Myosin VI is an actin-based retrograde motor protein that plays a crucial role in both endocytic and secretory membrane trafficking pathways. Myosin VI's targeting to and function in these intracellular pathways is mediated by a number of specific binding partners. In this paper we have identified a new myosin-VI-binding partner, lemur tyrosine kinase 2 (LMTK2), which is the first transmembrane protein and kinase that directly binds to myosin VI. LMTK2 binds to the WWY site in the C-terminal myosin VI tail, the same site as the endocytic adaptor protein Dab2. When either myosin VI or LMTK2 is depleted by siRNAs, the transferrin receptor (TfR) is trapped in swollen endosomes and tubule formation in the endocytic recycling pathway is dramatically reduced, showing that both proteins are required for the transport of cargo, such as the TfR, from early endosomes to the endocytic recycling compartment.
Endocytosis is essential for the uptake of vesicles containing receptor-bound nutrients and signalling receptors from the plasma membrane. After internalisation these vesicles are transported from the plasma membrane into the cell and fuse with the early endosome where cargo, such as signalling receptors, are sorted into multivesicular endosomes for lysosomal degradation. Membrane pumps, channels and nutrient receptors, such as the transferrin receptor (TfR), are sorted into a recycling pathway back to the plasma membrane, either via a rapid loop from an early endosome or via a longer indirect route through the perinuclear endocytic recycling compartment (ERC) (Maxfield and McGraw, 2004). Although the exact molecular mechanisms involved in the delivery of receptors to the recycling compartment and the transport back to the cell surface remain to be elucidated, the small GTPase Rab11 and the Eps15 homology domain (EHD) family of proteins have been implicated in controlling this membrane trafficking pathway (Ullrich et al., 1996; Sonnichsen et al., 2000; Naslavsky and Caplan, 2005; Naslavsky et al., 2006). The actin-based motor protein myosin Vb has also been reported to associate with Rab11 and to regulate transport of receptors out of the perinuclear ERC back to the cell surface (Hales et al., 2002; Lapierre et al., 2001; Roland et al., 2007).
We show here that another class of myosins is required for the delivery of cargo from the early endosome into the ERC. Class VI myosins play crucial role(s) in membrane trafficking pathways, because they are the only class that so far has been shown to move `backwards' towards the minus end of actin filaments – in the opposite direction to all other myosins so far characterised (Wells et al., 1999). Myosin VI is associated with secretory and endocytic membrane compartments (Buss et al., 2001; Buss et al., 1998; Warner et al., 2003; Aschenbrenner et al., 2003) and several binding partners, responsible for differential intracellular targeting and/or recruitment of myosin VI, have now been identified. In the early endocytic pathway, recruitment of myosin VI to clathrin-coated structures at the plasma membrane requires Dab2 (disabled-2) (Morris et al., 2002; Spudich et al., 2007) and to uncoated endocytic vesicles requires GIPC (GAIP-interacting protein C-terminus) (Bunn et al., 1999; Naccache et al., 2006). Myosin VI function in exocytic membrane trafficking pathways needs the Rab8 effector protein optineurin, which links myosin VI to the Golgi complex (Sahlender et al., 2005). In polarised epithelial cells myosin VI is required for the transport of newly synthesised membrane proteins containing a tyrosine-sorting motif to the basolateral domain. These membrane proteins are sorted on route to the basolateral plasma membrane in the endocytic recycling endosome (Ang et al., 2004), where myosin VI and optineurin, together with Rab8 and TfR have been localised (Au et al., 2007).
In this study we have identified lemur tyrosine kinase 2 (LMTK2, also known as cprk, KPI-2, BREK, Lmr2, AATYK2 and KIAA1079) as a myosin-VI-binding partner and have investigated the role(s) of the LMTK2–myosin-VI complex in endocytic and exocytic membrane trafficking pathways. LMTK2 is a member of the lemur kinase group and is a Ser/Thr-specific protein kinase with two predicted transmembrane domains at its N-terminus, followed by a kinase domain and a very long C-terminal tail domain (Wang and Brautigan, 2002; Kesavapany et al., 2003; Kawa et al., 2004). LMTK2 was previously identified as a binding partner of the p35-activator subunit for cyclin-dependent kinase 5 (Kesavapany et al., 2003), and of protein phosphatase 1C and its small inhibitor protein 2 (Wang and Brautigan, 2002). Several potential LMTK2 substrates have been identified: protein phosphatase 1C (Wang and Brautigan, 2002), phosphorylase b and the cystic fibrosis transmembrane conductance regulator (CFTR) (Wang and Brautigan, 2006). Mutations in the CFTR gene are linked to the genetic disorder cystic fibrosis (Rommens et al., 1989; Riordan et al., 1989). Interestingly, myosin VI is required for endocytosis of CFTR from the apical domain of polarised epithelial cells (Swiatecka-Urban et al., 2004; Ameen and Apodaca, 2007) and here we show that myosin VI binds directly to LMTK2, which can potentially phosphorylate CFTR. LMTK2 is the first transmembrane protein and kinase that binds directly to myosin VI. Our functional studies demonstrate that both proteins, myosin VI and LMTK2, are required for delivery of cargo, such as TfR, from the early endosome to the ERC.
LMTK2 is a novel myosin-VI-binding partner
Myosin VI was identified in a yeast two-hybrid screen as a LMTK2-binding partner by using the LMTK2 cytoplasmic domain (residue 67 to the end of the C-terminus) as a bait to probe a human brain cDNA library. About half of positive clones identified in this screen encoded the C-terminal tail of myosin VI. To confirm and further investigate the interaction between myosin VI and LMTK2, we used a mammalian two-hybrid assay. Using this system, we have previously mapped the binding sites for three other myosin-VI-binding partners at two independent sites in the C-terminal domain of the myosin VI tail: Dab2 binding requires a WWY motif, whereas the GIPC- and optineurin-binding site contains a RRL motif (Sahlender et al., 2005; Spudich et al., 2007). To determine the region on myosin VI involved in LMTK2 binding, we tested a range of point mutants and established that binding of LMTK2 to the myosin VI tail requires the WWY motif, the same site used by Dab2 (Fig. 1B,C). Mutations in the RRL motif in the myosin VI tail (the binding site for optineurin) did not affect its binding to LMTK2 (data not shown). By testing multiple LMTK2 deletion mutants we identified a minimal fragment of ∼200 amino acids (between aa 567 and 773) downstream of the kinase domain, which mediates interaction between the myosin VI tail and LMTK2 (Fig. 1A). Interestingly, myosin VI binds in a region that is different to the one identified for protein phosphatase 1 (PP1) and its small inhibitor protein 2 (Inh2; aa 1344-1450) (Wang and Brautigan, 2002) but in the same region as p35, the activator subunit of CDK5 (aa 391-632) (Kesavapani et al., 2003).
Myosin VI exists as four splice isoforms due to a large and a small insert in the tail region (Fig. 1B), and these isoforms have been shown to have distinct roles in membrane trafficking pathways (Buss et al., 2001; Au et al., 2007). So we tested the binding of LMTK2 to these different splice isoforms of myosin VI. LMTK2 binds to the myosin VI tail regardless of whether it contains the large (31 aa) insert or not; however, it shows slightly stronger binding when this insert is present (Fig. 1D). This result suggests that LMTK2, which is a ubiquitously expressed kinase, is able to interact with the different myosin VI isoforms that are expressed in a variety of different tissues and cells (Buss et al., 2001).
To confirm that myosin VI directly binds to LMTK2 we used the glutathione-S-transferase (GST) pull-down assay. An in-vitro-translated [35S]methionine-labelled LMTK2 fragment (aa 451-1095) was incubated with either GST alone, wild-type myosin VI tail tagged to GST or the mutant myosin VI tail (WWY→WLY) tagged to GST (Fig. 2A). Whereas LMTK2 strongly binds to the wild-type myosin VI tail, no binding to GST alone and only very weak binding to the mutant myosin VI tail (WWY→WLY) was observed. These results support the observation that binding of LMTK2 and Dab2 to the myosin VI tail involves the WWY motif, and that myosin VI binds directly to LMTK2.
Finally, we confirmed the interaction between myosin VI and LMTK2 in vivo by co-immunoprecipitation using affinity-purified polyclonal antibodies against either myosin VI tail or the cytoplasmic tail of LMTK2. As shown in Fig. 2B, endogenous myosin VI co-immunoprecipitates with a small amount of LMTK2; however, overexpression of GFP-tagged myosin VI leads to a substantial increase in the amount of LMTK2 found in a complex with myosin VI. In the reciprocal experiment, antibodies against LMTK2 can also immunoprecipitate myosin VI from HeLa cells that express GFP–myosin-VI (Fig. 2C). These results demonstrate that myosin VI and LMTK2 exist together in a protein complex within the cell.
Intracellular localisation of LMTK2 and myosin VI
To investigate whether myosin VI and LMTK2 localise to the same intracellular compartment, GFP-tagged myosin VI and untagged LMTK2 were coexpressed in HeLa cells; cells were then labeled with a polyclonal antibody against LMTK2 and a monoclonal antibody against GFP. LMTK2 was present at the plasma membrane, and shows a vesicular staining pattern throughout the cell, but is concentrated in cell extensions and in the perinuclear area (Figs 3, 4). GFP-tagged myosin VI was found to colocalise with LMTK2 in small punctate structures throughout the cell (Fig. 3A). Myosin VI and LMTK2 only partially colocalise, probably because LMTK2 is not the only myosin-VI-binding partner present in the cell and both proteins might have overlapping but distinct intracellular functions. To see whether endogenous myosin VI also colocalises with LMTK2, LMTK2-transfected cells were stained with antibodies against myosin VI. Endogenous myosin VI colocalised with LMTK2 on small vesicles (Fig. 3B). In the reverse experiment, however, we were unable to detect endogenous LMTK2, which probably reflects the low expression level of this kinase.
Are the vesicles containing myosin VI and LMTK2 involved in the endocytic or the exocytic membrane trafficking routes? Previously, we demonstrated that myosin VI is present at and around the Golgi complex, and plays a role in constitutive secretion (Warner et al., 2003; Sahlender et al., 2005). LMTK2 has also been reported to be concentrated in the perinuclear region of the cell at the Golgi complex (Kesavapany et al., 2003). However, we found very little overlap between LMTK2 and GM130, a marker protein of the Golgi matrix (supplementary material Fig. S1A). In RNA interference (RNAi) experiments, cells depleted of myosin VI showed (as expected) a dramatic reduction in constitutive secretion of the reporter molecule, secreted alkaline phosphatase (SEAP), however, cells depleted of LMTK2 displayed no reduction in the level of SEAP secretion when compared to control cells (supplementary material Fig. S1B). Similarly, the transport of tsVSV-G from the Golgi complex to the cell surface was not affected in LMTK2 knockdown (KD) cells (supplementary material Fig. S1C). These results indicate that LMTK2 does not play a major role in the secretory pathway.
To test the involvement of LMTK2 in the endocytic pathway, GFP-tagged LMTK2 was transfected into HeLa cells, which were pulsed with fluorescently labeled transferrin for 20 minutes to allow uptake of transferrin into peripheral endocytic structures and the ERC. Images in Fig. 4a-c demonstrate that LMTK2 is present on transferrin-positive endocytic structures. Double labelling experiments show that LMTK2 is present on a subset of early endosomes, as revealed by colocalisation with Rab5 and EEA1 (Fig. 4d-j), and also associated with the Rab11-positive ERC (Fig. 4k-m). In agreement with previous studies using unpolarised cells, we observed myosin VI concentrating on uncoated vesicles, where it colocalises with GIPC and the early endosomal marker Rab5, but hardly any myosin IV colocalises with EEA1 (supplementary material Fig. S2). These results show that LMTK2 can be found along the early endocytic and recycling pathway, where it colocalises with Rab5, EEA1 and Rab11, whereas myosin VI displays a more restricted localization, being concentrated in the early endocytic pathway on Rab5-positive early endosomes.
Loss of myosin VI and LMTK2 leads to enlarged endosomes
To address the function of LMTK2 and myosin VI in the endocytic and recycling pathway, we reduced the cellular expression levels of these proteins using small interfering RNA (siRNA). HeLa cells were transfected twice with either control oligonucleotide, or siRNAs targeting LMTK2 or myosin VI. Immunoblot analysis showed that 48 hours after the second transfection the expression levels of LMTK2 and myosin VI were down to less than 5% as compared to control cells (Fig. 5A). To study the effects of LMTK2 or myosin VI depletion, we processed the KD and control cells for immunofluorescence microscopy to visualise the steady-state localisation of endocytic organelles. The fixed cells were labeled with antibodies against the early endosomal marker EEA1 and against vps26, a component of the mammalian retromer complex, which is present on early and late endosomes and is important for retrieval of transmembrane proteins from endosomes back to the Golgi complex. In control cells early and late endosomes are represented by small vesicles dispersed throughout most of the cell, with a slightly higher concentration in the perinuclear region (Fig. 5B). However, in cells transfected with siRNAs specifically targeting myosin VI or LMTK2, EEA1 and vps26 are now present in large swollen, ring-like endosomes that often aggregate in the perinuclear region. This phenotype was most dramatic in myosin VI KD cells, where hardly any small endocytic structures were found in the cell periphery. LMTK2 KD cells displayed identical changes in endosome morphology; however, the redistribution into a tight spot in the perinuclear area was not always as dramatic as in myosin-VI-depleted cells.
To follow the uptake of transferrin into control and KD cells we pulsed the cells for 20 minutes with fluorescently labeled transferrin. In control cells transferrin was found in peripheral endocytic structures and was also present in the perinuclear ERC. In myosin VI KD and LMTK2 KD cells, however, the transferrin was trapped in swollen endocytic vesicles (Fig. 5C) that were positive for EEA1 (supplementary material Fig. S3) and vps26 (data not shown).
Myosin VI and LMTK2 are required for transport of transferrin from early endosomes to the ERC
After endocytosis and delivery to an early endocytic compartment the TfR is either recycled directly from the early endosome back to the plasma membrane or it is delivered to a morphologically distinct ERC (Fig. 9) (for review see Maxfield and McGraw, 2004). This compartment has been described as a perinuclear collection of tubular/vesicular membranes containing Rab11 (Hopkins and Trowbridge, 1983; Yamashiro et al., 1984). Since loss of myosin VI and LMTK2 expression causes the accumulation of TfR in swollen organelles that contain the early endosomal marker EEA1, we tested whether this receptor is still delivered to the Rab11-positive ERC in myosin VI and LMTK2 KD cells. A stable cell line expressing GFP-tagged Rab11 was either mock-transfected, or transfected with myosin VI or LMTK2 siRNA, and immunostained with antibodies against GFP and the transferrin receptor. As shown in Fig. 6 in mock-transfected cells the endogenous TfR is present in the cell periphery, but also in the perinuclear region and on tubules emanating from the centre towards the periphery of the cell. These control cells show a high degree of colocalisation between the TfR and Rab11, a marker of the ERC. In myosin VI and LMTK2 KD cells the TfR is trapped in swollen endosomes, consistant with the results from the fluorescently labelled transferrin uptake experiment, but now shows very little colocalisation with Rab11 and the ERC (Fig. 6A), indicating that the TfR does not reach the ERC.
In addition, we investigated the effect of myosin VI and LMTK2 depletion on the late endocytic pathway. We observed no accumulation of the lysosomal marker LAMP1 in transferrin-positive endosomes and also no defects in ligand-induced EGFR degradation in either of the KD cells (supplementary material Fig. S4). These results suggest that myosin VI and LMTK2 play a role in the transport of cargo from the early endosome to the perinuclear recycling compartment.
Loss of myosin VI and LMTK2 causes a delay in recycling of TfR to the plasma membrane
We used FACS analysis to measure intracellular trafficking of TfR in control and siRNA-transfected cells (see details in Materials and Methods and corresponding figure legends). To determine whether the loss of myosin VI or LMTK2 expression does not only inhibit delivery of the TfR to the ERC but also affects its recycling back to the plasma membrane, we pulsed mock- and siRNA-transfected cells with fluorescently labelled transferrin at 37°C, followed by a chase for different times in the presence of excess unlabelled transferrin to allow recycling. Comparing the remaining amount of transferrin in cells revealed small but consistent defects in TfR recycling to the cell surface in both myosin-VI- and LMTK2-depleted cells (Fig. 6B).
To determine the possible role of myosin VI and LMTK2 in the fast recycling pathway, we used two approaches. First, we loaded cells with Tf–Alexa-Fluor-647 at 16°C and allowed them to recycle at 37°C. Temperatures below 20°C have been shown to accumulate TfR in sorting endosomes and prevent its transport to perinuclear recycling compartments (Ren et al., 1998). Second, we pulsed cells with transferrin at 37°C for 5 minutes before allowing recycling at 37°C. During the short pulse, transferrin only reaches the early endocytic vecicles and not the perinuclear recycling compartment. Using either method we observed a decrease in fast recycling in myosin-VI-depleted cells and no decrease but even a slight increase in fast recycling in cells depleted of LMTK2 (supplementary material Fig. S5A,B).
We also observed a reduction in the total amount of transferrin internalised in myosin VI KD and LMTK2 KD cells (supplementary material Fig. S5C,D), which could reflect accumulation and trapping of TfR in the swollen endocytic compartment. However, the total amount of TfR measured by FACS (supplementary material Fig. S5E) and by western blot (data not shown) appeared to be reduced in both myosin-VI- and LMTK2-depleted cells, which could explain the observed reduction in total transferrin internalised.
Tubule formation in the endocytic recycling pathway is inhibited in the absence of myosin VI or LMTK2
In the cell line stably expressing GFP-Rab11 the TfR is present in an array of long tubules – probably transport intermediates – that contain Rab11 and emerge from the perinuclear region, and extent into the periphery of the cell. In myosin VI and LMTK2 KD cells we noticed a dramatic reduction of Rab11-positive tubules emanating from the perinuclear area as well as a collapse of Rab11 into a tighter perinuclear spot. Apart from this, the overall distribution of Rab11 in the cell periphery and in cell extensions at the plasma membrane was unchanged (Fig. 7A). To quantify this effect the number of cells containing Rab11-positive tubules was scored in mock-treated cells and compared with the number in myosin VI or LMTK2 KD cells. In control cells about half the cells (46%) had obvious Rab11 tubules, whereas only 13% of the myosin VI KD and 26% of the LMTK2 KD cells contained any Rab11-positive tubules (Fig. 7B). These results suggest that tubule formation for cargo recycling back to the plasma membrane is dramatically inhibited in myosin VI and LMTK2 KD cells.
To verify this observation we investigated the formation of EHD1- or EHD3-containing tubules in cells depleted of myosin VI or LMTK2. EHD1 and EHD3 belong to the Eps15 homology domain (EHD) family of proteins that are part of the Rab11-mediated endocytic recycling pathway (Naslavsky and Caplan, 2005). Whereas EHD1 appears to regulate transport of a number of different receptors from the ERC to the plasma membrane (Caplan et al., 2002; Lin et al., 2001; Naslavsky et al., 2004; Picciano et al., 2003), EHD3 plays a role in the delivery of cargo from the early endosome to the ERC (Naslavsky et al., 2006). Both proteins are found on vesicles and long membrane tubules associated with the ERC and the Rab11 recycling pathway. To determine whether loss of myosin VI or LMTK2 affects formation of EHD1- or EHD3-positive tubules, we used HeLa cell lines stably expressing either GFP-EHD1 or GFP-EHD3. In GFP-EHD1-expressing cells treated with myosin VI siRNA very few cells containing long EHD1 tubules were observed, instead EHD1 was present in large vesicular structures in the cell periphery. GFP-EHD1 cells transfected with LMTK2 siRNA also displayed a very dramatic reduction in the number of EHD1-positive tubules (Fig. 8A). Knocking down myosin VI and LMTK2 had a similar effect on EHD3-positive tubule formation in GFP-EHD3 cells (Fig. 8B). To quantify the reduction in the number of EHD3 tubules, the number of cells containing no tubules, few (<15) tubules or multiple (>15) tubules were counted (Fig. 8C). The majority of the mock-transfected cells contained multiple tubules, whereas more than 50% of the myosin VI KD cells contained no EHD3-positive tubules at all and the number of LMTK2 KD cells containing multiple tubules was dramatically reduced. We then looked at those cells that had tubules under all three experimental conditions (mock and both KDs), and established the number of tubules per cell and the length of individual tubules. As shown in Fig. 8D, in both myosin VI and LMTK2 KD cells there was a shift in the distribution towards having fewer tubules (dramatic in myosin VI KD and considerable in LMTK2 KD). Only minor differences of tubule length in KD cells compared with control cells were observed (Fig. 8E).
In conclusion our results indicate that LMTK2 is a novel binding partner of myosin VI. Both proteins are required for delivery of TfR to the ERC and are part of the membrane trafficking machinery in the endocytic recycling pathway (Fig. 9).
We identified LMTK2 as a novel myosin-VI-interacting protein, the first membrane protein that has been shown to directly bind to myosin VI. LMTK2 is a member of a family of Ser/Thr kinases with a N-terminal transmembrane domain and a long cytoplasmic tail containing the catalytic domain. The interaction between myosin VI and LMTK2 was first discovered in a yeast two-hybrid screen and confirmed in a variety of different protein-protein-interaction assays in vitro and in vivo. Our localisation studies show that LMTK2 is present on transferrin-, Rab5-, EEA1- and Rab11-positive structures along the endocytic and recycling pathway. Myosin VI is found in the early endocytic pathway close to the plasma membrane on Rab5-positive endosomes but, in contrast to LMTK2, very little myosin VI is present on EEA1- or Rab11-positive vesicles (supplementary material Fig. S2 and data not shown). The broader distribution of LMTK2 along the endocytic pathway and the restricted localisation of myosin VI in the early pathway might explain the limited colocalisation of these two proteins. In addition, both proteins were shown to have separate and probably independent functions in the cell (Sahlender et al., 2005; Kawa et al., 2004).
LMTK2 binds to the WWY motif in the C-terminal tail of myosin VI and, therefore, shares a binding site with Dab2. This binding site is distinct from the RRL motif, where optineurin and GIPC bind the C-terminal tail of myosin VI. Dab2 links myosin VI to the very first steps of the endocytic pathway, involving clathrin-mediated vesicle formation at the plasma membrane, whereas GIPC has been implicated in myosin-VI-driven transport of uncoated early endocytic vesicles through the cortical actin network in epithelial cells (Aschenbrenner et al., 2003). LMTK2 is thus the third myosin-VI-binding partner implicated in endocytic membrane trafficking.
Aschenbrenner et al. observed that ablating myosin VI activity by overexpressing of the dominant-negative myosin VI tail leads to accumulation of transferrin-positive vesicles in the cortical actin network (Aschenbrenner et al., 2003). In our experiments we also observed a slight delay in trafficking of transferrin-containing vesicles through the peripheral actin network in myosin VI KD cells (data not shown); however, the TfR still reached and accumulated in swollen early endosomes. The major defect we observed in myosin VI KD cells was therefore not the retention of uncoated endocytic vesicles in the cell periphery but a block in transport of TfR from the early endosomes to the ERC.
In the majority of the experiments there was a slightly more dramatic phenotype when myosin VI expression was suppressed than when LMTK2 expression was lost, although similar knockdown levels were achieved for both proteins (Fig. 5A). In myosin VI KD cells the loss of Rab11-, EHD1- and EHD3-positive tubules was more severe and transferrin recycling was inhibited to a greater extent. Since LMTK2 is a Ser/Thr kinase and only small amounts of it may be required to phosphorylate its target proteins and to fulfil its intracellular roles, it may be necessary to completely deplete LMTK2 in order to see a dramatic effect. However, overall we observed a very similar phenotype in myosin VI and LMTK2 KD cells, which strongly suggest that both proteins act in a functional complex in the endocytic pathway. Since LMTK2 is a transmembrane protein it might be involved in recruiting myosin VI to the surface of the endosome or, as a protein kinase, it could be involved in regulating myosin VI function by phosphorylation, or myosin VI might transport the kinase to the appropriate compartment so that it can phosphorylate its target proteins.
Several possible phosphorylation sites have been identified in myosin VI. Phosphorylation of two Thr residues (T1089 and T1092) in the C-terminal tail region (TINT) has been shown to regulate optineurin binding to the myosin VI tail (Sahlender et al., 2005). In addition, myosin VI contains a conserved Thr residue (T405) in the actin-binding interface of the motor domain and its phosphorylation plays a crucial role in regulating the activity of myosin I (Bement and Mooseker, 1995). Although phosphorylation at this site does not change the in vitro kinetic properties of the motor (Morris et al., 2003), it might modulate the myosin-VI–actin interaction, when phosphorylated in vivo (Buss et al., 1998; Naccache and Hasson, 2006). To investigate whether LMTK2 is able to phosphorylate myosin VI in vitro, we performed phosphorylation experiments using purified baculovirus-expressed myosin VI and LMTK2 immunoprecipitated from HeLa cells. So far, under the condition tested (see Materials and Methods), we have not been able to demonstrate that myosin VI is phosphorylated by LMTK2. However, the full-length LMTK2 tested may not be fully active after being released from membranes by using detergent and immunoprecipitated from HeLa cells. In addition LMTK2 activity has been shown to be regulated by nerve growth factor as well as by direct phosphorylation by CDK5 (Kawa et al., 2004; Kesavapany et al., 2003). The level of phosphorylation of GFP–myosin-VI immunoprecipitated from cells transfected either with wild-type LMTK2 or a kinase-dead LMTK2 was also tested, but so far we have been unable to detect significant differences in myosin VI phosphorylation (data not shown).
The major phenotype observed in myosin VI and LMTK2 KD cells is a block in TfR trafficking from early endosomes to the ERC. This defect is very similar to the phenotype observed in EHD3 KD cells, where delivery of internalised transferrin to the ERC is blocked and there is an accumulation of cargo in the EEA1-positive early endosome. In these cells, like in the myosin VI and LMTK2 KD cells, only a slight delay in transferrin recycling was measured (Naslavsky et al., 2006). Other studies have suggested that blocking TfR delivery to the ERC might redirect it into the direct fast recycling pathway from the early endosome back to the plasma membrane (Hirst et al., 2005) (see model in Fig. 9). We observed only very slight upregulation of fast recycling in LMTK2-depleted cells, which may partially compensate for the defect in trafficking through the ERC. However, in myosin VI KD cells both fast and slow recycling are slightly reduced, a phenotype similar to that observed in cells overexpressing dominant-negative Rab11S25N (Ullrich et al., 1996; Ren et al., 1998). These results indicate that myosin VI affects delivery into both fast and slow recycling pathways.
Our results suggest a role for myosin VI and LMTK2 at the early endosome for delivery of cargo to the ERC. Further support for this proposal is the observation that RNAi of myosin VI and LMTK2 has a dramatic effect on EHD3-tubule formation. However, whether loss of EHD3-tubule formation is the primary cause for reduced transport to the ERC needs to be established. The loss of EHD1 and Rab11 tubules emerging from the ERC in the myosin VI and LMTK KD cells suggests that these proteins either have a role at the recycling endosome or that this is a secondary effect due to the reduced transport of cargo into the ERC. However, we have recently shown that myosin VI functions in the recycling compartment in MDCK cells (Au et al., 2007). In these polarised epithelial cells myosin VI is present in recycling endosomes, where it is required for sorting of AP-1B-dependent cargo to the basolateral domain. Therefore, myosin VI might play a role in the recycling compartment not only in polarised and but also in nonpolarised cells.
Lmtk2-knockout mice are apparently normal at birth, but adult Lmtk2–/– males are infertile (Kawa et al., 2006). Closer investigation revealed that Lmtk2–/– mice have defects in spermatogenesis, particularly in the formation of the acrosome. Although the exact molecular mechanisms involved in acrosome formation are not known, it has been shown that membrane trafficking is required for acrosomal proteins to be sorted and delivered into the growing acrosome (Moreno et al., 2000; Ramalho-Santos et al., 2002). It would be interesting to look more closely at other specialised cell types in this knockout mouse and compare their phenotypes with those of the myosin VI (Snell's waltzer) knockout mouse.
Here, we have established that the myosin-VI–LMTK2 complex is a new player in the endocytic recycling pathway, because both proteins play a crucial role in trafficking of cargo from the early endosome to the ERC and in tubule formation from the ERC. Whether myosin VI and LMTK2 are involved in the formation of transport vesicles or tubules, whether they play a role in recruitment of proteins/cargo into transport carriers or whether myosin VI transports LMTK2 to its target proteins are some of the intriguing questions that remain to be answered.
Materials and Methods
Expression constructs and antibodies
The LMTK2/pCIneo (Cprk/pCIneo) construct expressing human LMTK2 amino acids 1-1443 has been previously described (Kesavapany et al., 2003) and the LMTK2 aa 1-1503 construct (pCMV-KPI-2) (Wang and Brautigan, 2002) was provided by D. Brautigan (University of Virginia). To create the LMTK2-GFP fusion protein, LMTK2 (aa 1-1503) was cloned into XhoI and BamHI sites of pEGFPN1 (Clontech). Human myosin VI constructs with (MyoVI-LI) and without (MyoVI-NI) the insert in the tail domain have been described previously (Sahlender et al., 2005; Spudich et al., 2007).
To generate the constructs for stable expression, human Rab11 cDNA (purchased from University of Missouri-Rolla cDNA Resource Center) and mouse EHD1 and EHD3 (EST cDNAs obtained from the IMAGE Consortium) were cloned into pEGFPC1 vector, the GFP-fusion cassettes were then subcloned into pIRESneo2 (Clontech).
The following antibodies were used: rabbit polyclonal against GFP (Molecular Probes), monoclonal against GFP (3E6, Qbiogene), monoclonals against EEA1, Rab5 and GM130 (BD Transduction Laboratories), monoclonal against transferrin receptor (TfR; Zymed), monoclonal against α-tubulin (Sigma), polyclonal against EGFR (Santa Cruz), monoclonal against LAMP1 (Developmental Studies Hybridoma Bank, University of Iowa), polyclonal against GIPC (Proteus Biosciences), polyclonals against myosin VI tail (Buss et al., 1998) and Vps26 (Seaman, 2004). Rabbit polyclonal antibodies against LMTK2 were generated by immunization with a GST-LMTK2 fusion protein (aa 996-1443) and affinity purified using antigen-CNBr-Sepharose chromatography (GE Biosciences).
The yeast two-hybrid screen was performed using the cytoplasmic domain of LMTK2 (aa 67-1443) as the bait cloned into pY1. A pre-transformed human brain MatchMaker library (Clontech) was screened using LMTK2 as the bait according to the manufacturer's instructions. The interaction between myosin VI and LMTK2 identified in the yeast two-hybrid screen was confirmed in the mammalian two-hybrid assay using the human myosin VI tail (containing the large insert, aa 840-1284) as the bait and the LMTK2 cytoplasmic tail as the prey. The human myosin VI tail was used as a template to generate point mutations WWY→WLY (aa 1191-1193) and RRL→AAA (aa 1115-1117) using a QuickChange protocol (Stratagene). The mutant tail fragments were cloned into the `bait vector' pM (Clontech). A series of LMTK2 deletion mutants (Fig. 1A) were cloned into the `prey vector' pVP16 (Clontech). CHO cells cultured on six-well plates were transfected with 1 μg of both bait and prey vector, together with 0.8 μg of each reporter plasmid – pG5luc (Clontech), encoding inducible firefly luciferase, and pRL-CMV (Promega), encoding the constitutively active Renilla luciferase. After 72 hours the relative luciferase activity was assayed using the Dual-Luciferase Reporter Assay System (Promega).
Cell culture and transfection
CHO, HeLa and RPE cells were cultured in F-12 HAM, RPMI 1640 and DMEM/F-12 medium respectively, containing 10% fetal calf serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin. Cells were transfected using FuGENE (Roche Diagnostics) according to the manufacturer's instruction. To generate cell lines stably expressing GFP-Rab11, GFP-EHD1 or GFP-EHD3, HeLa cells were transfected with the corresponding constructs in pIRESneo2 and clones were selected in complete medium containing 500 μg/ml G418 (Gibco).
HeLa cells growing in 10-cm plates were transfected with the GFP-MyoVI-NI construct and 36-48 hours later lysed in extraction buffer containing 50 mM TRIS pH 7.5, 100 mM NaCl, 20 mM NaF, 20 mM Na4P2O7, 5 mM MgCl2, 5 mM ATP, 1 mM activated Na-o-vanadate, 1% Igepal CA-630 and protease-inhibitor cocktail (Roche). The lysate was cleared by centrifugation (13000 g, 15 minutes) and immunoprecipitation was performed as described previously (Buss at el., 1998). For immunoprecipitation rabbit polyclonal antibodies recognising the tail domain of myosin VI or LMTK2 residues 996-1443 were used. Non-immune rabbit serum was used as a negative control in all experiments. Immunoprecipitated complexes were washed five times in extraction buffer and analysed by SDS-PAGE followed by immunoblotting. Blots were developed using ECL detection reagent (Amersham).
GST pull-down assay
Chicken myosin VI tail (aa 840-1277) fused to GST was expressed and purified as described (Buss et al., 1998). The LMTK2 fragment (aa 451-1095) was cloned into pcDNA3 (Invitrogen), and transcribed and translated in vitro in the presence of [35S]-methionine using the TNT-coupled Reticulocyte Lysate System (Promega). The [35S]-labelled LMTK2 fragment was incubated with 5 μg of GST or myosin VI tail fused to GST on glutathione beads in 10 mM HEPES pH 7.4, 150 mM NaCl, 1% Triton X-100 at 4°C for 2 hours. After extensive washing the protein complexes bound to glutathione beads were separated by SDS-PAGE and analysed by autoradiography.
OligofectAMINE (Invitrogen) was used for transfection of siRNA. For efficient knockdown of myosin VI and LMTK2, HeLa cells were transfected twice with siRNA, on days 1 and 3. On day 5, cells were processed for immunofluorescence and the efficiency of protein depletion was assessed by western blotting. Mock-treated cells and cells transfected with scrambled siRNA oligo were used as controls. All siRNA oligonucleotides were obtained from Dharmacon. Human myosin VI was targeted with ON-TARGETplus Smart pool siRNA; tested separately, all four oligonucleotides yielded the same myosin VI depletion and the same phenotype. LMTK2 was targeted with either oligonucleotide 2068 (5′-UCAGGAGCGUUGAACUUGAUU-3′) or one of the following ON-TARGETplus oligonucleotides, 1158 (5′-GCAGGUACAAGGAGGAUUAUU-3′), 1262 (5′-GCAGAUCAGACUAAGUAUAUU-3′) and 1972 (5′-GUAGUAACUUGGAGCUUGAUU-3′). All four oligonucleotides gave the same level of protein depletion and the same phenotype. Most LMTK2-knockdown experiments described here were performed using a mixture of these four oligonucleotides.
HeLa cells growing on coverslips were transfected with corresponding plasmids and 48 hours later were fixed with 4% PFA, permeabilised with 0.1% Triton X-100, blocked with 1% BSA in PBS and processed for indirect immunofluorescence using primary antibodies (specified in the figure legends) and secondary antibodies coupled with Alexa-Fluor-488 or Alexa-Fluor-555 (Molecular Probes). In RNAi experiments, the cells were plated on coverslips on day 4, and on day 5 fixed and processed for immunofluorescence as described above. The cells were visualised with a Zeiss Axioplan epifluorescence microscope or Zeiss LSM-510 confocal (Carl Zeiss MicroImaging Inc.). For transferrin-uptake experiments, cells were starved in RPMI containing 0.2% BSA for 2 hours, loaded with 5 μg/ml Tf–Alexa-Fluor-555 (Molecular Probes) in RPMI-0.2% BSA for 20 minutes at 37°C, washed in PBS and fixed.
To quantify Rab11- or EHD3-positive tubules, the images of cells containing tubules were taken with Zeiss Axioplan epifluorescence microscope using 100× objective. The number and length of individual tubules were measured manually. Branches were considered as individual tubules.
FACS-based endocytosis assays
Transferrin-uptake and -recycling assays were performed as previously described (Peden et al., 2004). Briefly, for recycling assays, the cells were incubated with transferrin coupled to Alexa-Fluor-647 (Tf–Alexa-Fluor-647; Molecular Probes) for 30 minutes at 4°C followed by internalisation at 37°C or 16°C in the continuous presence of Tf–Alexa-Fluor-647. Cells were then washed and incubated at 37°C in media supplemented with 100 μg/ml unlabelled transferrin for various times before fixation in 3.7% PFA. Cell-associated Tf–Alexa-Fluor-647 was determined by FACS analysis using BD FACSCalibur flow cytometer (BD Biosciences).
In vitro kinase assay
To determine whether LMTK2 can phosphorylate myosin VI, LMTK2 was immunoprecipitated from HeLa cells. The cells were lysed with IP buffer (containing 50 mM Tris pH 7.4, 100 mM NaCl, 20 mM β–glycerophosphate, 20 mM NaF, 20 mM Na4P2O7, 1 mM EGTA, 1 mM sodium o-vanadate, 1 mM DTT, 1% Igepal and protease-inhibitor cocktail) and LMTK2 was immunoprecipitated using antibodies against the cytoplasmic domain (aa 996-1443) coupled to protein A sepharose beads (GE Healthcare). The beads were washed five times with IP buffer and once with 50 mM HEPES, 10 mM MgCl2, 1 mM EGTA, 1 mM DTT. To test for phosphorylation the immunoprecipitated LMTK2 on beads was incubated with 5 μg baculovirus-expressed purified myosin VI in kinase buffer (containing 50 mM HEPES pH 7.5, 10 mM MgCl2, 1 mM EGTA, 1 mM DTT, 0.2 mM o-vanadate, 25 nM microcystin, 0.25 mM ATP and 0.2 μCi/μl [γ-32P]ATP) for 30 minutes at 30°C. Histone H1 (Calbiochem) was used as a positive control substrate. For a negative control the LMTK2 antibodies were substituted with a non-immune IgG for imunoprecipitation and for the following kinase reaction. Reactions were terminated by addition of SDS sample buffer and boiling. The samples were resolved by SDS-PAGE, the gels were fixed, Coomassie-stained, dried and subjected to autoradiography. No significant level of 32P-incorporation into the myosin VI band was detected.
We thank Valerie Cullen for performing the yeast-two hybrid screen that identified myosin VI as a LMTK2-binding partner, S. Arden for testing the interaction in GST pull-down assays, C. Puri for creating Rab11-GFP cell line, S. Gokool for providing GFP-EHD1 and GFP-EHD3 cell lines, A. Peden for help with transferrin uptake assays. We also would like to thank J. P. Luzio for help and advice. We are grateful to D. Brautigan (University of Virginia) for the full-length LMTK2 construct. This work was funded by a Wellcome Trust Senior Fellowship (F.B.), a MRC Senior Fellowship (M.N.J.S.) and supported by the Medical Research Council (J.K.-J. and C.C.M.), BBSRC, MNDA and European Union NeuroNE (C.C.M.). CIMR is in receipt of a strategic award from the Wellcome Trust.