Molecular Medicine 4: 665-670, 1998

Clioquinol-Zinc Chelate: A Candidate CausativeAgent of Subacute Myelo-Optic Neuropathy

Jack L. Arbiser,' Stine-Kathrein Kraeft,/ Robert van Leeuwen,3Selwyn J. Hurwitz,, Martin Selig,4 G. R. Dickersin,4Alan Flint,5 H. Randolph Byers,6 and Lan Bo Chen2

'Department of Dermatology, Emory University School of Medicine,Atlanta, Georgia, U.S.A.2Department of Cancer Biology, Dana-Farber Cancer Institute, Boston,Massachusetts, U.S.A.3Department of Dermatology, University of Leiden, Leiden, TheNetherlands4Department of Pathology, Massachusetts General Hospital andHarvard Medical School, Boston, Massachusetts, U.S.A.5Department of Genetics, Children's Hospital, Boston, Massachusetts, U.S.A.6Department of Dermatology, Boston University School of Medicine,Boston, Massachusetts, U.S.A.Communicated by J. Folkman. Accepted August 1, 1998.

Abstract

Background: 5 -chloro-7-iodo-8-hydroxyquinoline(clioquinol) was used clinically three decades ago as an

oral antiparasitic agent and to increase intestinal ab-sorption of zinc in patients with acrodermatitis entero-pathica, a genetic disorder of zinc absorption. Use ofclioquinol was epidemiologically linked to subacutemyelo-optic neuropathy (SMON), characterized byperipheral neuropathy and blindness, which affected10,000 patients in Japan. Discontinuation of oral clio-quinol use led to elimination of SMON, however, themechanism of how clioquinol induces neurotoxicity isunclear.Materials and Methods: We tested the effect of clio-quinol-metal chelates on neural crest-derived mela-noma cells. The effect of clioquinol chelates on cells was

further studied by electron microscopy and by a mito-chondrial potential-sensitive fluorescent dye.

Results: Of the ions tested, only clioquinol-zinc chelatedemonstrated cytotoxicity. The cytotoxicity of clioquinol-zinc chelate was extremely rapid, suggesting that its pri-mary effect was on the mitochondria. Electron microscopicanalysis demonstrated that clioquinol-zinc chelate causedmitochondrial damage. This finding was further confirmedby the observation that clioquinol-zinc chelate caused a

decrease in mitochondrial membrane potential.Conclusions: We demonstrate that clioquinol, in thepresence of zinc, is converted to a potent mitochondrialtoxin. The phenomenon of clioquinol mediated toxicityappears to be specific to zinc and is not seen with othermetals tested. Since clioquinol has been shown to cause

increased systemic absorption of zinc in humans, it islikely that clioquinol-zinc chelate was present in appre-

ciable levels in patients with SMON and may be theultimate causative toxin of SMON.

Address correspondence and reprint requests to: Dr. Jack L.Arbiser, Department of Dermatology, Emory UniversitySchool of Medicine, Woodruff Memorial Building, Rm.5309, Atlanta, GA 30322, U.S.A. Phone: (404) 727-5872;Fax: (404) 727-5878; E-mail: jarbise@emory.edu

Introduction5-chloro-7-iodo-8-hydroxyquinoline (clioqui-nol) has been used orally as an antiparasitic andin the treatment of acrodermatitis enteropathica,

Molecular MedicineC) 1998 The Picower Institute Press

666 Molecular Medicine, Volume 4, Number 10, October 1998

8000

6000

E

-

4000

2000

0

CON CLIO ZN CLIO/ZN

Fig. 1. Cell survival curve of MMAN cells after24 hr of treatment with clioquinol, zinc, combi-nation therapy, or vehicle alone. The number ofcells in the clioquinol-plus-zinc treatment is signifi-cantly different from the values in the other threecolumns. The error bars represent SEM. The asteriskrepresents p < 0.05.

an inherited inability to absorb zinc (1-5). Al-though clioquinol was effective in the treatmentof acrodermatitis enteropathica, oral use of clio-quinol was discontinued as it has been associatedwith subacute myelo-optic neuropathy (SMON),a condition characterized by peripheral neurop-

athy and blindness. Over 10,000 patients in Ja-pan were affected by SMON before clioquinolwas discontinued (2). The mechanism of howclioquinol causes SMON is currently unknown.Multiple theories have been ascribed, includingsynergistic toxicity with pesticides and viral in-fection, but none have been conclusively proven

(6,7).In order to determine the mechanism of

clioquinol toxicity, chelates of clioquinol withcommon ions were prepared, and cytotoxicitywas tested on neural crest-derived melanomacells. Of the metals tested, only clioquinol-zincchelate showed toxicity towards melanomacells. Zinc chelates with other agents, such as

penicillamiine and 1, 1 0-phenanthroline,showed no cytotoxicity. The rapid course ofcytotoxicity seen in the presence of clioquinol-zinc chelate suggested that the target of toxic-ity was the mitochondria. Using mitochondri-al-specific dyes, clioquinol-zinc chelate was

found to cause a rapid decrease in mitochon-drial membrane potential. Because clioquinolhas been shown to cause increased systemiclevels of zinc in humans due to formation of

clioquinol-zinc chelate, and clioquinol-zincchelate is a mitochondrial toxin, clioquinol-zinc chelate is a likely cause of SMON.

Materials and MethodsCell Culture

MMAN is a human melanoma cell line originat-ing in this laboratory, originally derived from alymph node metastasis (8). MMAN cells weremaintained in Dulbeccos modified Eagle medium(DMEM; Sigma Chemical Co., St. Louis, MO)supplemented with 10% fetal calf serum (Hy-clone, Logan, UT), penicillin, and streptomycin.Cells were subcultured weekly at a ratio of 1:4. Astock solution of 1 mg/ml 5-chloro-7-iodo-8-hy-droxyquinoline (Sigma) was prepared by rec-onstituting the drug in absolute ethanol or di-methylsulfoxide. Cells were grown in DMEMsupplemented with 5%o fetal calf serum. Zincchloride, calcium chloride, magnesium sulfate,and copper sulfate were reconstituted as a50-mM stock solution in water and filter steril-ized prior to use. Solutions of clioquinol andaqueous ion solutions were mixed together priorto addition of media to allow for optimal chelateformation.

Cell Survival Assay

Ten thousand MMAN cells were plated in 24-well dishes. Twenty-four hours after plating, me-dia were replaced with media containing clioqui-nol alone at a concentration of 7.5 ,ug/ml, zincchloride at 50 ,uM, the combination of bothdrugs, or vehicle alone (control cells). After 24 hrof treatment, cells were counted using a Coultercounter (Hialeah, FL). Each experimental treat-ment was performed in triplicate.

Electron Microscopy

Cells for electron microscopic studies weregrown in 35-mm plastic petri dishes and treatedwith test compounds for 24 hr. All steps wereperformed at room temperature unless other-wise noted. The cells were fixed for 45 min in K(IIfixative (2.5% glutaraldehyde, 2% formalde-hyde, 0.025% CaCl2 in a 0.1 M cacodylate buffer,pH 7.4) and washed in buffer. The cells werethen post-fixed with 1.2% osmium tetroxide incollidine buffer for 1 hr, dehydrated in gradedethanol solutions, stained en bloc with 2% ura-nyl acetate, infiltrated with ethanol/epon mix-

J. L. Arbiser et al.: Clioquinol-Zinc Chelate and SMON

Fig. 2. Electron microscopy of MMAN cellstreated with clioquinol and zinc. (A) Controlmelanoma cells (X 5800). The inset demonstratesmicrovilli present in the control melanoma cells(X 12,000). (B) Melanoma cells following treatmentwith 150 ,uM zinc chloride (X6000). (C) Melanomacells following treatment with 7.5 ,ug/ml clioquinol.The asterisks in the central photograph (X5300) in-dicate dilated endoplasmic reticulum. The arrows in

the upper inset (X 15,000) demonstrate focal adhe-sion plaques, and the lower inset (X 14,300) showsincreased melanosomes seen as a result of clioquinoltreatment. (D) Melanoma cells following treatmentwith 7.5 gg/ml clioquinol in combination with 150,uM zinc chloride (X7200). The inset (X 15,600)demonstrates damaged mitochondria; L, lipid drop-lets seen in necrotic cells.

tures, and embedded in 100% epon. Represen-tative sections were stained with Sato's lead andexamined with a Phillips 301 electron micro-scope at 40 kV (9).

Mitochondrial Membrane Potential

Prior to measurements, 5 X 104 cells were platedin 8-well coverglass tissue culture chambers(Nunc Inc., Naperville, IL). Cells were loadedwith the potential-sensitive dye MitoTracker RedCMXRos (50 nM, Molecular Probes, Eugene,OR) in DMEM containing 5% fetal calf serum for30 min (10). Cells were treated with vehicle,clioquinol alone (7 [kg/ml), zinc alone (50 ,tMZnCl2), and clioquinol-zinc (7 ,ug/ml clioqui-nol/50 ,M ZnCl2) at 370C for 30, 60, and 90 min.At the end of incubation, chambers weremounted on the microscope stage of an inverted

confocal laser scanning microscope (LSM41O0,Zeiss, Gottingen, Germany) equipped with anexternal argon-krypton laser. Images from rep-resentative fields of the different treatment con-ditions were taken under identical contrast andbrightness settings to allow for semiquantitativecomparison of the mitochondrial fluorescenceintensities. Images were printed with a Fujix Pic-trography 3000 color printer (Fujifilm, Tokyo,Japan) using Adobe Photoshop software (AdobeSystemis, Mountain View, CA).

Statistics

Significant differences between two groups weredetermined using an unpaired, two-tailed Stu-dent's t test. Results are expressed as the mean ±SEM.

667

668 Molecular Medicine, Volume 4, Number 10, October 1998

Fig. 3. Microscopic analysis of mitochondrialmembrane potential in MMAN cells. Cells wereloaded with the mitochondrial potential-sensitivedye MitoTracker Red CMXRos and incubated withvehicle alone (DMSO) (A), 50 ,uM ZnCl2 (B), clio-quinol (7 gtg/ml) (C), or clioquinol/ZnCI2 (7 ,ug/ml

clioquinol/50 ,uM ZnCl2) (D) for 90 min. Images ofrepresentative fields were recorded under identicalconditions. Color bar displays gray levels (relativefluorescence intensities) that correspond to colors.Bar, 25 ,rm.

ResultsClioquinol-ion chelates were formed with ionscommon in the intestine. These ions includezinc, calcium, magnesium, and copper. Onlyclioquinol-zinc chelate demonstrated apprecia-

ble cytotoxicity to melanoma cells, and this tox-icity appeared to be at least additive, as it oc-

curred at a concentration at which zinc alone or

clioquinol alone had little cytotoxicity (Fig. 1).Chelates.of clioquinol with calcium, magnesium,or copper did not cause appreciable cytotoxicity,nor did zinc in the presence of other chelatorssuch as penicillamine or 1,10 phenanthroline

(data not shown). The cytotoxicity observed bytreatment with clioquinol-zinc chelate was irre-versible when the melanoma cells were replatedin the absence of chelate.

In order to further study the morphologicchanges induced by clioquinol, zinc, and thecombination of both compounds, electron mi-croscopy of MMAN cells was performed. Thecombination of zinc and clioquinol caused mito-chondrial swelling, an increase in melanosomes,lipid droplet accumulation, hydropic vacuoliza-tion, and nuclear changes consistent with necro-

sis (Fig. 2). This cytotoxicity took place at doses

- ()

- IO()

I ,.i- I _i

- 20()

- 255

.-l

I

I9

I

l

J. L. Arbiser et al.: Clioquinol-Zinc Chelate and SMON 669

of clioquinol and zinc which by themselves werenot cytotoxic, implying at least additive toxicity.

The rapid onset of action of clioquinol-zincchelate and the changes visible on electron mi-croscopy suggested that the target of this chelatemight be the mitochondria. In order to confirmthis, melanoma cells were incubated in the pres-ence of a mitochondrial potential-dependentdye. Cells incubated in the presence of clioqui-nol-zinc chelate showed nearly complete loss ofmitochondrial potential after 90 min of incuba-tion (Fig. 3). Cells incubated in the presence ofvehicle alone, clioquinol alone, or zinc aloneshowed no loss of mitochondrial potential(Fig. 3).

Discussion8-hydroxyquinolines are compounds that havehad wide oral and topical use. Oral clioquinol hasbeen used as an antiparasitic agent and for thetreatment of acrodermatitis enteropathica, a ge-netic disorder of zinc absorption (1-6). Oral useof this agent has diminished in recent years be-cause of the epidemiologic association of oralclioquinol with subacute myelo-optic neuropa-thy (SMON), a neuropathy associated with par-esthesias and gait disturbances, as well as retinaltoxicity (1-6). Despite intensive study, themechanism of how clioquinol causes neurotox-icity has not been elucidated.

In order to cause neurotoxicity, clioquinolmust be converted to a lipophilic compound thatcan be orally absorbed and can enter neuronaltissue. Metabolism is an unlikely pathway asclioquinol is metabolized to its glucuronide andsulfate, which is not lipophilic and is readily ex-creted (1 1, 12). The compound must also demon-strate greater toxicity to intact cells than clioqui-nol itself. Magnesium-clioquinol chelate hasbeen reported to cause mitochondrial uncou-pling in cell-free systems (13,14), but we ob-served no cytotoxicity of magnesium-clioquinolchelate in intact cells. Finally, the clioquinol de-rived toxin must be present in humans. The che-late of clioquinol with zinc ions fulfills all of thesecriteria. Clioquinol-zinc chelate shows rapid cy-totoxicity to human cells and is lipophilic. Zincions have been shown to alter the conformationand to inhibit the biologic activity of neuropep-tides related to neuronal survival, includingnerve growth factor (15). The combination ofmitochondrial toxicity observed in this study andthe inhibition of nerve growth factor activity

seen in the presence of zinc makes zinc a likelyculprit in SMON. Clioquinol-zinc chelate hasbeen demonstrated to be present in humans byvirtue of correction of the zinc deficit in acroder-matitis enteropathica, and it has been demon-strated to facilitate uptake and distribution ofzinc to neuronal tissues (11,16). This is the firstreport showing that clioquinol-zinc chelate,which was of benefit in the treatment of zincdeficiency, is also a potent mitochondrial toxin.Thus, clioquinol-zinc chelate fulfills all of thecriteria of a cause for SMON.

AcknowledgmentsJ. L. A. was supported by grants from the Dennatol-ogy Foundation, Howard Hughes Medical Institute,the Thomas B. Fitzpatrick Research Award from theKAO Corporation, and grant R03AR44947 from theNational Institutes of Health.

References1. Tsubaki T, Honma Y, Hoshi M. (1971) Neurolog-

ical syndrome associated with clioquinol. Lancet 1:696-697.

2. Oakley GP. (1973) The neurotoxicity of the halo-genated hydroxyquinolines. JAMA 225: 395-397.

3. Nakae K, Yamamoto S, Igata A. (1971) Subacutemyelo-optico-neuropathy (SMON) in Japan. Acommunity survey. Lancet 23: 510-512.

4. Sturtevant FM. (1980) Zinc deficiency, acroder-matitis enteropathica, optic atrophy, subacute my-elo-optic neuropathy and 5,7-dihalo-8quinolin-ols. Pediatrics 65: 610-613.

5. Moynahan EJ. (1974) Acrodermatitis entero-pathica: a lethal inherited zinc deficiency disorder.Lancet 2: 399-400.

6. Toyokura Y, Takasu T. (1975) Clinical features ofSMON. Jpn. J. Med. Sci. Biol. 28S: 87-99.

7. Hanakago R, Uono M. (1981) Clioquinol intoxi-cation occurring in the treatment of acrodermati-tis enteropathica with reference to SMON outsideof Japan. Clin. Toxicol. 18: 1427-1434.

8. Vink J, Thomas L, Etoh T, Bruijn JA, Mihm MC Jr,Gattoni-Celli S, Byers HR. (1993) Role of beta-Iintegrins in organ specific adhesion of melanomacells in vitro. Lab. Invest. 68: 192-203.

9. Rossouw DJ, Cinti S, Dickersin GR. (1986) Lipo-sarcoma: an ultrastructural study of 15 cases.Am. J. Clin. Pathol. 85: 649-667.

10. Poot M, Zhang YZ, Kramer JA, et al. (1996) Anal-ysis of mitochondrial morphology and functionwith novel fixable fluorescent stains. J. Histochem.Cytochem. 44: 1363-1372.

11. Weismann K, Knudsen L. (1978) Effects of peni-cillamine and hydroxyquinoline on absorption of

670 Molecular Medicine, Volume 4, Number 10, October 1998

orally ingested 65zinc in the rat. J. Invest. Dermatol.71: 242-244.

12. Ohshima N, Kotaki H, Saitoh Y, Nakagawa F, TamuraZ. (1989) Sex difference of the metabolic disposition ofclioquinol in rats. J. Pharmacobiodyn. 12: 371-377.

13. Inouye B, Ogata M. (1979) Effect of chinoform onthe function of biological membranes. Physiol.Chem. Phys. 11: 49-57.

14. Hagihara M, Yagi K. (1975) Effect of albumin onuncoupling of oxidative phosphorylation by

chinoform in rat liver mitochondria. Experientia31: 1069-1070.

15. Ross GM, Shamovsky IL, Lawrance G, et al.(1997) Zinc alters conformation and inhibits bio-logical activities of nerve growth factor and relatedneurotropins. Nat. Med. 3: 872-878.

16. Tjalve H, Stahl K. (1984) Effect of 5-chloro-7-iodo-8-hydroxyquinoline on the uptake and dis-tribution of nickel, zinc, and mercury in mice. ActaPharmacol. Toxicol. 55: 65-72.