Neurobiology of Disease 42 (2011) 242–251

Contents lists available at ScienceDirect

Neurobiology of Disease

j ourna l homepage: www.e lsev ie r.com/ locate /ynbd i

Clioquinol induces autophagy in cultured astrocytes and neurons by acting as azinc ionophore

Mi-Ha Park a,b, Sook-Jeong Lee b, Hyae-ran Byun b, Yunha Kim d, Young J. Oh a,Jae-Young Koh b,c,⁎, Jung Jin Hwang d,⁎a Department of Biology, University of Yonsei, College of Life Science and Biotechnology, Seoul 120-749, Republic of Koreab Neural Injury Research Lab, University of Ulsan, College of Medicine, Asan Medical Center, Seoul 138-736, Republic of Koreac Department of Neurology, University of Ulsan, College of Medicine, Asan Medical Center, Seoul 138-736, Republic of Koread Institute for Innovative Cancer Research, University of Ulsan, College of Medicine, Asan Medical Center, Seoul 138-736, Republic of Korea

⁎ Corresponding author. J.-Y. Koh is to be contacted atDof Medicine, University of Ulsan, 388-1 Poongnap-DonRepublic of Korea. Fax: +82 2 483 5446. J.J. Hwang,Research, College of Medicine, University of Ulsan, Asan388-1 Poongnap-Dong Songpa-Gu, Republic of Korea. Fax

E-mail addresses: jkko@amc.seoul.kr (J.-Y. Koh), jjhw(J.J. Hwang).

Available online on ScienceDirect (www.scienced

0969-9961/$ – see front matter © 2011 Elsevier Inc. Aldoi:10.1016/j.nbd.2011.01.009

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 June 2010Revised 4 November 2010Accepted 2 January 2011Available online 8 January 2011

Keywords:Alzheimer's diseaseHuntington's diseaseAutophagosomeAutolysosomeNeuronal deathAmyloid

Recent studies have demonstrated that clioquinol, an antibiotic with an anti-amyloid effect, acts as a zincionophore under physiological conditions. Because increases in labile zinc may induce autophagy, weexamined whether clioquinol induces autophagy in cultured astrocytes in a zinc-dependent manner.Within1 hofexposure to0.1–10 μMclioquinol, the levels ofmicrotubule-associatedprotein1 light chain3(LC3)-II, amarker of autophagy, began to increase in astrocytes. Confocal live-cell imaging of GFP-LC3-transfected astrocytesshowed the formation of LC3(+) autophagic vacuoles (AVs), providing a further indication that clioquinol inducedautophagy. Addition of 3-methyladenine or small-interfering RNA against autophagy-related gene 6 (ATG6/Beclin-1) blocked clioquinol-induced increases in LC3-II. FluoZin-3 fluorescence microscopy showed that, like thezinc ionophore pyrithione, clioquinol increased intracellular zinc levels in the cytosol and AVs in an extracellularzinc-dependent manner. Zinc chelation with N,N,N′,N′-tetrakis-(2-pyridylmethyl) ethylenediamine (TPEN)reduced, and addition of zinc increased the levels of LC3-II and LC3(+) puncta, indicating that zinc influx plays akey role therein. Moreover, astrocytes and SH-SY5Y cells expressing mutant huntingtin (mHttQ74) accumulatedless aggregates when treated with clioquinol, and this effect was reversed by TPEN. These results indicate thatclioquinol-induced autophagy is likely to be physiologically functional.The present study demonstrates that clioquinol induces autophagy in a zinc-dependent manner and contributes toclearance of aggregated proteins in astrocytes and neurons. Hence, in addition to its metal-chelating effect in andaround amyloid beta (Aβ) plaques, clioquinol may contribute to the reduction of Aβ loads by activating autophagyby increasing or normalizing intracellular zinc levels in brain cells.

epartmentofNeurology, Collegeg Songpa-Gu, Seoul 138-736,Institute for Innovative CancerMedical Center, Seoul 138-736,: +82 2 483 5446.ang0829@amc.seoul.kr

irect.com).

l rights reserved.

© 2011 Elsevier Inc. All rights reserved.

Introduction

Autophagy is a macro-catabolic pathway conserved amongeukaryotic organisms (Klionsky, 2007; Levine and Klionsky, 2004)that serves as a mechanism for the bulk degradation of organelles andvarious intracytosolic protein aggregates. During activation of autop-hagy, a portion of cytoplasm is sequestered within a double-lipidmembrane-delimited vacuole termed an autophagosome, which even-tually fuseswith a lysosome to form an autolysosome. Following fusion,

the inner membrane of the autophagosome is digested and its contentsare mixed with lysosomal enzymes, resulting in the degradation of theautophagosomal contents (Klionsky, 2007; Levine and Kroemer, 2008;Mizushima and Klionsky, 2007).

A number of neurodegenerative diseases are associated withaccumulation of abnormal protein aggregates (Ross and Poirier, 2004),suggesting an important role for aberrant autophagy in their pathogenesis(Cuervo et al., 2004; Ravikumar et al., 2002; Ravikumar et al., 2004;Webbet al., 2003). Consistently, genetic ablation of autophagy-related genescauses neurodegeneration in mice (Hara et al., 2006; Komatsu et al.,2006). Thus, because autophagy is likely the singular mechanism fordegrading large cellularmacromolecules such as toxic protein aggregates,the accumulation of these protein aggregates is likely non-specificallyaugmented by inefficient autophagic degradation, regardless of themechanism of aggregation or the identity of proteins.

Clioquinol is a metal-binding compound that was used clinically asan anti-amoebic drug (Schaumburg, 2000). However, owing to a sideeffect, termed subacute myelo-optic neuropathy (SMON), affecting

243M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

large number of patients in Japan (Tateishi, 2000; Tsubaki et al., 1971),clioquinol has been withdrawn from the market. Bush and colleaguesshowed that clioquinol is highly effective in reducing amyloid beta(Aβ) plaque load in APP Tg2576 mice, a model for Alzheimer's disease(AD) (Cherny et al., 2001). It was initially suggested that the weakzinc-chelating property of clioquinol might underlie the anti-amyloideffect. Consistent with this, we demonstrated that genetic removal ofsynaptic zinc from the mouse brain greatly reduces plaque load (Leeet al., 2002), and a similar plaque-reducing effect has beendemonstrated for the membrane-permeant metal chelator DP-109(Lee et al., 2004). Collectively, these results indicate that removal ofzinc from extracellular Aβ plaques may help reduce the plaque load inthis AD model (Cherny et al., 2001).

Given the high membrane permeability and weak zinc-bindingproperty of clioquinol, recent studies have suggested that this drugmay instead act as a zinc ionophore (Colvin et al., 2008; Ding et al.,2008), shuttling free zinc in or out of cells, depending on the free zincconcentration gradient. Consistent with this concept, clioquinol has

Fig. 1. Clioquinol induces autophagy in astrocytes. A. LDH released from astrocytes after a 24compared with control; paired t-test,). B. Phase-contrast photomicrographs of astrocytes, sC.Western blot for LC3 in astrocytic cultures treated with the indicated concentrations of cliodependent increase in intracellular LC3-II levels. D. Western blot for LC3 in astrocytic cultureastrocytes were sham washed (CTL) or exposed to 5 μM clioquinol (ClioQ) for 6 h, and thenLC3 (left), LysoGreen (middle) and the merged image of the two (right) are shown. LC3(+) vbetween RFP and LysoGreen signals is evident, indicating that some AVs were fused to lysos

been shown to induce zinc-dependent cell death in certain cancer celllines (Ding et al., 2005).

We previously demonstrated a link between intracellular zinc andautophagy in astrocytes under oxidative stress (Lee et al., 2009) and inMCF-7 breast cancer cells treated with tamoxifen (Hwang et al.,2010), suggesting that if clioquinol increases intracellular zinc levelsby acting as a zinc ionophore, it might also activate autophagy. Herewe examined whether, in addition to its extracellular chelating effect,clioquinol induces autophagy by introducing zinc into astrocytes, andthereby helps reduce the build-up of protein aggregates such as Aβ inbrain cells.

Materials and methods

Materials

Clioquinol was purchased from Calbiochem (San Diego, CA).Pyrithione, 3-methyladenine (3MA), TPEN, zinc chloride, calcium

-h exposure to the indicated concentrations of clioquinol (mean±SEM, n=3; **pb0.01ham washed (CTL) or treated with 10 μM clioquinol (ClioQ) for 24 h. Scale bar, 50 μm.quinol for 6 h. β-actin was used as a loading control. Clioquinol induced a concentration-s at the indicated hours (h) after the addition of 5 μM clioquinol. E. RFP-LC3-transfectedloaded with LysoTracker Green (LysoGreen). Fluorescent confocal micrographs for RFP-acuoles increased in number and size after exposure to clioquinol. A substantial overlapomes. Images are from a z-series (seven z-section images, 1 μm thick). Scale bar, 10 μm.

Fig. 2. 3MA and ATG6-siRNA (ATG6si) inhibit clioquinol-induced autophagy.A. Western blots for LC3 in samples from control (CTL) cultured astrocytes orastrocytes treated for 6 h with 5 μM clioquinol alone (ClioQ), clioquinol and 5 mM 3MA(+3MA), or 3MA alone (3MA). β-actin was used as a loading control. Addition of 3MAblocked the clioquinol-induced increase in LC3 levels. B. Bars denote fold-increases indensitometric values of LC3-II bands relative to the density of corresponding β-actinbands. The value in sham-washed controls (CTL) was defined as 1 (mean±SEM, n=3;**pb0.01 compared with control; *pb0.05 compared with clioquinol alone; pairedt-test). C. Western blots for ATG6 and LC3 in samples obtained from culturedastrocytes. Astrocytes were transfected for 2 days with 50 nM scrambled (control)siRNA (Controlsi) or siRNA targeting ATG6 (ATG6si), and then exposed to 5 or 10 μMclioquinol for 6 h. β-actin was used as a loading control. Clioquinol increased ATG6levels, an effect that was completely blocked by treatment with ATG6si. Reduction ofATG6 also resulted in the failure of clioquinol treatment to induce an increase in LC3 IIlevels. D. Fluorescent photomicrographs of astrocytes transfected with GFP-LC3 for 16 hand then shamwashed (CTL, upper) or exposed to 5 μMclioquinol for 5 h (ClioQ, lower).In sister cultures, astrocytes were transfected with 50 nM Controlsi (left) or ATG6si(right) for 2 days prior to GFP-LC3 transfection. ATG6si significantly reduced thenumber of GFP-LC3(+) AVs induced by clioquinol. Scale bar, 50 μm.

244 M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

ethylenediaminetetraacetate (Ca2+-EDTA) were obtained from Sigma(St. Louis, MO). Chelex 100 resin was purchased from BioRad(Hercules, CA).

Cortical cell cultures

Mixed cortical cell cultures, containing both neurons and astrocytes,were prepared from fetal mice at 15 days of gestation, as describedpreviously (Choi, 1985). Briefly, dissociated cortical cells were platedonto a previously established astroglial cell monolayer at 2.5 hemi-spheres per 6-well plate (Nunc, Roskilde, Denmark) in plating medium(Dulbecco's modified Eagle medium [DMEM; Gibco BRL, Rockville, MD]supplemented with 20 mM glucose, 38 mM sodium bicarbonate, 2 mMglutamine, 5% fetal bovine serum and 5% horse serum). Cytosinearabinoside (Ara C, 10 μM)was added 5–6 days after plating to halt thegrowth of non-neuronal cells.

Near-pure neuronal cultures were prepared from same-aged fetalmice by plating dissociated cortical cells on laminin-coated plates (4hemispheres per 6-well plate) in plating medium and adding Ara C3 days after plating. Dissociated cells were used 7–8 days afterseeding.

Astroglial cultures were prepared from neocortices of newbornmice (postnatal day 1–3) and plated at 0.5 hemispheres per 24-wellplate in the same plating medium but supplemented with 7% fetalbovine serum and 7% horse serum (McCarthy and de Vellis, 1980).Cells were used 14–28 days after seeding.

SH- SY5Y cell cultures

SH-SY5Y, a neuroblastoma cell line, was obtained from ATCC(Manassas, VA). Cells were cultured and maintained in platingmedium (DMEM supplemented with 20 mM glucose, 38 mM sodiumbicarbonate, 10% fetal bovine serum).

Western blot analysis

Cells were lysed on ice with lysis buffer (20 mM Tris-HCl pH 7.4,150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mMsodium pyrophosphate, 1 μM Na3VO4, 1 μg/ml leupeptin, 1 μg/mlaprotinin, and 1 mM PMSF). The lysates were centrifuged and theprotein concentration in supernatants was determined using the DcProtein Assay Reagent (BioRad). Equal amounts of proteins wereloaded and electrophoresed on SDS-polyacrylamide gels and thentransferred to PVDF membranes. Membranes were incubated over-night at 4 °Cwith primary antibodies, and then probedwith secondaryantibodies for 1 h at room temperature. Immunoreactive proteins werevisualized using ImmobilonWestern Chemiluminescent HRP Substrate(Millipore, Billerica, MA) and quantitatively analyzed by densitometry(Kodak Molecular Imaging software). The antibody to microtubule-associated protein 1 light chain 3 (LC3) was from Novus (1:1000;Littleton, CO), the antibody to β-actin (1:5000) and huntingtin(Htt;1:500) was from Sigma and Millipore, and the antibodies toautophagy-related gene 6 (1:1000; ATG6/Beclin-1) and GFP (1:2000)were fromCell Signaling (Beverly,MA) and Santa Cruz (Santa Cruz, CA).

Confocal live-cell imaging

For confocal imaging, astrocyteswere culturedonpoly-L-lysine-coatedglass slides. Astrocytes were transiently transfected with RFP-LC3 orGFP-LC3 plasmid for 16 h using Lipofectamine2000 (Invitrogen, Carlsbad,CA) according to the manufacturer's instructions. The GFP-LC3 plasmidwas constructed as previously described (Lee et al., 2009). The RFP-LC3plasmid was a generous gift from Drs Maria Colombo and MichelRabinovitch (Universidad Nacional de Cuyo, Mendoza, Argentina). Theastrocyteswere treatedwith 5 μMclioquinol for the indicated times, afterwhich they were stained with 5 μM FluoZin-3, 75 nM LysoTracker Red

and 75 nM LysoTracker Green (Invitrogen) in minimum essentialmedium (MEM) for 5–30 min in a CO2 incubator, and transferred toHanks balanced salt solution. Confocal images were obtained using anUltra View Confocal Live-Cell Imaging System (PerkinElmer, Waltham,MA) with an ECLIPSE TE2000 microscope (Nikon, Melville, NY).

245M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

Clearance of GFP-tagged huntingtin Q74 aggregates

Astrocytes and SH-SY5Y cells were transiently transfected withGFP-tagged huntingtin Q74 (GFP-mHttQ74) using Lipofectamine2000(Invitrogen) according to the manufacturer's instructions. Six hoursafter transfection with GFP-mHttQ74, cells were washed with MEM forastrocytes orwith serum-freeDMEMfor SH-SY5Y cells, and treatedwiththe indicated drugs. Cells were observed under a fluorescencemicroscope (Olympus IX70, Tokyo, Japan) during a 18-hour periodafter drug exposure. The size and number of GFP-positive aggregates inastrocytes were analyzed using ImagePro (Media Cybernetics, SilverSpring, MD). Quantitative analysis for Htt and GFP blots was performedaccording to the methods mentioned above. GFP-mHttQ74 plasmidDNA was a generous gift from Dr. David C. Rubinsztein (University ofCambridge, Cambridge, UK).

Transfection with small interfering RNA for ATG6

ATG6 knockdown was accomplished by transfecting astrocyteswith 50 nM On-TARGETplus small-interfering RNA (siRNA) targetingATG6 (Dharmacon, Lafayette, CO) using Lipofectamine RNAi MAX(Invitrogen); non-targeting (scrambled) siRNAwas used as a negativecontrol. Cells were incubated for 2 days and then treated with 5 μMclioquinol for 5 h.

Zinc-deficient media cultures

Zinc-deficient culture conditions were generated by incubatingMEMwith 10% Chelex 100 resin for 1 h with constant orbital shaking.After centrifugation to remove resin, 1.8 mM CaCl2 and 0.8 mMMgSO4 (zinc-free) was added to Chelex 100-treated MEM. Cells werethen incubated in zinc-deficient media with the indicated drugs for1 h, stained with 5 μM FluoZin-3 for 30 min, and observed under afluorescence microscope.

Propidium iodide staining

The disruption of membrane integrity was observed by stainingcells with 2 μg/ml propidium iodide for 10 min at 37 °C in a 5% CO2

incubator and observed under a fluorescence microscope at awavelength of 568 nm.

Quantification of cell death

Overall neuronal cell death was quantitatively assessed bymeasuring lactate dehydrogenase (LDH) activity released into thebathing medium from damaged cells (Koh and Choi, 1987). Aftersubtracting the mean background value in culture cells withoutneuronal damage (defined as 0%), each LDH value in mixed cultureswas scaled to the mean value in sister positive-control culturestreated with 300 μM glutamate for 24 h (defined as 100%); this lattertreatment induced nearly complete neuronal death in the absence ofglial damage. LDH values in astrocytic cultures were obtained in asimilar manner. In brief, after subtracting the mean background valuein cultured astrocytes without any injury (defined as 0%), each LDHvalue in sister positive-control cultures treated with 300 μM zinc for24 h (defined as 100%).

Immunocytochemistry

Cells were fixedwith 4% paraformaldehyde for 1 h and permeabilizedwith 0.2% Triton X-100 for 15 min. After blocking with bovine serumalbumin, cells were incubated with anti-MAP2 (1:750; Sigma) or anti-GFAP (1:750; Millipore) antibodies, and Alexa Fluor-conjugated second-ary antibodies (1:1000; Invitrogen).

Statistics

All data are presented as mean±SEM. Paired t-tests were used toanalyze differences between two groups. Values of pb0.05 wereconsidered statistically significant.

Results

Clioquinol induces autophagy in astrocytic cultures

Cultured astrocytes were exposed to 0.1–10 μM clioquinol for 24 h.At low concentrations (0.1–5 μM), clioquinol did not induce conspicu-ous astrocytic cell death. However, at 10 μM, clioquinol induced amodest, but significant, level of cell death, which was accompanied byaccumulation of cytoplasmic vacuoles (Figs. 1A and B). We thenexamined whether clioquinol treatments, including exposure tosublethal concentrations, induced autophagy in cultured astrocytes.Western blot analyses showed that clioquinol induced a concentration-and time-dependent increase in the levels of the lipidated form of LC3(LC3-II), which is a necessary for autophagophore formation (Figs. 1Cand D). The formation of autophagic vacuoles (AVs) in live cells wasexaminedmorphologically byobservingRFP-LC3-transfected astrocytesunder a confocal microscope. Although some RFP-LC3(+) AVs werepresent in control astrocytes, 5 μM clioquinol treatment for 6 hsignificantly increased their number and size (Fig. 1E, left panel).Double staining showed that the majority of LC3(+) AVs were alsostained with LysoTracker, indicating that the majority of AVs had fusedwith lysosomes to form autolysosomes (Fig. 1E, middle and rightpanels). Clioquinol-induced increases in LC3-II levels were blocked byaddition of 3MA, an inhibitor of class III phosphoinositide 3-kinase(PI3K) and an autophagy inhibitor (Figs. 2A and B). Moreover,siRNA-mediatedknockdownofATG6 (alsoknownasBeclin-1) inhibitedthe increases in LC3-II levels (Fig. 2C) and attenuated the increase inthe number of GFP-LC3(+) vacuoles in clioquinol-treated cells (Fig. 2D).

Clioquinol acts as a zinc ionophore and introduces zinc into AVs

Because clioquinol has been reported to act as a zinc ionophore(Colvin et al., 2008; Ding et al., 2008), we examined whether it alsoacts in this capacity in the current experimental context. Staining ofcortical cell cultures with FluoZin-3-AM revealed that treatment withclioquinol (ClioQ, 5 μM) in normal exposure media slightly increasedintracellular labile zinc (Fig. 3A, upper panel). Pyrithione (50 μM), awidely used zinc ionophore, also modestly increased intracellularlabile zinc levels in cortical cell cultures. We then tested the effectsof altering extracellular zinc levels. Addition of zinc (500 nM) tothe medium alone did not increase intracellular labile zinc levels.However, in this high zinc media, both clioquinol and pyrithionefurther increased cellular zinc levels compared with that inducedin normal medium (Fig. 3A, middle panel). In contrast, the increasein labile zinc levels induced by clioquinol or pyrithione wassubstantially blunted in zinc-deficient media (Fig. 3A, low panel),prepared as described in Materials and methods. These resultsindicate that, under the present experimental conditions, the mediacontains a certain amount of zinc, and clioquinol, like pyrithione, actsas a zinc ionophore.

Next, we examined the subcellular localization of labile zinc inclioquinol-treated astrocytes. Staining of RFP-LC3-transfected astro-cytes with FluoZin-3 revealed that exposure to 5 μM clioquinol for6 h resulted in the accumulation of zinc in autophagosomes (Fig. 3B).Double staining with LysoTracker and FluoZin-3 demonstrated asubstantial overlap between the two signals, indicating that lyso-somes, likely autolysosomes, contained high levels of labile zinc(Fig. 3C).

246 M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

Fig. 4. Clioquinol-induced autophagy is dependent on zinc. A. Fluorescent confocalimages of RFP-LC3-transfected astrocytes after a 6 h exposure to 5 μM clioquinol(ClioQ), 5 μM clioquinol and 500 nM TPEN (+TPEN), or 5 μM clioquinol and 1 μM zinc(+Zn). Whereas chelation of zinc by TPEN decreased autophagy, addition of zincpotentiated clioquinol-induced autophagy. Images are from a z-series (six z-sectionimages, 1 μm thick). Scale bar, 25 μm. B. Western blot for LC3 in astrocytes treated with5 μM clioquinol alone (ClioQ), 5 μM clioquinol and 500 nM TPEN (+T), 5 μM clioquinoland 1 μM zinc (+Z), 500 nM TPEN alone (T), or 1 μM zinc alone (Z) for 6 h. β-actin wasused as a loading control. Clioquinol increased the level of LC3 II in a zinc-dependentmanner. C. Western blot for LC3 in cells treated with 1 μM pyrithione (Pyri), 1 μMpyrithione and 500 nMTPEN (+T), or 1 μMpyrithione and 1 μMzinc (+Z) for 3 h.β-actinwas used as a loading control.

247M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

Zinc plays a role in clioquinol-induced autophagy

Having observed an anatomical correlation between labile zincand AVs following clioquinol treatment, we examined whetherintracellular labile zinc played a causal role in the formation of AVs.To test this, we used TPEN, a cell membrane-permeant, high-affinityzinc chelator (Canzoniero et al., 2003). We found that the substantialincrease in the formation of RFP-LC3(+) AVs induced by clioquinol(Fig. 1, upper panels) was completely blocked by the addition of TPEN(Fig. 4A, lower left panel). Conversely, addition of zinc to the mediumfurther increased clioquinol-induced formation of LC3(+) AVs(Fig. 4A, lower right panel); high zinc medium (500 nM) alone,which did not increase intracellular zinc levels, did not increase LC3(+) AV formation (data not shown). Consistent with these morpho-logical observations, Western blot analyses showed that TPENattenuated, and excess zinc potentiated, clioquinol-induced increasesin LC3-II levels (Fig. 4B). The same pattern of changes in LC3-II levelswas obtained with the zinc ionophore pyrithione (Fig. 4C).

Clioquinol reduces accumulation of huntingtin aggregates

Because autophagy is activated by clioquinol, we examinedwhetherclioquinol-induced autophagy can increase the clearance of abnormalproteins, such as mutant huntingtin protein with 74 polyQ repeats(mHttQ74). In GFP-mHttQ74 transfected astrocytes, GFP(+) proteinaggregates formed from 12 h after transfection. Compared withcontrols, addition of clioquinol substantially reduced the accumulationof aggregates (Figs. 5A and B). Consistent with the observation thatclioquinol-induced autophagy was dependent on labile zinc, additionof TPEN completely abrogated the inhibitory effect of clioquinol onGFP-mHtt aggregates (Figs. 5A and B). Moreover, treatment with TPENalone significantly increased aggregate formation, but addition of zincreduced TPEN-induced aggregate formation, indicating that labile zincmight reduce the accumulation of protein aggregates by supportingbaseline autophagy (Figs. 5D and E). Western-blot analyses using ananti-GFP antibody and anti-Htt antibody confirmed these morpholog-ical findings (Figs. 5C and F). In addition to astrocytes, clioquinol andTPEN showed same effects on mHtt aggregate formation in neuron-derived SH-SY5Y cells, indicating that clioquinol acts on both neuronand astrocytes by the same zinc-dependentmechanism(Figs.5G andH).

Clioquinol induces autophagic cell death in neurons

Finally, we examined the effect of clioquinol on neuronal cells.Exposure of mixed cortical cell cultures containing neurons andastrocytes to 10 μM clioquinol for 24 h induced extensive death ofneurons (Figs. 6A and B), but little death of background astrocytes(Fig. 1B). Propidium iodide-staining of live cells (Fig. 6A) and anti-MAP2-staining of fixed cells indicated that most dead cells wereneurons. Measurement of LDH released into media from dead cellswas used to quantify cell death. Whereas 1 μM clioquinol was nottoxic, 5 and 10 μM clioquinol led to cell death in 41% and 78% ofcells, respectively (Fig. 6C). The addition of zinc chelators (TPEN orCa2+-EDTA) inhibited the neuronal death induced by 10 μM clioquinol;in contrast, administration of zinc potentiated clioquinol-induced celldeath (Fig. 6D). These results indicate that, similar to its induction of

Fig. 3. Clioquinol induces zinc accumulation in AVs. A. Fluorescent photomicrographs ofenvironments. Top and middle rows: Cells were sham washed (CTL) or treated with 5 μM cl(ClioQ+Zn), or 50 μM pyrithione and 500 nM zinc (pyrithione+Zn) for 1 h in normal expopyrithione-induced increases in intracellular zinc levels, which were only modestly increasincrease intracellular zinc levels. Bottom row: In zinc-deficient media prepared with Chintracellular zinc levels beyond those in controls. Scale bar, 100 μm. B. Fluorescent confocal mtransfected astrocytes were sham-washed (CTL) or treated with 5 μM clioquinol (ClioQ) fortreated astrocytes. Images are from a z-series (five z-section images, 1 μm thick). Scale bar, 10stained astrocytes incubated in the presence (ClioQ) or absence (CTL) of 5 μM clioquinol foincreased in lysosomes after exposure to clioquinol. Images are from a z-series (seven z-se

autophagy in astrocytes, clioquinol-induced neuronal death was alsodependent on extracellular zinc. Clioquinol-induced neuronal death wasinhibited by 3MA, showing that it occurs mainly through an autophagicmechanism (Fig. 6E).Moreover, the levels of LC3-II in near-pure neuronalcultureswere increased in a clioquinol-concentration-dependentmanner(Fig. 6F) and addition of 3MA blocked the increase in LC3-II induced byclioquinol (Fig. 6G and H).

FluoZin-3-stained mixed neuronal and astrocytic cultures exposed to different zincioquinol (ClioQ), 50 μM pyrithione, 500 nM zinc (Zn), 5 μM clioquinol and 500 nM zincsure medium. Addition of zinc to the culture medium greatly increased clioquinol- anded by clioquinol or pyrithione alone; zinc alone at this concentration (500 nM) did notelex 100, 1 h treatment with 5 μM clioquinol or 50 μM pyrithione failed to increaseicrographs of RFP-LC3 (left), FluoZin-3 (middle) and merged images (right). RFP-LC3-6 h. There was a substantial overlap between zinc and LC3(+) vacuoles in clioquinol-μm. C. Fluorescent confocal micrographs of LysoTracker Red (LysoRed)- and FluoZin-3-r 6 h. The levels of labile zinc were low in the cytosol of control astrocytes, and werection images, 1 μm thick). Scale bar, 10 μm.

Fig. 5. Reduction of huntingtin aggregates by clioquinol. A. Fluorescent photomicrographs of GFP-tagged mutant huntingtin Q74 (GFP-mHttQ74)-transfected astrocytes. Aftertransfection with GFP-mHttQ74 for 6 h, astrocytes were treated with 2.5 μM clioquinol, or clioquinol and 500 nM TPEN for 18 h. Arrows indicate GFP-positive aggregates. Scale bar,50 μm.. B. Left bars denote the number of aggregate-positive cells as a percentage of total GFP-positive cells. Right bars indicate relative changes in the size of GFP-positive aggregates.Values for individual bars were normalized to control values, defined as 100% (mean±SEM; n=4; **pb0.01 compared with CTL or clioquinol; two-tailed t-test,). C. Western blotanalysis of GFP-mHttQ74 using an anti-GFP antibody and anti-Htt antibody. Transfected cells were sham washed (CTL) or treated with 2.5 μM clioquinol, or clioquinol and 500 nMTPEN for 18 h. The bars represent the ratio of GFP-mHttQ74 bands to the corresponding β-actin bands, normalized to the ratio in CTL, defined as 1 (mean±SEM; n=4; *pb0.05compared with CTL or clioquinol; two-tailed t-test). Clioquinol noticeably reduced mHtt aggregation, and the addition of TPEN blocked the clearance of aggregates by clioquinol.D. Fluorescent photomicrographs of GFP-mHttQ74-transfected astrocytes. After transfection with GFP-mHttQ74 for 4 h, astrocytes were sham-washed (CTL) or treated with 500 nMTPEN or with the combination of 500 nM TPEN and 25 μM zinc chloride for 18 h. Arrows indicate GFP-positive aggregates. Scale bar, 50 μm. E. Left bars show the number ofaggregate-positive cells as a percentage of total GFP-positive cells. Right bars represent relative changes in the size of GFP-positive aggregates. Values for individual bars werenormalized to the control values, defined as 100% (mean±SEM; n=4; *pb0.05, **pb0.01 compared with CTL or TPEN; two-tailed t-test). F. Western blot analysis of GFP-mHttQ74using an anti-GFP antibody and anti-Htt antibody. Cells were sham washed (CTL) or treated with 500 nM TPEN or with the combination of 500 nM TPEN and 25 μM zinc chloride for18 h. The bars represented the ratio of GFP-mHttQ74 bands to the corresponding β-actin bands, normalized to the ratio in CTL (mean±SEM; n=4; **pb0.01 compared with CTL orTPEN; two-tailed t-test). TPEN markedly accelerated formation of aggregated mHtt. Addition of zinc to TPEN-treated cells reversed effect of TPEN alone. G. Western blot analysis ofGFP-mHttQ74 using an anti-GFP antibody and anti-Htt antibody. GFP-mHttQ74-transfected SH-SY5Y cells were shamwashed (CTL) or treated with 2.5 μM clioquinol, clioquinol plus500 nM TPEN or 500 nM TPEN alone for 18 h. H. Left bars represent the ratio of GFP-mHttQ74 bands to the corresponding β-actin bands, normalized to the ratio in CTL (mean±SEM;n=4; **pb0.01 compared with CTL or clioquinol; two-tailed t-test). Clioquinol markedly reduced formation of aggregated mHtt. Addition of TPEN reversed clioquinol-inducedclearance of mHtt aggregates. Right bars depict the ratio of Htt bands to the corresponding β-actin bands, normalized to the ratio in CTL (mean±SEM; n=4; **pb0.01 comparedwith CTL or clioquinol; two-tailed t-test).

248 M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

Discussion

Clioquinol is the prototype of compounds that target metal-Aβinteractions in AD. Its efficacy in reducing plaque loads has beendemonstrated in APP Tg2576 mice (Cherny et al., 2001). However,despite initial promising results (Ritchie et al., 2003), human clinical

trials were halted, reportedly due to contaminants introduced duringchemical synthesis (Adlard et al., 2008). Instead, PBT-2, a structuralanalogue of clioquinol, is currently in clinical trials (Lannfelt et al., 2008).

Because clioquinol is a weak zinc chelator (Kd≈10−7), the amyloidplaque-reducing effect of clioquinol was initially attributed to metalchelation (Cherny et al., 2001); because metals such as zinc and copper

Fig. 6. Clioquinol induces autophagy and autophagic cell death in neuronal cultures. A. Phase-contrast and propidium iodide-stained photomicrographs of mixed cortical cultures24 h after sham wash (CTL) or exposure to 10 μM clioquinol (ClioQ). Clioquinol induced cell death almost exclusively in neurons. Scale bar, 100 μm. B. Differential interferencecontrast and anti-GFAP (upper panel) or anti-MAP2 (lower panel) fluorescent photomicrographs of mixed cortical cultures 24 h after sham wash (CTL) or exposure to 10 μMclioquinol (ClioQ). Anti-MAP2 staining showed that clioquinol induced neuronal cell loss. Scale bar, 100 μm. C. LDH released frommixed cortical cultures after a 24 h exposure to theindicated concentrations of clioquinol (mean±SEM, n=3; **pb0.01 compared with control; paired t-test). D. Bars represent LDH released frommixed cortical cultures after a 24 hexposure to 10 μM clioquinol (ClioQ), clioquinol and 500 nM TPEN (+TP), clioquinol and 1 mM Ca2+-EDTA (+CaE), or clioquinol and 500 nM zinc (+Z). Clioquinol inducedneuronal death in a zinc-dependent manner. E. Bars denote LDH released from mixed cortical cultures after a 24 h exposure to 10 μM clioquinol (ClioQ), clioquinol and 5 mM 3MA(+3MA), or 5 mM3MA alone (3MA). (mean±SEM, n=3; **pb0.01 compared with clioquinol; paired t-test). F. Western blot for LC3 in near-pure neuronal cultures treated with theindicated concentrations of clioquinol for 6 h. β-actin was used as a loading control. G. Western blot for LC3 in nearly pure neuronal cultures treated with 1 μM clioquinol alone(ClioQ) or clioquinol plus 5 mM 3MA (+3MA) for 6 h. H. Bars denote densitometric measurements of the LC3II bands from nearly pure neuronal cultures sham washed (CTL) ortreated with 1 μM clioquinol alone (ClioQ), or clioquinol and 5 mM 3MA (+3MA) for 6 h, relative to the corresponding β-actin bands. All values were normalized to the ratio incontrol, defined as 1 (mean±SEM, n=3; *pb0.05 compare with control; **pb0.01 compared with clioquinol alone; paired t-test).

249M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

250 M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

are bound to aggregate Aβ in plaques, chelating these metals mightrender Aβ plaques more susceptible to degradation and clearance(Adlard and Bush, 2006; Bush et al., 1994; Lee et al., 2002). Consistentwith this idea, clioquinol-induced reductions in plaque loads in thebrainwere associated with concurrent increases in the levels of Aβ in theblood (Cherny et al., 2001). Moreover, our laboratory showed thatanother lipophilic chelator, DP-109, alsomarkedly reduced plaque loadsin Tg2576 mice (Lee et al., 2004). However, recent studies haveindicated that clioquinolmight act as a zinc ionophore rather than a zincchelator in physiological solutions (Ding et al., 2008). This may reflectthe fact that clioquinol is a weak zinc chelator with high membranepermeability. Hence, it will act to transfer zinc down its concentrationgradient. Because intracellular free zinc levels are quite low (Colvinet al., 2008), it is likely that clioquinol would carry zinc from theextracellular to the intracellular space under normal conditions. Thiseffect might be increased in the AD brain, where Aβ plaques provide agreater reservoir of extracellular zincand intracellular zinc levelsmaybelower (Cherny et al., 2001).

Our results support this hypothesis and suggest another interest-ing possible mechanism of clioquinol: induced plaque reduction.According to this mechanism, clioquinol may activate autophagy inneurons and astrocytes, and thereby help reduce the accumulation ofAβ aggregates. Although the precise mechanism responsible forthe formation of Aβ aggregates is still a matter of intense inves-tigation, recent evidence indicates that abnormal autophagy may be acontributing factor. In AD, the levels of ATG6/beclin-1 in the brainare substantially reduced and the autophagy-related gene ATG8 isdownregulated (Pickford et al., 2008). Hence, various activatorsof autophagy, such as rapamycin and lithium are considered can-didate drugs for treating AD (Sarkar et al., 2009). In fact, a recentreport showed that rapamycin is effective in reducing Aβ loads andimproving cognitive deficits in Tg PDAPP mice (Spilman et al., 2010).The induction of functional autophagy in cultured astrocytes byclioquinol, demonstrated here, is interesting in this context becauseit suggest that this effect of clioquinol might contribute to its anti-amyloid effects in Tg 2576 mice (Cherny et al., 2001).

From a mechanistic standpoint, our data indicate that dynamicmovements of labile zinc across the cell membrane may play a crucialrole in clioquinol-induced autophagy. In previous studies, we havedemonstrated that labile zinc accumulates not only in the cytosol andnuclei of astrocytes under oxidative stress conditions, but also in AVs,including autophagosomes and autolysosomes (Lee et al., 2009). Asshown in the present study, chelation of intracellular zinc with TPENmarkedly reduced the level of autophagy induced by oxidative stress.The only difference may be the source of labile zinc. In the case ofoxidative stress, the source is likely intracellular zinc-binding proteinssuch as metallothioneins (Maret and Vallee, 1998), whereas withclioquinol and other zinc ionophores, the source of zinc is likelythe extracellular compartment. Another important question is themechanism of autophagy inhibition by zinc chelators such as TPEN.Although TPEN reduced zinc accumulation in AVs, our results do notprove that this is the underlying mechanism. However, because TPENblocked the conversion of LC3 to LC3-II in both our previousand present studies, it is possible that labile zinc is involved in thesignaling events leading to AV formation.

Although clioquinol is promising as a treatment for AD, its well-documented side effect, SMON, represents a major concern for its usein humans. Although a vitamin B12 deficiency has been proposed as amechanism, it is not yet clear how clioquinol produces SMON inhumans, and the possibility that clioquinol itself is neurotoxic has notbeen fully investigated.

Our results clearly demonstrate that short-term treatment (24 h)with clioquinol can induce neuronal death at high concentrations(1–10 μM). However, under chronic-exposure conditions, toxic con-centrations may be significantly lower. Although it is unclear whetherthe clioquinol toxicity seen in the present study is relevant to SMON

in humans, the possibility that zinc levels may affect the toxicpotential of clioquinol is intriguing. In this regard, our results andreports from others support the idea that clioquinol mainly acts as azinc ionophore, shuttling zinc ions in and out of cells depending on theintra–extracellular concentration gradients. This action may underliethe zinc-dependent anti-cancer effect of clioquinol observed inhuman prostate cancer cells (Yu et al., 2009). Another intriguingaspect of clioquinol-induced neuronal death is that it occurs at least inpart via the autophagic mechanism, as evidenced by the significantreduction in clioquinol-induced neuronal death in the presence of theinhibitor of autophagy 3MA.

In conclusion, clioquinol or its structural analogues, like otherchemicals, may be a double-edged sword. When present in excess, itmay induce cell death, especially of neurons, by introducing zinc andactivating excess autophagy. However, at an appropriate concentration,clioquinol may have dual beneficial effects on AD pathology. In theextracellular space, clioquinol may chelatemetals from extracellular Aβaggregates, making them more friable for degradation. Entry ofclioquinol-metal complexes, especially clioquinol-zinc complex, intobrain cells increases intracellular zinc levels, promoting effective,concurrent activation of autophagy. This dual mechanism may helpreduce intracellular Aβ accumulation as well as remove toxic wasteproducts or organelles. Hence, while caution is warranted to minimizethe possible side effects of clioquinol, at adequate doses clioquinol mayhave multiple beneficial actions in the AD brain.

Acknowledgments

This work was supported by a Korea Healthcare Technology R&DProject (MIHWAF A084270) funded by Ministry of Health and FamilyAffairs by Ministry of Health and Welfare.

References

Adlard, P.A., Bush, A.I., 2006. Metals and Alzheimer's disease. J. Alzheimers Dis. 10,145–163.

Adlard, P.A., et al., 2008. Rapid restoration of cognition in Alzheimer's transgenicmicewith 8-hydroxy quinoline analogs is associated with decreased interstitial Abeta. Neuron 59,43–55.

Bush, A.I., Pettingell, W.H., Multhaup, G., d Paradis, M., Vonsattel, J.P., Gusella, J.F.,Beyreuther, K., Masters, C.L., Tanzi, R.E., 1994. Rapid induction of Alzheimer A betaamyloid formation by zinc. Science 265, 1464–1467.

Canzoniero, L.M., Manzerra, P., Sheline, C.T., Choi, D.W., 2003. Membrane-permeantchelators can attenuate Zn2+ -induced cortical neuronal death. Neuropharmacol-ogy 45, 420–428.

Cherny, R.A., Atwood, C.S., Xilinas, M.E., Gray, D.N., Jones, W.D., McLean, C.A., Barnham,K.J., Volitakis, I., Fraser, F.W., Kim, Y., Huang, X., Goldstein, L.E., Moir, R.D., Lim, J.T.,Beyreuther, K., Zheng, H., Tanzi, R.E., Masters, C.L., Bush, A.I., 2001. Treatment with acopper-zinc chelator markedly and rapidly inhibits beta-amyloid accumulation inAlzheimer's disease transgenic mice. Neuron 30, 665–676.

Choi, D.W., 1985. Glutamate neurotoxicity in cortical cell culture is calcium dependent.Neurosci. Lett. 58, 293–297.

Colvin, R.A., Bush, A.I., Volitakis, I., Fontaine, C.P., Thomas, D., Kikuchi, K., Holmes, W.R.,2008. Insights into Zn2+ homeostasis in neurons from experimental and modelingstudies. Am. J. Physiol. Cell Physiol. 294, C726–C742.

Cuervo, A.M., Stefanis, L., Fredenburg, R., Lansbury, P.T., Sulzer, D., 2004. Impaireddegradation of mutant alpha-synuclein by chaperone-mediated autophagy.Science 305, 1292–1295.

Ding, W.Q., Liu, B., Vaught, J.L., Yamauchi, H., Lind, S.E., 2005. Anticancer activity of theantibiotic clioquinol. Cancer Res. 65, 3389–3395.

Ding, W.Q., Yu, H.J., Lind, S.E., 2008. Zinc-binding compounds induce cancer cell deathvia distinct modes of action. Cancer Lett. 271, 251–259.

Hara, T., Nakamura, K., Matsui, M., Yamamoto, A., Nakahara, Y., Suzuki-Migishima, R.,Yokoyama,M.,Mishima, K., Saito, I., Okano, H.,Mizushima, N., 2006. Suppression of basalautophagy in neural cells causes neurodegenerative disease in mice. Nature 441,885–889.

Hwang, J.J., Kim, H.N., Kim, J., Cho, D.H., Kim, M.J., Kim, Y.S., Kim, Y., Park, S.J., Koh, J.Y.,2010. Zinc(II) ion mediates tamoxifen-induced autophagy and cell death in MCF-7breast cancer cell line. Biometals 23, 997–1013.

Klionsky, D.J., 2007. Autophagy: from phenomenology to molecular understanding inless than a decade. Nat. Rev. Mol. Cell Biol. 8, 931–937.

Koh, J.Y., Choi, D.W., 1987. Quantitative determination of glutamate mediated corticalneuronal injury in cell culture by lactate dehydrogenase efflux assay. J NeurosciMethods. 20, 83–90.

Komatsu, M., Waguri, S., Chiba, T., Murata, S., Iwata, J., Tanida, I., Ueno, T., Koike,M., Uchiyama, Y., Kominami, E., Tanaka, K., 2006. Loss of autophagy in the

251M.-H. Park et al. / Neurobiology of Disease 42 (2011) 242–251

central nervous system causes neurodegeneration in mice. Nature 441,880–884.

Lannfelt, L., Blennow, K., Zetterberg, H., Batsman, S., Ames, D., Harrison, J., Masters, C.L.,Targum, S., Bush, A.I., Murdoch, R., Wilson, J., Ritchie, C.W., 2008. Safety, efficacy,and biomarker findings of PBT2 in targeting Abeta as a modifying therapy forAlzheimer's disease: a phase IIa, double-blind, randomised, placebo-controlledtrial. Lancet Neurol. 7, 779–786.

Lee, J.Y., Cole, T.B., Palmiter, R.D., Suh, S.W., Koh, J.Y., 2002. Contribution by synaptic zincto the gender-disparate plaque formation in human Swedish mutant APPtransgenic mice. Proc. Natl Acad. Sci. USA 99, 7705–7710.

Lee, J.Y., Friedman, J.E., Angel, I., Kozak, A., Koh, J.Y., 2004. The lipophilic metal chelatorDP-109 reduces amyloid pathology in brains of human beta-amyloid precursorprotein transgenic mice. Neurobiol. Aging 25, 1315–1321.

Lee, S.J., Cho, K.S., Koh, J.Y., 2009. Oxidative injury triggers autophagy in astrocytes: therole of endogenous zinc. Glia 57, 1351–1361.

Levine, B., Klionsky, D.J., 2004. Development by self-digestion: molecular mechanismsand biological functions of autophagy. Dev. Cell 6, 463–477.

Levine, B., Kroemer, G., 2008. Autophagy in the pathogenesis of disease. Cell 132, 27–42.Maret, W., Vallee, B.L., 1998. Thiolate ligands in metallothionein confer redox activity

on zinc clusters. Proc. Natl Acad. Sci. USA 95, 3478–3482.McCarthy, K.D., de Vellis, J., 1980. Preparation of separate astroglial and oligodendroglial

cell cultures from rat cerebral tissue. J. Cell Biol. 85, 890–902.Mizushima, N., Klionsky, D.J., 2007. Protein turnover via autophagy: implications for

metabolism. Annu. Rev. Nutr. 27, 19–40.Pickford, F., Masliah, E., Britschgi, M., Lucin, K., Narasimhan, R., Jaeger, P.A., Small, S.,

Spencer, B., Rockenstein, E., Levine, B., Wyss-Coray, T., 2008. The autophagy-relatedprotein beclin 1 shows reduced expression in early Alzheimer disease and regulatesamyloid beta accumulation in mice. J. Clin. Invest. 118, 2190–2199.

Ravikumar, B., Duden, R., Rubinsztein, D.C., 2002. Aggregate-prone proteins withpolyglutamine and polyalanine expansions are degraded by autophagy. Hum. Mol.Genet. 11, 1107–1117.

Ravikumar, B., Vacher, C., Berger, Z., Davies, J.E., Luo, S., Oroz, L.G., Scaravilli, F., Easton, D.F.,Duden, R., O'Kane, C.J., Rubinsztein, D.C., 2004. Inhibition ofmTOR induces autophagyand reduces toxicity of polyglutamine expansions in fly and mouse models ofHuntington disease. Nat. Genet. 36, 585–595.

Ritchie, C.W., Bush, A.I., Mackinnon, A., Macfarlane, S., Mastwyk, M., MacGregor, L.,Kiers, L., Cherny, R., Li, Q.X., Tammer, A., Carrington, D., Mavros, C., Volitakis, I.,Xilinas, M., Ames, D., Davis, S., Beyreuther, K., Tanzi, R.E., Masters, C.L., 2003. Metal-protein attenuation with iodochlorhydroxyquin (clioquinol) targeting Abetaamyloid deposition and toxicity in Alzheimer disease: a pilot phase 2 clinicaltrial. Arch. Neurol. 60, 1685–1691.

Ross, C.A., Poirier, M.A., 2004. Protein aggregation and neurodegenerative disease. Nat.Med. 10 (Suppl), S10–S17.

Sarkar, S., Ravikumar, B., Floto, R.A., Rubinsztein, D.C., 2009. Rapamycin andmTOR-independent autophagy inducers ameliorate toxicity of polyglutamine-expanded huntingtin and related proteinopathies. Cell Death Differ. 16,46–56.

Schaumburg, H.H., 2000. Experimental and clinical neurotoxicology. Oxford UniversityPress, New York.

Spilman, P., Podlutskaya, N., Hart, M.J., Debnath, J., Gorostiza, O., Bredesen, D.,Richardson, A., Strong, R., Galvan, V., 2010. Inhibition of mTOR by rapamycinabolishes cognitive deficits and reduces amyloid-beta levels in a mouse model ofAlzheimer's disease. PLoS ONE 5, e9979.

Tateishi, J., 2000. Subacute myelo-optico-neuropathy: clioquinol intoxication inhumans and animals. Neuropathology 20 (Suppl), S20–S24.

Tsubaki, T., Honma, Y., Hoshi, M., 1971. Neurological syndrome associated withclioquinol. Lancet 1, 696–697.

Webb, J.L., Ravikumar, B., Atkins, J., Skepper, J.N., Rubinsztein, D.C., 2003. Alpha-Synuclein is degraded by both autophagy and the proteasome. J. Biol. Chem. 278,25009–25013.

Yu, H., Zhou, Y., Lind, S.E., Ding, W.Q., 2009. Clioquinol targets zinc to lysosomes inhuman cancer cells. Biochem. J. 417, 133–139.