Dissecting the nature of long range interactions probed by RDCs in αS

As mentioned before, in folded proteins RDCs report on the orientation of internuclear vectors relative to a global molecular alignment tensor (Bax, 2003). In unfolded states of polypeptides they are believed to provide information about the degree of orientation that the internuclear vector adopts, with respect to the orientation of the persistence length fragment in which the residue is immersed (Dyson and Wright, 2004). Thus, RDCs are suggested to be valuable reporters of both local and long-range conformational propensities, and indeed they have proved to be exquisitely sensitive in the study of αS conformations. However, the true nature of what RDCs are able to detect is somehow still unclear, and thus it is instructive to comment on how the studies of the αS system may help to shed more light on this issue.

At the starting of this thesis work, the application of RDCs for unfolded proteins had just been introduced. Almost 5 years ago, much of a debate was started when Shortle explained the unexpected observation of RDCs in chemically denatured staphylococcal nuclease as evidence of the persistence of secondary structure, contrary to the expectation of vanishingly weak couplings due to conformational averaging (Shortle and Ackerman, 2001).

Theoretical studies by Annila refuted that explanation stating that conformational restrictions on the backbone of the polypeptide chain were sufficient to give non-vanishing RDCs (Louhivuori et al., 2003; Louhivuori et al., 2004), and later, experimental data from the Poulsen and Wright laboratories conclusively demonstrated the ability of RDCs to probe conformational propensities in chemically denatured proteins (Mohana-Borges et al., 2004).

However, it has not been until very recently that ensemble-averaged computer simulations provided conclusive evidence on the nature of RDCs in unfolded states of proteins.

In order to understand the origin of conformational restrictions probed by RDCs, Sosnick and collegues (Jha et al., 2005a) generated an unfolded state ensemble by using a self-avoiding statistical coil model that was based on backbone conformational frequencies in a coil library, a subset of the Protein Data Bank (Jha et al., 2005b). This ensemble of unfolded chains is shown to predict the experimental RDCs in chemically denatured apomyoglobin, ubiquitin, staphylococcal nuclease and eglin C, while retaining the global conformation of the unfolded state (Rg). The authors found that local structural propensities, in particular extended β and PII conformations, contributed strongly to the RDC signal (Jha et al., 2005a).

Blackledge and colleagues (Bernado et al., 2005b), independently generated a statistical ensemble of coil conformations and calculated averaged RDCs over the entire population. Each conformer from the ensemble was built sequentially using randomly selected φ/ψ pairs drawn from a database of amino acid specific conformations present in loop regions of high-resolution X-ray structures, and a simple volume-exclusion model was added to avoid steric overlap. RDCs were calculated for each NH vector of the conformer and DNH from each site were then averaged over 50000 conformers to ensure convergence. The authors successfully reproduced RDCs from the native unfolded domain of protein PX and the chemically unfolded states of apomyoglobin and staphylococcal nuclease. The outcome of this study is that non-vanishing RDCs will be expected for polypeptides sampling randomly the Ramachandran space, and suggest that RDCs are an interpretable structural signature of the unfolded state (Bernado et al., 2005b).

A collaboration with the Blackledge laboratory allowed us to apply this strategy on the RDCs studies of αS (Bernado et al., 2005a). An ensemble of conformers was generated for αS and RDCs were calculated for the protein aligned in C8E5/octanol anisotropic media (Figure 8.3.A). The range and fine structure of RDCs in αS were reproduced for the central part of the protein (residues 30 to 110) suggesting local conformational restrictions caused by random sampling of residue-specific φ/ψ distributions (Bernado et al., 2005b). However, the ensemble failed to reproduce the high couplings at the N- and C-terminus of the protein.

When a long-range interaction constraint between the N-and C-terminus was imposed to the creation of the ensemble, a remarkably close agreement with the experimental value was obtained in the RDCs calculations (Figure 8.3.B). Variations in long-range contacts were assayed, but the closest match remained to be in the N- to C-terminal contact. The results indicate that the proposed contact is present in the native ensemble of conformations

populated by αS in solution, although they do not exclude the presence of other contacts within the same conformer. Indeed, imposing as constraints simultaneous contacts between the C-terminal domain and N-terminal and NAC regions provides an ensemble that fits the data almost as closely (χ²= 59).

Figure 8.3. Long range order in αS probed by ensemble-averaged simulations of RDCs. A , B. Ensemble-averaged RDCs simulated (red) without long-range contacts (red, A) and when long long-range contacts between the N-terminus and C-terminus are set as restraints (red, B). In blue the experimental RDCs measure in Pf1 phages as alignment media are shown. Simulated data are scaled to maximize fit in the region 22-112. C. Effect of long-range contacts on capacity of conformational ensembles to reproduce experimental RDCs from αS. Figures denote χ² = Σ(Dij,calc - Dij,meas)² that compares experimental RDCs measured in αS aligned in Pf1 phage with calculated averages over 50000 conformers having long-range contacts (<15 Å)

between Cβ in the specified ranges. Adjacent domains are equivalent to ensembles with no specified contact shown in A (χ²= 129) and are not shown.

This theoretical approach is therefore complementary to the PRE-based detection of long-range contacts in αS, and unequivocally shows that tertiary interactions are present in the native state of αS. We are working currently in an atomic resolution complementation to this alignment prediction strategy, which may deliver more precise detail on the conformations populated by αS, and comparison with the ensemble of conformers that we previously generated by PRE will further provide valuable insights.

Intriguingly, when we measured RDCs in the homologous protein βS, which lacks the central hydrophobic domain and does not evidence tertiary interactions as the one present in αS, we observed that the C-terminus of this protein also displayed high couplings. However, in contrast to the profile obtained for αS, high couplings in βS were exclusively located at that region, which has previously displayed a strong tendency to populate extended conformations with the PII signature. The occurrence of 8 Pro residues and the contribution of electrostatic repulsion between negatively charged side chains (net charge of -16) favor the adoption of extended conformations at the C-terminus. Thus, in this particular case, RDCs probe locally encoded conformational restrictions in the ensemble of conformations populated by the natively unfolded protein βS.

Taken together these data supports the novel interpretation of RDCs as simultaneous reporters of long-range structural order and local conformational sampling in an ensemble of rapidly interconverting conformations.