Contribution to Current Knowledge - Characterisation of an interaction involved in viral replic

5.2.1 Dynamics

It is perhaps somewhat surprising that a globular protein is so dynamic on a millisecond to microsecond time-scale. The work described in chapters 3 and 4 suggests that communication between the active site of 3Cpro_{and the RNA–binding face takes place. In addition to the work}

described here, a number of sources have postulated or provided evidence for allostery between the two functional regions of the protein and related proteins in other picornaviruses (Peterset al., 2004; Wang and Johnson, 2001; Shihet al., 2004; Bjorndahlet al., 2007). Bjorndahl and co-workers specifically mention that the structure of the RNA–binding face is unchanged in HRV-14 3Cpro

upon inhibitor binding but leave the possibility open for a change in the RNA–binding face to affect catalytic activity.

5.2.2 Allostery

Allostery (Gr. αλλος: other andστερεος: shape) refers to an alteration in the activity of one

protein domain upon binding of an effector to a distal region of the protein. Allosteric interactions in proteins have been studied for a long time. The second crystal structure of a protein to be solved was that of haemoglobin (Perutzet al., 1960), the archetypal allosteric protein. An early model to be proposed for allostery in haemoglobin was that of Monod, Wyman and Changeux (MWC) (Monodet al., 1965). In this model oxygen molecules bind to subunits in haemoglobin and alter the equilibrium between two quaternary states, designated tense (T) and relaxed (R), where T and R forms of haemoglobin have low and high affinity for oxygen respectively. The in- crease in affinity for binding the second, third and fourth molecules of O2to a single haemoglobin

molecule is due to the shift of the equilibrium towards the R state. Another model, that of se- quential binding was proposed for haemoglobin allostery. In this, the KNF model (Koshland

et al., 1966), binding of each O2changes the structure of a neighbouring unbound subunit to the

R form, with higher affinity for O2. It is possible for hybrid TR intermediates to occur.

Despite the thermodynamic implications of the MWC model, the majority of proteins have, until recently, been largely considered to have little or no allosteric function. However, several

strands of evidence are beginning to suggest that most, if not all, non-fibrous proteins have the potential for allosteric control. Another way of putting this, in keeping with the classical view of allostery, is that distribution of the conformational ensemble is also not confined to allosteric proteins, it is a characteristic of proteins in general (Gunasekaranet al., 2004; Smock and Gierasch, 2009). A seminal paper by Cooper and Dryden (Cooper and Dryden, 1984) examined the idea that changes in protein dynamics could be of functional consequence. The internal structure of globular proteins is dominated by hydrophobic interactions. Protein folding is often described in terms of minimisation of exposure of hydrophobic residues to the bulk solvent. NMR studies of the dynamics of side chains within the protein interior show that the side chains sample many different conformations and that these states exchange on several time-scales. Upon cooling to around 200K, the exchange between states is ’frozen’ and only ps–ns timescale motions (such as bond librations) occur (Lee and Wand, 2001). This so-called ’glass transition’ indicates that proteins in solution contain a large amount of residual entropy even when well folded (Wand, 2001b). Although crystal structures, as well as averages of NMR ensembles, suggest a static, well-defined fold for proteins, the reality is that proteins constantly sample the conformational space around the bottom of the ‘folding funnel’ (Dobsonet al., 1998; Lazaridis and Karplus, 1997). Functional consequences of this dynamic situation are beginning to be appreciated. In retrospect it would seem rather surprising if proteins did not take advantage of this large amount of entropy available to them for aspects of their function.S2_{data on folded proteins indicates that backbone}

motion is low compared to the side-chains and that the backbone functions as a relatively rigid ’skeleton’. This is not to say that large motions involving the backbone do not occur. Many functional motions involve backbone rearrangement. Side chain motions on a ps–ns time-scale are extensive in all proteins (Wand, 2001a; Huet al., 2003). Interestingly, the degree of motion in the backbone appears to be poorly correlated to the degree of motion in the attached side-chains (Wand, 2001b).

5.2.3 Allostery in single domain proteins

It is now becoming clear that proteins exhibit allosteric behaviour even within domains. Re- cent NMR studies have shown that mutations in proteins can affect the distribution of states in an ensemble and that this can lead to effects at distal sites. An elegant, NMR-based study of allosteric response in a single domain protein is provided by the work of Volkman and co-workers (Volkmanet al., 2001). NtrC is a member of the ‘two-component system’ signalling family (Stock

et al., 2000). Phosphorylation of the response-regulator region of NtrC promotes oligomer for- mation in the full length protein, which then stimulates transcription by an RNA polymerase. Mutant NtrC with partial activity when unphosphorylated shows altered chemical shifts in the region that changes structure upon phosphorylation. Both the apo and mutant forms of the protein show significantRexin this region; upon phosphorylation theRexis largely abrogated. This leads to the conclusion that the active state is sampled by the apo form but only transiently. The partially active mutant samples the active state more frequently. The conclusion drawn from these data is that the protein samples the active state even in the apo form; phosphorylation of the protein merely changes the equilibrium of a pre-sampled conformational ensemble. It appears that “active” conformers are present in “inactive” ensembles at lower concentrations. The authors of this study also suggest that theμs–ms time-scale motions seen in this study are

likely to be common as ‘many biological processes occur in micro- to millisecond time scales’. Conversely another study of dynamic pathways in a globular protein found that networks of re- sponses took place mainly on a ps–ns time-scale. Clarkson and co-workers (Clarksonet al., 2006) showed that, by mutating valine residues in eglin C to alanine, they were able to alter the side chainS2_and

τeparameters of distal residues. There were found to be two types of response: a

‘contiguous network response’ involving changes inS2_{, transmitted residue to residue down a}

pathway and a ‘disperse network response’ where rotomer populations (measured byτe) were

altered in residues up to 10 Å from the mutated residue with no obvious chain of contact. The signals were not necessarily bidirectional. It was also found that the entropic couplings predict enthalpic couplings, indicating, that even if the changes in dynamics (and therefore entropy in the protein core) are not the root cause of allostery, they may be a good indicator of the presence of allosteric interactions. The difference in time-scales found to be important in these two studies is indicative of two different facets of the interactions. It seems likely that the residual entropy in a protein, as identified by side-chain motions on a ps–ns time-scale has been utilised by proteins to transmit information between distal sites. However, as mentioned above, many biological ac- tivities occur on a longer time-scale. It is possible that the time-scale of motions relevant to the allosteric interaction is much faster than the time-scale of the process affected by those motions (Wand, 2001b).

5.2.4 Possibilities of networks for 3C

pro

What form does the allosteric interaction take in 3C? There is a small amount of perturbation of the active site upon RNA–binding. I have suggested previously that this could be mediated by a chain of residues, largely followingβ-strands through the centre of the protein. The network of chemical-shift perturbations is not entirely contiguous. However, it has been suggested that transduction of motion or energy may occur through immobile residues (Ota and Agard, 2005). Figure 5.1 shows the pathway through the centre of the protein of residues that exhibit chemical shift perturbation upon RNA binding. The study of eglin C dynamics mentioned previously (Clarksonet al., 2006), has shown that, although it is possible for contiguous transmis- sion of dynamic changes to cause allosteric effects in proteins, this is by no means necessary. Chemical shifts are probably coupled to selected motions on a ps–ns time-scale(Berjanskii and Wishart, 2005). Examination ofRex found by the cross-correlation method reveals that many of the residues in the two β-strands exhibit significant motion on the milli- to microsecond timescale. It is possible that a change in the rate of exchange in these residues upon binding of the RNA causes changes in the equilibrium population of states, leading to the observed changes in chemical shift. This is certainly suggested by the change in distribution ofRCC

ex between 3C and

3CI. The RNA-binding region is very dynamic on a ms–μs time-scale, more so than the opposite

face. It would be interesting to see whether, in common with previously mentioned work (Volk- manet al., 2001), there is a decrease in this motion upon RNA binding. Thus, in 3Cpro_{, it would}

seem likely that substrate or RNA binding can induce changes in ps–ns dynamics in the protein core, and this can alter the equilibrium ofμs–ms motion at either the catalytic or RNA–binding

face of the protein.

In document Characterisation of an interaction involved in viral replication : submitted in partial fulfilment of the requirements of the degree of Doctor of Philosophy, Institute of Fundamental Sciences, Massey University (Page 142-145)