Converging cellular themes for the hereditary spastic paraplegias

Abstract

Hereditary spastic paraplegias (HSPs) are neurologic disorders characterized by prominent lower-extremity spasticity, resulting from a length-dependent axonopathy of corticospinal upper motor neurons. They are among the most genetically-diverse neurologic disorders, with >80 distinct genetic loci and over 60 identified genes. Studies investigating the molecular pathogenesis underlying HSPs have emphasized the importance of converging cellular pathogenic themes in the most common forms of HSP, providing compelling targets for therapy. Most notably, these include organelle shaping and biogenesis as well as membrane and cargo trafficking.

Introduction

Hereditary spastic paraplegias (HSPs) are a large group of inherited neurologic disorders with prominent lower extremity spasticity due to a length-dependent axonopathy of long corticospinal neurons (Figure 1), whose axons can extend to 1 m in length, with or without additional prominent clinical findings [1]. HSPs are among the most genetically-diverse Mendelian disorders, with >80 distinct spastic gait genetic loci (SPG1-80 plus others) and over 60 gene products identified. Prominent lower extremity spasticity is also observed in inherited diseases classified outside of the spastic gait loci (SPG) nomenclature, including some hereditary motor and sensory neuropathies, spinocerebellar ataxias, motor neuron diseases, parkinsonism, spastic ataxias, and disorders of cognition. This daunting clinical complexity is tempered by the fact that most HSP gene products compellingly segregate into a relatively small group of cellular themes that play keys roles in axon development and maintenance. These include organelle shaping and biogenesis, membrane and cargo trafficking, mitochondrial function, nucleotide metabolism, and lipid/cholesterol metabolism [1]. Since many HSP proteins are involved in the biogenesis and/or shaping of membrane compartments, including the most prevalent forms, this review will focus primarily on this central theme, though others are outlined in both Figure 2 and Table 1.

Figure 1.

Human pyramidal motor system, emphasizing the long corticospinal neurons affected in HSP. Adapted from [1].

Figure 2.

Common pathogenic themes in the HSPs, emphasizing where the listed proteins are known or believed to function (adapted from [1]). Some proteins in Table 1 are not listed, pending further studies of their main sites of action.

TABLE 1.

Identified HSP gene products (SPG loci only), grouped by function

Disease/genea (HUGO)	Protein name	Inheritance	Cellular functions
Membrane traffic, organelle shaping and biogenesis
SPG3A/ATL1	Atlastin-1	AD	ER morphogenesis, BMP signaling, LD regulation
SPG4/SPAST	Spastin (M1 and M87 isoforms)	AD	Microtubule severing, ER morphogenesis, endosomal traffic, BMP signaling, LD biogenesis, cytokinesis
SPG6/NIPA1	NIPA1	AD	Endosomal traffic, Mg2+ transport, BMP signaling
SPG8/KIAA0196	Strumpellin	AD	Endosomal traffic, cytoskeletal (actin) regulation
SPG10/KIF5A	Kinesin heavy chain 5A	AD	Microtubule-based motor protein, anterograde axon transport
SPG11	Spatacsin	AR	Endosomal traffic, lysosomal biogenesis, autophagy
SPG12/RTN2	Reticulon 2	AD	ER morphogenesis
SPG15/ZFYVE26	Spastizin/ZFYVE26/FYVE-CENT	AR	Endosomal traffic, lysosomal biogenesis, autophagy
SPG17/BSCL2	Seipin/BSCL2	AD	LD biogenesis at ER
SPG18/ERLIN2	Erlin2	AR	ER-associated degradation, lipid raft-associated
SPG20	Spartin	AR	Endosomal traffic, BMP signaling, cytokinesis, LD turnover, ?mitochondrial regulation
SPG21	Maspardin	AR	Endosomal traffic (late)
SPG30/KIF1A	Kinesin family member 1A	AR	Microtubule-based motor protein, axon transport
SPG31/REEP1	REEP1	AD	ER morphogenesis, microtubule interactions, LD regulation
SPG33/ZFYVE27	Protrudin	AD	ER morphogenesis, endosome interactions
SPG47/AP4B1	AP-4 β1 subunit	AR	Endocytic adaptor protein complex
SPG48/KIAA0415	AP-5 ζ1 subunit	AR	Endocytic adaptor protein complex
SPG50/AP4M1	AP-4 μ1 subunit	AR	Endocytic adaptor protein complex
SPG51/AP4E1	AP-4 ε1 subunit	AR	Endocytic adaptor protein complex
SPG52/AP4S1	AP-4 σ1 subunit	AR	Endocytic adaptor protein complex
SPG53/VPS37A	VPS37A	AR	Retromer component
SPG57/TFG	Trk-fused gene	AR	ER morphology, vesicle biogenesis
SPG58/KIF1C	Kinesin family member 1C	AR, AD	Motor protein, retrograde Golgi-to-ER transport
SPG59/USP8	Ubiquitin-specific peptidase 8	AR	Deubiquitination enzyme
SPG60/WDR48	WD repeat domain 48	AR	Regulation of deubiquitination
SPG61/ARL6IP1	ARL6IP1	AR	ER morphogenesis
SPG62/ERLIN1	Erlin1	AR	ER-associated degradation, lipid raft-associated
SPG67/PGAP1	GPI inositol-deacylase	AR	Transport of GPI-anchored proteins from ER to Golgi apparatus
SPG69/RAB3GAP2	RAB3 GTPase-activating protein 2	AR	ER morphogenesis
SPG72/REEP2	REEP2	AR, AD	ER morphogenesis, microtubule interactions
SPG78/ATP13A2	ATP13A2	AR	Endosomal and lysosomal traffic
No SPG/TECPR2 (OMIM SPG49)	TECPR2	AR	Autophagy
Mitochondrial regulation
SPG7	Paraplegin	AR	Mitochondrial m-AAA ATPase
SPG13/HSPD1	HSP60	AD	Mitochondrial chaperonin
SPG55/C12orf65	C12orf65	AR	Mitochondrial protein translation
SPG74/IBA57	IBA57	AR	Mitochondrial iron-sulfur cluster assembly pathway
SPG77/FARS2	Phenylalanine tRNA synthetase 2	AR	Mitochondrial protein translation
Myelination and lipid/sterol modification
SPG2/PLP1	Proteolipid protein	X-linked	Major myelin protein
SPG5/CYP7B1	CYP7B1/OAH1	AR	Cholesterol metabolism
SPG26/B4GALNT1	β-1,4-N-acetyl-galactosaminyl transferase	AR	Ganglioside biosynthesis
SPG28/DDHD1	Phospholipase A1	AR	Phosphatidic acid metabolism, membrane traffic
SPG35/FA2H	Fatty acid 2-hydroxylase	AR	Myelin lipid hydroxylation
SPG39/PNPLA2	NTE/PNPLA6	AR	Phospholipid homeostasis, BMP signaling
SPG42/SLC33A1	Solute carrier family 33, acetyl-CoA transporter, member 1	AD	Acetyl-CoA transporter, BMP signaling
SPG44/GJC2	Connexin-47	AR	Intercellular gap junction channel
SPG46/GBA2	Glucocerebrosidase 2	AR	Lipid metabolism
SPG54/DDHD2	Phospholipase A1	AR	Phosphatidic acid metabolism, membrane traffic
SPG49/CYP2U1 (OMIM SPG56)	CYP2U1	AR	Long-chain fatty acid metabolism
SPG73/CPT1C	Carnitine palmitoyltransferase 1C	AD	Lipid metabolism, ceramides
SPG75/MAG	Myelin-associated glycoprotein	AR	Cell membrane adhesion molecule
Axon Pathfinding
SPG1/L1CAM	L1CAM	X-linked	Cell adhesion and signaling
Nucleotide metabolism
SPG63/AMPD2	Adenosine monophosphate deaminase 2	AR	Deaminates AMP to IMP (purine metabolism)
SPG64/ENTPD1	Ectonucleoside triphosphate diphosphohydrolase 1	AR	Hydrolyzes ATP and other nucleotides (purinergic transmission)
SPG65/NT5C2	Cytosolic 5′-nucleotidase	AR	IMP hydrolysis, purine/pyrimidine nucleotide metabolism
Other/Unknown
SPG22/SLC16A2	Solute carrier family 16 (monocarboxylic acid transporter) member 2	X-linked	Thyroid hormone (T3) transporter
SPG23/DSTYK	Dusty protein kinase	AR	Cell death regulation
SPG43/C19orf12	C19orf12	AR	Unknown
SPG66/ARSI	Arylsulfatase I	AR	Sulfate ester hydrolysis, hormone biosynthesis
SPG68/FLRT1	Fibronectin leucine-rich transmembrane protein 1	AR	Regulation of cell adhesion and fbroblast growth factor signaling
SPG70/MARS	Methionyl-tRNA synthetase	AR	Cytosolic methionyl-tRNA synthesis
SPG71/ZFR	Zinc-finger RNA binding protein	AR	Unknown
SPG76/CAPN1	Calpain-1	AR	Calcium-activated, non-lysosomal, thiol protease
SPG79/UCHL1	Ubiquitin C-terminal hydrolase L1		Thiol protease of peptidase C12 family
SPG80/EXOSC3	Exosome component 3	AR	RNA exosome complex

Abbreviations: AD, autosomal dominant; AR, autosomal recessive; BMP, bone morphogenetic protein; LD, lipid droplet.

^aWhere different from disease name. Adapted from [1].

Mutations in genes responsible for SPG4, SPG3A, and SPG31 comprise about half of autosomal dominant HSP cases, and their respective protein products play keys roles in organelle shaping. Furthermore, loss-of-function mutations in the two most common autosomal recessive HSPs, SPG11 and SPG15, affect proteins that also clearly fall within this admittedly broad category [1].

Numerous HSP proteins cooperatively shape the ER network

The SPAST gene is mutated in SPG4, by far the most common HSP, and encodes spastin, an AAA ATPase that catalyzes internal breaks in microtubules, regulating cellular functions dependent on microtubules. There are two main spastin isoforms, M1 and M87, generated via alternative translation initiation. Cellular distributions of spastin are isoform-specific, with the larger, membrane-bound M1 isoform found predominantly in endoplasmic reticulum (ER) membranes but the soluble M87 isoform in cytoplasm and at structures such as midbodies and endosomes. A hydrophobic hairpin domain lies within M1 spastin’s N-terminal region that is absent from the smaller M87 isoform and mediates its membrane localization as well as some protein interactions [2].

The ER is a continuous membrane system comprising the double-membrane of the nuclear envelope, ribosome-studded sheets, and a polygonal, interconnected network of smooth tubules extending throughout the cell, with periodic dense tubular matrices [2,3]. Tubular ER extends well into axons (Figure 3), including the very long axons impaired in HSPs, and comprises tubules far narrower than typically encountered in other cell types, with diameters of only 20–30 nm [4,5]. Importantly, M1 spastin interacts with the atlastin-1 protein mutated in SPG3A, the second most common HSP. Atlastin-1 is a member of a superfamily of membrane-bound, ER-localized GTPases that mediate homotypic fusion of ER tubules, generating the three-way junctions that give peripheral ER its characteristic polygonal appearance. Remarkably, numerous other HSP proteins also localize to the tubular ER and function to sculpt the distinctive features of the tubular ER network. These proteins include REEP1 (SPG31), reticulon 2 (SPG12), ARL6IP1 (SPG61), RAB3GAP2 (SPG69), and REEP2 (SPG72). Several of these propel the high curvature of tubular ER membranes, most notably those of the REEP and reticulon protein families plus ARL6IP1; these proteins harbor elongated, hydrophobic segments which form hairpins that insert into the ER membrane – generating high curvature through hydrophobic wedging and scaffolding in addition to serving as protein interaction motifs [2].

Figure 3.

Diagram of an axon, emphasizing organelle interactions with the cytoskeleton (top) and with one another (bottom). Though not shown, multiple motor types can be found on the same organelle.

REEPs comprise six members in humans, REEP1-6, with prominent phylogenetic and structural differences between REEP1-4 and REEP5-6, though all have predicted hydrophobic hairpins. REEP1-4 proteins also interact with microtubules [6,7], and both REEP1 and REEP2 are mutated in different forms of HSP. In highly-polarized cells such as neurons, the distribution of the ER is tightly coordinated with cytoskeletal dynamics, mostly involving microtubules [2]. Interactions among REEPs, reticulons, atlastins, and M1 spastin provide an attractive mechanism for coupling ER membrane remodeling to microtubule dynamics [6]. So how could disruptions in ER shaping and network formation potentially give rise to axonal pathology in HSPs? For starters, in zebrafish spastin is important for proper axon outgrowth during embryonic development [8], reminiscent of effects of atlastin-1 depletion in rat cortical and human forebrain neurons [9]. Also, SPG4 mice expressing mutant spastin exhibit axon transport defects, as do neurons derived from iPSCs and olfactory stem cells from SPG4 patients [10–12].

At the organelle level, there is accumulating evidence for direct effects of HSP-related alterations on ER morphology in axons. In Drosophila, hydrophobic hairpin-containing ER proteins are required for shaping and continuity of axonal ER, suggesting direct roles for ER modeling in axon maintenance and function [13]. These morphological changes have several known impacts on neuronal function, such as affecting synaptic structure and function and altering store-operated calcium entry [14,15]; there are likely others yet to be uncovered. Although precisely how dynamic changes in ER morphologies are linked to the maintenance of axons remains unknown, spastin-, atlastin-, and microtubule-dependent transport and distribution of ER tubules appear critical.

ER-organelle contacts in HSP pathogenesis

Direct effects of HSP mutations in directly altering ER morphology within axons prefigures a prominent role in disease pathogenesis, but how would this be mediated? The interactions of ER with microtubules just discussed is one way. Another related possibility is through impairing proper ER interactions with other organelles, an area of intense interest given the extensive contacts ER makes with a plethora of organelles [16,17]. For instance, contacts between endosomes and ER promote endosomal tubule fission, and interaction between the MIT domain of spastin (representing both M1 and M87 isoforms) and the ESCRT-III protein IST1 at ER-endosome contacts drives this process. A failure of fission results in defective sorting of the mannose 6-phosphate receptor, leading to disrupted lysosomal enzyme trafficking and abnormal lysosomal morphology in neurons and other cell types [18].

Similar lysosomal abnormalities are seen in cellular models lacking the SPG31 protein REEP1, which localizes to tubular ER, as well as upon depletion of the SPG8 protein WASHC5/strumpellin [18], which is part of the pentameric WASH complex that associates with endosomes via interaction with VPS35, a component of the trimeric retromer complex. WASH regulates the actin network on early endosomes by activating Arp2/3 to nucleate new actin filaments, branching off from existing filaments. Increased tubulation upon WASH complex depletion may reflect a dearth of the actin-mediated forces required for fission of tubular transport intermediates from endosomes [19]. Furthermore, the ER protein protrudin (SPG33), which binds many other HSP proteins [20], is crucial for promoting neurite outgrowth through its role in ER-endosome contacts [21]. Finally, the ER shaping protein reticulon-3 orchestrates the creation of plasma membrane (PM)-ER contact sites required for the formation of non-clathrin-dependent invaginations, as well as Ca²⁺ release triggered by IP₃-dependent activation of ER Ca²⁺ channels, to internalize EGFR via this path [22]. Together, these data implicate impairment of ER-endosome contacts in corticospinal axonopathy and suggest that coupling of ER-mediated endosomal tubule fission to lysosome function links different classes of HSP proteins previously considered functionally distinct into a compelling common pathway for axonal degeneration (Table 1).

The ER has a few other possible HSP-relevant roles in cells. Models of HSP linked to alterations in ER shaping proteins also impact mitochondrial distribution, function, and contacts [23–27]. In addition, lipid droplets (LDs) are prominent organelles formed from ER, and they are major fat storage organelles in eukaryotes [28]. In C. elegans and Drosophila, atlastin loss-of-function mutations trigger not only changes in ER morphology, but also reductions in tissue triglycerides and LD size. In mammalian cells, co-overexpression of atlastin-1 and REEP1 generates large LDs [29], and REEP1 plays a key role in LD regulation in both cell culture and in vivo [30]. Similarly, increasing M1 spastin levels affects LDs, decreasing number but increasing size. Overexpression of Drosophila spastin leads to larger and less numerous LDs in fat bodies, plus increased triacylglycerol levels. By contrast, spastin overexpression increases LD number when expressed selectively in skeletal muscles or nerves. Depleting spastin in flies decreases LD number in skeletal muscle, nerves and fat bodies, as well as reducing triacylglycerol levels in larvae [31]. In aggregate, these studies demonstrate an additional, evolutionarily-conserved role for these ER network shaping proteins in LD formation and size regulation.

Potential pathogenic significance of LD changes for HSP pathogenesis is supported by studies of several other known HSP proteins. SPG20 is a rare HSP caused by loss of spartin protein function. In addition to roles in cytokinesis and epidermal growth factor (EGF) receptor trafficking, spartin also regulates LD biogenesis by promoting atrophin-1 interacting protein 4 (AIP4)-mediated ubiquitination of LD proteins in cells and in mouse models [32,33]. The SPG17 protein seipin also regulates ER-LD contacts and cargo delivery, further indicating that alterations in LD biogenesis or turnover could affect lipid distribution, organelle shaping, or signalling important for axon maintenance [34]

Impairments in lipid metabolism

A key function of tubular ER is the synthesis, metabolism, and distribution of sterols and lipids. As shown in Figure 2 and Table 1, many HSP proteins are enzymes known to be involved in related lipid and sterol biosynthetic pathways. For example, SPG42 results from mutations in SLC33A1 that encodes the acetyl-CoA transporter; acetyl-CoA is essential for maintaining a balance between fat and carbohydrate metabolism. SLC33A1 transports acetyl-CoA into the Golgi apparatus and has been directly linked to the growth of axons, since depletion of slc33a1 in zebrafish causes defective outgrowths from spinal cord [35].

Another example is provided by patatin-like phospholipase domain-containing protein 6 (PNPLA6), an integral membrane protein of the ER in neurons that is mutated in autosomal recessive SPG39. PNPLA6 deacylates phosphatidylcholine, the major membrane phospholipid. Mutation of the PNPLA2 gene changes membrane composition, resulting in distal degeneration of long axons. Another HSP protein, cytochrome P450-7B1 (CYP7B1), is mutated in autosomal recessive SPG5 and functions in cholesterol metabolism; in patients with SPG5 there is a dramatic increase in oxysterol substrates in plasma and cerebrospinal fluid, providing a clear biomarker for therapeutic trials [36,37].

Endoysosomal and autophagic dysfunction

The most common autosomal recessive HSPs are SPG11 and SPG15, virtually identical clinical disorders with prominent additional features including early-onset parkinsonism, cataracts, retinal abnormalities, ataxia, TCC, and characteristic white matter changes. This clinical spectrum is quite different from the mostly pure lower extremity spasticity that characterizes SPG3A, SPG4, and SPG31. The SPG11 and SPG15 proteins spatacsin and spastizin, respectively, bind one another and function together in endolysosomal pathways [38]. One function that appears particularly relevant for HSP pathogenesis is autophagic lysosome reformation (ALR), a pathway that generates new lysosomes [39]. Autophagy allows cells to adapt to changes in their environment by coordinating degradation and recycling of cellular constituents. Lysosomes are critical for terminating autophagy through their fusion with mature autophagosomes to generate autolysosomes that degrade autophagic materials. Thus, maintenance of the lysosomal population is essential for autophagy-dependent cellular clearance. Loss of spatacsin and spastizin, as occurs in SPG11 and SPG15, results in depletion of free lysosomes competent to fuse with autophagosomes and an accumulation of autolysosomes, reflecting failure in ALR. Mechanistically, spastizin and spatacsin are essential components for initiation of lysosomal tubulation [39]. Mouse models for these disorders have similarly shown lysosomal abnormalities [40,41]

Spastizin and spatacsin co-precipitate with the SPG48 protein KIAA0415/AP5Z1, a subunit of a hetero-tetrameric adaptor protein complex, AP-5, also involved in endolysosomal dynamics [42–44]. Finally, mutations in multiple proteins of the AP-4 adaptor protein complex cause autosomal recessive syndromes (SPG50-SPG52), with clinical features ranging from intellectual disability to progressive spastic paraplegia. ATG9A, the only multi-spanning membrane component of the autophagy core machinery, has recently been identified as a specific cargo of AP-4, further implicating dysfunctional autophagy in some forms of HSP [45].

Concluding remarks

There are clearly key converging pathogenic pathways for the most common HSPs related to ER morphology and endolysosmal/autophagic pathways, and studies have suggested links between these categories that could be pathologically relevant. Animal and cellular models of disease may help to clarify the most promising targets for therapies. Importantly, microtubule-binding agents can rescue axonal phenotypes in a variety of model systems for the most common HSPs–Drosophila mutant spastin neurons, mouse Spast knock out neurons, and neurons derived from both SPG3A and SPG4 forebrain neurons differentiated from patient iPSCs [9,10,46]. These types of agents have also been shown to be effective in models of spinal cord injury [47], and ER shaping plays a key role in axon regeneration after injury [48], increasing the potential impact of such therapies. The very compelling link to endolysosomal and autophagic dysfunction for many HSPs positions these pathways as important areas to develop therapeutically as well [49]. With persuasive cellular pathogenic mechanisms identified, pharmacologic manipulations and evaluations in cellular and animal pre-clinical models are increasingly promising.