A Cell-Permeant Amiloride Derivative Induces Caspase-Independent, AIF-Mediated Programmed Necrotic Death of Breast Cancer Cells

Amiloride is a potassium-sparing diuretic that has been used as an anti-kaliuretic for the chronic management of hypertension and heart failure. Several studies have identified a potential anti-cancer role for amiloride, however the mechanisms underlying its anti-tumor effects remain to be fully delineated. Our group previously demonstrated that amiloride triggers caspase-independent cytotoxic cell death in human glioblastoma cell lines but not in primary astrocytes. To delineate the cellular mechanisms underlying amiloride’s anti-cancer cytotoxicity, cell permeant and cell impermeant derivatives of amiloride were synthesized that exhibit markedly different potencies in cancer cell death assays. Here we compare the cytotoxicities of 5-benzylglycinyl amiloride (UCD38B) and its free acid 5-glycinyl amiloride (UCD74A) toward human breast cancer cells. UCD74A exhibits poor cell permeability and has very little cytotoxic activity, while UCD38B is cell permeant and induces the caspase-independent death of proliferating and non-proliferating breast cancer cells. UCD38B treatment of human breast cancer cells promotes autophagy reflected in LC3 conversion, and induces the dramatic swelling of the endoplasmic reticulum, however these events do not appear to be the cause of cell death. Surprisingly, UCD38B but not UCD74A induces efficient AIF translocation from the mitochondria to the nucleus, and AIF function is necessary for the efficient induction of cancer cell death. Our observations indicate that UCD38B induces programmed necrosis through AIF translocation, and suggest that its cytosolic accessibility may facilitate drug action.


Introduction
Most currently employed cancer therapeutics initiate apoptotic death in actively proliferating cancer cells. While such agents exhibit a significant degree of efficacy, two key barriers to more effective cancer treatment remain. First, because therapeutic regimens tend to target rapidly proliferating cells, any subset of cells that are dormant or that proliferate slowly can evade therapeutic intervention and give rise to either primary tumor recurrence or the emergence of metastatic lesions [1]. Second, tumor cells commonly activate potent anti-apoptotic pathways to promote their survival and drug resistance [2]. Thus, the development of small molecules that act independently of cell cycle progression to engage non-apoptotic cell death mechanisms offers a particularly attractive approach to thwart tumor progression [3].
Numerous studies in recent years point to the existence of programmed cell death (PCD) mechanisms that are distinct from caspase-dependent (type I) apoptosis [4][5][6][7]. For example, autophagy, or type II PCD, is a stress-induced salvage pathway employed by cells experiencing limited nutrients. Through this mechanism cells sequester bulk cytoplasm and organelles into double membrane vesicles, which ultimately fuse with lysosomes to mediate the degradation of their contents and provide nutrients to support cell survival [8,9]. However, if the stressful conditions become overwhelming the type II PCD pathway can trigger caspase-independent cell death.
Historically, necrosis has been conceptualized as a non-specific cell death process, involving the disruption of the plasma membrane and extrusion of the cytosolic contents, with the potential induction of inflammatory response. However, very recent studies indicate that some necrotic processes, such as programmed necrosis (type III PCD), are under the control of the cell and contribute to both physiological and pathological processes [10,11]. While many of the molecular and cellular details of programmed necrosis remain to be elucidated, it is now recognized that programmed necrosis can be orchestrated by key cellular factors such as the mitochondrial flavoprotein apoptosisinducing factor (AIF), and is characterized by the swelling of organelles such as mitochondria and the endoplasmic reticulum followed by the loss of plasma membrane integrity.
Amiloride, an FDA-approved diuretic that acts on epithelial sodium channels, has been demonstrated in numerous studies to suppress the growth and metastasis of a variety of tumor types in rat and mouse models (reviewed in [12]). We have demonstrated that high-dose amiloride treatment is cytotoxic toward cultured glioma cell lines but does not affect primary rat astrocytes at the same concentration, and that cytotoxicity is caspase-independent and independent of amiloride's inhibitory activities toward the type 1 sodium-proton exchanger (NHE1) and the sodium-calcium exchanger (NCX) [13,14]. Moreover, the amiloride derivatives 5-(N-ethyl-N-isopropyl)-amiloride (EIPA) and hexamethylene amiloride (HMA) have been reported to reduce the growth, viability, motility and invasiveness of hepatocellular carcinoma cells and xenografts [15][16][17]. HMA has also been shown to induce cell death in leukemic cells, while not affecting the viability of normal hematopoietic cells [18]. Taken together, these observations suggest that amiloride and its derivatives exhibit selective anticancer cytotoxicity independent of its ion channel inhibitory activity, making this class of drugs attractive for future clinical evaluation.
A major challenge to repurposing amiloride as an anti-cancer therapeutic is its low potency in cytotoxicity assays. In the present study we have examined the breast cancer cell cytotoxicity of two amiloride derivatives modified at the C(5) position with different substituents. Surprisingly, we observed that the more potent of these derivatives induces cell death via AIF-mediated programmed necrosis, raising the possibility that such amiloride derivatives may be employed to attack tumor cells inherently resistant to apoptosisinducing therapeutics. Moreover, its cytotoxic activity appears to correlate with its cell permeability, suggesting the existence of a novel intracellular target for amiloride and its derivatives.

Reagents and Chemicals
UCD74A and UCD38B were synthesized, purified and handled as previously described [19]. Amiloride (Sigma, St. Louis, MO), UCD74A, and UCD38B were dissolved in DMSO as 200 mM stocks, cisplatin (Calbiochem) as a 20 mM stock, and ionomycin (Calbiochem) as a 5 mM stock. Compounds were stored in the dark at 220uC. The pancaspase inhibitor z-VAD-fmk (Biovision, Mountain View, CA) was used at 1 mM concentration and treatments were carried out for 24 hours. The Green Multi-Caspase Staining kit was purchased from PromoKine and was employed according to the manufacturer's instructions. The autophagy inhibitors chloroquine and compound C (Sigma) were used at 10 mM and 25 mM, respectively, and treatment was carried out for 24 hours.

Cell Culture
All cell lines were obtained from American Type Culture Collection. MDA-MB-231 and MCF-7 cells were cultured in Dulbecco's modified Eagle's medium, while SKBR3 cells were cultured in McCoy's 5A medium. Media contained 10% bovine fetal calf serum and 2% v/v penicillin/streptomycin. Cells were maintained at 37uC in a 10% CO 2 humidified incubator.

Immunoblot Analysis
After treatment with inhibitors, cells were lysed in SDS sample buffer, lysates were resolved by 12% SDS-polyacrylamide gel electrophoresis, and proteins transferred to nitrocellulose. Membranes were blocked for 1 hour with 5% milk in Tris-buffered saline/0.05% Tween-20 (TBST) followed by overnight 4uC incubation in primary antibodies. Blots were then washed with TBST and incubated with HRP-conjugated secondary antibody (goat anti-mouse or anti-rabbit IgG, Invitrogen-Molecular Probes)

Live Cell (MTT) Tetrazolium Assay
Cells were seeded in 24-well plates and grown overnight, and then treated with various compounds for an additional 24 hours. Cells were then treated with 0.5 mg/ml 3-(4,5-Dimethyl-thiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) for 2 hours at 37uC, media was removed, and formazan crystals were dissolved with 300 ml of acidic isopropyl alcoholic solution. 200 ml of sample was transferred to a 96-well plate, and absorbance was measured at 570 nm with 655 nm reference. The absolute optical density was normalized to time-matched, vehicle-treated control cells and expressed as percent viability.

Cytotoxic Lactate Dehydrogenase (LDH) Assay
Cells were seeded in 24-well plates and grown overnight. After treatment with and without inhibitors, 100 ml of supernatant per well was transferred into a 96-well plate and cellular LDH release

Fluorescence Microscopy
The three amiloride derivatives exhibit similar intrinsic fluorescence (unpublished observations), and their visualization was carried out as previously described [20] using an Olympus BX61 deconvolving fluorescence microscope with excitation at 350 nm and emission at 460 nm. For AIF localization studies, cells were cultured in chamber slides overnight, and then treated with DMSO or 250 mM UCD38B, UCD74A or amiloride for 2 hours at 37uC. Cells were washed with PBS and fixed with 4% paraformaldehyde at room temperature for 15 minutes, then washed with PBS and permeabilized with 80% methanol in PBS for 5 minutes. Cells were blocked with 3% non-fat dried milk/1% BSA in PBS for 1 hour at room temperature and incubated with 1:100 anti-AIF mAb in 0.1% BSA/PBS overnight at 4uC. Alexa Fluor 488-conjugated anti-mouse IgG (Invitrogen-Molecular Probes, Eugene, OR) was used at 1:500 dilution in 0.1% BSA/ PBS for 1 hour at room temperature. After PBS wash, slides were mounted with mounting media containing 49,6-diamidino-2phenylindole (DAPI; Invitrogen-Molecular Probes). Imaging was carried out using a confocal microscope (Olympus).

Electron Microscopy
Cells were plated in 8-well chambered slides overnight and then treated with 250 mM UCD38B, UCD74A or amiloride for 2 hours at 37uC. Cells were fixed in Karnovsky fixative and were imaged using a Philips 120 BioTwin electron microscope at 80 KV equipped with a Gatan MegaScan model 794/20 digital camera.

Cellular Fractionation
After treatment with inhibitors, cells were trypsinized, spun at 3006g for 5 minutes, resuspended in hypotonic buffer (20 mM Tris-HCl pH 7.4, 10 mM NaCl, 3 mM MgCl 2 ), transferred to a pre-chilled microcentrifuge tube and incubated on ice for 15 minutes. 25 ml 10% NP-40 was added and the tube was vortexed for 10 seconds. The homogenate was centrifuged at 18006g at 4uC and both the supernatant (cytosolic fraction) and pellet (nuclear fraction) were collected. The nuclear fraction was resuspended in 50 ml extraction buffer (100 mM Tris/HCl, pH 7.4, 2 mM Na 3 V0 4 , 100 mM NaCl, 1% Triton X-100, 1 mM EDTA, 10% glycerol, 1 mM EGTA, 0.1% SDS, 1 mM NaF, 0.5% deoxycholate, 20 mM Na 4 P 2 O 7 , 1 mM AEBSF, 4 mg/ ml pepstatin A, 4 mg/ml leupeptin, 4 mg/ml aprotinin) for 30 minutes on ice with vortexing for 5 second every 10 minutes. The sample was then centrifuged for 30 minutes at 14,0006g at 4uC and the supernatant (nuclear extract) was collected. Cytosolic and nuclear samples were treated with SDS sample buffer and immunoblotting analysis was performed as described above.

siRNA Knockdown
Cells were seeded in 24-well plates and were grown for 16 hours. Cells were then transfected with 100 nM ON-TARGETplus SMARTpool RNAi oligonucleotides (Dharmacon) directed to AIF, cyclinD1, or beclin1, or with nontargeting siRNA per manufacturer's instructions. Proliferation assays and immunoblotting were carried out after 48 hours as described above.

FACS Assessment of Necrosis
MDA-MB-231 cells were treated with DMSO, 250 mM UCD38B, or 50 mM cisplatin (CP) for 24 hours, and suspended cells were incubated with FITC-Annexin V and propidium iodide (PI) from BD Pharmingen FITC Annexin V Apoptosis Detection Kit 1 as described by the manufacturer. Cells were sorted using a Becton Dickinson FACScan flow cytometer, data were collected using Cell Quest version 3.3 software and analyzed using FloJo version 7.6.5 software.

Statistical Analysis
Values are expressed as mean 6 standard error and were calculated from experiments carried out in triplicate. P values were determined using the independent two sample t-test with values less than 0.05 considered statistically significant (*, P,0.05; **, P,0.001).

Cytotoxicity of Amiloride Derivatives Correlates with Cell Permeability
To begin to develop amiloride derivatives that exhibit greater potency in tumor cell death assays, we have replaced the C(5)amino group with glycine to make compound UCD74A, or additionally added benzyl alcohol to the glycine carboxylate to make the more hydrophobic compound UCD38B [20] (see Figure 1A). Surprisingly, we observed that the cytotoxic potencies of these compounds correlate with their cell permeabilities. Using the intrinsic fluorescence of all three compounds to assess their cellular uptake by fluorescence microscopy ( Figure 1B), we observed that amiloride modestly accumulates within MDA-MB-231 breast tumor cells, UCD38B is far more permeant, while UCD74A appears to be largely excluded from the cell interior.
These observations indicate that modification of the 5-amino group of amiloride can markedly alter the cell permeability of analogs.
To assess the impact of the derivatives on tumor cell viability, we first examined a panel of cultured human breast cancer cell lines using the MTT assay. Breast cancer is a general descriptor applied to several genetically and pathologically distinct diseases that may be distinguished using the biomarkers HER2, estrogen receptor (ER), and progesterone receptor (PR), and strategies for treatment can vary markedly depending on classification. SKBR3 cells are a model of HER2-positive breast cancer, MCF7 cells are a model of ER/PR-positive breast cancer, and MDA-MB-231 cells are a model of triple-negative or basal breast cancer. While we observed some variability among the model cell lines, each line was sensitive to amiloride ( Figure 2) at concentrations previously observed to kill cultured human glioma cells [13,14], suggesting that the drug may act broadly toward multiple tumor types and subtypes. We observed that amiloride exhibits a modest potency toward each of the cell lines, eliciting half-maximal effects at 100-200 mM and full effects at .500 mM. UCD38B is more than twice as potent as amiloride in all cell lines, while UCD74A is relatively inert (Figure 2). These observations indicate that the cytotoxic potency of amiloride can be modulated by modification of its C(5) position. Moreover, the results illustrated in Figures 1B and 2 strongly suggest that cell permeability may play a significant role in the cytotoxicity of the drug.

UCD38B-induced Cell Death Occurs Independently of the Cell Cycle
The majority of anti-cancer chemotherapeutics target cycling cells, limiting their efficacy toward dormant cells that could reinitiate primary breast tumors or seed metastatic lesions. To determine whether UCD38B cytotoxic effects are cell cycledependent, we examined the impact of cyclin D1 knockdown on UCD38B-induced cytotoxicity. Since amiloride and its derivatives exhibited qualitatively similar effects on all three breast cancer cell lines, subsequent cell death studies were limited to the highly malignant MDA-MB-231 line. siRNA-mediated cyclin D1 knockdown suppresses cyclin D1 protein levels in MDA-MB-231 cells by greater than 95% ( Figure 3A), decreases the fraction of cells in Sphase from 14% to 5% (not shown), and suppresses proliferation as measured by MTT assay by greater than 50% ( Figure 3B). However, impeding cell cycle progression at G1/S has no discernable effect on the ability of UCD38B to induce cell death ( Figure 3C), indicating that proliferation is not a requirement for the cytotoxicity of this class of drugs.

Apoptosis and Autophagy do not Account for UCD38B Cytotoxicity
To elucidate the mechanism of UCD38B-induced cell death, we first asked whether amiloride or UCD38B induce canonical caspase-dependent apoptosis. Consistent with our previous observations that amiloride promotes the caspase-independent death of cultured human glioblastoma cells [13,14], we observed that amiloride does not induce PARP and caspase-7 cleavage in MDA-MB-231 breast tumor cells, and that co-incubation with the caspase inhibitor z-VAD-fmk does not significantly attenuate the cytotoxicity of amiloride (Figures 4A and 4B). Likewise, UCD38B induces only modest PARP and caspase-7 cleavage compared to the apoptosis-inducing agent cisplatin, and its cytotoxic potency is unaffected by z-VAD-fmk ( Figure 4C and 4D). Under these conditions, z-VAD-fmk efficiently binds to caspases within cells ( Figure 4E) and inhibits tamoxifen-induced MDA-MB-231 cells ( Figure 4F) as previously described [21]. In addition, treatment of cells with UCD38B does not significantly promote staining of cells with FITC-Annexin V, a marker of apoptosis, but efficiently induces the uptake of propidium iodide, a marker of cell permeability and death ( Figure 5). Taken together, these observations indicate that even though UCD38B very modestly induces PARP and caspase-7 cleavage, it acts through a cell death pathway that is largely independent of canonical apoptosis.
Autophagy is a homeostatic cellular process that attempts to preserve cell viability under environmental stress, but can initiate cell death if conditions become too harsh [22]. UCD38B induces an autophagic response in MDA-MB-231 cells, as reflected in the elevation of the autophagy protein LC3 and its lipid modification to the LC3-II form ( Figure 6A). However, inhibition of the autophagic process by the AMPK inhibitor compound C or the lysosomotropic agent chloroquine [23,24] (Figures 6A and 6B), or by a greater than 90% knockdown of the autophagy protein beclin-1 ( Figures 6C and 6D), does not disrupt UCD38B-induced cell death. Taken together, these observations indicate that autophagic cell death likely does not account for UCD38B anticancer cytotoxicity.

UCD38B does not Induce Necrotic Cell Death through Calcium-mediated Calpain Activation
We have observed that UCD38B but not UCD74A triggers the formation of large intracellular vacuoles, membranous structures derived from intracellular organelles that are frequently formed in response to cell stress [6]. To determine whether UCD38Binduced cytotoxicity might involve structural changes to specific intracellular organelles, we carried out transmission electron micrographic analysis of MDA-MB-231 cells treated with DMSO control or with UCD74A or UCD38B for two hours. We observed that UCD38B (Figures 7D-F) but not UCD74A ( Figure 7C) induces the massive dilation of the endoplasmic reticulum (ER) by one to two orders of magnitude, resulting in the formation of large cytosolic vacuoles in essentially every treated cell. Amiloride treatment induces similar changes as UCD38B ( Figure 7B), except that morphological differences are far less pronounced. These observations indicate that the disruption of ER integrity is associated with UCD38B cytotoxicity.
We have recently demonstrated that cytosolic free calcium levels are elevated in glioblastoma cells relative to primary astrocytes, and that further calcium elevation induces cell death [14]. These observations raise the possibility that the observed structural alterations in the ER could trigger an increase in cytoplasmic free calcium, leading to calpain activation the induction of necrosis. This mechanism has been implicated in the dehydroeburicoic acid-induced necrotic death of glioblastoma cells [25]. To determine whether calcium-dependent necrosis contributes to UCD38B-dependent cytotoxicity, we examined cell death using the lactate dehydrogenase (LDH) release assay, a measurement of the extrusion of cytosolic contents from necrotic cells. We observed that UCD38B does indeed provoke the release of LDH from dying cells ( Figure 8A), indicating that the compound likely induces a necrotic mechanism of cell death. However, pre-treatment of cells with the calcium chelator BAPTA-AM did not affect UCD38Binduced cell death under conditions where it inhibited ionomycininduced cell death ( Figure 8A). Moreover, treatment of cells with RIP1 inhibitor necrostatin-1 [26] did not affect cell death (Figures 8B) under conditions where it inhibits necroptic cell death by shikonin ( [27] Figure 8C). Together, these observations strongly suggest that UCD38B induces a form of necrosis distinct from that mediated by calcium-induced calpain and RIP1mediated necroptosis.

UCD38B Induces AIF Nuclear Translocation
Since it has been previously suggested that AIF translocation from mitochondria to the nucleus initiates caspase-independent necrosis [28], we examined the localization of AIF without and with amiloride, UCD74A or UCD38B treatment by cellular fractionation (Figures 9A and 9B), and by confocal immunofluorescence microscopy ( Figure 9C). Both assays revealed a significant translocation of AIF from the cytosol, presumably the mitochondria, to the nucleus specifically upon treatment with cytotoxic concentrations of 38B.
To determine whether AIF contributes to cytotoxicity, we examined the impact of AIF knockdown on UCD38B-induced cell death. siRNA-mediated AIF knockdown led to 65-75% suppression of the protein over a course of 2-5 days (Figures 10A and  10B), and led to a reproducible reversal of UCD38B cytotoxicity  Figure 10C). While the incomplete reversal of UCD38B-induced cell death is likely the result of incomplete AIF knockdown, these observations indicate that AIF contributes substantially to UCD38B cytotoxicity.

Discussion
The current study builds upon our previous observations that amiloride selectively induces the caspase-independent death of glioma cells relative to non-transformed glial cells [13,14]. Here we extend our analysis to examine the impact of two amiloride derivatives on breast cancer cell lines, allowing us to draw several key conclusions concerning the suitability of amiloride derivatives in attacking malignant tumors.
Our comparison of amiloride-based congeners UCD38B and UCD74A demonstrate that the modification of the C(5) amino group of the amiloride parent compound can significantly impact its cytotoxic activity; modification with the 5-benzylglycinyl moiety potentiates cytotoxicity while conversion of the C(5) amino group to glycine inhibits this function. These observations suggest that bulkier or more hydrophobic substituents at this position may augment cytotoxicity. Our future studies will be directed toward further augmenting the cytotoxic potential of amiloride by evaluating the potency of other C(5) derivatives.
Surprisingly, we observed that although UCD38B induces an autophagic response, its mechanism of cytotoxicity is via AIFdependent programmed necrosis or type III PCD. While necrotic cell death has traditionally been conceptualized as a non-specific process, a series of recent studies has revealed that necrosis can be divided into different subtypes, some of which are regulated and may be exploited for therapeutic benefit [29]. Many of the morphological and biochemical characteristics of these subtypes are reflected in UCD38B-treated breast cancer cells, however, none of the previously described programmed necrosis mechanisms account for all of the changes that we observe in breast tumor cells after UCD38B treatment.
Parthanatos and necroptosis are recently described forms of type III PCD that are characterized by AIF translocation, but they differ in the mechanisms underlying flavoprotein release from the mitochondria. In parthanatos PARP processing promotes the formation of PAR polymers, which translocate to the mitochondria from the nucleus to promote AIF release [30]. AIF release in necroptosis is triggered by various stimuli including calpains, cathepsins, reactive oxygen species, and Bax, and is dependent on the RIP1 kinase [29,31]. However, our observations that 38Binduced cell death is not affected by PARP or calpain inhibitors (not shown), or the RIP1 inhibitor necrostatin-1 ( Figures 8B and  8C), rule out these two mechanisms.
Paraptosis is a type III PCD that is characterized by extensive vacuolation and swelling of the endoplasmic reticulum and mitochondria, and caspase-independent cell death [8]. Methuosis, another type III PCD, is also characterized by the presence of large cytosolic vacuoles that are formed through macropinocytosis by clathrin-independent endocytic vesicle trafficking [32]. In contrast, amiloride has been demonstrated to inhibit macropinocytosis through its inhibition of NHE1 [33]. Like UCD38B, both paraptosis and methuosis can produce an autophagic response that is not required for cell death, however a role of AIF in these cell death processes has not been described [34,35]. Thus, the precise mechanism of UCD38B-induced programmed necrosis remains to be delineated.
A key question that emerges from our studies concerns the molecular target of UCD38B action. The anti-cancer effects of amiloride have been ascribed to its inhibition of sodium-hydrogen exchanger (NHE1) and urokinase plasminogen activator (uPA) [12], however, our previous studies have ruled out NHE1. While uPA remains a possibility, our observations point to a strong correlation between cytotoxic potency of the amiloride derivatives and their cell permeability. Thus, either UCD38B acts on an intracellular uPA form that harbors protease activity and is essential for tumor cell viability, or an unknown intracellular target remains to be identified. Our current studies are directed toward resolving these possibilities.
In summary, our studies provide the impetus for the development of new amiloride-based anti-cancer small molecules that initiate cell cycle-independent type III cell death in cancer cells. As these agents act on cells independent of their proliferative state, we predict that they would act synergistically with radiation therapy and conventional chemotherapies that target proliferating cancer cells. Moreover, slowly proliferating or dormant cancer cells could be effectively targeted by this class of compounds. Finally, cancer cells that are resistant to apoptosis-inducing drugs, including tumor-initiating or cancer stem cells, may also be targeted by this class of necrosis-inducing agents.