Endo and exo conformations of the proline ring.

2D ¹³C-¹³C DQ-SQ correlation spectra of mouse calvarial bone (green), in vitro osteoblast extracellular matrix (blue) and model collagen peptide (orange). All spectra were obtained at 10 kHz MAS and 297 K, and only the proline carbon regions are shown in this figure (full spectra can be found in Supplementary Fig. ). The peptide and osteoblast ECM samples are isotopically enriched specifically in glycine and proline residues. In the bone sample, ~20% of essential amino acids and glycine are U-¹³C, ¹⁵N labelled. The GPO proline signals are labelled in the figure and the ¹³C-¹³C correlations between them are traced (purple line) for the model peptide spectrum, starting with the amide carbon at around 172 ppm. The same trace is overlaid on the bone and osteoblast matrix spectra to show how we arrived at the assignment of the collagen GPO proline signals in those spectra, with a slight difference in that the Cγ signal in the spectra obtained from biologically-derived samples is at 24.8 ppm rather than 25.3 ppm in the peptide spectrum (pink line). The GPY (where Y ≠ O) connectivity is shown to be separate from that of GPO (grey line). The arrows on the bone spectrum indicate the two sets of signals corresponding to Cβ-Cγ, where the difference between the GPO (purple) and GPY (grey) signals can be most clearly observed. The black dotted line in each spectrum indicates the DQ-SQ diagonal.

Overlay of the 2D ¹³C-¹³C DQ-SQ correlation spectra of mouse calvarial bone (green) and model collagen peptide (orange), emphasising the identification of the GPO P_X population in the mouse bone spectrum.

Slices taken from the DQ-SQ correlation spectra of mouse calvarial bone (green), in vitro osteoblast matrix (blue) and model collagen peptide, showing the ratio of GPO to GPY (where Y ≠ O) signals. GPO slices are shown with solid lines (some with purple arrows), GPY slices are shown with dashed lines (some indicated with grey arrows). The Cγ assignment in Fig.  is shown with a green arrow on the bone GPO Cβ-Cγ slice. Vertical scaling was normalised on the larger signal within a GPO/GPY pair, maintaining the intensity difference between GPO and GPY to scale, but the vertical scale between pairs of GPO/GPY and between different samples are not shown to a unified scale. All spectra are presented on the same horizontal scale, enabling comparison of the chemical shift and also the linewidth at half height.

The energy landscape simulation results presented as Ramachandran plots. For each different, accessible conformational structure, the dihedral angles of two consecutive residues, proline and the residue subsequent to proline, are plotted. These backbone conformations arise from perturbing a single Pro residue in (PAG)₁₂ and (POG)₁₂ whilst maintaining the overall ring conformation as either endo or exo. For (PAG)₁₂, only four accessible structures were found, three of which were very similar in backbone dihedral angles in the current scaling, and therefore overlap nearly perfectly. (a) shows the backbone dihedral angles for the alanine residue of these four structures and (b) shows the backbone dihedral angles for the proline residues, showing a clear separation in backbone dihedral angles for endo and exo ring conformations. For (POG)₁₂, 64 accessible structures were found. (d) shows the backbone dihedral angles for the hydroxyproline for all 64 structures, split into two populations according to the conformation of the hydroxyproline ring. The preceding proline ring conformations were plotted for the case where hydroxyproline is exo (e) and endo (f). Panel (c) illustrates the dihedral angle change in terms of overall backbone conformation using structures generated for a GGG tripeptide (glycine was used for clarity) of constant dihedral angle using Avogadro 1.1.1, ranging from a fully extended backbone conformation (180°, 180°) to a much more compressed coiled conformation (−20°, 120°). It is clear that as ϕ and ψ increase, the peptide backbone is increasingly extended, while as the dihedral angles tend towards lower values, the backbone is increasingly compressed. Although the dihedral angle changes presented in the simulation represent a level of change that is smaller than that shown in the middle two peptides in (c) (a length difference of under 5%), the absolute extent of expansion/compression will scale with the length of the peptide. As previously reported, the idealized 7/2 (tighter) and 10/3 (looser) triple helices both have dihedral angles that more closely match endo P_X in (b,e and f).

The distribution of GPO sites within a type I collagen fibril, from diffraction data (a) and in the consensus sequence (b). In (a), the collagen fibrillar structure (from PDB 3HR2) is modelled from X-ray diffraction data on rat tail tendon with the Cα atoms in GPO triplets represented as red spheres. In (b) the consensus sequence based on the most frequent amino acid at each position is shown. The three chains of collagen are staggered by one residue with respect to each other, based on a previous study on the VWF A3 domain binding. The D period arrangement shows a good match to the experimentally-derived fibril structure model above. Highly conserved (75%+) GPO triplets are highlighted in red; others in pink. High affinity integrin binding sites are in blue and the DDR/ VWF binding site in green.