Many complex cellular processes in the cell are catalysed at the expense of ATP hydrolysis. The enzymes involved bind and hydrolyse ATP and couple ATP hydrolysis to the catalysed process via cycles of nucleotide-driven conformational changes. In this review, I illustrate how smFRET (single-molecule fluorescence resonance energy transfer) can define the underlying conformational changes that drive ATP-dependent molecular machines. The first example is a DEAD-box helicase that alternates between two different conformations in its catalytic cycle during RNA unwinding, and the second is DNA gyrase, a topoisomerase that undergoes a set of concerted conformational changes during negative supercoiling of DNA.
- ATP hydrolysis
- C-terminal domain
- fluorescence resonance energy transfer (FRET)
- RNA-binding domain
- RNA-recognition motif
Conformational cycles underlie the function of molecular machines
Many endergonic reactions in the cell are catalysed using the energy liberated in the exergonic reaction of ATP hydrolysis. The chemical energy from ATP hydrolysis is coupled to these processes via enzymes that bind and hydrolyse ATP. While the molecular details of the energy coupling remain elusive, it is commonly accepted that cycles of nucleotide-driven conformational changes in these molecular machines are key to the coupling process. In line with this concept, structural studies of ATP-dependent enzymes have revealed distinct conformations in different nucleotide states. In the following, two examples will illustrate how ATP-dependent conformational changes drive enzymes that modify nucleic acid structure. The first example is a DEAD-box helicase that alternates between two different conformations in its catalytic cycle during RNA unwinding, and the second is DNA gyrase, a topoisomerase that undergoes a set of concerted conformational changes during negative supercoiling of DNA.
The DEAD-box protein YxiN: a conformational switch mediates ATP-driven RNA unwinding
RNA helicases catalyse the ATP-dependent structural rearrangement of RNA and RNA–protein complexes, and are implicated in virtually all RNA-mediated processes (reviewed in ). DEAD-box helicases are the largest family of RNA helicases (for a review see ). They share a helicase core comprising two flexibly linked RecA domains. The core carries all conserved motifs involved in ATP binding and hydrolysis, RNA binding and coupling of ATP hydrolysis to RNA unwinding and constitutes a minimal helicase. Many DEAD-box proteins contain N- and C-terminal regions flanking the helicase core that mediate interactions with their RNA substrate or with proteins, and may be directly involved in duplex separation [3–8].
Conformational changes in the nucleotide cycle
Structural studies on DEAD-box proteins have mainly been restricted to the helicase core. The structures show a similar fold of the two RecA-like domains, but the juxtaposition of these domains differs substantially between different DEAD-box proteins. YxiN, a DEAD-box helicase from Bacillus subtilis, is involved in ribosome biogenesis. Its substrate specificity is mediated by an RBD (RNA-binding domain) following the helicase core . This domain specifically binds to hairpin 92 in the 23S ribosomal RNA via an RRM (RNA-recognition motif) . Using smFRET (single-molecule fluorescence resonance energy transfer), and YxiN labelled with donor and acceptor dyes on opposite sides of the inter-domain cleft in the helicase core (Figure 1), we have shown that the helicase core of the DEAD-box protein YxiN adopts an open conformation in the absence of RNA and nucleotide with an open cleft between the RecA domains [11,12]. This conformation resembles the structure of the DEAD-box protein DeaD from Methanococcus jannaschii , and is significantly more closed than many other DEAD-box protein structures . Co-operative binding of ATP and RNA induces a conformational change in the YxiN helicase core that leads to the closure of the cleft between the RecA domains. Distances determined from smFRET experiments demonstrate that the closed conformation resembles the structures of various DEAD-box proteins in complex with single-stranded RNA and ATP analogues [8,14–19]. The formation of a compact structure leads to extensive contacts at the interface of the two RecA domains. In this closed conformation, both the catalytic site for ATP hydrolysis and a continuous RNA binding are formed by residues from both domains, rationalizing the co-operative binding of RNA and ATP and the RNA-stimulated ATPase activity common to DEAD-box proteins. In smFRET experiments, we have followed the conformation of the YxiN helicase core through all functionally relevant states in the nucleotide cycle (Figure 1). Using ADP·BeFx and ADP·MgFx as substrate and product state analogues, we could demonstrate that the helicase core maintains the closed conformation during ATP hydrolysis . In contrast, the helicase core is predominantly in the open conformation in the presence of ADP and RNA. Consequently, phosphate release subsequent to ATP hydrolysis must trigger re-opening the inter-domain cleft. Phosphate release has previously been described as one of the rate-limiting steps in the catalytic cycle of the homologous DEAD-box protein DbpA from Escherichia coli . Re-opening of the inter-domain cleft disrupts the bipartite RNA-binding site formed by both RecA domains, and the concomitant reduction in RNA affinity leads to RNA release. Overall, DEAD-box proteins are thus nucleotide-dependent switches that are in the ‘on’-conformation in the ATP state, and in the ‘off’-conformation in the ADP state [12,20,22].
Rationalizing the effects of mutation: coupling
Structural data indicate that formation of the closed conformer is coupled to a local distortion of the RNA, and this local destabilization has been suggested as a first step of RNA unwinding . Using smFRET, we have demonstrated that mutations preventing closure of the helicase core lead to unwinding deficiency , possibly because the initial local destabilization of the RNA substrate is lacking. Strikingly, an ATPase-deficient mutant of YxiN still adopts the closed conformation, but is unable to catalyse RNA unwinding. Thus the closure of the cleft in the helicase core is not sufficient for RNA unwinding to occur , indicating that a critical step subsequent to the conformational change is required. Such an ‘activation step’ has previously been proposed from biochemical data , and it has been proposed that nucleotide analogues that do not support RNA unwinding are not capable of inducing the transition to the activated complex. However, a recent study has revealed virtually identical structures of a DEAD-box helicase in complex with p[NH]ppA (adenosine 5′-[β,γ-imido]triphosphate), ADP·BeFx and ADP·AlFx , and yet only ADP·BeFx supports single turnover RNA unwinding . The molecular nature of this activation thus remains unclear.
The current knowledge can be integrated into a model for the mechanism of RNA unwinding by DEAD-box helicases (Figure 1) that links nucleotide-driven conformational changes with structural changes in the RNA, and rationalizes the effect of mutations and nucleotide analogues on the unwinding reaction . The conformational change in the helicase core, with an ‘on’-switch upon ATP and RNA binding, and an ‘off’-switch upon phosphate release, is at the heart of ATP-dependent RNA unwinding by DEAD-box proteins.
Finding the missing piece: localization of the RBD
The helicase core constitutes a degenerate nucleotide-regulated conformational switch that supports a basic RNA-unwinding activity. Domains flanking the helicase core define the specific function of a particular DEAD-box protein, e.g. by anchoring it to a specific RNA substrate. In addition to high-affinity RNA binding, these extensions may also position the RNA substrate for unwinding by the helicase core [6,24] or actively contribute to local destabilization of RNA substrates .
The YxiN RBD C-terminal to the helicase core is required for high-affinity binding of hairpin 92 in 23S ribosomal RNA . The crystal structure of the isolated RBD revealed an RRM fold , and the structure of the RBD in complex with a ribosomal RNA fragment has recently revealed the molecular details of RNA binding . While studies have addressed the global shape of YxiN  and its E. coli homologue DbpA , the orientation of the RBD with respect to the helicase core remained unknown. To provide a framework for understanding the functional co-operation of the YxiN helicase core and the RBD, we mapped the orientation of the RBD in smFRET experiments . Distance restraints between a cysteine residue in the N- and C-terminal RecA domains and various cysteine residues introduced into the RBD were determined in smFRET experiments. The structure of the isolated RBD (PDB code 2G0C) was positioned relative to the helicase core, taking into account the distance information obtained. Two different orientations of the RBD satisfy the experimental distance restraints, but only one orientation is favoured by the available biochemical data. In this orientation, the RBD covers a flexible patch on the C-terminal RecA domain formed by loops. The same patch mediates the interaction of eIF4G (eukaryotic initiation factor 4G) with eIF4A . It may thus constitute an adaptable interface for interactions of the helicase core with flanking domains or with protein partners.
Outlook: regulation of the helicase core
smFRET experiments have provided a clear picture of global conformational changes in the catalytic cycle of DEAD-box proteins. Future experiments will have to address the modulation of the helicase core activity by additional domains or its regulation by other factors.
DNA gyrase: a gated complex
DNA topoisomerases catalyse the inter-conversion of DNA topoisomers and affect key cellular events such as replication, recombination and transcription . Bacterial DNA gyrase introduces negative supercoils into DNA in an ATP-dependent reaction [30,31]. In the active form of the enzyme, a heterotetramer formed by two GyrA and two GyrB subunits, two cavities are delimited by three (transient) protein interfaces, termed gates. Supercoiling occurs by a strand passage mechanism [32–34]: First, a dsDNA (double-stranded DNA) segment binds to the DNA-gate in the centre of the heterotetramer, in the vicinity of the catalytic tyrosine residues. Both strands of this DNA (G-segment) are cleaved, and the 5′-ends remain covalently attached to the tyrosine residues. A second dsDNA segment (T-segment) is captured in the upper cavity by a nucleotide-induced closure of the N-gate, formed by the GyrB subunits. Opening of the DNA-gate then generates a gap in the G-segment, through which the T-segment is transported into the lower cavity. Finally, the T-segment leaves the lower cavity through the C-gate. The G-segment is religated, and dissociates from the enzyme. Directed strand-passage thus requires the regulated opening and closing of three protein interfaces, the N-gate, the DNA-gate and the C-gate. In addition, it may be guided by the CTDs (C-terminal domains) of GyrA. These domains adopt a β-propeller type structure, and wrap DNA by binding it along their circumference . The mechanism of co-ordinating these concerted conformational changes is currently unknown.
Gate-DNA and the DNA-gate
To address conformational changes of the DNA-gate that has to open to allow for strand passage, we designed a model DNA with donor and acceptor dyes flanking a preferred cleavage site for gyrase. Using this model substrate, we have recently shown that DNA bound to the DNA-gate of gyrase is distorted from regular B-form DNA  (Figure 2). The DNA distortion is coupled to DNA cleavage: A cleavage-deficient gyrase mutant still binds the gate-DNA, but does not alter its DNA conformation. On the other hand, gyrase cleavage complexes can be populated with DNA substrates containing a phosphorothiolate-modification that promotes the formation of cleavage-complexes by disfavouring ligation. Such a phosphorothiolate-modified DNA shows a larger fraction of the distorted conformer in complex with gyrase compared with non-modified DNA. Interestingly, the distortion of the DNA is already observed in the absence of nucleotides, indicating that it precedes nucleotide-induced closure of the N-gate. Ciprofloxacin, a gyrase inhibitor that populates cleavage complexes, counteracts the DNA distortion .
The conformation of the gyrase DNA-gate was directly monitored in experiments with gyrase that was labelled with donor and acceptor on opposite sides of the DNA-gate . These experiments revealed that the default state of the DNA-gate is the closed conformation, as evidenced by GyrA structures [37,38]. The DNA-gate remains predominantly closed when gyrase binds to the short gate-DNA . Experiments with short DNA oligonucleotides reflect a situation in the absence of T-segment, and thus in the absence of strand passage, limiting the possible conclusions for the functional cycle of gyrase. Therefore we also followed the conformation of the gyrase DNA-gate during DNA relaxation and during ATP-dependent DNA supercoiling . Again, the DNA-gate is predominantly in the closed conformation (Figure 2). Altogether, our results thus suggest that the DNA-gate only transiently opens to allow for strand passage. This is in agreement with biochemical data that find very low fractions of cleavage complexes during the gyrase reaction . Presumably, the distortion of the gate-DNA is accompanied by local conformational adjustments of GyrA in the region surrounding the DNA-gate that do not lead to a measurable change in FRET efficiency. Via such conformational adjustments, the DNA distortion may be linked to an ‘unlatching’ of the DNA-gate, preparing it for transient release as soon as a T-segment is captured. Such a hierarchical sequence of events would provide a strict coupling of gate opening to strand passage. However, it is currently unknown how the distortion of the DNA and the ‘unlatching’ and opening of the DNA-gate are co-ordinated with N-gate closure and T-segment capture, or C-gate opening and T-segment release.
Outlook: co-ordination of opening and closing of different gates
To dissect the individual conformational changes and to gain insight into their co-ordination, we have investigated the conformational dynamics of the N-gate (A. Gubaev and D. Klostermeier, unpublished work). The crystal structure of a GyrB fragment in complex with p[NH]ppA shows an intertwined dimer , suggesting a nucleotide-induced dimerization. Subsequent biochemical studies have established that the N-gate functions as an ‘ATP-operated clamp’ , and support its conceived role in the postulated strand passage mechanism [32,42]. smFRET experiments with gyrase labelled with donor and acceptor dyes at the N-gate confirm that the N-gate closes upon binding of the nonhydrolysable ATP analogue p[NH]ppA to the GyrB subunits. Preliminary results point towards an active role of the CTDs in presenting the T-segment and in co-ordinating N-gate closure (A. Gubaev, M.A. Lanz and D. Klostermeier, unpublished work). Future studies that address the sequence of the individual conformational changes and their co-ordination will reveal how strand passage towards negative DNA supercoiling is achieved.
Work in my laboratory was funded by the VolkswagenStiftung, the Swiss National Science Foundation and the NCCR Nanoscale Sciences.
I thank all the members of my laboratory for their contributions.
NACON VIII: 8th International Meeting on Recognition Studies in Nucleic Acids: An Independent Meeting held at The Edge, University of Sheffield, Sheffield, U.K., 12–16 September 2010. Organized by Mike Blackburn, Mark Dickman, Jane Grasby, David Hornby, Chris Hunter, John Rafferty, Jim Thomas, David Williams and Nick Williams (Sheffield, U.K.).
Abbreviations: CTD, C-terminal domain; dsDNA, double-stranded DNA; eIF, eukaryotic initiation factor; p[NH]ppA, adenosine 5′-[β,γ-imido]triphosphate; RBD, RNA-binding domain; RRM, RNA-recognition motif; smFRET, single-molecule fluorescence resonance energy transfer
- © The Authors Journal compilation © 2011 Biochemical Society