Allelic heterogeneity and geographical distribution of human TAAR1 SNVs.
A list of rare missense TAAR1 SNVs was collated from the NCBI dbGaP database. To identify the geographical distribution of each SNV, the reference SNP cluster ID (rsID) were classified based on five WHO regions, African Region (AR), South-East Asian Region (SEAR), Region of the Americas (ROA), European Region (ER), and Western Pacific Region (WPR). A total of 290 individual TAAR1 SNVs were identified, all of which are segregated at varying capacity, covering all domains of TAAR1 including TM, ICL, ECL, N-terminal extension and C-terminal tail. To visualise the position of individual SNVs and identify allelic heterogeneity at each residue, SNVs were mapped onto a TAAR1 snake plot from GPCRdb (Fig. 1) (40). As highlighted, the current dataset shows 108 residues with a single SNV, which constitutes approximately 37% of this dataset. The remaining 63% are segregated as residues with double, triple, and quadruple allelic heterogeneity. Notably, the highest heterogeneity (four variants) was displayed at six positions; V943.23 (Superscript is position according to Ballesteros & Weinstein numbering system, which indicates helix position of residue relative to the most conserved amino acid within a TM (44)), R1213.50 and M135ICL2, G181ECL2, T2716.55 and M3027.51. In addition, SNVs in two post translational modification sites (Predicted PTM site, identified using GPCRdb) were identified, T672.48, K134ICL2 (40). Overall, this dataset presents with SNVs covering approximately 55% (186 residues) of human TAAR1.
To characterise the geographical distribution and consequently define shared and unshared SNVs, a Venn analysis was performed. As shown in Fig. 2A,C, 20 SNVs are shared across all five geographical regions (referred to as “common SNVs” henceforth) and a total of 75 SNVs are unique. Notably, unique SNVs were most pervasive in SEAR and AR (31 unique SNVs in each) and no unique SNVs were identified in the ROA. To elucidate the prevalence of TAAR1 SNVs regionally, we defined the burden of TAAR1 SNVs by tallying all the unique and shared SNVs present in each region (Fig. 2B). In this dataset, the total burden of TAAR1 mutations were the highest in AR (n = 212) and lowest in the WPR (n = 50).
To best illustrate our overall data, a river plot was utilised (Supplementary Figs. 1–4). Briefly, all plots were composed of two complementary horizontal axes, top axis with the WHO region and bottom axis with the SNVs. Both axes are bridged by up to five groups of “streams”, each representing a WHO region. In this case, the stream can be seen to originate from the axis of WHO region and links to a single/multiple SNVs that is in association with the region. The stream also can be interpreted to originate from the axis of SNVs and linking to a WHO region/s to precisely locate the region/s in association with SNV. Thus, the relative axial area, i.e. the size of each box a given SNV, or a WHO region occupies, represents the sum of associations with its complementary variable. In addition, shared SNVs are presented using the grey shade and unique SNVs are presented with varying colour shades corresponding to the WHO region (e.g. V2887.37A in AR with sky blue).
SNVs at the ligand binding regions of human TAAR1.
The potential impact of identified SNVs on important aspects TAAR1 function, such as agonist binding and signalling remain unexplored. Hence, our analysis focused on studying the SNVs present in the TAAR1 agonist binding and signalling domains. With the emergence of various agonist bound TAAR1 cryo-EM structures, residues critical for differential agonist binding and signalling are beginning to unfold (27–30). Considering this, we performed identical structural mapping analysis using the available cryo-EM structures to map the distribution of SNVs across TAAR1 ligand binding pockets. This was conducted in conjunction with the geographical data to analyse and define the influence of SNVs present in each region.
‘Orthosteric SNVs’ are defined as SNVs at residues surrounding the binding site of various TAAR1 agonists, identified from experimentally determined TAAR1 structures (Fig. 3) including RCSB entries 8W8A, 8W89, 8W88, 8W87, 8JSO, 8JLN, 8JLO, 8JLP, 8JLQ, 8JLR, 8WC8, 8WCA and 8UHB. A total of 19 orthosteric SNVs, affecting 16 residues were identified (Fig. 3A, supplementary table 1). Three common orthosteric SNVs were identified at two positions, I1043.33 (variant; V1043.33) and T1975.45 (variants; I1975.45, S1975.45), which are crucial for the binding of several exogenous and endogenous TAAR1 agonists (Fig. 3B, C). Specifically, residue I1043.33 forms interactions with 3-iodothyronamine (T1AM), amphetamine and other synthetic agonists, while residue T1975.45 interacts with ralmitaront (Fig. 3C) (27–30). As shown in Fig. 3B, six unique orthosteric SNVs were found, including I1043.33S, S1083.37P, F1123.41L, V1504.52I, S183ECL2F and T1945.42A. All residues form key interactions with ralmitaront, with residue specific interactions with other agonists such as T1AM (I1043.33, S1083.37, S183ECL2, T1945.42), ulotaront (I1043.33, and S183ECL2F), amphetamine (I1043.33 and S183ECL2F), methamphetamine (I1043.33 and T1945.42), β-phenylethylamine (PEA) (I1043.33 and T1945.42) and Ro5256390 (I1043.33 and T1945.42) (27–30) as illustrated in Fig. 3C. Apart from F1123.41L and S183ECL2F (both found in ER), all unique orthosteric SNVs belong to SEAR. Notably, shared orthosteric SNVs such as S802.61G (WPR/SEAR), H993.28R (ROA and others, Fig. 3B) and V184ECL2L (SEAR/ER) may influence the binding of endogenous compounds PEA and T1AM (27). Furthermore, residues D1033.32 and W2646.48 interact with all TAAR1 agonists tested thus far (Fig. 3C), with a single shared SNV at each residue: D1033.32N (WPR and SEAR) and W2646.48L (AR, SEAR, and ROA) as illustrated in Fig. 3B (27–30).
SNVs at the micro-switch domains of human TAAR1.
SNVs found at the highly conserved micro-switch domains may influence TAAR1 functionality. Micro-switches are structural motifs conserved across the Class A GPCR superfamily that link ligand binding with conformational changes associated with G protein coupling and receptor activation (29). From the cryo-EM data, it was evident TAAR1 has the typical micro-switches for the classical class A GPCR helical rearrangement (27–30). These micro-switches include the DRY (D1203.49, R1213.50 and Y1223.51), PIF (P2025.50, I1113.40 and F2606.44), CWxP (C2636.47, W2646.48, C2656.49 and P2666.50), and NPxxY motifs (N3007.49, P3017.50, M3027.51, V3037.52 and Y3047.53).
Using structural mapping analysis, we classified SNVs within the four micro-switch domains as ‘micro-switch SNVs’ (Fig. 4A, Supplementary table 2). A total of 16 micro-switch SNVs were identified, located within the DRY (seven SNVs), CWxP (six SNVs), PIF (one SNV) and NPxxY (two SNVs) motifs (Fig. 4B). When considering geographical distribution, three common micro-switch SNV were identified: R1213.50C/S affecting the DRY and N3007.49K affecting NPxxY motif. Furthermore, a total of seven unique micro-switch SNVs were identified; four in AR, two in SEAR and one in the ER as shown in Fig. 4B. Shared variants in micro-switch regions were also identified, one of each affecting DRY, PIF, NPxxY and three affecting CWxP (Fig. 4B).
Although the importance of the micro-switch regions in class A receptor rearrangement and activation are well known, the functional implications of SNVs in these motifs on TAAR1 activity remains largely unknown. Notably, putative deleterious effects of R1213.50C/L, C2636.47G, W2646.48L, P2666.50S/A and N3007.49S/K were previously described by GPCRdb using SIFT and PolyPhen algorithms. Therefore, to characterise the putative effects of TAAR1 micro-switch SNVs, four different in silico algorithms were employed to predict the effect of variants as either tolerable or damaging; SIFT4G, PROVEAN, Mutation Assessor and MutationTaster 2. All micro-switch SNVs were predicted to be damaging by all four selected algorithms (Fig. 4C).
SNVs affecting the G protein-coupling residues of human TAAR1.
Cryo-EM structures of TAAR1 in complex with structurally diverse ligands have revealed TAAR1 agonists engender divergent G protein coupling and downstream signalling outcomes, with mutagenesis revealing residues involved in this preferential G protein activation (27–30). As such, we characterised 9 SNVs at five positions within TAAR1 identified as critical for divergent G protein activation and signalling transduction (referred to as signalling SNVs) (Fig. 5A, B). Two common signalling SNVs were identified, one affecting a residue important for TAAR1-Gαs subunit activity (H55ICL1R), and T2526.36A critical TAAR1-Gαi activity. Two unique SNVs were present in the AR; one affecting residue critical for TAAR1-Gαs activity (Q2205.68E), and one, V1253.54A, involving a residue implicated with creating selective bias for Gαi over Gαs coupling (29). Additional signalling SNVs were shared across multiple regions and may influence TAAR1-Gα-protein interactions based on sites from previous experimental studies, as summarised in Fig. 5B (29).