Characterization and Subcellular Localization of Human Pmel17/silver, a 100-kDa (Pre)Melanosomal Membrane Associated With 5,6,­ Dihydroxyindole-2-Carboxylic Acid (DHICA) Converting Activity

Zang H. Lee,~ Ling Hou, Gisela MoellmalUl,* Elizabeth Kuklil1ska,* Kathleen Al1tol;t Malcolm Fraser,i" Ruth Halabal1, * and Byoung S. Kwon Department of Microbiology and Immunology, and Walther Oncology Center, Indiana University School of Medicine, IndianapoLis, Indiana; llDepartmcnt of Oral Microbiology and hnmullology, Chosun University School of Dentistry, Kwang joo, Korea; *Department of Dermatology, Yale University School of Medicine. New H aven. Connecticut; and tDepartmcnt of Biolol,';cal Science. University of Notre Dame, South Bend, Indiana, U.S.A.

Pmel17 is preferentially expressed in pigment cells in istry, the antigen was localized to the limiting mem­ a tnanner suggestive of involvenlent in melanin bio­ branes of pretnelanosomes and presumed premelano­ synthesis. The is identical to the silvey (si) pig­ genic cytosolic vesicles and, to a minor extent, in the tnentation locus in mice. We now have produced a premelanosomal matrix. In an ill vityo assay, both the recombinant glutathione-S-transferase-human Pmel natural and the recombinant Pmel 17 accelerated the 17 fusion protein and raised polyclonal antibodies conversion of DHICA to melanin. This activity was against it to confinn the ultrastructural location and inhibited by the anti-Pmel 17 polyclonal antibodies, presumed site of action predicted by the deduced indicating that the acceleration of DHICA conver­ primary structure of Pmel 17/silveY, and to authenti­ sion by the natural protein is genuine and cannot be cate the specificity of the DHICA converting func­ due to contaminating complexed . We sug­ tion as inherent to the silvey-locus protein. Full­ gest that ill situ Pluel 17/silvev is a component of a length Pmel 17 eDNA also was produced in insect postulated premelallosomal/melanosomal complex cells in a baculovirus expression vector to ensure that of melubrane-bound melanogenic oxidoreductive en­ activity did not originate from a co-precipitated pro­ zymes and cofactors, in analogy to the electron transfer tein. Natural hPmel 17 from human cells chain in ntitochondria. KI!J' woyds: melalloma a II tige 11/ has an approxintate tnolecular size of 100 kDa. By ",elallocyte/tyl'osirlase-yelated p,'oteills/NK1-beleblHMB 45. itnmunoperoxidase electron microscopic cytochem- ] l,west Devmalo1106:605-610, 1996

//lei 17, originaUy isolated from a human foreskin thesis distal to tyrosinase (Kwon et nl, 1987). Like tyrosinase Pmel melanocyte Agtll cDNA expression Ijbrary, is spc­ 17 tnRNA expression, in murine Cloudman S-91 and human citicaUy produced in melanocytes and maps near the melanotic melanoma ceUs is induced by agents that increase sillier coat color locus on mouse 10 and melanization (Kwon et nl, 1987). On the basis of sequence homol­ to region q12- q13 on human (Kwon ogy, we have proposed that at least two gene families regulate Pet ai, 1991).' [n normaJ human melanocytes, Pmcl17 mRNA levels m e lanin biosynthesis (Kwon, 1993): the cyrosinase gene family, correlate with melanin content better than does expression of whjch includes tyrosinase, TRP1/gp75, and DOPAchrome tau­ tyrosinase, the key enzyme of m e lanogenesis, suggesting that the tomerase or TRP2; and the Pmel 17 gen e family, wluch includes silver protein functions as a positive regulator of melanin biosyn- Pmel 17/gp100 (see below) of mammalian melanocytes, MMp115, a chick en melanosomal matrix protein (Mochii el ai, 1991), and Manuscript received September 2, 1995; revised November 13, 1995; Rl'E1, a bovine retinal pigment epitheljal protein homologous to accepted for publication November 20, 1995. Pmel 17 (Kjm and Wistow, 1992). In its general structure, Pmel 17 Reprint requcsts to: Dr. Byoung S. Kwon, Deparlll1 CIlt of Microbiology, is similar to tyrosinase, TRPl, and TRP2 inasmuch as it possesses a Indiana University School of Medic inc, 635 Barnhill Drive. Indianapolis, IN putative transmembrane domain and a cytosolic tail (Kwon et ai, 46202-5120. 1991); its amino acid sequence is similar to that of MMp115, with A bbreviations: AcNPV, A lliograplw calijomia nuclear polyhedrosis bacu­ the exception that the latte r has no putative tnU1smembrane lovirus; GST-hPmel 17, glutathionc-S-transfcrase-human Pmel 17 fusion domajn. protein; Sf-21 , SpadoJ'lerajillgiperda-21. R ecently, we demonstrated that PIliI'I 17 from si lsi nuce contains I Kim K-K, Heng HHQ, Slti XM, Tsui L-C, Lce ZH, YOUL1 13S, Pickard RT, Kwon BS: Genomic organization and FISH mapping of human Pm cl an (A) insertional mutation in the putative cytosolic tail, indicating 17, the putativc silver locus. P(glllelli Cell Res (iJ1 press). that Pmel17 protein is indeed the product of the sillier locus (Kwon

0022-202X/96/S10.50 • Copyright © 1996 by The Society for Invcstigativc Dermatology, Inc.

605 606 LEE ET A L TH E JO URN AL OF INVESTIGAT IVE DER.MATOLOGY ct aI , 199 4; Kwon et aI , 1995). StiU in question, however , are the precise subcellular loc ation and func tio n of Pme l 17. 1 2 3 4 5 kDa MATERIALS AND METHODS

Production of HUlnan Ptnel 17 in Bacteria and Insect Cells For expressio n in bacteria , a midportio l1 of human Pmcl 17 cDNA encoding amino acids 141-435, spanning the repeti tio n m o tif (Kwo n ci aI, 1. 991), was fu sed in fr ame with the glutathio ne-S-transferase (GST ) gene using the pGEX vector (Pharmacia, Piscataway, NJ). The fu sio n protcin GST -human -106 Pmel 17, expressed in Esell erirhifl co li strain T op 1 (Stratagene, La Jolla. CAl , was purified by all-ini ty chromatography over glutathione beads and used to raise polyclo nal anti-Pmel! 7 antibodi es (sec below). A bacu! o virus express io n vecto r containing the full-size hPmcl 17 was 80 constructed fo r producti on of a full-l ength rhPmel 17 in no nmelanogeni c insect cells exprcssing no o ther melanogeni c pro tein . A 2 .0-kb EcoR 1 fragment of hPmcl 17 cDNA encoding the complete protein was inserted into the EcoR1 site of the PVL 1393 vecto r (a gift from Dr. Max Summers, Tex

was detected in nonpigment cell s, such as !-leLa cells or THP-l presence of immunopurified native Pmel ] 7. Because it facili tates monocytes. Some aitinity-purified fractions were resolved as two the formation of melanin and is located in the limiting membranes reactive bands: the 100-kDa b;lI1d and another of 52 kDa (data not of known melanogenic subcellular organell es, Pmel 17 may be a shown). T he small er protein could represent a degradation product constituent of a postulated membrane-bound melrulogenic complex of Pmel :17, but more like ly is anti-Pmel 17 JgG heavy chain leached of oxidoreductive enzymes (Orlow el ai, 1994) that include tyrosi­ from the column. nase (A/billo or c-locus protein) and the tyrosinase-related proteins T IU~ l (gp7SIIJ-locus protein Oacksol1, 1988; Halaban ruld Moell­ Subcellular Localization of Hun~an PInel 17 The abundant m~nn, 1990; Tomita ci al. 1991]) and TRP2 (DOPAchrome distribution of Pmel 17 gene product in human m e lanocytes is tautomerasels laty- locus protein [Krompo uzos et ai, 1994]), in anal­ illustra ted in Fig 2 . We used tyrosinase-positive albino meh1l10cytes 3 ogy to the electron transfer chain in mitochondria. Matrix com­ to avoid interference by melanin with both the accuracy of ponents bound the an ti -hPme l 17 antibodies in a distribution localization and the clarity of visualization of thc immunoperoxi­ consistent with vesiculog lobular bodies Oimbow and Fitzpatrick, dase reaction product. The ul trastructure of the amelanotic mela­ 1994). nocytes in a control preparation that was incubated without the T he m embrane localizations confirm predictions based on the primary antibodies is shown in Fig 2a. T he cell s contailled deduced ruluno acid sequence of Pmel ] 7 (Kwon el aI, 1991). This well-fon11ed premelanosomes of the e umelanosome type, with sequence calls for a glycoprotein with a transmembrane domain and l.irrtiting membranes devoid of peroxidatic reaction product 01' a cytosoli c tail, with the majority of the glycosylated chain extend­ melanin. T he myriad small electron-de nse cytoplasmic vesicles may ing into an extracytosolic space, such as a vesicular lumen (e.g., be microperoxisomes, with peroxidase activity due to catalase m e lanosome), or, in the case of vesicular exocytosis, an extracel­ (Novikoff el aI, ] 973). Vesicles of the same size in controls not lular space where the protein may become antigenic. In f.1Ct, a incubated for peroxidase activity were devoid of such contrast (not protein analogous to Pmel 17, referred to as gpl00 (Adem a el ai, illustmted). Somc of the micropel'oxidatic vesicles had been inter­ 1994) and recognized by m onoclonal ;l11tibodies NK] I beteb, nalized by premelanosomes or were present in multivesiculate HMB-45, and HMB-50 (Schaumburg-Lever et ai, 1991; Adema el bodies. aI, 1993; Taatjes ct aI, 1993), has been identified as a potent Bound polyclonal ;l11ti-Pmel 17 antibodies, on the other hand tumor-derived antigen in m elanoma patients of the human leuko­ (Fig 2b), were localized to the limiting membranes of the (pre-) cyte antigen-A2 haplotype (Cox el ai, 1994; K awakami el ai, 1994). melanosomes and to cytoplasmic vesicles, many of which were of Pme l 17 and gplOO are translation products of alternatively spliced the coated va l'iety and larger than the presumed microperoxisomes. mRNA (Adema el ai, 1994). T he membrane locali zation ofPmel1 7 T he unreactive membranes of mitochondria and endoplasmic re­ agrees w ith the binding sites ofNKl/beteb (Winder el al. 1994). an ticulum served as built-in contro ls . antibody raised against a membrane preparation from a human The reaction product was located at the inte rnal surf.1ce of the m etastatic melanoma (Vennegoor ci ai, 1988), and with the immu­ (pre)melanosomal and vesicular membranes (Fig 2c), consiste n t nogold data obtained with rabbit polyclonal antibodies raised with an epitope or epitopes along the N-terminal extracytosolic against a m elanosomal fraction from murine B16 melanoma cell s extension of Fmel ] 7, which contains the presumptive g lycosyla­ (Orlow cl al. 1993b; Z hou et aI, 1994). T he antigenic sites were tion sites. Again, a few (pre)melaJlosomes also contained reaction locali zed to the extracytosolic domain of the membranes, consistent product in the form of g lobules within the matrix. Preincubation with an epitope on the postulated intramelaJlosom aJ glycosylated with H 0 , in an attempt to inhibit selectively native catalase, 2 2 N-terminal extension of Pmel 17. In all cases, the otientation was markedly diminished antibody binding but unde rscored the mem­ identical to one obtained with anti-PEP2, an antiserum specific for brane and matrix-globular locali zatio ns (Fig 2d). the predicted luminal portion of the melanosomaJ membrane DHICA Converting Activity of Pmel 17 Because Pmel 17 protein TRP1 (Orlow el aI, 1993). Our data, as supported by others, from pigment cell s may be complexed w ith other melanogenic including the immunogold data of Orlow ci al (1993b), are proteins (Orlow el aI, 1994) and during immunopurification may consistent with the sillier-l ocus product being a I)() II a fide membrane tag along tyrosin ase, TR.P1, and TRP2, and perhaps others not yet protein. known, recombinant hPme l 17 expressed in the no n melanogenic T he impo rtance ofPmel 17 to pigmentation is indicated from the Sf-21 insect cell s was immunopurified and tested for DHICA-to­ chan ges in melanocyte survival in mice carrying the silsi (sillier) melanin converting 'lctivity. In contrast to parental cell s or cells mutation (Quevedo el aI, ] 98]). T he sing le nucleotide insertion is infected with an irrelevant recombinant baculovirus, the iJllmUnO­ expected to alter the last 24 amino acids of the original C-terminus purified baculovirus-ex"jJressed Pmel 17 tested positive (Fig 3a). and length e n the prote in by 12 residues. T hese changes are likely to This activity was protein dose dependent. Furthermore, fractions misguide the mutant protein away from melanogenic membra~les containing immunoaffinity-purified native hPmcl 17, as identifi ed because they bring about the loss of a melanosomaJ targetmg by we tern blot analysis (Fig 1), were pooled and assayed for sequence located in the C- tenninal extension ofPmei 17 and other DHTCA-to-melanin converting activity in the presence and absen ce melanogeluc membrane proteins (Vijayasaradh.i el aI, 1995). T he of anti-Pmel 17 antibodies. The DHICA converting activity was premature death of follicular melanocytes in the sillIer mice sugges~s blocked by the antibodies (Fig 3b). that Pmel-17-accelerated conversion of DHICA to melruun IS required for the speedy removal of cytotoxic intermediates derivir~g DISCUSSION from earlier oxidation-reduction reactions in the melanoge111c We have shown that human Pmcl J 7 I sililcr, a protein of 100 kDa, pathway. Genetic support for this notion comes from the observa­ is located in vesicular and (pre)mclanosom;ll limiting m embranes tion that in Ago/./li lsihw · and in Ago llli l yel/ol" ls il,, (~ r mice, si lvering and that ill Ililro , this protein accelerates th e conversion of DHICA d ecreases rather than increases with age (Si.lvers, ] 979). AgO lIl1 and to melanin. T he enzymatic activity was detected in immunopre­ Agollii/yel/olli mice sufi:er .fi·om a mutation that results in a blockade cipitates of melanoma cell extracts using polyclonal anti-Pmel 17 of the melanocytic receptor for melanocyte-stimulating hormone antibodies and in recombinant Pmel 17 produced in nonmelano­ (melanotropin) (Lu et ai, 1994), thereby decreasing expression of genic insect cell s. T he latter ensured that the DHICA conversion tyrosinase (Halaban el aI, 1984) and Pmel 17 (Kwon et aI, 1987). It activity did not originate from a co-precipitated melanogenic is reasonable to postulate that in the above mutants, with a protein. Our data complement those of C hakraborty el al, 2 who constitutional decrease in eumelanogenic cytotox;c intermediates, a reported superoxide-dependent polymerization of DHICA in the

2 Chakraborty AK, Platt JT, Kim K-K, Kwon 135, Bennett DC, Pawelek JM: Po lymctization of 5,6-dihyd roxyindolc-2-carboxyli c acid to melanin by J As suggested by Moellm3nn, XIV International Pigmcnt Cell Confe r­ the Pmel 171sil ver loclI s prote in . Ellrop en ll J Bioc!t clII. 1996. ill press. "net:. Kobe, J apan , 199(). 608 LEE ET AL THE JO URN AL OF INVESTIGAT I V E DERM AT O LOGY VOL. 106. NO.4 APRIL 1996 C H AI1-ACTERIZ ATION O F T H E HUMAN S ILtiER PROTEIN 609

pattern similar to that in sil si mice (Queved o cl ai, 1981). The Bit mutation consists of a single amino acid substitution near the a N-terminus ofTRP1, close to the putative signal-peptide cleavage 0.20 site Oohnson and Jackson, 1992), and may adversely affect the insertion of the n ascent protein into the endoplasmic reticulum, and hence melanosom es. The gene tic evidence that two l11 elanosol11al proteins, which for different reasons may have difficulty entering Ec a 0.15 premelanosomes, produce similar phenotypes strengthen s the con­ a clusion that both have similar functions. N evertheless, the relative "

Diseases Resea rel, Cellter Grnllt A R41942 /C) RE T igelanr, Ynle Skill Discnscs R cscnrc/, Ce'ller (E K , GM, RH). We tlmllk Dr, Max S 'IIIIII, ers, Texns A&M Ulli,'crsi')" jiJr p"ollidillg lir e PilL 1393 IIector nlld tlr" SJ:2 J illSecl cells, nlld Drs. AsllOk C lmkrabOll), alld JOIrIl PnI/,c/ck, Ynle UlliIJw'il)" for DHICA nlld ,lre protorol.fl,r tl,e DHICA colIl'crsioll nssn)'. ' fie gml,j idl), n[kllOldellg" tir e tec/Illiml! o 0.5 1.0 1.5 2.0 2.5 3.0 selTetnrinl nssistO/,re of Mr. Riclmrd T. Pick(//'d, Ms. S IISil lI Eicldllllrz, Dr. Y ll s ltnll.~ Xie, nlld Ms. A lldrey C OI"SOIl. nil 4 IIIdinrm Ullillersit)'; nlld Mr. Jark ScI""'iber of TIME (hr) Ynll' U"illersi,y.

Figure 3. Facilitation of conversion of 5,6-dihydroxyindole-2-car­ boxylic acid (DHICA) into soluble melanin by recombinant as weU as native human Pmel 17. a) DHICA converting activity of immunopu­ R E FERENCES rified recombinant human Pmel 17 (rhPmel 17 ; 0 ). whose synchesis was Adcl1w GJ. de Uocr AJ. vall't l-IullcllIIar R. Dcnijll M. Ruiter DJ . VogcJ AM. ,Figd.o r directed by a recombinanc baculovirus in SJ~2 1 insect ceUs; idc ntical CG: Md;lII ocytc lineage-specific :1 llt"igCIlS recogni zed hy mo noclomd antibodies columns and procedures were used with control wi ld-type virus-infected NK I-bercb. HMU-50. :lIld HMI3-45 arc cncoded by a sin gle eDNA. A lii) POlh,,1 Sf-21 cell s (0 ) and ,'if-2 1 cells infected with an unrelated re combinant 1~ 3 :1 579- 1 5HS . 1993 bacul oviru s (.). Dinl/w"ds. buffer alone. b) DHICA converting activity in Ade l1l;} CJ . dc Uocr AJ . Vogcl AM. Loell ell \ X/ A. Figdor CG: Mo lecular ch:tr3cl,criza­ ti011 o f the melanocyte lineagc-spccitit: antige'l g-p I 00 . .1 Bioi Chi'lll 269:20 126- il11m ull opurified nati ve hPmcl 17 from YU SITl melanoma cells . • . hPmel 20133. 1994 17; .0 hl'mel1 7 plus anti-hPmel 17 antibodies: 6 , buller alone. Concen­ Cox AL. Skipper J. C hcll Y. Hende rson R. Darrow T. Sll:IbanowitzJ. Engelhard V. trations as in Mnterinls tII/(1 Metlwds. Each poin t represents mea n ::': SD. Hunt D, Sling-luff" C: Iden tifi c:tli OIl of a peptide recognizcd by five mc lanom .t­ specifi c cytotoxic T cell lines. Science 264 :7 16-7 18. 1994 I-Ialahan R . Lnll gdo ll R • .Birchall N. C UOIl O C . Baird A. ScOlt G. Moclll1l :1I11l G. McGuire J: U;lsic fihro b last grow th (;\ctor ofker;\tillocytcS is ;1 11<1 ru rOl l mitogen for superimposed sillIer mutation IS not as m elanocytotoxic as in norl11all11c1anocytcs.j Cdl BiI,1 107: 1611-1(, 19. 198R eumelanotic, wild-type mice. Halaball R . Moellmanll G ; Murine and huma n I,- Iocus pi gmclwttion gcncs encode a glycopro lcin (gp7S) with cat;!l:lsc activity. Pr"r NlJll l iI"mf S'r; USA 87:48(1)-.18 13. The sillier mutatio n has long been known to afFe ct coat color 1990 dilution, 1110st noticeably when expressed o n a Bi b (genetic locu s I-Ialaban ll. Po m erantz SI-I, Marshall S. Lerner AB: T YI'osin;1sc :lcri viry an d abund:II H.·c encoding TR.]>1) background (Silvers, 1979). Pmel 17 l11ay, indeed , ill C loudm :ln m clal1 01nn cells . .'1 '1'11 [ji()r/I('III B;opl,)),( 230:383-387. 198-1 . complement most strongly w ith TL~P'l, the l11 e lanosomal O1el11- .Jackson I.J: A e DNA encoding: tyrosilt:lse-re hucd protein nt:l pS to rhe browlt loctl s til m ouse Ipublished c r r;ltUHl. J'f(I( N atl Arlld Sci US; I B6:997. 1995 I· Pror N a,1 A ~' a d brane protein that appears to have bo th DH ICA synthetic activity Sri USA 8S:'l3 n-43%. I'IH8 (dopachrome tautomerase [Winder cl ai, '1993, 1994]) and its own Jilltbow K. Fitz patrick TB: C h:tr:tL"lcriz:lti(Ht Of:l11 CW l1t Cl:tlt Oso l1lai srnu.: t~lr a l COl1lp O­ DHICA conve rting activity (DI-IICA oxidase 1J0hnso n and Jack­ n c nt- t1tc vcsic uloglo bular Il lldy-hy CotlV c lttio n a l tr:U1 Slllissio n. Iltgh-volmge. son, 1992;Jimcnez-Cervantes cl ai, 1994; Kobayashi e/ al. 1994]), as and scanning e lectro n microscopy . .1 Ultmsll1lrt UI'S 4R: 26 1)-~ 83 . 1974 .Jimcttcz-Ccrv:1I1tcs C. Solano F. f{ o h;, yO\ shi T. U rahc K. Hcarlng ~'J . Lozano J A. well as being associated with catalase activity (I-Ialaban and Moe ll­ G:trci a- l3orrb n J C: A ne w e ll zV lIlat ic 1i.1IlCtiOll ill lht! I11 c lallog:c lII C p:tdtway: rh e mann, 1990). The genetically dominant mouse TRPI muwtion 5.6-dih ydroxyindo le-2-carbox)dic acid oxidase ac(ivil'Y ofryrosinasc-rc latcd pro­ I Light (B ,) causes premature m elanocyte deatJl in a phenotypic '"c;I1-1 (TR!'I).) BioI C iI .. ", 2 ('~: 1 7~93- I R O(J1. 1 99 ~

Figurc 2. Subcellular localization of hPnlcl 17 protein to premelanosOlnal and prcsulnptivc prcll1clanogcnic vesicular membranes. A. control in cubation (anti-PmcI17 antibodics oll1itted). T he pre l11 elanosol11cs and their limiting mcmbranes arc devoid ofinllll""opcroxid ase reaction product (cxal11ple between lm:,!c arrollls). The In yriad srnall electron- dense vesicles l11ay b e 111i c r opc r ~x isolllCS containing: native peroxidatic acti vity. due to carala.sc. So nIc of these vesicles have b een illtcl'naiiz cd by prcnlcial1 oS0 1l1~ S (small ( 11,."t~d atT(JIIJS) and Inultivesicular bodies (sw tl ll c",.,,,:d nn·lIf/!s ). B. lInnll1~l O r ea~t lv c preparation. Allti-Pmcl 17-inllnullopcroxidase product outlincs the lillli ting membranes of premclanosornes (I"' :~l ' nn''''''s) and prclncl,mogelllc veSIcles (cxanlplcs at n".'!"1tll' arrows). The latter arc larger th"111 the. preSt1l11ed micro pcroxisol11CS. III;, Illitocho l1dria: 1111 , nuc.:! e:lr I11 c nlbr i ll~l e , C. detail frol11 immunoreactive preparation B. The 'mti-Pmcl-17 immunopcroxida sc product is found on t.he extracytoplasmic (intcrnal) surf;,cc 0 1 (p re)mclanosomal limitin g 111CI11branes and on thc nlelnbranes o f prcslIl1wd pretllcianogcnic vesicles (OpCII an'I'/I)s) . Sontc of the reactive prcrnclanosomcs. have p.c r~xlda sc­ po ;tive vesicl es within thcir matrix (sIIInll (/I,."ed IIIT'''"S); others appear as bein g approached by cytoso lic vesicles (I",:~c b,.", OIm",s). D . antIbody btndmg was diminished by preincubation of thc ccll s with H 20, (to ina ctivate intrinsic peroxidase due to ca ta lase) , but the localization of anti-Pmel 17 to (prc)111 clanosolllal IlI c lnbrancs and vcsicul ogloblilar b odit.:s persisted . Settle bflrs, U.S J-Lillo 610 LEE ET AL THE J OURNAL OF INVESTIGATIVE DERMATOLOGY

joJulSon R, Jackson Jj: Li.,.\'ltt is a dominant mouse l11U[ation resulting in prcl11