Each helix has its own characteristic degree of conservation and number of group-conserved residues [26
], and its own distribution of polar and hydrophobic residues in accordance with its role in conferring stability, ligand binding, activation and G-protein coupling. Thus, for each helix, the various measures, taken in isolation, contribute differently to the identification of the preferred alignment, which can be inferred directly or preferably indirectly via GCR1. Here, our primary measure is the consensus score (j–l
). For TM1, TM5, TM6 and TM7, the consensus score also indicates plausible alternative alignments (). For this reason, additional measures have been used to distinguish between the preferred alignment and the plausible alignments, including the mutagenesis results for CLR (electronic supplementary material, table S1) and for other class B GPCRs (electronic supplementary material, table S2). Where new mutants have been made, these were initially chosen to probe the role of conserved amino acids across the TM domains. Additional mutants have been made to investigate particular regions or motifs in more detail. This approach has resulted in a good distribution of mutants across each helix.
Figure 1. The scores for the TM3 class A–class B alignment. The top row shows the scaled Blosum 62 profile alignment scores for the class A–class B (a), class B-GCR1 (b) and class A-GCR1 (c) alignments; points below the dotted line have scores less (more ...)
Key data relevant to alternative alignments. Data that support the preferred alignment by indicating against the alternatives shown by the consensus are shaded in grey.
In some alternative alignments, the variability pattern or the position of charged residues is incompatible with the receptor topology and this is indicated in . The alignment of class A group-conserved positions with class B group-conserved positions (electronic supplementary material, table S3—this resembles the cold spot method) has been assessed by comparison with random distributions of an equivalent number of residues, and found to be statistically superior to random with a p-value of 0.04 or less for all transmembrane helices (electronic supplementary material, table S4, with a summary in ), except for TM4 where there are few group-conserved residues; the p-values are also generally superior to those for our plausible alternative alignments () and to those for alternative alignments in the literature. The mutagenesis data collected over 266 residues have been interpreted using the four CLR homology models. Our prime focus in analysing the mutagenesis data for a given residue has been to assess whether it is consistent with the alignment; seeking to understand, the function of the residue has been a secondary focus. In reporting the alignment, our focus has been on the motifs shared between class A and class B GPCRs, and so much of the other information is reported in the electronic supplementary material. For the majority of this mutagenesis data, it is not possible to distinguish between correct and incorrect alignments, either because the mutation had no effect, because the residue would move to an equivalent position in the plausible alternative alignment or occasionally because the homology models do not yield a consensus on the residue environment (e.g. internal versus external, helical versus loop). However, for 27 residue positions, the mutation data have been instrumental in confirming the alignment since the result is difficult to explain except in the preferred alignment; none of the mutation data are contrary to the preferred alignment. The measures in that help to define the alignment by indicating against the alternatives are shaded in grey.
The alignment for TM3 is discussed first because it is extremely clear, as shown by the consensus results in , being defined by the conserved Cys3.26 that forms a disulphide bond with EL2. It is also characterized by a clear positive peak in a–c for each of the three Blosum 62-based profile alignments, indicating that alignment 0 () is the preferred alignment according to both the direct class A–class B alignment and according to the indirect method involving alignments with GCR1, as summarized in , rows 2–4. Based on the entropy correlations, d–f, the class A–class B alignment gives a clear preference for alignment 0 (row 5). While not every measure supports alignment 0, when the three criteria (Blossum 62 profile alignment, entropy and hydrophobicity) are multiplied together, the weaker peaks in a–i are down-weighted, resulting in a strong prediction for the 0 alignment (j–l and , rows 11–14). Since there are no other peaks, there are no plausible alternative alignments.
Figure 2. The class A–class B-GCR1 alignments (selected sequences). (a) TM1, (b) TM2, (c) TM3, (d) TM4, (e) TM5, (f) TM6 and (g) TM7. The most conserved positions in class A are marked by a vertical bar and correspond to position 50 in each helix, e.g. (more ...)
There are a number of highly conserved residues in TM3 (electronic supplementary material, table S3), most strikingly E3.46
. The latter forms the class B positional equivalent of the class A DRY3.51
]. However, the YLH motif does not appear to have the same function as the class A DRY motif as the mutation effects are less marked [29
]. The evidence from our data is that mutation has an effect on cell surface expression [30
], and there are precedents for this with Y3.51
of the DRY motif. The YLH motif seems to be part of an extended hydrophobic network also involving the proximal part of IL2 that helps maintain the inactive receptor in a closed state, although IL2, but not Y3.51
, also contacts Gs. E3.46
can interact with R1732.39
in TM2 in our alignment; together they may form the functional equivalent of the DRY motif, as will be discussed below.
For the remaining helices, we will focus largely on the consensus results (rows 11–14) of .
For TM1, the individual measures, profile alignment, entropy and hydrophobicity, suggest several alternatives, but 0 is the main alignment indicated by the consensus, with –3 arising as an additional lower scoring possibility from the direct class A–class B alignment (; electronic supplementary material, figure S1; for TM2–TM7, see electronic supplementary material, figures S2–S7). However, the –3 alignment can be eliminated by many of the remaining measures (). The mutagenesis data on CLR and on other class B GPCRs () are especially relevant as they both support the 0 alignment and suggest how TM1 and IL1 can support G-protein interactions and receptor stability. Residues K1.61 and L1.63 are part of the KKLH1.64 motif that is shared between class A and class B (; electronic supplementary material, tables S3 and S5); the first four residues in CLR are K1671.61SLS. Our models and class A X-ray structures show that L1.63 interacts with V3918.50, which in turn holds Y3267.53 in the inactive conformation (cf. the class A NPXXY motif) in its inactive conformation (electronic supplementary material, figure S8). K1.61 also interacts with Gβ in our model of the CLR–G-protein complex—but could also interact with E8.49. Similar stabilizing interactions between IL1 and H8 are seen in most inactive GPCR X-ray crystal structures (e.g. rhodopsin pdb code 1U19, but not the CXCR4, pdb code 3OE6, as it unusually has a positive residue at position 8.49). None of these interactions are possible in the −3 alignment and these interactions probably underlie the loss of function on mutation of K1.61 and L1.63. With some exceptions, e.g. for splice variants, IL1 is highly invariant in length, and so the alignment for TM1 essentially defines the alignment for TM2 and hence the KKLH motif can be considered as a continuation of TM1 and/or TM2.
Figure 3. Class B GPCR mutation data (singles or doubles). (a) CLR mutation data. Green, orange and red shading denote a less than 10-fold, 10–100-fold and greater than 100-fold decrease, respectively, in potency for cAMP production; blue indicates no significant (more ...)
Class A motifs and their class B counterparts.
For TM2, the consensus score strongly favours the 0 alignment (; electronic supplementary material, figure S2). There are essentially no alternatives, as the next highest scoring alignment (−3) is well below the 70 per cent threshold. Strong support for the 0 alignment also comes from the alignment of group-conserved residues, the variability (electronic supplementary material, figure S9), the alignment of the KKLH1.64
motif and from the seven distinguishing mutations. Small group-conserved residues [26
], which allow closer helical packing, align in TM2 and TM3 in the preferred alignment but not in the alternative alignment; in TM1 and TM7 small group-conserved residues also align in the alternative alignments. In addition, our alignment is consistent with a recent Cys-scan of the glucagon receptor which suggests that TM2 remains helical up to Q2022.68
For TM4, the consensus score strongly favours the 0 alignment, and indeed W4.50 aligns in all published class A–class B alignments. There are essentially no alternatives, as the next highest scoring alignment is well below the 70 per cent threshold (; electronic supplementary material, figure S4).
For TM5, the indirect approach via GCR1 clearly favours the preferred 0 alignment. The alternative −2 and +2 alignments arise from the consensus for the direct class A–class B alignment (; electronic supplementary material, figure S5). The most significant factor in determining the alignment arises out of the common interaction with the G-protein through the shared hydrophobic [I/L]xxL5.65
motif at the intracellular end of TM5 (). Residues 5.61 and 5.65 contact the transducin C-terminal peptide in the opsin structures [32
], but a larger range of residues contact the G-protein in the β2
-AR 3SN6 structure. Mutations, particularly to polar amino acids, at position 5.61 and 5.65 in class A [33
] and class B [37
] GPCRs inhibit G-protein coupling. It is important to consider the G-protein interaction when seeking to understand the mutagenesis results for residues at the intracellular end of TM5 Here, the 3SN6 β2
-AR–G-protein complex is a reasonable model since both CLR and the β2
-AR couple to Gs.
In all, the mutation data for positions 5.43, 5.50, 5.57, 5.61, 5.63 and 5.64 (from CLR, CRF, GLP-1, PTH1 and secretin receptors) are consistent with the 0 alignment and not the ±2 alternatives (which arose from the direct alignment). The indirect alignment approach via GCR1, which does not yield any alternatives, is probably more successful than the direct class A–class B approach because of the greater divergence between class A and class B in TM5 (and in TM6).
For TM6, 0 is the preferred alignment, with +3 as the alternative, but while all published class A–class B alignments align the conserved aromatic at position 6.48, the alignment is not as trivial as this match may imply. The class A CWLP6.50
motif has been much discussed as a possible activation microswitch, but since this residue does not change conformation as predicted in class A active GPCRs [20
], its role in class B GPCRs may be equally less dramatic. Most of the mutational evidence against the +3 alignment is discussed in the electronic supplementary material, but the highly conserved T6.37
(electronic supplementary material, table S3) appears to be a key motif shared between class B and some class A GPCRs. This residue contacts R1732.39
in the inactive models but not in the active models (). Mutation of T3386.37
gives rise to constitutive activation in many class B GPCRs [40
is part of the proposed class B DRY equivalent [7
] and like T6.37
is highly conserved (electronic supplementary material, table S3). For these reasons, it is possible that the R1732.39
interaction contributes towards a class B equivalent of the class A R3.50
ionic lock [44
] (see and discussion for further consideration of this residue). This polar lock does not form in the alternative alignments.
Figure 4. The explicit membrane inactive CLR models showing four key activation motifs in spacefill. The YLH3.51 motif is shown with green carbon atoms; the class B DRY equivalent, R2.39, H2.43 and E3.46 is shown with grey carbon atoms. F7.53 corresponding to Y7.53 (more ...)
The TM6 alignment is partly complicated because the conserved prolines do not align. The conserved proline in class B GPCRs is at position 6.42 and like its class A counterpart at position 6.50 is likely to introduce a key bend in TM6, which is important during activation [6
]; it also helps to define the alignment since in the 0 alignment it has the same orientation as P6.50
but any attempt to move the orientation of this proline in CLR results in a loss of function [6
]. The KxxK6.35
motif may interact with the C-terminus of Gαs, aided by flexing of TM6 around P6.42
(class B) or P6.50
(class A). This interaction is born out by MD simulations, but the analogous residues in the β2
-AR–Gs complex (PDB code 3SN6) are poorly resolved and so the interpretation should be used with care, despite the observations from mutagenesis experiments that implicate K6.32
in G-protein coupling in both class A [36
] and class B [39
] GPCRs. For this reason, it is important to note that the alignment is based on mutagenesis results throughout the transmembrane region and not in just one region, such as the G-protein coupling region.
For TM7, the consensus strongly favours the 0 alignment. Two motifs are apparent: the class B equivalent of the NPXXY7.53 motif is VAVLY7.53, of which the Y7.53 appears to be the most significant (F7.53 in CLR) and the NxE[F/V]xxxL8.54 motif on helix 8. N7.57 lies at the interface between TM7 and helix 8; it is conserved in most class B GPCRs (electronic supplementary material, table S3) and many class A GPCRs, but is a gap in the β2-AR. The models illustrate a possible stabilizing role as the highly conserved N3887.57 contacts R1732.39 of the class B DRY equivalent in both the active and inactive models, but would be unable to do so in the +4 alignment. The E394A8.49 mutant in the VPAC1 receptor decreases cAMP production; simulations show that this residue may interact with K1.61 in the inactive receptor (a similar interaction is seen between these positions in approximately 70 per cent of class A GPCR X-ray crystal structures), but it would not be possible in the alternative +4 alignment.