GTP and ATP hydrolysis in biology

ATP is the universal energy coin of the cell and the cleavage of one of the phosphoanhydride bonds drives a large number of metabolic reactions and other energy requiring processes. The chemical mechanism of NTP hydrolysis has been and still is a matter of intensive debate between the “associative” and the “dissociative” communities. While ATP/GTP hydrolysis in water and without enzymes has been shown to clearly proceed by a dissociative mechanism, this is less clear for enzymatic phosphoryl transfer, and may not be proven one way or the other experimentally. Much less focus has been directed towards the structural design of the active site of phosphor-transfer enzymes, the comparison between different types and families of NTP converting enzymes1, the biological implication of such conversion, the nature of the rate-limiting step and the structure at the active site close to the transition state. In this special issue we focus on the comparison between catalytic mechanisms of proteins hydrolyzing ATP and GTP, putting together as many examples as suitable for one issue of the Journal. We neglect metabolizing enzymes and highlight systems which regulate and drive important biological processes such as signal transduction, motility or biosynthesis and degradation. At first sight the biological consequences of GTP or ATP hydrolysis seem to be different, since GTP hydrolysis is mostly used for controlling signal transduction and trafficking pathways, with no force generation or any other obviously energy consuming process involved. GTP binding proteins do this by regulating, via the GDP-GTP cycle, the interaction with effectors, whereas ATP hydrolysis is used for force generation or other energy requiring process. However, the proteins of the dynamin superfamily also use the energy of GTP hydrolysis for generating the force required for remodeling membranes. A notable aspect of common features of ATP versus GTP hydrolysis systems is that the ATP hydrolyzing proteins myosin and kinesin are closely related to the GTP-binding proteins, also by sequence comparison. Both use a similar structural change in the switches for either regulation of interaction with effectors or by using the conformational change and the energy difference between the ATP and ADP-bound states for the power stroke (see article by Müller and Goody). These two motility generating systems carry out the power stroke by switching between low and high affinities states to actin and microtubules, respectively, albeit with interesting strategic differences. The most frequent motif found in the proteins described here is the P-loop (Saraste et al.,1990)2 (see Figure on the cover of this issue), with the conserved sequence motif GxxxxGKS/T. It is also called the Walker A motif, since it was first described by John Walker (Walker et al., 1982)3, 4. It is a conserved structural motif connecting a β-strand with an α-helix and wrapping around and neutralizing the negative charge of the phosphates. Almost all the GTP and ATP hydrolyzing systems presented here belong to this class and are obviously evolutionarily related. All proteins described here share a three-layered structure in their NTP binding domain, with a mostly parallel β-sheet and α-helices on both sides (Leipe et al., 2002)4. The rates of hydrolysis between different P-loop proteins can vary considerably. While the metabolizing enzyme adenylate kinase (also a P-loop protein) has phosphoryl transfer rates on the order of 1000 s−1, regulatory proteins such as Ras hydrolyse GTP with a rate of around 4x10−4 s−1 or in some cases even lower, with intermediate rates for most of the systems presented here. However, in the biological context the nucleotide turnover rates can be stimulated dramatically, in the case of small G proteins of the Ras superfamily by the GTPase activating proteins (GAPs), by interaction of myosin with actin, of kinesin with microtubules, by homo or hetero-oligomerization in the case of dynamin or AAA+ ATPases. In all these cases the firing of the hydrolysis reaction or the release of products is tightly regulated to avoid futile cycles of NTP binding and hydrolysis. In the case of most small G proteins of the Ras superfamily, two residues involved in GTP hydrolysis are almost universally conserved as Gln and the so-called arginine finger (see Mishra and Lambright). However there is variation in the way these residues come together in the active site when GAPs interact with their cognate G proteins. There one finds almost all combinations of Gln being in cis or trans. In heterotimeric G proteins the Arg finger is intrinsic to the Gα protein and GTPase acceleration is achieved by a particular GAP called RGS (see Sprang). All translation factors involved in protein biosynthesis are regulated by GTP binding proteins (trGTPases) such EF-Tu, EF-G, RF3 (see Maracci and Rodnina). They contain a His in the switch II position in place of a Gln which is believed to perform the same function in catalysis, i.e. orienting the catalytic water. Their intrinsic hydrolysis rates are extremely slow and stimulation comes from interaction with the ribosome. Hydrolysis by EF-Tu, which has been a model G protein since 30 years and is one of the most heavily investigated, is stimulated by more than six orders of magnitude, but only if the ribosome is loaded with the correct codon-anticodon paired AA-tRNA in the ribosomal A site. This brings the SRC loop of the large ribosomal subunit close to the active site such that it can stabilize and/or rearrange the catalytic machinery. Myosin and Kinesin are more related to the P-loop G proteins than the proteins discussed below. However their mechanism of ATP hydrolysis is more divergent. In the case of kinesin (see Cross) there are two water molecules in the active site, one of which is positioned for nucleophilic attack while the other is believed to accept the proton from the former. Activation of nucleophilic water involves additional interactions with Ser, Mg and Gly. Microtubuli accelerate the rate-limiting ADP release such that the weakly bound ADP state of kinesin is allowed to form the strongly bound microtubule-kinesin-ATP complex. This is opposite to the case with myosin and actin (see Geeves) where the ATP-state is the weakly actin bound state. ATP hydrolysis is very similar in that it involves a two-water mechanism with a very similar activation process. Here the Pi release is the rate-limiting step of the reaction. Both myosin and kinesin form a strong salt bridge between an Arg on switch I and a Glu on switch II in the closed ATP-bound state whose opening is coupled to the major conformational change and the power stroke. According to sequence analysis, P-loop proteins can be divided into two major classes, the first class including the GTPases and the related proteins kinesin and myosin. The second is called ASCE, because they contain an additional b-strand between the P-loop and the strand containing a catalytic Glu/E that is involved in the catalytic mechanism by proton abstraction4. Most proteins of this type form homo- or hetero-oligomers with varying stoichiometry. One of the ASCE clades is the superfamily of ABC (ATP binding cassette) proteins, a very large group which are mechano-chemical engines and switches that function in a broad range of biological processes. The most well-known are the ABC transporters which catalyze the import or export of small molecules across membranes. In the most general form their NBD (nucleotide binding) domains dimerize face to face, where one protomer via its “signature motif” induces and stabilizes the nucleotide binding site of the other (see Hopfner). This is another unifying principle in many of the NTP hydrolysis proteins discussed here in that the catalytic machinery of the NTP binding is incomplete and the missing catalytic residues are put in place by homo- or heterodimerization and that the firing of the reaction is coupled to the biological process, in this case transport of solute. Another major class of P-loop proteins is formed by the AAA+ superfamily (Iyer et al., 2004)5, which is an extension of the originally described AAA proteins (ATPases associated with a variety of cellular activities) family (Erdmann et al., 1991)6 after it had been realized that a large number of additional proteins have the conserved sequence elements and work by the same principle. This now includes a large class of evolutionarily related proteins (Lakshminarayan et al., 2004). In the most general terms AAA+ proteins are mechano-chemical units which disassemble proteins, protein complexes or DNA double strands as in helicases. Their defining feature is the formation of a (usually) hexameric ring formed by a varying number of different polypeptides. The sequence analysis shows, apart from the P-loop and the Walker B glutamic acid residue, a number of additional conserved motifs such as the sensor 1 and sensor 2 motifs with a catalytic arginine which are involved in regulating oligomerization, ATP binding and/or hydrolysis. The nucleotide binding site is in the interface between subunits, and a totally conserved arginine from one subunit acts as a catalytic arginine finger for the other. ATP-dependent proteases of the AAA+ class contain a combination of ATP binding and protease domains. They are involved in quality control and preventing the accumulation of aberrantly folded proteins. The number of subunits that form the hexameric ring is variable, as is the number of active protomers in the assembly. ATP hydrolysis is proposed to induce translocation into the hydrolysis chamber, but the number of ATP molecules and whether or not ATP hydrolysis is an ordered, sequential or probabilistic event is at the focus of current research (see Bittner et al.). One of the many questions surrounding the action of the hexameric ring AAA+ ATPases is the order of events in the coupling of ATP hydrolysis and disassembly . A case in point is the action of NSF. It disassembles the SNARE complexes which mediate the fusion between vesicles and membranes in vesicular transport. The models being discussed are distributive, involving several rounds of interactions, processive, involving a stepwise reaction around the ring, whereas global disassembly involves no intermediates. The single molecule measurements described here (see Ryu et al.) describe how global ATP hydrolysis and Pi release leads to a strained conformation whereby the SNARE complex is ripped apart in one coordinated event. Such a mechanism may be shared by many other AAA+ members. The clamp loader is a device which loads the ring-like sliding clamp onto double-stranded DNA which in turn induces attachment of polymerase and other functional units. It is an outlier of the AAA+ superfamily and it is now believed not be a motor. Unlike most of its cousins it does not form a hexameric ring but rather a pentameric spiral. Its ATP hydrolysis is believed to be responsible for inducing release from DNA and the clamp after it has loaded the latter while the hexameric ring AAA+ ATPases might function as processive motors (see Kelch). As with the proteases, where the substrate induces ATP hydrolysis, the presence of DNA bound to the clamp loader is required for setting up the active site for catalysis. Compared to the clamp loader, much less is known about the microtubule severing enzymes spastin and katanin (see Ross). They form monomers with ADP and form a hexameric ring on MTs in the presences of ATP, but detailed structural and mechanistic investigation are yet to be done in order to understand the relation between ATP hydrolysis and splitting of the filament. Dynein is a particularly complex and very large AAA+ machine of 0,7 to 1,8MDa consisting of several polypeptides which moves along microtubules in the minus end direction. The dynein heavy chain is formed by six AAA modules coded for by a singly polypeptide. It also contains a number of extensions that are required for the power stroke and the interaction with microtubules. Although the sequence formally includes six ATP binding modules, only four contain the conserved catalytic residues, ie. an aparagine and arginine from sensor 1 and 2, respectively, a Glu for activating water and an arginine finger from the neighboring subunit. However, only the first AAA subunit is required for motility and recent structures have convincingly shown how the ATP/ADP conversion triggers the conformational change that is transmitted to the microtubule binding domain for attachment and release (see Schmidt and Carter). ATP hydrolyzing enzymes are by no means the only ones to oligomerize. Recently we have proposed the name GADs (GTPase activated by dimerization, Gaspar et al.,7) for such proteins. MnmE is a particular interesting G protein in that it is both an enzyme and a regulator for modification of the tRNA wobble position (see Fislage et al.). It belongs to the class of G proteins called HAS (Hydrophobic Amino acid Substituted for catalytic glutamine) which lack the residue important for catalysis of small and heterotrimeric G proteins (Anand et al., 2010)8. Dimerization is absolutely necessary for hydrolysis and requires the presence of potassium ions. Analysis of the active site in the transition state shows that dimerization induces the proper localization of catalytic residues and positions the potassium ion in exactly the place where small G proteins receive the amidinium group of the arginine finger from GAPs. The superfamily of dynamin related proteins features another class of proteins, the most prominent of which is dynamin itself (see Daumke and Praefcke). Dynamin forms a helical spiral around the neck of a budding vesicle. GTP hydrolysis occurs when two neighboring dynamin monomers in adjacent rings of the spiral form a head-to-head dimer which stabilize residues in the active site and position a small cation, in this case sodium, into a catalytic competent position analogous to the potassium in MnmE. Other members of the dynamin superfamily also require dimerization for catalysis but each of them show variation in the catalytic mechanism. The ATPase Hsp90 (see Pearl) belongs to the emerging family of ATPases called GHKL (Dutta et al.,)9, which is involved in the folding and/or stabilization of certain client proteins. It does not contain a P-loop, but we have included it here because of the similarities with the proteins mentioned. It contains three domains, a nucleotide binding domain (NBD), a middle and a C-terminal domain. Like MnmE, it is a constitutive dimer via the C-terminal domain. In the absence of nucleotide, the NBDs are far apart, but in the presence of ATP a dramatic conformational change occurs, which brings the NBDs into close proximity to each other such that they form a hydrolysis competent state. The middle domain contains a conserved arginine, which after dimerization is close to the γ-phosphate and could thus act as an arginine finger. A further interesting feature is that there are a number of cofactors which are specific negative or positive modulators of the ATPase activity in the presence of the specific client proteins. Thus, in summary, we have assembled here a number of proteins with vastly different biological function yet similar mechanisms of NTP hydrolysis, where the hydrolysis of either ATP or GTP is coupled to a fundamental biological reaction. These reviews reveal functional and structural mechanisms and will hopefully stimulate further discussion on similarities and differences between these and other systems. I thank all the contributors for their enthusiasm in support of this endeavor.