AAA+ ATPases: Achieving Diversity of Function with Conserved Machinery



AAA+ adenosine triphosphatases (ATPases) are molecular machines that perform a wide variety of cellular functions. For instance, they can act in vesicle transport, organelle assembly, membrane dynamics and protein unfolding. In most cases, the ATPase domains of these proteins assemble into active ring-shaped hexamers. As AAA+ proteins have a common structure, a central issue is determining how they use conserved mechanistic principles to accomplish specific biological actions. Here, we review the features and motifs that partially define AAA+ domains, describe the cellular activities mediated by selected AAA+ proteins and discuss the recent work, suggesting that various AAA+ machines with very different activities employ a common core mechanism. The importance of this mechanism to human health is demonstrated by the number of genetic diseases caused by mutant AAA+ proteins.

AAA+ adenosine triphosphatases (ATPases) (ATPases associated with diverse cellular activities) were first defined as a subset of P-loop ATPases in the early 1990s, based on the homology within the ATP-binding domain. Conserved from prokaryotes to humans, these proteins are macromolecular machines that use the energy released by ATP hydrolysis to remodel their target substrates in various ways. For example, AAA+ proteins are involved in membrane fusion, protein degradation and DNA replication. While AAA+ proteins are highly homologous within the ATPase domain, usually located at the C-terminus of the protein, other regions of these proteins typically show little sequence similarity. As AAA+ domains are so similar, there has been considerable interest in understanding whether AAA+ proteins employ a common mechanism, and if so, how this common mechanism is adapted by the divergent portions of a AAA+ protein to provide specificity of function. In this review, we will discuss AAA+ proteins that are involved in fundamental eukaryotic cell biological processes. We also examine the recent work concerning the most well-studied AAA+ proteins, the prokaryotic Clp family, considering whether these findings can be applied to other AAA+ proteins. In addition, we will look at how adaptor proteins regulate AAA+ proteins. Finally, we will explore diseases that result when AAA+ proteins are mutated.

Conserved Features of the AAA+ Domain

The AAA+ family of ATPases shares several conserved structural motifs. These include sequences important for ATPase activity, discussed in detail below, and two other motifs that do not contribute directly to ATPase activity but comprise residues in the AAA+ domain: the N-linker, which may transduce energy from ATP hydrolysis to the rest of the protein, and the loops that line the pore of a AAA+ oligomer (1–6) (Figure 1B and Figure 4). Some AAA+ proteins, such as N-ethylmaleimide sensitive factor (NSF), have two AAA+ domains per monomer (called D1 and D2), whereas others, such as spastin, have only one (Figure 1A,C). Members of this family form oligomers that are predominantly hexameric and are always ring shaped with a central cavity (reviewed in 7, Figure 1C).

Figure 1.

Conserved features of AAA+ proteins. A) Domain organization. Class II AAA+ proteins have one AAA+ domain, while Class I AAA+ proteins have two AAA+ domains in tandem in the same polypeptide. Some atypical AAA+ proteins, like dynein, have six AAA+ domains covalently linked rather than forming non-covalent homohexamers. B) Conserved motifs. Enlargement of AAA+ domain showing relative location of features (for details, see text). PL = pore loop, SRH = second region of homology. C) Oligomerization. The active form of most AAA+ proteins is a ring-shaped hexamer. AAA+ proteins with two AAA+ ATPase domains have a stacked or double-ring appearance.

Figure 4.

Location of pore loops in a representative AAA+ ATPase, p97/VCP. A) Top view of the crystal structure of murine p97, rendered as a hexamer. Each p97 monomer contains two AAA+ domains with each contributing two pore loops. In the side view (see C), the structure resembles stacked rings with the D1 AAA+ domain forming the top ring and the D2 AAA+ domain forming the bottom ring. The D1 pore loop 1 is shown in red, D1 pore loop 2 in green, D2 pore loop 1 in orange and D2 pore loop 2 in yellow. Note that all the four pore loops project into the central pore. Renderings based on the structure of murine p97 (Protein Data Base #1OZ4). B) Closer view of pore region showing pore loops. Note that part of D2 pore loop 2 is disordered, and the complete loop is not seen in this structure. C) Cutaway side view of the hexamer. This view highlights the inside surface of pore cavity. Note that D1 loop 1 (red) lies closer to the mouth of the pore, whereas loop 2 (green) is deeper in the pore. The same is true for the D2 loops (loop1, orange; loop 2, yellow).

The ATPase domain of AAA+ proteins is 200–250 amino acids long and is characterized by a number of domains that are important for ATP binding and hydrolysis. These include the Walker A P-loop, Walker B motif, sensor-1 and sensor-2 motifs, arginine finger and a second region of homology that differentiates classically defined AAA proteins such as NSF from the broader AAA+ family that includes members such as dynein (8–10) (Figure 1B). The conserved residues lie in two subdomains, a ‘large AAA’ domain composed of an α/β Rossman fold and a ‘small AAA’ domain that is predominantly α-helical. The arrangement of the β strands in the large AAA domain distinguishes AAA+ nucleotide-binding domains from those of the other proteins using an α/β Rossman fold for nucleotide binding (11).

ATP binding and hydrolysis are obviously required to provide the energy AAA+ proteins need to remodel their substrate proteins. However, ATP binding may also play a role in oligomer formation and stability. The ATP-binding pocket is formed by residues from adjacent monomers in the active oligomeric form of a AAA+ protein, and therefore, ATP binding may regulate the assembly of some AAA+ proteins. The stability of the oligomeric form seems to vary among AAA+ proteins. Oligomer formation also allows for the possibility of co-operative ATP hydrolysis, which has been observed for some AAA+ proteins such as Hsp104 and p97/VCP (12,13).

Cell Biological Functions of AAA+ proteins

AAA+ proteins figure in a broad array of cellular processes (Figure 2 and Table 1). In this section, we will demonstrate this variety by discussing the functions of a few representative AAA+ proteins. We have grouped them into three categories to illustrate how AAA+ proteins use common mechanisms to accomplish different purposes. The two largest categories are those AAA+ family members that remodel protein complexes without unfolding or destroying their target proteins and those that are involved in protein quality control and do unfold their targets. Finally, we have included a third group of microtubule-interacting proteins because their activities are unique in terms of the larger AAA+ family.

Figure 2.

Subcellular localization of AAA+ proteins. This is not an exhaustive diagram of all AAA+ proteins but illustrates the location of the proteins highlighted in the text. For simplicity, proteins that are found in multiple locations are only shown in one. For the function of the proteins at the indicated locations, please see the text and Table 1.

Table 1. AAA+ proteins involved in common cell biological functions
ProteinSubcellular localizationFunction
NSFCytosol, vesicles, Golgi and endosomesDisassembly of SNARE complexes for vesicular transport
VPS4Multivesicular body and endosomesInvolved in disassembly of ESCRT-III complex during budding into the multivesicular body
PEX1 and PEX6Peroxisome and cytosolDislocation of PTS receptor from the peroxisomal membrane for recycling into the cytosol
Clp/Hsp100 familyCytosolProtein quality control (can disaggregate and unfold proteins); some members are protease associated
m-AAA protease (AFG3L2 and paraplegin)Mitochondrial inner membrane facing matrixMaturation of mitochondrial proteins through cleavage or membrane dislocation and mitochondrial protein quality control
i-AAA protease (YME1L1)Mitochondrial inner membrane facing intermembrane spaceMitochondrial protein quality control and translocation of proteins into the intermembrane space
p97/VCPCytosol, ER and GolgiRetrotranslocation of misfolded substrates during ERAD, maintenance of the ER, nuclear envelope and Golgi reassembly after mitosis
Cytoplasmic dyneinMicrotubules, spindle, endosomes, lysosomes, vesicles and GolgiMinus-end-directed microtubule-based motor protein
KataninCentrosome and spindleMicrotubule severing
SpastinGolgi, centrosome, spindle and nucleusMicrotubule severing
FidgetinCentrosome, spindle and nucleusMicrotubule severing
TorsinALumen of ER and nuclear envelopeUnknown

Non-destructive recycling

Disassembly of protein complexes by AAA+ proteins plays an important role in several cell biological functions including membrane dynamics and organelle assembly. In these cases, the target protein is not completely unfolded but merely disassociated from a larger complex and recycled for future use. NSF was one of the first eukaryotic AAA+ proteins to be studied in detail and was discovered as an essential component of vesicular transport (14). It was subsequently shown that NSF-mediated disassembly of cis SNARE complexes is required in order to regenerate free SNAREs, allowing fusion of vesicle and target membranes (15) (Figure 3A). Another type of membrane dynamics, budding into the lumen of endosomes in the multivesicular body pathway, also involves a AAA+ protein, VPS4. Once cargo selection is complete, VPS4 is critical for the disassembly of the ESCRT-III complex (endosomal sorting complex required for transport) found on the endosome, allowing intralumenal vesicle formation (16). However, the mechanism by which VPS4 catalyzes disassembly is not clear.

Figure 3.

AAA+ proteins use similar mechanisms to accomplish different purposes in the cell. A) Non-destructive recycling. The NSF is involved in the disassembly of the coiled-coil SNARE complex that is formed upon vesicle fusion. Disassembled SNAREs are ready for further rounds of fusion. B) Quality control. Some AAA+ proteins, such as ClpA, unfold proteins and feed them into an associated protease.

In another case, two AAA+ disassembly machines are used in organelle biogenesis. PEX1 and PEX6 are involved in the import of peroxisomal matrix proteins. Many peroxisomal proteins contain a peroxisomal targeting sequence (PTS) that is bound by a receptor. PEX1 and PEX6 form a complex that is required to dislocate PTS–receptor complex from the membrane back to the cytosol, so that it can be reused for further rounds of import (17).

Quality control

Numerous AAA+ proteins are involved in regulating the unfolding, refolding or proteolysis of proteins that are either damaged, targeted for destruction or need to be posttranslationally modified. Some AAA+ proteins also function as chaperones. For instance, in the Clp/Hsp100 family, ClpB dislocates polypeptides from aggregates and unfolds them, allowing them to interact with co-chaperones, such as the DnaK complex, and be refolded (18). ClpA and ClpX, other members of the same subfamily, associate with a ring-shaped protease, ClpP. They unfold protein substrates by translocating them through their central cavity and into the protease (19,20) (Figure 3B). This unfoldase–protease interaction ensures that only substrates recognized by ClpA or ClpX will be degraded. In the same way, the AAA+ proteins that form the cap of the 26S proteasome also control access to its proteolytic interior (21).

The mitochondrial inner membrane contains two proteases that are unique because their AAA+ domain and peptidase domain are found in the same polypeptide. These closely related proteases differ mainly in their topology: the m-AAA protease faces the matrix, and the i-AAA protease faces the intermembrane space. These two proteases, in addition to cleavage of specific mitochondrial proteins, are also utilized for quality control of respiratory chain complexes and degrade non-assembled membrane proteins. The m-AAA protease is composed of two AAA+ proteins, paraplegin and AFG3L2 (22,23). Posttranslational cleavage by this protease is required for the maturation of several important mitochondrial proteins, such as OPA1 (24). Paraplegin can also control the maturation of proteins through membrane dislocation rather than proteolysis; it has recently been shown to correctly position cytochrome C peroxidase for intramembrane cleavage by another protease (25). The i-AAA protease of mitochondria is also composed of a AAA+ protein, YME1L1, that homooligomerizes (26). This protease also has a newly discovered role in the translocation of polynucleotide phosphorylase into the intermembrane space (27).

Another AAA+ protein, p97/valosin-containing protein (VCP), is involved in protein quality control through the endoplasmic reticulum-associated degradation (ERAD) pathway. To ensure that only properly folded proteins enter the secretory pathway, misfolded proteins are retrotranslocated out of the ER, multiubiquitylated and eventually degraded by the proteasome (28). p97/VCP contains a ubiquitin-binding domain and for at least some substrates, is thought to provide the driving force for membrane dislocation. p97/VCP also has a role in maintenance of the ER, and in nuclear envelope and Golgi reassembly after mitosis (29–31). p97/VCP accomplishes its different tasks by interacting with several sets of adaptor proteins (32). For instance, in ERAD, p97/VCP forms a complex with Ufd1 and Npl4, while in membrane fusion it interacts with p47 (28–30).

Microtubule-associated AAA+ proteins

Motor proteins

One of the most well-known AAA+ proteins is cytoplasmic dynein. Dynein is a minus-end-directed microtubule motor protein that carries numerous cellular cargos and is important in a wide variety of cellular processes such as transport of vesicles and organelles, maintenance of the Golgi and orientation of the mitotic spindle (33–35). The dynein heavy chain is an atypical AAA+ protein in that its six AAA+ domains are linked in a single polypeptide (Figure 1). Also unusual is that only four of its six AAA+ domains have consensus Walker A sequences, which are required for nucleotide binding (8). Dynein binds to microtubules in an ATP-sensitive manner through the globular tip of a coiled-coil stalk that is located between the fourth and the fifth AAA+ domains (36–38). The N-terminal portion of the heavy chain, called the stem, binds many accessory factors (39,40). Relative movements of the stalk, ring and stem through the ATPase cycle are thought to generate the power stroke that enables dynein to move along a microtubule (41).

Microtubule severing enzymes

Three other AAA+ proteins, katanin, spastin, and fidgetin, are associated with microtubules in a different way; they are microtubule severing enzymes that make internal breaks in microtubules (42–44). Katanin is important for severing in the mitotic spindle and also plays a role in the meiotic spindle in Caenorhabditis elegans(45,46). In addition, spastin and fidgetin have recently been shown to stimulate microtubule depolymerization during anaphase in cultured Drosophila cells (47). Microtubule severing has also been hypothesized to generate non-centrosomal microtubule arrays, such as those found in the axons (48). It is unclear how these enzymes use their ATPase activity to sever microtubules. It is possible that they translocate and unfold tubulin like the quality control proteins ClpA and ClpX. Alternatively, ATPase activity may be coupled to a conformational change, popping a tubulin heterodimer out of the microtubule lattice without any unfolding.

Mechanism of Action: Pore Loops

As we have seen, AAA+ proteins are involved in a plethora of cellular processes. However, their common hexameric structure and conserved features such as pore loops suggest that they may use similar mechanisms. In this section, we will discuss the mechanism of the best studied group of AAA+ proteins, those that translocate substrate through their central pore, such as ClpA (Figure 3B). We will then examine how these findings might apply to other AAA+ proteins.

Pore loops in protein unfolding

To study the mechanism of translocation, a number of groups have explored the function of loops lining the central cavity of the AAA+ oligomer (Figure 4). The first loop to be examined lines the mouth of the pore and contains a conserved YVG or aromatic-hydrophobic-glycine sequence in many AAA+ proteins. Mutations in pore loop 1 affect substrate binding and processing without substantially interfering with oligomerization or ATPase activity (3–6,13,49). With ClpA and ClpB, cross-linking studies have shown that substrate proteins are in direct contact with this loop (6,50). A second loop deeper in the cavity (pore loop 2) has been examined in ClpA. Point mutations in this loop abolished binding of one type of target protein and degradation of another (50). For ClpA, three loops, pore loop 1 and 2 in D1 and pore loop 1 in D2, are equally important for substrate binding. This suggests that the three loops act in concert rather than sequentially.

Binding of substrates to flexible loops suggests a possible mechanism for translocation. For ClpA, Hinnerwisch et al. propose that translocation of substrate is driven by the movement of the D2 loop from a proximal position to a distal position while it is attached to substrate. This is consistent with the data from the RecA family bacteriophage Φ12 packaging motor P4, where the equivalent of pore loop 1 has been shown to alternate between ‘up’ and ‘down’ conformations throughout the ATPase cycle (51).

Pore loops in other types of AAA+ proteins

Subsequently, the role of pore loops has been examined in AAA+ proteins that are not obligate translocators but still contain the conserved aromatic-hydrophobic-glycine motif in pore loop 1. Conservation of this motif suggested that pore loops might also be important for substrate recognition and enzyme function in other AAA+ proteins. In support of this hypothesis, pore loop 1 has been cross-linked to substrate in p97/VCP (13), despite being substantially shorter in p97/VCP than in ClpA (50). Consistent with the results from Clp family AAA+ proteins, mutation of the aromatic residue of pore loop 1 in spastin and p97/VCP does not abolish ATPase activity but still interferes with the enzyme function (13,52). In addition, a pore loop 1 mutation in VPS4 abolishes retroviral budding in a cell-based assay (53). The aromatic residue in pore loop 1 in Yme1 is also required for substrate binding (of properly folded substrates), unfolding and translocation (54). In p97/VCP, substitution of residues in pore loop 2 of both D1 and D2 reduces co-operativity of ATP hydrolysis and impairs ERAD (13). Analysis of pore loop 2 in spastin reveals that it plays a critical role in recognition and severing of its substrate, tubulin (52). Thus, even in enzymes that do not have multiple substrates and that may not translocate their substrate, pore loops are critical for target protein recognition and for function.

Models for Co-ordination Among Subunits

As AAA+ oligomers have multiple ATPase domains, how might ATP hydrolysis be co-ordinated among subunits? One possibility is that a circular shape allows AAA+ proteins to act as rotary motors, with sequential ATP hydrolysis around the ring driving the substrate deeper into the pore (51). However, other methods of ATP hydrolysis, such as concerted hydrolysis by all subunits, are also possible. In addition, all ATP-binding sites within a hexamer might not be in the same state at the same time (55). Martin et al. recently used an elegant method to examine ATP hydrolysis in a hexamer by covalently linking six monomers through a flexible linker. They demonstrated that for ClpX, although efficiency is reduced, only one monomer with ATPase activity is required for translocation activity, and that, if multiple active monomers are included, they may be located at any position in the ring. Thus, for ClpX, a probabilistic method of ATP hydrolysis is likely. This is logical for a protein with numerous different substrates, in that whichever monomer is in contact with the substrate could hydrolyze its ATP, limiting the amount of non-productive hydrolysis (56). ClpX’s residual activity even with only one active ATPase domain is reminiscent of dynein, which does not have six active AAA+ domains, suggesting that dynein shares more with its AAA+ relatives than was originally apparent.

ATP hydrolysis at different subunits in the ring was also examined in a recent paper by Doyle et al. (57). They showed that when ATPγS binds at some ATP-binding sites, slowing hydrolysis, ClpB and Hsp104 can act independently of the chaperones they normally require. The authors suggest that this illustrates the difference between holding a substrate, which requires ATP binding, and unfolding, which requires ATP hydrolysis. Presumably binding by ATPγS allows a substrate to be held without the aid of a chaperone (57).

Regulation of AAA+ Proteins: Adaptor Proteins

Given the commonalities between AAA+ proteins, one must question how they achieve functional specificity. The answer likely lies within the divergent N-terminal domains that they utilize to interact with adaptor proteins. These adaptors may localize, regulate or determine the specific cellular function of a AAA+ protein, as discussed earlier for p97/VCP. For example, spastin binds through its N-terminal domain to atlastin, which localizes spastin to the Golgi (58). VPS4 localizes to endosomes through the interactions with CHMP1B/Did2, which also regulates its dissociation of the ESCRT-III complex (53,59). Finally, cytoplasmic dynein is a prototypical example of a AAA+ protein that uses a large number of adaptor proteins. In addition to the interactions of the heavy chain with the light, light intermediate and intermediate chains, dynein’s interactions with dynactin, LIS1 and a great variety of other proteins help to account for its myriad of cell biological roles (60–64).

More recently, adaptor proteins that bind to unique elements in the AAA+ domain and regulate ATPase activity have also been discovered. In yeast, Vta1 binds to the Vps4 β domain (a three-stranded antiparallel β sheet, inserted into the small AAA+ subdomain) and stabilizes the Vps4 oligomer (53,65). This stabilization increases the ATPase activity of Vps4; conversely, loss of Vta1 in vivo compromises Vps4 function (65,66). Cdc48, the yeast homolog of p97/VCP, binds to two adaptor proteins, Ufd2 and Ufd3, through its D2 domain. Ufd2 and Ufd3 compete for the same binding site on Cdc48 and have opposing activities, with Ufd2 adding further ubiquitin moieties, promoting degradation and Ufd3 inhibiting degradation (67). In prokaryotes, the DnaK chaperone system regulates ClpB ATPase activity through the M domain, a unique region composed of four helices found in the D1 ATPase domain of ClpB. These helices, which are inserted at the same location as the VPS4 β domain, appear to undergo large motions through the ATPase cycle (68,69). Although a physical interaction between the M domain and the DnaK/DnaJ/GrpE (KJE) complex has not been demonstrated, regulation of ATPase activity by the M domain only occurs during KJE-dependent disaggregation reactions (68).

AAA+ Proteins and Disease

The important role that AAA+ proteins play in cellular function is illustrated by the number of different human genetic diseases caused by dysfunctional AAA+ proteins (Table 2). Given the similarities in mechanism among many AAA+ proteins, it will be particularly important to determine shared features that may allow a common therapeutic approach.

Table 2. AAA+ proteins mutated in human genetic disease
Gene (protein)DiseaseInheritancePhenotypeCommentsReferences
  1. AD, autosomal dominant; AR, autosomal recessive.

DNAH5 (ciliary dynein heavy chain 5)Primary ciliary dyskinesia/Kartagener syndromeARBronchiectasis, sinusitis, dextrocardia and infertilityMutant protein does not correctly localize to ciliary axoneme, leading to outer dynein arm defects. (70)
PEX1 and PEX6Zellweger’s syndrome spectrumAR Infantile: hypotonia, inability to feed and liver dysfunction; Childhood: developmental delay, sensorineural hearing loss, retinal dystrophy and liver dysfunctionMutations affect peroxisomal import. Homozygosity or compound heterozygosity for frameshift mutations leads to more severe phenotype. (71–74)
BCS1LMitochondrial complex III deficiencyARNeonatal tubulopathy, encephalopathy and liver failure  (88)
GRACILE syndromeARGrowth retardation, amino aciduria, cholestasis, iron overload, lactic acidosis and early death  (89)
Bjornstad syndromeARSensorineural hearing loss and pili tortiUsually have at least one allele with partial activity. (75)
TOR1A (torsinA)Torsion dystoniaADInvoluntary twisting movementsMay function at nuclear envelope. Most individuals, regardless of ethnic background, have same mutation (ΔE302/3). (76)
SPG7 (paraplegin)Hereditary spastic paraplegiaARProgressive lower extremity spasticity and weakness  (23)
SPG4 (spastin)Hereditary spastic paraplegiaADProgressive lower extremity spasticity and weaknessMost mutants impaired in microtubule severing. May affect the ability to form normal axonal microtubule array. (82)
VCP (p97)IBMPFDADInclusion body myopathy, Paget disease of bone and frontotemporal dementiaAccumulation of misfolded proteins in inclusion body seen histologically. Most mutations in ubiquitin-binding domain. Family with mutation in D1 AAA+ domain has more aggressive disease. (86)

The first type of disease that is caused by mutant AAA+ proteins is developmental. These conditions are typically severe and present in childhood. For instance, while cytoplasmic dynein has not been found to cause genetic disease in humans, the closely related axonemal dyneins cause primary ciliary dyskinesia when they are mutated (70). Three proteins that are involved in the organelle assembly also cause disease when dysfunctional. PEX1 and PEX6 are both involved in the import of proteins into the matrix of the peroxisome during peroxisomal development. Loss of their function leads to Zellweger’s syndrome spectrum, which presents in infancy or in childhood with severe dysfunction in many organ systems (71–74). Mutations in BCS1L, which is a chaperone required for the assembly of mitochondrial complex III, cause three different autosomal recessive diseases. These diseases, called complex III deficiency, GRACILE syndrome and Bjornstad syndrome, form a clinical spectrum, the severe end of which involves multisystem abnormalities and death in childhood. While nonsense or frameshift mutations leading to protein truncation have been seen in all three phenotypes, patients with the milder Bjornstad syndrome have at least one allele that is predicted to have partial activity (such as a missense mutation). In addition, mutations in the AAA+ domain that cause the more severe GRACILE syndrome and complex III deficiency are in residues that are predicted to interact with ATP or magnesium. In contrast, mutations causing Bjornstad syndrome map to the outer face of the AAA+ domain and are more likely to disrupt protein–protein interactions (75).

The second type of disease that is caused by mutations in a number of different AAA+ proteins is neuronal. Mutation of the TOR1A locus encoding torsinA leads to torsion dystonia, a childhood-onset disorder characterized by involuntary twisting movements (76). TorsinA is a peripheral membrane protein that is found in the lumen of the ER and nuclear envelope (77,78). Its function is still being characterized; however, its localization to the nuclear envelope in its ATP-bound state suggests that its substrate is located there (79). Mutant torsinA (ΔE302/3) also localizes to and disrupts the nuclear envelope (80). Two homologous transmembrane proteins have been found to interact with torsinA and may be its substrates: lamina-associated polypeptide 1 in the nuclear envelope and LULL1 in the ER (81).

Two forms of hereditary spastic paraplegia (HSP) are caused by altered AAA+ proteins. In HSP, axons of spinal neurons in the dorsal columns and cortical spinal tract degenerate over time, with clinical symptoms resulting from this pathology. HSP can be caused by mutation or deletion of spastin, a microtubule severing enzyme (23,82). Although spastin was recently shown to play a role in the mitotic spindle, its function in post-mitotic neurons is most likely the generation of non-centrosomal microtubules, which may be critical for maintaining the specialized axonal microtubule array (83). Spastin severing may also generate short microtubules at neuronal branch points. Furthermore, loss of spastin severing may impair microtubule-based axonal transport.

Mutations in paraplegin, a subunit of the m-AAA ATPase of mitochondria, also lead to HSP. Paraplegin has been shown to have two functions that together may contribute to axonal degeneration. Paraplegin’s role in quality control of respiratory chain complexes (it removes non-assembled or misfolded subunits) is believed to be important for maintaining respiratory competence (84). Recently, it was found that paraplegin plays an additional role in respiration. It can cleave mitochondrial targeting sequences, including that found on MrpL32, a subunit of the mitochondrial ribosome. Therefore, paraplegin is required for ribosome assembly (85). Mitochondrial ribosomes translate the mitochondrially encoded subunits of the respiratory complex; this is an additional way in which respiration can be impaired by mutant paraplegin. Because synapses are regions that require large amounts of energy and contain many mitochondria, paraplegin mutations may cause synaptic defects. Pathologically, paraplegin knockout mice show enlarged mitochondria trapped in swollen axons, and it is possible that this phenotype is also because of the respiratory insufficiency of the mitochondria (84).

Another type of neurodegeneration that is combined with phenotypes outside the nervous system, inclusion body myopathy with Paget disease of bone and frontotemporal dementia (IBMPFD), is caused by mutations in p97/VCP. In patients with this disease, p97/VCP stains aggregates found in muscle (86). Most IBMPFD mutations in p97/VCP are in the N-terminal ubiquitin-binding domain of the protein, and one of these mutants is impaired in ERAD (87). Together, these results suggest that p97/VCP’s role in ERAD, as opposed to its other activities, is particularly important in disease pathogenesis, and that a dysfunctional ERAD pathway can lead to aggregates of misfolded proteins and ultimately to degenerative disease.

Concluding Remarks

In this review, we have attempted to address whether AAA+ proteins operate by a universal mechanism. One aspect of this issue is whether the translocation mechanism that has been studied in such detail for the Clp proteins occurs in other AAA+ proteins. Even though pore loops have been shown to be important, translocation has not been demonstrated in other types of AAA+ proteins. It is possible that proteins that do not unfold their substrate translocate it partially and then stop. It is easier to understand that how some form of translocation might act in some AAA+ proteins, such as microtubule severing enzymes, while in others, such as dynein, it is more difficult to imagine.

The discovery of increasing numbers of adaptors for AAA+ proteins and unique domains within AAA+ proteins suggests that while a common mechanism may be coming to light, there are many details still to be determined regarding the mechanics of these amazing molecular machines. And, while it seems that for ClpX, a probabilistic mode of ATP hydrolysis occurs within the hexamer, it is not clear that this is the case for all AAA+ proteins.

Another question remains regarding the nature of disease mutations in AAA+ proteins. Interestingly, some autosomal dominant diseases, such as HSP caused by spastin mutations, and some autosomal recessive diseases, such as mitochondrial complex III deficiency, are both caused by mutations in the AAA+ domain that are thought to affect ATP binding and hydrolysis. Most likely it is not just that neurons are exquisitely sensitive to the loss of AAA+ proteins because paraplegin mutations are autosomal recessive. In addition, it is not clear whether for autosomal dominant diseases, the mutations act as dominant negatives or if loss of function results in haploinsufficiency. It is clear that there are many avenues of study to be explored before we fully understand AAA+ ATPases and their place in the cell.


We thank Dr Lawrence Shapiro for help with rendering of crystal structure images. We would also like to thank the members of the Gundersen laboratory for critical reading of the manuscript. This work was funded by National Institutes of Health/National Institute of Neurological Disorders and Stroke, the Spastic Paraplegia Foundation and the Irma T. Hirschl Trust.