Pigmentary glaucoma is a significant cause of human blindness.Abnormally liberated iris pigment and cell debris enter the ocu-lar drainage structures, leading to increased intraocular pres-sure (IOP) and glaucoma1–3. DBA/2J (D2) mice develop a form ofpigmentary glaucoma involving iris pigment dispersion (IPD)and iris stromal atrophy (ISA)4,5. Using high-resolution map-ping techniques, sequencing and functional genetic tests, weshow that IPD and ISA result from mutations in related genesencoding melanosomal proteins. IPD is caused by a prematurestop codon mutation in the Gpnmb (GpnmbR150X) gene, asproved by the occurrence of IPD only in D2 mice that arehomozygous with respect to GpnmbR150X; otherwise, similar D2mice that are not homozygous for GpnmbR150X do not develop

IPD. ISA is caused by the recessive Tyrp1b mutant allele and res-cued by the transgenic introduction of wildtype Tyrp1. Wehypothesize that IPD and ISA alter melanosomes, allowingtoxic intermediates of pigment production to leak frommelanosomes, causing iris disease and subsequent pigmentaryglaucoma. This is supported by the rescue of IPD and ISA in D2eyes with substantially decreased pigment production. Thesedata indicate that pigment production and mutant melanoso-mal protein genes may contribute to human pigmentary glau-coma. The fact that hypopigmentation profoundly alleviatesthe D2 disease indicates that therapeutic strategies designed todecrease pigment production may be beneficial in human pig-mentary glaucoma.

Fig. 1 Mapping and identification of ipd. The D2-derivedipd mutation was mapped using intersubspecific crossesbetween D2 and CAST. Mice were assayed by slit-lampexamination (a–c). a, Wildtype CAST eye with complex irismorphology without pigment dispersion (15 mo). b, Par-tially dilated eye showing the IPD mutant phenotype in aD2-CAST genetic background (15 mo). Dispersed pigment isclearly visible on the lens as well as on the iris (arrowheads).ISA is wildtype. c, Representative eye showing the charac-teristic phenotype resulting from the combined presence ofIPD and ISA (8 mo). Homozygosity for ipd mutant alleles isindicated by the presence of dispersed pigment on the lensand iris (arrowheads) and by the overall degree of iris atro-phy. IPD phenotypes in this mouse are clearly evident at 8 mo(versus 15 mo in b), because of the earlier age of onset pre-cipitated by the additional presence of ISA (which wasintentionally fixed within mapping mice for this purpose;see Methods). The dispersed pigment in this eye appearssomewhat unfocused owing to corneal haziness and calcifi-cation, which are common to aged mice with a D2 geneticbackground. d, High-resolution genetic and physical map-ping of ipd. Filled boxes represent D2 allele; open boxes,CAST allele. The number of mice with each haplotype islisted at the bottom of each column. The horizontal linesindicate the positions of markers within individual BACs.For simplicity, only a subset of markers within the intervaland BACs flanking the critical region are shown (BAC namesbeing abbreviated; see Methods). Using mice with informa-tive recombinations, the ipd critical region was narrowed tothe 281T–KOC13 interval spanned by a single BAC, 281N17(solid bar). The location of genes with confirmed positionsin mice and humans are indicated. e, Sequence comparisonof Gpnmb products amplified from wildtype and D2 eyesindicating the presence in D2 mice of a premature stopcodon mutation, GpnmbR150X. f, Representation of GPNMBprecursor protein and the predicted truncated proteinencoded by GpnmbR150X. Only motifs conserved betweenthe mouse and human proteins are indicated. SS, signalsequence; RGD, RGD tripeptide; PKD, polycystic kidney dis-ease repeat; TM, transmembrane; LL, dileucine melanoso-mal sorting signal; *, potential N-glycosylations.

letter

nature genetics • volume 30 • january 2002 81

Mutations in genes encoding melanosomal proteinscause pigmentary glaucoma in DBA/2J mice

Michael G. Anderson1,2, Richard S. Smith1,2, Norman L. Hawes2, Adriana Zabaleta2, Bo Chang2, Janey L.Wiggs3 & Simon W.M. John1,2,4

1The Howard Hughes Medical Institute and 2The Jackson Laboratory, Bar Harbor, Maine 04609, USA. 3Department of Opthalmology, Harvard MedicalSchool, Massachusetts Eye and Ear Infirmary, Boston, Massachusetts 02114, USA. 4Department of Ophthalmology, Tufts University School of Medicine,Boston 02111, Massachusetts, USA. Correspondence should be addressed to S.W.M.J. (e-mail: swmj@jax.org).

Published online: 17 December 2001, DOI: 10.1038/ng794

a b c

d

e f

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

82 nature genetics • volume 30 • january 2002

Subsequent to a progressive iris disease, inbred D2 micedevelop a form of pigmentary glaucoma involving increasedIOP, retinal ganglion cell loss and optic nerve head excavation4.The D2 iris disease results from two genetically separable traits,IPD and ISA, whose combined interactions result in severe irisdepigmentation. Observed independently, IPD and ISA areassociated with related but distinct clinical and histologic mani-festations5. IPD is characterized by a deterioration of the poste-rior iris pigment epithelium, slit-like transillumination defectsand pronounced pigment dispersion (Fig. 1b). These features ofIPD resemble human pigment dispersion syndrome (PDS,OMIM 600510), an apparently common cause of glaucoma inhumans whose molecular basis remains unknown1,3,6,7. ISA isassociated with deterioration of the anterior iris stroma, result-ing in a loss of clinically detectable iris stromal complexity andan accumulation of stromal pigment and cell debris in the ocu-lar drainage structures5. Mice that are homozygous with respectto both ipd (the IPD-causing gene) and isa (the ISA-causinggene) have an earlier-onset, more severe iris disease and moresevere glaucoma than either single mutant (Fig. 1c)5. We previ-ously localized ipd to mouse chromosome 6 (position 25.5 cM)and isa to mouse chromosome 4 (position 34.4 cM)5. To identifythe causative mutations at these loci, we carried out intrasub-specific mapping crosses.

For ipd, a high-resolution genetic map was generated andused to identify mice with informative recombinations (Fig. 1d).Ageing and clinical analysis of these mice narrowed the ipdcritical region to a 0.13-cM interval, for which we generated aBAC-derived physical map. New markers generated from thephysical map further restricted ipd to a single BAC, 281N17.Analysis of 100 sequences generated from randomly selected

281N17 fragments identified exons of two genes within thisBAC, Igf2bp3 and Gpnmb. Analysis of a sequenced human con-tig containing the IGF2BP3 and GPNMB genes identified twoadditional putative genes (NLP1 and PRO2738), potentiallywithin or flanking the critical region. Of these candidates, onlyGPNMB is known to be expressed in pigmented cells or theeye8,9. Sequencing of Gpnmb from D2 mice detected a prema-ture stop codon mutation, GpnmbR150X, within the fourth of 11exons (Fig. 1e). The truncated protein encoded by GpnmbR150X

was predicted to lack a carboxy-terminal dileucine melanoso-mal sorting motif10, as well as the PKD domain that potentiallyinfluences protein–protein interactions11 (Fig. 1f ).

We found that some D2 stocks separated from the modern D2lineage during the 1970s and 1980s have the wildtype Gpnmballele on an otherwise D2 genetic background (Fig. 2). Sandy(sdy), for example, is a recessive coat-color mutation that sponta-neously arose on the D2 genetic background in 1983 (ref. 12). Itwas maintained at The Jackson Laboratory as a closed breedingcolony (DBA/2J-sdy) using sdy/+ obligate heterozygotes and wassubsequently cryopreserved. We recovered the DBA/2J-sdy stock,retaining its closed breeding status, and found that wildtype andGpnmbR150X alleles segregated in this stock. Similarly, two stocksderived from D2 in the 1970s (DBA/2J-hotfoot, a spontaneousmutation that occurred on the D2 background13, and AKXD-28/Ty, a recombinant inbred line carrying D2 alleles across thechromosomal region flanking Gpnmb14,15) have wildtype Gpnmballeles. This indicates that GpnmbR150X arose during the early1980s and subsequently became fixed in the ancestors of themodern D2 strain.

To test functionally whether GpnmbR150X is ipd, we aged andanalyzed normally pigmented DBA/2J-sdy heterozygotes withdifferent Gpnmb genotypes. Because inbred mouse stocks arespecifically bred to be genetically identical, these mice are almostcertainly identical except for their Gpnmb genotypes. Only Gpn-mbR150X homozygotes developed IPD (Fig. 2a–d; 6 affected of 6tested GpnmbR150X/R150X, none of 13 tested GpnmbR150X/+ andnone of 4 tested Gpnmb+/+; all the mice were 6–8 mo). In addi-tion, our colonies of DBA/2J-hotfoot and AKXD-28/Ty carrywildtype Gpnmb alleles and do not develop IPD (Fig. 2e,f). Theseexperiments prove that GpnmbR150X causes IPD.

For isa, our initial experiments indicated the candidate geneTyrp1 (ref. 5). To further evaluate Tyrp1 as a candidate for isa, weanalyzed a total of 1,237 meioses. In these aged mice, the ISAphenotype and Tyrp1 showed complete concordance, placing thealleles at the same genetic position (95% confidence interval0.0–0.1 cM). The Tyrp1 allele of D2 mice (Tyrp1b) encodes amutant protein containing two amino-acid substitutions, com-pared with the Tyrp1 allele of C57BL/6J mice (which have normalirides). Tyrp1b results in a brown instead of black coat color16,

Fig. 2 Only GpnmbR150X homozygotes develop IPD. Clinical slit-lamp examina-tion of genetically identical DBA/2J-sdy mice in which GpnmbR150X is the onlyknown segregating factor (a–d). All eyes are from normally pigmented oblig-ate sdy heterozygotes that are homozygous for isa and for D2 microsatellitemarkers flanking Gpnmb. a, An 8-mo GpnmbR150X/R150X eye showing IPD andISA. Dispersed pigment characteristic of IPD is visible on the lens and iris(arrowhead), and the stromal atrophy typical of ISA is distinct at the pupillaryborder (arrow). b, An 8-mo GpnmbR150X/+ eye with ISA but wildtype withrespect to IPD. c, Transillumination (red regions corresponding to reflectedlight passing through the iris) of the GpnmbR150X/R150X eye shown in a, indica-tive of diseased iris caused by the presence of IPD and ISA. d, Lack of transillu-mination from the GpnmbR150X/+ eye shown in b, supporting the absence ofIPD. e, Mice from our DBA/2J-hotfoot colony that are not homozygous withrespect to GpnmbR150X do not develop IPD, as shown here at 16.5 mo. Asexpected, this eye has the stromal atrophy (arrow) characteristic of ISA (n=9mice, 10–19 mo). f, AKXD-28/Ty mice lack GpnmbR150X and do not develop IPD,as shown here at 13 mo. Stromal atrophy (arrow) is particularly pronounced atthe pupillary border (n=30 mice, 6–22 mo).

a b

c d

e f

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

nature genetics • volume 30 • january 2002 83

and other mutant Tyrp1 alleles caused melanocyte death17. Totest whether ISA is caused by the mutant Tyrp1b allele, we gener-ated and analyzed a D2 stock of mice transgenic for a BAC con-taining wildtype Tyrp1 (D2-Tg(Tyrp1); Fig. 3). The BAC has aninsert of approximately 88 kb. Analysis of the genomic regionsflanking human and mouse Tyrp1 with public and proprietary(Celera) databases indicated that Tyrp1 is probably the only genein the transgenic insert. Both databases predicted a lack ofexpressed sequences within 61 kb proximal or distal to humanTYRP1, indicating that if a similar gene spacing exists in themouse, the insert would not contain additional genes. This con-clusion was also supported by direct analysis of the insertsequence against Celera’s largely completed mouse database,indicating that Tyrp1 was the only gene within the rescuing BACinsert. Clinical analysis of aged mice hemizygous with respect tothe transgene showed that wildtype Tyrp1 rescued the ISA phe-notype, confirming that Tyrp1b causes ISA (Fig. 3c,d; n=11 mice,aged 6–23 mo).

The predicted full-length GPNMB and TYRP1 proteins con-tain several motifs common to melanosomal proteins, both pro-teins sharing similarity to tyrosinase and each other. Based onthese similarities, it has been suggested that GPNMB and TYRP1belong to a common gene family that also includes tyrosinaseand silver9. BLAST and CLUSTAL analyses both indicate signifi-cant similarity between GPNMB and si, the product of the silverlocus in mice, which has been implicated as a structural compo-nent of the melanosomal matrix18. TYRP1 is reported to be themost abundant melanosomal glycoprotein19, influencesmelanosome structure20 and is required for the stabilization of amembrane-bound melanogenic protein complex21. MouseTyrp1 has various enzymatic activities, including catalase22 and5,6-dihydroxyindole-2-carboxylic acid (DHICA) oxidase activ-ity23; human TYRP1, however, lacks the latter activity24. Thestructural functions of TYRP1 may therefore reflect a more evo-lutionarily conserved function.

We hypothesized that the GpnmbR150X and Tyrp1b mutationsalter melanosomes, allowing pigment production to occur whilecytotoxic intermediates of pigment production escape, inducingiris disease and subsequent glaucoma. In support of this forTyrp1b, we demonstrated that albino mice with completely abol-ished pigment production did not show ISA5. To test ourhypothesis for both GpnmbR150X and Tyrp1b, we carried out anepistasis experiment examining the consequences of reducedpigment production on the expression of both IPD and ISA(Fig. 4). This experiment also tested the potential therapeuticbenefit of reduced, but not abolished, pigment production. Pearl(pe) is a coat-color mutation occurring on the D2 background(DBA/2J-pe)25. DBA/2J-pe homozygotes have hypopigmented

coats and eyes. Despite homozygosity for GpnmbR150X andTyrp1b, aged homozygous DBA/2J-pe mice did not show IPD orISA phenotypes (Fig. 4a–f). In a similar experiment, hypopig-mented eyes from mice homozygous for sdy and GpnmbR150X

showed no signs of IPD or ISA (data not shown). Histologicanalysis of a subset of eyes from DBA/2J-pe mice confirmed theclinical results (Fig. 4g,h). These hypopigmentation mutationsalso prevented the induction of glaucoma by the GpnmbR150X

and Tyrp1b mutations (Fig. 4i,j). These results showed that theglaucomatous IPD and ISA phenotypes were dependent uponthe level of active pigment production occurring in the adult eyeand indicated that treatments to reduce pigment productionmight be beneficial in pigmentary glaucoma.

The molecular basis of pigmentary glaucoma in humans isunknown. Our data suggest that pigment production andmutant melanosomal protein genes may contribute to humanpigmentary glaucoma. We have sequenced the GPNMB codingregion from the affected individuals of four families segregatingfor PDS, but have not detected any mutations (data not shown).Further studies analyzing additional PDS patients and the fullGPNMB locus (including the regulatory sequences) are under-way. If a similar molecular etiology underlies human pigmentaryglaucoma, several possibilities warrant mention. First, eventhough prostaglandin analogs (such as latanoprost) may lowerIOP in the short term, the induction of melanogenic processesand increased iris pigmentation26,27 by these commonly usedIOP-lowering glaucoma medications may have adverse long-term effects for individuals with pigmentary glaucoma. Second,because ultraviolet light induces melanogenic pathways (at leastin skin)28, exposure to bright sunlight may acutely exacerbate irispigment dispersion. Third, our results predict that conditionsdecreasing iris melanogenesis should offer protection againstpigmentary glaucoma.

Fig. 3 Transgenic rescue of ISA. Clinical slit-lamp examination of D2-Tg(Tyrp1)mice shows the rescuing effects of wildtype Tyrp1 (a–d). a, A 2-mo D2 iris show-ing normal predisease morphology. b, A 12-mo D2 iris with pronounced deteri-oration resulting from the interaction of the ISA and IPD phenotypes. Thepresence of ISA is indicated by stromal atrophy that is especially pronounced atthe pupillary border (arrow). c, A 12-mo iris from a D2-Tg(Tyrp1) transgenicmouse hemizygous for the insertion of a wildtype Tyrp1–containing BAC, dis-playing rescue of ISA. The iris is indistinguishable from those of young mice. Asthe onset of IPD is delayed when ISA is not present5, signs of IPD are also lack-ing at this age. d, A 16-mo D2-Tg(Tyrp1) iris continuing to show rescue of ISAwith early signs of IPD, now distinct as clumps of dispersed pigment on the iris(arrowheads). Histological analysis supports the clinical demonstration of res-cue (e–g). e, Wildtype iris from a 2-mo predisease D2 eye with robust anteriorstromal (s) and posterior pigment epithelial (p) layers separated by a thin dila-tor muscle (arrowheads). f, A 12-mo D2 iris with pronounced iris atrophy. g, A12-mo D2-Tg(Tyrp1) iris showing robust iris stroma. h, D2-Tg(Tyrp1) mice (right)have darker coats than D2 mice (left), indicating the functional expression ofTyrp1 from the transgenic BAC. (×400, e–g).

a b

c d

e

f

g

h

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

84 nature genetics • volume 30 • january 2002

Known human TYRP1 mutations cause OCA3 (refs 29,30), aform of oculocutaneous albinism (OMIM 203290), with noreported increased risk of pigmentary glaucoma. TYRP1 is amember of a multiprotein complex required for the stabiliza-tion of tyrosinase (the enzyme that catalyzes the first commit-ted step in pigment production)21. Because OCA3 mutationscause oculocutaneous albinism, they may be ‘self-rescuing’with respect to pigmentary glaucoma, as the mutant TYRP1protein may not stabilize tyrosinase, resulting in greatlydecreased pigmentation. As shown here, decreased productionof cytotoxic intermediates of melanogenesis spared D2 micefrom pigment dispersion and glaucoma. It has yet to be deter-mined whether alleles permissive for normal levels of iris pig-mentation, such as Tyrp1b, contribute to human oculardisease. Future studies with D2 mice will help further dissectthe biologic functions of Gpnmb and Tyrp1, and the linkbetween melanogenesis and pigmentary glaucoma.

MethodsAnimal husbandry and stocks. We carried out all experiments in compli-ance with the Association for Research in Vision and Opthalmology state-ment on the use of animals in ophthalmic and vision research. We housedmice under previously reported conditions4,5. The Jackson Laboratory’sroutine surveillance program routinely screened all mouse strains used forthis study for select pathogens. The mouse strains and abbreviations ofmice used in this study include DBA/2J (D2), C57BL/6J (B6) and CAST/Ei(CAST). Unless specifically stated, all references to D2 refer to modern D2mice. We obtained DBA/2J-sdy, DBA/2J-hotfoot (Grid2ho-4J) and DBA/2J-pe (Ap3b1pe-8J) mice from The Jackson Laboratory Cryopreservation Ser-vice. We ascertained the Gpnmb genotype of these stocks by assaying forthe presence or absence of a unique PvuII site (CAGCTG) created by theGpnmbR150X mutation within PCR products amplified from genomicDNA. The Jackson Laboratory Microinjection Service generated DBA/2J-Tg(Tyrp1)280G21Sj (D2-Tg(Tyrp1)) mice using a Tyrp1-containing BAC280G21 obtained from a C57BL/6-derived mouse BAC library (GenomeSystems). We confirmed the presence of a wildtype Tyrp1 allele within280G21 by direct sequencing.

Human subjects. We sequenced the GPNMB gene of two affected membersof each of four families affected by an autosomal dominant form of pig-ment dispersion syndrome7. We purified genomic DNA from a leukocytepellet and selectively amplified, using PCR, each exon and splice junctionof the GPNMB gene.

Clinical examinations. We examined the eyes of mice aged 2–24 mo with aslit-lamp biomicroscope (Haag-Streit) and photographed them with a ×40objective lens. Phenotypic assessment of iris stromal atrophy, dispersedpigment and transillumination followed previously described criteria4,14.

Locus mapping. We mapped the isa and ipd loci with similar strategiesusing mapping mice fixed for the D2 allele of either isa or ipd and segregat-ing for the other. As a consequence, we phenotyped recombinants for thepresence or absence of the phenotype of interest and the severe conse-quences resulting from the presence of both alleles. The current mappingcrosses were derived from intersubspecific crosses of D2 and CAST. Theipd mapping included 2,539 meioses from an N2 or N3 backcross((D2CAST × D2)N2 × D2) or ((D2CAST × D2)N3 × D2) and 3,470 meios-es from an N3 intercross ((D2CAST × D2)N3 × (D2CAST × D2)N3). Ourmapping of isa is based on 870 progeny of an N2 backcross ((D2CAST ×D2)N2 × D2) and 367 previously reported meioses5. We aged and clinical-ly examined mice containing informative recombinations at 4–8-wk inter-vals until the mice reached 12 mo. Because the presence of both alleles wasapparent at 5–6 mo, we used only mice examined at an age of over 7 mo tosubsequently narrow the critical intervals. We used all recombinations forgenerating the genetic maps. Mice with recombinations generated fromthe intercross but homozygous for CAST alleles were progeny tested by re-mating to D2 for phenotypic interpretation.

Fig. 4 Genetic disruption of the pigment production pathway rescues IPD, ISAand glaucoma. The pe mutation was used to decrease flow through the pig-ment production pathway. All mice are homozygous for the GpnmbR150X andTyrp1b mutations. pe/+ mice (left column) are normally pigmented, whereaspe/pe mice (right column) have hypopigmented eyes and coats. Slit-lampexamination of these aged mice shows the rescuing effects of a decreasedlevel of pigmentation (a–f). a, A 2-mo pe/+ iris showing normal pigmentationand wildtype morphology (n=7 mice, 1–4 mo). b, A 2-mo pe/pe iris that ishypopigmented, resulting in a reddish hue but otherwise wildtype morphol-ogy (n=8 mice, 1–4 mo). c, A 13-mo pe/+ iris showing pronounced deteriora-tion and significant pigment dispersion characteristic of inbred D2 mice (n=7mice, 7–13 mo). d, A 12-mo pe/pe iris showing an intact iris with no indicationof ISA or IPD (n=6 mice, 7–13 mo). e, A 22-mo pe/+ iris that has deteriorated tothe point of transparency (n=7 mice, 15–22 mo). f, A partially dilated 22-mope/pe iris continuing to show rescued iris morphology (n=3 mice, 15–22 mo).Histological analysis of hypopigmented D2 irises confirms the clinical observa-tions (g,h). g, A 16-mo pe/+ iris showing atrophy of the iris stroma and iris pig-ment epithelium. h, A hypopigmented 12-mo pe/pe iris with normal irismorphology (n=3 mice of 8–9 mo, n=3 mice of 12–16 mo). Histological analysisof the retina and optic nerve head shows that hypopigmentation spares ani-mals from glaucomatous damage (i,j). i, A 16-mo pe/+ peripapillary regionshowing a pronounced excavation of the optic nerve head and loss of retinalganglion cells. j, A 16-mo pe/pe peripapillary region showing no signs of glau-comatous damage. (×400 (g,h), ×100 (i) and ×200 (j)).

a b

c d

e f

g h

i j

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

nature genetics • volume 30 • january 2002 85

We isolated BACs flanking ipd from a 129Sv-derived mouse CITB library(Research Genetics). As the identity given to each BAC correlates to theplate position it was isolated from, these identities may not correlate pre-cisely with those assigned by Research Genetics. The full names of the BACsflanking ipd are 454F16, 547F2, 281N17, 365K21, 385L17 and 230O24. Thesequence of BAC 454F16 is publicly available. We purchased polymorphicmicrosatellite markers from Research Genetics. We used BAC end-sequences to generate additional markers, including polymorphicmicrosatellites, single nucleotide polymorphisms and sequence tagged sites(STSs). The sequence and reaction parameters of all unique markers areavailable upon request. For ipd, we isolated a total of eight overlappingBACs spanning the F161–385S interval and examined each for the presenceof all expected STS markers to ensure the use of nonrecombinant BACs. Weamplified Gpnmb RT–PCR products from D2 and B6 eyes and sequencedthe complete coding region. The Jackson Laboratory Microchemistry Facil-ity carried out all the sequencing of mouse DNA. The amino-acid positionsfor Gpnmb and Tyrp1 are given based on the predicted precursor proteins.

Histologic analysis. We fixed enucleated eyes for plastic sectioning (0.8%paraformaldehyde and 1.2% glutaraldehyde in 0.08 M phosphate buffer,pH 7.4, or 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.2) orparaffin sectioning (3.2% formaldehyde, 0.7 M acetic acid, 61% ethanol)as previously described4,14. We classified iris lesions using previouslyreported criteria4,14.

Gene abbreviations. The official full names of genes used in this study are(with NCBI LocusLink ID numbers in parentheses): GPNMB, glycopro-tein (transmembrane) nmb (10457); TYRP1, tyrosinase-related protein 1(7306); and ABP1, amiloride-binding protein 1 (amine oxidase (coppercontaining)) (26). The following sequences have only interim names listedby NCBI LocusLink: IGF2BP3, IGF-II mRNA-binding protein 3 (10643);NLP_1, nucleoporin-like protein 1 (11097); PRO2738, hypothetical pro-tein PRO2738 (55495); and LR8, LR8 protein (28959). Gpnmb, Tyrp1, si,Igf2bp3, and Abp1 are official names listed with MGD.

Accession numbers. Human contig containing IGF2BP3 and GPNMB,NT_007749; human contig flanking TYRP1, NT_015047; mouse BAC454F16 flanking ipd, AC006949.

AcknowledgmentsWe thank F. Farley and J. Smith for assistance preparing the manuscript, C.Fickett and J. Martin for animal care; O. Savinova for technical assistance,and G. Cox, J. Naggert, and D. Gould for critical reading of the manuscript.This work was supported in part by grants from the National Institutes ofHealth. Scientific support services at The Jackson Laboratory are funded by agrant from the National Cancer Institute. Major funding was provided by theHoward Hughes Medical Institute. S.W.M.J. is an Assistant Investigator ofThe Howard Hughes Medical Institute.

Received 12 June; accepted 11 October 2001.

1. Epstein, D.L. in Chandler and Grant’s glaucoma (eds. Epstein, D.L., Allingham, R.R.& Schuman, J.) 220–231 (Williams & Wilkins, Baltimore, 1997).

2. Sugar, H.S. Pigmentary glaucoma. A 25-year review. Am. J. Ophthalmol. 62,499–507 (1966).

3. Campbell, D.G. & Schertzer, R.M. in The Glaucomas Vol. 2 (eds Ritch, R., Shields,M.B. & Krupin, T.) 975–991 (Mosby, St Louis, 1996).

4. John, S.W.M. et al. Essential iris atrophy, pigment dispersion, and glaucoma inDBA/2J mice. Invest. Ophthalmol. Vis. Sci. 39, 951–962 (1998).

5. Chang, B. et al. Interacting loci cause severe iris atrophy and glaucoma in DBA/2Jmice. Nature Genet. 21, 405–409 (1999).

6. Ritch, R., Steinberger, D. & Liebmann, J.M. Prevalence of pigment dispersionsyndrome in a population undergoing glaucoma screening. Am. J. Ophthalmol.115, 707–710 (1993).

7. Andersen, J.S. et al. A gene responsible for the pigment dispersion syndromemaps to chromosome 7q35–q36. Arch. Ophthalmol. 115, 384–388 (1997).

8. Turque, N. et al. Characterization of a new melanocyte-specific gene (QNR-71)expressed in v-myc-transformed quail neuroretina. EMBO J. 15, 3338–3350(1996).

9. Weterman, M.A. et al. nmb, a novel gene, is expressed in low-metastatic humanmelanoma cell lines and xenografts. Int. J. Cancer 60, 73–81 (1995).

10. Vijayasaradhi, S., Xu, Y., Bouchard, B. & Houghton, A.N. Intracellular sorting andtargeting of melanosomal membrane proteins: identification of signals forsorting of the human brown locus protein, gp75. J. Cell Biol. 130, 807–820 (1995).

11. Ibraghimov-Beskrovnaya, O. et al. Strong homophilic interactions of the Ig-likedomains of polycystin-1, the protein product of an autosomal dominantpolycystic kidney disease gene, PKD1. Hum. Mol. Genet. 9, 1641–1649 (2000).

12. Swank, R.T., Sweet, H.O., Davisson, M.T., Reddington, M. & Novak, E.K. Sandy: anew mouse model for platelet storage pool deficiency. Genet. Res. 58, 51–62(1991).

13. Lalouette, A., Guenet, J.L. & Vriz, S. Hotfoot mouse mutations affect the delta 2glutamate receptor gene and are allelic to lurcher. Genomics 50, 9–13 (1998).

14. Anderson, M.G. et al. Genetic modification of glaucoma associated phenotypesbetween AKXD-28/Ty and DBA/2J mice. BMC Genet. 2, 1 (2001).

15. Taylor, B.A. in Genetic Variants and Strains of the Laboratory Mouse Vol. 2 (eds.Lyon, M.F., Rastan, S. & Brown, S.D.M.) 1597–1659 (Oxford University Press,Oxford, 1996).

16. Zdarsky, E., Favor, J. & Jackson, I.J. The molecular basis of brown, an old mousemutation, and of an induced revertant to wild type. Genetics 126, 443–449(1990).

17. Johnson, R. & Jackson, I.J. Light is a dominant mouse mutation resulting inpremature cell death. Nature Genet. 1, 226–229 (1992).

18. Kobayashi, T. et al. The Pmel 17/silver locus protein. Characterization andinvestigation of its melanogenic function. J. Biol. Chem. 269, 29198–29205(1994).

19. Tai, T., Eisinger, M., Ogata, S. & Lloyd, K.O. Glycoproteins as differentiationmarkers in human malignant melanoma and melanocytes. Cancer Res. 43,2773–2779 (1983).

20. Moyer, F.H. Genetic variations in the fine structure and ontogeny of mousemelanin granules. Am. Zool. 6, 43–66 (1966).

21. Kobayashi, T., Imokawa, G., Bennett, D.C. & Hearing, V.J. Tyrosinase stabilizationby Tyrp1 (the brown locus protein). J. Biol. Chem. 273, 31801–31805 (1998).

22. Halaban, R. & Moellmann, G. Murine and human b locus pigmentation genesencode a glycoprotein (gp75) with catalase activity. Proc. Natl Acad. Sci. USA 87,4809–4813 (1990).

23. Kobayashi, T. et al. Tyrosinase related protein 1 (TRP1) functions as a DHICAoxidase in melanin biosynthesis. EMBO J. 13, 5818–5825 (1994).

24. Boissy, R.E., Sakai, C., Zhao, H., Kobayashi, T. & Hearing, V.J. Human tyrosinaserelated protein-1 (TRP-1) does not function as a DHICA oxidase activity in contrastto murine TRP-1. Exp. Dermatol. 7, 198–204 (1998).

25. Feng, L. et al. The β3A subunit gene (Ap3b1) of the AP-3 adaptor complex isaltered in the mouse hypopigmentation mutant pearl, a model forHermansky–Pudlak syndrome and night blindness. Hum. Mol. Genet. 8, 323–330(1999).

26. Wistrand, P.J., Stjernschantz, J. & Olsson, K. The incidence and time-course oflatanoprost-induced iridial pigmentation as a function of eye color. Surv.Ophthalmol. 41, S129–S138 (1997).

27. Stjernschantz, J., Ocklind, A., Wentzel, P., Lake, S. & Hu, D.N. Latanoprost-inducedincrease of tyrosinase transcription in iridial melanocytes. Acta Ophthalmol.Scand. 78, 618–622 (2000).

28. Nylander, K. et al. Transcriptional activation of tyrosinase and TRP-1 by p53 linksUV irradiation to the protective tanning response. J. Pathol. 190, 39–46 (2000).

29. Manga, P. et al. Rufous oculocutaneous albinism in southern African Blacks iscaused by mutations in the TYRP1 gene. Am. J. Hum. Genet. 61, 1095–1101(1997).

30. Boissy, R.E. et al. Mutation in and lack of expression of tyrosinase-related protein-1 (TRP-1) in melanocytes from an individual with brown oculocutaneous albinism:a new subtype of albinism classified as OCA3. Am. J. Hum. Genet. 58, 1145–1156(1996).

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m