Generic GPCR Residue Numbers - Aligning Topology Maps Minding The Gaps

Abstract

Generic residue numbers facilitate comparisons of e.g. mutational effects, ligand interactions, and structural motifs. The class A GPCR residue numbering by Ballesteros and Weinstein has more than 1100 citations, and the recent crystal structures for class B, C and F now call for community consensus in residue numbering within and across these classes. Furthermore, the structural era has uncovered helix bulges and constrictions that offset the generic residue numbers. The use of generic residue numbers depends on convenient access by pharmacologists, chemists and structural biologists. We review the generic residue numbering schemes for each GPCR class, as well as a complementary structure-based scheme, provide illustrative case stories and GPCRDB web tools to number any receptor sequence or structure.

G Protein-coupled Receptors

G protein-coupled receptors (GPCRs) constitute the largest family of human cell surface receptors [1]. They respond endogenously to ions, neurotransmitters, lipids, carbohydrates, nucleotides, amino acids, peptides and proteins; and also sense light, pain, tastes and odours [2]. Their abundance in human physiological systems, accessibility and druggability have made them a major drug target family – ~30% of the marketed drugs act on GPCRs [3]. The GPCRs are typically classified into the classes A–F [4], or with the alternative system, GRAFS, for the human receptors [5] (Table 1). In the last years, the number of GPCR crystal structures has grown exponentially and today are available for all of the human classes (A–C, F), except the Taste type 2 receptors. These structures have revealed common conformational changes during receptor activation, allosteric modulation by ions, lipids, cholesterol, and water; as well as interactions with the G protein [6–8].

Table 1.

Human GPCR classification and crystal structures

Class [1]	GRAFS [2] Family	Note	# Human receptors [3]	Crystal Structures (7TM domain)
A	Rhodopsin	Includes ~400 olfactory receptors [4]	282 (+ 400 olfactory)	2000: rhodopsin [5], 2007: β2-adrenoceptor; and today: 23 receptors in total [6]
B	Secretin	Also referred to as B1	15	2013: Corticotropin-releasing factor receptor 1 [7] and Glucagon receptor [8]
	Adhesion	Also referred to as B2	33	-
C	Glutamate	Includes Taste type 1 receptors	22	2014: Metabotropic glutamate receptor 1 [9] and 5 [10]
F	Frizzled		11	2013: Smoothened [11]
O (Other)	Taste2	Re-classified as own family [12], although originally grouped with Frizzled [2]	24	-

Note: Classes D and E do not exist in human and are fungal mating pheromone receptors and cyclic AMP receptors, respectively.

Generic Residue Numbers - Maps To Navigate GPCR Topology

All GPCRs share a structural core of seven transmembrane (7TM) helices, making up the machinery for signal transduction across the cell membrane. The 7TM domain contains or is part of the binding site of class A and B1 receptor ligands, and serves as a site for allosteric modulation of class B2, C and F GPCRs [19,20]. So far, 109 GPCRs have been drugged [21], the vast majority with ligands binding within the transmembrane region [22]. The conserved 7TM scaffold allows for the alignment of sequences or structures to identify the corresponding residues, which are indexed with a generic residue number. Such generic residue numbers allow for comparison of e.g. mutant effects, ligand interactions and structure features across receptor subtypes, species orthologs or ligand families. For example, such residue numbers have been used to define the shared ligand-accessible residue positions within the transmembrane bundle [23]. These conserved positions are common placeholders, while receptor sequences vary to compose the unique mosaic responsible for ligand affinity and selectivity.

Class A GPCR Residue Numbering

The Ballesteros–Weinstein numbering scheme [24] is based on the presence of highly conserved residues in each of the seven transmembrane (TM) helices. It consists of two numbers where the first denotes the helix, 1–7, and the second the residue position relative to the most conserved residue, defined as number 50. For example, 5.42 denotes a residue located in TM5, eight residues before the most conserved residue, Pro5.50. The residue numbers can be counted directly within the receptor protein sequence (alignment), however the reference residues are not conserved in all receptors and the numbering is therefore not always straightforward (Class A conservation: N1.50: 98%, D2.50: 90%, R3.50: 95%, W4.50: 97%, P5.50: 78%, P6.50: 99%, P7.50: 88% [25]).

Alternative class A numbering schemes have been presented by Oliveira [26], Baldwin [27,28] and Schwartz [29,30] et al.. They have a common basis by enumerating residue positions from the helix extracellular ends aiming to assign residues located at the same depth in the membrane with the same numbers, e.g. 3.16 and 6.16 (this reverses the TM2, TM4 and TM6 sequences). However, none of the schemes, which use different starting points and numbers, succeeded as GPCR crystal structures have uncovered extensive variations in the length and inclination of transmembrane helices. The alternative schemes also differ by format: Oliveira numbers (the oldest numbering scheme) omit the dot separator to make the numbers computationally more accessible, and Baldwin and Schwartz helix numbers are denoted with roman numerals, I–VII.

Class B, C and F GPCR Residue Numbering

Class B, C and F schemes have been established using the same procedure as the class A Ballesteros-Weinstein system, but use unique reference positions (X.50) so that the residue numbers can be counted directly within the receptor protein sequence (alignment). The class B GPCR Wootten [31] scheme is based on the B1/Secretin subclass, but the reference residues are the most conserved also for five of the B2/Adhesion receptor helices and the remaining two, TM3–4, still have a high conservation (E3.50 58% and W4.50 42%) [25]. It was used in the publications of the crystal structures of both the human glucagon receptor [14] and corticotropin-releasing factor receptor 1 [13]. The class C GPCR Pin [32] numbering was used in the publication of the metabotropic glutamate receptor 5 crystal structure [33]. The class F GPCR Wang scheme was introduced in the recent publication of Smoothened receptor crystal structures [34]. In humans, this is a small class with only 11 members, and in cases where a helix has more than one fully conserved position, the one structurally closest to the class A Ballesteros-Weinstein was used as the reference position. As all schemes use identical formatting, it has been suggested to append the class name (A–F) where clarification is needed, e.g. 3.50b for class B Wootten numbers [35].

Cross-class GPCR Residue Numbering

The low sequence conservation between the GPCR classes has hitherto hindered (correct) sequence alignments, although some inter-class receptor modeling studies correctly aligned the majority of the seven helices (e.g. [36–38]). The structural conservation is higher and the recent crystallographic data has opened up for structure-based sequence alignments from class A to B [13,14,35], C [15,16], and F [17,39]. Some helices display large inter-class lateral deviations or different bending, but as adjacent helices are often translated in the same direction, structural multi-residue motifs with a shared functional mechanism are often conserved across the classes. The published cross-class residue comparisons have utilized the Ballesteros-Weinstein numbers, and where needed together with a class-specific number, e.g. Y7.53a.57b. Furthermore, reference cross-class alignments, based on the available crystal structures, are available in GPCRDB (below). Table 2 shows the alignment of the class specific Ballesteros-Weinstein numbers based on structural alignment of crystal structures of representative receptors from class A (bRho), B (GCGR), C (mGluR1), and F (SMO).

Table 2.

Alignment of the class-specific Ballesteros-Weinstein numbers based on structural alignment of crystal structures of representative receptors from class A (bovine rhodopsin, 12 bRho), B (glucagon receptor, GCGR), C (metabotropic glutamate receptor 1, mGluR1), and F (smoothened receptor, SMO). Class-specific reference residue positions for each of the 7 TM helices are in bold and marked grey.

Helix	Class A (bRho)	Class B (GlucagonR)	Class C (mGlu1)	Class F (SMO)
TM1	G511.46a	S1521.50b	G6031.50c	T2411.43f
	N551.50a	L1561.54b	T6071.54c	T2451.47f
	T581.53a	A1591.57b	V6101.57b	T2481.50f
TM2	L762.43a	H1772.50b	C6311.39c	L2672.42f
	D832.50a	F1842.57b	I6381.46c	F2742.47f
	L842.51a	V1852.58b	F6391.47c	F2752.50f
	F882.55a	S1892.62b	Y6422.50c	S2782.53f
TM3	L1313.46a	E2453.50b	K6783.50c	A3273.46f
	R1353.50a	L2493.54b	I3543.54c	W3313.50f
TM4	W1614.50a	W2724.50b	I7144.40c	W3654.50f
	P1714.60a	W2824.60b	L7244.50c	I3754.60f
TM5	V2115.45x46a	P3105.42	Y7595.46c	P4075.50f
	P2155.50a	A3145.46b	L7635.50c	V4115.54f
	I2195.54a	N3185.50b	C7673.54c	G4155.58f
TM6	L2626.45a	G3596.50b	C7956.47c	L4645.45f
	W2656.48a	E3626.53b	W7986.50c	F4675.47f
	P2676.50a	V3646.55b	A8006.52c	C4695.49f
	Y2686.51a	F3656.56b	F8016.53c	H4705.50f
TM7	A2997.46a	G3937.50b	V8237.40c	F5267.46f
	P3037.50a	A3977.54	L8277.44c	I5307.50f
	-	-	P8337.50c	V5367.56f

Case Story 1: Class A/B common receptor activation motif in TM7

A Tyr residue Y^7.53a.57b, conserved in both class A (Y7.53) and class B (Y7.57) GPCRs, has been proposed to play an important role in the activation of both receptor families (Fig. 1) [40]. In the GCGR (class B GPCR) crystal structure [14] Y400^7.57b forms hydrogen bonds with the conserved T351^6.42b and E245^3.50b residues [35] in a conformation that in class A GPCRs is linked to activation and interaction with the G protein [6,41]. In CRF₁, the thermostabilizing mutation Y363^7.57bA contributes to the shift in the conformation of the receptor towards an inactive state [13].

Figure 1. Class A/B common receptor activation motif in TM7.

Intracellular view of structural superposition of CRF₁ (magenta, PDB: 4K5Y), GCGR (cyan, PDB: 4L6R) and β₂AR (yellow, PDB: 3SN6) receptors. The conserved Tyr residue Y^7.53a.57b and in CRF1₁ the thermostabilizing mutation Y363^7.57bA are shown in stick representation. In GCGR the polar interactions of Y400^7.57b with T351^6.42b and E245^3.50b residues are represented as dotted lines.

Case Story 2: Class A/C unique features of conserved residues in TM6 involved in ligand binding and receptor activation

In both Class A and C, TM6 contains a highly conserved tryptophan at position 6.48a.50c. This residue occupies an equivalent position in both classes, but in the mGluR1 [15] and mGluR5 [16] structures, TM6 has moved laterally away from TM7 compared to Class A. In contrast with Class A, this movement allows W^6.50c to create a hydrogen bond to the backbone carbonyl of 5.44c bridging TM6 to TM5. Despite this conformational difference, the aromatic residues W^6.48a.50c and F^6.51a.53c play a role in the ligand binding in both classes (allosteric ligands for class C), as shown in the class A [6,8] and C [15,16] crystal structures (Fig. 2). Their role in ligand binding and potentially in receptor activation is supported by mutation studies for several class A aminergic [42,43], chemokine [44], adenosine [45], lipid [46]; and class C mGluR1 [47], mGluR5 [48,49] and CASR [50,51] receptors.

Figure 2. Class A/C unique features of conserved residues in TM6 involved in ligand binding and receptor activation.

Side view of Rhodopsin (purple, PDB: 2X72) and mGluR1 (green, PDB: 4OR2). 11-cis-retinal and FITM are shown in line representation. L^5.44c and the aromatic residues W^6.48a.50c and Y/F^6.51a.53c are shown in stick representation. In mGluR 1 the polar interaction between W^6.50c and L^5.44c is represented as a dotted line.

Case Story 3: Class A/B/C/F common hydrophobic core

In the helical bundle, TM3 represents a structural and functional hub [6] being centrally placed with contacts to most all other helices except TM1. TM3 plays a central role in activation [6], which involves the upward movement of TM3 along its axis and a rotation of TM6 [7]. Comparisons of inactive and activated class A receptor structures has led to the proposal of a hydrophobic mechanism hindering activation consisting of the rearrangement of a core of highly conserved hydrophobic residues in 3.43a together with V/I/L6.40a, V/I/L6.41a and F6.44a [7]. This core is well conserved in all human GPCRs: all classes A, B, C and F have a highly conserved hydrophobic residue in position 3.43a.47b.47c.43f (mainly L, M, L and W; respectively) that forms contacts with at least two hydrophobic residues in the equivalent TM6 positions 6.40a.45b.42c.39f, 6.41a.46b.43c.40f and 6.44a.49b.46c.43f (recently described in Fig. 6 in Bortolato et al.) [25,39]. Furthermore, in all classes but C the other side of TM3 has hydrophobic contact with a highly conserved W4.50abf (same index in all classes) on TM4 [25,39].

Extracellular Loop 2 Residue Numbering

Sequence analysis shows that there is a large diversity in the lengths and compositions of the N- and C-termini as well as the extracellular and intracellular loops connecting the TM helices of GPCRs [52]. Nevertheless, for the extracellular loop 2 (EL2) a similar residue numbering scheme has been applied [47,53] in which EL2 residues are labeled 45.X, indicating the location between TM4 and TM5 (“45”) [53]. The reference position (X.50) is a conserved cysteine forming a disulfide bridge with a TM3 residue C3.25a.29b.29c.25f, which is fully (100%) conserved in the classes B, C and F and 88% conserved in class A [25,53]. The second extracellular loops of many GPCRs have approximately the same number of residues upstream and downstream of C45.50 [53]. However, the loop regions of GPCRs are structurally less conserved than the TM helices [6] and EL2 is no exception. Thus, in comparisons across ligand families the structural conservation is limited to the backbone of 45.50 – 45.52. This short stretch is still of high importance because the 45.52 sidechain is often pointing down into the 7TM binding pocket [53]. Longer loop modeling has to be reserved to receptors where there is a template with the same length in TM4-C45.50 and C45.50-TM5, respectively, and guided by experimental restraints. Moreover, the recent GPCR Dock competitions show that although the positions of specific EL2 residues close to C45.50 can be correctly predicted, computational modeling of complete EL2 structures is still highly challenging [54–56].

Mind The Gap – Introducing A Novel Complementary Crystal Structure-based GPCR Residue Numbering

Class A GPCRs have long been known to contain helix kinks in TM5–7 induced by highly conserved proline residues, which are unable to form backbone hydrogen bonds [57]. Upon receptor activation, they act as hinges allowing the helices to tilt inwards to tighten the ligand cavity and outwards to widen the G protein-binding pocket [7]. This does not affect the generic residue numbers, as the best corresponding alignment of residues is obtained knowing that receptor structures are dynamic and the same helix in two receptors can in most cases tilt to adopt a similar position. Consequently, the GPCR 7TM was assumed to be uniformly shaped leading to the long-standing principle to never insert gaps within sequence alignments of the transmembrane helices.

As we entered the structural era, we discovered that the GPCR 7TM frequently contains distortions of another type: bulges and constrictions that are local to one helix turn, but offset the generic numbers of all following residues in the helix. This is because there is one residue extra in a bulged (π-helical) and lacking in a constricted (3₁₀-helical) helical turn, respectively, causing a single unmatched position, i.e. a gap in structure and sequence, when compared to the corresponding undistorted helix. To date, more than 110 crystal structures of 27 unique GPCRs, have uncovered nine bulges (Fig. 3A–I) and six constrictions (Fig. 3J–O) in TM1, TM2, TM4, TM5 and TM7 [58]. These range all classes A, B, C and F; and two constrictions are shared by two classes (Figs. 3N–O). Some helical distortions affect only a specific receptor subtype (Fig. 3J), whereas others are shared by the majority (Figs. 3E,H and O) or all (Fig. 3I and O) members of a GPCR class. Of note, the highly conserved proline residues P6.50×50 and P7.50×50 are associated with kinks in all available class A GPCR structures, but encompass only few (12%, Fig. 3F) and no observed bulges, respectively, i.e. they are missing the backbone hydrogen bond but have no amino acid insertion.

Figure 3. Bulges and constrictions in GPCR class A, B, C and F crystal structures.

Bulged (A–I) and constricted (J–O) helical turns have one extra and lacking residue position, respectively, when superposed to an undistorted helix reference. The structural and sequence alignments of equivalent residues therefor contain a gap, which offsets all following generic residue numbers (grey). The GPCRDB generic residue numbers (black) are structure-based and accounts for bulges and constrictions by numbering an extra residue as the preceding followed by a 1, e.g. 1.40×411, and a skipping a lacking position. To facilitate cross-class comparisons, their alignments have been assigned GPCRDB numbers for all classes. The percentages to the lower right of the structures represent the conservation/frequency of the bulge or constriction within the class, as seen in the GPCRDB alignments [25]. A_2A has two bulges in TM5 (D) giving a double offset, e.g. 5.39 vs. 5×41. Spheres represent the alpha carbon of each residue. Receptors; 5-HT_2B (PDB: 4IB4), A_2A (PDB: 4EIY), β₁AR (PDB: 4AMJ), β₂AR (PDB: 2RH1), CRF₁ (PDB: 4K5Y), H₁ (PDB: 3RZE), P2Y₁₂ (PDB: 4NTJ), S1P₁ (PDB: 3V2Y), SMO (PDB: 4JKV) and mGlu₁ (PDB: 4OR2).

GPCRDB has implemented a complementary structure-based scheme that corrects for bulges and constrictions. The scheme is based on pairwise superposition of inactive receptor structures using ‘What IF’ [59]. Each helix is analyzed separately and a bulge or constriction is defined by comparison to an undistorted reference. The single bulge residue that protrudes the furthest (Fig. 3A–I) is assigned the same number as the preceding residue followed by a 1, e.g. a bulge after residue 46 gets the number 461. The lacking position in a constriction (Fig. 3J–O) is simply skipped in the residue numbering. In uncertain cases where a distortion could be a crystallization artifact, it is put on a “waiting” list of potential bulges/constrictions that require additional crystallographic evidence. GPCRDB numbers are assigned to all distortions, even those that seem to be shared by all members of a GPCR class. This has two reasons, firstly it allows for cross-class structure/sequence alignments, and secondly, it future-proofs the residue numbers by avoiding updates if another class-member with an undistorted helix is crystallized.

To distinguish the GPCRDB scheme, it uses a unique separator x, e.g. 5×46, to denote that it is based on experimental structures, which are predominantly X-ray structures although NMR structures are also taken into account when available. For clarity, it is recommended to use the combination of the sequence- and structure-based numbers, e.g. 2.58×57, unless the comparison only includes crystallized receptors.

Case Story 4: TM2 bulge affects chemokine and opioid receptor binding site residues

While most GPCR crystal structures contain an alpha bulge in TM2 [58] at position 2.55×551, chemokine receptors and opioid receptors contain a S/T^2.56XP^2.58 sequence motif that stabilizes a different helical conformation. Site-directed mutagenesis data probing the TM2–TM3 interface [60] and receptor–ligand interactions in chemokine receptors [44,61] were used to successfully predict the TM2 conformation of CXCR4 and interactions between W94^2.60 and D97^2.63 and the co-crystallized 1T1t ligand prior to the CXCR4-1T1t crystal structure [62] in the GPCR DOCK 2010 challenge [55,63] (Fig. 4A). CXCR4 models based on an ungapped sequence alignment between CXCR4 and the crystal structure templates available at the time of this GPCR modeling competition (Rhodopsin, β₁AR, β₂AR and A_2A) incorrectly oriented W94^2.60 and D97^2.63 towards the membrane layer instead of towards the ligand binding site [55,63]. Residues at positions 2.60 and 2.63 interact with co-crystallized ligands in the CCR5 (W^2.60, Y^2.63), κ (Q115^2.60, V118^2.63) [64], and NOP (Q107^2.60, D110^2.63) [65] crystal structures. Moreover, mutagenesis studies indicate that 2.63 (K108 and N129, respectively) plays a role in δ and μ receptor ligand selectivity [61,66]. These residues structurally align with residues L101^2.64×63 and H93^2.64×63 in β₁AR (Fig. 4B) and β₂AR, respectively, two examples of GPCRs that contain a bulge in TM2. L101^2.64×63 interacts with carmeterol, dobutamine, and carvedilol (Fig. 4B) in β₁AR crystal structures[67,68], while BI-167107 interacts with H93^2.64×63 and FAUC50 covalently binds the H93^2.64×63C cysteine mutant in β₂AR crystal structures [41,69].

Figure 4. TM bulges and constrictions orient different residue numbers in the ligand binding site in GPCR crystal structures.

Comparison of the T^2.56XP^2.58 stabilized region (orange) in TM2 of the CXCR4 chemokine receptor that orients W^2.60 and D^2.63 into the 1T1t binding site [62] (A), and the alpha-bulge in TM2 of the turkey β₁ adrenergic receptor that orients G^2.61×60 and L^2.64×63 into the carvedilol binding site [68] (B). Comparison of the constricted region (orange) in TM4 of the histamine H₁ receptor [70] that orients W^4.56×57 into the doxepin binding site (C), and TM4 of the M₂ muscarinic receptor [72] that orients W^4.57 into the 3-quinuclidinyl-benzilate binding site (D). Ligand carbon atoms are colored magenta. Residues W^6.48 and Y/F^6.51 in TM6 are shown as reference (note that TM6 in panels A-B is rotated ~120 degrees compared to TM6 in panels C–D).

Case Story 5: TM4 constriction affects the alignment of aminergic receptor binding site residues

The TM4 of the aminergic histamine H₁ receptor (H₁R) crystal structure [70] is constricted, directing W158^4.56×57 (an important residue for H₁R-ligand binding based on mutation studies [71]) towards the aromatic ligand binding pocket (Fig. 4C). The aminergic muscarinic M₂ and M₃ receptor crystal structures [72,73] are not constricted in TM4, and the structurally equivalent position of the conserved tryptophan is 4.57. This again lines the ligand binding site and makes hydrophobic/aromatic contacts with the co-crystallized ligand, in accordance with mutation studies of W400^4.57 in M₂ [74] (Fig. 4D). Furthermore, combined mutagenesis and protein-ligand modelling studies to explain ligand selectivity for histamine H₃ (H₃R) and H₄ (H₄R) receptors [75] and H₄R species variants [76] indicate that the residues at position 4.57 (Y167^4.57 in H₃R andN147^4.57 in H₄R) are directed toward the ligand binding pocket. This suggests that the TM4 of H₃R and H₄R are undistorted making the constriction unique for the H₁R subtype.

Numbering Made Easy – Residue Numbering Web Tools At GPCRDB

GPCRDB has been a major community resource for more than 20 years [25,77–79]. It contains reference data; such as the largest collection of receptor mutants, crystal structures, 3D structure models in the inactive and active states and sequence alignments for all species in UniProt. Recently, GPCRDB was equipped with a new suite of interactive web browser tools and diagrams; such as phylogenetic trees, sequence motif search (of e.g. a binding site) and receptor sequence plots (snake and helix box plots). Notably, for all of the above, correct alignment of residues in sequence and/or structure is crucial.

For e.g. pharmacologists, chemists and structural biologists to use generic residue numbers, a more convenient way is needed than the own manual generation of sequence alignments and assignment of residue numbers. Consequently, GPCRDB provides reference alignments and a series of residue numbering tools. The sequence alignments are based on the available crystal structures and gapped to account for bulges and constrictions (Fig. 5A). GPCRDB supports alignments of all classes displaying both the sequence-based Ballesteros-Weinstein (A), Wootten (B), Pin (C) and Wang (F) and structure-based GPCRDB residue numbers. The alignments are followed by a consensus sequence and statistics on residue/property conservation and can be downloaded for further analysis. A pre-defined set in the receptor selection allows for alignment or lookup (below) of only the crystallized receptors.

Figure 5. GPCRDB Residue numbering tools.

GPCRDB offers a suite of residue numbering tools. A) Sequence alignments are gapped to account for bulges and constrictions and list both the structure-based (GPCRDB) and sequence-based residue numbers, for each class. The example, obtain from the structure browser tool, shows the TM2 alignment for two crystallized class A-C receptors and one class F GPCR. B) Lookup tables show generic and receptor-specific residue numbers and can be used to compare the equivalent receptor residues. The example covers TM1 for the class A receptors included in the case stories. C) Structures can be browsed or uploaded in pdb format to assign up to two generic numbering schemes to the transmembrane residues. The example shows the carazolol binding site of the β2-adrenergic receptor (PDB: 2RH1).

Lookup tables show side-by-side listing of receptor-specific and generic residue numbers (Fig. 5B). The tables can list multiple receptors, e.g. subtypes or species orthologs to facilitate the inference of e.g. mutagenesis data or ligand interactions observed in crystal structure complexes. GPCR structures can be browsed or uploaded in pdb format to assign up to two generic numbering schemes to the transmembrane residues (Fig. 5C). The numbers are stored in the pdb file by replacing the B-factor of the C-alpha and carbonyl carbons and scripts are presented for visualization within PyMOL, Maestro and MOE. Finally, to allow for compatibility and re-interpretation of published literature, the Utopia PDF reader [80] can be downloaded to annotate receptor residues with information from GPCRDB.

Potential Exceptions And Future Directions

The sequence- and structure-based numbering systems share some limitations. Generic residues can only be assigned to receptor regions with conserved structural fold, i.e. the 7TM domain, whereas the termini and loops can only be compared, in the best case, within ligand families. Both systems also depend on sequence homology to produce sequence alignments, although this dependency decreases for the GPCRDB numbering as new crystal structure templates become available. In fact, there are 7 alignments – one for each helix. Most receptors and helices have a sufficiently high overall sequence similarity and the bulk of the others can be anchored on highly conserved residue positions (for example the X.50 references). The few remaining are mostly orphan receptors with atypical sequences [81].

The upcoming crystal structures are expected to reveal additional offsets of the sequence-based numbering, while improving the structure-based system. The flip side of the coin is that this requires the GPCRDB numbers to be updated for the given subset of receptors and offset residue positions. The GPCRDB numbering scheme has been designed so that the numbers are adjusted only when a newly determined GPCR structure reveals a new bulge or constriction that has not been predicted by homology. It should be pointed out that all abundant offsets, affecting many receptors, have most likely already been covered. Firstly, all bulges (Fig. 3A–I) have a proline residue 3–5 positions downstream, and in the largest class, A, there are no more highly conserved proline residues in the 7TM that allow for additional frequent bulges. Secondly, most bulges and constrictions have so far only been observed in a single crystal structure and have a low expected frequency, indicating that future adjustments will be limited to small ligand subfamilies or a subset of its members. Consequently, determination of structures covering novel ligand families will be the most valuable.

Conformational plasticity and flexibility is an intrinsic feature of individual receptors that has to be accounted in any analysis and generalization of structural information for GPCRs. So far, generic structure based numbering has shown good tolerance to conformational changes in 7TM region. Thus, our analysis shows that the bulges and constrictions do not change between multiple crystal structures of the same GPCR, when available, even though these structures were determined using different complexes, resolutions, stabilization constructs, and crystal packing forms. Even when relatively large conformational changes between different activation states are considered, the corresponding helical distortions do not involve formation of a full bulge/constriction. This does not preclude a possibility of such changes being found in some of the newly solved GPCR structures in the future – in this case careful analysis should be applied to understand the mechanism and exclude potential artifacts.

Concluding Remarks

The first crystal structures of the class B, C and F GPCRs have opened up the field for receptor function studies and drug design. Also cross-class sequence alignments can now be constructed enabling us to uncover the common cogs and cranks within the 7TM machinery. We have described the schemes for generic numbering of such residue hotspots for chemical structure-activity relationships, pharmacological effects of receptor mutants and structural mechanisms. Furthermore, the multiplicity of structures has uncovered helix bulges and constrictions, making it clear that the correct mapping of equivalent residues requires the gapping of sequence alignments based on crystallographic evidence. Herein, GPCRDB presents such reference structure-based alignments and residue numbers. It is recommended to use the combination of sequence-based and structure-based GPCRDB numbers, e.g. 2.58×57, unless the comparison only includes crystallized receptors.

Acceptance by the GPCR community will initially be determined by simplicity – i.e. counting numbers within the specific sequence (alignment) as compared to looking it up in the GPCRDB alignments or tables. In the longer term, it is more critical how many offsets accumulate in the sequence-based numbering and how many can be accepted whilst still preserving the utility of the system such that the same number actually refers to the equivalent residue across different GPCRs. The GPCRDB numbering system has been fully endorsed by the International Union of Basic and Clinical Pharmacology Committee on Receptor Nomenclature and Drug Classification (NC-IUPHAR). NC-IUPHAR, together with its database GuideToPharmacology [9], will be working closely with GPCRDB to encourage acceptance of this system by the wider GPCR community.

Box 1: Generic GPCR Residue Numbering How-To.

Sequence-based residue numbers are defined by class using the Ballesteros-Weinstein (A), Wootten (B), Pin (C) and Wang (F) schemes in the format X.50 where X is the helix (1–7) followed by the residue position relative to the most conserved (50). Where GPCR crystal structure superposition shows a numbering offset due to helical bulges or constrictions, the structure-based GPCRDB number can be appended separated by an x, e.g. 5.42×43; or used as a substitute, e.g. 5×43. In a similar way, it is possible to combine the equivalent numbers for the different GPCR classes after appending the class name, e.g. 3.43a.47b. Generic numbers can also be combined with receptor-specific numbers by using superscript, e.g. S348^5.47×48 or S5.47×48³⁴⁸. Generic residue numbers can be assigned to any sequence, structure or pdf article at GPCRDB.