Background
Bone metastases occur exceptionally often in cancers derived from the breast, the prostate and the kidney. The biological process of metastasis requires the capability of extravasation, migration and homing to the bone microenvironment. Tumor cells when metastasized to bone are able to activate osteoclasts by secreting osteoclast promoting factors [
1]. The latter is the basis of the classical concept of osteolytic metastasis, while the local secretion and/or activation of latent growth factors contribute to the development of osteoblastic metastases and autocrine tumor propagation [
2].
Anti-resorptive bisphosphonates (BP) accumulate in bone as they show a high affinity to hydroxylapatite and are incorporated by osteoclasts via phagocytosis [
3]. First generation BP like clodronate induce apoptosis by accumulating toxic ATP adducts whereas second generation amino-BP inhibit the mevalonate pathway enzyme farnesyl pyrophosphate synthase (FPPS) very specifically. As a consequence protein prenylation of small GTP binding proteins like Rab, Ras or lamins, which are important for cytoskeleton organization and cellular polarization, is inhibited and may initiate apoptosis [
4]. Additionally it was reported for zoledronic acid (ZA) and to a lesser extend for ibandronate (IBN) and risedronate (RIS) as well as for alendronate (ALN) that treatment of cells led to the accumulation of isopentenyl pyrophosphate (IPP) and produced a new endogenous ATP analogue (triphosphoric acid 1-adenosin-5′-yl ester 3-(3-methylbut-3-enyl) ester (ApppI)), which also caused apoptosis in osteoclasts by inhibiting the mitochondrial ADP/ATP translocase [
5].
BP have been developed for osteoporosis treatment where numerous clinical studies proved their efficacy in reducing the incidence of fragility fractures. When applied in higher cumulative doses than used for osteoporosis, BP effectively reduced the number of skeletal related events in patients with bone metastases [
6,
7], which has made them an important class of drugs in the treatment of osteolytic bone diseases [
8]. Besides the effects on their classical targets, cells of the myelomonocytic/macrophage lineage and especially osteoclasts, BP have been shown to induce apoptosis in a variety of benign and malignant cells, although in some cases μM concentrations were required [
3]. These
in vitro effects in concert with clinical studies have stimulated discussions about a putative clinically relevant anti-tumor effect of BP. Almost twenty years ago it was shown that adjuvant treatment with BP reduces the incidence of bone metastases and the overall mortality in patients suffering from breast cancer. These results were confirmed in the ABCSG-12 trial, where ZA was used only twice a year for the adjuvant treatment of estrogen receptor positive breast cancer patients. Positive long term effects from patients of the first cohort were reported in a second analysis more than ten years after the first publication [
9‐
11]. Moreover, a synergistic anticancer efficacy of ZA in combination with neoadjuvant chemotherapy was shown in breast cancer patients with respect to additional tumor shrinkage [
12]. These effects were confirmed by the ZO-FAST study, where ZA was associated with improved disease-free survival in postmenopausal women [
13]. However, the discussion is ongoing and presently a proven anti-tumor effect seems to be restricted to the postmenopausal high bone turnover subpopulation of women suffering from breast cancer [
14].
The detailed characterization of the molecular effects of modern BP like ZA stimulated research about their effects on both osteoblastic differentiation and on anti-tumor effects, but a prominent question remained to be solved, if local μM concentrations of BP can be achieved in the clinical setting [
15,
16]. Such high concentrations are needed because the cellular uptake is relatively poor in cells other than macrophages and osteoclasts as described for e.g. free ZA in ovarian tumor cells [
17]. However it was speculated that BP concentrations in the bone microenvironment and especially in the resorption lacuna can reach concentrations up to hundreds of μM [
18]. The two most prominent
in vitro effects of BP, which may add to their putative anti-tumor effects, are the capability of inducing apoptosis in tumor cells and eliciting an immune response. Stimulation of breast cancer cells with bisphosphonates and inhibition of the mevalonate pathway as a consequence leads to the accumulation of IPP and ApppI. IPP acts as phosphoantigen for γδT cells, which have the ability to attack the tumor cells [
19]. The mechanism by which IPP is secreted or transported to the outer surface of a cell is still unknown [
20,
21]. Channels and transporters for pyrophposphates or ATP might be responsible for mediating these effects and promising candidates are pannexin (PANX) hemichannels (especially PANX1), the progressive ankylosis protein homolog ANKH as well as organic anion transporters of the solute carrier family 22 (organic anion transporter SLC22A6, SLC22A8 and SLC22A11) and multidrug resistance associated protein 1 (ABCC1). For PANX1, which is a part of the purinergic receptor P2RX7 complex, participation in ATP release was shown [
22‐
24]. ANKH is a transmembrane protein and controls intra- and extracellular levels of pyrophosphate, which is important in bone mineralization [
25]. Solute carrier family 22 members are responsible for the transport of organic anions mainly in the kidney and liver [
26] whereas ABCC1, a member of the human ABC transporter family that is involved in multidrug resistance, mediates export of organic anions and drugs from the cytoplasm [
27]. All channels and transporters are sensitive to the anion transport blocker probenecid (Prob), whereas carbenoxolone (CBX) has no effect on ANKH but is effective in inhibiting PANX1 mediated release. Ibrutinib was described to block ABCC1 transport while novobiocin inhibits SLC22A6, 8 and 11 [
24,
28‐
31]. Therefore these substances can be used to distinguish between ANKH, PANX1, ABCC1 and SLC22A mediated effects.
Sustained effects of bisphosphonates on osteogenic differentiation upon treatment with low concentrations and intermittent treatment with high concentrations of ZA and alendronate were previously demonstrated [
32,
33], while permanent exposure to high doses induced apoptosis in both tumor cells and osteogenic precursors [
32,
34,
35]. In MCF-7 cells we identified ZA target genes as KLF2, KLF6 and Ki-67 and we assumed that IPP/ApppI accumulation might mediate this effect in cell populations that are largely insensitive to apoptosis induction [
15]. It is of major importance to unravel the differential potency of various BP on tumor cell growth and apoptosis and to describe the downstream targets in non-osteoclastic cells.
Here we show that breast cancer cell lines permanently exposed to various BP (zoledronic acid, ibandronate, alendronate, risedronate) undergo apoptosis (MDA-MB-231, to a lesser extend T47D) or show reduced viability (MCF-7). The relative potency of various BP mirrors their antiosteolytic potency with ZA inducing the greatest increase in apoptosis. Interestingly, all other BP tested were almost equally potent in reducing MCF-7 viability. Co-incubation with the anion transporter and channel blocking agent probenecid and novobiocin revealed a synergistic effect, which shows that accumulated pyrophosphates might be secreted to the extracellular space and according to previously described sensitivity renders SLC22A family members as good candidates for the sensitizing effects. Bisphosphonates have relevant effects on tumor cell biology and an adjuvant therapy with BP in combination with a respective sensitizer might be useful in the treatment of breast cancer.
Discussion
Apart from osteoclasts, BP may have clinically relevant effects on benign and malignant cells. We found variable efficacies of different BP on cell viability and caspase 3/7 activity of the breast cancer cell lines MDA-MB-231, T47D and MCF-7. The most potent BP in MDA-MB-231 cells with respect to caspase 3/7 activity induction was ZA, while other BP were markedly less effective in the descending order IBN > ALN > RIS when applied in equimolar concentrations. In the apoptosis insensitive cell lines the picture was different with ZA showing high efficacy on the reduction of cell viability in T47D cells followed by ALN, IBN and RIS in contrast to MCF-7 cells where ZA and ALN depicted comparable effects followed by the weaker compounds RIS and IBN. The observed differences cannot be explained by the rank order of BP in their potency to inhibit the target enzyme farnesyl pyrophosphate synthase (FPPS) with ZA and RIS depicting the highest potency followed by the much weaker inhibitors IBN and ALN [
4]. Differences in cellular BP uptake and retention might be responsible for these observations. Nothing is known if all BP are incorporated with the same efficacy, also the mechanism by which tumor cells take up BP is under discussion. The process of pinocytosis might be relevant but the transport through a channel protein cannot be excluded. At pH 7.4 the Amino-BP differ in their zeta potential as the R2 groups of ZA, ALN and IBN are positively charged in contrast to RIS, where the group is negatively charged [
4]. Analyses with nanoparticles revealed that positively charged particles are more likely engulfed by pinocytosis than negatively charged particles [
36] but also a channel protein or a transporter might distinguish between the different groups in favor of the positively charged BP. Both processes would lead to reduced RIS uptake possibly explaining the weak effects of this compound in tumor cells.
The determination of IPP accumulation and ApppI formation revealed differences between the analyzed breast cancer cell lines and the various BP. In T47D cells we detected high levels of IPP/ApppI and in MCF-7 cells high to moderate levels of IPP and low levels of ApppI as reported previously [
19]. In MDA-MB-231 cells IPP and ApppI were only measurable in single samples. ZA was the most potent BP in inducing IPP/ApppI followed by RIS and ALN and IBN being the weakest compound. Our data are not in line with observations in J774 macrophages where ApppI was highest after ZA treatment followed by RIS, IBN and ALN [
5], which is similar to their known order of affinity to FPPS and we again speculate that cells incapable of phagocytosis reflect mechanisms for BP uptake, which distinguish between differently charged BP.
Tumor cells are capable of releasing IPP to the extracellular space, which can bind to an unknown antigen-presenting molecule to be recognized by the T-cell receptor of γδT-cells [
20,
21]. The mechanisms by which IPP is secreted are unknown and we assumed that the pyrophosphate channels PANX1 and/or ANKH or organic anion transporters as ABCC1 and/or members of the organic anion transporter family SLC22A might mediate this release. All analyzed breast cancer cells depicted similar expression levels of PANX1 and ABCC1 whereas a considerable variability of ANKH and SLC22A11 expression was observed. At first our lead candidate was ANKH but by establishing ANKH transgenic T47D cells we were able to exclude its relevance. We further hypothesized that blocking the above mentioned channels and transporters and subsequently inhibiting the release of BP-induced pyrophosphates enhances IPP/ApppI accumulation, leading to an increase in the BP effect on tumor cell viability. Co-stimulation with the PANX1 inhibitor CBX or the ABCC1 inhibitor ibrutinib together with BP did not result in an appreciable synergistic effect in contrast to a co-stimulation with BP and the organic anion transporter and pyrophosphate channel blocking agent probenecid (Prob) or the SLC22A blocker novobiocin. Both probenecid and novobiocin revealed remarkable additive effects on BP-mediated cell viability reduction and caspase 3/7 activity induction in certain conditions. Therefore we hypothesize that solute carrier family 22 (organic anion transporter) members might be the main candidates to release IPP into the extracellular space. By blocking SLC22A members the described effects of BPs on tumor cells can be intensified.
Furthermore we tried to find out if the additive effect of Prob and BP on tumor cell viability is consistent with an increase in intracellular IPP and ApppI. The most remarkable induction of pyrophosphate accumulation was observed in samples showing low BP-induced IPP/ApppI levels like in IBN and ALN treated T47D cells. T47D cells are generally able to accumulate IPP/ApppI in high amounts as it was reported before [
19]. MCF-7 lack the expression of SLC22A11 while T47D show only low expression of ANKH in contrast to MDA-MB-231 cells. MDA cells produce comparably high levels of the three channels/transporters ANKH, PANX1 and SLC22A11 and this is a possible explanation why the intracellular levels of IPP and consecutively ApppI can not be measured.
Equimolar concentrations of IPP and AMP are necessary for the formation of ApppI, catalyzed by aminoacyl-tRNA synthase enzymes. The concentration of AMP is dependent on the cellular energy metabolism. ApppI formation sequestrates AMP, which is then not available for mitochondrial ATP regeneration and ApppI itself blocks the adenine nucleotide translocases, which catalyzes the exchange of cytoplasmic ADP with mitochondrial ATP across the mitochondrial inner membrane. The molecular consequences of ATP deficiency are a negative energy balance and either reduction of proliferation or apoptosis induction, the latter being dependent on the individual susceptibility of cells to induce the apoptosis program. This condition is perfectly reflected by the ATP-based proliferation measurement, which we used for the determination of cell viability. The intracellular pool of nucleotides for energy metabolism and nucleic acid synthesis appears to be different in the used cell lines. In apoptosis sensitive cells this leads to caspase 3/7 activity induction while in resistant cells proliferation is inhibited.
Our data may also shed light on the mechanisms of regulation of intracellular versus extracellular concentrations of phosphate compounds through channel-mediated release in general. As we showed earlier, ZA enhanced mineralization of osteogenic precursors
in vitro[
32]. Inorganic pyrophosphates are inhibitors of mineralization and upon inhibition of the delivery of these pyrophosphates to the cell surface through both stimulation of intracellular decoy mechanisms and inhibition of channel delivery mineralization should be increased around cells that are able to perform coordinated mineralization processes. Further research will have to unravel this putatively pathology-relevant role of channel activity.
Methods
Cell culture
All media were obtained from Life Technologies GmbH (Darmstadt, Germany), fetal calf serum (FCS) was obtained from Biochrom AG (Berlin, Germany). MCF-7, MDA-MB-231 and T47D cells were cultivated as described previously [
15]. As all experiments were performed with cell lines an ethical approval was not required.
Establishment of stable ANKH overexpressing T47D cells
2.5 × 105 T47D cells per well were seeded on 6well plates and transfected with 2.5 μg pCMV-ANKH (Sino Biological Inc., Beijing, PR China) or the empty pCMV vector, both linearized with SspI (New England Biolabs, Frankfurt, Germany), by using LipofectAMINE 2000 (Life Technologies GmbH, Darmstadt, Germany) according to the manufacturer’s instructions. As selection antibiotics 100 μg/ml hygromycin (Life Technologies GmbH) was added with every medium change.
Determination of cell viability and caspase 3/7 activity
For determination of effects of bisphosphonates on cell viability and caspase 3/7 activity MDA-MB-231, T47D and MCF-7 as well as T47D-pCMV-ANKH and T47D-pCMV control cells were seeded on 96-well plates with a density of 1000 cells/well and were stimulated with 5, 20, 50 and 100 μM zoledronic acid (ZA), ibandronate (IBN), alendronate (ALN) and risedronate (RIS) (AXXORA GmbH, Lörrach, Germany) for 72 h. To analyze effects of probenecid (Prob) co-treatment MCF-7, MDA-MB-231 and T47D cells were stimulated with 0.25 mM Prob (Sigma Aldrich GmbH) together with 20, 50 or 100 μM ZA, ALN, RIS and IBN, respectively. Additional co-stimulatory experiments were performed by using 50 μM carbenoxolone (CBX, Sigma Aldrich GmbH), 5 μM ibrutinib, 100 μM novobiocin (both Selleckchem, Houston, USA) together with 50 μM of each bisphosphonate. Cell viability and caspase 3/7 activity were determined after 72 h with the CellTiter-Glo Luminescent Cell Viability Assay and the Caspase-Glo 3/7 Assay (both Promega GmbH, Mannheim, Germany) according to the manufacturer’s instructions as described previously [
15]. Cytotoxicity was determined in MCF-7 and MDA-MB-231 cells after ZA treatment by using the CytoTox-Fluor™ Cytotoxicity Assay (Promega GmbH) according to the manufacturer’s instructions. Significances were calculated with the Mann–Whitney U Test by comparison of the untreated control to the stimulated values and by comparison of BP treated cells to BP/Prob or BP/CBX co-stimulated cells.
RT-PCR
Total RNA was isolated from MCF-7, T47D and MDA-MB-231 cells by using the NucleoSpin RNA II kit (Macherey-Nagel, Düren, Germany) according to the manufacturer’s instructions. Two micrograms of total RNA were reverse-transcribed with MMLV reverse transcriptase (Promega GmbH) in a volume of 25 μl. For amplification of ABCC1, ANKH, PANX1, SLC22A6, SLC22A8, SLC22A11 and the housekeeping gene EF1α 1 μl of cDNA was used as a template in a volume of 50 μl. Taq DNA polymerase was obtained from Promega GmbH and primers were obtained from biomers GmbH, Ulm Germany with the following sequences in 5′-3′ direction: ABCC1for GGATTTTTGCTGTGGATCGT; ABCC1rev ACCAGCCAGAAAGTGAGCAT; ANKHfor AAAGCCGTCCTGTGTATGGT; ANKHrev CAGGGATGATGTCGTGAATG; PANX1for AGAGCGAGTCTGGAAACC; PANX1rev CAAGTCTGAGCAAATATGAGG; SLC22A6for GTCTGCAGAAGGAGCTGACC; SLC22A6rev GTCCACAGCACCAAAGATCA; SLC22A8for CTGAGCACCGTCATCTTGAA; SLC22A8rev TGGTGTCCACCAGGATGATA; SLC22A11for CTGCCCTCTTGCTCAGTTTC; SLC22A11rev CACTGGCGTTGGAAAGAGTT; EF1αfor AGGTGATTATCCTGAACCATCC; EF1αrev AAAGGTGGATAGTCTGAGAAGC. PCR conditions were as follows: 30 s, 94°C; 30 s, annealing temperature (54°C EF1α, 55°C ANKH, 57°C PANX1, 60°C ABCC1, SLC22A6 and SLC22A11, 62°C SLC22A8), 30 s, 72°C; 35 cycles. PCR bands were analyzed by agarose gel electrophoresis.
Quantitative PCR
MDA-MB-231, T47D and MCF-7 cells were treated for 72 h with 20 or 50 μM ZA, RIS, IBA, ALN as indicated and co-treated with 0.25 mM probenecid. Quantitative PCR (qPCR) was performed in 20 μl by using 1 μl of the cDNA, which was previously diluted 1:5 and 10 μl of KAPA SYBR FAST qPCR Universal Mix (Peqlab Biotechnologie GmbH, Erlangen, Germany) and 2.5 μl of primer pairs for human
KLF2 or
GAPDH as housekeeping gene (Quantitect Hs_KLF2_1 and Hs_GAPDH_1_SG, Qiagen GmbH, Hilden, Germany), dissolved according to the manufacturer’s instructions. The primers for 36B4, which was used as housekeeping gene, and the primers for ABCC1, ANKH, and PANX1 (see above) were obtained from biomers.net GmbH, Ulm, Germany and were used in a concentration of 1 pmol each per reaction with the following sequences in 5′-3′ direction: 36B4_qFor: TGCATCAGTACCCCATTCTATCAT; 36B4_qRev: AGGCAGATGGATCAGCCAAGA [
37]. QPCR conditions were as follows: 95°C, 3 min; 40 cycles: 95°C, 15 s; 60°C, 15 s; 72°C, 20 s; followed by melting curve analysis for specificity of qPCR products. QPCR was performed with the Opticon DNA Engine (MJ Research, Waltham, USA). Data were obtained from three independent experiments and qPCRs were performed three times. Results were calculated with the Relative Expression Software Tool (REST 2009 V2.0.13) obtained from Qiagen GmbH [
38].
Immunocytochemistry for ANKH and PANX1
Breast cancer cells were seeded on coverslips in 6well plates, grown over night, washed thrice with PBS, fixed for 5 min with ice-cold methanol, dried and stored at −80°C until staining. Before staining cells were washed with PBS, permeabilized with PBS/0.05% Tween-20, washed again with PBS, and blocked with 3% BSA in PBS. Cells were incubated with the primary antibodies for ANKH (1:300 (sc-67242) and PANX1 (1:500 sc-49695), respectively (both Santa Cruz Biotechnology, Inc., Heidelberg, Germany) for 16 h at 4°C and a phycoerythrin-labeled secondary antibody (NorthernLights anti-mouse IgG-NL557, RnD Systems, NL007, 1:400) for 2 h at RT. The coverslips were transferred on slides with a drop of Vectashield with DAPI (LINARIS GmbH, Wertheim, Germany) and analyzed under a fluorescence microscope (Axioskop2, filters 1 and 20, Carl Zeiss MicroImaging GmbH, Jena, Germany).
Determination of IPP and ApppI in cell samples
IPP and ApppI were determined as described previously [
5]. Briefly, 1 × 10
6 MCF-7, MDA-MB-231 and T47D cells/well were seeded on 6well plates and incubated overnight. MCF-7 cells were stimulated with 20 μM ZA and 50 μM of all other BP (RIS, IBN, ALN), MDA-MB-231 and T47D cells were treated with 50 μM BP (ZA, RIS, IBN, ALN), 0.25 mM Prob or the combination of each BP and Prob for 24 h, washed with ice-cold PBS and harvested. IPP/ApppI was extracted with ice-cold acetonitrile (300 μl) and water (200 μl) containing 0.25 mM NaF and Na
3VO
4 as phosphatase inhibitors. IPP and ApppI were quantified with HPLC-ESI-MS, protein contents were determined by the BCA method (Perbio Science Deutschland, Bonn, Germany). Values were obtained from three independent experiments.
Competing interests
All authors have no conflicts interests except: Franz Jakob receives honoraria for lectures and consulting from Novartis, Procter & Gamble, Servier, Lilly, MSD, and Roche. Lorenz Hofbauer receives honoraria for lectures and consulting from Amgen, Novartis, Servier, and Merck. Tilman Rachner has received honoraria for consulting and unrestricted research grants from Novartis and MSD.
Authors’ contributions
RE participated in the design of the study and wrote the manuscript, JMW, SG, and BM carried out the proliferation and apoptosis assays, SZ carried out the qPCR analyses, JM, SA and SCS carried out the IPP/ApppI measurements. LH and TR drafted the manuscript, FJ conceived the study, participated in the design of the study and drafted the manuscript. All authors read and approved the final manuscript.