The cargo adaptor proteins RILPL2 and melanophilin co-regulate myosin-5a motor activity

Vertebrate myosin-5a is an ATP-utilizing processive motor associated with the actin network and responsible for the transport and localization of several vesicle cargoes. To transport cargo efficiently and prevent futile ATP hydrolysis, myosin-5a motor function must be tightly regulated. The globular tail domain (GTD) of myosin-5a not only functions as the inhibitory domain but also serves as the binding site for a number of cargo adaptor proteins, including melanophilin (Mlph) and Rab-interacting lysosomal protein–like 2 (RILPL2). In this study, using various biochemical approaches, including ATPase, single-molecule motility, GST pulldown assays, and analytical ultracentrifugation, we demonstrate that the binding of both Mlph and RILPL2 to the GTD of myosin-5a is required for the activation of myosin-5a motor function under physiological ionic conditions. We also found that this activation is regulated by the small GTPase Rab36, a binding partner of RILPL2. In summary, our results indicate that RILPL2 is required for Mlph-mediated activation of Myo5a motor activity under physiological conditions and that Rab36 promotes this activation. We propose that Rab36 stimulates RILPL2 to interact with the myosin-5a GTD; this interaction then induces exposure of the Mlph-binding site in the GTD, enabling Mlph to interact with the GTD and activate myosin-5a motor activity.

Vertebrate myosin-5a (Myo5a) is a processive motor that converts chemical ATP energy into mechanical power and moves continuously on the actin network (1). Myo5a can be structurally divided into four distinct regions: the N-terminal motor domain, six IQ motifs serving as the binding site for the light chains, the proximal tail comprising a series of coiled coils, and the C-terminal globular tail domain (GTD) 3 (2)(3)(4)(5). The motor domain is the active domain containing the ATP-bind-ing site and the actin-binding site. The tail portion, including the proximal tail and the GTD, serves as the binding site for the adaptor proteins. In addition, the GTD also functions as the inhibitory domain.
A tail inhibition model for the regulation of Myo5a motor function is widely accepted (2)(3)(4)(5). In this model, Myo5a motor activity is activated by Ca 2ϩ or cargo binding via adaptor proteins, and this activation is accompanied by a conformational transition of Myo5a from the folded conformation (ϳ14 S) of the off state to the extended conformation (ϳ11 S) of the on state (6 -9). In the folded conformation, the GTD interacts with the motor domain and inhibits its motor function, whereas in the extended conformation, the motor domain is dissociated from the GTD and is fully active.
The motor function of Myo5a can be activated by Mlph. Mlph stimulates the ATPase activity of Myo5a (18) and increases the number of processively moving Myo5a molecules (19). We demonstrated that Mlph-GTBM is sufficient for activation of the ATPase activity of Myo5a by disrupting the head-GTD interaction (9). However, all abovementioned activations of Myo5a by Mlph are observed only in the presence of elevated concentrations of salt (0.15-0.3 M KCl), suggesting that other unknown cellular factors might be required for the activation of Myo5a by Mlph (5). One candidate is RILPL2 (Rab-interacting lysosomal protein-like 2), a Rab-interacting lysosomal protein (RILP) family member. Mammalian cells express three RILP family proteins (RILP, RILPL1, and RILPL2), each of which contains two conserved regions, RH1 and RH2 (RILP homology region 1 and 2, respectively) (20). RILPL2 was first identified as a binding partner of Myo5a in the brain (21). Structural and biochemical studies show that the RH1 of RILPL2 can form a ternary complex with Myo5a-GTD and Mlph-GTBM (22), sug-gesting that RILPL2 might cooperate with Mlph in regulating the motor function of Myo5a.
In this work, we found that, similar to Mlph, RILPL2 is expressed in mouse melanocytes and demonstrated that both Mlph and RILPL2 are required for the activation of Myo5a motor function under physiological ionic strength. Moreover, the RILPL2/Mlph-GTBM activation of Myo5a motor function is stimulated by Rab36, a binding protein of RILPL2. These results indicate that RILPL2 and Rab36 play a key role in regulating Myo5a motor function.

RILPL2-RH1 substantially increased the interaction between Myo5a and Mlph-GTBM
We first examined the expression of three RILP family proteins (RILP, RILPL1, and RILPL2) in melanocytes. The cDNAs of the three RILP family proteins were amplified by RT-PCR from mouse melanocytes, melan-a cells. DNA sequencing showed that these cDNAs are consistent with the reported sequences, indicating that all three RILP family proteins are expressed in melan-a cells. We then cloned the cDNAs of these three RILP family proteins into bacterial expression vectors. All three RILP family proteins, containing an N-terminal GST tag or His tag, were expressed in Escherichia coli and purified by GSH-Sepharose or Ni-agarose chromatography. To test the interactions between RILP family proteins with Myo5a, we performed a GST pulldown assay. As shown in Fig. 1A, RILPL2, but not RILP or RILPL1, was specifically pulled down with GSTtagged Myo5a-GTD. These results are consistent with the previous finding that Myo5a-GTD specifically interacts with RILPL2-RH1 but not with RILP-RH1 (22).

Myo5a motor function is co-activated by RILPL2 and Mlph-GTBM
We hypothesized that RILPL2-RH1 and Mlph-GTBM might co-regulate the motor function of Myo5a. To test this possibility, we examined Mlph-GTBM activation of the actin-activated ATPase activity (hereafter referred to as ATPase activity) of Myo5a in the absence or presence RILPL2-RH1 under physiological ionic strength conditions, i.e. in the presence of 100 mM KCl. As shown in Fig  To further analyze the activation of Myo5a by RILPL2-RH1 and Mlph-GTBM, we examined the motility activity of Myo5a at the single-molecule level. Cy3B-labeled Myo5a was first mixed with or without 20 M Mlph-GTBM and/or 5 M RILPL2-RH1 and then introduced into a flow chamber decorated with Alexa 488 -labeled actin filaments. Images of Cy3B and Alexa 488 fluorescence were recorded under a total internal reflection fluorescence microscope ( Fig. 3A and Movies S1-S4). In the absence of Mlph-GTBM and RILPL2-RH1, very few fluorescent spots of Cy3B-Myo5a colocalized and moved along actin filaments, and addition of Mlph-GTBM and RILPL2-RH1 caused a 13.1 Ϯ 0.7-fold increase in the number of Myo5a molecules moving per time and length compared with Myo5a alone (Fig. 3B). On the other hand, addition of either Mlph-GTBM or RILPL2-RH1 alone did not greatly increase the number of Cy3B-Myo5a molecules moving along actin filaments. These results indicate that both Mlph-GTBM and RILPL2-RH1 are required for activation of Myo5a motor activity.
It is well established that activation of Myo5a ATPase activity by Ca 2ϩ or adaptor proteins is accompanied by a conformational transition of Myo5a from the ϳ14 S folded conformation in the off state to the ϳ11 S extended conformation in the on state (6 -9). We expected that RILPL2-RH1/Mlph-GTBM also induced the conformational transition of Myo5a. Therefore, we measured the S value of Myo5a in the presence of RILPL2-RH1 and/or Mlph-GTBM under physiological ionic strength (100 mM KCl) and in the absence of Ca 2ϩ (in the presence of 1 mM EGTA). As shown in Fig. 4, the S value of Myo5a was 14.4 S and decreased to 12.2 S by 5 M Mlph-GTBM and 5 M RILPL2-RH1. On the other hand, 10 M RILPL2-RH1 alone slightly increased the S value of Myo5a, and 10 M Mlph-GTBM alone had little effect on it. These results indicate that, under physiological ionic strength and in the absence of Ca 2ϩ , both Mlph-GTBM and RILPL2-RH1 are required for inducing Myo5a to form the extended conformation. Moreover, Myo5a might be more compact in the presence of RILPL2-RH1 than in its absence.
We then examined the effects of Rab36 on the interaction between RILPL2 and Myo5a-GTD using a competition pulldown assay. We found that 15 M Rab36-GTP␥S significantly increased the amount of RILPL2 pulled down but decreased that of RILPL2-RH1 (Fig. 5C, compare lane 9 and lane 10). With Rab36-GTP␥S increasing from 0 to 15 M, the molar ratio of RILPL2 versus RILPL2-RH1 gradually increased from 0.28 Ϯ 0.11 to 0.90 Ϯ 0.18 (Fig. 5C, right panel). These results indicate that Rab36 regulates the interaction between RILPL2 and Myo5a-GTD.

Discussion
It is well established that the inhibited Myo5a is in a folded conformation, in which the GTD folds back to interact with the head, and the activated Myo5a is in an extended conformation, in which the head-GTD interaction is disrupted by Ca 2ϩ or adaptor proteins (6 -8, 24, 25). Mlph is one of the best-characterized adaptor proteins that are able to activate Myo5a (18,19). However, under physiological ionic strength, Myo5a motor function was only weakly stimulated by Mlph or its truncation Mlph-GTBM (9,18). This raises the possibility that other unidentified cellular factors might be required for the activation of Myo5a by Mlph (5). In this work, we demonstrated that RILPL2 is required for Mlph-mediated activation of Myo5a motor function under physiological ionic strength.
Structural and biochemical studies indicate that Myo5a-GTD forms a homodimer (Fig. 6A, top) in which the N-terminal extension of the GTD interacts with the Mlph-GTBM binding site in the counterpart GTD (26,27). In other words, the Mlph-GTBM binding site is buried in the GTD-GTD interface. Thus, weakening the GTD-GTD interaction is expected to enhance

Co-regulation of myosin-5a
the activation of Myo5a by Mlph. Consistently, the activation of Myo5a by Mlph-GTBM is enhanced by the alanine mutations of the two conserved basic residues (Arg-1490 and Lys-1791), both of which participate in the GTD-GTD interaction (27). These findings lead us to propose that, in addition to the inhibited state and the activated state of Myo5a, there must be a third state, the preactivated state, which is similar to the inhibited state except that the Mlph-GTBM-binding site in the GTD dimer is exposed (27). We have proposed previously that the equilibrium of the inhibited state and the preactivated state of Myo5a is regulated by other unidentified cellular factors (5).

Co-regulation of myosin-5a
Based on the findings above and the three-state model of the regulation of Myo5a (27), we propose the following model for activation of Myo5a by Rab36, RILPL2, and Mlph (Fig. 6B). First, Rab36 binds to RILPL2-RH2, and this binding relieves the RILPL2-RH2 inhibition on RILPL2-RH1. RILPL2-RH1 then binds to Myo5a-GTD, exposing the Mlph-GTBM-binding site and transforming Myo5a from the inhibited state to the preactivated state. Finally, Mlph binds to Myo5a-GTD and alloster-ically inhibits the head-GTD interaction, transforming Myo5a from the preactivated state to the activated state.
A comparison of the crystal structures of the Myo5b-GTD dimer and the Myo5a-GTD/Mlph-GTBM/RILPL2-RH1 complex shows that the distance between the two putative motor domain-binding sites in the GTD dimer is decreased upon its binding to RILPL2-RH1 (Fig. 6A). This is consistent with the RILPL2-RH1-induced increase in Myo5a's S value (Fig. 4). In the inhibited state, the two Myo5a-GTDs form a dimer and bind to the two heads of Myo5a, forming a triangular folded conformation. RILPL2 contains two domains, RH1 and RH2, which interact with Myo5a-GTD and the GTP-bound Rab36, respectively. In the absence of Rab36, RH2 prevents RH1 from binding to Myo5a-GTD. Binding of GTP-bound Rab36 to RH2 relieves RH2's inhibition on RH1. RH1 then binds to the GTD dimer of Myo5a in the inhibited state and induces the GTD to expose the Mlph-GTBM-binding site, forming the preactivated state. Finally, Mlph-GTBM binds to the Mlph-GTBM-binding site of Myo5a-GTD in the preactivated state and induces Myo5a to transform to the activated state in the extended conformation.

Co-regulation of myosin-5a
Therefore, we expect that Myo5a in the preactivated state is more compact than in the inhibited state. The tail inhibition model for the regulation of myosin-5 by Ca 2ϩ and adaptor proteins is well accepted, and many adaptor proteins of myosin-5 have been identified. However, so far, only two adaptor proteins of myosin-5, Mlph (9,18,19) and Rab11 (28,29), have been shown to be able to activate myosin-5 motor function. Moreover, those activations are only observed under an elevated ionic strength. The current finding that both RILPL2 and Mlph are essential for activation of Myo5a motor function suggests that activation of class V myosin might require multiple adaptor proteins. Recently, it has been shown that Myo5a-GTD and two Myo5a adaptor proteins, Spir-2 and Rab11, might form a ternary complex (30). It is likely that Myo5a motor function is synergistically activated by Spir-2 and Rab11. Future experiments should clarify this issue.
Myo5a is essential for normal accumulation of melanosomes in dendritic tips in melanocytes, and its mutation causes perinuclear accumulation of melanosomes. Both Mlph and its interacting GTPase Rab27 are required for docking of Myo5a onto melanosomes and, lacking of either of them, a similar perinuclear accumulation of melanosomes (10 -14, 31). Our finding suggests that RILPL2 and its interacting GTPase Rab36 cooperate with Mlph and Rab27 for docking of Myo5a onto melanosomes. It is possible that lack of either RILPL2 or Rab36 might also cause similar perinuclear accumulation of melanosomes in melanocytes.
RILP has been reported to mediate the association of dynein with late endosomes (32)(33). Because both RILP and RILPL2 interact with Rab36 (23, 34), one possible scenario in melanocytes is that RILP-dynein and RILPL2-Myo5a compete with each other in associating with the melanosome-bound Rab36, regulating transport of melanosomes along microtubules and actin filaments. The roles of Rab36 and its effectors RILP and RILPL2 in the distribution and maturation of melanosomes in melanocytes deserve further investigation.
The recombinant proteins were expressed in BL21(DE3) E. coli as His-tagged proteins (in the pET30a vector) or GSTtagged proteins (in the pGEX-4T2 vector) and purified by Niagarose affinity chromatography or GSH-Sepharose affinity chromatography using standard procedures. FLAG-tagged Myo5a (full-length mouse melanocyte-type Myo5a) and GSTtagged Myo5a-GTD (comprising the C-terminal 410 residues of Myo5a) were prepared as described previously (36). Mlph-GTBM peptide (residues 176 -201 of Mlph) and GST-tagged Mlph-GTBM were prepared as described previously (9). All protein concentrations in this study refer to the mole of polypeptide regardless of their oligomerization states.

ATPase assay
The ATPase activity of Myo5a was measured as described previously at 25°C in an ATP regeneration system (36).

Single-molecule motility assay
The single-molecule motility assay was performed at room temperature (ϳ20°C) in a 10-l flow chamber in the absence of calcium as follows. Flow cells were incubated with 0.2 mg/ml N-ethylmaleimide-myosin in high-salt buffer (10 mM Hepes-KOH (pH 7.5), 500 mM KCl, and 5 mM MgCl 2 ) for 5 min, blocked with high-salt buffer plus 1 mg/ml casein for 5 min, and incubated with 0.04 M Alexa 488 -phalloidin-labeled actin filaments in low-salt buffer (20 mM Hepes-KOH (pH 7.6), 25 mM KCl, 2 mM MgCl 2 , and 1 mM EGTA) for 3 min. After being rinsed with low-salt buffer plus 1 mg/ml casein, the flow cham-

Co-regulation of myosin-5a
ber was infused with the assay solution (ϳ1 nM Cy3B-Myo5a, 12 M CaM, 2000 units/ml catalase, 40 units/ml glucose oxidase, 6 mg/ml glucose, 10 M ATP, 3.3 mM PEP, and 67 units/ml pyruvate kinase with or without 20 M Mlph-GTBM and/or 5 M RILPL2-RH1 in low-salt buffer). Images (512 ϫ 512 pixels) were recorded at a frame rate of 5 s Ϫ1 using an inverted Nikon Ti-E total internal reflection fluorescence microscope as described previously (37). Images were merged and analyzed using the Matlab and ImageJ programs. The Cy3B fluorescent spots that moved along Alexa 488 -phalloidin-labeled actin filaments were counted and normalized to the total length of actin filaments.
N-ethylmaleimide-myosin and Alexa 488 -phalloidinlabeled actin filaments were prepared as described previously (37). CaM-A73C, a human CaM mutant created by site-directed mutagenesis, was expressed in E. coli and purified as described previously (38). CaM-A73C was labeled by Cy3B maleimide (GE Healthcare) as described previously (37). Cy3B-Myo5a was prepared by exchanging endogenous CaM with a 1:6 mixture of Cy3B-labeled CaM-A73C and WT CaM as described previously (39).

GST pulldown assay
GST pulldown assays were performed as described previously (27). All pulldown assays were repeated at least three times, and typical images are presented. In the GST-Mlph-GTBM/Myo5a pulldown assay, GSH-Sepharose beads ( GST-RILPL2 pulldown with Rab36-GTP␥S or -GDP was performed as follows. Rab36 was preloaded with nucleotides by incubating 2 M His-Rab36 with 20 M GTP␥S or GDP in 1ϫ PBS (12 mM sodium phosphate (pH7.4), 137 mM NaCl, and 27 mM KCl) containing 1 mg/ml BSA, 1 mM DTT, and 5 mM EDTA at 37°C for 30 min before adding 10 mM MgCl 2 to stop the reaction. GST or GST-RILPL2 (20 M, 20 l) was bound onto GSH-Sepharose beads (10 l) and then incubated with 200 l of 2 M His-Rab36-GTP␥S or His-Rab36-GDP at 4°C for 1 h. The GSH-Sepharose beads were washed with 200 l of 1ϫ PBS containing 1 mM DTT and 5 mM MgCl 2 five times and then eluted by 10 mM GSH, 50 mM Tris-HCl (pH8.0), 1 mM DTT, and 0.2 M NaCl. The eluted proteins were analyzed by SDS-PAGE (4%-20%) and visualized with CBB staining.

Analytical ultracentrifugation
The conformation of Myo5a was monitored by sedimentation velocity as described previously (9) with minor modifications (see below). About 2 M Myo5a was first dialyzed against 20 mM MOPS-KOH (pH 7.0), 100 mM KCl, 1 mM DTT, and 1 mM EGTA at 4°C overnight. Prior to running, 1 mM MgCl 2 , 5 M His-RILPL2-RH1, and 5 M Mlph-GTBM peptide were added to the Myo5a sample. In control samples, 5 M His-RILPL2-RH1 and 5 M Mlph-GTBM peptide were replaced with H 2 O, 10 M His-RILPL2-RH1, or 10 M Mlph-GTBM peptide. The sedimentation velocity was run at 40,000 rpm (20°C) for 2 h in a Beckman Optima XL-I analytical ultracentrifuge. The sedimentation data were fitted to a continuous size distribution model in SEDFIT with a confidence level (F-ratio) of 0.68, where the frictional ratio (f/f 0 ) was allowed to float. The solvent viscosity and density were computed using SEDNTERP (40). The c(s 20,w ) distributions were plotted using OriginPro 9.1.