Diverse roles of MYO6 in the endocytic pathway

The process of endocytosis permits the uptake of basic nutrients or cell surface receptors from the plasma membrane. Several parallel uptake routes have been described including clathrin-mediated endocytosis, in which the protein clathrin and associated machinery initiate the formation of vesicles from invaginations in the plasma membrane. After internalization the receptor-containing vesicles are delivered to early endosomes and sorted either for lysosomal degradation or recycled back to the cell surface. Four splice isoforms of MYO6 are expressed in different cell types and tissues, which perform specialized roles at distinct steps along the endocytic pathway. These splice variants contain a large insert (LI) (adding 21–31 amino acids), a small insert (SI) (adding nine amino acids), both LI and SI or no insert (NI) (see Fig. 1D).

The MYO6 splice variant containing the large insert (LI) at the beginning of the cargo-binding domain (CBD) is predominantly expressed in polarized epithelial cells, where it associates with clathrin-coated pits or vesicles at the base of microvilli at the apical plasma membrane. Indeed, loss of MYO6 expression in knock out mice (Snell's waltzer mice) causes defects in endocytosis of the cystic fibrosis transmembrane conductance regulator (CFTR) from the apical domain of enterocytes, as well as in the uptake of the megalin-receptor in the kidney, thereby leading to proteinuria [23-25]. Interestingly, MYO6 has been shown to facilitate the movement of the Na⁺/H⁺ exchanger down the microvillus in intestinal cells, as well as the movement of sodium transporters in renal proximal tubule cells, to allow clustering in clathrin-coated pits and endocytosis [26, 27]. MYO6 targeting to clathrin-coated structures involves the LI splice variant, which constitutes a clathrin-binding domain [28], and the clathrin adaptor Dab2 (Disabled-2) [15]. The endocytic adaptor protein Dab2 contains a phosphotyrosine binding (PTB) domain that mediates binding to the cytoplasmic tail of cell surface receptors. In addition, Dab2 features several binding sites for other components of the endocytic machinery. Indeed, in situ proximity labelling using the MYO6 tail domain identified several proteins associated with clathrin-mediated endocytosis such as AP2, ITSN1, SYNJ1, PICALM, EPS15L1 and FCHO2 that are linked to MYO6 through Dab2 [18].

The isoform of MYO6 without any inserts (NI) is present on a subset of peripheral early endosomes, which are positive for the small GTPase Rab5 and the Rab5 effector APPL1 (Adaptor Protein Phosphotyrosine interacting with PH domain and Leucine Zipper 1) [29-31]. APPL1 is a multifunctional adaptor protein containing multiple domains that mediate phospholipid binding and allow interaction with many signalling receptors. At its C-terminus APPL1 binds to the PDZ domain of GIPC, which interacts with a large number of cell surface receptors including receptor tyrosine kinases, G-protein-coupled receptors and other transmembrane proteins. GIPC also binds directly to MYO6 and recruits this myosin to APPL1-positive early endosomes [32]. The GIPC-MYO6 complex has been suggested to translocate early endosomes away from the plasma membrane through the cortical actin network, since expression of a nonfunctional MYO6 rigor mutant inhibits endosome movement [31, 33, 34]. A crucial role for MYO6 in the spatial organization of the APPL1-signalling endosomes is also supported by the findings that the depletion of MYO6 leads to the premature maturation and displacement of APPL1 endosomes from the cell cortex in the perinuclear region of the cell [35] and expression of a mutant MYO6 that moves towards the plus end of actin filaments causes extreme cortical clustering of endosomes [36]. Finally, recent in situ proximity labelling experiments identified a tripartite complex of GIPC1, MYO6 and LARG, which is a GEF for RHO GTPases. LARG activation downstream of specific G-protein-coupled receptors may link Rho-dependent remodelling of the actin cortex to regulation of early endosome motility and retrograde movement of internalized receptors [18].

Additional MYO6 adaptor proteins on these early endosomes are TOM1 (Target of Myb protein 1) and the related TOM1L2 (Target Of Myb1 Like 2 Membrane Trafficking Protein). Both proteins contain a ubiquitin-binding GAT domain and a VHS domain, similar to the ESCRT-0 subunits Hrs and STAM. This family of proteins plays an important role in the sorting of ubiquitinated cargo destined for lysosomal degradation into late endocytic intraluminal vesicles [37]. Targeting of TOM1/L2 to endosomes involves binding to TOLLIP (Toll-Interacting Protein) and/or endofin, which contains a FYVE domain that binds PI(3)P present on early endosomes. Although the exact function of TOM1 and the MYO6-TOM1 complex is still unclear, TOM1 has been implicated in regulating interleukin-1 and tumour necrosis factor-alpha signalling and trafficking [38, 39].

Another MYO6 adaptor in the endocytic pathway is the transmembrane serine/threonine kinase LMTK2 (Lemur Tyrosine Kinase 2) [40]. This kinase colocalizes with MYO6 on Rab5-positive endosomes, where it appears to participate in the sorting of receptors from early endosomes into the recycling compartment [29]. Little is known about the function of this protein compared to other kinases although it has been shown to regulate endocytic trafficking, as well as signalling and apoptosis. Changes in LMTK2 expression and function are linked to neurodegeneration, cancer and defects in spermatogenesis [41-43]. At present it is not known whether MYO6 activity is regulated through phosphorylation by LMTK2 or whether MYO6 mediates the cellular localization and trafficking of this kinase.

Functions of MYO6 in the exocytic pathway

In addition to its role in the endocytic pathway, MYO6 appears to play a similarly important role in cargo transport in the secretory pathway. In this pathway, proteins synthesized at the endoplasmic reticulum are transported in tubular/vesicular carriers to the Golgi complex, where they are processed and sorted for delivery to the plasma membrane and incorporation or release from the cell. MYO6 localizes to vesicles in the perinuclear region around the Golgi complex [44, 45] and to vesicles close to the plasma membrane [16]. Functional studies using fibroblasts isolated from the Snell's waltzer mouse reveal a significant reduction in Golgi size compared to wild-type fibroblasts, combined with a defect in constitutive exocytosis [45]. Furthermore, MYO6 is required for the polarized delivery of newly synthesized transmembrane proteins to the basolateral plasma membrane domain in polarized epithelial cells and to the leading edge of migratory cells [46, 47]. The role of MYO6 in the secretory pathway is closely linked to optineurin and the small GTPase Rab8, an established optineurin-binding partner. The functions of optineurin and MYO6 in secretion involve secretory vesicle fusion and possibly fusion pore opening at the plasma membrane [48]. In neurosecretory cells the small insert (SI) isoform of MYO6 tethers secretory granules to the actin cortex regulated via c-Src phosphorylation [49]. MYO6-dependent tethering of these secretory granules to the cortical actin network involves ENA/VASP scaffolding proteins that are involved in actin polymerization in addition to interactions with other proteins [49, 50].

MYO6 and autophagy

Autophagy is an essential lysosomal degradation pathway for the breakdown and recycling of cytosolic components, which are sequestered in large double-membrane vesicles, called autophagosomes. After autophagosome closure, amphisomes are formed by fusion with early endosomes, which are believed to deliver essential machinery required for autophagosome–lysosome fusion. Autophagy maintains cellular homeostasis and growth under conditions of starvation in a nonselective process. This pathway also targets large cytosolic objects that are too large to be degraded via the proteasome such as protein aggregates, damaged organelles and invading pathogens. These objects are typically marked with a ubiquitin tag, which is recognized by selective autophagy receptors through their ubiquitin-binding domain. The autophagy receptors typically interact with LC3 (Microtubule-associated Protein 1A/1B-Light Chain 3), triggering the recruitment of LC3-positive autophagosomal membranes and thereby capturing the ubiquitinated object inside an autophagosome. MYO6 has been shown to bind directly to three of the well-known autophagy receptors, optineurin, NDP52 (Nuclear Dot Protein 52 kDa) and TAX1BP1 (Tax1-Binding Protein 1). In all three autophagy receptors the ubiquitin-binding site exactly overlaps with the MYO6-binding site, however, the binding of TAX1BP1 to MYO6 is stronger than the interaction of the autophagy receptor with ubiquitin [51]. These findings are consistent with a dual function of the autophagy receptors: first MYO6-independent recognition and capture of ubiquitinated cytosolic objects and second in the recruitment of this myosin to the outer surface of autophagosomes by binding simultaneously to MYO6 and autophagosome-associated LC3. The autophagy receptors and TOM1 bind to distinct sites in the MYO6 CBD; shown by pull down experiments demonstrating that the distance between these two binding domains allows simultaneous binding of the autophagy receptors and the endosome-anchor TOM1 [52]. These results support the model that a ternary complex of MYO6, TOM1 and the autophagy receptors provides a mechanism for the tethering of endosomes to autophagosomes to facilitate endosome–autophagosome fusion and maturation [30].

In addition, MYO6 has a crucial role in xenophagy for the autophagy-dependent clearance of pathogens such as Salmonella typhimurium from the cytosol of infected cells [51]. Furthermore, during mitophagy, the pathway for removal of nonfunctional mitochondria via autophagy, MYO6 is selectively recruited to damaged mitochondria. MYO6 recruitment does not depend on optineurin, TAX1BP1 and NDP52 but requires its ubiquitin-binding domain [53]. MYO6 induces the formation of actin cages around damaged mitochondria thereby isolating dysfunctional mitochondria destined for mitophagy from the healthy network. This novel role for MYO6 is unique to mitophagy, where the actin cages prevent the re-integration of damaged mitochondria into the functional mitochondrial network.

MYO6 adaptor proteins regulate actin track dynamics

Accumulating evidence indicates that myosin motors do not simply translocate along pre-existing actin filaments, but also actively induce actin tracks as required. For example, class IX myosins (MYO9) have RhoGAP domains in their tails, which regulate RhoGTPase-dependent modulation of actin dynamics; class V myosins (MYO5) have recently been shown to coordinate motor targeting with assembly of actin tracks by binding to the actin nucleator SPIRE2, and myosins of class I can interact directly with proteins that regulate actin patch formation at the plasma membrane [54-56]. The recent identification of MYO6-associated RhoGEF complexes, which link to different actin modulatory pathways, builds significantly on this regulatory theme and highlights the extent to which motor activity is synchronized with actin track assembly. In situ proximity labelling identified LRCH1 and LRCH3 of the LRCH (Leucine-Rich repeats Calponin Homology domain) family of proteins in a complex with MYO6. These proteins comprise a series of N-terminal leucine-rich repeats (LRRs), involved in protein–protein interactions, and a C-terminal calponin homology domain, which can directly bind to actin. LRCH1 and LRCH3 associate directly with MYO6 and DOCK7 (Dedicator Of CytoKinesis 7) thus constituting the linker between MYO6 and DOCK7 [57]. DOCK7 belongs to the DOCK180 family of guanine nucleotide exchange factors (GEFs), which catalyse the loading of GTP onto Rho GTPases through a DOCK homology region (DHR)-2 domain and thereby activating them [58]. In the case of DOCK7 it appears to act as a GEF towards prenylated membrane localized RAC1 (Ras-related C3 botulinum toxin substrate 1) and CDC42 (Cell division control protein 42), two major regulators of the actin cytoskeleton [59]. Functionally, DOCK7 acts predominantly in the nervous system and has been implicated in Schwann cell migration, myelination and neurite outgrowth [60, 61]. MYO6, LRCH3 and DOCK7 form a tripartite complex, the DISP (DOCK7-Induced Septin disPlacement) complex [18]. BioID data revealed possible links between LRCH3 and septins (SEPT), in particular SEPT7, but also SEPT8, SEPT9 and SEPT10. Septins are a large family of GTPases with phospholipid-binding activity that assemble into filamentous or ring-like structures, which associate with distinct subsets of actin filaments and microtubules, as well as membranes of specific curvature and lipid composition [62, 63]. Interestingly, overexpression of components of the DISP complex displaces the septins from actin filaments and induces the formation of cytosolic septin rings. Dynamic actin filaments in membrane ruffles, for example are devoid of septin structures, therefore, the DISP complex might regulate the removal of septin structures from actin filaments to enable actin filament reorganization through the DOCK7-stimulated RAC1 and CDC42 GTPase activity [18].

The second RhoGEF associated with MYO6 is LARG (Leukaemia-associated RhoGEF), also known as ARHGEF12, which is a RhoA-specific RhoGEF first identified as a fusion product with MLL in acute myeloid leukaemia [64]. LARG is among a family of RhoGEFs activated by the Gα_12/13 G protein subunits and thus mediates RhoA activation downstream of a number of receptors.

MYO6 adaptors for specialized functions

Myosins of class VI also appears to have a number of distinct tissue-specific functions. The interaction between MYO6 and SAP97 (synapse-associated protein 97) has only been reported in neuronal tissues so far [65]. Indeed, MYO6 has been identified in a complex with SAP97 and the AMPA receptor subunits, GluA1 and GluA2, and functionally appears to regulate the delivery or endocytosis of the receptors to or from the cell surface in an activity-dependent fashion [65-67].

In inner ear hair cells, MYO6 binds to otoferlin, a transmembrane protein and putative Ca²⁺ sensor of synaptic exocytosis. The otoferlin-MYO6 complex has been suggested to participate in targeting neurotransmitter vesicles to the specialized ribbon synapses at the basolateral membrane of these hair cells for exocytosis and signal transmission to ganglia neurons [68, 69]. Since mutations in both MYO6 and otoferlin cause deafness in humans, this suggests a complementary role for these proteins in normal hair cell function in the ear [70, 71].

A yeast two-hybrid screen has identified phospholipase C (PLCδ3), an important enzyme in phosphoinositide metabolism, as a MYO6-binding partner [72]. Although PLCδ3 is also expressed in inner ear hair cells, no hearing abnormalities were observed in the PLCδ3 KO mouse. The expression of MYO6, however, is decreased in the intestine of the PLCδ3 KO mouse and loss of the enzyme decreases the number of microvilli in the small intestine. Thus, a functional complex of PLCδ3 and MYO6 may be required for maintenance of apical microvilli in polarized epithelial cells [72]. In addition, in myogenic cells MYO6 forms a complex with AKAP9 [73] (A-Kinase Anchoring Protein 9), a scaffolding protein linked to protein kinase A signalling pathways.

Finally, MYO6 has also been suggested to function together with the putative transcription coactivator NDP52 in the nucleus and to regulate RNA polymerase II-dependent transcription [74, 75].

MYO6 ABD1 and MyUb domain encompassing the RRL motif

GIPC binds directly to the ABD1 [79]. In its inactive form GIPC exists as an autoinhibited dimer, which opens up upon cargo binding to interact with the RRL motif in MYO6 [20]. The GIPC-MYO6 complex can assemble into linear higher order oligomers, in which GIPC binds via its C-terminal GIPC-homology 2 (GH2) domain to the RRL motif, which is embedded in the middle of the second α-helix in MYO6 ABD2^1084–1128. The GH2 of the first GIPC molecule interacts with the two arginines (RR^1116,1117) of the RRL motif and an extended region to allow hydrophobic interactions [20]. The GH2 domain of the second GIPC molecule binds to the opposite side of the MYO6 CBD covering the leucine (L¹¹¹⁸) of the RRL motif allowing the formation of five, or even longer, linear GIPC-MYO6 oligomers. In vivo observations support the importance of MYO6 oligomerization induced by GIPC through a mutant MYO6+ that moves towards the plus end of actin filaments. MYO6+ induced filopodia formation only occurs in the presence of GIPC, which accumulates together with MYO6+ at the tips [36]. Interestingly, an artificially dimerized MYO6 without the CBD, can rescue this phenotype, which supports the importance of GIPC-mediated dimerization/oligomerization for MYO6 functions in vivo.

Direct binding of GIPC and other adaptor proteins to the ABD1 not only requires the RRL motif but also the neighbouring extended hydrophobic surface^1123–1131 (see Fig. 1D) [21]. Interestingly, this region is dispensable for ubiquitin binding, but essential for direct adaptor binding. In contrast to GIPC, NDP52, TAX1BP1 and optineurin not only interact directly with MYO6 [51, 52], but potentially also bind indirectly via ubiquitin chains to the MYO6 MyUb domain [21, 22]. MYO6 interacts with K48, K63, K29 and K11 linked ubiquitin chains, with a clear preference for the latter three [21]. Both MYO6 and TAX1BP1 have been shown to bind to K63 Ub⁴ chains simultaneous, thus forming a MYO6-ubiquitin-TAX1BP1 complex [22]. Therefore, the interaction of MYO6 with the autophagy receptors may have extra regulation via ubiquitin chains.

Finally, MYO6 can also directly interact with ubiquitinated proteins via the MyUb domain or the MIU-binding motif in the absence of any of the known cargo-adaptor proteins. MYO6 MyUb recognizes and directly interacts with ubiquitinated mitochondria marked for degradation by Parkin in vivo [22]. The MIU-binding motif shows no affinity or preference towards ubiquitin linkages; however, in cooperation with the MyUb domain it seems to specifically support K11-linked ubiquitin binding.

MYO6-adaptor interaction at the WWY-binding motif

The interaction of MYO6 with the clathrin adaptor Dab2 has been suggested to induce dimerization of MYO6 [76]. Targeting of MYO6 to clathrin-coated structures at the plasma membrane involves Dab2-binding via the WWY and IWE^1261-1263 protein-interaction motifs within the ABD2^1181-1263, and binding to PtdIns(4,5)P₂ via the phospholipid domain^1122–1131 (PIP₂) [52, 76, 79, 81]. In addition, MYO6 may interact directly with clathrin through a binding region that has been identified in the α2-linker helix of the LI splice isoform [28]. The MYO6 CBD contains two discrete spatially separated Dab2-binding sites, site I (including the WWY motif) and site II (including the IWE motif). Structural data suggest that Dab2 may induce dimerization of MYO6 by binding to the WWY motif on one MYO6 monomer and the IWE motif of another [76]. The remaining two binding sites on the two MYO6 monomers are occupied by a second Dab2 in the reverse orientation thereby stabilizing the dimer. Site I is the major binding site, allowing hydrophobic interactions between Dab2 and MYO6, whereas binding site II interacts with Dab2 through both charge and hydrophobic interactions thereby increasing the binding affinity [76]. Point mutations of the tryptophan of either binding motif, W¹²⁰² of WWY or W¹²⁶² of IWE, disrupt Dab2 binding to MYO6 and abolish cellular localization to clathrin-coated structures at the plasma membrane in vivo [52, 76]. The importance of site I was confirmed by in situ proximity labelling, which demonstrated the loss of Dab2 and other proteins associated with clathrin-mediated endocytosis upon mutation of the WWY site [18].

TOM1 also binds to the WWY motif in the ABD2, as determined by pull downs and mammalian 2-hybrid in vitro [30]. MYO6 may simultaneously interact, through its ABD1, with the LC3-binding autophagy receptors (NDP52, TAX1BP1 and optineurin) and, through the WWY motif in ABD2, with TOM1-positive endocytic vesicles [82]. Indeed, in vitro experiments using purified proteins have shown that two adaptor proteins such as TOM1 and NDP52 can bind simultaneously directly to the MYO6 ABD1 and the ABD2 [52]. Interestingly, in situ proximity labelling of MYO6 NI CBD has indicated that the mutation of the RRL motif also leads to the loss of TOM1, and its partner TOLLIP [18]. These results suggest that although the interaction between TOM1 and MYO6 occurs directly via the WWY motif, in vivo the binding between MYO6 and TOM1 may also be indirect via ubiquitin using the ubiquitin-binding TOM1 VHS domain and the MYO6 MyUb domain. A detailed structural and biochemical analysis is required, to distinguish whether the MYO6-TOM1 interaction requires ubiquitin chains or occurs by direct binding.

Actin track composition

The actin filament track composition provides the first layer of regulation as myosin activity is responsive to the actin isoform involved, the actin-binding proteins present and the nucleotide-state of the actin monomer.

Actin filaments are highly dynamic structures that assemble by the addition of new ATP-actin monomers to the plus-end of filaments at the plasma membrane or the surface of internal membranes. Through the constant addition of new ATP-actin the filaments undergo retrograde flow and sequential ATP-hydrolysis, referred to as filament ageing. Although ATP-actin and ADP-Pi actin filaments have very similar properties, the ADP-actin accumulating at the minus-end of the filament is less rigid and stable. Interestingly, in vitro MYO5 shows increased run lengths on ADP+Pi-rich filaments, whereas MYO6 which prefers ‘old’ ADP-rich filaments [6]. Recent cryo-EM structures confirm minor conformational changes between nucleotide states of F-actin [84], as well as within the actin-binding interface of the motor domain of nonmuscle myosin class IIc (MYH14) and MYO6 [77, 85].

From the six different actin isoforms expressed in mammalian cells, four isoforms are only present in muscle cells. The two cytoplasmic actin isoforms (β-cyto and γ-cyto) are found in most cell types and tissues. Although these two cytoplasmic actins only differ by four residues in their N-terminal sequence, they can assemble into distinct actin structures in nonmuscle cells and preferentially associate with different classes of myosins. At present it is not known whether MYO6 is differentially activated by different actin isoforms, however, MYH14 shows increased kinetics when interacting with cytoplasmic β-actin, whereas ATPase activity of MYO7A results in greater stimulation by γ-actin [86].

Activation of select myosin motors is not only influenced by the actin isoforms that make up a filament, but is also regulated by a large number of actin-binding proteins. In this respect, the most important is probably tropomyosin, which has been well-characterized together with troponin for its role in Ca²⁺ regulation of sarcomeric myosins of class II in striated muscle. In nonmuscle cells 40 tropomyosin (Tm) isoforms have been identified, which spatially and temporally decorate distinct actin filament assemblies [87, 88]. The presence of these tropomyosin isoforms on different actin structures modulates the affinity of myosin motors and leads to the select recruitment of specialized myosins to functionally distinct actin filaments. For example, myosin class Ic (MYO1C) does not bind to actin filaments containing Tm2, whereas MYO5A selectively engages with Tm3.1 –decorated actin filaments [89, 90]. Future studies will show whether MYO6 preferentially binds to different actin filament populations containing distinct tropomyosin isoforms.

Regulation of MYO6 through phosphorylation

In addition to the regulation of motor activity through crosstalk with the actin filament track, post-translational modifications of the myosin heavy chain, or associated light chains, are another important regulatory mechanism. For example, nonmuscle MYO2s are regulated by phosphorylation of the regulatory light chains by a variety of kinases, which trigger dramatic structural changes of the myosin from a folded inactive state to an unfolded active state [91]. In Acanthamoeba, several of the myosin class I family members (myosin Ia, b and c) have been shown to be regulated by phosphorylation at a single serine or threonine in the motor which enhances the actin-activated ATPase [92]. This site obeys the TEDS rule (it must either be a Thr, Glu, Asp or Ser) for myosin motor domain phosphorylation postulated by Bement and Mooseker [93]. MYO6 is also phosphorylated at Thr405 in the cardiomyopathy loop in the actin-binding region, and this phosphorylation event affects the K_ATPASE but not the V_max [9]. Phosphorylation in the MYO6 motor domain occurs downstream of growth factor receptor activation and is accompanied by its recruitment to membrane ruffles [44]. Moreover, Salmonella virulence effectors, such as SopE, have been suggested to activate MYO6 through PAK-dependent phosphorylation at Thr405 [94].

Backfolding regulates ATPase kinetics

Backfolding as a way to inactivate ATP hydrolysis is a recurring theme in the myosin family (II, V, VI, VII and X) [95-97]. MYO5 has been shown to take a compact conformation, where the tail back-folds onto the motor domain and in this conformation the ATPase activity is reduced 50-fold [98]. In the case of MYO5, this backfolding is regulated by calcium binding to, and changing the conformation of, calmodulin bound to the neck domain.

MYO6 has two calmodulins bound at the neck domain. The first calmodulin is a structural calmodulin that is ‘locked’ into the molecule. The second calmodulin has been shown to change conformation depending on the number of Ca²⁺ ions bound. Upon binding Ca²⁺, one lobe detaches from the heavy chain and rebinds to a higher affinity binding site present in the 3-helix bundle at the beginning of the tail [97]. This has the effect of ‘breaking’ the lever arm, so the molecule cannot translocate cargo, and simultaneously the tail domain enters into a ‘primed’ position making it more readily able to bind cargo [97]. In the absence of cargo, and in a return to the low Ca²⁺ condition, the molecule will retake a compact backfolded conformation. If cargo is bound, the restored lever arm can now facilitate active translocation. The backfolded molecule can also unfold in the presence of high concentrations of binding partners similar to MYO5 [99]. This may simply be a matter of shifting equilibriums between the inactive and active state, however, on its own this is probably not an effective method of regulation within the cell.