Spastin is a highly conserved protein whose complex functions are attributed to its multiple modular domain structure and different isoforms. Full-length spastin consists of four functional domains: The hydrophobic domain (HD) situated between residues 1-87, the MT-interacting and trafficking domain (MIT) spinning (residues 116-194), the MT-binding domain (MTBD) situated in residues 270-328 and the adenosine triphosphatases associated with diverse cellular activities (AAA) domain spinning (residues 342-599). The HD has a high affinity for the membrane structure and determines the subcellular distribution of full-length spastin. MIT is essential for spastin's interaction with endosomal sorting complexes required for transport (ESCRT)-III proteins responsible for cytokinesis completion, endosomal trafficking and other related cellular processes. Similarly, MTBD mediates spastin's interaction with tubulin, which is a prerequisite for severing. By contrast, the AAA domain mediates hexamer formation and catalyzes the severing of MTs (3,30,31).

M1 and M87 are localized in different tissues and subcellular locations. M1 is largely distributed in the cytosol due to the strong nuclear export signal (NES) sequence of 50-87, while M87 is distributed in the whole cell. Spastin also has a nuclear export signal sequence situated in residues 195-204 that spans the MIT and exon 4 region and a nuclear localization signal (NLS) sequence (32,33). Fig. 1 presents the spastin domain architecture and highlights the NES and NLS locations. Although spastin is found in the nucleus, its function in the nucleus remains elusive. A recent study postulated that extracellular histone 2B is able to recruit spastin at the midbody during the late stages of abscission (34). By using proteomics analysis, our group also identified the interaction between spastin and numerous histone isoforms (not published). Since histones are vital for the morphology of chromosomes, it may be inferred that spastin is involved in the consolidation of chromosomes or in the regulation of protein transcription. However, this assumption remains to be experimentally validated.

Studies that determined that spastin's structure binds with peptides have revealed its severing mechanism. Spastin forms hexamers through its AAA region in the presence of ATP. The hexamer structure has a central spiral ring structure that is positively charged. The loop subsequently engages with the negatively charged C-termini of tubulin, thereby mechanically extracting tubulin from the lattice using the energy provided by ATP hydrolysis (35-37). Mechanistically, spastin forms open spirals, as the subunits are mutated in an ATP-binding manner. The hexamer moves up to form a closed structure when the spastin structure with adenosine diphosphate-beryllium fluoride represents a 'broken spiral' in which the sixth AAA is in the apo state (36).

Questions to be addressed are whether MT severase is able to depolymerize or amplify MTs, and how severase controls MT length and mass. Recent studies provide answers to these questions and postulate that spastin acts as an amplifier of total MT length and mass (29,38). However, spastin's potential to induce MT regrowth or catastrophe (i.e. the conversion of a growing microtubule into a shrinking one depends on the MT concentration (29,39,40). To date, two major theories for MT rescue after severing action by spastin have been postulated. According to the first theory, spastin grips the C-terminus of tubulin in an adenosine triphosphate (ATP) hydrolysis manner, followed by refilling the damaged lattice with guanosine triphosphate-tubulin. The MTs severed by spastin are then stabilized and act as seeds to mediate MT regrowth (38). In the second theory, spastin concentrates at the newly generated plus end of MTs to block depolymerizing activity, similar to other MT-associated proteins (MAPs), such as calmodulin regulated spectrin associated protein family member 3. The key evidence for this theory is that the stimulation of increasing the MT length and mass is independent of ATP hydrolysis (29). In general, both theories indicate that spastin is a dual-function enzyme that promotes MT regrowth or catastrophe and provide novel insight regarding its severing activity during physiological conditions.

Over the past decades, spastin was thought to cut long MTs into short MTs to regulate MT dynamics (3). A question may be put forward regarding how MTs behave after severing and how severed MTs contribute to cellular functions. The observations discussed above provide an intuitionistic model and form a foundation to understand MT behavior after severing and provide comprehensive insight into numerous molecular and cellular mechanisms of spastin. The cellular functions of spastin introduced below are based on this model.

Spastin is an MT-severing enzyme containing MTBD and AAA domains with adequate severing activity. In addition, the HD and MIT domains interact with various proteins and recruit spastin to specific subcellular localizations to sever the spatial MTs. Understanding the domain-based interactions may provide insight into spastin-severing activities at the cellular level. The cellular functions of spastin based on its domain interactions are described below and illustrated in Fig. 2.

M1 spastin has an HD region that forms a hairpin loop that embeds into the membrane lipid bilayer, causing spastin to be wedged into the membrane. Three other HSP-related proteins, receptor expression enhancing proteins (REEPs), alastins and reticulons, also have a similar hairpin loop and are also recruited in the ER. Thus, these proteins shape the tubular ER (12,41).

Spastin shapes the ER network and affects Ca²⁺ handling (42). However, the mechanisms of spastin in ER shaping remain to be explored, although numerous HSP-related proteins involved in ER shaping have been largely investigated. The ER sheets in Drosophila melanogaster neurons are larger and have much fewer tubules when the dominant-negative variant of the MT-severing protein spastin is used to induce MT stabilization. Furthermore, the store-operated calcium entry (SOCE) process is inhibited, affecting the flies' flight ability. However, promoting MT dynamics by applying the MT-destabilizing drug vinblastine restores flies' ER morphology, SOCE and flight ability (42). These results suggest that the MT dynamics provided by spastin have a decisive role in forming ER tubules. An important question is whether the severing activity mediated by spastin affects ER morphology. Currently, there is no direct evidence of how spastin affects MTs surrounding the ER. As discussed above, MT dynamics have an important role in ER morphogenesis. There are two conditions under which MTs direct the fate of ER morphology: i) Kinesin-MT-mediated transport drives ER tubules; and ii) the ER tubules move along the growing MT by docking onto the plus-end through a tip attachment complex (43-45). Combined with the innovative hypothesis of the severing model, it is reasonable that spastin promotes ER tubule formation by increasing MT fragments with more MT-plus ends (38). Furthermore, the increase in the total length of the MT potentially leads to more kinesin loading, resulting in more efficient transport.

Another potential mechanism of spastin-mediated ER shaping is through interactions with protrudin. Protrudin is also an ER-resident protein and regulates the sheet-to-tubule balance (46,47). Kinesin-related protein 5 (KIF5) belongs to the kinesin motor family, which drives long-distance cargos along the MT 'tracts' in the anterograde direction using the energy generated by ATP hydrolysis (48,49). Protrudin interacts with KIF5 (16,49), thus influencing ER morphology in an MT-dependent mode. Furthermore, protrudin extracts lipids from the ER by interacting with PDZ domain-containing protein 8 (PDZD8), thus proving that these proteins coordinate with each other to influence ER morphology (50,51). M1 spastin is able to interact with protrudin, thereby regulating ER morphology, probably by modulating the MT dynamics necessary for protrudin-mediated kinesin motor activities (47,52).

Overall, spastin contributes to ER tubule formation and inhibits the SOCE process. Although the clear attribution of the ER sheet-tubule balance to cellular functions remains largely elusive, it is now known that Ca²⁺ signaling is affected. Furthermore, casein kinase 2 (CK2) is also activated upon spastin deficits (19). Whether other signaling pathways are activated or inhibited when the ER sheet-tubule balance is affected by spastin remains to be explored. Given that the ER is distributed throughout neurons, how spastin affects ER shaping, the calcium pathway and other signaling pathways in neural development and disease is an important issue in understanding the physiological and pathological process, which still requires further investigation. Answers to these questions will further elucidate the pathogenesis of HSP.

LDs are highly dynamic and complex organelles responsible for the assembly and storage of neutral lipids (53). They are specialized compartments of the tubular ER, from which they are generated in a stepwise process (54). Lipids from LDs are transported to the cell membranes and regulate intracellular signals (55,56). Various studies postulated that spastin anchors to LDs through its hydrophobic domain (57-86). Spastin depletion or overexpression of the dominant-negative variant impairs the formation of LDs in the nerves and skeletal muscles of fruit flies, while overexpression of spastin promotes the formation of LDs (11,57). These results suggest that M1 spastin is associated with the formation of LDs. Although there is no direct evidence indicating how spastin affects the formation of LDs, it may be hypothesized that the newly formed LDs may be ER shaping products, as LDs are specialized compartments of the tubular ER. However, how LDs function remains largely elusive. Whether LDs transported to the cell membrane or intracellular signals are changed still requires further research. In addition, it has not been demonstrated that dysfunction of LDs is involved in HSP pathogenesis. Whether LD abnormalities are a common defect or rather a readout of different dysfunctional pathways remains elusive. Therefore, LD function should be investigated in patients with HSP to further the understanding of HSP and offer novel therapeutic opportunities.

The MIT domain is adequate for binding two ESCRT-III proteins: increased sodium tolerance-1 (IST1) and charged multivesicular body protein 1B (CHMP1B) (17,58,59). This interaction allows the recruitment of spastin to specific subcellular localizations, thus regulating spatial MT dynamics. The ESCRT-III machinery is vital for membrane scission in numerous membrane remodeling processes, including viral budding from the cell surface, budding of endosomes to form vesicles of late endosomes and cytokinetic abscission (14,60,61). In addition, the ESCRT-III proteins recruit spastin to regulate cytokinetic abscission, endosomal trafficking and fatty acid (FA) metabolism.

The role of spastin during cytokinesis has been extensively elucidated and reviewed in various studies (30,58,62). In the present review, the process is briefly discussed. During the late stages of the anaphase, the nuclear envelope, which is frequently intersected by spindle MTs, is progressively reassembled around the reforming daughter nuclei. In this process, spastin recruitment by IST1 helps to remove the remaining spindle MTs for nuclear envelope sealing. Failure of this process leads to the accumulation of DNA damage.

Similarly, the daughter cells remain connected by an inter-cellular bridge with the midbody ring overlapping with MTs at its center during the late stages of abscission. CHMP1B thus recruits spastin to remove the MTs in the intercellular bridge. Depletion of spastin results in abscission failure, causing the daughter cells to remain connected by elongated MT-filled bridges.

Endosomes are sorting organelles that mediate the internalization of extracellular transmembrane molecules and ligands. Endosomes are categorized into early, recycling and late endosomes based on their internalization stage. Internalized molecules may be recycled into the cell membrane using recycling endosomes or degraded using late endosomes (63,64). Endosomal trafficking regulates plasma membrane receptor concentrations, which are critical extracellular responses. The ESCRT machinery recruits spastin into endosomes to regulate endosome fission and trafficking pathways (65,66).

Existing evidence suggests that spastin regulates endosome trafficking in two ways. i) Spastin may promote the tubulation and fission of endosomes, thereby inhibiting their pathway to the lysosome (17,18,67). This process is accomplished by the actin-dependent and MT-dependent pulling force and membrane scission by dynamin (68,69). In this process, the ESCRT-III protein increased sodium tolerance 1 recruits spastin into the endosome through the interaction of its MIT domain. Previously, severing activities mediated by spastin were thought to generate more MTs (38), thus ensuring that MTs were loaded with more dynein to enhance the driving force, thereby resulting in an enhancement of endosome tabulation and fission. However, this point of view lacks experimental evidence. Of note, a recent study provided direct evidence that the ER and endosome membrane contact sites (MCSs) are vital for endosome tubulation and fission (18). Overall, a better explanation for the tubulation and fission of endosomes may be the result of spastin-mediated ER reorganization. The changed ER network reconstructs the MCSs and contributes to promoting endosome tabulation and fission, while how spastin affects the MCSs between the ER and endosomes remains to be elucidated. ii) Spastin is able to regulate endosome trafficking by antagonizing protrudin, thereby regulating endosome trafficking in a generalized manner (16). Protrudin is an ER-resident protein that interacts with Rab7 and phosphatidylinositol 3-phosphate. These interactions facilitate the transfer of KIF5 to the late endosomal motor adaptor FYVE and coiled-coil domain containing 1, thus promoting endosomal transport to the plasma membrane (70,71). Although the interaction of spastin with KIF5 remains to be proven, studies have indicated that it is able to interact with protrudin (52). Regardless of whether it is direct or indirect, this interaction may recruit spastin to KIF5, resulting in MT 'railway breakage', which blocks the late endosome toward the cell periphery.

To date, the process of endosomal behaviors mediated by spastin has remained largely elusive. As described above, spastin promotes endosome tabulation and fission to inhibit the lysosome pathway. Spastin regulates endosome trafficking in a generalized manner. Of note, the destination of the endosome cargo remains to be determined. Taking BMP receptors as an example, it is unlikely that the BMP receptor retransports to the cell periphery, as its level on the membrane increases upon deficits of spastin. Furthermore, spastin inhibits the lysosomal pathway, and it remains to be determined how cargo is metabolized in the cell. In addition, the specific cargos transported through spastin-mediated endosomal trafficking remain to be identified. Answering these questions may lead to the elucidation of the specific mechanisms of endosomal trafficking mediated by spastin and pave the road for the development of therapies for HSP. With the rapid progression of biological analysis techniques, it may be promising to identify endosome components and trace these cargos, thus elucidating the endosomal trafficking mediated by spastin and revealing the molecular and cellular pathogenic mechanisms leading to HSP.

Membrane contact sites between LDs and peroxisomes have been observed for decades. However, the molecular underpinning that connects these organelles remains largely elusive (72,73). Peroxisomes harvest FAs from LDs for energy production and maintain cellular homeostasis by recycling lipids and protecting cells from oxidative stress and damage (74,75). It has been postulated that spastin tethers to peroxisomes through the interaction between the MIT domain and the peroxisome resident protein ATP-binding cassette sub-family D member 1 and mediates FA flux from LDs to peroxisomes (13). Furthermore, overexpression of M1 spastin reduces peroxidized lipid accumulation in cells exposed to oxidative stress. On the other hand, the dominant-negative variant mutant spastin (K388R) is not able to promote LD peroxisome tethering and fails to lower peroxidized lipid levels in LDs (13,76). In addition, HSP patient-derived cells with mutations in spastin also indicated impaired peroxisome movement and distribution. These observations have established an intuitionistic model to describe the mechanism of cellular organelle crosstalk between LDs and peroxisomes and the process of FA trafficking and metabolism. However, there is still a lack of evidence that FA impairment is involved in patients with HSP, which requires further validation. Overall, this mechanism strongly suggests that LD peroxisome tethering has an important role in the neuropathology of HSP and may provide an opportunity of targeting FA metabolism for HSP therapy.

Neurogenesis is active during the embryonic period, as it is responsible for producing new neurons. In this period, neural stem and progenitor cells go through symmetric proliferative division until a sufficient population of neural stem cells (NSCs) is produced and subsequently develops into neurons (88,89). It has been reported that both M1 and M87 spastin is enriched in proliferative NSCs and embryonic brain tissues. Spastin depletion using short hairpin RNA led to a decrease in NSC proliferation and neuronal lineage differentiation. Of note, this phenomenon was partly reversed using the MT-destabilizing drug nocodazole, indicating that NSC proliferation requires the dynamic MTs provided by spastin (90). This observation suggests that spastin participates in neurogenesis by promoting MT dynamics. As the Sonic hedgehog (Shh) signaling pathway is responsible for regulating and coordinating cellular growth and development in the embryo, this study also revealed that the Shh signaling pathway was inhibited by spastin depletion (83). However, the inhibition mechanism of the Shh signaling pathway by spastin depletion should be further evaluated.

Growth cones at the end of an axon mediate its extension and turning after axon differentiation. They sense the extracellular environment and regulate axonal development and turning. They also have a vital role in intercellular signaling transmission (91,92). In general, axon development undergoes three stages: Protrusion, engorgement and consolidation. In the growth cone, the MTs are highly dynamic in contrast to the axonal draft. Highly dynamic MTs contribute to the high sensitivity of the cones to guidance cues. Furthermore, filopodia and lamellipodia consolidation also requires the mechanical force provided by dynamic MTs (93,94).

MTs form highly stable and polarized bundles in axons, providing structural support and 'railways' to guide MT-dependent transport. During axon branching, the cytoskeleton must be remodeled, which requires actin filament accumulation to form axonal filopodia. Soon after that, the MTs then penetrate the actin-rich filopodia, allowing the maturation and continuous extension of the branches (95,96). This phenomenon proves that MT dynamics are a vital factor during axon branching and formation.

As described above, M1 spastin is distributed in the cytosol and mainly at the membrane sites, such as the ER and LD. However, M87 spastin is distributed in the whole cell and was reported to be enriched in the growth cone (97), and part of the axonal branch points are likely to generate new branches (24,98). There is abundant evidence suggesting that spastin contributes to axonal branching and elongation. Furthermore, the new MT severing theory suggests that spastin and other severing enzymes may act as amplifiers to generate more MT seeds during regrowth (29,38). Cumulative evidence has indicated that neurons are sensitive to severing activities. Appropriate severing by spastin and katanin, which are highly expressed in neurons during development, promotes axonal elongation (77,99-101). Of note, overexpression of spastin significantly promotes the formation of axonal branches, whereas katanin does not. The accumulation of spastin in the branch points strongly suggests that branch formation requires the accumulation of actin filaments and MTs (100). Thus, it may be inferred that spastin potentially affects the actin cytoskeleton by interacting with related actin-associated proteins through MIT and MTBD domains. However, the actual mechanism involved in spastin-mediated remodeling of the actin cytoskeleton should be further evaluated.

The formation and maturation of dendritic spines contribute to the connections between neighboring neurons, which is vital for responding to novel stimuli throughout neural development and adulthood (102,103). In the last decade, research on spastin has primarily focused on axonal behavioral mechanisms. It is worth noting that M87 spastin is distributed in the entire cell compartment, including dendrites and dendritic spines (26,100). How M87 spastin affects dendrite behaviors remains largely elusive.

A decade ago, spastin was reported to regulate MT dynamics during the elimination of neuromuscular junctions in fly knockout models (27). Recent studies further indicated that spastin is closely related to learning and memory. Furthermore, it participates in the formation and maturation of dendritic spines and regulates the transport of the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) subunit GluA1 (26) and GluA2 (25). Lopes et al (25) established a spastin-knockout mouse model and determined that the formation and maturation of dendritic spines were inhibited upon spastin depletion and that KIF5-mediated GluA2 transport was also blocked upon spastin depletion. AMPAR trafficking may induce long-term potential or long-term depression, thereby regulating the maturity of dendritic spines (104,105). As such, spastin potentially regulates dendritic spine morphology by altering MT-dependent transport. However, another study suggested that overexpression of spastin leads to an increase in the proportion of immature dendritic spines and leads to a decrease in GluA1 membrane expression (26). That study also indicated that GluA1 trafficking is closely related to spastin-mediated endosomal trafficking. The different effects of spastin on GluA1 and GluA2 trafficking may be attributed to the different trafficking pathways. GluA1 membrane trafficking is primarily dependent on the endocytosis or exocytosis of endosomes, while GluA2 binds directly to glutamate receptor interacting protein (GRIP) and is then subjected to KIF5-mediated transport (106). Of note, GRIP is able to bind the GluA2 subunit but not to GluA1 or GluA4 (107). Furthermore, neural activities are sensitive to the dosage of spastin levels, and it still requires to be elucidated how spastin works during synapse formation under physiological conditions. Collectively, these studies postulate that spastin-mediated cargo delivery is closely related to the formation and maturation of dendritic spines by regulating AMPAR trafficking. Furthermore, studies have also confirmed that KIF5 is also able to transport GABA receptors toward the dendritic spine (108). However, whether spastin is able to regulate this process requires further exploration.

Although the actin cytoskeleton is the primary regulator of spine morphology, MTs may also penetrate the spine transiently. Invasive MTs are able to transport relevant cargo proteins, such as p140Cap, to promote the polymerization of actin filaments (109). It may be assumed that spastin contributes to actin polymerization by interacting with actin-binding proteins, as it acts as an amplifier to generate more MT seeds during regrowth. However, this assumption should be verified.

Axon transport is essential for regulating axon composition, neural development, maintenance and survival (110,111). The process is mediated by motor molecules (kinesin or dynein) to deliver different cargos, such as organelles, signal molecules and RNAs, along the MTs in an ATP-dependent manner. For instance, the kinesin motor mediates the anterograde transport responsible for delivering RNAs, proteins and organelles toward the growth cones and synapses. Similarly, dynein-mediated retrograde transport delivers nutrient factors and proteins that require degradation by the cell body, thus responding to the extracellular environment (112-114). This process is highly dependent on the MTs. As such, spastin is potentially involved, as it alters the MT dynamics and arrays.

Numerous studies have reported spastin involvement in MT-mediated cargo transportation. Of note, spastin mutations identified in patients with HSP lead to axonal swellings and axonal transport abnormalities (19,20,115). To date, spastin has also been reported to regulate the transportation of mitochondria (20,116), peroxisomes (117,118), amyloid precursor protein (20), vesicle-associated membrane protein 7 (VAMP7) (119) and BMPII receptors (120). In particular, M1 spastin is crucial during fast axonal transport using isolated squid axoplasm but not M87 (6,19). Consistent with these observations, M1 spastin has a pivotal role in VAMP7 transport in cortical neurons but not M87 (121). However, Connell et al (16) postulated that both M1 and M87 antagonize KIF5-mediated BMPII transport to the cell membrane. These studies confirm that both M1 and M87 spastin are involved in axonal transport despite the differences in each cargo transportation mechanism and their transport direction.

M1 and M87 are different in the process of axon transportation. M1 has an additional hairpin loop domain that determines its location in the ER and LDs (32). It was also reported that M1 spastin is more efficient in promoting endosomal tubule fission. (17). Thereafter, M1 spastin regulates cargo transportation in ER-dependent pathways, while M87 primarily directs cargo transport by altering MT dynamics. Furthermore, M1 spastin is able to potentially recruit M87 from the cytosolic pool, as spastin requires hexamerization to sever MTs (28). Therefore, it may be hypothesized that M1 is more efficient than M87 in axonal transport via ER-dependent pathways.

As described previously, M1 spastin contributes to tubular ER formation by regulating ER-surrounded MT dynamics. The ER is distributed throughout the whole cell and is conducive to the physical continuity of axons (122). The ER and other organelle membrane contact sites allow communication with each other, thereby regulating intracellular membrane trafficking (123,124). For instance, Allison et al (18) revealed that spastin promotes endosomal fission and regulates its trafficking using ER-endosome MCSs. Furthermore, ER-resident proteins, such as protrudin and PDZD8, potentially contribute to endosome trafficking (50) after ER morphology is altered by M1 spastin. ER morphology alteration is closely associated with calcium flux (125,126). The process is required to maintain axonal transport (126). Of note, spastin also potentially influences the calcium signaling pathway in axons, thus indirectly contributing to normal axonal transport. As such, the M1 spastin-mediated mechanism involved in the regulation of axonal transport may be ER-dependent, which requires further experimental confirmation.