A protein quality control system, consisting of molecular chaperones and proteases, controls the folding status of proteins and mediates the refolding or degradation of misfolded proteins. Ring-forming AAA+ (ATPase associated with various cellular activities) proteins play crucial roles in both processes by co-operating with either peptidases or chaperone systems. Peptidase-associated AAA+ proteins bind substrates and thread them through their axial channel into the attached proteolytic chambers for degradation. In contrast, the AAA+ protein ClpB evolved independently from an interacting peptidase and co-operates with a cognate Hsp70 (heat-shock protein 70) chaperone system to solubilize and refold aggregated proteins. The activity of this bi-chaperone system is crucial for the survival of bacteria, yeast and plants during severe stress conditions. Hsp70 acts at initial stages of the disaggregation process, enabling ClpB to extract single unfolded polypeptides from the aggregate via a threading activity. Although both classes of AAA+ proteins share a common threading activity, it is apparent that their divergent evolution translates into specific mechanisms, reflecting adaptations to their respective functions. The ClpB-specific M-domain (middle domain) represents such an extra feature that verifies ClpB as the central disaggregase in vivo. M-domains act as regulatory devices to control both ClpB ATPase activity and the Hsp70-dependent binding of aggregated proteins to the ClpB pore, thereby coupling the Hsp70 chaperone activity with the ClpB threading motor to ensure efficient protein disaggregation.
The AAA+ protein family
In all organisms, many vital cellular processes, including membrane fusion, cell cycle regulation, organelle biogenesis, protein repair and degradation, are controlled by members of the AAA+ (ATPase associated with various cellular activities) superfamily [1–3]. The activity of AAA+ proteins relies on their ability to use the energy of ATP hydrolysis to generate a mechanical force, leading to the remodelling of bound substrates. ATP binding and hydrolysis is mediated by the conserved AAA domain, which additionally drives the oligomerization of AAA+ proteins, leading to the formation of barrel-shaped oligomers with a central channel. The protein superfamily can be divided further into two distinct classes, on the basis of the number of AAA domains present in the protein. Class I proteins contain two AAA domains, referred to as AAA-1 and AAA-2, separated by a linker sequence of variable length. In contrast, class II proteins contain only one AAA domain, which is homologous with AAA-2. Functional diversity of AAA+ proteins is caused by the presence of additional domains that are either fused to or inserted into the AAA domain. Such extra domains are present in some family members, but are lacking in others. They either interact with substrate proteins directly or serve as docking sites for adaptor proteins that deliver their bound cargo to their cognate AAA+ protein [4,5]. In the present paper, we describe and compare the activities of prokaryotic AAA+ proteins that have crucial functions in protein quality control systems. A special emphasis is given to family members that function as a disaggregase by solubilizing stress-induced protein aggregates.
AAA+ proteins in protein degradation: feeding peptidases with substrates
The activity of every protein depends on the correct folding into its native structure. The structural integrity of a protein depends on physical and chemical parameters. Extracellular stress factors, such as changes in temperature or oxidative stress, can interfere with the folding state of proteins, leading to protein inactivation and the accumulation of misfolded protein species that can affect cellular processes. To cope with these stresses, all cells have evolved quality control systems consisting of molecular chaperones and ATP-dependent proteases. Both of these protein classes monitor the folding state of substrates and either refold or degrade them. Proteolysis of misfolded proteins in the cytosol of prokaryotes is mainly mediated by proteasome-like machines that consist of an AAA+ protein (e.g. ClpA, ClpC or ClpX) and a proteolytic component that is either covalently attached (e.g. Lon) or diffusible (e.g. ClpP) . These proteolytic systems are also operative in controlling signal transduction pathways by degrading native regulatory proteins that harbour specific destruction tags. The peptidase ClpP consists of two heptameric rings that form a barrel-shaped proteolytic core with the active sites hidden in an interior chamber. Access to these sites is controlled by narrow pores that do not allow the passage of folded polypeptides . Complex formation between ClpP and the co-operating AAA+ protein requires a conserved tripeptide ‘P-element’ (Ile-Gly-Leu/Phe) that is located on an exposed loop at the bottom of the AAA domain . This mode of complex formation generates a continuous channel starting at the pore entrance of the AAA+ protein and ending within the proteolytic chamber of ClpP (Figure 1A).
Functions and mechanisms of bacterial AAA+ proteins in protein quality control systems
Substrate selection is mediated by the AAA+ protein component. To allow transfer of bound substrates into the associated peptidase, AAA+ proteins possess an ATP-dependent threading activity. This threading activity is largely mediated by conserved aromatic residues that are located on mobile loops at the central pore of the AAA+ protein. These residues are proposed to act as molecular clamps by binding and releasing substrates in a co-ordinated fashion. It is suggested that nucleotide-driven movements of the aromatic residues generate a pulling force, leading to substrate threading and unfolding in a coupled event, enabling the transport of the unfolded polypeptide through the axial channel (Figure 1A) . AAA+ protein variants that carry mutations of the aromatic residues have lost both threading and unfolding activity and cannot feed associated peptidases with substrates [9–11]. Notably, the AAA+ components of proteolytic complexes can also function independently, and exhibit chaperone activity in the absence of peptidases. However, previous studies have demonstrated mutual communication between both co-operating partners, pointing to an interdependent evolution of AAA+ proteins and their respective peptidases [12,13].
ClpB: an AAA+ protein which rescues proteins from the aggregated state
Severe stress conditions, such as extreme heat, can overburden the capacity of cellular protein quality control systems, leading to accumulation of misfolded proteins, followed by their aggregation. Protein aggregation has been viewed for a long time as a dead-end road in the life of proteins. However, cells are able to solubilize and even reactivate aggregated proteins, as demonstrated first by Lindquist and co-workers for Saccharomyces cerevisiae . Central to this remarkable activity is the co-operation of two chaperones, the Hsp (heat-shock protein) 70 chaperone system and the AAA+ protein ClpB, which together form a bi-chaperone system to reverse protein aggregation [16–19]. It is important to note that each component of this bi-chaperone system alone exhibits only weak (Hsp70) or no (ClpB) disaggregation activity. The disaggregation activity of ClpB–Hsp70 is essential for the ability of cells to survive transient extreme stress conditions, a phenomenon referred to as thermotolerance [20,21].
Genes encoding orthologues of ClpB are found in most bacteria, as well as in parasitic protozoa, yeast (S. cerevisiae Hsp104) and plants (Arabidopsis thaliana Hsp101). Thermotolerance development in all these organisms relies on the disaggregation activity of ClpB [16,20,22]. The lack of large-scale mobility and more severe exposure to environmental stress conditions can be interpreted as the common property of organisms harbouring a ClpB orthologue. Intriguingly, ClpB orthologues are absent from animal genomes. Expression of Hsp104 in human T-cells results in both increased resistance to heat stress and increased recovery of heat-aggregated firefly luciferase, suggesting that human cells possess only limited disaggregation activity .
Substrate threading: a conserved, unifying activity of AAA+ proteins involved in protein quality control
In contrast with AAA+ proteins involved in proteolysis, ClpB does not harbour the P-element and, accordingly, does not associate with the peptidase ClpP. Consequently, it was unclear whether ClpB possesses a threading activity, like peptidase-co-operating AAA+ proteins, or whether its unique function in protein disaggregation and co-operation with an Hsp70 chaperone relies on a different mechanism for substrate remodelling. An initial hint as to the nature of the operative mechanism came from analysis of the disaggregation process. Protein disaggregation results in the continuous release of single unfolded polypeptides from the aggregate, a process that can be explained by one-by-one extraction of substrates as a result of ClpB-dependent threading . Meanwhile, direct evidence for the existence of a ClpB threading activity has been independently provided by several groups. First, construction of BAP (ClpB/ClpA/P-loop), a ClpB variant that contains the P-element, which could associate and co-operate with ClpP, resulting in the degradation of soluble misfolded model substrates . The ease with which ClpB could be converted from a refolding into a degrading AAA+ protein reflects the close structural and functional relationship between members of this protein class and demonstrates the existence of a general threading activity (Figure 1B). The demonstration of a conserved threading activity that is shared by ClpB and AAA+ proteins which function in proteolysis also explains the ability of ClpP-co-operating ClpA/ClpC to degrade aggregated proteins in vitro [35,36]. Furthermore, processing of soluble substrates by BAP–ClpP complexes was independent of Hsp70, indicating that substrate threading must be an integral and autonomous function of ClpB. Secondly, conserved pore-located aromatic residues that have crucial functions in substrate threading by peptidase-co-operating AAA+ proteins also exist in each AAA domain of ClpB (Tyr251and Tyr653 respectively in Escherichia coli ClpB) [24–26]. Mutagenesis of these residues demonstrated their crucial function in both substrate threading and aggregate solubilization [24,25]. Comparable results were obtained by Glover and co-workers, who demonstrated that Hsp104 variants with mutations in conserved pore residues are affected in protein disaggregation . Thirdly, cryo-EM (cryo-electron microscopy) analysis of Thermus thermophilus ClpB hexamers in different nucleotide states recently revealed conformational changes of conserved aromatic residues (Tyr251 in E. coli ClpB) positioned at the entrance pore of the translocation channel . It has been demonstrated that Tyr251 becomes exposed at the surface of the AAA-1 domain in the ATP-activated state, forming a single high-affinity substrate-binding site [25,26]. The structure of p[NH]ppA (adenosine 5′-[β,γ-imido]triphosphate)-bound ClpB, which is suggested to represent a transition state of ATP hydrolysis, does not show exposure of Tyr251, but instead reveals a large opening of the central pore from 13 to 28 Å (1 Å=0.1 nm) . This conformational change is potentially accompanied by a downward motion of Tyr251 and might lead to threading and unfolding of bound substrates. Finally, an unfolding activity of ClpB–Hsp104 towards tightly folded GFP (green fluorescent protein) carrying an unfolded polypeptide tag for AAA+ protein recognition has been reported recently [28,29]. This ClpB–Hsp104 activity is reminiscent of the function of peptidase-co-operating AAA+ proteins in the unfolding of native proteins containing degradation tags [30–32]. However, in contrast with AAA+ proteins involved in proteolysis, specific conditions were required to unleash the unfolding activity of ClpB–Hsp104. Thus substrate processing was only observed in the presence of mixtures of ATP and its non-hydrolysable analogue ATP[S] (adenosine 5′-[γ-thio]triphosphate). Alternatively, ClpB–Hsp104 mutants with defective ATPase activity at either the AAA-1 or AAA-2 domain could trigger substrate remodelling [28,29]. Since processing of the soluble GFP substrate was not observed for wild-type ClpB–Hsp104 proteins in the presence of ATP, it is suggested that the specific conditions mentioned above achieve a balance of substrate binding, holding and unfolding, ensuring that substrates are not prematurely released without unfolding. Surprisingly, conditions that allowed for substrate unfolding by ClpB–Hsp104 were shown to largely impair protein disaggregation and thermotolerance development [33,34]. Consequently, it is unclear whether the unleashed ClpB unfolding activity towards folded domains is functional during protein disaggregation, or whether it reflects an off-pathway that has been erased during evolution for wild-type ClpB–Hsp104 to allow co-operation with an Hsp70 chaperone. In this context, it will be interesting to monitor the destiny of native GFP as part of an aggregate in ClpB–Hsp70-mediated disaggregation.
Hsp70: the partner makes the difference
Incubation of aggregated proteins with ClpB alone does not lead to detectable changes within an aggregate. Instead, aggregate solubilization by ClpB is only achieved in the additional presence of the Hsp70 chaperone system. The coupling of both chaperones is still poorly understood; however, recent findings have started to illuminate their co-operative mechanism.
Various independent findings support an essential function of Hsp70 at initial stages of the disaggregation reaction (Figure 1B). First, degradation of aggregated proteins by BAP–ClpP still requires the presence of Hsp70 . Along the same line, binding of aggregated proteins to the central pore of either the AAA-1 or AAA-2 domain of ClpB is Hsp70-dependent. Notably, the interaction of soluble misfolded substrates (e.g. casein) with the ClpB translocation channel and their degradation by BAP–ClpP did not require Hsp70, demonstrating that the co-operating Hsp70 chaperone has a specific function in the processing of aggregated proteins [24,38]. Secondly, kinetic analysis revealed that the interaction of Hsp70 with aggregated proteins represents the rate-limiting step of the initial disaggregation process . Thus pre-incubation of protein aggregates with DnaK, the major prokaryotic Hsp70, and its DnaJ co-chaperone resulted in faster subsequent ClpB–DnaK-mediated substrate refolding. Thirdly, Hsp70, in contrast with ClpB, displays disaggregation activity by itself and can reactivate some aggregated proteins [16,40–43]. The structural characteristics of a protein aggregate that determine the additional need of ClpB for solubilization are poorly understood. It had been suggested originally that aggregate size was a crucial parameter, since Hsp70 was capable of solubilizing small-sized aggregates alone, but not large-sized aggregates of the same model substrate . In contrast, recent results indicate that not only aggregate size, but also the conformational properties of the aggregated substrate determines whether ClpB is required for the disaggregation process: Liberek and co-workers noticed a correlation between the content of β-structures in aggregates and ClpB requirement . Whereas disordered aggregates of luciferase were efficiently solubilized by Hsp70 alone, aggregates with a high β-structure content were only processed in the additional presence of ClpB . Such aggregates may also be generated during heat stress in vivo, explaining the crucial function of ClpB in protein disaggregation and thermotolerance development.
The role of Hsp70 is probably not restricted to the first steps of the disaggregation process. Hsp70 might also be required for binding with substrates after they have been threaded through the axial channel of ClpB. Such activity would further ensure reactivation of aggregated proteins in a cellular context, which is obligatory for thermotolerance [15,24].
The precise mode of ClpB–Hsp70 co-operation is largely unknown. Both chaperones may work in a consecutive manner: the initial action of Hsp70 could lead to the remodelling of aggregates that are processed further by ClpB at a second stage. However, the theory that a physically linked bi-chaperone complex acts as the active disaggregation machinery is supported by the observation that co-operation between the two chaperone systems exists only for proteins of the same species [16,45,46]. Evidence for such a complex was provided for T. thermophilus ClpB–Hsp70; however, the region(s) that mediate ClpB–Hsp70 interaction remain to be identified .
M-domains: coupling the ClpB threading motor with the Hsp70 activity
Given that other AAA+ proteins unfold and thread substrates, but are not involved in protein disaggregation in vivo, substrate threading might not be sufficient to verify ClpB as the central cellular disaggregase. It is tempting to speculate that the divergent evolution of ClpB resulted in the possession of a specific mechanism, which reflects an adaptation to its particular function. Clearly, ClpB must possess extra features which enable it to be such a potent disaggregase.
ClpB harbours two additional domains, the N-domain (N-terminal domain) and the unique M-domain, which are prime candidates for providing such extra activity. N-domains are connected via flexible linkers to the AAA-1 domain ring and provide additional anchor points for aggregate binding, thereby facilitating protein disaggregation [47,48]. However, N-domains are not essential for ClpB function [34,49–51]. In contrast, the ClpB-specific M-domain is strictly required for protein disaggregation [34,52–55].
The M-domain is inserted into the AAA-1 domain and forms a large coiled-coil structure that is positioned adjacent to AAA-1 and located on the outer surface of the ClpB molecule [26,55]. M-domains exhibit nucleotide-controlled conformational changes during the ClpB ATPase cycle. They become visible as long spokes in the AAA-1 domain ring of p[NH]ppA-bound ClpB and splay out radially from the ClpB barrel . In the ATP-activated state, M-domains adopt a more leaning position, whereas they are not visible in cryo-EM structures of ADP-bound ClpB. Whereas the mobility of M-domains has been demonstrated to be crucial for protein disaggregation, it is still unknown how these dramatic conformational changes affect substrate processing or Hsp70 co-operation [38,55,56].
M-domains were originally proposed to act as a molecular ‘crowbar’, enabling ClpB to convert larger protein aggregates into smaller ones [16,55]. However, direct evidence for interactions between M-domains and substrates, and for the formation of smaller protein aggregates by a crowbar mechanism, is lacking. Furthermore, the co-operation of Hsp70 and ClpB in the monomerization of dimeric substrates also argues against M-domain-mediated fragmentation activity [33,57]. Instead, M-domains could either control substrate threading or the co-operation with Hsp70.
The sequence homology among ClpB M-domains is rather low and restricted to two short regions, termed motifs 1 and 2 . Previous studies have illuminated the specific function of M-domains by analysing the consequences of alterations in motif 2 of the M-domain [38,53,54]. Such ClpB variants were affected in the ATPase cycle and exhibited a dominant gain-of-function phenotype, unleashing a toxic activity that is not observed for wild-type ClpB. These findings suggest a regulatory function of M-domains, and point to intimate contacts between M-domain motif 2 and the ClpB AAA domains. Indeed, this region is positioned between the AAA-1 domains of neighbouring subunits, potentially facilitating co-operative interactions in the ClpB ring . Intriguingly, ClpB M-domain mutants were only affected in protein disaggregation, whereas the processing of soluble misfolded polypeptides remained unaffected, demonstrating that the basic ClpB threading motor for substrate unfolding remained functional . This finding provided the first evidence for a function of M-domains in establishing co-operation with Hsp70. Further analysis revealed that M-domains are required to specifically allow the Hsp70-dependent shuffling of aggregated proteins to the ClpB pore site. Such a mechanism establishes the M-domain as a regulatory device that acts by coupling the initial Hsp70 chaperone activity with the ClpB threading motor, leading to efficient protein disaggregation .
Seventh International Meeting on AAA Proteins: Independent Meeting held at the Royal Agricultural College, Cirencester, U.K., 9–13 September 2007. Organized and Edited by John Mayer (Nottingham, U.K.) and Paul Freemont (Imperial College London, U.K.).
This work was supported by grants from the Deutsche Forschungsgemeinschaft (Bu617/14-3/4), the Fond der Chemischen Industrie to B.B., and a Heisenberg Fellowship of the DFG (Deutsche Forschungsgemeinschaft) to A.M.