Driving autophagy – the role of molecular motors Free

Molecular motors play a crucial role in facilitating the passage of various cellular components in vesicular transport pathways along cytoskeletal tracks. The cytoskeletal tracks, composed of microtubules and actin, provide a structural network through which motor proteins move cargoes, contributing to diverse cellular activities. One such extensively studied vesicular transport pathway is autophagy, an evolutionarily conserved process that is essential for maintaining cellular homeostasis. Autophagy involves the formation of a double-membraned structure around cargo, which is destined for lysosome-mediated degradation (Aman et al., 2021). Over the years, significant progress has been made in understanding autophagic flux, with an appreciation for the role of the endoplasmic reticulum (ER) in the biogenesis of the autophagosomal membrane (Nakatogawa, 2020). The initial isolation membrane, or phagophore, then expands through the addition of membranes from various cellular reservoirs, forming the autophagosome enclosing the cargo (Melia et al., 2020). Subsequently, upon fusion of the autophagosome with the late endo-lysosomal compartment, lysosomal hydrolytic enzymes degrade the cargo (Nakatogawa, 2020). Autophagosome formation, maturation and fusion are orchestrated by the hierarchical assembly of various autophagy factors onto the autophagosome membrane. As autophagosomes progress through their maturation stages, they interact with cytoskeletal tracks and molecular motors, which facilitate their journey. Notably, considerable research has focused on the role of cytoskeletal tracks in autophagic flux; it has been shown that drugs that promote microtubule disassembly, such as nocodazole and Taxol, prevent autophagosome formation, suggesting that microtubules have a role in autophagosomal biogenesis (Köchl et al., 2006; Geeraert et al., 2010). However, the role of microtubules in later autophagic steps such as autophagosome–lysosome fusion remains controversial (Fass et al., 2006; Mackeh et al., 2013). Whereas microtubules and their associated motors participate in the transport of autophagosomes, actin has been shown to provide structural support during autophagic flux (Kast and Dominguez, 2017). Indeed, starving cells treated with actin-depolymerising compounds, such as cytochalasin D and latrunculin B, fail to generate autophagosomes (Aplin et al., 1992). The actin-branching factor actin-related protein 2 and 3 (Arp2/3) participates in the shaping of autophagosomes during the initial stages, suggesting that a branched actin network drives the formation of autophagosomes (Monastyrska et al., 2008; Campellone et al., 2023). A number of actin-nucleation-promoting factors, which are responsible for maintaining the activity of Arp2/3, also regulate autophagy – including WASP homologue associated with actin, Golgi membranes and microtubules (WHAMM) (Kast and Dominguez, 2015, 2017; Kast et al., 2015); junction mediating and regulatory protein (JMY) (Coutts and La Thangue, 2015; Hu and Mullins, 2019; Liu and Klionsky, 2019); and members of the Wiskott–Aldrich syndrome protein and SCAR homologue (WASH) complex (Dupont and Codogno, 2013; Zavodszky et al., 2014) – and they colocalise with several proteins involved in autophagosomal biogenesis, such as zinc finger FYVE domain-containing protein 1 (DFCP1, also known as ZFYVE1), ATG5, Beclin-1 and sequestosome-1 (SQSTM1, also known as p62), indicative of their role in early stages of autophagy. The association of cytoskeletal tracks with autophagy components at different stages of the process is mediated by molecular motors, namely, dynein, kinesins and myosins. In this Review, we explore the roles of molecular motors in autophagic flux, with emphasis on post-translational modifications, the mode of motor recruitment and their interactions with different components of autophagic flux. Furthermore, we discuss motor-mediated strategies for autophagosomal movement and the role of motors in disease-related dysregulation of the autophagic pathway, while also addressing outstanding questions in the field.

Dyneins belong to the AAA+ family of proteins and are motors that move towards the minus end of microtubules. There are three dynein motors: cytoplasmic dynein-1; dynein-2, which is involved in intraflagellar transport; and axonemal dynein (Reck-Peterson et al., 2018). Dynein-1 (hereafter referred to as dynein) is a giant protein that mostly functions in complex with its co-factor dynactin. A subunit of dynactin, p150^Glued (also known as DCTN1), enables the recruitment of the motor complex onto the microtubule, while the other subunits, which include lissencephaly-1 (LIS1, also known as PAFAH1B1) and nuclear distribution protein nudE-like 1 (NDEL1) fine-tune motor activity (Kardon and Vale, 2009; Roberts et al., 2013; Xiang and Qiu, 2020). Dynein has been implicated in autophagic flux, as mutations in dynein impair autophagosome–lysosome fusion owing to reduced autophagosome transport towards the lysosomes in the perinuclear region (Jahreiss et al., 2008). Inhibiting dynein activity using erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA) or genetic manipulation of dynein motor complex components results in impaired clearance of autophagic substrates, an effect that has been observed consistently across different model systems (Ravikumar et al., 2005; Batlevi et al., 2010; Kimura et al., 2008). Furthermore, two independent reports using live-cell microscopy and microinjection techniques have strengthened the notion of the involvement of dynein in autophagic flux by coming to the consensus that autophagosomes engage in continuous bi-directional migration followed by vectorial centripetal movement towards lysosomes in the microtubule-organising centre (MTOC) region (Jahreiss et al., 2008; Kimura et al., 2008). However, the interplay between the dynein motor complex and the autophagosome, as well as its directed movement towards the lysosome, turns out to be more complicated than initially perceived. Much like other cellular cargo, autophagosomes navigate the dense cytoplasm to reach their assigned destination within the cell. Consequently, the dynein-mediated trajectories adopted by autophagosomes frequently show stationary pauses, stops, fusion with other organelles, track modifications and associations with various dynein adaptors (Cai and Ganesan, 2022). A recent study employing mathematical modelling as well as experiments has illuminated the journey of autophagosomes within neuronal cells, underscoring the significance of autophagosome maturation within a spatiotemporal framework (Cason et al., 2022). Autophagic flux in neurons is more compartmentalised compared to that in other cell types, with the majority of autophagosome biogenesis occurring at the distal neurite tip, from where autophagosomes then travel towards the cell soma, where lysosomes are enriched (Maday et al., 2012; Maday and Holzbaur, 2016; Kulkarni and Maday, 2018). However, local autophagosome biogenesis has also been observed in axons, particularly in the case of neuronal mitophagy, where co-transport of PTEN-induced kinase 1 (PINK1) mRNA with mitochondria facilitates dendritic and axonal translation of PINK1 to help the clearance of damaged mitochondria (Harbauer et al., 2022). It has therefore been speculated that this local translation of PINK1 and subsequent PINK1–parkin (PRKN)-mediated mitophagy might further require the assembly of downstream autophagic machinery for efficient capture and degradation of damaged mitochondria. Both dynein and kinesin co-fractionate with axonal autophagosomes (Maday et al., 2012). The retrograde movement of autophagosomes from the distal tip is mediated by the dynein motor complex. Indeed, the dynein inhibitor EHNA disrupts the retrograde movement of axonal autophagosomes toward the cell body, emphasising the role of dynein in this process (Maday et al., 2012). Moreover, expression of the CC1 domain of p150^Glued, which acts a dominant-negative dynein inhibitor known to disrupt the dynein–dynactin complex, delays retrograde movement, whereas the presence of the dominant-negative tail of the kinesin-1 heavy chain KIF5C does not, suggesting a preferential association of autophagosomes with the dynein complex for retrograde transport (Maday et al., 2012; Maday and Holzbaur, 2014).

Different dynein-associated adaptors control the conformation of dynein that is needed for movement along microtubules by stabilising the dynein–dynactin interaction and regulating motor engagement with diverse cargoes (Xiang and Qiu, 2020). The scaffolding proteins JNK-interacting protein-1, -3 and -4 (JIP1, -3 and -4, also known as MAPK8IP1, MAPK8IP3 and SPAG9, respectively), and Huntingtin-associated protein 1 (HAP1) participate in autophagosome biogenesis, maturation and fusion with lysosomes, as has been observed in mouse primary neuronal cultures (Fu and Holzbaur, 2013,; Fu et al., 2014; Kulkarni and Maday, 2018; Cason et al., 2021; Rafiq et al., 2022) (Fig. 1, Table 1). The sequential binding of JIP1, HAP1, and a complex containing JIP3 and/or JIP4 (hereafter JIP3/4) occurs as autophagosomes egress the distal axonal tip and steer towards the soma for fusion with lysosomes. This process might be governed by the incorporation or dissociation of additional factors on the autophagosomal membrane, which in turn is dictated by the maturation status of the autophagosomes in the axon (Cason et al., 2021) (Fig. 1). Various other adaptors, such as Rab-interacting lysosomal protein (RILP; Wijdeven et al., 2016; Khobrekar and Vallee, 2020; Khobrekar et al., 2020), oxysterol-binding protein homologue (ORP1L, also known as OSBPL1A; Wijdeven et al., 2016), the adaptor protein 2 (AP2) complex (Kononenko et al., 2017), SNAP-associated protein (SNAPIN), which is a component of the BLOC-1 complex (Cai and Sheng, 2011), and the HTT–HAP1 complex interact with different components of the dynein complex and autophagosomes, thereby promoting retrograde movement in different cell types (summarised in Table 1). RILP, through its three LC3-interacting region (LIR) motifs in its C-terminal domain, directly interacts with autophagosomes. Additionally, RILP interacts with ATG5 to regulate the maturation of autophagosomes. Depletion of RILP results in the accumulation of autophagosomes with impaired retrograde movement (Khobrekar et al., 2020). The endocytic AP2 complex transports autophagosomes containing brain-derived neurotrophic factor (BDNF) and/or tropomyosin-related kinase receptor type B (TRKB, also known as NTRK2) in a retrograde manner from the distal axon. Two distinct AP2 complex subunits, AP2α_A (also known as AP2A1) and AP2β (also known as AP2B1), interact with LC3 proteins (MAP1LC3A, MAP1LC3B and MAP1LC3C; hereafter collectively termed LC3) and p150^Glued to facilitate the attachment of autophagosomes to the dynein motor complex (Kononenko et al., 2017) (Fig. 1, Table 1).

The interaction of dynein with adaptors and autophagosomes is also controlled by various post-translational modifications. For instance, BCL2-like 11 (BCL2L11, also known as BIM) and Beclin-1-regulated autophagy protein 1 (AMBRA1) directly interact with microtubules through the dynein light chain 1 (DYNLL1) (Di Bartolomeo et al., 2010; Luo et al., 2012). Under basal conditions, these proteins recruit the Beclin-1–VPS34 (PIK3C3) complex to microtubules (Di Bartolomeo et al., 2010; Luo et al., 2012; Luo and Rubinsztein, 2013). However, upon autophagic induction, BCL2L11 and AMBRA1 undergo phosphorylation, causing the Beclin-1–VPS34 complex to detach from microtubules, subsequently initiating the formation of autophagosomes (Di Bartolomeo et al., 2010; Luo et al., 2012; Luo and Rubinsztein, 2013). In addition, an RNAi-based screen in Drosophila melanogaster has identified the striatin-interacting phosphatase and kinase (STRIPAK) complex as a regulator of retrograde movement of autophagosomes in Drosophila neurons. Connector of kinase to AP-1 (Cka) and protein phosphatase 2A (PP2A), two Drosophila STRIPAK complex components, participate in autophagosomal movement and have been suggested to regulate dynein activity and its interaction with various proteins through phosphorylation-dependent modifications (Neisch et al., 2017). Lysosomal Ca²⁺ signalling modulates autophagic flux via the TFEB axis (Medina et al., 2015; Medina, 2021). Additionally, dynein-mediated motility is activated in response to Ca²⁺ ion changes in the cell (Wang et al., 2019). During starvation, lysosomes are trafficked from the peripheral reserves to the perinuclear region to facilitate fusion with autophagosomes (Li et al., 2016). Starvation-mediated efflux of lysosomal Ca²⁺ ions through the transient receptor potential channel mucolipin 1 (TRPML1, also known as MCOLN1) on the lysosome is sensed by the lysosomal Ca²⁺ sensor apoptosis-linked gene 2 (ALG2, also known as PDCD6), which then physically interacts with the dynein–dynactin complex to drive the perinuclear migration of lysosomes, promoting their fusion with autophagosomes (Li et al., 2016). Therefore, exploring different regulatory mechanisms dictating the participation of dynein in autophagic flux would help delineate molecular nuances in the pathway.

The kinesin (KIF) superfamily of proteins constitutes another class of microtubule-associated molecular motors that transport cargo to either the minus end or plus end of microtubules (Hirokawa et al., 2009). Based on phylogenetic studies, the fifteen subfamilies that make up the kinesin superfamily can be broadly divided into three groups: N-kinesins, M-kinesins and C-kinesins (Hirokawa et al., 2009; Ali and Yang, 2020). Post-translational modifications (Hammond et al., 2010), Ca²⁺ signalling (Vinogradova et al., 2004), Rab GTPase activity (Delevoye and Goud, 2015) and binding to adaptors all influence the activity of kinesins and their cargo engagement (Tempes et al., 2020; Canty et al., 2023). Kinesins have also been investigated for their role in the autophagic flux (Fig. 2). For instance, depletion of kinesin 5B (KIF5B), a member of the kinesin-1 family, and the kinesin-3 family protein kinesin 1B (KIF1B) in HeLa cells results in perinuclear accumulation of both autophagosomes and lysosomes, which is accompanied by inefficient clearance of autophagosomes (Cardoso et al., 2009). Post-translational modifications on microtubules regulate the motility of dynein and kinesin motors (Janke and Chloë Bulinski, 2011; Sirajuddin et al., 2014). Interestingly, by using super-resolution imaging techniques, it has been discovered that lysosomes localise to a considerably greater extent on de-tyrosinated microtubules than on other microtubule populations in a kinesin-1-dependent way (Mohan et al., 2019). Furthermore, de-tyrosinated microtubules are platforms for the fusion of autophagosomes with lysosomes, wherein reduced lysosomal motility coupled with frequent pauses enhances contact with the autophagosomes, thereby increasing the likelihood of fusion (Mohan et al., 2019). In this scenario, genetic silencing of KIF5B leads to a decrease in the autolysosomal population due to decreased autophagosome–lysosome fusion (Mohan et al., 2019). In mammalian cells, ATG9A, an integral autophagosomal membrane protein, shuttles between the perinuclear trans-Golgi network, autophagosomes and peripheral regions, and this is facilitated by various proteins including the adaptor protein 4 (AP4) complex with its accessory subunit RUN and SH3 domain-containing protein 2 (RUSC2) (Bielska et al., 2014; Davies et al., 2018; Mattera et al., 2020). Binding of ATG9A to kinesin-1 has recently been shown to be mediated by an interaction of RUSC2 with kinesin light chain 2 (KLC2) (Guardia et al., 2021). Klp98 (also known as Klp98A), an orthologue of human KIF16B, has been identified as having a role in autophagy based on an RNAi screen in Drosophila (Mauvezin et al., 2016). Depletion of Klp98 in Drosophila fat bodies leads to a decrease in the number of autophagic vesicles, which is accompanied by an abnormal clustering of autophagosomes in the perinuclear region. Klp98 interacts with Rab14 on lysosomes and Atg8A on autophagosomes, and has therefore been speculated to mediate autophagosome–lysosome fusion by bringing the two compartments in close proximity (Mauvezin et al., 2016; Mauvezin and Neufeld, 2017). As mentioned above, autophagic flux in neurons is compartmentalised, and the KIF1A homologue UNC-104 is involved in the compartmentalisation of autophagy in Caenorhabditis elegans AIY interneurons (Stavoe et al., 2016). Here, the KIF1A homologue regulates localised biogenesis of autophagosomes by directing the anterograde migration of ATG-9 to the synaptic area where autophagosome biogenesis takes place. In atg-9 and unc-104 mutant worms, a substantial decrease in the rate of autophagosome biogenesis is observed, further supporting a role for the KIF1A homologue in trafficking of ATG-9 (Stavoe et al., 2016) (Fig. 2A).

Kinesins interact with autophagosomes via adaptor proteins, including FYVE and coiled-coil domain containing 1 (FYCO1), which binds to LC3A (MAP1LC3A) and LC3B (MAP1LC3B) through an extended LIR motif in its C-terminal region (Olsvik et al., 2015; Cheng et al., 2016). Rab7 (herein referring to Rab7A) is known to localise on the membranes of lysosomes and autophagosomes, facilitating their fusion (Gutierrez et al., 2004; Hyttinen et al., 2013). The coiled-coil region of FYCO1 undergoes dimerisation and forms a complex with Rab7, as well as interacting with phosphatidylinositol 3-phosphate (PI3P) via a FYVE domain. A centrally located kinesin binding site on FYCO1 likely directs the plus end-directed motility of FYCO1 on microtubules (Pankiv and Johansen, 2010). In HeLa cells, ninein-like protein (NLP, also known as NINL) overexpression enhances the colocalisation between autophagosomes and lysosomes, whereas its depletion has the opposite effect (Xiao et al., 2021). The NLP–FYCO1–Rab7 complex has thus been suggested to control autophagosome loading onto kinesins, driving their motility towards the microtubule plus end (Xiao et al., 2021) (Fig. 2B). The binding of FYCO1 to LC3 is regulated by serine/threonine kinase 4 (STK4)-dependent phosphorylation of LC3 at its T50 residue during starvation. In the absence of this LC3 phosphorylation, autophagosomes move aberrantly towards the cell periphery, which reduces the frequency at which they encounter and subsequently fuse with lysosomes (Nieto-Torres et al., 2021a,b). This phosphorylation event decreases the affinity of FYCO1 for autophagosomes and shifts the motility of autophagosomes towards the perinuclear region of the cell, possibly by the disengagement of kinesins and subsequent binding of dynein to the autophagosomal membrane (Nieto-Torres et al., 2021a,b) (Fig. 2B). Another protein complex, BLOC-1-related complex (BORC), which comprises eight subunits, has been previously reported to regulate kinesin-dependent positioning of lysosomes in non-neuronal cells (Pu et al., 2015). Fusion of the autophagosome with the lysosome is facilitated by the orchestration and interaction between several protein complexes, including SNARE proteins, syntaxin 17 (STX17), the homotypic fusion and protein sorting (HOPS) complex, Rab proteins and vesicle-associated membrane proteins (VAMPs), among others (Nakamura and Yoshimori, 2017; Lorincz and Juhász, 2020). By assisting in recruitment of the HOPS complex to the lysosome and by regulating the assembly of STX17–SNAP29 on autophagosomes, BORC might mediate productive encounters between autophagosomes and lysosomes, thereby facilitating their eventual fusion (Jia et al., 2017) (Fig. 2C). Another protein complex consisting of trafficking kinesin proteins 1 and 2 (TRAK1 and TRAK2, collectively referred to here as TRAK1/2) and mitochondrial Rho-GTPase (Miro; herein referring to Miro1, also known as RHOT1) acts as an adaptor to connect kinesin heavy chain (KHC) proteins to the mitochondrial surface (Klosowiak et al., 2016; Canty et al., 2023). This complex regulates the positioning, motility and degradation of damaged mitochondria via a selective autophagy pathway, called mitophagy (Baltrusaitis et al., 2023; Canty et al., 2023; Jetto et al., 2022). Damage to mitochondria triggers mitochondrial depolarisation, which results in recruitment of parkin to the outer mitochondrial membrane (OMM) and its phosphorylation by PINK1; PINK1 then phosphorylates Miro on the OMM, promoting its proteasomal degradation (Wang et al., 2011; Pickles et al., 2018). Ca²⁺ binding to Miro, coupled with phosphorylation, results in degradation of Miro and subsequent detachment of the KHC, halting the movement of damaged mitochondria (Wang and Schwarz, 2009; Wang et al., 2011). This arrest enhances recruitment of the autophagic machinery, promoting mitochondrial engulfment and degradation (Fig. 5B). Another kinase, leucine-rich repeat kinase 2 (LRRK2), also mediates phosphorylation and subsequent degradation of the Miro protein complex by forming a complex with it. The Parkinson's disease (PD)-linked LRRK2 variant G2019S disrupts phosphorylation of Miro and delays the arrest of damaged mitochondria, subsequently perturbing the mitophagy pathway (Hsieh et al., 2016). After autophagosome–lysosome fusion, the lysosomal compartments are recycled back via a process called autophagic lysosome reformation (ALR) (Chen and Yu, 2017). Here, clathrin, along with KIF5B, drives autolysosomal tubulation (Liu and Klionsky, 2018). In addition, the BORC subunit biogenesis of lysosomal organelles complex 1 subunit 1 (BLOC1S1) facilitates ALR by recruiting KIF5B, KLC1, pleckstrin homology domain-containing family M member 2 (PLEKHM2) and ADP-ribosylation factor-like protein 8B (ARL8B) onto lysosomal tubules. In addition, WHAMM, an actin-associated protein, aids in the process by providing traction forces for tubulation (Wu et al., 2021) (Fig. 2D). Kinesins thus enhance the last stage of autophagosomal fusion by coordinating protein recruitment onto the autophagosomal membrane and bringing the autophagosomal and lysosomal membranes into close apposition. Kinesins are also involved in autophagosome biogenesis, as well as in selective autophagy pathways like mitophagy.

As discussed above, autophagosomes utilise both dynein and kinesin motors, and display initial bi-directional movements, pauses and subsequent directed motility towards lysosomes. Dynein mediates retrograde autophagosome movement, and kinesins (including KIF5B, KIF1B, KIF1A and KIF16B) control anterograde motion (Fig. 3A–C). The recruitment of motors to autophagosomal membranes, which enables autophagosome motility, is followed by a series of events that includes motor inactivation, detachment and reactivation, as well as interaction of motors with adaptors and competitive binding with either adaptors or vesicles, resulting in a tug-of-war that is dictated by post-translational modifications (Fig. 3A) (Hancock, 2014). It has been observed that inhibition of either dynein or kinesin results in reduced autophagosomal motility in both directions in cultured neurons (Maday et al., 2012; Cason et al., 2021), which could be a result of cooperative binding of motors to autophagosomes. The scaffolding protein JIP1 interacts with both KHC proteins and p150^Glued (Fu et al., 2014), which initiates the initial bi-directional movement of autophagosomes as observed in the distal region of axons (Maday and Holzbaur, 2014). Binding of JIP1 to KHC releases the kinesin from its auto-inhibited state, inducing its processivity via a mechanism that is mediated by phosphorylation of S421 in the JIP1 protein (Fu and Holzbaur, 2013). The shift from initial bi-directional movement to uni-directional retrograde motility has been suggested to be the result of coordinated inactivation of kinesin and binding of dynein motor complex, which subsequently promotes the retrograde movement of autophagosomes from the distal tip of neurons. First, JIP1 interacts with LC3 via its putative LIR motif, which has been shown in vitro to suppress interaction of JIP1 with kinesin (Fu and Holzbaur, 2013). Second, the phosphatase MKP1 (also known as DUSP1) has been shown to colocalise with autophagosomes that are positive for JIP1 at the distal tip of neurons, which might maintain JIP1 in a non-phosphorylated state, further reducing its ability to interact with KHC. This, in turn, leaves JIP1 free to interact with dynein, thereby promoting the retrograde movement of autophagosomes towards the cell soma (Fu et al., 2014). Whether the interaction of JIP1 with dynein results in the activation of dynein motor complex is currently unknown and requires further investigation. Thus, here, inactivation of one motor (kinesin) promotes the binding of another motor (dynein) to the autophagosome, regulating its transport in neurons (Fig. 3D). Similarly, JIP3/4 bind to either dynein or kinesin when in a complex with ARF6 or Rab10, respectively, regulating the motility of autolysosomes in primary neuronal cells (Cason and Holzbaur, 2023; Dou et al., 2023). In this context, ARF6 inhibits the binding of JIP3/4 to kinesin, likely due to steric hindrance (Cason and Holzbaur, 2023), whereas it enhances the interaction of JIP3/4 with the dynactin subunit p150^Glued; this shifts the interaction from a JIP3/4–Rab10–kinesin complex to a JIP3/4–ARF6–dynein complex at autolysosomes, possibly as a result of local activation of ARF6. This shift results in displacement of kinesin and promotes the retrograde motility of autolysosomes (Fig. 3E). LRRK2 is known to regulate the movement of autophagosomes in axonal processes (Boecker et al., 2021; Dou et al., 2023), and a constitutively active variant of LRRK2, G2019S, is strongly associated with the pathophysiology of PD (Chang et al., 2021). This hyperactivation of LRRK2 leads to the phosphorylation of Rab10, resulting in abnormal recruitment of JIP4 to autolysosomes, which further enhances the activation and binding of kinesins. This results in an ineffective tug-of-war between the bound kinesin and dynein motors, leading to impaired autolysosomal motility accompanied by frequent pausing (Fig. 3E) (Boecker et al., 2021; Cason and Holzbaur, 2023). Furthermore, overexpression of ARF6 results in release of the JIP3/4–Rab10 complex from kinesin, shifting autolysosomal motility towards dynein-mediated transport; this rescues the motility defects in autophagic flux observed as a consequence of LRRK2 hyperactivation (Cason and Holzbaur, 2023; Dou et al., 2023). Thus, efficient autophagic flux relies on a balanced competition between microtubule-associated motors and perturbations in motor engagement with the autophagosomes. The involvement of motors in vesicular transport has traditionally been investigated in isolation, which prevents a comprehensive knowledge of the routes involved in transport processes, including autophagy. In addition to dynein and kinesin, myosins, which are actin-associated motors, are also essential for efficient autophagic flux, as will be discussed in the next section.

Myosins are ATP hydrolysis-driven molecular motors that move along the actin cytoskeleton. The myosin superfamily is made up of ∼40 genes, which are divided into 12 classes based on molecular architecture (Foth et al., 2006; Fili and Toseland, 2020). Although an extensive body of work has demonstrated a pivotal role for microtubule-mediated transport in autophagic flux, there is also evidence that the actomyosin network is important in modulating the autophagy pathway. In Drosophila, Atg1, which is a crucial autophagy protein involved in the initiation of autophagic flux, initiates the activation of myosin II, promoting autophagosome formation. Atg1-induced autophagy correlates with the reorganisation of filamentous actin, suggesting an important role for actomyosin (Chen et al., 2008; Tang et al., 2011). In this context, Spaghetti squash activator (Sqa) acts as a crucial intermediary, linking Atg1 to myosin II activation. In mammalian cells, UNC-51-like kinase 1 (ULK1) and zipper-interacting protein kinase (ZIPK, also known as DAPK3) – homologues of Atg1 and Sqa, respectively – regulate myosin II activity during starvation-induced autophagy, as indicated by a decrease in myosin II activation upon ULK1 and ZIPK depletion (Tang et al., 2011). Reduction of myosin II also impairs Atg1-mediated cycling of Atg9 during starvation in Drosophila, suggesting that myosin II might act as a motor involved in the trafficking of Atg9 (Tang et al., 2011; Wrighton, 2011) (Fig. 4A). Vesicular transport in the budding yeast Saccharomyces cerevisiae is heavily dependent on the actomyosin network. The cytoplasm-to-vacuole targeting (Cvt) pathway in yeast is a selective autophagy-related pathway that transports vacuolar hydrolases into the vacuole (Lynch-Day and Klionsky, 2010). Precursor of apurinic/apyrimidinic endonuclease 1 (pre-Ape1), which is the precursor of one the vacuolar peptidases, forms large multimeric complexes that are loaded onto Atg9 vesicles by Atg19 and the adaptor protein Atg11 (Yamasaki and Noda, 2017). Myo2 then loads pre-Ape1 complexes onto actin filaments, directing their movement towards the pre-autophagosomal structure (Kumar et al., 2017). Several class I myosins are known to bind to membrane lipids. Myosin 1C (MYO1C) is targeted to the membrane via a pleckstrin homology domain present in its C-terminal tail region (Lu et al., 2015; Nevzorov et al., 2018). Perturbation of MYO1C activity by genetic silencing (using siRNA) or pharmacological means (using pentachloropseudilin, also known as PCIP) results in cholesterol lipid raft mis-trafficking from peripheral membrane stores to membrane compartments near the perinuclear region (Brandstaetter et al., 2014). The discrepancy in lipid trafficking might also have an effect on the integrity of the autophagosomal membrane, as evidenced by a defect in autophagosome–lysosome fusion resulting in the accumulation of autophagic substrates (Brandstaetter et al., 2014) (Fig. 4B). Liquid–liquid phase separation (LLPS) has received considerable attention as a driving force of autophagic flux (Danieli and Martens, 2018; Noda et al., 2020). A recent study has addressed the role of actomyosin dynamics in LLPS of SQSTM1 bodies into larger aggregates, which could plausibly increase their recognition and subsequent degradation by the autophagy machinery (Feng et al., 2022). The condensation of SQSTM1 bodies is affected by disruption of Arp2/3-mediated actin branching or myosin 1D (MYO1D) depletion. MYO1D may actively recruit small nanoscale SQSTM1 particles to branching actin networks (Feng et al., 2022). This recruitment of smaller SQSTM1 bodies onto branched actin tracks increases their proximity, facilitating their eventual coalescence into bigger phase-separated microscale SQSTM1 bodies (Fig. 4C) (Feng et al., 2022).

Myosin VI (MYO6) is a ubiquitously expressed motor that is involved in a plethora of cellular functions (Geisbrecht and Montell, 2002; Warner, 2003). Unlike other myosins, it moves towards the pointed minus end of actin filaments (Sweeney and Houdusse, 2010). MYO6 has been extensively studied for its role in autophagy (reviewed in Tumbarello et al., 2013). MYO6 is known to aid in autophagosomal maturation, and its fusion with the endo-lysosomal compartments via its interaction with TOM1, a component of the ESCRT-0 complex (Tumbarello et al., 2012; Hu et al., 2019), and autophagy adaptors such as optineurin (OPTN) (Sahlender et al., 2005), Tax1-binding protein 1 (TAX1BP1) and nuclear dot protein 52 kDa (NDP52, also known as CALCOCO2) (Morriswood et al., 2007) (Fig. 4D). Interaction of MYO6 with TOM1 on the endo-lysosomal compartments via its WWY motif facilitates the fusion of autophagosomes with endo-lysosomal compartments, likely by bringing the two compartments into close proximity (Tumbarello et al., 2012; Hu et al., 2019) (Fig. 4D). By acting as a bridge adaptor, MYO6 can simultaneously bind to autophagy receptors through its RRL motif and to TOM1 via its WWY motif (He et al., 2016; Hu et al., 2019). The C-terminal ubiquitin-binding domain of MYO6 (MyUb), which contains the RRL motif, binds to K63, K11 and K29 ubiquitin linkages. A hydrophobic patch on helix two of the MyUb domain crucially binds adaptors like OPTN and NDP52 (He et al., 2016). The autophagy adaptors TAX1BP1, OPTN and NDP52 share overlapping ubiquitin- and myosin-binding regions (Sundaramoorthy et al., 2015; Hu et al., 2018). TAX1BP1 interacts with MYO6 via its zinc finger motifs UBZ1 and UBZ2, with UBZ2 being pivotal for interactions (Hu et al., 2018). Accordingly, point mutations in these motifs reduce the interaction between TAX1BP1 and MYO6 (Hu et al., 2018).

Xenophagy, the autophagy-mediated degradation of intracellular pathogens, is an important immune defence mechanism (Sharma et al., 2018). The autophagic machinery and cytoskeletal remodelling during infection have been extensively explored using intracellular pathogen models (Kishi-Itakura et al., 2020). Such studies have shown that xenophagy is coordinated through the recognition of ubiquitylated pathogens by autophagy adaptors such as TAX1BP1 (Tumbarello et al., 2015) and NDP52 (Verlhac et al., 2015; Viret et al., 2018), which initiate the assembly of the autophagy machinery for lysosomal targeting and degradation of the pathogen (von Muhlinen et al., 2012; Sorbara and Girardin, 2015; Kimmey and Stallings, 2016). In this context, MYO6 associates with NDP52, mediated by the C425 residue of NDP52, and thus aids in the completion of the autophagic process during pathogen invasion, plausibly by facilitating autophagosomal maturation and fusion with lysosomes, as mentioned above (Fig. 5A) (Verlhac et al., 2015). Interestingly, MYO6 also aids in the entry of pathogenic Salmonella bacteria during their invasion (Brooks et al., 2017) (Fig. 5A). The Salmonella effector proteins SopE and SopB modulate the activity of MYO6 for its cellular uptake. During invasion, SopE activates host Rho GTPases, which then activates p21-activated kinase (PAK) proteins, causing downstream actin reorganisation and consequent recruitment of MYO6 to the macropinocytic cup. SopB together with MYO6 then activates phosphoinositide 3-kinase (PI3K) signalling at the entry point, which alters the lipid and protein content of the membrane, facilitating pathogen entry (Brooks et al., 2017).

Loss of MYO6 results in the accumulation of dysfunctional mitochondria (Kruppa et al., 2018). Additionally, upon induction of mitophagy, MYO6 localises to the surface of mitochondria; this localisation is mediated via the MYO6 MyUb domain and RRL motif, and occurs in a parkin-dependent manner, as C431S mutation of parkin diminishes the translocation of MYO6 to the mitochondrial surface (Kruppa et al., 2018). Mitophagy is marked by two distinct actin waves around the damaged mitochondrial fragment; the first transient wave of actin filament formation during the initial stages of mitophagy enhances mitochondrial fission (Li et al., 2015; Kruppa et al., 2018). Myosin II aids in this process by providing the force required for constriction of the mitochondrial surface during fission (Korobova et al., 2014). The second actin polymerisation wave is facilitated by the recruitment of MYO6 to the mitochondrial surface. Formation of an actin cage around the damaged mitochondrion results in ‘bite-sized’ mitochondrial pieces. Such an isolation of damaged mitochondrial fragments prevents their fusion with the healthy mitochondrial pool, promoting their capture by the autophagic machinery (Kruppa et al., 2018). Another unconventional myosin, MYO19, is involved in regulating mitochondrial morphology and docks onto the OMM, either through an interaction with Miro proteins, or directly through its tail domain (Oeding et al., 2018). As Miro has a role in PINK1-mediated mitophagy (Wang et al., 2011), loss of Miro from the OMM upon proteasomal degradation results in the detachment of kinesin and MYO19, and arrests movement of damaged mitochondria (Wang et al., 2011; Oeding et al., 2018; Baltrusaitis et al., 2023) (Fig. 5B). This arrest of the movement of damaged mitochondria, accompanied by bursts of actin filament formation around them, promotes their isolation, eventually triggering recruitment of autophagic machinery, which culminates in mitophagy. Overall, the structural plasticity and diversity in the interacting partners of myosin motor proteins brings different components of autophagic flux into close proximity, facilitating the maturation of autophagosomes and their subsequent fusion with lysosomes.

As discussed above, autophagic flux is regulated by several different motors, eventually culminating in the efficient degradation of cellular cargoes. Accumulation of misfolded and aggregated proteins is a classic signature of neurodegenerative disorders, and autophagy has been extensively investigated for its neuroprotective role (Menzies et al., 2015; Guo et al., 2018). Perturbation of motor components can also give rise to pathogenesis. For instance, the expression of p150^Glued has been shown to be downregulated in motor neurons of individuals with sporadic amyotrophic lateral sclerosis (ALS) (Ikenaka et al., 2013). In C. elegans, loss of function of DNC-1 (the p150^Glued homologue) leads to defective autophagosome accumulation in motor neurons, causing axonal degeneration and severe locomotion deficits (Ikenaka et al., 2013). Furthermore, mutation or downregulation of motor protein components has been found to perturb vesicular transport in axons (Gunawardena et al., 2014). Alzheimer's disease (AD) is marked by an accumulation of amyloid β (Aβ) aggregates. Expression of dynein intermediate chain has been observed to be downregulated in the brains of people with AD and in transgenic mice models of AD, accompanied by impaired autophagic flux (Zhou et al., 2020). Accumulation of aggregates of the 42-amino-acid form of Aβ (Aβ42) at the axonal terminals as observed in AD disrupts the interaction of SNAPIN with the dynein intermediate chain, thereby impairing retrograde movement of autophagic vesicles and increasing autophagic stress in neurons (Fig. 1) (Tammineni et al., 2017). In ALS and stressed neurons, NDEL1 phosphorylation by hyperactive cyclin-dependent kinase 5 (CDK5) increases the affinity of dynein for microtubules, blocking ATP-driven dynein motility. This prolongs the attachment of dynein to microtubules and decreases dynein-mediated motility of autophagosomes, thereby compromising autophagic flux (Klinman and Holzbaur, 2015). Additionally, ALS-linked variants in the autophagy adaptor OPTN have been shown to limit interaction of OPTN with MYO6, thereby adversely affecting autophagy-mediated clearance of protein aggregates (Sundaramoorthy et al., 2015). Tar DNA-binding protein 43 (TDP43) fragments, which are crucial markers in ALS and frontotemporal dementia, localise to mitochondria, where they trigger mitochondrial depolarisation and subsequent parkin recruitment, thus promoting mitophagy (Jun et al., 2020). Myosin IIB (which has a heavy chain encoded by MYH10) colocalises with aggregates of TDP25 (a C-terminal fragment of TDP43), indicating a role in mitochondrial degradation of pathogenic TDP25 fragments (Jun et al., 2020). Genetic silencing of myosin IIB results in accumulation of damaged mitochondria, affecting neuronal viability and mitophagy in mouse cortical neurons (Jun et al., 2020), which is suggestive of an involvement in the pathophysiology of ALS. Thus, dysregulation of motor function in disease scenarios disrupts autophagic flux, and identification of pathogenic variants in motor proteins that lead to perturbation of the autophagic pathway would provide a better understanding of disease pathophysiology in the context of autophagy.

As discussed here, dynein, kinesins and myosins coordinate the movement of the autophagic machinery, influencing autophagosome biogenesis and maturation, as well as acidification and fusion. In addition, whereas adaptor proteins initiate or facilitate the association between motors and the autophagosome membrane, the recruitment and assembly of the motors themselves are modulated by post-translational modifications and binding to LIR motifs. Adaptors that bind to different motor proteins in response to specific cues govern their activation or detachment, subsequently modulating the autophagic flux. As discussed above, some adaptors, such as RILP (Khobrekar et al., 2020), BORC (Jia et al., 2017) and ORP1L (Wijdeven et al., 2016), associate with both autophagosomal and lysosomal membranes. Therefore, studying the specific contribution of adaptor proteins and the post-translational modifications that affect their binding in isolation is experimentally challenging. Advances in microscopy techniques coupled with computational simulations might help to address and identify the factors that fine-tune the different intricate steps of autophagic flux. Here, we have discussed the movement of autophagosomes as a consequence of motor activity, but it is reasonable to assume that the association of some motors could also act as a molecular brake, allowing the loading or unloading of proteins important for biogenesis, maturation or fusion of autophagosomes as they traverse the complex cytoskeletal network, and identification of motors in this regard would help us to understand the nuances of the pathway. De-tyrosination of microtubules has been shown to reduce the movement of autophagosomes and lysosomes, increasing the likelihood of contact and subsequent fusion, but more research is needed in this area, which should also take into account the interplay between cytoskeletal components and associated motors. Furthermore, dysregulation of motor proteins or the presence of motor protein pathogenic variants has been documented in several neurodegenerative disorders, as reviewed extensively elsewhere (Gunawardena et al., 2014; Brady and Morfini, 2017). Investigating the perturbations in autophagic flux in the context of such dysfunctional motor proteins might help to better understand disease pathogenesis and progression, as well as to develop therapeutic strategies. Numerous questions surround the role of motors in autophagic flux, including factors affecting the loading and unloading of effector proteins from the motor complex. It is still unclear whether and how the maturation status of autophagosomes influences this process. While microtubules, actin and their motors regulate autophagic flux, what biases the occurrence of specific events on a cytoskeletal track is still unknown. Comprehensive experimentation is thus required to ascertain the signalling cues and motors that could prompt changes in cytoskeletal tracks during the course of the autophagic pathway.

We are thankful to the members of the Autophagy laboratory (Jawaharlal Nehru Centre for Advanced Scientific Research), especially Irine Maria Abraham, Mallika Bhat and Aparna Hebbar for their critical review of the manuscript. We thank those researchers who have contributed to the area and respectfully apologise to those whose contribution could not be recognised owing to space limits. The figures were generated using BioRender.com.