative gradients of in vivo tumors, provide a more reliable prediction of the response to a possible treatment [55
]. In this perspective, we tested the efficacy of DCA on 3D culture obtained from PANC-1 and BXPC-3 and showed that DCA treatment compromised the structure and viability of already formed spheroids and compromised spheroid formation from both cell lines. In particular, at the higher dose DCA was able to almost completely inhibit spheroid formation from PANC-1. Consistent with our observations in 2D cultures, BXPC-3 spheroids were less sensitive then PANC-1 spheroids to DCA treatment. Likewise, in 2D cultures a significant Lin28 downregulation was also observed in spheroids from both cell lines. Although the above reported changes in the expression level of broadly recognized CSC markers in both 2D and 3D cultures do not imply conclusive evidence of the effect of DCA on the stemness compartment of PDAC, nevertheless, they provide hitherto unappreciated clues deserving further investigations. At the in vivo level, DCA treatment caused a slower, although not significant, cancer growth in BxPC-3-luc tumor-bearing mice as compared to the control mice, as assessed by reduced photon counting and decreased tumor volume.
To rationalize the puzzling effects of DCA on the PDAC cell lines reported in our study, we put forward the following hypothetical sequence of events purposed to stimulate further investigations (Figure 10
). We suggest that consequently to an enhanced oxidation of pyruvate more reducing equivalents are transferred to the mitochondrial respiratory chain with generation of ROS. The respiratory complexes are both producers and target of ROS [57
], thereby fostering a vicious cycle leading to progressive inhibition of the functional electron transfer throughout the respiratory chain promoting further ROS-genic diversion of electrons to O2
. In addition, damping of the respiratory activity may cause accumulation of intermediates of the tricarboxylic acid cycle as well as of acetyl-CoA. When the latter accumulates it is known to cause lysine-acetylation and inhibition of function of a number of mitochondrial proteins including respiratory chain complexes [58
]. The ensuing progressive mitochondrial damage/dysfunction is counteracted by upregulation of mitophagy. How and if these DCA-mediated mitochondrial alterations cause growth arrest and rewiring of the cell phenotype, in particular of the cancer stem cell compartment, remains to be established. However, a number of reported evidences indicate that a pro-oxidative state causes stem cell to exit from their undifferentiated state and induces/favors commitment [60
]. Moreover, epigenetic modifications, such as those causing chromatin remodeling, regulate the balance between pluripotency and differentiation of stem cells [62
]. Possibly the stalled TCA cycle may cause efflux of citrate in the cytosol where it releases acetyl-CoA, thus enhancing its availability for histone acetylation. Obviously, other uncharacterized DCA-targets may contribute or even dominate the observed drug effects.
In conclusion, our results clearly indicate that the efficacy of DCA in inhibiting cancer cell growth is not always causally related to its documented stimulatory effect on the PDH activity and consequently reverse Warburg effect. Other off-targets must be mechanistically considered depending on the cell phenotype. In this context, the evidence, emerging from this study, that the CSC compartment in PDAC-derived cell lines might be affected by DCA treatment is relevant. It would be worthy to verify if this occurs in other cancer cell types and work is in progress in our lab in this direction. Recent research projects developing surface-functionalized, dichloroacetate-loaded nanoparticles [64
], and multifunction drugs obtained by chemotherapeutic agents with DCA as ligand [65
] could help to design the right targeted pharmacological formulation for developing new effective therapeutic strategies to fight pancreatic as well as other types of cancer.