ORIGINAL ARTICLE See related Commentary on page vi Interactions of Human Myosin Va Isoforms, Endogenously Expressed in Human , Are Tightly Regulated by the Tail Domain

Wendy Westbroek, Jo Lambert, Philippe Bahadoran,n Roser Busca,n Marie Chantal Herteleer, Nico Smit,w Mieke Mommaas,z Robert Ballotti,n and Jean Marie Naeyaert n Department of Dermatology, Ghent University Hospital, B-Gent, Belgium; UniteŁ de Recherches sur la Biologie et la Physiopathologie de la Peau, INSERM, FaculteŁ de MeŁ decine, Nice, France; wDepartment of Dermatology, Leiden University Medical Center, the Netherlands; zCenter for Electron Microscopy, Leiden University Medical Center, the Netherlands

Primary human epidermal melanocytes express six en- Our data show that isoforms lacking exon F but con- dogenous isoforms of the human actin-associated myo- taining exon D are associated with vesicles near the sin Va motor protein, involved in transport. Golgi area. Myosin Va isoforms containing exon F are As isoforms containing exon F are most abundant in able to colocalize with and in£uence melanosome dis- melanocytes, we hypothesized that these isoforms tribution by indirect interaction with rab27a and direct probably have a -speci¢c function. To interaction with melanophilin. These results indicate uncover the biologic role of the six isoforms we intro- that the myosin Va medial tail domain provides the duced enhanced green £uorescent protein (eGFP)-myo- globular tail domain with organelle-interacting speci¢- sin Va tail constructs in human melanocytes. We found city. Key words: myosin VA/Rab27a/melanophilin/alternative that the medial tail, undergoing alternative splicing, has splicing/exon F/primary human melanocyte/Griscelli/pigmen- to be expressed in combination with the globular tail in tation. J Invest Dermatol. 120:465 ^475, 2003 order to obtain clear colocalization with .

yosin Va encoded by the dilute gene (d)inmice encoding 3, 27, and 25 amino acid long stretches (Lambert et al, and the Griscelli locus in humans, is a dimer that 1998a). Interestingly, the full-length transcripts containing exons through its mechanochemical N-terminal head ABCDEF and ACDEF were most abundant in human melano- domain facilitates movement of a cargo along cytes and leukocytes, types involved in the human autosomal subcortical actin ¢laments (Nascimento et al, recessive (Griscelli et al, 1978). Indeed, in a M1996; Cheney, 1998). Its carboxyterminal globular tail is asso- minority of Griscelli patients myosin Va mutations were detected ciated, at least indirectly, with subcellular organellar structures, (Pastural et al, 1997, 2000; Lambert et al, 2000), which leads to an such as the (Dekker-Ohno et al,1996; in vivo and in vitro aberrancy in the positioning of melanosomes, Takagishi et al, 1996), the (Nascimento et al,1997; which are the pigment-bearing organelles in human melanocytes Bizario et al, 2002), (Catlett and Weisman, 1998), and (Lambert et al, 1999). Normally, as melanosomes mature they are synaptic vesicles (Prekeris and Terrian, 1997), with - transported towards the periphery of the melanocytic rich domains (Wu et al, 1998a) and melanosomes (Lambert et al, and transferred to the surrounding . Even distribu- 1998b). Moreover, the tail domain is targeted by the microtu- tion of in the cytoplasm of the latter cells gives rise to bule-associated motor (Huang et al, 1999) and a human skin pigmentation. When myosin Va is malfunctioning, light chain (Benashski et al, 1997; Naisbitt et al,2000;Loet al, melanosomes display a central perinuclear accumulation in mi- 2001), allowing £uent transition of vesicles from to crotubule-rich perinuclear areas of the melanocyte (Bahadoran actin ¢laments (Westbroek et al, 2001). The C-terminal domain of et al, 2001). Phenotypically, this leads to pigment dilution of the mouse myosin Va was already shown to undergo alternative hair, and to a lesser extent, of the skin (Mancini et al,1998).Pre- splicing in its medial tail domain, with exon combination viously,Wu et al (1998a) formulated a ‘‘capture hypothesis model’’ ACDEF reported to be a melanocyte-speci¢c isoform, whereas stating that melanosomes move quickly up and down the den- lack of exons D/F appeared brain-speci¢c (Seperack et al,1995; drite along microtubules mediated by microtubule motors until Huang et al, 1998). In 1998, our group reported the existence of they are captured in the peripheral subcortical actin network by ¢ve endogenously expressed human myosin Va alternative tran- myosin Va, ready for transfer towards the neighboring keratino- scripts, with alternative usage of exons B, D, and F, respectively, cyte target cells. The necessity of the C-terminal part of myosin Va in melanosome tethering was demonstrated by introduction of a dominant negative myosin Va-eGFPtail construct in healthy Manuscript received May 29, 2002; revised September 24, 2002; accepted for publication October 30, 2002 mouse melanocytes. This gave rise to an abnormal perinuclear ac- Reprint requests to: J.M. Naeyaert M.D., Ph.D., Department of cumulation of mature melanosomes. Thanks to recent genetic, Dermatology, Gent University Hospital, De Pintelaan 185, B-9000 Gent, morphologic, and biochemical data from dilute, ashen,andleaden Belgium. Email: [email protected] mouse mutant melanocytes, all showing the phenotype of central

0022-202X/03/$15.00 . Copyright r 2003 by The Society for Investigative Dermatology, Inc.

465 466 WESTBROEK ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY melanosome accumulation, a cooperative tripartite complex for staining, and confocal microscopy on living cells were obtained from recruitment of myosin Va on the melanosome membrane has neonatal foreskin and established as described previously (Naeyaert et al, been proposed (Marks and Seabra, 2001). Rab27a, a Ras-like 1991; Smit et al, 1998). Immunoelectron microscopy was also performed small guanosine triphosphate GTPase encoded by the ashen on the human metastatic melanoma cell line MM-AN, maintained as previously described (Byers et al, 1991). For the coimmunoprecipitation locus (Wilson et al, 2000) and mutated in the majority of assays, mouse melanoma B16 cells were cultured in Dulbecco minimal described Griscelli patients (MeŁ nascheŁ et al, 2000), directly Eagle’s medium with 10% fetal bovine serum and penicillin/streptomycin. associates with mature melanosomes in its GTP-bound form (Bahadoran et al, 2001). Melanophilin, encoded by the leaden locus, is a rab27a e¡ector, which is recruited by rab27a-GTPto Constructs cDNA was prepared from total RNA using the Superscript enzyme (Invitrogen Merelbeke, Belgium). The di¡erent myosin Va cDNA the melanosomal surface by direct interaction with the N-term- regions of our interest were ampli¢ed with Extaq (Takara Boechout, inal part of melanophilin. Eventually, myosin Va is recruited to Belgium) using the following primers based on GenBank accession the melanosome membrane via a direct interaction between the number Y07759 for the medial tail domain containing the di¡erent exon C-terminal portions of myosin Va and melanophilin (Kuroda combinations: (i) myoVKpn_3833: 50 -CGGGGTACCCCGCAAATTAGA et al, 2001; Matesic et al, 2001; Fukuda et al,2002;Wuet al,2002a). GGTGCAGAACT-30, and (ii) myo V4562_gg_ST: 50 -TTAGGCAAATAC In this report we want to uncover the biologic role of six endo- TTTCAGTTG-30; for the medial tail domain with the di¡erent exon genously expressed myosin Va isoforms in human melanocytes. combinations plus the C-terminal globular tail: (i) myoVKpn_3833: 50 - CGGGG TACCCCGCAAATTA GAGGTGCAGAACT-30, and (ii) The phenomenon of creating di¡erent biologic functions for 0 0 motor proteins by means of alternative splicing is already well myoV_5818: 5 -TCAGACCCGTGAAATAAAGCCC-3 ; for the C- terminal globular tail alone: (i) myoVK pn_4505: 50 -CGGGGTACCCC illustrated for kinesin (Khodjakov et al, 1998; Gyoeva et al,2000) GCTTGAAAAACAGGATAAGACG-30, and (ii) myoV_5818: 50 -TCAGAC and cytoplasmic dynein (Nurminsky et al, 1998; Lee et al, 1999).To CCGTGAAATAAAGCCC-30. These myosin Va fragments were investigate whether the di¡erent myosinVa transcripts are of phy- directionally cloned into the KpnI/SmaI sites of pEGFP-C1 (Clontech, siologic relevance, we fused several tail sequences to eGFPand Erembodegem, Belgium) in frame with eGFP( Fig 1). For two-hybrid transfected them into primary human melanocytes. This study screening, the following bait plasmids were constructed by directional shows that the presence of exon F in myosin Va is a prerequisite insertion of myosin Va polymerase chain reaction products into the for colocalization with melanosomes and exon D mediates target- EcoRI^SmaI sites of the pBD-Gal4 vector (Stratagene, Amsterdam, The ing to Golgi-derived vesicles. Netherlands), using the following primers for the myosin Va control bait: (i) myoVEcoRI_4961: 50 -CCGGAATTCCGGTACAGTGGAGAAGA GGGCT-30, and (ii) myoVgg_5272_ST: 50 -GGTTAGACCGAGTGGAAGGAGT (Huang et al, 1999). For the medial tail domain with the di¡erent myosinVa MATERIALS AND METHODS exon combinations plus the globular tail domain: (i) myoVEcoRI_3833: 50 - CCGGAATTC CGGCA AATTAGAGGTGCAGAACT-30, and (ii) Cell culture The primary human epidermal melanocytes used for myoV5818_ST: 50 -GGTTAGA CCG AGTGGAAGGAGT-30. The following mRNA, cDNA preparation, immunoelectron microscopy, immuno- primers were used to clone human melanophilin (GenBank accession

Figure1. Overview of the di¡erent eGFP fused myosinVa constructs. Schematic drawing of the di¡erent domains of full length myosinVa. The ¢rst set of constructs consisted of six di¡erent exon combinations in the medial tail fused to eGFP. In the second set of constructs we also included the C- terminal globular tail of myosin Va. Finally, we fused the globular tail to eGFP. VOL. 120, NO. 3 MARCH 2003 HUMAN MYOSIN V SPLICE FORMS TARGET MELANOSOMES 467 number NM_024101) in pAD-Gal4^2.1 and pBD-Gal4Cam (both GFPantibody (gift from J. Fransen, Nijmegen) (1:1000) (Cuppen et al, Stratagene) via the EcoRI cloning site: mlph_50 -CCGGAATTCCGGT 1999), followed by 10 nm protein A-gold incubation. The grids were CCAAGCTCACTGATGAAGAG-30 and 50 -CCGGAATTCCG GTCTC contrasted with uranyl acetate and viewed with a Philips (Eindhoven, The TCTGTCCTGTCCCGTTA-30. The following target plasmids were Netherlands) EM 410 electron microscope. Several negative controls were constructed by BamHI/EcoRI directional cloning into the pAD-Gal4^2.1, performed. Labeling was performed excluding the polyclonal anti-GFPAB using the following primers: kinesin control target based on GenBank but including the 10 nm gold particles. The complete immuno-labeling accession number X65873: kinBamHI_2444 50 -CGCGGATCCGCGA was also performed on nontransfected cells. Finally, the experiments CTCATCAAAAACAGATCA GTAG-30 and kinEcoRI_ST_3202 50 - themselves also included a negative control, this due to the fact that not CCGGAATTCCGGTTACACTTGTTTGCCT CCTCCAC-30; rab27a every melanocyte was electroporated. All the negative controls showed no wild-type GenBank accession number NM_004580: rab27a_BamHI 50 - aspeci¢c gold label. CGCGGATCCGCGA TTATGTCTGATGGAGATTATG-30 and rab27a EcoRI 50 -CCGGAATTCCGGTTATGTCGTTACTTGACTTCTCA-30. Yeast two-hybrid screening and b-galactosidase assay The Gal4 Rab27a T23N [guanosine diphosphate (GDP) bound form] and rab27a based yeast two-hybrid vector system (Stratagene) in combination with Q78L (GTP-bound form) were created using polymerase chain reaction yeast strain Saccharomyces cerevisiae PJ69-4 (James et al,1996)wasset based mutagenesis (Stratagene Quickchange mutagenesis kit). We used up to detect a physical interaction between the myosin Va transcripts and rab27a wild type in pAD-Gal4^2.1 as a template in combination with rab27a or melanophilin. All constructs were tested for autoactivation by the following primers: rab27a T23N (forward) 50 -GACTCTGGTG cotransfection with empty pAD-Gal4^2.1 or pBD-Gal4 vector into TAGGGAAGAACAGTGTACT TTACCAATAT-30, rab27a T23N (reverse) PJ69-4a. Interactions were scored using SD-Leu-Trp-Ade, SD-Leu-Trp- 50 -CTGAGACCACATCCCTTCTTGTCA CATGAA-30, rab27a Q78L His and b-galactosidase ¢lter assay. As an additional positive control we (forward) 50 -CAGTTATGGGACACAGCAGGGCTGGA GAGGTTTCG used human kinesin (2444^3202 bp) and myosin Va (4961^5272 bp) TAGCTTA AC-30 and rab27a Q78L (reverse) 50 -GTCAATACCCTGT (Huang et al, 1999). The internal positive control, p53 and pSV40, GTCGTCCCGACCTCTCCAAAGCATCGAATTG-30. The cloning and negative control, pLamin and pSV40 from Stratagene were also strategies were initially performed in silico, using the PDRAW 32 acaclone assayed. program, a freeware program developed by Kjeld Olesen (http:// medlem.tripodnet.nu/acaclone/). Proper folding of the created fusion Immunoprecipitations Twenty hours after transfection with eGFP- proteins was tested with SMART (Simple Modular Architecture ACDEF-GT, eGFP-ABCEF-GT, eGFP-ABCE-GT, and eGFP-ABCDE- Research Tool, http://smart.embl-heidelberg.de). Constructs were GT, mouse B16 cells were washed with phosphate-bu¡ered saline and sequenced by Baseclear (Leiden, the Netherlands). To check the correct lyzed (1% Nonidet P40, 50 mM Tris pH 8, 150 m M NaCl and a protease sequence of exon F, which is not included in GenBank accession number inhibitor mix). Cells were scraped from the plate and centrifuged at Y07759, we used GenBank accession number AF090424. 13,000 r.p.m. (10,000 g) After coupling anti-GFPpolyclonal antibody (Abcam, Cambridge, UK) to protein A-Sepharose beads (Amersham Transfection and electroporation The human melanocytes were plated Biosciences), cell extracts were mixed with 1% Triton X-100 washed on glass coverslips and, at 70^80% con£uency, transfected with the beads, and incubated for 1 h at 41C. The beads were washed six times in appropriate ratio of plasmid DNA, enhancer, and e¡ectene according to lysis bu¡er and boiled in Laemmli bu¡er for 5 min. The entire eluates the manufacturer’s instructions (Qiagen Leusden, The Netherlands). were loaded on to a 5% and a 12.5% sodium dodecyl sulfate- Mouse B16 melanoma cells were transfected with the appropriate amount polyacrylamide gel electrophoresis gel (respectively to detect rab27a and of DNA mixed with lipofectAMINE (Invitrogen). Confocal microscopy the eGFP-myosin Va constructs) and transferred to nitrocellulose was performed on living and ¢xed cells between 24 h and 48 h after membranes (Hybond þ , Amersham Biosciences). transfection. For electroporation the primary melanocytes were collected in an amaxa certi¢ed cuvette (amaxa GmBH, Cologne, Germany). A Immunoblotting Membranes obtained from 5 and 12.5% sodium mixture of 100 ml nucleofector solution (amaxa GmBH) and 5 mg plasmid dodecyl sulfate-polyacrylamide gel electrophoresis were probed with a DNA was added. The cells were electroporated in the nucleofector monoclonal anti-GFPantibody (1: 3000) (Transduction Laboratories) and electroporator (amaxa GmBH) with the U_16 speci¢c nucleofector a monoclonal anti-rab27a antibody 1: 2000 (gift from M. Seabra, program. Oxford), respectively. Horseradish peroxidase-labeled anti-mouse antibody (1:4500) (Dako) was used as secondary antibody. The antigen-antibody Indirect immuno£uorescence and confocal microscopy Primary complexes were detected with the enhanced chemiluminescence kit epidermal melanocytes or B16 melanoma cells were transfected with the (Amersham Biosciences). described eGFP-constructs and ¢xed after 24^48 h of transfection in 3% paraformaldehyde for 20 min, incubated in 50 mM NH4Cl for 10 min and permeabilized with 0.2% Triton X-100 for 5 min. Cells were RESULTS labeled overnight at 41C with the following antibodies: mouse monoclonal NKI beteb antibody (1:40), aimed against the (pre)melanosomal silver The di¡erent medial tail exon combinations and the protein (Monosan, Uden, The Netherlands); polyclonal PEP1 antibody globular tail of myosin Va, expressed separately, do not (1: 200) made by Eurogentec (Angers, France) and aimed against the COOH terminus of the murine tyrosinase-related protein 1 (TRP-1); display a speci¢c subcellular distribution In 1998, we mouse monoclonal anti-rab27a (1:100) (Transduction Laboratories, showed that a speci¢c domain in the medial stalk domain of the Lexington, UK); mouse anti-human monoclonal Golgi antibody (1:40) human myosin Va gene undergoes alternative splicing through (Chemicon, Hampshire, UK), which stains Golgi zone in human cells; alternate usage of three exons termed exon B, D, and F. Five and anti-melanophilin antibody (1:100), made by Eurogentec by endogenously expressed isoforms, ABCDEF, ACDEF, ABCEF, immunizing rabbits with two immunogenic peptides at the N-terminal ABCDE, and ABCE, were picked up in human leukocytes and of the mouse Mlph. Because this anti-melanophilin antibody only a number of primary human skin cell types, including recognizes mouse melanophilin, we had to use B16 mouse melanoma cells melanocytes (Lambert et al, 1998a). The observation that exon B instead of human melanocytes. The secondary antibody was a biotinylated is speci¢c to the brain-spliced isoform in mice cannot be goat anti-mouse antibody or goat anti-rabbit (Amersham Life Science, Orsay Cedex, France) that was labeled with streptavidin Texas red (1: 50) extended to human melanocytes (Wu et al, 2002a). Moreover, (Amersham Life Science). Cells were mounted on glass slides with we could detect isoforms lacking exons D and F, which are mounting £uid containing Dabco anti-fading agent (Dako, Heverlee, considered to be brain-speci¢c, in human melanocytes. Later on, Belgium). The living and ¢xed melanocytes were viewed with a Leica we also detected the endogenous expression of a sixth isoform TCS SP2 confocal microscope using a HCX PL APO 63.0 Â1.20 W namely ACE in human melanocytes (unpublished). Although CORR UV water^immersion objective. Mid Z-plane cell sections are isoforms ACDE and ACEF could theoretically occur, we never shown. detected them in several normal cell types studied. To investigate the role of the six endogenously expressed alternative splice Immunoelectron microscopy Primary melanocytes electroporated variants, we cloned the medial tail containing exon with and MM-AN melanoma cells transfected with eGFP-ABCDEF-GT, eGFP-ACDEF-GT, eGFP-ABCEF-GT, eGFP-ABCE-GT, eGFP-ACE-GT, combinations ABCDEF, ACDEF, ABCEF, ABCDE, ABCE, and eGFP-ABCDE-GT were ¢xed in 0.1% glutaraldehyde for 1 h at room and ACE in frame to the C-terminal part of eGFP. Transfection temperature, pelleted, and embedded in 12% gelatin. The pellet was cut in primary human epidermal melanocytes gave an average 3 into 1 mm cubes, cryoprotected in 2.3 M sucrose and snapfrozen in liquid transfection e⁄ciency of about 5%, which was satisfying to nitrogen. Ultrathin cryosections were labeled with a rabbit polyclonal anti- obtain reproducible results. Confocal microscopy on living 468 WESTBROEK ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 2. The myosinVa medial tail domain or globular tail domain alone is not su⁄cient to interact with organelles. (A) Confocal imaging of living human melanocytes transfected with eGFP-ABCDEF gave a di¡use localization in the cytoplasm, low £uorescence could be observed in the nucleus but nucleoli were completely excluded. (B) Phase contrast image showed that melanosomes exhibit a normal peripheral distribution. Scale bars:20mm. (C) eGFP-globular tail localized as a distinct perinuclear dot. (D) Introducing this construct had no in£uence on melanosome distribution. Mid cell sections are shown. Scale bars:20mm. melanocytes was performed at 24 and 48 h after transfection. We Overexpression of the six di¡erent exon combinations plus decided to perform the initial localization experiments on living the C-terminal globular tail sequence creates a speci¢c cells instead of ¢xed cells to exclude misinterpretation of subcellular localization pattern The fact that overexpressing localization due to ¢xation artifacts. the myosin Va region of alternate exon usage and the globular The six di¡erent exon combinations fused to eGFPall gave tail alone did not reveal a speci¢c cytoplasmic pattern, brought the same homogeneous distribution pattern throughout the along the idea to overexpress a second set of constructs. whole cytoplasm. A less denser £uorescence was observed in Overexpression of myosin Va tail fragments was already shown the nucleus but the nucleoli were excluded (Fig 2A). No to block vesicle transport in other species (Wu et al,1998;Brown abnormalities in melanosome distribution (Fig 2B) were et al, 2001) and the tail domain is known to include certain motifs observed. These data suggest that the alternate isoforms created important in targeting binding partners (Reck-Peterson et al, in this medial tail region are by themselves not able to bind to 2000). Knowing this we generated six eGFPfusion constructs speci¢c subcellular structures. Picking out only a partial myosin containing the six exon combinations followed by the Va stretch by reverse transcription^polymerase chain reaction C-terminal globular tail domain (eGFP-ABCDEF-GT, eGFP- could of course cause misfolding of the derived protein structure ACDEF-GT, eGFP-ABCEF-GT, eGFP-ABCDE-GT, eGFP- and hence create a shift in function; however, this was carefully ABCE-GT, and eGFP-ACE-GT). Moreover, overexpression of excluded using the SMART program. Transfection of the myosin Va headless fragments could gradually compete with globular tail alone fused to eGFPalways resulted in targeting of endogenous full-length myosin Va for interaction with possible a perinuclear structure, presumably the microtubule organizing binding partners. As these fragments lack the actin-binding and center (MTOC) (Fig 2C), melanosome distribution appeared to adenosine triphosphate-binding motif, targeted structures might be normal (Fig 2D). gradually lose track from the actin creating a VOL. 120, NO. 3 MARCH 2003 HUMAN MYOSIN V SPLICE FORMS TARGET MELANOSOMES 469 dominant-negative e¡ect on the positioning of these targets. 24 h technical di⁄culties we had to perform our initial after transfection, the constructs eGFP-ACE-GT (Fig 3A), eGFP- immunoelectron microscopy experiments on MM-AN ABCE-GT (data not shown), and eGFP-ABCDE-GT (data not melanoma cells. Later we were able to repeat these experiments shown) localized in distinct dots over the entire cytoplasm of by electroporation of primary human melanocytes in which the the living melanocytes. For all three constructs, after 48 h of e⁄ciency was about 60%. We could conclude that the transfection, these dots had gradually bundled into perinuclear observations made for each construct were similar in both cell localization (Fig 3C,E) of the melanocytes. It was striking that types. Immunoelectron microscopy on electroporated primary the localization pattern of eGFP-ACE-GT and eGFP-ABCE-GT, human melanocytes with eGFP-ABCDE-GT clearly showed which was more like a dot, seemed di¡erent as eGFP-ABCDE- clustering of vesicles in the Golgi area (Fig 4C,D). GT, which was more stretched. Immuno-£uorescence confocal Immunoelectron microscopy on MM-AN melanoma cells microscopy on transfected primary melanocytes showed transfected with eGFP-ABCE-GT (Fig 4B) and incubated with retention in the after 48 h for eGFP-ABCDE- gold-conjugated anti-GFPantibody revealed localization on GT (Fig 3E^G). The other ¢ve constructs did not colocalize small vesicles, which were clustered around the MTOC. Similar with the Golgi marker. None of these three constructs induced images were observed for eGFP-ACE-GT (data not shown). The an alteration in melanosome position (Fig 3D,H). Owing to excessive amount of gold particles present around the MTOC is

Figure 3. Phenotypic e¡ects of myosin Va tail constructs lacking exon F. Overexpression of eGFP-ACE-GT in living melanocytes (A) gave a punc- tate pattern throughout the cytoplasm after 24 h of transfection. eGFP-ABCE-GT (C) gave a clearly visible perinuclear £uorescent aggregate after 48 h of transfection due to a dominant negative e¡ect on vesicular transport. Cells transfected with eGFP-ABCDE-GT (E) were ¢xed after 48 h and stained with anti-Golgi AB conjugated to Texas red (F). The merged image in (G) showed a clear colocalization pattern. None of these constructs showed alteration in melanosome distribution (B, D, H) Mid cell sections are shown. Scale bars:10mm. 470 WESTBROEK ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 3. Continued.

probably due to the overexpression of the eGFPconstruct. No (Fig 5E), eGFP-ACDEF-GT (data not shown), and eGFP- gold particles were detected on melanosomes. When primary ABCEF-GT (Fig 5I) were labeled with NKI-beteb (Fig 5F), an melanocytes were transfected with eGFP-ABCEF-GT (Fig 5A), antibody detecting the melanosomal silver protein and an anti- eGFP-ACDEF-GT, and eGFP-ABCDEF-GT (data not shown), TRP1 antibody (Fig 5J), speci¢c for end-stage melanosomes. £uorescence after 24^36 h was accentuated in a punctate The merged confocal microscopy images revealed partial peripheral rim (Fig 5A), with a less intense punctate pattern all colocalization of the eGFP-label with the melanosome markers over the cytoplasm. Melanosome distribution seemed normal in melanocytes where the dominant-negative e¡ect had not (Fig 5B). This pattern turned into a striking perinuclear occurred yet (data not shown) and nearly full colocalization accumulation of £uorescence (Fig 5C) and melanosomes in the where the dominant-negative e¡ect was already established (Fig living melanocytes (Fig 5D) after 48 h of transfection. 5G,K). The phase contrast pictures clearly show the accumulation This indicated that eGFP-ABCDEF-GT, eGFP-ACDEF- of melanosomes around the nucleus (Fig 5H,L). To support these GT, and eGFP-ABCEF-GT were able to induce a dominant- observations obtained at light microscopy level, we also negative ‘‘Griscelli-like’’ phenotype on melanosome transport. performed immunoelectron microscopy. Immunogold labeling The fact that these constructs have exon F in common indicates of electroporated primary human melanocytes using an anti- a speci¢c role for this exon in melanosome targeting by GFPantibody con¢rmed the presence of eGFP-ABCDEF-GT myosin Va. (Fig 4A), eGFP-ACDEF-GT, and eGFP-ABCEF-GT (data not shown) on the melanosomal surface. Taken together these data Myosin Va tail constructs with exon F associate with strongly indicate that the presence of exon F in the C-terminal melanosomes To determine the role of exon F in melanosome part of myosin Va is of major importance for, either direct or targeting, melanocytes transfected with eGFP-ABCDEF-GT indirect, association with melanosomes. VOL. 120, NO. 3 MARCH 2003 HUMAN MYOSIN V SPLICE FORMS TARGET MELANOSOMES 471

Figure 4. Immunoelectron microscopy shows colocalization of exon F tail constructs with melanosomes. Ultrathin cryosections of primary human melanocytes and MM-AN human melanoma cells, electroporated or transfected with various myosin Va eGFP-GT constructs, were labeled with anti-GFPantibody detected with 10 nm gold particles. Electroporation of eGFP-ABCDEF-GT ( A) in primary human melanocytes shows gold labeling on membranes of melanosomes, which are readily recognizable as electron-dense structures and are indicated by an arrow. Ten nanometer gold particles were rarely observed on other organelles or membranous structures. These observations are representative for all myosin Va tail constructs containing exon F. MM-AN melanoma cells transfected with eGFP-ABCE-GT (B) and immuno-labeled with gold conjugated anti-GFPantibody.Ten nanometer gold parti- cles were mainly located on vesicles, which were clustered around the MTOC, indicated by an asterix. Arrowheads indicate Golgi stacks. These observations are also representative for eGFP-ACE-GT. Primary human melanocytes electroporated with eGFP-ABCDE-GT (C) give labeling in the Golgi area, which is indicated by the arrows. A magni¢cation (D) of the rectangle in C clearly shows 10 nm gold particles on vesicles localized near the Golgi area. Scale bars: 0.2 mm.

Rab27a colocalizes and coimmunoprecipitates but does not whether the exon F containing myosin Va tail constructs interact directly interact with the myosin Va transcripts containing with rab27a to target melanosomes; therefore, we investigated exon F Rab27a is known to be associated with the whether these transcripts were able to coimmunoprecipitate melanosomal membrane in human (Bahadoran et al,2001)and with endogenously expressed rab27a. B16 mouse melanoma cell mouse (Wu et al,2001;Humeet al, 2002) melanocytes. extracts were made after transfection with several eGFP Moreover, a partial colocalization between rab27a and constructs. As shown in Fig 7, eGFP-ACDEF-GT (lane 3) endogenous myosin Va has been shown in primary human and eGFP-ABCEF-GT (lane 4) coimmunoprecipitated with melanocytes (Bahadoran et al, 2001). Melanocytes were rab27a using an anti-GFPantibody, whereas neither the tail transfected with eGFP-ABCDEF-GT, eGFP-ACDEF-GT constructs lacking exon F, namely eGFP-ABCE-GT (lane 5) and (Fig 6A), and eGFP-ABCEF-GT, ¢xed after 36 h and labeled eGFP-ABCDE-GT (lane 2), nor the eGFP-C1 (lane1) did. with an anti-rab27a antibody (Fig 6B). Colocalization with It is striking that eGFP-ACDEF-GT (lane 3) always seemed to rab27a could be observed for the myosin Va tail constructs with coimmunoprecipitate in a slightly less e⁄cient manner than exon F. At ¢rst, colocalization was partial (not shown), but eGFP-ABCEF-GT (lane 4). To address the fact whether this became complete (Fig 6C) after the melanosomes had moved interaction between the exon F containing transcripts and towards the perinuclear area (Fig 6D). Apparently, the exon F rab27a is direct, we performed yeast two-hybrid screening myosin Va tail constructs not only had a dominant-negative with rab27a wild-type, rab27a-GDPand rab27a-GTPas a prey e¡ect on melanosome distribution but also on the rab27a and the six medial tail exon combinations as well as the cytoplasmic distribution. These observations were never made exon combinations, including the globular tail as a bait, did not for myosin Va constructs lacking exon F. The question rose reveal a direct interaction between these two proteins (Tab l e I ). 472 WESTBROEK ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 5. Phenotypic e¡ects of myosin Va tail constructs containing exon F. Melanocytes transfected with eGFP-ABCEF-GT were live observed after 24 h (A) and 48 h (C). After 24 h a peripheral punctate localization was observed (A), whereas melanosome distribution was normal (B). After 48 h we detected £uorescent (C) as well as melanosomal clustering (D) around the nucleus. This indicated the occurrence of a dominant-negative e¡ect on melano- some transport. Primary human melanocytes, transfected with the exon F tail construct eGFP-ABCDEF-GT (E) and eGFP-ABCEF-GT (I)were¢xed, permeabilized, and incubated with NKI-beteb (F) or anti-TRP1 antibody (J). The merged image showed full colocalization of eGFP-ABCDEF-GT with NKI-beteb melanosomal marker after generation of a ‘‘Griscelli like’’phenotype (G) and full colocalization of eGFP-ABCEF-GT with anti-TRP-1 (K). The dominant negative e¡ect on melanosome transport is clearly visible on the phase contrast images (H,L). Again, these images are representative for what was observed with all myosin Va tail constructs containing exon F. Mid cell sections are shown. Scale bars:10mm.

Melanophilin directly interacts and colocalizes with the mouse melanoma cells instead of human melanocytes. myosin Va transcripts containing exon F Apparently, rab27a Colocalization between the myosin Va exon F constructs and and myosin Va transcripts with exon F are functioning in a melanophilin was observed in mouse B16 melanoma cells complex, but are not directly interacting. With the ¢nding that (Fig 6F, G ). These results taken together show that, in humans, the leaden locus encodes a rab e¡ector, termed melanophilin, it the myosin Va transcripts containing exon F undergo association has been suggested that rab27a would recruit myosin Va to the with the melanosome by direct interaction with melanophilin, melanosomes aided by melanophilin (Marks and Seabra, 2001; which binds directly to rab27a, the actual melanosome Matesic et al, 2001). We investigated the direct interaction of receptor. melanophilin with the several human myosin Va transcripts by using yeast two-hybrid screening. When testing the medial tail domain with the six exon combinations alone no interaction DISCUSSION with melanophilin was seen. Myosin Va bait constructs containing exon F and the globular tail, being ABCDEF-GT, In 1998, we found that human myosin Va undergoes alternative ACDEF-GT, and ABCEF-GT, gave a speci¢c interaction with splicing in its medial tail domain in a cell type-speci¢c manner melanophilin. Because the anti-melanophilin antibody could (Lambert et al, 1998a). The goal of this study was to gain better only recognize mouse melanophilin (Fig 6E), we had to use B16 insight into the physiologic role of all six endogenously expressed VOL. 120, NO. 3 MARCH 2003 HUMAN MYOSIN V SPLICE FORMS TARGET MELANOSOMES 473

Figure 6. Myosin Va exon F transcripts are associated with rab27a and melanophilin. Human melanocytes were transfected with eGFP-ACDEF- GT (A) and after ¢xation incubated with anti-rab27a antibody (B). After 36 h of transfection the dominant-negative e¡ect was not only visible on melano- some transport (D) but also on rab27a distribution (C). Indirect immuno£uorescence staining with the anti-melanophilin antibody on mouse B16 cells shows a speci¢c punctate pattern in the cytoplasm and an accumulation in the tips of the dendrites (E). After transfection with eGFP-ABCEF-GT (G) and ¢xation, mouse B16 melanoma cells were incubated with anti-melanophilin antibody (F). The same observation as for rab27a could be made for mel- anophilin. Introducing a dominant-negative exon F myosin Va tail construct completely redistributed melanophilin (F) towards the microtubule-rich peri- nuclear region (indicated by arrows). Mid cell sections are shown. Scale bars:15mm.

Figure 7. MyosinVa exon F transcripts are associated with rab27a in vivo. (A) Coimmunoprecipitation of eGFP-ABCDE-GT (lane 2), eGFP-ACDEF- GT (lane 3), eGFP-ABCEF-GT (lane 4), and eGFP-ABCE-GT (lane 5) transfected in B16 melanoma cells with endogenous rab27a, was performed with an anti-GFPantibody and blotted with anti-rab27a. Cells transfected with an empty peGFP-C1vector ( lane 1) gave no positive result nor did the nontransfected cells (lane 6). Only exon F constructs were able to coimmunoprecipitate. The band of the construct containing exon D and F (eGFP-ACDEF-GT) was less intense than the band of the construct only containing exon F (eGFP-ABCEF-GT). (B) As an internal loading control the same amount of each B16 lysate was put on gel. 474 WESTBROEK ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Table I. Exon F myosin Va transcripts directly interact with group revealed an interaction between melanophilin and melanophilin and not with rab27a in yeast two hybrid rab27a, and between melanophilin and myosin Va. Their screening. The di¡erent myosin Va medial and globular tail con- in vitro binding assay, however, revealed that the myosin Va structs were cloned into the pBD-Gal4 vector, whereas full-length globular tail as such is able to undergo a weak interaction with human melanophilin, rab27aWT, rab27a-GTP, and rab27a-GDP into melanophilin (Fukuda et al, 2002), indicating the possible interac- the pAD-Gal4-2.1 vector. After cotransfection into PJ69-4a and as- tion of the globular tail alone with melanosomes. As illustrated saying on SD-Leu-Trp-Ade, SD-Leu-Trp-His and a-galactosidase by our data, we were never able to see colocalization of the ¢lter lift, constructs containing exon F directly interacted with mel- globular tail domain with melanosomes. Transfection in human anophilin whereas the others did not. None of the constructs inter- melanocytes gave a low di¡use cytoplasmic signal but next to acted with rab27a WT, rab27a-GTP, and rab27a-GDP. The positive that also a distinct perinuclear dot, possibly the MTOC. This is controls myosin Va (bp4961-5272) plus kinesin (bp2444-3202) and comparable with transfections of this domain performed in B16 p53 plus pSV40 gave the expected positive interaction. The negative mouse melanoma cells (Nascimento et al, 1997) but in contradic- control pSV40 plus pLamin gave no growth on both SD-Leu-Trp- tion with recent data where the globular tail transfected in mouse Ade and SD-Leu-Trp-His melanocytes showed a rather unspeci¢c colocalization (Wu et al, 2002a). An explanation for interaction of the globular tail with SD-Leu- SD-Leu- -galactosidase melanophilin could be that the globular tail exhibits low binding Trp -Ade Trp -His a⁄nity towards melanosomes, which was beyond the detection level of our assays and those of other groups. Our data strongly ABCDEF-GT þmelanophilin þþ þ indicate that exon F mediates the speci¢c interaction of the ACDEF-GT þmelanophilin þþ þ globular tail with melanosomes but is not able by itself to bind ABCEF-GT þmelanophilin þþ þ melanosomes. Perhaps narrowing the domain containing the ABCDE-GT þmelanophilin exact binding site on myosin Va (now broadly limited to the 620 ABCE-GT þmelanophilin C-terminal residues of mouse myosin Va) for melanophilin, and ACE-GT þmelanophilin thus for melanosomes, might help to obtain better insight into this discrepancy. The three transcripts lacking exon F, colocalized with other ve- myosin Va transcripts (ABCDEF, ACDEF, ABCEF, ABCDE, sicular structures in melanocytes than melanosomes. Transfection ABCE, and ACE) in primary human epidermal melanocytes. of eGFP-ABCDE-GT, eGFP-ABCE-GT, and eGFP-ACE-GT re- Transfection of the alternate splice region in the medial tail fused sulted in a displacement of vesicles towards the perinuclear area, to eGFPin melanocytes revealed that this region as such is not indicating the appearance of a dominant-negative e¡ect on vesicle su⁄cient to colocalize with any particular structure at light mi- transport and the involvement of these isoforms in vesicle posi- croscopy level. Confocal microscopy on human primary melano- tioning. eGFP-ABCDE-GT colocalized with the Golgi-apparatus cytes transfected with the medial tail alternate splice forms after appearance of the dominant-negative e¡ect on vesicle trans- combined with the globular tail domain, however, revealed colo- port. Our data suggest that the presence of exon D in the non- calization with and redistribution of melanosomes in melano- exon F isoforms mediates speci¢c interaction with Golgi-derived cytes with exon F containing constructs and colocalization of vesicles through an as yet unidenti¢ed receptor. Looking at the Golgi-derived vesicles with constructs lacking exon F but con- coimmunoprecipitation assays of the eGFPconstructs with taining exon D. rab27a repeatedly revealed a better coimmunoprecipitation of Recently two models of melanosome transport were sug- the construct containing exon F but lacking exon D (eGFP-AB- gested. The ¢rst one suggested that melanophilin recruits rab27a CEF-GT) compared with the construct containing exon D and F to the melanosome, thereby acting as a rab27a stabilization mole- (eGFP-ACDEF-GT). A possible explanation could be that the cule. Rab27a in turn would recruit myosinVa to the melanosome, presence of exon D, which induces a Golgi-derived vesicle inter- hereby acting as a myosin Va ‘‘receptor’’.The second model sug- action, negatively regulates the interaction of exon F with rab27a gested that rab27a recruits melanophilin to the melanosome, and therefore also with melanosomes. These observations might which in turn targets myosin Va (Marks and Seabra, 2001). Using indicate that the exons display a mutual regulatory e¡ect. immuno£uorescence, confocal microscopy, and immunoelectron Immuno£uorescence confocal microscopy and immunoelec- microscopy on transfected melanocytes or melanoma cells, with tron microscopy revealed that, after 48 h of transfection, eGFP- we found that exon F is absolutely required for myosin Va to in- ABCE-GT and eGFP-ACE-GT clustered around the MTOC on teract with melanosomes in human melanocytes. Our yeast two- small, yet unidenti¢ed, vesicles. Transfection assays with eGFP- hybrid screening and coimmunoprecipitation assays also revealed ABCE-GT compared with eGFP-ACE-GT and also eGFP-ABC- a direct interaction of melanophilin and an indirect interaction of DEF-GT compared with eGFP-ACDEF-GT show that exon B, rab27a, with exon F transcripts. These data indicate that in human which is three amino acids long, does not appear to have any in- melanocytes the second melanosome transport model, where £uence on localization of the constructs. Our data also clearly rab27a serves as a myosin Va receptor on the melanosome mem- illustrate that ABCE, or the so-called ‘‘brain-speci¢c transcript’’ brane with melanophilin acting as a bridging molecule, is applic- (Seperack et al, 1995; Huang et al, 1998), does not show molecular able (Marks and Seabra, 2001). At the time of submission of our interaction or colocalization with melanosomes. This is in con- data, studies from two independent groups concerning the hier- trast to what is stated by Bizario et al (2002), who claim that trans- archy of interaction of the myosin Va, rab27a, and melanophilin/ fection of ABCE in mouse melanoma B16 cells can induce a Slac-2 complex with melanosomes, at least in mice, were pub- dominant negative e¡ect on melanosome motility. The fact that lished. The ¢rst group (Wu et al, 2002a,b) demonstrated, by using they used mouse B16 cells instead of human melanocytes cannot engineered myosin Va isoforms, that in mouse melanocytes exon be the reason for this di¡erence as our transfection experiments in F is also required for myosin Va to target melanosomes. Their human melanocytes were also simultaneously performed in yeast two-hybrid and immunoprecipitation assays with the engi- mouse B16 melanoma cells (data not shown), we never observed neered tail constructs show that mouse melanophilin directly and melanosomal accumulation around the nucleus after transfection mouse rab27a indirectly interacts with exon F transcripts, hereby with eGFP-ABCE-GT. Furthermore, all six myosin Va isoform also supporting the second model of melanosome transport. In- globular tail constructs showed the same (co)localization pattern terestingly, similar to our ¢ndings with human myosin Va iso- in human melanocytes and mouse B16 cells. forms, Wu et al (2002a) also saw that the interaction of myosin In summary, our data provide evidence, for the ¢rst time, that Va with melanosomes and melanophilin only occurred when the in humans part of the pigmentation process, namely melanosome carboxy-terminal globular tail domain was present in partnership transport, is speci¢cally regulated through three physiologically with the exon F-containing medial tail domains. The second expressed exon F-containing isoforms. These isoforms guide VOL. 120, NO. 3 MARCH 2003 HUMAN MYOSIN V SPLICE FORMS TARGET MELANOSOMES 475 their interaction with melanosomes through protein complex Lambert J, Onderwater J, Vander Haeghen Y,Vancoillie G, Koerten HK, Mommaas formation with melanophilin and rab27a. Through its nonexon AM, Naeyaert JM: Myosin V colocalizes with melanosomes and subcortical actin bundles not associated with stress ¢bers in human epidermal melano- F isoforms, myosin Va is able to undergo an interaction with cytes. JInvestDermatol111:835^840, 1998b Golgi-derived vesicles, putatively mediated by a receptor, which Lambert J,Vancoillie G, Naeyaert JM: Molecular motors and their role in pigmenta- remains to be identi¢ed. Our ¢ndings also indicate that the tion. Cell Mol Biol 45:905^918, 1999 region of myosin Va displaying alternative splicing alone is not Lambert J, Naeyaert JM, De Paepe A,Van Coster R, Ferster A, Song M, Messiaen L: su⁄cient to interact with or in£uence distribution of organelles. Arg-Cys substitution at codon 1246 of the human myosin Va gene is not asso- This suggests that the alternative splice region itself is not ciated with Griscelli syndrome. J Invest Dermatol 114:731^733, 2000 Lee S, Wisniewski JC, Dentler WL, Asai DJ: Gene knockouts reveal separate func- responsible for direct organelle interaction but rather provides tions for two cytoplasmic in Tetrahymena thermophila. MolBiolCell the globular tail with organelle-interacting speci¢city. 10:771^784, 1999 Lo KW, Naisbitt S, Fan JS, Sheng JS, Zhang M: The 8 kDa dynein light chain binds to its targets via a conserved ‘‘K/R-X-T-Q-T’’ motif. J Biol Chem 276:14059^ We thank Coby Out, Frans Van Nieuwpoort, and Martine De Mil for culturing of 14066, 2001 Mancini AJ, Chan LS, Paller AS: Partial with immunode¢ciency: Griscelli melanocytes and the melanoma cells; Jos Onderwater for assisting with electron micro- syndrome. report of a case and review of the literature. JAmAcadDermatol scopy; Lambert Verschragen for preparation of electron micrographs, January Slats for 38:295^300, 1998 assistance with the confocal microscope; Dr Valerie Frankard (Cropdesign NV,Ghent Marks MS, Seabra MC: The melanosome. membrane dynamics in black and white. Belgium) for providing the PJ69A4 yeast strain and the Stratagene control plasmids. Nature Rev Cell Biol 2:738^747, 2001 This work was supported by the National Fund for Scienti¢c Research (FWO, gran- Matesic LE, Yip R, Reuss AE, et al: Mutations in Mlph, encoding a member of the tnumber:G005300). Rab e¡ector family, cause the melanosome transport defects observed in lea- den mice. Proc Natl Acad Sci USA 98:10238^10243, 2001 MeŁ nascheŁ G, Pastural E, Feldmann J, et al: Mutations in RAB27A cause Griscelli syn- drome associated with haemophagocytic syndrome. Nature Genet 25:173^176, 2000 Naeyaert JM, Eller M, Gordon PR, Park HY, Gilchrest BA: Pigment content of cul- tured human melanocytes does not correlate with tyrosinase message level. Br J REFERENCES Dermatol 125:297^303, 1991 Naisbitt S, Valtschano¡ J, Allison DW, et al: Interaction of the postsynaptic density- Bahadoran P, Aberdam E, Mantoux F, et al: Rab27a: a key to melanosome transport 95/guanylate kinase domain-associated protein complex with a light chain of in human melanocytes. J Cell Biol 152:843^849, 2001 myosin-V and dynein. J Neurosci 20:4524^4534, 2000 Benashski SE, Harrison A, Patel-King RS, King SM: Dimerization of the highly Nascimento AA, Cheney RE, Tauhata SB, Larson RE, Mooseker MS: Enzymatic conserved light chain shared by dynein and myosin V. JBiolChem272:20929^ characterization and functional domain mapping of brain myosin V. JBiol 20935, 1997 Chem 271:17561^17569, 1996 Bizario JC, Nascimento A, Casaletti L, Patussi EV, Chociay MF, Larson RE, Esprea- Nascimento AA, Amaral RG, Bizario JC, Larson RE, Esprea¢co EM: Subcellular ¢co EM: Expression of constructs of the neuronal isoform of myosin-Va inter- localization of myosin-V in the B16 melanoma cells, a wild-type cell line for feres with the distribution of melanosomes and other vesicles in melanoma the dilute gene. Mol Biol Cell 8:1971^1988, 1997 cells. Cell Motil Cytoskeleton 51:57^75, 2002 Nurminsky DI, Nurminskaya MV, Benevolenska EV, Shevelyov YY, Hartl DL, Brown JR, Simonetta KR, Sandberg LA, Sta¡ord P, Langford GM: Recombinant Gzovdev VA: Cytoplasmic dynein intermediate-chain isoforms with di¡erent globular tail fragment of myosin-V blocks vesicle transport in nerve cell targeting properties created by tissue-speci¢c alternative splicing. MolCellBiol extracts. Biol Bull 201:240^241, 2001 18:6816^6825, 1998 Byers HR, Etoh T, Doherty JR, Sober AJ, Mihm MC: Cell migration and actin or- Pastural E, Barrat FJ, Dufourcq-Lagelouse R, et al: Griscelli disease maps to chromo- ganization in cultured human primary, recurrent cutaneous and metastatic some 15q21 and is associated with mutations in the myosin-Va gene. Nat Genet melanoma. Time-lapse and image analysis. Am J Pathol 139:423^435, 1991 16:289^292, 1997 Catlett NL, Weisman LS: The terminal tail region of a yeast myosin-V mediates its Pastural E, Ersoy F,Yalman N, et al: Two genes are responsible for Griscelli syndrome attachment to membranes and sites of polarized growth. Proc Natl Acad at the same 15q21 locus. Genomics 63:299^306, 2000 Sci USA 95:14799^14804, 1998 Prekeris R, Terrian DM: Brain myosin V is a synaptic vesicle-associated motor pro- Cheney RE: Puri¢cation and assay of myosin V. Methods Enzymol 298:3^18, 1998 tein: evidence for a Ca2 þ ^dependent interaction with the synaptobrevin-sy- Cuppen E, Wijers N, Schepens J, Fransen J, Wieringa B, Hendriks WA: FERM do- naptophysin complex. J Cell Biol 137:1589^1601, 1997 main governs apical con¢nement of PTP-BL in epithelial cells. JCellSci Reck-Peterson SL, Provance DW,Mooseker MS, Mercer JA: Class V myosin. Biochem 112:3299^3308, 1999 Biophys Acta 1496:36^51, 2000 Dekker-Ohno K, Hayasaka S,Takagishi Y, et al: Endoplasmic reticulum is missing in Seperack PK, Mercer JA, Strobel MC, Russell LB, Mooseker MS, Copeland NG, dendritic spines of Purkinje cells of the ataxic mutant rat. Brain Res 714:226^ Jenkins NA: Retroviral sequences located within an intron of the dilute gene 230, 1996 alter dilute expression in a tissue-speci¢c manner. EMBO J 14:2326^2332, Fukuda M, Kuroda TS, Mikoshiba K: Slac2-a/Melanophilin, the missing link be- 1995 tween Rab27 and myosin Va: implications of a tripartite protein complex for Smit NP, Kolb RM, Lentjes EG, et al: Variations in melanin formation by cultured melanosome transport. J Biol Chem 277:12432^12436, 2002 melanocytes from di¡erent skin types. Arch Dermatol Res 290:342^349, 1998 Griscelli C, Durandy A, Guy-Grand D, Daguillard F, Herzog C, Prunieras MA: A Takagishi Y, Oda S, Hayasaka S, Dekker-Ohno K, Shikata T, Inouye M, Yamamura syndrome associating partial albinism and immunode¢ciency. Am J Med H: The dilute-lethal (dl) gene attacks a Ca2 þ store in the dendritic spine of 65:691^702, 1978 Purkinje cells in mice. Neurosci Lett 215:169^172, 1996 Gyoeva FK, Bybikova EM, Minin AA: An isoform of kinesin light chain speci¢c for Westbroek W, Lambert J, Naeyaert JM: The dilute locus and Griscelli syndrome: the Golgi complex. JCellSci113:2047^2054, 2000 Gateways towards a better understanding of melanosome transport. Pigment Huang J, Mermall V,Strobel MC, Russell LB, Mooseker MS, Copeland NG, Jenkins Cell Res 14:320^327, 2001 NA: Molecular genetic dissection of mouse unconventional myosin-VA. tail Wilson SM,Yip R, Swing DA, et al: A mutation in Rab27a causes the vesicle trans- region mutations. Genetics 148:1963^1972, 1998 port defects observed in ashen mice. Proc Natl Acad Sci USA 97:7933^7938, Huang J, Brady S, Richards B, Stenoien D, Resau J, Copeland N, Jenkins N: Direct 2000 interaction of microtubule-and actin-based transport motors. Nature 397:267^ Wu X, Bowers B, Rao K, Wei Q, Hammer JA: Visualization of melanosome dy- 270, 1999 namics within wild-type and dilute melanocytes suggests a paradigm for myo- Hume AN, Collinson LM, Hopkins CR, et al: The leaden gene product is required sin V function in vivo. JCellBiol143:1899^1918, 1998a with Rab27a to recruit myosin Va to melanosomes in melanocytes. Tr a ⁄ c Wu X, Kocher B, Wei Q, Hammer JA: Myosin Va associates with microtubule-rich 3:193^202, 2002 domains in both interphase and dividing cells. Cell Motil Cytoskeleton 40:286^ James P, Halladay J, Craig EA: Genomic libraries and a host strain designed for 303, 1998b highly e⁄cient two- hybrid selection in yeast. Genetics 144:1425^1436, 1996 Wu X, Rao K, Bowers MB, Copeland NG, Jenkins NA, Hammer JA: Rab27a en- Khodjakov A, Lizunova EM, Minin AA, Koonce MP, Gyoeva FK: A speci¢c light ables myosin Va-dependent melanosome capture by recruiting the myosin to chain of kinesin associates with mitochondria in cultured cells. Mol Biol Cell the organelle. JCellSci114:1091^1100, 2001 9:333^343, 1998 Wu X, Rao K, Zhang H, et al: Identi¢cation of an organelle receptor for myosin-Va. Kuroda TS, Fukuda M, Ariga H, Mikoshiba K: The Slp homology domain of synap- Nature Cell Biol 4:271^278, 2002a totagmin-like proteins 1^4 and Slac2 functions as a novel Rab27A binding Wu X, Wang F, Rao K, James R, Sellers R, Hammer JA: Rab27a is an essential domain. JBiolChem277:9212^9218, 2001 component of melanosome receptor for Myosin va. Mol Biol Cell 13:1735^1749, Lambert J, Naeyaert JM, Callens T, De Paepe A, Messiaen L: Human myosin V gene 2002b produces di¡erent transcripts in a cell type-speci¢c manner. Biochem Biophys Res Commun 252:329^333, 1998a