Chance, destiny, and the inner workings of ClpXP

Abstract

AAA+ proteases are responsible for protein degradation in all branches of life. Using single molecule and ensemble assays, Cordova et al. (2014) investigate how the bacterial protease ClpXP steps through a substrate’s polypeptide chain and construct a quantitative kinetic model that recapitulates the interplay between stochastic and deterministic behaviors of ClpXP.

All organisms face the challenge of degrading a large number of diverse protein substrates with exquisite selectivity. The proteases that are used to meet this challenge feature a central proteolytic chamber capped by a ring of six ATPase subunits of the AAA+ superfamily. To function properly, the ATPases must unfold protein substrates that differ in amino acid sequence, structure, and stability, and translocate them into the proteolytic chamber without releasing partially processed intermediates. Although a wealth of structural and biochemical studies have provided insights into these processes (Nyquist & Martin 2014), we still lack a coherent picture of how a chambered protease works. In this issue, Cordova et al. (2014) present an elegant set of experiments that describe an intriguing behavior of the bacterial protease ClpXP and define a quantitative kinetic model for protein unfolding and translocation.

At the mechanistic level, the best understood chambered protease is ClpXP. The operation of an AAA+ ATPase depends on coupling between the chemical steps of ATP hydrolysis and the mechanical steps that produce directional motion. In the archetypal F1-Fo ATPase, ATP hydrolysis and mechanical steps are tightly coupled, meaning conformational changes do not occur when ATP hydrolysis is blocked and vice versa (Kinosita et al. 2000; Stinson et al. 2013). This feature allows the ATPase to convert energy very efficiently. In contrast, the ClpXP motor functions with much looser coupling. The motor typically undergoes many futile ATPase cycles as it attempts to unfold a substrate (Kenniston et al. 2003), and it can hydrolyze ATP even when conformational changes are impaired by a covalent crosslink, indicating that ATP hydrolysis and conformational changes can be uncoupled (Stinson et al. 2013). The ATPases do not function in a strictly sequential manner, and the motor can hydrolyze ATP even with dead subunits in the ring (Martin et al. 2005). Nevertheless, the subunits clearly interact with each other because mutations in one subunit affect the ATPase activity of others and reduce the ability of ClpXP to unfold and degrade proteins (Stinson et al. 2013; Martin et al. 2005; Cordova et al. 2014.).

Here, Cordova et al. combine single molecule optical trapping and single turnover and steady state ensemble approaches to characterize how ClpXP moves along a protein substrate. They measure the unfolding of a model protein substrate and the kinetics of individual steps during translocation. These steps vary in size from 1 nm to 4 nm and do not occur in a defined order. However, the motor appears to have some memory because steps of similar sizes cluster so that a 1 nm step is more likely to be followed by another 1 nm step. Another surprise was that the machine can take 4 nm steps, presumably reflecting conformational changes in four subunits, even with only two ATPase subunits that can hydrolyze ATP. These observations lead to a model with an intriguing interplay between stochastic and deterministic behaviors. The disparate sizes of translocation steps and the lack of a defined order imply stochastic behavior, which is reflected in the model as a translocation step beginning with ATP hydrolysis by any of the three or four subunits that bind ATP at any given time (Figure 1). Based on the ClpXP structure, the ATPase-linked conformational changes in a single subunit are likely to produce only 1 nm translocation steps. Thus, the larger steps are proposed to arise from coordinated conformational changes in neighboring subunits that produce additional, unresolved 1 nm substeps, each accompanied by ATP hydrolysis. The key feature of the model is that the ATPase activity and conformational changes propagate directionally around the ring. Consequently, the initial stochastic selection of a subunit determines how many additional subunits can participate in the coordinated step and thus produces the range of sizes for the translocation steps. The memory effect is also reflected in the model. Steps of 2 nm or more have to be followed by conformational changes and ATP-binding to reset the ring before the next step can be taken, whereas a 1 nm step can be followed rapidly by another 1 nm step without the ring resetting.

Figure 1. Substrate Degradation by the ClpXP protease.

The ClpX ATPase ring is shown with ATP-bound subunits in red and free subunits in grey. The dice faces symbolize the finding that any subunit loaded with ATP can initiate a chain of ATP hydrolysis reactions. The proteolytic ClpP rings are depicted in grey; their proteolytic sites are buried at the center and therefore not visible.

What do these properties mean for the functions of ClpXP? It is possible that the large steps could increase the overall translocation rate, which might be important when the initial translocation steps for some substrates are in kinetic competition with local protein refolding, whereas the small steps could be useful for maintaining a tighter grip on certain substrates. It is also possible that the stochastic variations in translocation step sizes and rates, together with the imperfect coupling, allow the motor to accommodate irregular spacing of the features it recognizes in its substrates. This behavior appears to differ fundamentally from that of related ATPase hexamers such as the papillomavirus DNA helicase E1, which uses a continuous cycle of highly coordinated conformational changes to translocate DNA through the ring, perhaps because DNA presents a more uniform track (Enemark & Joshua-Tor 2006). It will be interesting to determine whether the basic features of the Cordova model are present in other chambered proteases. The stochastic and deterministic behaviors of ClpXP are well established although other mechanistic explanations have been proposed (Sen et al. 2013), but studies on the archaeal proteasome (Smith et al. 2011) and mitochondrial proteases (Augustin et al. 2009) have emphasized coordinated behavior. Future investigations will show whether these discrepancies are due to true differences in the way the proteases operate or reflect emphases on different aspects of the same general mechanism.

The paper also raises new questions. Intriguingly, Cordova et al. notice that the degradation rate measured in single molecule experiments is consistently higher than that in ensemble experiments. In the single molecule experiments the polypeptide substrate is prethreaded into the ATPase ring, implying that the rate-limiting step in solution is a commitment step at the protease fully engages its substrate. The commitment step may provide a way to balance specificity and processivity and reduce the risk of the protease becoming blocked by an efficiently targeted by hard-to-unfold substrate. Further work will be necessary to probe the mechanics of this step and its functional roles.

In addition, the authors point out that although their detailed kinetic model recapitulates the known behaviors of ClpXP, it is not strictly determined by the data and some of the rate constants in the model are not constrained by individual experiments but chosen ad hoc to maximize the agreement with observations. Indeed, a key value of this concrete model is that it provides the framework for critical tests of the proposed mechanism and will serve as a roadmap for further dissection of this fascinating protein machine.