Defects of the myosin VIIa motor protein cause deafness and retinal anomalies in humans and mice. We report on the identification of a novel myosin-VIIa-interacting protein that we have named MyRIP (myosin-VIIa- and Rab-interacting protein), since it also binds to Rab27A in a GTP-dependent manner. In the retinal pigment epithelium cells, MyRIP, myosin VIIa and Rab27A are associated with melanosomes. In transfected PC12 cells, overexpression of MyRIP was shown to interfere with the myosin VIIa tail localization. We propose that a molecular complex composed of Rab27A, MyRIP and myosin VIIa bridges retinal melanosomes to the actin cytoskeleton and thereby mediates the local trafficking of these organelles. The defect of this molecular complex is likely to account for the perinuclear mislocalization of the melanosomes observed in the retinal pigment epithelium cells of myosinVIIa-defective mice.
Mutations in the MYO7A gene, encoding myosin VIIa, cause auditory and retinal defects in mice and humans (for a review, see Petit, 2001). The primary structure of myosin VIIa is highly conserved between vertebrates and invertebrates, with a motor head domain, a neck region composed of five IQ (isoleucine–glutamine) motifs, and a long tail region [1359 amino acids (aa) in man]. The tail begins with a dimerization domain, followed by two large repeats of ∼460 aa, each containing a myosin tail homology 4 (MyTH4) and a 4.1, ezrin, radixin, moesin (FERM) domain, separated by a src homology type 3 (SH3) domain (Petit, 2001).
The myosin-VIIa-defective shaker-1 mice are characterized by a progressive disorganization of the hair cell stereocilia, i.e. the stiff actin-filled microvilli forming the mechanoreceptive structure to sound stimulation (Self et al., 1998). In the retina of shaker-1 mice, melanosomes of the pigment epithelium cells are abnormally concentrated in the perinuclear region (Liu et al., 1998), and opsin transport in the photoreceptor cells is delayed (Liu et al., 1999; Wolfrum and Schmitt, 2000). Together, these observations point to the existence of several different roles for this unconventional myosin.
To gain insight into the cellular functions of myosin VIIa, we searched for proteins interacting with the myosin VIIa tail using the yeast two-hybrid system (Kussel-Andermann et al., 2000a). We report here that myosin VIIa binds to a novel Rab effector protein that directly interacts with Rab27A, a Rab GTP-binding protein involved in the trafficking of skin melanosomes (for a review, see Marks and Seabra, 2001).
To identify proteins interacting with the tail of myosin VIIa, the C-terminal MyTH4 and FERM domains of the protein (last 464 aa) were used as the bait to screen a human retinal two-hybrid cDNA library (Kussel-Andermann et al., 2000a). A prey composed of 418 aa encoded by clone A7 was considered to be a specific ligand of myosin VIIa, since no interaction could be observed with two control proteins, namely lamin C and merlin/schwannomin (which also possesses a FERM domain; see Supplementary figure 1). Using the A7 cDNA as a probe, a 5.2 kb transcript was detected in most human and mouse tissues tested, namely brain, kidney, heart, liver, muscle and pancreas (data not shown). A full-length cDNA, A7FL (DDBJ/EMBL/GenBank accession No. AF396687), was reconstituted from a human retinal cDNA library (see Methods). The deduced amino acid sequence predicts a 859 aa (96 kDa) protein (Figure 1A), hereafter named MyRIP (myosin-VIIa- and Rab-interacting protein) based on functional considerations (see below).
Sequence analysis of MyRIP revealed the existence of an N-terminal FYVE (Fab1p, YOTB, Vac1P, EEA1) domain (Figure 1A and B). The MyRIP FYVE domain, which is highly conserved between the human and murine proteins (95% aa identity), possesses the eight conserved cysteine residues (Figure 1B) that are known to coordinate the binding to two Zn2+ ions. Homology search analysis revealed an homology between the MyRIP FYVE domain and the Rab3-binding domain of Rabphilin3A, Rim and Noc2; and with the Slp homology domain (SHD) of two members of the synaptotagmin-like protein (Slp) family, Slp2 and Slp3 (30% aa identity), and Slac2a/melanophilin (50% aa identity; Figure 1B). The crystal structure of the Rabphilin3A FYVE domain complexed to Rab3A (Ostermeier and Brünger, 1999) has highlighted the involvement of a motif, SGAWF(F/Y), in the interaction with Rab3A (Figure 1B). Interestingly, a similar, although not identical, motif (SLEWFY) is present in MyRIP (Figure 1B). Accordingly, the MyRIP amino acid sequence can be subdivided into three regions, the N-terminal FYVE domain (aa 1–134), the myosin-VIIa-binding region (myBD, aa 143–560) and the C-terminal region (aa 561–859).
In order to confirm the interaction between myosin VIIa and MyRIP, HEK293 mammalian cells were cotransfected with constructs encoding the myosin VIIa tail and either the full-length MyRIP or the MyRIP–myBD truncated form. Coimmunoprecipitation of the myosin VIIa tail with either MyRIP or MyRIP–myBD (Figure 1C) was demonstrated on cell extracts incubated with an anti-MyRIP antibody (anti-hA7; see Methods). Furthermore, in a pull-down experiment, the myosin VIIa tail bound to the GST-tagged MyRIP–ΔCt (aa 1–561) fragment but not to GST alone (data not shown). In addition, an in vitro binding assay demonstrated a direct interaction between the 464 aa tail fragment of myosin VIIa corresponding to the original bait and a MyRIP fragment (Figure 1D). Taken together, these results demonstrate that MyRIP specifically binds to myosin VIIa.
The presence of a putative Rab-binding motif in the MyRIP sequence prompted us to search for Rab GTPases interacting with MyRIP. The interaction of different Rab proteins with immobilized GST-tagged MyRIP–RabBD, MyRIP–ΔCt or GST was measured by means of a filter binding assay. Significant bindings of Rab27A–GTP to MyRIP–RabBD and MyRIP–ΔCt were detected, as compared with immobilized GST (Figure 2A and B). Since Rab3 isoforms A, B, C and D, Rab4, Rab5, Rab6, Rab11 and Rab26 did not bind to MyRIP–RabBD (Figure 2A), the binding of Rab27A–GTP to MyRIP–RabBD was considered specific. This binding reached saturation, and a Kd of ∼1 μM was derived (data not shown). Furthermore, Rab27A–GDP did not bind to MyRIP–RabBD, indicating that MyRIP specifically interacts with the GTP-bound form of Rab27A (Figure 2B). MyRIP–ΔCt exhibited no GTPase-activating protein (GAP) or guanine nucleotide exchange factor (GEF) activity (data not shown). The crystal structure of Rabphilin3A–Rab3A complex indicates that the SGAWFF motif (Figure 1B) constitutes one interface between Rabphilin3A and Rab3A molecules (Ostermeier and Brünger, 1999). Mutating this motif in Noc2 (SGAWFY to SGAAAA) impairs Noc2–Rab3A binding (Haynes et al., 2001). In contrast, the mutation of the MyRIP SLEWFY motif to SLEAAA was not detrimental to the binding of MyRIP to Rab27A (Figure 2C).
Specific antibodies were generated against MyRIP (see Methods) and were used to analyse the distribution of the protein on mouse tissue sections. MyRIP was detected in a variety of tissues, including brain, skin, heart, adrenal medulla, pancreas, intestine, liver, kidney, muscle and testis (data not shown). The distribution of MyRIP was further analysed, using the anti-hA7 antibody, in the mouse inner ear and retina, i.e. the two target organs of the myosin VIIa defect in mice and humans. In the mouse inner ear, MyRIP was detected in the cochlear and vestibular hair cells, where it was colocalized with myosin VIIa in the synaptic region and along the hair cell bundle (data not shown). In the retina, MyRIP was present in the photoreceptor cells and pigment epithelium cells (Figure 3A), i.e. the two cell types where myosin VIIa has been detected (Hasson et al., 1995; El-Amraoui et al., 1996). In the photoreceptor cells, MyRIP was localized in the synaptic region. Ultrastructural analysis of MyRIP localization showed that the protein was present in both pre- and post-synaptic areas (data not shown). In the pigment epithelium cells, MyRIP and Rab27A were observed in the microvilli that surround the tips of the photoreceptor outer segments (Figure 3A and B; see Supplementary data). In addition, analysis of the subcellular distribution of MyRIP by immunoelectron microscopy showed the presence of gold particles on melanosomes (Figure 3C and D). Myosin VIIa (Figure 3E; Liu et al., 1998) and Rab27A (Figure 3F) were also associated with retinal melanosomes.
Since MyRIP was present also in the synaptic areas of retinal photoreceptor cells (Figure 4A) and inner ear hair cells, we used the PC12 neuroendocrine cell line to determine whether MyRIP is able to recruit the myosin VIIa tail upon cotransfection. In differentiated PC12 cells producing GFP–myosin-VIIa tail, a punctate labelling of the cell body was observed, whereas the neurites were almost unstained (Figure 4A). In contrast, in PC12 cells expressing both the GFP–myosin-VIIa tail and myc-MyRIP, the myosin VIIa tail was enriched significantly at the tip of the neurites (Figure 4B). This enrichment appeared specific, since it was not observed upon cotransfection with a construct encoding either a MyRIP truncated form, lacking the Rab-binding protein, or the myc-tagged fruitfly luciferase (data not shown). We then examined the involvement of each of the three MyRIP regions in the protein targeting (Figure 4C). We found that, in PC12 cells expressing a MyRIP–RabBD construct, MyRIP–RabBD labelling was observed essentially at the tip of the neurites (Figure 4E, arrows), reminiscent of that observed in cells producing full-length MyRIP. In contrast, no such enrichment was observed at the tips of the neurites in cells producing either MyRIP–myBD, MyRIP–ΔRabBD or MyRIP–Ct (Figure 4D and F). Together, these results show that the targeting of MyRIP to the synaptic regions at the tip of the neurites is mediated by its Rab-binding domain. They also show that MyRIP is able to promote the recruitment of myosin VIIa at these emplacements.
We have identified a novel Rab effector protein, MyRIP, that binds to myosin VIIa. Several lines of evidence argue in favour of MyRIP providing a link between the actin cytoskeleton and organelles: (i) MyRIP directly interacts with the actin-based motor protein myosin VIIa and with Rab27A; (ii) MyRIP, Rab27A and myosin VIIa are detected on retinal melanosomes; and (iii) MyRIP is able to interfere with myosin VIIa tail localization and, in particular, to promote its recruitment, via the Rab-binding domain, to the tip of the neurites in differentiated PC12 cells.
Previous studies have implicated Rab27A in the transport of the skin melanosomes (Marks and Seabra, 2001). Mutations of Rab27a were found in ashen mutant mice, which exhibit clumping of melanosomes in the perinuclear region of skin melanocytes (Wilson et al., 2000), and in patients affected by Griscelli syndrome, which associates with partial albinism and immunodeficiency (Menasche et al., 2000). Myosin Va has also been shown to be involved in melanosome trafficking, since Myo5a mutations have been identified both in dilute mice, which exhibit defects in the trafficking of skin melanosomes (Mercer et al., 1991), and in patients affected by Griscelli syndrome (Pastural et al., 2000). Moreover, several studies on skin melanocytes have led to the conclusion that Rab27A recruits myosin Va onto melanosomes and thereby enables the capture and local actin-based movement of these organelles towards the cell periphery (Bahadoran et al., 2001; Hume et al., 2001; Wu et al., 2001). These studies thus demonstrate that myosin Va and Rab27A are part of a melanosome trafficking machinery. However, it is not yet clear whether these molecules bind directly to each other. Our results show that MyRIP provides a physical link between Rab27A, on the retinal melanosome membrane, and the actin-based motor molecule myosin VIIa. We propose that myosin VIIa, MyRIP and Rab27A act together to coordinate the short range peripheral movement of melanosomes along the actin-rich microvilli of retinal pigment epithelium cells (Figure 5). As a corollary, the loss of a functional myosin-VIIa–MyRIP–Rab27A complex is likely to account for the perinuclear accumulation of the melanosomes observed in the retinal cells of myosin-VIIa-defective shaker-1 mice (Liu et al., 1998). However, shaker-1 mice do not exhibit visual impairment, which indicates that the proper localization of the retinal melanosomes is not required for retinal viability.
During the completion of this study, Matesic et al. (2001) reported that leaden, another coat pigmentation mouse mutant, results from a defect in Slac2a/melanophilin, and Kuroda et al. (2002) found that Slac2a/b and Slp1-4 interact with Rab27A. MYRIP and melanophilin show no homology, and melanophilin lacks the region that, in MyRIP, binds to myosin VIIa. However, the two proteins shared similar FYVE domains, both of which contain two Zn2+-binding sites followed by a SLEW(F/Y)Y motif (see Figure 1B). Slac2b and Slp1-4 also possess a similar motif in their SHD2 domain, e.g. SGQWFY in Slp2 protein (Kuroda et al., 2002). We show here that mutating the SLEWFY motif of MyRIP does not impair its binding to Rab27A. The MyRIP–Rab27A interaction thus may involve the N-terminal α-helical segment, upstream of the Zn2+-binding sites, a region that has been shown to mediate Rim–Rab3A binding (Sun et al., 2001; Wang et al., 2001). Interestingly, the melanophilin defect in leaden mice is due to the deletion of 7 aa (REEERLQ) between aa 31 and 37, i.e. within the corresponding α-helical segment of the protein (Matesic et al., 2001). Similar amino acids are observed in MyRIP (KEEERLS, aa 31–37), Rim (TEEERNI, aa 29–35), Slac2b and Slps, thus suggesting that the Rab-binding sites of all these proteins are located within their N-terminal α-helical domain (or SHD1 domain, according to Kuroda et al., 2002). By analogy with the interactions described here, we consider that the leaden phenotype is likely due to a disruption of melanophilin binding to Rab27A. In any case, the finding that leaden results from a defective MyRIP-related protein supports our proposal that MyRIP is involved in melanosome trafficking. Therefore, peripheral melanosome trafficking in retinal pigment epithelium cells and in skin melanocytes would depend on the same Rab protein, Rab27A, but would involve two different motor proteins, myosin VIIa and myosin Va, respectively.
The presence of MyRIP in the synaptic areas of inner ear hair cells and retinal photoreceptor cells suggests that this protein is involved in the trafficking of other organelles, including synaptic vesicles. In PC12 cells, MyRIP is able to target the myosin VIIa tail to the tips of the neurites, where secretory vesicles are concentrated. Preliminary data suggest that MyRIP is associated with secretory vesicles in these cells. Interestingly, when the myosin VIIa tail is coexpressed with MyRIP lacking the Rab-binding domain, MyRIP–ΔRabBD, the two proteins perfectly colocalized in PC12 cells, but no enrichment was observed at the tips of the neurites. Therefore, MyRIP–RabBD likely mediates the targeting of MyRIP–myosin-VIIa complex to the tip of the neurites. Although we still do not know whether Rab27A is present at the tip of these neurites, in the retinal pigment epithelium cells, the localization of the three proteins at the apical cell microvilli strongly suggests that Rab27A recruits the MyRIP–myosin-VIIa complex via an interaction with MyRIP–RabBD. Finally, MyRIP is expressed in wide range of tissues, among which some lack Rab27A and/or myosin VIIa (i.e. muscle, heart and brain). The here-described dynamic module and its likely variant, made up myosin-V–melanophilin–Rab27A, led us to put forward the idea that unconventional myosin/myosin- and Rab-binding protein/Rab form a sort of basic recognition unit. Heterogeneity of this unit may underlie its contribution to different steps of vesicle trafficking. Since we propose Rab27A as a component of two of them, the possibility that also melanophilin and MyRIP belong to different modules is worth considering.
MyRIP and other expression constructs.
Yeast two-hybrid screening of a human retinal library was performed as described previously (Kussel-Andermann et al., 2000a). The cDNA of clone A7 was used to isolate the MyRIP full-length cDNA from a retinal cDNA library in λgt11 (Clontech). The mouse MyRIP cDNA (DDBJ/EMBL/GenBank accession No. AF396688) was obtained by 5′ RACE–PCR using an inner ear Marathon cDNA. Oligonucleotides designed from the rat EST226177 clone were used for amplification.
The MyRIP full-length coding sequence (aa 1–859) was PCR-amplified from human retina Marathon cDNA and subcloned in the pCMV–tag3B vector (Clontech). The following myc-tagged MyRIP truncated constructs were generated: pCMV3B–MyRIP–RabBD (aa 1–134); pCMV3B–MyRIP–ΔRabBD (aa 138–859); pCMV3B–MyRIP–myBD (aa 138–568); and pCMV3B–MyRIP–Ct (aa 561–859). The mutant form of the MyRIP–RabBD was generated by PCR amplification. The GFP–myosin-VIIa tail construct has been described previously (Kussel-Andermann et al., 2000b).
Three different anti-MyRIP polyclonal antibodies were generated in the rabbit: anti-hA7 was raised against a His6-tagged A7 fusion protein derived from the human MyRIP sequence (aa 186–383); and anti-mA7 was raised against a His6-tagged A7 fusion protein derived from the murine MyRIP sequence (aa 110–305); anti-hA7P was raised against a peptide, hA7P: EAPSRQPRDQGQHPRC, derived from the human MyRIP sequence (aa 310–324). The specificity of the three immunopurified antibodies (anti-hA7, anti-hA7P and anti-mA7) was checked by immunoblot and immunofluorescence analyses.
For immunoprecipitation experiments, the myosin VIIa tail (aa 847–2215) was cloned in the pcDNA3 vector (Invitrogen). For the production of GST–MyRIP fusion proteins, the following constructs were subcloned in the pGEX-4T1 vector (Pharmacia): pGEX-RabBD (aa 1–134) and pGEX-MyRIP–ΔCt (aa 1–568). For the production of biotinylated MyRIP, a fragment of MyRIP–myBD corresponding to aa 138–403 was subcloned into the pXa3 vector (Promega). The in vitro binding and pull-down experiments were performed as described previously (Kussel-Andermann et al., 2000a).
Immunohistofluorescence and electron microscopy analysis.
PC12 cells and mouse tissues were fixed with 4% paraformaldehyde and processed for indirect immunofluorescence microscopy as described previously (El-Amraoui et al., 1996). Immunoelectron microscopy was performed on LR White embedded ultra-thin sections of mouse retina as described previously (Wolfrum and Schmitt, 2000). The monoclonal mouse anti-Rab27A was obtained from Transduction Laboratories.
Filter binding assay.
Purified Rab proteins (40–120 nM) were loaded with [3H]GTP (7.7 Ci/mmol, Amersham) and incubated with 150 pmoles of GST or GST-fusion proteins immobilized on glutathione–Sepharose in 20 mM Tris–HCl pH 8.0, 150 mM NaCl, 1 mM MgCl2, 1 mM dithiotreitol and 50 μM Zn acetate. The mixture (30–50 μl) was kept on ice for 15 min. The reaction was terminated by the addition of 3 ml washing buffer (20 mM Tris–HCl pH 8.0, 100 mM NaCl, 1 mM MgCl2) and filtration through glass-fibre filters (Whatman GF/C) preincubated in washing buffer supplemented with 0.5% bovine serum albumin. After three washes, the filters were processed for scintillation counting.
We thank J.-P. Hardelin and S. Safieddine for critical reading of the manuscript, J. Levilliers for her continuous help and R. Schäfer for technical assistance. We thank G. Baldini, M. Cormont, J. Salaméro, M. Zerial, B. Goud, H. Shimomura and G. de Saint Basile for their kind gift of vectors encoding Rab3D, Rab4, Rab5, Rab6, Rab11, Rab26 and Rab27A, respectively. This work was supported by grants from EC (QLG2-CT-1999-00988), R. and G. Strittmatter, Retina-France, A. and M. Suchert, Forschung contra Blindheit-Initiative Usher Syndrome, Deutsche Forschungsgemeinschaft (Wo548/3 and /4) and FAUN-Stiftung (to U.W.).