Proc. Nati. Acad. Sci. USAVol. 89, pp. 11421-11425, December 1992Medical Sciences

Malignancy of eye melanomas originating in the retinal pigmentepithelium of transgenic mice after genetic ablation ofchoroidal melanocytes

(Ki/W, WY mutant alleles/programmed cell death/neural crest-derived pigment cells/brain-derived pigment cells)

BEATRICE MINTZ* AND ANDRES J. P. KLEIN-SZANTOt*Institute for Cancer Research and tDepartment of Pathology, Fox Chase Cancer Center, Philadelphia, PA 19111

Contributed by Beatrice Mintz, September 11, 1992

ABSTRACT Eye tumors of the retinal pigment epithelium(RPE) have been thought generally to be benign, whereaschoroidal ones are malignant. To test this asmptio in mice,the W/W' (Kit) mutant genotype was introduced Into mdano-ma-prone tansgenlc mice whose recombinant simian virus 40transforming sequences are specifically expred in pigmentcells. WIWv causes prammed death of neural crest-derivedpigment cells, includin choroidal ones, but leaves Intact thebrain-derived pigment cells, such as those in theRPE. Dysplasticcells arose in the RPE, contiguous with frank melanotic neo-plasms. Invasion of the optic nerve, and tumor growth outsidethe orbit, attested to the malignancy of these RPE-derivedmelanomas. The widespread mela previously seen In micewith this transgene was absent when W/W' was added, thusvalidating its chief origin from neural crest cells.

Ocular melanomas may arise from any of the pigmentedtissues of the eye: the choroid, ciliary body, or iris (togethercomprising the uveal tract), or the retinal pigment epithelium(RPE). In humans, choroidal melanomas are the most com-mon type and are generally malignant (1, 2). Human RPEtumors are said to be difficult to diagnose (3), especially as eyetumors tend to be seen at late stages; most ofthem are benignand their status as melanomas has been questioned. In themelanoma-susceptible transgenic mice that we have pro-duced, malignant eye melanomas appeared to originate fromthe RPE as well as from the other possible target tissues (4).[The transgene, designated Tyr-SV40E in those experiments,is activated by the pigment cell-specific tyrosinase promoter,which drives the simian virus 40 (SV40) early-region se-quences, including the T-antigen transforming gene.] Never-theless, in advanced disease in the mice, multiple displacedtumors often coexisted in an eye, and contributions from theinitially neighboring RPE and choroidal layers could then nolonger be distinguished. It was thus of interest to determinewhether mouse eye tumors arising from the RPE could beclearly shown to be malignant and to compare the potential forRPE malignancy in mouse and human.We decided upon genetic ablation of choroidal melano-

cytes in the transgenic mice, as this would obviate occur-rence of choroidal melanomas. Such a strategy is feasiblebecause choroidal and retinal pigment cells develop fromseparate sources and are separately affected by specificmutant genes (5). The neural crest normally provides migra-tory dendritic melanocytes to the skin and hair follicles, to thechoroid layer of the eye, and to several other tissues (6). Thebrain is the other source ofpigment cell precursors. The opticvesicles arise as paired bulb-like outpocketings of the fore-brain. Each indents to form a double-layered cup whoseinnermost layer develops into the highly complex neural

retina while the outer one gives rise chiefly to the single-cell-layer pigment epithelium. Melanoblasts originating in theneural crest populate the choroid on the opposite flank of theRPE. The RPE then serves in part as a polarized transportsystem, functionally linking the vascular network of thechoroid with the neural retina. Some mutations at the W(white-spotting) locus lead to programmed death of neuralcrest-derived-but not brain-derived-pigment cells, failureof germ-cell proliferation and survival, and macrocytic ane-mia (7). W/Wv is one of the genotypes of mice with blackeyes, white coat, and severe anemia (6). The animals areusually inviable unless reconstituted with healthy hemato-poietic tissue; they are also presumptively sterile. The locus,now termed Kit, has been found to encode a transmembranegrowth factor receptor with tyrosine kinase activity; both Wand Wv alleles have been isolated and sequenced (8, 9).

MATERIALS AND METHODSTo incorporate WIWv into our Tyr-SV40E transgenic mice(4), the sterility due to WIWv requires that the W and Wvalleles be introduced separately in a series of matings.Moreover, the limited viability of WIWv due to severemacrocytic anemia (6) necessitates early rescue by repopu-lation with normal hematopoietic cells.

All mice in the experiment were congenic with theC57BL/6 inbred strain. Individuals either hemizygous for theT-antigen-encoding transgene (Tag!-) or homozygous (Tag!Tag) were all from line 9, described in the earlier report (4).The transgene in this line is not linked to the W locus (datanot shown). In the following series of matings, only thesegregants of interest are indicated:

(i) W/+ x Tag/Tag -* W/+;Tag/-, etc.(ii) WV/+ x Tag/Tag -* Wv/+;Tag/-, etc.(iii) W/+;Tag/- x Wv/+;Tag/- -+ W/Wv;Tag/Tag,

W/Wv;Tag/-, etc. WIWv mice, recognizable by their ex-treme pallor, were reconstituted at birth by intravenousinoculation with adult bone marrow cells, as described (10).Marrow donors with the 8-globin Hbbd genetic variant wereused, to distinguish donor-derived erythropoiesis from that ofthe recipients (Hbbs), in electrophoretic analyses of blood tomonitor successful repopulation. Healthy WIWv survivorsthat were hemi- or homozygous for the transgene wereidentified by DNA analysis of tail tissue (4).

Tissues prepared for histology were fixed in formalin andembedded in paraffin; the sections were stained with hema-toxylin and eosin.

RESULTSCandidate Mice. The transgene was previously found to

reduce the content of both eumelanin (black/brown) andphaeomelanin (yellow) pigments in the coat (4). The

Abbreviations: RPE, retinal pigment epithelium; SV40, simian virus40.

11421

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

June

2, 2

021

11422 Medical Sciences: Mintz and Klein-Szanto

C57BL/6 mice of transgenic line 9 are uniformly gray with avery slight brownish cast, rather than the wild-type black.The addition of W/+ or WV/+ caused a further overallreduction in coat color, especially in the case ofWV/+, whichitself causes a dilution of color; the animals also had thecharacteristic white feet, white belly spot, small white headspot, and some dorsal white areas (Fig. 1) devoid of pigmentcells. W/Wv mice with the transgene were all-white withblack eyes (Fig. 1), thus resembling W/Wv controls. Suc-cessful hematopoietic reconstitution at birth by intravenousinjection with adult bone marrow cells (10) overcame theanemia. Hematopoiesis in rescued individuals ultimatelycomprised 83-92% of red blood cells derived from thenonanemic congenic donor strain, as determined with a3-globin (Hbbd) marker. Three nonanemic WIWv survivorswith the transgene were obtained; one was hemizygous(Tag!-) for the transgene and two were homozygous (Tag/Tag). These were sacrificed at 17-24 weeks of age and theirhistopathology was compared with that of similar transgenicmice (4) that were wild-type at the Wlocus. As nontransgeniccontrols, three WIWv and three wild-type mice on the sameinbred-strain background were examined in the same age-range as the experimentals; none had tumors. All threetransgenics with W/WV, but not the WIWv or wild-typecontrols, developed eye tumors in the same time period aspreviously found in the transgenics lacking WIWv. However,the new experimental animals had no cutaneous melanomasor other tumors.RPE Pigumentary Change with Age. During histological

comparisons of eyes from mice of different ages up to 28weeks, a sharp decrease with age in the number of melano-somes was unexpectedly seen in the RPE cells. While asignificant decline in these melanin-containing granules wasknown in the aging human RPE (reviewed in ref. 11), thischange was not noted in a study of aging of the C57BL/6mouse RPE (12) but is clearly a feature of normal aging (Fig.2). RPE melanosomes in adult mice are larger and moreirregular than those in the choroid (13); their somewhatcoarse appearance is readily discernible, even at low mag-

FiG. 1. Changes in coat color due to genes affecting pigment-cellviability and tumorigenicity. All mice are on the C57BL/6 inbred-strain background. The black coat characteristic of the wild-type(+/+, top animal) becomes markedly lighter (in the middle animal)when two genetic changes are made: the Tyr-SV40E transgenecauses reduction in both eumelanin and phaeomelanin pigments; theWv gene (here Wv/+) causesp an overall dilution of color and alsonsome white areas devoid ofpigment cells. The further addition oftheW gene (W/Wv;Tag/-) (bottom animal) results in an all-white coatlacking pigment cells (which are of neural crest origin in the coat)while the eyes remain black (due to pigment cells of brain origin).

_wo~~~~~~~ a of 0

a bFiG. 2. Normal decrease with age in numbers ofmelanin granules

in the RPE (arrowhead) ofC57BL/6 control mice shown at 14 weeks(a) and at 28 weeks (b). The pigment granules are less numerous inthe older mouse, thus allowing their polarized distribution in relationto the cell nucleus to be more readily seen. The melanin content ofthe choroid layer (densely black, below the RPE) appears similar atboth ages. (Hematoxylin/eosin stain; x220.)

nification, when the number of granules declines. The basisfor the decline is not clear. The activity of tyrosinase-therate-limiting enzyme in melanin synthesis-has been docu-mented in the fetal mouse RPE but has been thought to beabsent in the adult (14). However, based on nonmurinemammalian studies, there is some recent evidence for a slowproduction of melanin throughout life (11).Eye HEsopathokogy in WIW' Mice with the Tyr-SV40E

Tlangene. Tumorigenesis in the eyes is first externallydetectable as an eye enlargement, often at different times onthe two sides (Fig. 3). The normal C57BL/6 RPE, with itsbasement membrane, is separated by a slight space from thecomplex choroidal layer containing a dense brushwork ofheavily pigmented melanocytes (Fig. 2 and Fig. 4a). In theWIWv controls and in the transgenics with WIWv, thechoroid was devoid of melanocytic cells but appeared oth-erwise normal (Fig. 4 b-d). Thus, the pigment cell-specifictyrosinase promoter of the transgene would be expected toactivate the SV40 transforming sequences in the RPE and notin choroidal cells of the experimental animals. RPE melano-mas, and no choroidal tumors, did in fact occur in both eyes

FIG. 3. A W/Wv;Tag/- mouse in which the slight enlargementof the right eye is caused by a growing tumor, later found to haveoriginated in the RPE.

Proc. Nad. Acad. Sci. USA 89 (1992)

Dow

nloa

ded

by g

uest

on

June

2, 2

021

Proc. Nati. Acad. Sci. USA 89 (1992) 11423

} B

isAbsJ&Jh1 s -.

FIG. 4. (a) Normal eye section of a 14-week control C57BL/6 mouse showing the neural retina and, below it, the RPE (left arrowhead) andthe melanocytic cells of the choroid (right arrowhead), with intense melanin deposition in both. (b) Eye of a 24-week W/Wv-C57BL/6nontransgenic animal. No melanocytic cells are present in the choroid (left arrowhead). The RPE (right arrowhead) is less pigmented than inyounger W/Wv or wild-type mice. (c) Retina ofa 24-week W/Wv mouse hemizygous for the Tyr-SV40E transgene (Tag/-). A hyperplastic area(arrowhead) ofthe hypomlcanotic RPE is contiguous with the hypomelanotic early melanoma above. (d) In a 17-week WIWv mouse homozygousforthe transgene (Tag/Tag), a melanoma has arisen in the papillary area ofthe RPE and has caused retinal detachment. Tumor cells have invadedthe optic nerve (arrowhead). (e) Higher magnification ofthe same tumor, showing invasion in the optic nerve. Some melanoma cells are intenselypigmented. (f) Tumor nodule in the base of the skull of the same animal. (Hematoxylin/eosin stain; a-c, and e, x100; d, x20; f, x40.)

of all three transgenic mice with W/WV. The melanomas werehypopigmented and arose in focally hypopigmented anddysplastic regions of the RPE. The contiguity of an earlymelanoma and a hyperplastic RPE region is evident in Fig.4c. The melanomas were poorly differentiated and werechiefly epithelioid, with some spindle and glanduliform areas,and with sporadically distributed pigmentation.Malignancy ofRPE Melanomas. At more advanced stages,

the RPE melanomas had invaded the optic nerve. In the case

shown (Fig. 4 dand e), the main tumor mass caused extensiveretinal detachment and displacement. Tumor had also in-vaded, or metastasized to, the base of the skull (Fig. 4f).Lack of Myanosis in Tyr-SV40E Mice with WIW'. The

occurrence of melanized cells in tissues ordinarily lackingmelanin, or producing relatively small amounts, is referred toas melanosis or melanocytosis. Apart from the usual pres-ence of melanin-containing cells in the skin and eyes, mod-erate numbers of pigmented cells are often found in mice in

Medical Sciences: Mintz and Klein-Szanto

Dow

nloa

ded

by g

uest

on

June

2, 2

021

11424 Medical Sciences: Mintz and Klein-Szanto

the nictitans, meninges of the brain, harderian glands, par-athyroids, thymus (5), inner ear (15), and spleen. Only in onemouse strain, PET (now extinct), have pigmented cells beenseen more widely distributed, in connective tissues through-out the body (16). The existence of melanosis in Tyr-SV40Etransgenic mice (17) therefore came as a surprise, especiallyas the transgene caused a reduction rather than an increasein coat pigmentation, and the ocular and cutaneous melano-mas in the same animals tended to be hypo- rather thanhyperpigmented (4). In those individuals, melanosis wasseen, with decreasing frequency, in the following novel sites:nasal mucosa, endocardium, lungs, peripheral nervous sys-tem, lymph nodes, central nervous system, genital organs,skeletal muscle, oral mucosa, tooth enamel organ, pinealgland, choroid plexus, mammary glands, larynx, salivaryglands, bladder, and urethra (17). In many cases, the pig-mentation per se was in dendritic cells embedded in othertissue (e.g., in muscle). Even more striking was the fact thatan array of neoplasms, generally also containing some mel-anin, occurred in association with some of the melanotictissues. Included were tumors of the choroid plexus, en-docardium, peripheral nerve sheath (schwannomas), co-chlea, pineal gland, salivary gland, and nasal mucosa. Thesetumors appeared to be distinct from peripheral melanotictumors judged to be metastases of eye melanomas (4, 17).We have proposed (17) that most of the examples of

melanosis and associated tumors represent neural crest-derived melanocytes that had increased in number, and hadin some cases become transformed, under the influence ofthetransgene; the melanocytic schwannomas may have comefrom developmentally bipotential neural crest cells. Theother examples could be ascribed to epithelia and endotheliawith some developmental plasticity, especially those of neu-roectodermal origin; to epithelia receiving melanin granulesinjected by melanocytes (which ordinarily inject granules intohairs and keratinocytes); to tissues exhibiting metaplasticconversion; or to phagocytic cells.The W/Wv;Tag/- (or W/WvTag/Tag) mice in the present

study afford an opportunity to test those interpretations, asWIWv leads to programmed cell death (18, 19) of melano-cytes originating in the neural crest, while expression of theTyr-SV40E transgene causes transformation of cells capableof melanization. The animals examined to date do in fact lackmelanosis as well as the tumors previously associated withthat condition. Thus the new evidence lends support to theview of melanosis as primarily an aberration of neural crestdevelopment. The lack of cutaneous melanomas reflects thedeath of skin melanoblasts in WIWv mice. The transgene hasapparently failed to override the deleterious effect of W/Wvon melanoblast viability, possibly due to a greater expressionof W/W".

DISCUSSIONThe results unambiguously demonstrate that melanotic neo-plasms can originate from the RPE in vivo and can becomemalignant. Dysplastic cells, representing premalignantchanges of the RPE, were seen contiguous with full-fledgedmalignant melanomas, under genetic conditions precludingorigin from the choroid. Invasion of the optic nerve andpresence of tumor outside the orbit attest to the malignantnature of the RPE-derived tumors. Their designation asmelanomas is supported by their origin from melanized cellsand is more specific than a broader designation based on theirepithelial origin, especially as the RPE is not a conventionallining or glandular epithelium.The RPE melanomas seen here and previously (4) in the

transgenic mice were predominantly hypomelanotic and ep-ithelioid. Nevertheless several other patterns characterizedby glanduliform, fusiform, or small-cell differentiation were

observed. The diversity in morphologic architecture of hu-man melanomas is well known and has been reviewed byNakhleh et al. (20), who have concluded that "malignantmelanomas may assume the histologic guise of adenocarci-nomas, small cell carcinomas, and sarcomas." Most of thetumors reviewed by those investigators reacted strongly withthe reagents for S-100 protein and the HMB-45 antigen. Thefact that these same histological and histochemical charac-teristics were found in both skin and RPE tumors in ourtransgenic mice, before the WIWv genotype was introduced(4), strongly supports the view that the RPE neoplasms aremelanomas. Further evidence for this view is their invasiveand metastatic potential. It should be noted that no tumor ofthe RPE, except melanoma, is characterized by this aggres-sive biological behavior. For example, RPE adenocarcino-mas, which can have some features in common with mela-nomas, are known to be less aggressive and usually nonme-tastasizing (21). In addition, the adenocarcinomas producecytokeratins whereas melanomas, including the eye tumorsfound in our transgenic mice, are negative for keratin.The disparity between our observations on mice and re-

ports of human RPE tumors may have several possibleexplanations: species differences; incorrect or ambiguousdiagnoses of at least some human eye tumors (3), due to theadvanced stages of detection; or differences in the develop-mental stages at which the mouse and human tumorsarise-at relatively younger and older ages, respectively. Inthe experimental mice, the tyrosinase promoter undoubtedlycauses the transgene to start functioning in fetal life, wheneye pigmentation first appears. At that time, cells of thedeveloping eye are most actively proliferating (12) and thusprovide many potential target cells for transformation (22);the fact that RPE melanomas were more common thanchoroidal ones in the original (non-W/WV) transgenics (4)may be further attributable to differences in transcriptionalstrength of this promoter in RPE vs. choroid. In contrast, thetypically later-onset human eye tumors originate when thereis little RPE cell division (23), yet choroidal melanocytes maystill be frequently dividing. The difference between humanand mouse in occurrence of malignant RPE melanomas maytherefore be more apparent than fundamental, with thebalance of melanoma formation simply shifting among thedifferent pigmented eye tissues with age, and the RPE inolder individuals becoming more likely to form only benigntumors. Nevertheless, any agent capable of stimulating RPEcell division in vivo or in vitro would be expected to reactivatethe potential for malignant RPE melanomas at any age. Thisexpectation is borne out by the fact that explanted RPE cellsof adult human and other mammalian species resume activedivision in culture and can be transformed with virions oroncogenes (24, 25).Two aspects of RPE aging contribute to an understanding

of eye melanoma formation in Tyr-SV40E transgenic mice:pigmentary and mitotic changes. The higher prenatal thanpostnatal expression of tyrosinase in the RPE would meanthat the transgene, driven by a tyrosinase promoter, is likelyto initiate RPE melanomas at an early age; melanomas wereindeed already present when the eyes were first examinedhistologically, at 4 weeks (4). Mitotic activity in the C57BL/6mouse RPE declines soon after birth and cell numbers arestabilized by 2 weeks of age, with some further increase inarea due to cell enlargement (12). In light of the age-relatedreduction in melanization (Figs. 2 and 4b), it follows that astimulus to cell proliferation-in this case, the transformingSV40 sequences-would in effect distribute the availablemelanosomes among an increasing number of cells, therebyleading to hyperplastic nodules that are hypomelanotic, andto hypomelanotic tumors. This did in fact occur (Fig. 4c; alsosee figure 2 in ref. 4) and was puzzling when first observed.

Proc. NatL Acad. Sci. USA 89 (1992)

Dow

nloa

ded

by g

uest

on

June

2, 2

021

Proc. Natl. Acad. Sci. USA 89 (1992) 11425

W/WV animals containing the integrated Tyr-SV40E trans-gene can now be used as a source of material with whichseveral questions can be examined. Ostensibly normal RPEcells, destined to become malignant, can be explanted fromvery young mice to define genetic changes as the cellsprogress toward malignancy. Whereas unavoidable contam-ination ofRPE cultures with choroidal melanocytes has untilnow posed a problem for studies in vitro (26), no suchambiguity would persist here. The animals themselves willalso enable metastatic propensities specific to RPE melano-mas to be identified in the absence of any choroidal or skinmelanomas.

We thank Nancy Loughery for excellent technical assistancethroughout the study and Monika Bradl for DNA verification of thetransgene copy number. This work was supported by U.S. PublicHealth Service Grants CA42560 to B.M. and CA06927 and RR05539to the Fox Chase Cancer Center, as well as an appropriation to theCenter from the Commonwealth of Pennsylvania.

1. Foos, R., Straatsma, B., Gardner, K., Zakka, K. & Omphroy,C. (1983) in Intraocular Tumors, eds. Lommatzsch, P. K. &Blodi, F. C. (Springer, Berlin), pp. 51-57.

2. Newell, F. W. (1992) Ophthalmology Principles and Concepts(Mosby Year Book, St. Louis), pp. 284-317.

3. Tso, M. 0. M. (1979) in The Retinal Pigment Epithelium, eds.Zinn, K. & Marmor, M. (Harvard Univ. Press, Cambridge,MA), pp. 267-276.

4. Bradl, M., Klein-Szanto, A., Porter, S. & Mintz, B. (1991)Proc. NatI. Acad. Sci. USA 88, 164-168.

5. Markert, C. L. & Silvers, W. K. (1956) Genetics 41, 429-450.6. Silvers, W. K. (1979) The Coat Colors ofMice (Springer, New

York).7. Mintz, B. & Russell, E. S. (1957) J. Exp. Zool. 134, 207-238.

8. Nocka, K., Majumder, S., Chabot, B., Ray, P., Cervone, M.,Bernstein, A. & Besmer, P. (1989) Genes Dev. 3, 816-826.

9. Nocka, K., Tan, J. C., Chiu, E., Chu, T. Y., Ray, P., Trakt-man, P. & Besmer, P. (1990) EMBO J. 9, 1805-1813.

10. Capel, B. & Mintz, B. (1989) Exp. Hematol. 17, 872-876.11. Boulton, M. (1991) Prog. Retinal Res. 11, 125-151.12. Bodenstein, L. & Sidman, R. L. (1987) Dev. Biol. 121, 192-204.13. Guillery, R. W. & Price, S. D. (1985) in Biological, Molecular

and Clinical Aspects of Pigmentation, eds. Bagnara, J. T.,Klaus, S. N., Paul, E. & Schartl, M. (Univ. Tokyo Press,Tokyo), pp. 303-312.

14. Miyamoto, M. & Fitzpatrick, T. B. (1957) Science 126, 449-450.

15. Deol, M. S. (1970) Proc. R. Soc. London Ser. A 175, 201-217.16. Nichols, S. E. & Reams, W. M., Jr. (1960) J. Embryol. Exp.

Morphol. 8, 24-32.17. Klein-Szanto, A., Bradl, M., Porter, S. & Mintz, B. (1991)

Proc. Nat!. Acad. Sci. USA 88, 169-173.18. Mintz, B. (1970) Symp. Int. Soc. Cell Biol. 9, 15-42.19. Mintz, B. (1971) Symp. Soc. Exp. Biol. 25, 345-370.20. Nakhleh, R. E., Wick, M. R., Rocamora, A., Swanson, P. E.

& Dehner, L. P. (1990) Am. J. Clin. Pathol. 93, 731-740.21. Tso, M. 0. M. & Albert, D. M. (1972) Arch. Ophthalmol. 88,

27-38.22. Mintz, B. & Fleischman, R. A. (1981) Adv. Cancer Res. 34,

211-278.23. Tso, M. 0. M. (1968) Arch. Ophthalmol. 80, 214-216.24. Albert, D. M., Rabson, A. S. & Dalton, A. J. (1968) Invest.

Ophthalmol. 7, 357-365.25. Dutt, K., Scott, M., Del Monte, M., Agarwal, N., Sternberg,

P., Srivastava, S. K. & Srinivasan, A. (1990) Oncogene 5,195-200.

26. Wang, C.-W., Roque, R. S., Defoe, D. M. & Caldwell, R. B.(1991) Curr. Eye Res. 10, 1081-1086.

Medical Sciences: Mintz and Klein-Szanto

Dow

nloa

ded

by g

uest

on

June

2, 2

021