Two entry tunnels in mouse TAAR9 suggest the possibility of multi-entry tunnels in olfactory receptors

Orthosteric binding sites of olfactory receptors have been well understood for ligand-receptor interactions. However, a lack of explanation for subtle differences in ligand profile of olfactory receptors even with similar orthosteric binding sites promotes more exploration into the entry tunnels of the receptors. An important question regarding entry tunnels is the number of entry tunnels, which was previously believed to be one. Here, we used TAAR9 that recognizes important biogenic amines such as cadaverine, spermine, and spermidine as a model for entry tunnel study. We identified two entry tunnels in TAAR9 and described the residues that form the tunnels. In addition, we found two vestibular binding pockets, each located in one tunnel. We further confirmed the function of two tunnels through site-directed mutagenesis. Our study challenged the existing views regarding the number of entry tunnels in the subfamily of olfactory receptors and demonstrated the possible mechanism how the entry tunnels function in odorant recognition.

Supplementary Figure S11. Elimination of negative charge of some glutamate residues in ECL2 of TAAR9 decreases its ligand binding efficacy. Table S1. Residues in the 21 tunnels predicted by MOLE2.5. Table S2. Residues around two tunnels. Table S3. Values of properties of two tunnels offered by MOLE website. Table S4. Locations and features of 17 residues specific to each tunnel.
Supplementary Data. Amino acid sequences of 50 mouse aminergic receptors. interaction with TAAR5 than with β2AR. (a) Inter-helical non-covalent interactions and interactions with ECLs in β2AR. The similar overall inter-helical interaction pattern in β2AR comparing with that in TAAR9, without interactions between TM1 and TM2. Residues interacting with extracellular domains located in TM2, 3, 4, 5, 7 of β2AR, and TM1, 2, 3, 5, 6 of TAAR9. (b) Inter-helical non-covalent interactions and interactions with ECLs in TAAR5. Apart from interactions that same as Fig. 1b, pication interaction (green) also exists. TM4 only interact with TM3 in TAAR5, while it interacts with TM2, 3, 5 in TAAR9. Other interaction patterns within TMs are analogous with TAAR9. Residues in TM7 interact with extracellular domains in TAAR5 instead of residues in TM1 in TAAR9. (c) The level of conservation in intra-receptor interactions among TAAR5, TAAR9, and β2AR. 4 common interactions are present among these three receptors. In addition, other 10 interactions common in TAAR5 and TAAR9, 4 common in TAAR9 and β2AR, 1 common in TAAR5 and β2AR are observed.
TAAR9, TAAR5, and β2AR have 18, 19, and 16 specific interactions, respectively. Figure S2. 21 tunnels of TAAR9 predicted by MOLE2.5. We applied MOLE2.5 to explore intra-receptor space in TAAR9 and retrieved a result of 21 tunnels (red, green, yellow). We examined these tunnels in detail and found two tunnels which are open to the extracellular space and reach the orthosteric binding sites (blue). We named them Tunnel 1 (green) and Tunnel 2 (yellow), respectively.    Figure S4. Conservation of residues along two tunnels of TAAR9 in aminergic receptors. (a) Radical layout of phylogenetic tree of 50 aminergic receptors. TAAR family is clearly delineated from other aminergic receptors. (b) Conservation of 27 residues along two tunnels of TAAR9 among TAAR family and other aminergic receptors. The level of conservation of a specific site was calculated by the ratio of receptor having this site to the total number of receptors. (c) Pairwise comparison of conservation between TAAR and other aminergic receptors. Residues specific to Tunnel 1 and Tunnel 2 or common to two tunnels all show significant differences (p < 0.05) between TAAR and other aminergic receptors. (d) Details of all 27 residues in each receptor. These residues are more conserved in TAARs than in other receptors, except for D 3.32 , W 7.40 , and Y 7.43 . Some residues are specific in TAARs, such as R 2.64 , S 2.65 , and H 3.28 . Figure S5. Binding poses of spermine predicted by multistep induced-fit docking. (a) Docking of another TAAR9 ligand, spermine, into the receptor revealed three similar binding pockets as those described in Fig. 4a. (b) Spermine forms non-covalent bonds with residues in the orthosteric binding pocket in a similar pattern to spermidine. All binding sites of spermidine are involved in binding of spermine. (c) Spermine can also be docked into vestibular binding pocket 1 and forms salt bridge with D281 6.58 . 16 residues surrounding spermidine in vestibular binding pocket 1 are also observed in the docking posture of spermine. (d) Spermine can also be docked into vestibular binding pocket 2, interacting with E294 7.36 , S93 2.65 , S96 2.68 , and G188 in ECL2. Those residues are also observed to function as binding sites of spermidine in vestibular binding pocket 2. The only difference is that the extracellular residue, E95 2.67 , instead of Q191 45.51 in ECL2 is observed in spermine binding. 23 residues of vestibular binding pocket 2 are within 5 Å range of spermine, that include all 19 residues within 5 Å range of spermidine.

Supplementary
docking. (a) Cadaverine, another TAAR9 ligand, was also docked into three similar binding pockets as those described in Fig. 4a. (b) Cadaverine binding pose in the orthosteric binding pocket is similar to spermidine and spermine. Cadaverine can interact with D112 3.32 , but not with E294 7.36 . Except for Y293 7.35 , other residues with aromatic rings which are within 5 Å range of spermidine also exist in docking result of cadaverine. 13 of 14 residues within 5 Å range of cadaverine are the same as those 13 in the docking of spermidine. (c) Cadaverine can form salt bridges and hydrogen bonds with D281 6.58 in vestibular binding pocket 1. Other interactive residues in TAAR9 including Y293 7.35 and T288 in ECL3 are also found in docking of spermidine. 11 of 13 residues located within 5 Å range of cadaverine are the same as spermidine in vestibular binding pocket 1. (d) Cadaverine can also bind to E294 7.36 in vestibular binding pocket 2. Hydrogen bonds with three residues, S96 2.68 , G188 in ECL2, and Q191 45.51 in ECL2, that are observed in docking results of cadaverine also exist in the docking results of spermidine. 14 of 16 residues located within 5 Å range of cadaverine are observed in docking of spermidine in vestibular binding pocket 2.
induced-fit docking. (a) Docking of 1-(2-aminoethyl)piperidine into TAAR9 shows three similar binding pockets as those described in Fig. 4a. (b) 1-(2-aminoethyl)piperidine can be docked to orthosteric binding pocket which is in common with cadaverine. Another residue, V297 7.39 , was also noted to function in binding 1-(2-aminoethyl)piperidine. 14 of 18 residues located within 5 Å range of 1-(2-aminoethyl)piperidine are the same as those in the docking of spermidine. (c) 1-(2aminoethyl)piperidine can interact with D281 6.58 , Y293 7.35 , and T288 in ECL3 in vestibular binding pocket 1. Another aromatic ring, Y274 6.51 , which is not observed in docking result of spermidine, exist in that of 1-(2-aminoethyl)piperidine. 15 residues are located within 5 Å range of 1-(2aminoethyl)piperidine in vestibular binding pocket 1. (d) 1-(2-aminoethyl)piperidine can bind to E294 7.36 in vestibular binding pocket 2. It can also form hydrogen bond with Q191 45.51 in ECL2. All of the 15 residues located within 5 Å range of 1-(2-aminoethyl)piperidine in vestibular binding pocket 2 are the same as those in the docking of spermidine. Figure S8. Most of the residues within 5 Å of the orthosteric binding pocket of four ligands are in common. Presence of the critical residues within 5 Å of the orthosteric binding pocket (red), vestibular pocket 1 (green), and vestibular pocket 2 (yellow) are shown from the docking results of four TAAR9 ligands. 13 residues in the orthosteric binding pocket are among docking results of four ligands are in common. C116 3.36 and E294 7.36 are common among docking results of three ligands. These 15 residues are defined as the residues constituting the orthosteric binding pocket. 6 residues, including R92 2.64 and 3 residues in ECL2, are common in spermidine and sperminine docking results. Other 4 residues are specific for one ligand docking. In vestibular binding pocket 1, the numbers of residues common for docking results of 4 ligands, 3 ligands, 2 ligands, and 1 ligand are 9, 2, 9, and 5. Hence, there are 11 residues considered as the residues constituting vestibular binding pocket 1. In vestibular binding pocket 2, the numbers of residues common for docking results of 4 ligands, 3 ligands, 2 ligands, and 1 ligand are 11, 7, 2, and 4. Hence, there are 18 residues considered as the residues constituting vestibular binding pocket 2. SPD, spermidine; SPN, spermine; CAD, cadaverine; AEP, 1-(2-aminoethyl)piperidine.