Methods to account for movement and flexibility in cryo-EM data processing.

Recent advances in direct electron detectors and improved CMOS cameras have been accompanied by the development of a range of software to take advantage of the data they produce. In particular they allow for the correction of two types of motion in cryo electron microscopy samples: motion correction for movements of the sample particles in the ice, and differential masking to account for heterogeneity caused by flexibility within protein complexes. Here we provide several scripts that allow users to move between RELION and standalone motion correction and centring programs. We then compare the computational cost and improvements in data quality with each program. We also describe our masking procedures to account for conformational flexibility. For the different elements of this study we have used three samples; a high symmetry virus, flexible protein complex (∼1MDa) and a relatively small protein complex (∼550kDa), to benchmark four widely available motion correction packages. Using these as test cases we demonstrate how motion correction and differential masking, as well as an additional particle re-centring protocol can improve final reconstructions when used within the RELION image-processing package.

Simple rules for efficient assembly predict the layout of a packaged viral RNA.

Single-stranded RNA (ssRNA) viruses, which include major human pathogens, package their genomes as they assemble their capsids. We show here that the organization of the viral genomes within the capsids provides intriguing insights into the highly cooperative nature of the assembly process. A recent cryo-electron microscopy structure of bacteriophage MS2, determined with only 5-fold symmetry averaging, has revealed the asymmetric distribution of its encapsidated genome. Here we show that this RNA distribution is consistent with an assembly mechanism that follows two simple rules derived from experiment: (1) the binding of the MS2 maturation protein to the RNA constrains its conformation into a loop, and (2) the capsid must be built in an energetically favorable way. These results provide a new level of insight into the factors that drive efficient assembly of ssRNA viruses in vivo.

Structures of unliganded and ATP-bound states of the Escherichia coli chaperonin GroEL by cryoelectron microscopy.

We have developed an angular refinement procedure incorporating correction for the microscope contrast transfer function, to determine cryoelectron microscopy (cryo-EM) structures of the Escherichia coli chaperonin GroEL in its apo and ATP-bound forms. This image reconstruction procedure is verified to 13-A resolution by comparison of the cryo-EM structure of unliganded GroEL with the crystal structure. Binding, encapsulation, and release of nonnative proteins by GroEL and its cochaperone GroES are controlled by the binding and hydrolysis of ATP. Seven ATP molecules bind cooperatively to one heptameric ring of GroEL. This binding causes long-range conformational changes that determine the orientations of remote substrate-binding sites, and it also determines the conformation of subunits in the opposite ring of GroEL, in a negatively cooperative mechanism. The conformation of GroEL-ATP was determined at approximately 15-A resolution. In one ring of GroEL-ATP, the apical (substrate-binding) domains are extremely disordered, consistent with the high mobility needed for them to achieve the 60 degrees elevation and 90 degrees twist of the GroES-bound state. Unexpectedly, ATP binding also increases the separation between the two rings, although the interring contacts are present in the density map.

ATP-bound states of GroEL captured by cryo-electron microscopy.

The chaperonin GroEL drives its protein-folding cycle by cooperatively binding ATP to one of its two rings, priming that ring to become folding-active upon GroES binding, while simultaneously discharging the previous folding chamber from the opposite ring. The GroEL-ATP structure, determined by cryo-EM and atomic structure fitting, shows that the intermediate domains rotate downward, switching their intersubunit salt bridge contacts from substrate binding to ATP binding domains. These observations, together with the effects of ATP binding to a GroEL-GroES-ADP complex, suggest structural models for the ATP-induced reduction in affinity for polypeptide and for cooperativity. The model for cooperativity, based on switching of intersubunit salt bridge interactions around the GroEL ring, may provide general insight into cooperativity in other ring complexes and molecular machines.

Multivalent binding of nonnative substrate proteins by the chaperonin GroEL.

The chaperonin GroEL binds nonnative substrate protein in the central cavity of an open ring through exposed hydrophobic residues at the inside aspect of the apical domains and then mediates productive folding upon binding ATP and the cochaperonin GroES. Whether nonnative proteins bind to more than one of the seven apical domains of a GroEL ring is unknown. We have addressed this using rings with various combinations of wild-type and binding-defective mutant apical domains, enabled by their production as single polypeptides. A wild-type extent of binary complex formation with two stringent substrate proteins, malate dehydrogenase or Rubisco, required a minimum of three consecutive binding-proficient apical domains. Rhodanese, a less-stringent substrate, required only two wild-type domains and was insensitive to their arrangement. As a physical correlate, multivalent binding of Rubisco was directly observed in an oxidative cross-linking experiment.

Chaperonins.

The molecular chaperones are a diverse set of protein families required for the correct folding, transport and degradation of other proteins in vivo. There has been great progress in understanding the structure and mechanism of action of the chaperonin family, exemplified by Escherichia coli GroEL. The chaperonins are large, double-ring oligomeric proteins that act as containers for the folding of other protein subunits. Together with its co-protein GroES, GroEL binds non-native polypeptides and facilitates their refolding in an ATP-dependent manner. The action of the ATPase cycle causes the substrate-binding surface of GroEL to alternate in character between hydrophobic (binding/unfolding) and hydrophilic (release/folding). ATP binding initiates a series of dramatic conformational changes that bury the substrate-binding sites, lowering the affinity for non-native polypeptide. In the presence of ATP, GroES binds to GroEL, forming a large chamber that encapsulates substrate proteins for folding. For proteins whose folding is absolutely dependent on the full GroE system, ATP binding (but not hydrolysis) in the encapsulating ring is needed to initiate protein folding. Similarly, ATP binding, but not hydrolysis, in the opposite GroEL ring is needed to release GroES, thus opening the chamber. If the released substrate protein is still not correctly folded, it will go through another round of interaction with GroEL.

Asymmetry, commitment and inhibition in the GroE ATPase cycle impose alternating functions on the two GroEL rings.

The ATPase cycle of GroE chaperonins has been examined by transient kinetics to dissect partial reactions in complexes where GroEL is asymmetrically loaded with nucleotides. The occupation of one heptameric ring by ADP does not inhibit the loading of the other with ATP nor does it prevent the consequent structural rearrangement to the "open" state. However, ADP binding completely inhibits ATP hydrolysis in the asymmetric complex, i.e. ATP cannot by hydrolysed when ADP is bound to the other ring. This non-competitive inhibition of the ATPase by ADP is consistent with a ring-switching, or "two-stroke", mechanism of the type: ATP:GroEL --> ADP:GroEL --> ADP:GroEL:ATP --> GroEL:ATP --> GroEL:ADP, i.e. with respect to the GroEL rings, ATP turns over in an alternating fashion. When the ATP-stabilized, "open" state is challenged with hexokinase and glucose, to quench the free ATP, the open state relaxes slowly (0.44 s-1) back to the apo (or closed) conformation. This rate, however, is three times faster than the hydrolytic step, showing that bound ATP is not committed to hydrolysis. When GroES is bound to the GroEL:ATP complex and the system is quenched in the same way, approximately half of the bound ATP undergoes hydrolysis on the chaperonin complex showing that the co-protein increases the degree of commitment. Thus, non-competitive inhibition of ATP hydrolysis, combined with the ability of the co-protein to block ligand exchange between rings has the effect of imposing a reciprocating cycle of reactions with ATP hydrolysing, and GroES binding, on each of the GroEL rings in turn. Taken together, these data imply that the dominant, productive steady state reaction in vivo is: GroEL:ATP:GroES --> GroEL:ADP:GroES --> ATP:GroEL:ADP:GroES --> ATP:GroEL:ADP --> GroES:ATP:GroEL:ADP --> GroES:ATP:GroEL for a hemi-cycle, and that significant inhibi tion of hydrolysis may arise through the formation of a dead-end ADP:GroEL:ATP:GroES complex.

Binding, encapsulation and ejection: substrate dynamics during a chaperonin-assisted folding reaction.

Mitochondrial malate dehydrogenase (mMDH) folds more rapidly in the presence of GroEL, GroES and ATP than it does unassisted. The increase in folding rate as a function of the concentration of GroEL-ES reaches a maximum at a stoichiometry which is approximately equimolar (mMDH subunits:GroEL oligomer) and with an apparent dissociation constant K' for the GroE acceptor state of at least 1 x 10(-8) M. However, even at chaperonin concentrations which are 4000 x K', i.e. at negligible concentrations of free mMDH, the observed folding rate of the substrate remains at its optimum, showing not only that folding occurs in the chaperonin-mMDH complex but also that this rate is uninhibited by any interactions with sites on GroEL. Despite the ability of mMDH to fold on the chaperonin, trapping experiments show that its dwell time on the complex is only 20 seconds. This correlates with both the rate of ATP turnover and the dwell time of GroES on the complex and is only approximately 5% of the time taken for the substrate to commit to the folded state. The results imply that ATP drives the chaperonin complex through a cycle of three functional states: (1) an acceptor complex in which the unfolded substrate is bound tightly; (2) an encapsulation state in which it is sequestered but direct protein-protein contact is lost so that folding can proceed unhindered; and (3) an ejector state which forces dissociation of the substrate whether folded or not.

Chaperonins can catalyse the reversal of early aggregation steps when a protein misfolds.

Chaperonins use energy derived from ATP hydrolysis to enhance the efficiency of protein folding by a mechanism which remains a matter of debate. Here, we show that the kinetics of spontaneous and assisted folding of mitochondrial malate dehydrogenase are quantitatively described by a simple physical model. The protein folds from non-native chains by the slow formation of native-like monomers, which then dimerize to form the active enzyme. Misfolding proceeds by two phases of aggregation: the first is slowly reversible, the second is irreversible. Chaperonins accelerate the dissociation of the first-formed, unstable aggregates through a repeated binding-and-release cycle coupled to ATP hydrolysis. By this catalytic action, they supply the productive folding pathway with monomers, and block the irreversible phase of aggregation, thereby maintaining optimal folding yields even when present in sub-stoichiometric quantities. The hydrolytically active chaperonin is required until the substrate protein has completed the slow transition to its native-like, monomeric state. Both the observed rate of folding and the yield are increased by this mechanism without changing real rates in the productive pathway.

The origins and consequences of asymmetry in the chaperonin reaction cycle.

The binding of nucleotides and chaperonin-10 (cpn10) to the Escherichia coli chaperonin-60 (cpn60) and their effect upon the molecular symmetry has been examined both kinetically and at equilibrium. ATP binds tightly and is hydrolysed on only one heptameric ring of the cpn60 tetradecamer at a time, thus inducing asymmetry in the cpn60 oligomer even in the absence of cpn10. In the absence of cpn10 these seven ATP molecules hydrolyse to form a cpn60:ADP7 complex in which ADP is tightly bound (Kd = 2-7 microM); further ADP binding to form a cpn60:ADP14 complex is weak (K1/2 = 2.3 mM). We conclude that symmetrical nucleotide complexes (with 14 ATP or 14 ADPs) are unstable, demonstrating negative co-operativity between the rings. When cpn60 is mixed with cpn10 and ATP the resultant cpn60:ATP7:cpn10 complex is formed rapidly (the rate constant for cpn10 association is > 4 x 10(7) M-1 s-1) and before ATP is hydrolysed (k = 0.12 s-1 per active subunit) to produce an extremely stable cpn60:ADP7:cpn10 complex. This allows ATP association on the unoccupied ring and nucleotide asymmetry in the double toroid is preserved. In "trapping" experiments, where the cpn60:ADP7:cpn10 is challenged with ATP, cpn10 was observed to dissociate at a rate identical to that of steady-state ATP hydrolysis in the presence of cpn10 (k = 0.042 s-1 per active subunit). The spontaneous decay of cpn60:ADP7:cpn10 and any of the major steady-state complexes, under conditions where free nucleotides had been removed, occurred at a rate tenfold lower than ATP hydrolysis. Since the binding of the non-hydrolysable analogue AMP-PNP was unable to induce dissociation of the co-chaperonin it was concluded that a transient state following ATP hydrolysis is necessary for the rapid dissociation of cpn10, which occurs once in every cycle. Trapping experiments using sub-stoichiometric concentrations of cpn10, relative to cpn60, show an unchanged rate of cpn10 exchange upon ATP hydrolysis, indicating that the formation of a symmetric, "football"-shaped complex in which two molecules of the co-chaperonin are bound to cpn60, is not an obligatory intermediate in the exchange process.

Location of a folding protein and shape changes in GroEL-GroES complexes imaged by cryo-electron microscopy.

Protein folding mediated by the molecular chaperone GroEL occurs by its binding to non-native polypeptide substrates and is driven by ATP hydrolysis. Both of these processes are influenced by the reversible association of the co-protein, GroES (refs 2-4). GroEL and other chaperonin 60 molecules are large, cylindrical oligomers consisting of two stacked heptameric rings of subunits; each ring forms a cage-like structure thought to bind polypeptides in a central cavity. Chaperonins play a passive role in folding by binding or sequestering folding proteins to prevent their aggregation, but they may also actively unfold substrate proteins trapped in misfolded forms, enabling them to assume productive folding conformations. Biochemical studies show that GroES improves the efficiency of GroEL function, but the structural basis for this is unknown. Here we report the first direct visualization, by cryo-electron microscopy, of a non-native protein substrate (malate dehydrogenase) bound to the mobile, outer domains at one end of GroEL. Addition of GroES to GroEL in the presence of ATP causes a dramatic hinge opening of about 60 degrees. GroES binds to the equivalent surface of the GroEL outer domains, but on the opposite end of the GroEL oligomer to the protein substrate.