Exploring the folding pathways of annexin I, a multidomain protein. II. Hierarchy in domain folding propensities may govern the folding process.

In the context of exploring the relationship between sequence and folding pathways, the multi-domain proteins of the annexin family constitute very attractive models. They are constituted of four approximately 70-residue domains, named D1 to D4, with identical topologies but only limited sequence homology of approximately 30%. The domains are organized in a pseudochiral circular arrangement. Here, we report on the folding propensity of the D1 domain of annexin I obtained from overexpression in Escherichia coli. Unlike the D2 domain, which is only partially folded, the isolated D1 domain exhibits autonomous refolding in pure aqueous solution. Similarly, the D3 domain and D2-D3 module were obtained from expression in E. coli but were found to be largely unfolded. No conclusion could be drawn for the D4 domain because it was not possible to extract it from the bacterial inclusion bodies. The data allow us to propose a plausible scenario for the annexin I folding. This working model states that firstly the D1 domain folds, and the D2 and D3 domains remain partly unfolded, facilitating the docking of the D4 domain to the D1 domain. In a second step, the D1 and D4 domains dock, and D4 may fold if already not folded. The final step starts with the stabilization of the D1-D4 module. This stabilization is crucial for allowing the non-native local interactions inside the still partially unfolded D2 domain to switch to the native long-range interactions involving D4. This switch allows the complete folding of D2 and D3. The model proposes a sequential and hierarchical process for the folding of annexin I and emphasizes the role of both native framework and non-native structures in the process.

Exploring the folding pathways of annexin I, a multidomain protein. I. non-native structures stabilize the partially folded state of the isolated domain 2 of annexin I.

Proteins of the annexin family constitute very attractive models because of their four approximately 70 residue domains, D1 to D4, exhibiting an identical topology comprising five helix segments with only a limited sequence homology of approximately 30%. We focus on the isolated D2 domain, which is only partially folded. A detailed analysis of this equilibrium partially folded state in aqueous solution and micellar solution using 15N-1H multidimensional NMR is presented. Comparison of the residual structure of the entire domain with that of shorter fragments indicates the presence of long-range transient hydrophobic interactions that slightly stabilize the secondary structure elements. The unfolded domain tends to behave as a four-helix, rather than as a five-helix domain. The ensemble of residual structures comprises: (i) a set of native structures consisting of three regions with large helix populations, in rather sharp correspondence with A, B and E helices, and a small helix population in the second part of the C helix; (ii) a set of non-native local structures corresponding to turn-like structures stabilized by several side-chain to side-chain interactions and helix-disruptive side-chains to backbone interactions. Remarkably, residues involved in these local non-native interactions are also involved, in the native structure, in structurally important non-local interactions. During the folding process of annexin I, the local non-native interactions have to switch to native long-range interactions. This structural switch reveals the existence of a sequence-encoded regulation of the folding pathways and kinetics, and emphasizes the key role of the non-native local structures in this regulation.

A conformational equilibrium in a protein fragment caused by two consecutive capping boxes: 1H-, 13C-NMR, and mutational analysis.

The conformational properties of an 18 residues peptide spanning the entire sequence, L1KTPA5QFDAD10ELRAA15MKG, of the first helix (A-helix) of domain 2 of annexin I, were thoroughly investigated. This fragment exhibits several singular features, and in particular, two successive potential capping boxes, T3xxQ6 and D8xxE11. The former corresponds to the native hydrogen bond network stabilizing the alpha helix N-terminus in the protein; the latter is a non-native capping box able to break the helix at residue D8, and is observed in the domain 2 partially folded state. Using 2D-NMR techniques, we showed that two main populations of conformers coexist in aqueous solution. The first corresponds to a single helix extending from T3 to K17. The second corresponds to a broken helix at residue Ds. Four mutants, T3A, F7A, D8A, and E11A, were designed to further analyze the role of key amino acids in the equilibrium between the two ensembles of conformers. The sensitivity of NMR parameters to account for the variations in the populations of conformers was evaluated for each peptide. Our data show the delta13Calpha chemical shift to be the most relevant parameter. We used it to estimate the population ratio in the various peptides between the two main ensembles of conformers, the full helix and the broken helix. For the WT, E11A, and F7A peptides, these ratios are respectively 35/65, 60/40, 60/40. Our results were compared to the data obtained from helix/coil transition algorithms.

A computer program to determine a protein sequence from an amino acid analysis.

We have developed a computational method for analyzing the proteolytic products of a protein, knowing its sequence and the amino acid percentages of its products. For all fragments, amino acid percentages are calculated and compared to the experimental results (calculating the error within the experiment). The program keeps the best fitted fragment using a least squared method. This program was written to determine the sequence of the proteolytic products that appeared during the purification of annexin I domain 2. The reliability of the method was verified in this case. However the latter depends on the length and on the amino acid composition of the entire protein and of its fragments. This program may be suitable for analyzing the sequence of the products in any protease digestion, whether designed or accidental.

Folding properties of an annexin I domain: a 1H-15N NMR and CD study.

The annexin fold consists of four 70-residue domains with markedly homologous sequences and nearly identical structures. Each domain contains five helices designated A to E. Domain 2 of annexin I was obtained by chemical synthesis including ten specifically labeled residues and studied by 1H-15N NMR and circular dichroism (CD). In pure aqueous solution this annexin domain presents, at most, 25% of residual helix secondary structure compared to 75%-85% for the native helix content and thus does not constitute an autonomous folding unit. Dodecylphosphocholine (DPC) micelles were used to provide the annexin domain with non-specific hydrophobic interactions. The structuring effect of micelles was thoroughly investigated by CD and 1H-15N NMR. Most, but not all, of the native helix secondary structure was recovered at DPC saturation. NMR data made it possible to determine the intrinsic helix propensity hierarchy of the different helix segments of the domain: A approximately B approximately E > C, D. This hierarchy is remarkably well correlated with the location of the helices in the native protein since A, B, and E helices are those in contact with the remaining parts of the protein. This result tends to support the view that, for large proteins like annexins (35 kDa), high intrinsic secondary structure propensities, at least helix propensity, in selected protein segments is necessary for a correct folding process. As a consequence this also indicates that important information concerning the folding pathway is encoded in the protein sequence.

Dodecylphosphocholine micelles as a membrane-like environment: new results from NMR relaxation and paramagnetic relaxation enhancement analysis.

To further examine to what extent a dodecyl-phosphocholine (DPC) micelle mimics a phosphatidylcholine bilayer environment, we performed 13C, 2H, and 31P NMR relaxation measurements. Our data show that the dynamic behavior of DPC phosphocholine groups at low temperature (12 degrees C) corresponds to that of a phosphatidylcholine interface at high temperature (51 degrees C). In the presence of helical peptides, a PMP1 fragment, or an annexin fragment, the DPC local dynamics are not affected whereas the DPC aggregation number is increased to match an appropriate area/volume ratio for accommodating the bound peptides. We also show that quantitative measurements of paramagnetic relaxation enhancements induced by small amounts of spin-labeled phospholipids on peptide proton signals provide a meaningful insight on the location of both PMP1 and annexin fragments in DPC micelles. The paramagnetic contributions to the relaxation were extracted from intra-residue cross-peaks of NOESY spectra for both peptides. The location of each peptide in the micelles was found consistent with the corresponding relaxation data. As illustrated by the study of the PMP1 fragment, paramagnetic relaxation data also allow us to supply the missing medium-range NOEs and therefore to complete a standard conformational analysis of peptides in micelles.