Introduction
Recognition that breast cancer is a heterogeneous disease has helped shape advances in therapy, leading to more targeted therapeutic strategies and improved survival rates in discrete disease subgroups [
1]. This is exemplified by the advent of therapeutic agents targeting estrogen-receptor positive (ER
+) and HER2-positive (HER2
+) breast cancers, which make up approximately 70% of all breast tumours [
2,
3]. Despite these improvements, however, tumours often relapse due to innate or acquired resistance to the therapeutic insult. At the centre of this problem lies additional tumour heterogeneity whereby a small population of cells within, or possibly outside, the tumour are both resistant to drugs and provide the source of new tumour growth [
4,
5]. These cells also contribute directly to the seeding of secondary tumours at distal sites, the primary cause of mortality in breast cancer patients [
6]. These drug resistant cancer initiating cells, often referred to as breast Cancer Stem Cells (bCSCs), have been demonstrated functionally for both human and mouse mammary tumours and tumour cell lines [
7‐
13]. Experiments on human breast tumours in mouse models, for example, indicate that when these cells were deleted, the remaining cells were unable to sustain new tumour growth [
11,
13,
14]. There is, therefore, considerable interest in targeting CSCs within tumours with cytotoxic agents as a cure for breast and other cancers and where possible to broaden the specificity of therapeutic agents to treat as wide a patient group as possible.
Tumour Necrosis Factor (
TNF)-
Related
Apoptosis
Inducing
Ligand (TRAIL) is a promising anticancer agent that exhibits tumour specificity with only mild side effects observed in clinical trials for the treatment of colorectal cancer, non-small cell lung carcinoma and non-Hodgkins lymphoma [
15,
16]. In breast cancer, however, its therapeutic potential is limited by the fact that the majority of breast cancer cell types are resistant to TRAIL [
17,
18]. This has prompted much interest in identifying agents that might increase TRAIL sensitivity in a larger cohort of breast cancer patients. Moreover, stem cells, including cancer stem cells, are documented to be resistant to TRAIL [
16,
19,
20], suggesting that without further sensitization of the tumour-initiating cell sub-population, patients are likely to relapse following TRAIL therapy.
TRAIL targets tumour cells for instructive cell death via the cell-surface receptors TRAIL-R1 (DR4) and TRAIL-R2 (DR5), which initiate the formation of death inducing signalling complexes (DISCs) ultimately leading to the activation of the caspase cascade [
21]. A number of studies have described agents that sensitize one or more breast cancer subtypes to TRAIL, the majority of which implicate components of the apoptosis regulatory machinery as the underlying causes of sensitization [
22‐
35]. Common to a number of these studies is the observation that the endogenous inhibitor of death receptor killing,
cellular
FLICE-
Like
Inhibitory
Protein (c-FLIP), is down-regulated during the sensitization process [
22,
26,
28,
30‐
32,
34]. c-FLIP is a non-redundant antagonist of caspases -8 and -10, preventing these caspases from binding to the DISC and thus inhibiting autolytic cleavage and subsequent activation of downstream executioner caspases [
36] following stimulation by TRAIL. The suppression of c-FLIP has been shown to sensitize some breast cancer cell lines to TRAIL mediated killing, raising the possibility that such a mechanism could be targeted in breast cancer patients [
22,
31,
32,
37‐
39]. However, several questions concerning the specificity of c-FLIP in breast cancer remain that would significantly impact on its prospects as a therapy for breast cancer. These include: whether suppressing c-FLIP in non-tumour cells compromises their viability; whether a broad range of breast cancer subtypes are affected by c-FLIP sensitization; and of particular clinical significance, whether the normally chemo-resistant CSC sub-populations within each of these heterogeneous subtypes are sensitive to de-repression of this apoptotic pathway.
Here we addressed each of these clinically relevant questions by selectively targeting c-FLIP in pre-clinical models of breast cancer. We looked at the effects of suppressing c-FLIP in non-tumourgenic cells, and showed that c-FLIP exhibited tumour cell specificity, similar to that previously ascribed to TRAIL in other tumour types [
15,
40,
41]. Moreover, we demonstrated that the de-repression of TRAIL by c-FLIP inhibition selectively eliminated breast cancer stem cells (bCSCs) from tumour cell populations, irrespective of their HER2/ER receptor status and despite CSC plasticity within the surviving tumour cell population. These observations were then confirmed in
in vivo models of breast cancer whereby primary tumourgenesis was reduced by 80% and the seeding of new tumour growth at distal sites, leading to metastatic disease, was almost completely inhibited. These findings demonstrate potent cellular responses to TRAIL sensitization that have important clinical implications for the advent of new therapeutic strategies for breast cancer patients.
Materials and methods
All experiments were performed with the approval of the University of Cardiff School of Biosciences Ethics Committee and animal work was performed in accordance with the Home Office Animals (Scientific Procedures) Act 1986 under project licence 30/2849.
Cell culture
Four human breast cancer cell lines BT474ER+HER2+, SKBR3ER-HER2+, MCF-7 ER+HER2-, MDA-MB-231ER-HER2-; a murine mammary tumour cell line, N202.1A (from P-L Lollini, Sezione di Cancerologia, Bologna, Italy); the non-tumourigenic cell lines human MCF10A (from T. Stein, University of Glasgow, UK) and murine EPH4 (from C. Watson, University of Cambridge, UK) were maintained in DMEM (MDA-MB-231, EPH4), or RPMI 1640 medium (SKBR3, MCF-7 and BT-474), supplemented with 10% foetal bovine serum, 1% penicillin-streptomycin and 0.5% L-glutamine at 37°C in 5% CO2. Monolayer MCF10A cells were cultured in DMEM/F12, 5% horse serum, 20 ng/ml EGF, 0.5 mg/ml hydrocortisone, 100 ng/ml cholera toxin, 10 μg/ml insulin.
siRNA
Small interfering RNAs (siRNA) targeting two unique sequences in human c-FLIP (FLIPi - Sense: GGAUAAAUCUGAUGUGUCCUCAUUA, Anti-Sense: UAAUGAGGACACAUCAGAUUUAUCC) and a non-specific scrambled control (SCi - Sense: GGACUAAUAGUUGUGCUCCAAUUUA, Anti-Sense: UAAAUUGGAGCACAACUAUUAGUCC) RNA were used in reverse transfections (Invitrogen Life Technologies Ltd, Paisley, UK). Cells were trypsinised and resuspended at a density of 1 × 105 cells/ml and seeded into wells containing 20 μl of 100 nM siRNA in serum free Optimem (Invitrogen Life Technologies Ltd, Paisley, UK)) in a volume of 100 μl per well together with 0.3 μl of Lipofectamine (Invitrogen Life Technologies Ltd). Cells were cultured in the presence of siRNA for 48 hours (MCF-7, MCF10A, EPH4, N202.1A and MDA-MB-231) or 72 hours (SKBR3 and BT474) prior to subsequent assay.
TRAIL treatment of target cells
Cells were treated with soluble human recombinant TRAIL (SuperKillerTRAIL, Enzo Life Sciences, Exeter, UK) at a concentration of 20 ng/ml for 18 hours at 37°C in 5% CO2. For mouse target cells, soluble mouse recombinant TRAIL (Enzo Life Sciences) was added at a concentration of 100 ng/ml for 18 hours.
Western blot assays
Western blots of cell lysates were performed using the following antibodies: cFLIP (Enzo Life Sciences, NP6, ALX-8040428), ER α (Santa Cruz Biotechnology, Santa Cruz, CA, USA, sc-7207), ErbB2 (Abcam, Cambridge, UK, ab2428), Tubulin (Abcam, ab6160).
In vitro caspase inhibition
Functional blocking of caspases was assessed by co-incubation of cells with siRNA and the caspase inhibitors IETD (1 μM), LEHD (10 μM) and AEVD (10 μM) (R&D Systems, Abingdon, Oxford, UK) to inhibit caspases 8, 9 and 10 respectively. After 48 to 72 hours co-incubation, cells were analysed using Annexin-V APC apoptosis assay (eBioscience Ltd, Hatfield, UK).
Cell viability and cell death assays
In heterotypic cell culture assays: siRNA treated cells were treated with 0.25% trypsin for 10 minutes, washed and stained with PKH67 or PKH26 (Sigma-Aldrich, Gillingham, Dorset, UK). PKH67+ve FLIPi cells and PKH26+ve SCi cells were mixed 1:1 and cultured overnight with or without TRAIL and subsequently resuspended in 4 μl of 1:10 fixable near-IR live/dead stain (Invitrogen) and incubated for 15 minutes at 4°C. Cells were then gated for PKH staining versus live/dead staining using a FACS Canto (Becton Dickinson, Oxford, UK). For detailed protocol, see supplementary data. In homotypic cell culture assays: CellTiter blue cell viability assay (Promega UK Ltd, Southampton, UK) and Caspase-Glo assay (Promega) were performed according to the manufacturer's instructions and fluorescence/absorbance/luminescence was assessed using a FluoStar Optima plate reader, while annexin-V APC labelled cells (eBioscience) were analysed by FACS Canto.
Mouse mammary gland tissue histology and primary culture
All procedures were performed in accordance with the Animals (Scientific Procedures) Act 1986 and approved by the UK Home Office. c-FLIP
fl/fl mice [
42] were crossed with blg-Cre animals [
43] to conditionally delete c-FLIP from mammary epithelium. Mammary tissues from 12-week old and 14-day pregnant blg-Cre/c-FLIP
fl/fl females and blg-Cre/c-FLIP
+/+ littermate controls were harvested and fixed in 4% paraformaldehyde/PBS (pH 7.4) overnight, and embedded in paraffin. Paraffin sections (5 μm) were placed on slides, de-waxed and stained with H&E. For primary cell culture, mid-pregnant animals were sacrificed and abdominal mammary glands excised and washed in 70% ethanol. Lymph nodes were removed and finely minced tissue was then processed as described [
44]. Primary cells were maintained in 5% CO
2, 5% O
2 at 37°C.
Mammosphere culture
Cell lines were dissociated into single cell suspensions and plated in ultra-low attachment plates (Corning Life Sciences, Amsterdam, Netherlands) at a density of 20,000 cells/ml in a serum-free epithelial growth medium (MEBM, Lonza Walkersville, MD, USA), supplemented with B27 (Invitrogen), 20 ng/ml EGF (Sigma-Aldrich), Insulin (Sigma-Aldrich), β-mercaptoethanol and hydrocortisone. After seven days mammospheres were collected by gentle centrifugation (1,100 rpm), dissociated in 0.05% trypsin, 0.25% EDTA and re-seeded at 10,000 cells/ml for subsequent passages.
Aldefluor (ALDH1) assay
Surviving cell populations were harvested in 0.25% trypsin and collected by gentle centrifugation (1,100 rpm). Cell pellets were then washed twice in PBS prior to Aldefluor assay (Stem Cell Technologies, Grenobles, France) as previously described [
45].
Mouse tumourigenicity assays
In vivo tumour initiating capability of siRNA treated cells was assessed by orthotopic mammary fat pad transplantation and tail vein injections of BT474 and MDA-MB-231 cell lines, respectively. BT474 siRNA treated cells were harvested using 1 mM EDTA, washed and resuspended at a density of 5 × 106 cells/ml in serum-free L15 media. A 1.5 mg, 60-day slow release 17-β estradiol pellet (Innovative Research of America, Sarasota FL, USA) was inserted subcutaneously above the right scapula of anaesthetised athymic nude mice. A total of 1 × 106 cells were orthotopically injected directly into the abdominal mammary fat pad, with or without 100 ng/ml TRAIL. Mice were then monitored, and when palpable, tumour volume measured twice weekly. MDA-MB-231 cells treated with siRNA were harvested and prepared for injection in the same manner as BT474 cells. Cells were then injected into the mouse tail vein, with or without TRAIL in a volume of 200 μl and mice were sacrificed six weeks post-injection.
Statistical methods
Throughout the article, data are represented as mean +/- standard error taken over a minimum of three independent experiments, unless otherwise stated. Statistical significance was measured using parametric testing, assuming equal variance, in the majority of experiments with standard t tests for two-paired samples used to assess difference between means.
Discussion
Tumour heterogeneity is a major obstacle to therapy. Recent insights into the hierarchical organisation of tumour cell populations highlights the potential importance of targeting the minority tumour-initiating (cancer stem) cell population associated with cancers in order to radically improve patient outcome. The problem is that cancer stem cells are inherently resistant to chemotherapeutic challenge.
Here we have shown, using complementary
in vitro and
in vivo functional assays, that inhibition of c-FLIP (FLIPi) overcomes resistance of breast cancer stem cells (bCSCs) to the anti-cancer agent TRAIL, resulting in the selective elimination of stem cell characteristics in all of the cell lines tested, independent of hormone receptor status. This potentially broadens the range of breast cancer subtypes that could benefit from a TRAIL-based therapy [
18]. Formation of the DISC is a limiting factor in the initiation of the extrinsic apoptotic cascade [
50,
51]. We have confirmed that c-FLIP antagonises this cascade through the inhibition of either of the extrinsic initiator caspases, which cross-talk to the intrinsic pathway (caspase 9) [
38]. The ability to de-repress either caspase-8 or -10 via FLIPi helps to explain the broad range of breast cancer cell types affected.
We found that combined TRAIL/FLIPi treatment
ex-vivo had a marked impact on tumour seeding
in vivo, resulting in a comprehensive suppression of lung metastases arising from circulating tumour cells (Figure
6). Significantly this occurred when TRAIL was co-injected with cells that had previously not been subjected to TRAIL while in cell culture. Despite this, however, a residual tumour initiating capacity persisted in the TRAIL/FLIPi cohort. This may be explained by our
in vitro observations suggesting that bCSCs were marginally more resistant (Figure
5) and exhibited cellular plasticity (Figure
6B) in the nurturing microenvironment of adherent culture. The observation of functional plasticity in mammosphere culture supports a previous study using surrogate markers of bCSCs in breast cancer cell lines [
52]. Crucially, however, we show that the newly acquired MFU activity remained responsive to re-administration of TRAIL/FLIPi. A similar sensitivity to repeat treatments has previously been observed for the Akt inhibitor perifosine, in a xenograft model of Sum159 cells [
45]). These observations have important implications for the future prevention of disease relapse in the clinical setting as they demonstrate that the tumourigenic cell population may be targeted without selecting for resistant cells.
It has been suggested that tumour cells in their natural context do not necessarily exhibit the sensitivity to TRAIL monotherapy as observed
in vitro, implying that a combined therapy would be required to re-sensitize to TRAIL [
53]. We have used RNAi to demonstrate the proof of principle that suppression of c-FLIP expression in combination is sufficient to sensitize breast cancer cells to TRAIL. In light of this, a key future objective is to establish whether long-term suppression of c-FLIP
in vivo - perhaps following the cessation of TRAIL treatment - might help prevent the recurrence of tumours. Despite limitations in drug design due to structural homology between c-FLIP and caspases, agents with broad specificity for c-FLIP have been described, each with anti-tumour properties [
22,
26,
28,
30,
54,
55]. It remains to be determined if these agents exhibit selective targeting of cancer stem cells and whether this is recapitulated
in vivo in the absence of off target effects.
The breadth of the breast tumour cell types affected here raises the question of the potential ubiquity of FLIPi/TRAIL treatment in targeting other cancer types
in vivo. Of the few studies that have addressed the sensitivity of cancer stem cells to TRAIL [
16], the majority, including medulloblastoma [
56], glioblastoma [
19] and lymphoma [
57]-derived stem cells, are resistant, with the exception of colorectal cancer cell lines in which a FACS sorted side-population was shown to be TRAIL responsive [
58]. Sensitization of cancer stem cells to TRAIL has only previously been demonstrated in haematological cancers, including AML [
55] and T-cell lymphoma cells [
57], both of which have implicated, but not functionally proven, a role for c-FLIP in the process. TRAIL sensitization has not previously been described in solid tumour stem cells. Our study, therefore, is the first demonstration, to our knowledge, of TRAIL-mediated loss of functional stem cell activity in a solid tumour cell type and the first indication that CSC activity is directly influenced by c-FLIP.
Other mechanisms for targeting breast cancer stem cells have been described. Notably, a recent study demonstrated reduced stem cell activity in response to Notch1 or Notch4 suppression using the same breast cancer cell lines described herein [
9], which supports the use of gamma-secretase inhibitors in clinical trials [
47]. The Akt/Wnt pathway inhibitor, perifosine, reduces breast cancer stem cell numbers [
45] and incidentally is responsible for the reduction in c-FLIP levels in AML stem cells [
55]. Furthermore, it has been suggested that breast cancer stem cells may selectively express HER2 [
59,
45]) and that inhibition of this pathway could have beneficial consequences for breast cancer patients with both HER2-positive and HER2-negative disease [
47,
48]. As we have seen significant responses of CSCs to combined FLIPi/TRAIL, independent of HER2 receptor status, it will be of interest in the future to establish whether primary human tumour stem cell populations are equally susceptible and whether this is due to amplification of a DISC-related mechanism.
We have shown that the apoptosis observed following c-FLIP inhibition is, like TRAIL, a phenomenon that is relatively cancer-specific. Analysis of non-transformed mammary tissues from c-FLIP-deficient mice indicated that the absence of c-FLIP was not detrimental to normal tissue and did not sensitise normal tissue cells to TRAIL induced apoptosis. It has not been established, however, whether normal stem cells of the breast are affected by either intervention. Neural progenitor cells are resistant to TRAIL in a c-FLIP independent manner [
60] and we are currently investigating whether murine mammary stem cells are similarly refractory.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
LP was responsible for the design of the experiments, assembly of data and manuscript writing. NO was responsible for the design of experiments, data collection and manuscript writing. SMP developed a new assay. ME handled data analysis and interpretation, funding of research and manuscript writing. RC was responsible for the conception and design of the study, data analysis and interpretation, manuscript writing, final approval of manuscript and funding of research. All authors have read and approved the manuscript for publication.