Background
Despite significant advances in both our understanding of cancer, and in the development of new treatments, annual world-wide cancer-related fatalities remain high, with 9.6 million deaths (accounting for 1 in 6 deaths) alone in 2018 (World Health Organization). This is particularly true for tumors driven by cancer stem cells (CSCs), which have been shown to be responsible for tumor heterogeneity, metastasis, chemoresistance, and tumor relapse in a large number of cancers [
1‐
4]. Thus, from a medicinal perspective, targeting the CSC population represents a very appealing anti-cancer strategy. Unfortunately, progress in the development of anti-CSC agents has been very slow, and these types of compounds are still very far from reaching the clinic (reviewed in [
5‐
7]).
We and others have studied CSCs at the genetic, epigenetic, transcriptional, proteome and metabolic levels to identify their targetable weaknesses [
8‐
11], allowing us to discover that CSCs of many tumor types [
12,
13], such as pancreatic ductal adenocarcinoma (PDAC), preferentially use aerobic oxidative phosphorylation (OXPHOS) over anaerobic glycolysis to meet their energy requirements. As a consequence, CSCs exhibit increased mitochondrial mass and membrane potential (reflection of mitochondrial function) [
14]. While OXPHOS involves a significantly greater number of biochemical reactions, it is almost 20 times more efficient than glycolysis in terms of generation of ATP per unit of glucose. Considering this dependence on mitochondrial respiration, it is reasonable to predict that targeting OXPHOS in CSCs may represent an effective approach for treating cancer.
Some of us have recently demonstrated that ruthenium complexes containing a bipyridine and a terpyridine ligand, and one exchangeable (reactive) coordinating position ([Ru(terpy)(bpy)X]
n+) can react with solvent accessible guanines in DNA [
15,
16]. In contrast to other standard DNA metalating agents, such as cisplatin, these types of complexes present a kinetically controlled reactivity with DNA, likely due to the bulkiness provided by their ruthenium ligands, which also allows for exquisite chemoselectivity. Likewise, these complexes were found to smoothly ruthenate solvent-exposed guanines present in adjacent positions of four-stranded guanine DNA quadruplexes (GQs) [
15]. GQs and related secondary DNA structures play important physiological roles in controlling telomere association, recombination and replication, or in regulating transcription [
17]. These structures are present in nuclear DNA, but can also be found throughout the mitochondrial DNA (mtDNA), where they can contribute to the regulation of mitochondrial gene expression [
18], and thus cellular metabolism and respiratory functions [
19].
Considering the smooth reactivity of the aforementioned ruthenium complexes with DNA secondary structures, and given that their lipophilic and positively charged nature may facilitate a mitochondrial accumulation [
20‐
22], we questioned whether they could target the mtDNA of CSCs and thereby affect OXPHOS. This might eventually translate into compelling anticancer effects.
Using pancreatic CSCs (PaCSCs) as a model system, we herein demonstrate that ruthenium complexes of type ([Ru(terpy)(bpy)X]
n+) present a remarkable ability to reduce the self-renewal, invasive and tumorigenic capacity of PaCSCs by shutting down the transcription of their mtDNA protein-encoding genes and compromising their OXPHOS-dependent respiration. In contrast to other reported bioactive metal complexes [
21,
22], our compounds are not cytotoxic, and do not induce ROS or apoptosis. Our current data suggest that the biological effect is associated to a metalating interaction of the ruthenium complexes with specific guanines present in the D-loop region of the mtDNA, the regulatory region for mtDNA replication and transcription. More importantly, the ruthenium complexes exhibited an impressive effect to halt and even reduce tumor growth of pancreatic and colon cancer patient-derived xenografts (PDXs) in pre-clinical
in vivo models. We also present preliminary data which demonstrate activity for the treatment against an osteosarcoma (OS) PDX.
In brief, we have unveiled a new anti-cancer approach based on targeting key mitochondrial functions of CSCs, and validated the preclinical potential of designed ruthenium complexes in three different types of tumor entities.
Methods
Synthesis of ruthenium complexes – General
Chemical synthesis procedures, detailed protocols and characterization of all the compounds are described below (see also Fig. S
1 and Fig.
5). NMR and high-resolution electrospray ionization mass spectrometry (HR-ESI-MS) analysis was performed on all the synthesized compounds used in this article.
Synthesis of the parent ruthenium chloride [Ru(terpy)(bpy)Cl]Cl (Ru0)
RuCl3 • 3 H2O (2 g, 7.65 mmol) and 2,2’:6’,2’’ terpyridine (1.78 g, 1 equiv) were dissolved in EtOH:H2O 1:1 (80 mL), and the mixture was heated under reflux over 4 h in the dark. The resulting precipitate was washed with EtOH (x 3) and Et2O (x 1) to give Ru(terpy)Cl3 in a 75% yield (2.53 g, 5.74 mmol). The brown solid was used directly in a second step. Ru(terpy)Cl3 (550 mg, 1.25 mmol), 2,2’-bipyridine (195.2 mg, 1 equiv) and NEt3 (0.52 mL) were dissolved in EtOH:H2O 3:1 (120 mL) and the resulting solution was heated under reflux for 4h. The reaction mixture was filtered, and the solvents removed under reduced pressure to near dryness (~ 30 mL). The solution was refrigerated for 48h and the resulting precipitate was collected and washed with Et2O (x 1) to give [Ru(terpy)(bpy)Cl]Cl salt (Ru0) as a brownish red powder in a 60% yield (421 mg, 0.75 mmol).
Synthesis of the ruthenium aquo complex Ru1
Aqueous solutions of complex [Ru(terpy)(bpy)Cl]Cl salt (
Ru0) (either 1 or 5mM) were irradiated with blue LED light (455nm, 40-50W) for 1-2 hours to give the aquo derivative complex [Ru(terpy)(bpy)(H
2O)]Cl
2 (
Ru1). Concentration of the aqueous solutions of complex [Ru(terpy)(bpy)(H
2O)]Cl
2 were calculated by using UV-VIS absorption measures [λ
max (477 nm) ε = 9600] [
23].
Cell lines, primary human PDAC cells, patient samples
PDAC PDXs were obtained from Dr. Manuel Hidalgo under a Material Transfer Agreement with the Spanish National Cancer Centre (CNIO), Madrid, Spain (Reference no. I409181220BSMH). All PDAC PDX tumors contained G12D mutations in KRAS as determined by PCR sequencing as described in [
24]. To establish low-passage primary PDX-derived
in vitro cultures, PDX tumors were minced, enzymatically digested with collagenase (Stem Cell Technologies) for 60 min at 37°C, clarified via multiple rounds of filter purification with 100µm and 40µm Fisherbrand™ Sterile Cell Strainers (FisherScientific, Cat no. 11517532 and 11587522), and after centrifugation for 5 min at 1800 rpm, the cell pellets were resuspended and cultured in RPMI (Invitrogen) supplemented with 10% FBS (Invitrogen), 50 units/ml penicillin/streptomycin and fungizone (Invitrogen). PDX-derived cultures are referred to by a random number designation (e.g., Panc185, PancA6L, Panc215, Panc253, Panc265 or Panc354). Primary cultures were tested for Mycoplasma at least every 4 weeks.
CRC01 and OS170921 were obtained via the Hospital Ramón y Cajal-IRYCIS BioBank (PT13/0010/0002), integrated in the Spanish National Biobanks Network, under the RG-BIOB-54 Transfer Requests nº208 and nº198 and MTAs AC179 and AC168, and processed following standard operating procedures with the appropriate approval of the Ethical and Scientific Committees (Dictum 140/22 and 280/22), with informed consent and according to Declaration of Helsinki principles. CRC01 and OS170921 were subcutaneously implanted in immunocompromised female 6-week-old NU-Foxn1nu nude mice (Janvier, France) and passaged in vivo to establish PDX CRC01 and PDX OS170921.
Cellular toxicity assay
Ru1-mediated cellular toxicity was determined using the Toxilight BioAssay kit (Lonza, Walkersville, MD) according to the manufacturer's instructions.
Pancreatic CSC spheres were generated by culturing primary pancreatic cancer cells (5,000-20,000 cells/ml) in ultra-low attachment plates (Corning) using serum-free DMEM/F12 (Invitrogen) supplemented with B27 1:50 (Invitrogen), 20ng/mL bFGF (PAN-Biotech) and 50 U/mL penicillin/streptomycin (Thermo Fisher Scientific). Seven days later, spheres were harvested for subsequent assays or counted with an inverted EVOS FL microscope (Thermo Fisher Scientific) using a 10X objective with phase contrast. For serial passaging, spheres were harvested using a 40µm cell strainer (Fisher), trypsinized into single cells and re-cultured for another 7 days. Sphere counts are represented as number (no.) of spheres/ml or the fold change in spheres no./ml.
Colony assay
For colony formation assays, 500 cells were seeded in 24-well plates. Ru1 was added 24-48 h post seeding. Cells were cultured in RPMI 1640 containing 10% FBS at 37°C, 5% CO2. After 10-12 days, cells were fixed with PFA 4% (Paraformaldehyde, 16% w/v aq. soln., methanol free, Alfa Aesar™, Cat no. 11400580) for 10 min, washed with PBS and stained with Crystal violet (Sigma, Cat no. C3886-100G) for 1 h. Wells were digitalized and colonies/total area were quantified by lysing stained colonies in 1XPBS with 1%SDS followed by colorimetric absorbance analysis using a Synergy™ HT Multi-Mode Microplate Reader (BioTek, Winooski, Vermont, USA).
Flow cytometry and FACS
Cells or digested tumors (as described above) were resuspended in Flow buffer [1X PBS; 3% FBS (v/v); 3mM EDTA (v/v)] before analysis with a 4-laser Attune NxT Acoustic Cytometer (Thermo Fisher Scientific). For cell surface marker expression, refer to antibodies listed in Supplementary Table S
1. For Annexin-V staining, floating and attached cells were pooled and resuspended in 1X Annexin-V staining buffer containing Annexin-V-FITC diluted 1:20 (Biotium, Freemont, CA) and incubated for 20 min at room temperature prior to flow cytometric analysis. For autofluorescent detection, cells were excited with blue laser 488nm and selected as intersection with emission filters 530/40 (BL1) and 580/30 (BL2) or, in case of sorting, emission filter for FITC. For cell sorting, a FACS Vantage SE Flow Cytometer was used and data analyzed with BD FACSDiVa software.
For mitochondrial membrane potential measurement, CMX-ROS (M7512, Invitrogen), CM-H
2XRos (M7513, Invitrogen) or Mitoblue [
25,
26] were used. Probes were incubated with cells for 20 min at 37°C at a concentration of 10nM, 100nM or 10µM, respectively and fluorescence was detected using the filters (Ex561nm/Em585/16) YL1 for CMX-ROS, (Ex561nm/Em620/15) YL2 for CM-H
2XRos or (Ex405nm/Em512/25) VL2 for Mitoblue. For mitochondrial mass, 10-N-Nonyl acridine orange (NAO, A7847, Sigma Aldrich) was used at 0.1µM for 20 min at 37°C, and fluorescence was detected using the filters (Ex488nm/Em530/30) BL1. For ROS production measurement, MitoSOX (M36008, Invitrogen) was used at 1µM for 10 min at 37°C and detected with laser (Ex561nm/Em585/16) YL1.
For all assays, 2mg/ml DAPI (Sigma) or 2µl/ml 7-Amino-Actinomycin D (7-AAD, BD, Cat no. 51-68981E) was used to exclude dead cells, and fluorescence was detected using the filters (Ex405nm/Em440/50) VL1 or (Ex561nm/Em695/40) YL3, respectively. Data were analyzed with FlowJo 9.3 software (Tree Star Inc., Ashland, OR.).
Zebrafish maintenance and xenograft assays
Zebrafish embryos were obtained by mating adult zebrafish (
Danio rerio, wild-type), maintained in 30L tanks with a ratio of 1 fish per liter of water, with 14h/10h light/dark cycle and a temperature of 28.5°C according to published procedures [
27]. All the procedures used in the experiments as well as fish care were performed in agreement with the Animal Care and Use Committee of the University of Santiago de Compostela and the standard protocols of Spain (Directive 2012-63-UE). At the final point of the experiments, zebrafish embryos were euthanized by tricaine overdose.
For zebrafish xenograft assays and image analyses, zebrafish embryos were collected at 0 h post-fertilization (hpf) and incubated until 48 hpf at 28.5°C. At 48 hpf, hatched embryos were anesthetized with 0.003% of tricaine (Sigma). mCherry-H2B-labelled Panc185 cells were treated with Ru1 (100µM) for 24 h, trypsinized, resuspended and concentrated in an eppendorf at 106 cells per tube for each condition. Cells were then resuspended in 10µL of PBS with 2% PVP (Polyvinylpyrrolidone) to avoid cellular aggregation. Borosilicate needles (1mm O.D. x 0.75mm I.D.; World Precision Instruments) were used to perform the xenograft assays in the zebrafish embryos. Between 100 and 150 cells were injected into the circulation of each fish (Duct of Cuvier) using a microinjector (IM-31 Electric Microinjector, Narishige) with an output pressure of 34 kPA and 30 ms of injection time per injection. Subsequently, the injected embryos were incubated at a temperature of 34°C for 6 dpi in 30ml Petri dishes for each condition with SDTW (salt dechlorinate tap water). Imaging of the injected embryos was performed using a fluorescence stereomicroscope (AZ-100, Nikon) at 1 and/or 6 dpi in order to measure the proliferation, migration and invasion of the Panc185 injected human cancer cells inside the zebrafish circulation in each of the conditions assayed.
The image analysis of the injected embryos was carried out using Quantifish software v2.1 (University College London, London, UK) in order to obtain the proliferation ratio of the cells in the region of the caudal hematopoietic tissue (CHT) of the embryos, where the cells proliferate and metastasize. This program measures in each of the images provided the intensity of the fluorescence and the area of the positive pixel above a certain threshold of the cells. With these parameters, an integrated density value is obtained allowing one to compare different times between images to reach a proliferation ratio.
In vivo toxicity and tumorigenicity assays
All mice were housed according to institutional guidelines and all experimental procedures were performed in compliance with the institutional guidelines for the welfare of experimental animals approved by the Universidad Autónoma de Madrid Ethics Committee (CEI 60-1057-A068 and CEI 103-1958-A337) and La Comunidad de Madrid (PROEX 335/14 and 294/19) and in accordance with the guidelines for Ethical Conduct in the Care and Use of Animals as stated in The International Guiding Principles for Biomedical Research involving Animals, developed by the Council for International Organizations of Medical Sciences (CIOMS). Briefly, mice were housed according to the following guidelines: a 12 h light/12 h dark cycle, with no access during the dark cycle; temperatures of 65-75°F (~18-23°C) with 40-60% humidity; a standard diet with fat content ranging from 4 to 11%; sterilized water was accessible at all times; for handling, mice were manipulated gently and as little as possible; noises, vibrations and odors were minimized to prevent stress and decreased breeding performance; and enrichment was always used per the facility’s guidelines to help alleviate stress and improve breeding.
For toxicity and preliminary pharmacokinetics (PK) analyses, 10-week-old CD-1 mice (Janvier, France) of approximately 25-30g were treated with Ru1 via two routes of administration: 1) oral gavage (o.g., 100µl) or 2) retro-orbital (r.o.) injection (100µl). Ru1 was resuspended in physiological saline (0.45% NaCl) for r.o. injections or in H2O for o.g., to a concentration of approximately 0.5mM, such that mice were treated daily with a dose of Ru1 equivalent to 1.4mg/kg. At indicated time post treatment initiation, mice were weighed. Six and 24h post r.o. injection, and on day 28 (o.g.) or 29 (r.o.), mice were sacrificed, weighed, blood was collected in EDTA tubes (Aquisel, Cat no. 107545) for hematocrit analysis (Element HT5, Veterinary Hematology Analyzer, scil animal care company GmbH, Madrid, Spain), and organs were excised and weighed, photographed, fixed in 4% PFA and processed for histological analysis or analyzed by inductively coupled plasma mass spectrometry (ICP-MS), as described below. For preliminary PK analyses, a second group of CD-1 mice were injected r.o. with Ru1 (0.14 mg/kg) and at the indicated time points, blood was collected in EDTA tubes (Aquisel, Cat no. 107545) and analyzed by ICP-MS, as described below.
Indirect calorimetry analyses were carried out using a 16-chamber TSE PhenoMaster monitoring system (TSE Systems GmbH, Bad Homburg, Germany). Full access to food and water was continuously available, and their intake was monitored using built-in devices located within each cage. Calorimetry measurements were carried out during a period of 72 h, according to animal weight, to exclude changes in body weight that would contribute to differences in energy expenditure measurements [
28]. Seven days prior to introducing mice into the PhenoMaster monitoring system, 10-week-old C57Bl6 mice (Janvier, France) were subcutaneously implanted (in their back), with Micro-Osmotic Pumps (Azlet® model 1002, which release 0.25µl/hour over the course of 14 days) containing 100µl of 5mM
Ru1 or physiological saline (i.e., Sham). Mice were introduced into individual chambers and were on a 12-hour light-dark cycle (lights on at 07:00am) during the course of the experiment, with a maintained room temperature of 22 ± 2˚C. Oxygen consumption and CO2 release was measured. From these values, respiratory exchange ratio (RER) was determined as VCO2/VO2 and energy expenditure (EE) was calculated as = (3.185+ 1.232 x RER) x VO2.
For
in vivo tumor growth and Limiting Dilution Analysis (LDA) assays with PDAC cells, female 6- to 8-week-old NU-Foxn1nu nude mice (Janvier, France) were injected subcutaneously with dilutions of
Ru1-treated (100µM for 24h) or untreated PDAC cells in 50µl Matrigel (Corning) per injection. Tumor growth was monitored bi-weekly for up to 4 months. Mice were sacrificed and tumors were weighed, photographed, and part of each tumor was fixed in 4% PFA and processed for histological analysis. CSC frequencies were calculated using the ELDA software
https://bioinf.wehi.edu.au/software/elda/.
For PDX in vivo treatment experiments, tumors were initially established by subcutaneously implanting (with Matrigel (Corning)) tumor pieces of the indicated PDXs in the right and left flanks of 6- to 8-week-old NU-Foxn1nu nude mice (Janvier, France). 4-5 weeks post implantation, tumors were excised, cut into identical pieces of approximately 50mm3 and implanted (with Matrigel (Corning)) subcutaneously into the left and right flanks of 6- to 8-week-old NU-Foxn1nu nude mice (Janvier, France). Three weeks later, tumors were measured to ensure volumes of 125-150 mm3, mice were weighed to calculate treatment concentrations per Kg, randomized into treatment groups (5-6 mice per group) and treatments were initiated for approx. three consecutive weeks. Ru1 was resuspended in physiological saline (0.45% NaCl) to a concentration of approximately 0.5mM such that mice were treated with a volume of Ru1 equivalent to 1.4mg/kg. Initially three routes of administration for Ru1 were tested: 1) orally (100µl daily), 2) via retro-orbital injection (100µl daily) or 3) subcutaneously into the tumor (100µl twice per week). Gemcitabine (Accord Healthcare, S.L.U.) was administered twice a week (50 mg/kg i.p.) and 5FU (Sigma) was administered twice a week (30 mg/kg i.p.). Tumor volumes were determined twice per week by caliper measurements. At the time of sacrifice, mice were weighed, blood was collected in EDTA tubes (Aquisel, Cat no. 107545) and tumors and organs were excised and weighed, photographed, fixed in 4% PFA and processed for histological analysis or analyzed by inductively coupled plasma mass spectrometry (ICP-MS), as described below.
RNA sequencing analysis
Total RNA was isolated by the guanidine thiocyanate (GTC; VWR AMRESCO Chemicals, Cat no. K965-250ML) method using standard protocols [
29]. PolyA+ RNA fraction was processed as in Illumina’s ‘‘TruSeq RNA Sample Preparation v2 Protocol’’. The resulting purified cDNA library was applied to an Illumina flow cell for cluster generation (TruSeq cluster generation kit v5) and sequenced on the Genome Analyzer IIx with SBS TruSeq v5 reagents by following manufacturer’s protocols. RNA-seq data sets were analyzed using the tool Nextpresso [
30]. Nextpresso is comprised of four basic levels: 1. Quality check, 2. Read cleaning and/or down-sampling, 3. Alignment, and 4. Analysis (gene / isoform expression quantification, differential expression, gene set enrichment analysis and fusion prediction. Gene signatures (Hallmark gene sets) were downloaded from GSEA - Molecular Signature Database for Gene set enrichment analysis. Data deposited in the NCBI SRA database (Accession: PRJNA832709).
RNA Preparation and Real-Time PCR
Total RNA from human PDX-derived cell lines, PDX tumors or mouse organs was isolated by the GTC method using standard protocols [
29]. One microgram of purified RNA was used for cDNA synthesis using the Thermo Scientific Maxima First Strand cDNA Synthesis Kit (ThermoFisher Scientific) according to manufacturer´s instructions, followed by SYBR green RTqPCR (PowerUp™ SYBR™ Green Master Mix, ThermoFisher Scientific) using an Applied Biosystems StepOnePlus™ real-time thermocycler (ThermoFisher Scientific). Thermal cycling consisted of an initial 10 min denaturation step at 95°C followed by 40 cycles of denaturation (15 sec at 95°C) and annealing/extension (1 min at 60°C). mRNA copy numbers were determined relative to standard curves comprised of serial dilutions of plasmids containing the target coding sequences and normalized to ß-actin levels. Primers used are listed in Supplementary Table S
2.
Probabilistic graphical models
From FPKM data from PDX models, control or treated with
Ru1 or
Ru1-met, the 2,000 most variable genes were selected. These genes were used to build a probabilistic graphical model without other a priori information based on correlation as associative measurement. Probabilistic graphical model was constructed using R v3.2.5 and
grapHD package [
31]. The result is an undirected graph with local minimum Bayesian Information Criterion (BIC) based on the subsequent steps: first, the spanning tree with the maximum likehood is found and, second, the graph is customized by the adding of edges that reduce BIC and preserve the decomposability [
32]. The resulting network was analyzed searching for a functional structure as previously described [
33]. Gene ontology analyses were performed using DAVID webtool [
34], selecting “homo sapiens” as background and KEGG, Biocarta and GOTERM-FAT as categories. Functional node activities were calculated as the mean expression of the genes included in one functional node related to its overrepresented function. Differences between conditions were assessed by a non-parametric Kruskal-Wallis test using Graph Pad v6.
Flux Balance Analysis (FBA) is a method to model metabolic networks and to estimate tumor growth [
35]. For FBA, the whole human metabolic reconstruction Recon3D [
36] and COBRA Toolbox library [
37] were used. Recon3D contains information about 10,600 metabolic reactions, 5,835 metabolites and 5,939 Gene-Protein-Reaction rules (GPRs) which contain information about what genes are involved in each metabolic reaction as Boolean expressions. GPRs were solved as described in previous studies [
38,
39] using a modification of Barker et al. algorithm [
40] and incorporated into the model using a modified E-flux algorithm [
39,
41]. Briefly, “OR” operators were solved as the sum and “AND” operators were solved as the minimum. Then, GPRs were normalized using a Max-min function to an interval [0,1] and introduced into the model as the reaction bounds. As objective function the biomass reaction included in the Recon3D was used, as representative of tumor growth. The 10,600 reactions are grouped into 103 metabolic pathways. The mathematical problem was solved using linear programming. To compare metabolic activity between conditions, flux activities were calculated as the sum of fluxes of the reactions included in a concrete metabolic pathway defined in Recon3D. Then, a delta was calculated subtracting
Ru1 to control flux activity for each metabolic pathway.
Oxygen Consumption Rate (OCR) measurements
Sphere-derived Panc185, PancA6L and Panc215 cells were plated in XF HS Miniplates (Seahorse Bioscience) at a cellular density of 5,000 cells/well. For OCR determination, cells were incubated in Seahorse XF DMEM media (103680, Agilent) supplemented with 2mM glutamine, 10mM glucose, and 1mM pyruvate for 1 h, prior to the measurements using the Seahorse XFp Cell Mito Stress Kit (103010, Agilent). After an OCR baseline measurement, the minimum oxygen consumption was determined adding 1.5µM oligomycin (O) and the maximal respiration rate was assessed by adding 1µM FCCP (F). At the end of the experiment the non-mitochondrial oxygen consumption was evaluated adding both 0.5µM rotenone (R) and antimycin (A). Experiments were run in a XF HS Mini analyzer (Seahorse Agilent), and raw data were normalized to total protein using BCA protein assay kit (Cat. no. 23225, Thermo Scientific).
Lactate production assay
Supernatant from indicated cells was collected to evaluate the changes in the levels of lactate production. The analysis was performed using the Lactate Assay Kit (Sigma-Aldrich, St. Louis, Missouri, USA) according to the manufacturer’s instructions. The optical density was determined using a Synergy™ HT Multi-Mode Microplate Reader (BioTek, Winooski, Vermont, USA) at a wavelength set to 570nm. Data were normalized to total protein using the BCA protein assay kit (Thermo Scientific).
ATP determination assay
Lysate pellets of cells from control and treated cells were collected to evaluate the changes in the levels of ATP. The analysis was performed using the ATP Bioluminiscense Assay Kit CLS II (Cat. no. 11699695001, Roche) according to the manufacturer’s instructions. Bioluminiscence was determined using a Synergy™ HT Multi-Mode Microplate Reader (BioTek, Winooski, Vermont, USA). Data were normalized to total protein using the BCA protein assay kit (Thermo Scientific).
Immunostainings and confocal analysis
For fluorescence confocal microscopy, indicated cells were seeded in 8-well IbiTreat (ibidi chamber slides, Cat no. 181009/1) in RPMI (Gibco) containing 10% FBS (Thermo Fisher Scientific) at 37°C, 5% CO2. After indicated treatments with Ru1 or Ru-TMR and indicated time points, the medium was removed, and cells were stained with Mitotracker Green (MTR-G, M7514, Invitrogen) at a final concentration of 20nM in serum-free RPMI for 30 min at 37°C, washed with PBS, and then overlaid with fresh RPMI containing 10% FBS. The fluorescent images were collected immediately afterwards with a laser scanning confocal microscope Zeiss 710 40X Apochromatic and analyzed using the software ZEN 2009. For fluorescence confocal microscopy of Tetramethylrhodamine, ethyl ester, perchlorate (TMRE) and CellROX DeepRed, indicated cells growing in glass-bottom cell culture plates were treated with 100µM Ru1 or Ru1-met. After 24 hours of incubation, cells were washed twice with RPMI containing 10% FBS. 1µM TMRE (Cat no. T669, ThermoFisher Scientific) or 10 µM CellROX-DR (Cat no. C10422, ThermoFisher Scientific) reagent in RPMI containing 10% FBS were added for 20 or 30 min, respectively. Then, two new washes with RPMI containing 10% FBS were performed, and cells were observed in an Andor Dragonfly Spinning Disc confocal system attached to a Nikon Eclipse TiE using a 60X apochromatic objective and adequate filter settings.
Electron microscopy analysis
PaCSC-enriched spheres were trypsinized and centrifuged for 5 min at 400×g. Cell pellets were fixed using a solution of 2.5% glutaraldehyde in cacodylate buffer 0.1M for 60 min. Cell pellets were post-fixed in osmium tetroxide, dehydrated through ascending concentrations of ethanol and embedded in epoxy resin. Ultra-thin sections were obtained at 0.1μm, counterstained with uranyl acetate and lead citrate prior to image acquisition with a JEOL JEM1010 (100 kV) transmission electron microscope equipped with a Gatan Orius 200 SC camera. Images were processed using DigitalMicrograph (Gatan, Inc).
Western blot analysis
Cells were harvested in RIPA buffer (Sigma-Aldrich) supplemented with a protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN). Fifty micrograms of protein were resolved by SDS-PAGE and transferred to PVDF membranes (Amersham Pharmacia, Piscataway, NJ). Membranes were sequentially blocked with 1X TBS containing 5% BSA (w/v) and 0.5% Tween20 (v/v), incubated with a 1:500-1:1000 dilution of indicated antibodies (see Supplementary Table S
1) overnight at 4ºC, washed 5 times with 1X TBS containing 0.5% Tween20 (v/v), incubated with horseradish peroxidase-conjugated goat anti-rabbit or goat anti-mouse antibody (Amersham), and washed again to remove unbound antibody. Bound antibody complexes were detected with SuperSignal chemiluminescent substrate (Amersham) and images were obtained using MyECL Imager (Thermo Fisher Scientific). Densitometry histograms were obtained by measuring the intensity of the bands and normalized by their housekeeping loading control by ImageJ software. Blots are accompanied by the locations of molecular weight/size markers (M
r(K)), as determined using commercially available protein ladders (Novex®Sharp Pre-Stained Protein Ladder Cat no. LC5800 or PageRuler™ Prestained Protein Ladders Cat nos. 26616 or 26619, all from ThermoFisher Scientific).
Mitochondrial gradient purification
Enrichment of mitochondria prior to density gradient purification was performed following the protocol by Fernández-Vizarra
et al., [
42]. Briefly, 8×10
7 cells (untreated or treated for 24 h with 100µM
Ru1) were mechanically broken in IB buffer [35mM Tris-HCL pH 7.8; 5mM MgCl
2 (v/v); 25mM NaCl (v/v)] prior to density gradient isolation following the protocol described by Frezza C.
et al., [
43]. Gradient isolated mitochondria were then resuspended in IBC buffer [0.1M Tris-MOPS; 2mM MgCl
2 (v/v); 0.2M Sucrose (v/v)] pH 7.4.
Three downstream analyses were performed with purified mitochondria. 1) To determine the amount of Ru1 in purified mitochondria isolated from control- and Ru1-treated cells, density gradient-purified mitochondria were diluted with Nitric Acid (HNO3) to a final concentration of 60% for ICP-MS analysis, as described below. 2) To determine the capacity of Ru1 to enter directly into mitochondria and interact with mtDNA, density gradient-purified mitochondria from untreated cells were incubated for 2 h with 100µM Ru1, 100µM Ru1-met or an equal volume of H20 (diluent), all in IBC buffer. Before incubation, 0.1µg/µl of DNAse (Cat no. DN25, Sigma-Aldrich) was added to the purified mitochondria to avoid binding of Ru compounds to mtDNA from broken mitochondria. After incubation of purified mitochondria with Ru compounds, mitochondria were resuspended with an equal volume of saturated phenol (pH 8) for subsequent DNA extraction and ICP-MS analysis, both as described below. 3) To determine the capacity of Ru1 to directly interact with mtDNA, phenol-chloroform-extracted mtDNA from density gradient-purified mitochondria isolated from untreated cells were treated for 1h with diluent or 10µM of Ru1 or Ru1-met prior to PCR, as described below.
DNA extraction and PCR
DNA was isolated using standard phenol-chloroform extraction methods. For PCR amplification of the D-loop or
RNR2 regions of the mtDNA, 0.1 or 0.01ng of untreated or treated purified mtDNA was used as a template with primers specific to the two aforementioned regions (Supplementary Table S
3). Thermal cycling, using a SimpliAmp (ThermoFisher Scientific), consisted of an initial 7 min denaturation step at 95°C, 35 cycles of denaturation (30 sec at 95°C), annealing/extension (1 min at 54°C) and elongation (2 min at 72°C), and a final elongation of 10 min at 72°C. PCR products were resolved on a 1% Agarose/TBE gel for 1 h at 100V, and size verification was performed comparing SybrSafe-stained experimental bands to molecular weight bands of the 1kb Plus DNA Ladder (Cat no. 12308-011 Invitrogen, ThermoFisher Scientific).
Inductively coupled plasma mass spectrometry (ICP-MS)
To quantify the presence of ruthenium, mtDNA extracted from untreated or Ru1-treated mitochondria or cell pellets were dissolved in 100μl of 65% nitric acid in water and then analyzed by ICP-MS. For ICP-MS analysis of extracted PDX354 tumors and/or organs (liver, kidney and brain), weighed tissue samples were cut into small pieces and homogenized in a 15mL falcon tube. Sixty-five percent nitric acid was added to completely cover the tissue, and the mixture was digested overnight. After digestion, water was added to 3 mL, and the solution was centrifuged at 3000-4000 X g at 4°C for 10 min. The pellet was discarded, and the supernatant transferred to an Eppendorf tube for ICP-MS analysis. For serum samples, approximately 30-50µl of serum was dissolved in 65% nitric acid, and water was added to 3 mL for ICP-MS analysis.
ICP-Mass was performed at the CACTUS-Campus Lugo facility of the University of Santiago de Compostela using an ICP-MS Agilent 7700x with a Peltier (2°C) cooled sample introduction system based on a glass low-flow MicroMist Nebulizer and a quartz torch double pass spray chamber for aerosol filtering. The Ru calibration standards were prepared from a 1g/L commercial standard (Merck). Ir (Merck) was used as an internal standard.
Statistical analyses
Results are presented as means ± standard error of the mean (sem) unless stated otherwise. Pair-wise multiple comparisons were performed with one-way ANOVA (two-sided) with Bonferroni or Dunnett adjustment, as indicated in the figure legends. Student’s t-test were used to determine differences between means of groups. P values <0.05 were considered statistically significant. All analyses were performed using GraphPad Prism version 6.0c (San Diego California USA).
Discussion
Cis-platinum and derivatives are reactive metal complexes that have shown impressive utility as anti-cancer chemotherapeutic agents by inducing cancer cell apoptosis. However, these compounds present a promiscuous reactivity, and hence elicit many secondary toxic and resistance effects [
55]. Attempts have been made to develop related anti-cancer metal complexes with better selectivity than platinum derivatives, and in this context, ruthenium has been especially attractive owing to the ligand tuning possibilities and accessible coordination geometries. In fact, two ruthenium complexes (NAMI-A and NKP1339) have even entered clinical trials [
56], although apparently, they work by targeting proteins and/or altering the cellular redox state rather than by interacting with DNA. However, as with most metallodrugs, these ruthenium complexes are also quite promiscuous in terms of reactivity, which makes it difficult to control their biological fate and targeting profile [
57,
58].
Therefore, a major challenge in the field has been the discovery of metallo-derivatives with kinetically controlled reactivity and increased selectivity with regard to their biological targets. In this context, we recently found that the ruthenium complex
Ru1 is capable of metalating solvent exposed guanine residues, such as those present in adjacent positions of GQs, with high selectivity and low toxicity [
15]. This curious combination of controlled reactivity with DNA and lack of cytotoxicity, prompted us to explore its potential biological applications. We have now discovered that
Ru1 exhibits a potent inhibitor effect on PaCSCs, by targeting genes involved in OXPHOS. More importantly, the compound exerts impressive anticancer activity in vivo. Preclinical evaluation of
Ru1 in 6 different subcutaneous PDX models of PDAC, including an orthotopic PDAC tumor model, as well as CRC and OS PDXs, showed potent cytostatic activity, inhibiting tumor proliferation as early as 1-2 days post treatment initiation. This effect is comparable to what others have accomplished using toxic combination therapies (e.g., inhibitors that target upstream (SHP2 or SOS1) and downstream (MEK) mediators of KRAS signaling [
59,
60]), but without the toxic or resistance-associated side-effects. Moreover, while an additive effect was observed for some tumors when
Ru1 was combined with gemcitabine, reduced tumor re-growth was observed for all PDAC tumors treated with the combination approach compared to gemcitabine alone, which we attribute to a reduction in the non-CSC population as well as the CSC population, the latter being the drivers of disease relapse.
Regarding the mechanism of action, control experiments with
Ru1-met or
Ru1-py (analogs of
Ru1 that lack a kinetically labile coordination position) revealed that these compounds are inert under the same conditions, suggesting that the activity of
Ru1 is mediated by displacement of the labile aquo ligand by some nucleophilic component of a biological molecule, most probably nucleic acid guanines. Indeed, detailed mechanistic experiments with PaCSCs revealed that
Ru1 can reach the mitochondria and interact with their mtDNA. We were able to map this interaction to the D-loop region, an area of the mtDNA that contains the main regulatory sites for transcription initiation [
54]. Consequently, RNAseq analysis of PancA6L CSCs treated with
Ru1 showed modulation of only mtDNA encoded transcripts, suggesting that the functional effect of
Ru1 is indeed mitochondriotropic. Nonetheless, we cannot completely rule out that
Ru1 could interact with other regions of the mtDNA and/or nuclear DNA, although confocal microscopy analysis of PaCSCs treated with Ru-TMR did not show signal in the nuclei (Fig.
9A). Along these lines, Panc185 cells showed modulation of more genes compared to PancA6L in our RNAseq analyses, including nuclear genes that encode OXPHOS components (e.g.,
COX5), but only at concentrations of 100µM (Fig.
8G). These differences between Panc185 and PancA6L may be due to differences in the amount of
Ru1 that enters each cell line, with Panc185 up taking more
Ru1 over time (Fig.
9B). Thus, while we cannot exclude other mechanisms of action contributing to the biological effects observed in this study, it is clear that
Ru1 reduces the mRNA of all 13 mtDNA protein-encoding genes, which provokes a decrease in oxygen consumption, mitochondrial membrane potential, and ATP production, as well as a decrease on the members of the OXPHOS complex, all of which are necessary for PaCSCs, which depend on mitochondrial respiration to meet their energy requirements and are therefore more susceptible to mitochondriotropics compared to non-CSCs [
12,
61]. Examples of other inorganic complexes that work as mitochondriotropics have been described [
21,
62‐
64]; however, their anti-cancer activity is associated to mitochondrial-induced apoptosis, very different than that of
Ru1, which at the concentrations used in this study do not induce apoptosis (Fig.
1H) or increase ROS (Fig.
8D). Organic compounds, such as the benzene-1,4-disulfonamide compound 23 (DX3–213B), have shown promising results at the level of tumor growth inhibition in a PDAC syngeneic in vivo model, by disrupting ATP generation; however, its mechanism of action has not been elucidated, but most likely it is not mediated by inhibiting CSCs [
65].
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.