Fig 1.
Flowchart showing the classification of homologous helix pairs and families.
Pair classification uses the angles of the two helices at the most disrupted site (θmax, θmin), angle difference (Δθ) and the error on each angle (ϵ). Family classification uses analogous statistics to obtain the same classes: the median angle (θmedian), standard deviation (σθ), and mean error (μϵ) of the angles in the family.
Fig 2.
Two examples of helix pairs, which are (A) not significantly different and (B) significantly different.
PDB code, chain identifier and residue numbers are given for each helix. The black residues are at the most disrupted site (see Methods) in each helix pair. rn and rc give the quality of the cylinder fit (see Eq (1)) to the backbone atoms on the N- (red) and C- (blue) terminal sides of the kink site. θ is the angle measured between the two cylinders. ε is the estimated error of the angle measurement, calculated from rn + rc using Eq (4). If θmax − θmin > ε1 + ε2, the confidence intervals do not overlap therefore we consider the angles to be significantly different.
Table 1.
The number of aligned helix pairs in each class, and occurrence of proline within that class.
Fig 3.
The difference in angle at the most disrupted site between helices in homologous helix pairs (|θmax − θmin|) plotted against the sequence identity between A) the homologous helix sequences, B) the residues in spatial proximity to the homologous helices (neighbouring residues) and C) the homologous chain sequences.
Data from the redundant membrane protein set is shown so that the full range of sequence identity can be seen.
Table 2.
The number of helix families greater than or equal to each group size for the soluble and membrane protein sets.
Fig 4.
Illustrations of a homologous helix family from each of the three main classes.
The number of families in each class is shown in brackets (membrane set / soluble set). 113 out of 1,258 soluble families and 2 out of 45 membrane families were classified as ‘Other’. (a) and (b) Side and top view of helices in a family superimposed by aligning the residues prior to the kink site. (c) Boxplots to show the variation in angle after smoothing at each site in the helix. The grey box indicates the site of maximum disruption used to classify the helix (see Methods). In the top left of each graph, the classification, σθ (standard deviation of angles), and με (mean error) of the most disrupted site is given.
Fig 5.
Distributions of angles measured at each site of the seven transmembrane helices in the GPCR family, after smoothing.
The label at each site shown on the x-axis is the Class A numbering used in the GPCRDB [24]. The broken grey line at 20° is the threshold for the definition of a kink. The most disrupted site in each helix (see Methods) is shown in grey. In the top left of each graph, the classification, σθ (standard deviation of angles), and με (mean error) of the most disrupted site is given.
Fig 6.
GPCR kink angle variation shown on the PDB structure 1F88, chain A.
Colouring is by a) mean and c) standard deviation of angles at each site in the GPCR family, on a spectrum from the lowest values in blue to the highest in red. Grey residues have no angles measured as they are loop regions or within 6 residues of the end of the consensus helix (the minimum for a cylinder fit). b) is coloured by rainbow from N-terminus (blue) to C-terminus (red). The conserved kink angles in TMH 6 and TMH 7 and the correlated kink angles at sites in TMH 3 and TMH 5 are labelled.
Fig 7.
A) Angle distribution at position 1x43 in all GPCR structures. Angles from rhodopsin structures are shown in blue (n = 32); angles from all other structures shown in red (n = 83). B) Angle distribution at position 3x28 in the human adenosine A2A receptor. Agonist-bound receptors are shown in green (n = 3); antagonist-bound receptors in orange (n = 8). Fig M in S1 Document displays the errors for the angle data from both histograms.