Axon Pathfinding

Among the first HSP mutations described were in the L1CAM gene, encoding a cell surface glycoprotein of the immunoglobulin (Ig) superfamily. Loss-of-function mutations in L1CAM are implicated not only in X-linked, early-onset, complicated HSP (SPG1), but also in other X-linked syndromes including MASA (for mental retardation, aphasia, shuffling gait, and adducted thumbs), hydrocephalus, and agenesis of the corpus callosum ( Jouet et al. 1994, Weller & Gärtner 2001). Each of these disorders displays clinical and pathological evidence of corticospinal tract impairment; the disorders are considered together along a disease spectrum known as L1 disease or CRASH syndrome (for corpus callosum hypoplasia, retardation, adducted thumbs, spastic paraplegia or shuffling gait, and hydrocephalus) (Soderblom & Blackstone 2006).

The L1CAM protein is more than 1200 amino acid residues in size, with a large extra-cellular segment harboring 6 Ig-like domains and 5 fibronectin type III domains, a single transmembrane domain, and a short cytoplasmic tail. L1CAM participates in a complex set of extracellular and intracellular interactions, binding not only other L1CAM molecules but also a host of extracellular ligands—including other cell adhesion molecules, integrins, and proteoglycans—as well as intracellular proteins such as ankyrins. Disease mutations are found throughout the protein, and partial or complete loss of L1CAM function seems critical for the L1 disease phenotype. In L1CAM null mice, the corticospinal tracts are abnormal, arising from a conspicuous failure to decussate within the medulla (Dahme et al. 1997, Cohen et al. 1998).

How does this pathfinding defect occur? In the developing CNS, L1CAM associates with neuropilin-1 (Nrp1), which itself interacts with Plexin-A proteins to form the Semaphorin3A (Sema3A) receptor complex. Upon Sema3A binding to Nrp1, L1CAM and Nrp1 are cointernalized in a L1CAM-dependent manner. Sema3A is a repulsive guidance cue released from cells in the ventral spinal cord to steer corticospinal neurons away from the midline spinal cord/medullary junction, and L1CAM mutations may affect Sema3A signaling when axons are crossing the midline by interfering with receptor internalization and signaling at growth cones (Castellani et al. 2004). In fact, the association of Nrp1 with L1CAM mediates the activation of a focal adhesion kinase-mitogen-activated protein kinase pathway controlling a critical aspect of the repulsive behavior, the disassembly of adherent zones in growth cones and their subsequent collapse (Bechara et al. 2008). This compelling role of L1CAM in axon pathfinding during development is consistent with the early onset of SPG1.

Myelination

A distinguishing feature of axons in the central and peripheral nervous systems is an insulating myelin sheath, a specialization important for increasing the speed of electrical impulse propagation. Schwann cells supply myelin for peripheral neurons, whereas oligodendrocytes myelinate axons of CNS neurons. Spastic paraplegia as a manifestation of abnormal myelination in the CNS is not uncommon; for example, this occurs in multiple sclerosis and a variety of acquired and inherited leukodystrophies. Mutations in the PLP1 gene encoding the tetraspan integral membrane proteolipid protein (PLP) and its smaller DM20 isoform give rise to two major diseases along a clinical spectrum: a pure or complicated HSP (SPG2) and the generally much more severe Pelizaeus-Merzbacher disease (PMD) (Inoue 2005).

Although PLP and DM20 are the major protein constituents of CNS myelin (~50% of the total protein), PLP1 duplications paradoxically cause more severe disease than do deletions, whereas complete absence of PLP/DM20 is typically associated with SPG2 or mild presentations of PMD. Plp1 null mice in particular have been widely studied as a model for SPG2. Unexpectedly, in these mice the myelin sheath maintains its normal thickness, though with subtle anomalies of the intraperiod lines. In the underlying axons, anterograde transport is impaired, and cargoes undergoing retrograde transport become stuck at distal juxtaparanodal regions (Edgar et al. 2004, Gruenenfelder et al. 2011). It seems reasonable to postulate that oligodendrocytes modulate the activity of motor proteins involved in intracellular cargo transport via signaling cascades in the underlying axon and that this modulation is sensitive to PLP/DM20 (Gruenenfelder et al. 2011).

Mutations in a more recently identified HSP gene similarly define a disease spectrum comprising HSP and PMD-like disease where cell-cell communication is altered. The slowly progressive, complicated SPG44 is caused by homozygous mutations in the GJC2 gene encoding connexin 47 (CX47). Connexins (typically numbered based on predicted molecular weight) are oligomeric proteins forming gap junction channels, which establish connections between apposed cell membranes to permit the intercellular diffusion of ions and small molecules (typically <1000 Da). CX47 forms connections between astrocytes and oligodendrocytes in concert with CX43. Because CX47/CX43 heterotypic channels appear essential for the maintenance of CNS myelin, alterations in CX47 that result in CX47/CX43 channel dysfunction likely underlie SPG44 (Orthmann-Murphy et al. 2009).

A third HSP with a compelling link to dysmyelination is autosomal recessive SPG35. This disorder sits along a disease spectrum spanning neurodegeneration with brain iron accumulation, leukodystrophy, and HSP and results from loss-of-function mutations in the fatty acid-2 hydroxylase gene FA2H (Dick et al. 2010, Schneider & Bhatia 2010). The FA2H protein is a nicotinamide adenine dinucleotide phosphate (NADPH)–dependent monooxygenase that converts free fatty acids to 2-hydroxy fatty acids. These are incorporated into myelin galactolipids containing hydroxy fatty acid as the N-acyl chain, which maintains the myelin sheath. Fa2h null mice have been developed as a model for SPG35, and these animals exhibit significant demyelination, axon loss or enlargement, cerebellar defects, and spatial learning and memory deficits. Animals lacking Fa2h only in oligodendrocytes and Schwann cells do not exhibit memory deficits, indicating that some neurological manifestations may derive from a lack of FA2H in other cell types (Potter et al. 2011).

Taken together, this subgroup of HSPs most exemplifies a noncell-autonomous disease pathogenesis. In this regard, oligodendrocytes from Plp1 null mice were able to induce a focal axonopathy when transplanted into the dorsal columns of the myelin-deficient shiverer mouse (Edgar et al. 2004). Thus, HSP-associated alterations in oligodendrocyte-mediated myelination can directly cause changes in the underlying axon, impairing corticospinal tract function.

Endoplasmic Reticulum Network Morphology

Cellular organelles have diverse but characteristic morphologies that are evolutionarily conserved, indicating that the function of an organelle is fundamentally related to its form. An obvious corollary is that disruption of form can give rise to disease, and in fact, this has been shown for a number of organelles. For instance, mitochondrial morphology is shaped by the opposing processes of fission and fusion, and multiple large GTPases involved in regulating this balance are mutated in autosomal dominant neurological disorders, including optic atrophy type 1 (OPA1) and Charcot-Marie-Tooth type 2A neuropathy (MFN2) (Westermann 2010).

An analogous situation occurs in the endoplasmic reticulum (ER), and the roles of common HSP gene products in shaping the tubular ER network have indicated this may be the most common pathogenic theme (Park et al. 2010, Montenegro et al. 2012). The ER is among the most distinctive organelles because of its large size, morphological heterogeneity, and extension throughout the cell. Although it is a continuous membrane-bound luminal system, it comprises the distinct morphologies of the nuclear envelope (with thousands of specialized pores), peripheral sheet-like structures studded with polyribosomes, and a polygonal network of interconnected smooth tubules distributed widely throughout the cell. Concordant with this structural heterogeneity, the ER is a multifunctional organelle involved in the synthesis, modification, quality control, and trafficking of integral membrane and secreted proteins. It is critical as well for Ca²⁺ sequestration and release, signaling, sterol synthesis, and lipid synthesis and distribution. In neurons, the ER plays crucial roles in the massive polarized membrane expansion that occurs during axon and dendrite genesis and as an in-tracellular Ca²⁺ store integrated with pre- and postsynaptic signaling pathways (Verkhratsky 2005, Park & Blackstone 2010, Renvoisé & Blackstone 2010, Lynes & Simmen 2011).

The three most common autosomal dominant HSPs—SPG3A, SPG4, and SPG31—as well as the less common SPG12 result from mutations in proteins directly implicated in the formation of the tubular ER network, which is overwhelmingly smooth ER (Park et al. 2010, Montenegro et al. 2012). Mutations in the SPG3A gene ATL1 are the second most common cause of HSP and are the most common cause of early-onset disease. The SPG3A protein atlastin-1 is a member of a family of large oligomeric GTPases related to the dynamin superfamily. Atlastin-1 is one of three homologous proteins in mammals (atlastin-1, -2 and -3) thought to be paralogs, but it is the only form expressed highly in the CNS (Zhu et al. 2003, Rismanchi et al. 2008). Atlastin-related GTPases are found in all eukaryotic cells and include Sey1p in Saccharomyces cerevisiae and root hair defective 3 (RHD3) in Arabidopsis (Hu et al. 2009). In contrast with mammals, species such as S. cerevisiae and Drosophila melanogaster have only a single atlastin ortholog. Across species, the atlastins can diverge considerably at the sequence level, but all share a similar domain organization: a large cytoplasmic N-terminal domain containing a tripartite GTP-binding domain, two very closely spaced hydrophobic segments, and a small cytoplasmic C-terminal tail (Figure 3). These multimeric, integral membrane GTPases localize predominantly to the tubular ER but are also found in the ER-Golgi intermediate compartment and in the cis-Golgi apparatus in some cell types (Zhu et al. 2003, 2006; Rismanchi et al. 2008).

Open in a separate window

Figure 3

Spastin, atlastin, REEP, and reticulon proteins interact and shape the endoplasmic reticulum (ER) network. (Left) Schematic diagram of a neuron showing the distribution of different ER domains. Below this schematic is an enlargement of a three-way tubular ER junction. REEP and reticulon proteins form large oligomers to shape the tubular ER. Atlastin proteins are enriched in puncta along the tubules, including at three-way junctions. (Right) Proposed membrane topologies for protein families involved in generating the tubular ER network. GTP, atlastin GTPase domain; MTB, microtubule-binding domain.

Atlastin GTPase activity is required for the formation of the three-way junctions in ER tubules in a wide range of species by directly mediating homotypic fusion of ER tubules (Rismanchi et al. 2008, Hu et al. 2009, Orso et al. 2009, Bian et al. 2011, Byrnes & Sondermann 2011, Chen et al. 2011, Moss et al. 2011). Consistent with this role, atlastins localize to discrete sites along ER tubules, including at three-way junctions. Depletion of atlastin-1 by shRNA in cultured cortical neurons inhibits axon elongation (Zhu et al. 2006), and there is a link between proper ER morphology and the formation and maintenance of long cellular processes such as axons and plant root hairs (see sidebar, ER Shaping: Plant Roots to Axons).

Mammalian atlastins and yeast Sey1p (synthetic enhancement of yop1) interact directly with the reticulon and Yop1p/DP1/REEP families of ER-shaping proteins in the tubular ER (Hu et al. 2009, Park et al. 2010). Members of these families each have two long hydrophobic stretches that form intramembrane hairpin domains predicted to partially span the lipid bilayer, inducing and/or stabilizing high-curvature ER tubules via hydrophobic wedging (Voeltz et al. 2006, Hu et al. 2008, Shibata et al. 2009, West et al. 2011). Mutations in REEP1 cause an autosomal dominant, pure HSP known as SPG31. REEP1 belongs to a family of related proteins (REEP1–6 in mammals) and was originally identified on the basis of its ability to promote trafficking of olfactory receptors to the plasma membrane surface (Saito et al. 2004). REEP1 localizes to the tubular ER and interacts with atlastin-1 via its predicted hydrophobic hairpin motifs. Some members (REEP1–4) also have an extended C-terminal domain (relative to REEP5–6) that binds microtubules, establishing REEP1 as a member of a subfamily of REEPs (REEP1–4) involved not only in ER shaping but also in interactions of ER tubules with the microtubule cytoskeleton (Park et al. 2010, Blackstone et al. 2011). Very recently, mutations in the RTN2 gene encoding the ER-shaping protein reticulon 2 have been identified in families with autosomal dominant SPG12 (Montenegro et al. 2012).

The fundamental link between the tubular ER and microtubules has been appreciated for decades (Terasaki et al. 1986), and although the microtubule cytoskeleton is not absolutely required for ER network formation (Shibata et al. 2009), microtubule-based ER motility is key for the proper organization and distribution of ER tubules. This is achieved through a variety of mechanisms: membrane sliding, in which ER tubules slide along microtubules using motor activity; microtubule movement, in which ER tubules latch on to moving micro-tubules; and the tip attachment complex, in which ER attaches to growing microtubule plus ends (Waterman-Storer & Salmon 1998).

Impairment of this relationship between ER tubules and the microtubule cytoskeleton as a pathogenic mechanism for HSPs is supported further by the fact that the SPG4 protein spastin, a microtubule-interacting and severing AAA ATPase, binds atlastin-1 and REEP1 as well as the ER-shaping protein reticulon 1 (Evans et al. 2006, Mannan et al. 2006, Sanderson et al. 2006, Connell et al. 2009, Park et al. 2010). Spastin occurs as two main iso-forms generated by differential use of AUG start codons: a 60-kDa form and a 67-kDa form (Mancuso & Rugarli 2008). The larger iso-form has an additional 86 amino acid stretch at the N-terminus containing a hydrophobic segment predicted to insert in the ER membrane as a partially membrane-spanning hairpin (Park et al. 2010). Interactions of M1 spastin with REEP1 and atlastin-1 appear to be mediated largely through this hairpin, though flank-ing regions may also participate (Evans et al. 2006, Sanderson et al. 2006). The larger spastin isoform is particularly enriched in the spinal cord, and a dysfunctional M1 spastin polypep-tide, but not one representing the shorter M87 form (M85 in rodents), was deleterious to axon growth in cultured neurons (Solowska et al. 2008), strengthening the evidence linking M1 spastin to HSP pathogenesis. By comparison, M87 spastin appears involved in cytokinesis, secretion, and possibly endocytosis through its interactions linking microtubule dynamics to membrane modeling in these compartments (Yang et al. 2008, Connell et al. 2009). In sum, there appear to be strong physical and functional links between M1 spastin, atlastin-1, reticulon 2, and REEP1 involved in shaping the tubular ER network in concert with the microtubule cytoskeleton. Even so, the ER has a large number of functions, and it remains unclear which are most pathogenically relevant for HSPs.

Because several other HSP proteins also localize to the ER, understanding these may clarify the role of ER in HSP pathogenesis. Mutations in the Berardinelli-Seip congenital lipodystrophy protein 2 gene BSCL2, which encodes an integral membrane of the ER known as BSCL2/seipin, cause two distinct diseases. Heterozygous gain-of-function mutations in an N-linked glycosylation site give rise to a disease spectrum encompassing autosomal dominant SPG17 (Silver syndrome), with distal amyotrophy as a significant feature in addition to spastic paraparesis, and distal hereditary motor neuropathy type V, characterized by more prominent distal spinal muscular atrophy (Windpassinger et al. 2004). In contrast, auto-somal recessive loss-of-function mutations give rise to Berardinelli-Seip congenital lipodys-trophy, without spasticity or amyotrophy. The seipin protein and its yeast ortholog, Fld1p (few lipid droplets 1), regulate the size of lipid droplets (LDs), explaining the loss-of-function lipodystrophy presentation (Cui et al. 2011; Fei et al. 2011a,b; Tian et al. 2011). The missense changes underlying SPG17 have resulted in misfolding of seipin, forming aggregates and triggering ER stress (Ito & Suzuki 2009, Yagi et al. 2011), which may play a role in HSP pathogenesis. LDs accumulate under various cellular stress conditions, and a number of unfolded protein response pathways have been implicated in LD formation (Hapala et al. 2011). However, it will be important to investigate any effects of misfolded seipin on other aspects of ER structure and function.

Recently, the gene for the complicated, autosomal recessive SPG18 was identified in a consanguinous Saudi family as ERLIN2 (Alazami et al. 2011). A second study independently identified loss-of-function ERLIN2 mutations in a family with motor dysfunction, joint contractures, and intellectual disability (Yıldırım et al. 2011). The erlin2 protein resides in the ER and contains a SPFH domain—named for its presence in stomatin, prohibitin, flotillin, and HflC/K. SPHF domain–containing proteins share the ability to assemble into large oligomers, and they localize preferentially to cholesterol-rich domains, including lipid rafts (Browman et al. 2006). Erlin2 has been functionally linked to ER-associated degradation (ERAD), a multistep degradative pathway encompassing ubiquitin-proteosome-mediated degradation of ER proteins, thereby regulating levels of proteins such as the inosi-tol trisphosphate receptor (Pearce et al. 2009). Also, erlin2 binds gp78, a membrane-bound ubiquitin ligase similar to HRD1 that mediates sterol-accelerated ERAD of 3-hydroxy-3-methyl-glutaryl (HMG)-CoA reductase, a key enzyme in the biosynthesis of cholesterol ( Jo et al. 2011).

Lipid Synthesis and Metabolism

These latter two HSP proteins, erlin2 and seipin, present compelling insights into how defects in ER shaping and distribution might cause HSP. A key function of the ER is the synthesis, metabolism, and distribution of lipids and sterols, employing both vesicular and nonvesicular mechanisms, and other HSP proteins fit into this pathogenic theme. The complicated, autosomal recessive HSP Troyer syndrome (SPG20) is caused by mutations resulting in a loss of spartin protein (Bakowska et al. 2008). Spartin localizes to a variety of cellular structures and has been implicated in a number of functions, including cytokinesis and epidermal growth factor (EGF) receptor trafficking (Robay et al. 2006, Bakowska et al. 2007, Renvoisé et al. 2010, Lind et al. 2011). Spartin also regulates LD biogenesis by promoting atrophin-1 interacting protein 4 (AIP4)-mediated ubiquitination of LD proteins (Eastman et al. 2009, Edwards et al. 2009, Hooper et al. 2010) and by recruiting PKC-ζ via the PKC-ζ-interacting proteins ZIP1 (p62/sequestosome) and ZIP3 to LDs (Urbanczyk & Enz 2011). Little is known about any roles for LDs in axons. However, because the SPG17 protein seipin regulates LD formation, alterations in LD biogenesis or turnover could affect lipid distribution, organelle shaping, or signaling pathways important for axonal health.

Although not directly implicated in LD biogenesis, other HSP proteins are enzymes involved in related lipid and cholesterol biosynthetic pathways. SPG42 is caused by mutations in the SLC33A1 gene encoding the acetyl-CoA transporter. In animals, acetyl-CoA is essential for maintaining the balance between carbohydrate and fat metabolism. Under normal circumstances, acetyl-CoA from fatty acid metabolism enters the citric acid cycle, contributing to the energy supply of the cell. SLC33A1 transports acetyl-CoA into the Golgi apparatus lumen and has been directly linked to the growth of axons because knock down of slc33a1 in zebrafish causes defective outgrowth from the spinal cord (Lin et al. 2008).

Last, neuropathy target esterase (NTE) is an integral membrane protein of neuronal ER that is mutated in autosomal recessive SPG39, a complicated HSP with prominent amyotrophy. NTE deacylates the major membrane phospholipid, phosphatidylcholine. Mutation of the NTE gene PNPLA2 or chemical inhibition of NTE with organophosphates alters membrane composition and causes distal degeneration of long spinal axons in mice and man (Rainier et al. 2008, Read et al. 2009). Another HSP protein, cytochrome P450-7B1 (CYP7B1), is mutated in autosomal recessive SPG5 (Tsaousidou et al. 2008) and functions in cholesterol metabolism; in patients with SPG5 there is a dramatic increase in oxysterol substrates in plasma and cerebrospinal fluid (Schüle et al. 2010). Given the fundamental roles played by lipids and sterols in neuronal functions, it seems very likely that more genes will be identified within this category.

Endosomal Dynamics

Although changes in ER morphology encompass proteins altered in the majority of patients with HSP, more HSP proteins have been implicated in endosomal dynamics; however, an immediate caveat is that the cell seems to coopt some of these proteins and machineries for other functions (Blackstone et al. 2011). As noted earlier, the SPG20 protein spartin functions in LD turnover, but it is also required for efficient EGF receptor degradation and likely regulates EGF signaling (Bakowska et al. 2007). Spartin may also regulate a variety of signaling pathways via ubiquitin modification through its interactions with E3 ubiquitin ligases such as AIP4 and AIP5. Spartin harbors an MIT domain as well and interacts selectively with IST1, a component of the ESCRT-III complex. ESCRT comprises a series of cytosolic protein complexes, ESCRT-0, ESCRT-I, ESCRT-II, and ESCRT-III, and the sequential activities of these complexes are required to recognize and sort ubiquitin-modified proteins into internal vesicles of multivesicular bodies (Hurley & Hanson 2010). More recently, ESCRT proteins have been implicated in other cellular functions, including viral budding and cytokinesis, and spartin has been shown to participate in cytokinesis (Renvoisé et al. 2010, Lind et al. 2011).

This link to ESCRT-III is shared by the SPG4 protein spastin. Although discussed earlier in the context of its interactions with REEP1 and atlastin-1 to coordinate ER membrane modeling and microtubule interactions (Park et al. 2010), spastin harbors an MIT domain that binds the ESCRT-III subunits CHMP1B and IST1 to couple the severing of microtubules with membrane scission. These ESCRT interactions are crucial for spastin’s role in severing microtubules to complete the abscission phase of cytokinesis (Yang et al. 2008, Connell et al. 2009, Guizetti et al. 2011). ESCRT proteins are also involved in centro-some stability (Morita et al. 2010), raising the possibility that axon genesis and regulation by centrosomes could be a function of these proteins. Other possibilities include roles for the ESCRT-III interactions with spastin and spartin in the delivery and downregulation of cell surface receptors to regulate signaling in axons.

Mutations in the KIAA1096 (SPG8) gene that encodes strumpellin cause a severe, pure HSP (Valdmanis et al. 2007). Strumpellin contains few motifs or domains of identified function, aside from a region of putative spectrin repeats. The main clues to its possible endosomal functions stem from its identification as a subunit of the WASH complex (Derivery & Gautreau 2010). This complex, comprising seven subunits (five core components), connects tubular endosomes of retrograde cargo sorting to the cytoskeleton, and it associates with endosomes via an interaction with vacuolar protein sorting–associated protein 35 (VPS35) (Harbour et al. 2010). Along with VPS26 and VPS29, VPS35 is a component of retromer, an endosomal complex responsible for sorting cargoes from endosomes to the trans-Golgi network (Bonifacino & Hurley 2008). Depletion of members of the WASH complex cause increased tubulation at early endosomes, impairing trafficking through early endosomal compartments (Derivery et al. 2009, Gomez & Billadeau 2009, Jia et al. 2010). Several members of the complex—WASH1, FAM21 (family with sequence similarity 21), and actin-capping protein—regulate actin dynamics. The WASH complex helps generate an actin network on early endosomes, for instance by activating the Arp2/3 complex to nucleate new actin filaments branching off extant filaments. The increased tubulation associated with WASH-complex depletion may reflect a lack of actin-mediated forces that are required for fission of tubular transport intermediates from the endosome (Campellone & Welch 2010). Thus, the WASH complex (containing strumpellin) exemplifies another HSP protein functioning in coordinating membrane modeling and cytoskeletal organization. Strumpellin interacts with valosin-containing protein (VCP/p97), an AAA ATPase mutated in frontotemporal dementia with Paget’s disease of bone and inclusion body myopathy (Clemen et al. 2010), and VPS35 is mutated in a number of families with autosomal dominant, late-onset Parkinson disease (Vilariño-Güell et al. 2011, Zimprich et al. 2011). A causative mutation in the WASH subunit SWIP has recently been identified for autosomal recessive intellectual disability (Ropers et al. 2011). Thus, roles of the WASH-retromer axis in neurological disease clearly extend beyond the HSPs and represent a very important area for investigation.

Another emerging HSP-related complex related to endocytic trafficking comprises the SPG15 protein spastizin/FYVE-CENT, the SPG11 protein spatacsin, and the SPG48 protein KIAA0415. Clinically, SPG11 and SPG15 share a number of characteristics; both are frequently associated with thin corpus callosum, and both can present with juvenile parkinsonism. Spastizin and spatacsin colocalize in cytoplasmic structures and were identified as proteins that coprecipitate with the SPG48 protein KIAA0415 (Slabicki et al. 2010, Murmu et al. 2011). KIAA0415 was originally proposed as a DNA helicase on the basis of sequence predictions (Slabicki et al. 2010), but a very recent study provides compelling evidence that it is a subunit of a new adaptor protein complex, AP-5, involved in endosomal dynamics (Hirst et al. 2011). Spastizin/FYVE-CENT contains a FYVE domain and binds the lipid PI(3)P, functioning along with ESCRT proteins in cytokinesis (Sagona et al. 2010). Along these lines, mutations in multiple proteins of the AP-4 complex, which is involved in trafficking of amyloid precursor protein from the trans-Golgi to endosomes (Burgos et al. 2010), cause autosomal recessive syndromes with prominent clinical features ranging from intellectual disability to progressive spastic paraplegia (Abou Jamra et al. 2011, Moreno-De-Luca et al. 2011). Thus, these adaptor protein complexes appear highly relevant for pathogenesis of HSPs and other neurological diseases.

Together, the expanding number of HSP genes implicated in endosome dynamics are already revealing new relationships among protein complexes, with implications that extend beyond the primary HSPs. Indeed, a number of patients with familial amyotrophic lateral sclerosis (ALS) resulting from an autosomal recessive mutation in the ALS2 gene encoding the alsin protein, a guanine nucleotide exchange factor (GEF) for the small GTPases Rab5 and Rac1, have a disease presentation more similar to the HSPs than to ALS. Examination of Als2 null mice revealed motor impairments and a distal axonopathy of the corticospinal tract. Rab5-dependent endosomal fusion is impaired in neurons from these mice, whereas alsin overexpression in neurons stimulates Rab5-dependent endosomal fusion, resulting in enlarged endosomes (Devon et al. 2006, Deng et al. 2007, Hadano et al. 2007).

Motor-Based Transport

The identification of mutations in the KIF5A gene encoding kinesin heavy chain 5A (known also as kinesin-1A) in families with SPG10, a pure or complicated HSP, has provided direct evidence for motor-based transport impairments underlying HSPs (Reid et al. 2002, Goizet et al. 2009). KIF5 proteins are ATP-dependent motors that move cargoes in the anterograde direction along axons, and most mutations are missense changes in the motor domain. Drosophila harboring mutations in the KIF5 ortholog Khc have posterior paralysis, with organelle-filled axon swellings jammed with cargoes (Hurd & Saxton 1996). In mammals, the KIF5A motor protein shuttles neurofilament subunits along axons and possibly other anterograde cargoes such as vesicles. KIF5 also regulates transport of cargoes in den-drites and has roles in a number of membrane traffic pathways. The efficiency of cargo transport to the distal axon is thought to be affected either because the mutated KIF5A are slower motors or because they have reduced micro-tubule binding affinity and act in a dominant-negative manner by competing with wild-type motors for cargo binding (Ebbing et al. 2008).

Bone Morphogenetic Protein Signaling

One compelling candidate that crosses HSP categories and is widely implicated in neurodegenerative diseases is bone morphogenetic protein (BMP) signaling (Bayat et al. 2011). HSP-associated mutations are found in at least four proteins—atlastin-1, NIPA1 (nonim-printed in Prader-Willi/Angelman syndrome 1; SPG6), spastin, and spartin—that function as inhibitors of BMP signaling. In Drosophila and mammals, BMP signaling functions in regulating axonal growth and synaptic function, and impairment of BMP signaling in Drosophila leads to axon transport defects (Wang et al. 2007). In rodents, BMP signaling is upregulated following lesioning of the corticospinal tract, and suppression of this upregulation can promote regrowth of axons (Matsuura et al. 2008).

Of the HSP proteins known to inhibit BMP signaling, the best characterized mechanistically is NIPA1, an integral membrane protein with 9 predicted TMDs that localizes to endo-somes and the plasma membrane and functions in Mg²⁺ transport (Goytain et al. 2007). Drosophila larvae lacking spichthyin (NIPA1 or-tholog) have increased synaptic boutons at neu-romuscular junctions and increased phosphorylated MAD (mothers against decapentaplegic), a downstream messenger of BMP signaling. These changes can be suppressed with genetic alterations that inhibit BMP signaling (Wang et al. 2007). NIPA1/spichthyin is thought to inhibit BMP signaling by promoting the internalization of BMP type II receptors and their subsequent lysosomal degradation, and NIPA1 missense changes found in SPG6 patients interfere with this process, upregulating signaling. Similarly, depletion of spartin or spastin, which both localize partially to endosomes, upregulates BMP signaling (Tsang et al. 2009).

Dysregulated BMP signaling linked to axonal abnormalities has also been demonstrated for atlastin. Depletion of atlastin-1 in zebrafish resulted in abnormal spinal motor axon morphology, with increased branching and decreased larval mobility. BMP signaling was up-regulated in these larvae, and pharmacological or genetic inhibition of BMP signaling rescued the atl1 null phenotype (Fassier et al. 2010). In sum, these results suggest that abnormal BMP signaling, probably caused by abnormal BMP receptor trafficking in many cases, could be a unifying mechanism for some classes of HSP, including the two most common, SPG4 and SPG3A, which comprise almost 50% of patients. Investigating relevant HSP animal models will be critical to determine whether inhibition of BMP signaling using small-molecule inhibitors, several of which are available, can rescue disease phenotypes.

I thank Alan Hoofring for preparing the figures. Work in the author’s laboratory is funded by the Intramural Research Program of the National Institute of Neurological Disorders and Stroke, National Institutes of Health.