Hostname: page-component-586b7cd67f-l7hp2 Total loading time: 0 Render date: 2024-11-24T19:16:29.000Z Has data issue: false hasContentIssue false
  • English
  • Français

Predicted structure of fully activated human bitter taste receptor TAS2R4 complexed with G protein and agonists

Published online by Cambridge University Press:  08 April 2021

Moon Young Yang Open the ORCID record for Moon Young Yang [Opens in a new window] ,
Amirhossein Mafi Open the ORCID record for Amirhossein Mafi [Opens in a new window] ,
Soo-Kyung Kim Open the ORCID record for Soo-Kyung Kim [Opens in a new window] ,
William A. Goddard III Open the ORCID record for William A. Goddard III [Opens in a new window]  and
Brian Guthrie Open the ORCID record for Brian Guthrie [Opens in a new window]
Moon Young Yang
Affiliation:
Materials and Process Simulation Center, California Institute of Technology, Pasadena, CA91125, USA
Amirhossein Mafi
Affiliation:
Materials and Process Simulation Center, California Institute of Technology, Pasadena, CA91125, USA
Soo-Kyung Kim
Affiliation:
Materials and Process Simulation Center, California Institute of Technology, Pasadena, CA91125, USA
William A. Goddard III*
Affiliation:
Materials and Process Simulation Center, California Institute of Technology, Pasadena, CA91125, USA
Brian Guthrie
Affiliation:
Cargill Global Food Research, Wayzata, MN55391, USA
*
*Author for correspondence: William A. Goddard III, E-mail: [email protected]
Rights & Permissions [Opens in a new window]

Abstract

Bitter taste is sensed by bitter taste receptors (TAS2Rs) that belong to the G protein-coupled receptor (GPCR) superfamily. In addition to bitter taste perception, TAS2Rs have been reported recently to be expressed in many extraoral tissues and are now known to be involved in health and disease. Despite important roles of TAS2Rs in biological functions and diseases, no crystal structure is available to help understand the signal transduction mechanism or to help develop selective ligands as new therapeutic targets. We report here the three-dimensional structure of the fully activated TAS2R4 human bitter taste receptor predicted using the GEnSeMBLE complete sampling method. This TAS2R4 structure is coupled to the gustducin G protein and to each of several agonists. We find that the G protein couples to TAS2R4 by forming strong salt bridges to each of the three intracellular loops, orienting the activated Gα5 helix of the Gα subunit to interact extensively with the cytoplasmic region of the activated receptor. We find that the TAS2Rs exhibit unique motifs distinct from typical Class A GPCRs, leading to a distinct activation mechanism and a less stable inactive state. This fully activated bitter taste receptor complex structure provides insight into the signal transduction mechanism and into ligand binding to TAS2Rs.

Keywords

G protein-coupled receptorGEnSeMBLE methodGPCRgustducin G proteinmetadynamicsmolecular dynamics
Type
Research Article
Information
QRB Discovery , Volume 2 , 2021 , e3
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
© The Author(s) 2021. Published by Cambridge University Press

Introduction

Bitter taste perception is generally assumed to protect an organism against potentially toxic substances that are sensed by bitter taste receptors (TAS2Rs). TAS2Rs belong to the G protein-coupled receptors (GPCRs) superfamily with 25 expressed in humans. The TAS2Rs interact with a large and diverse group of compounds. Surprisingly, TAS2Rs are expressed not only in the oral cavity but also in many extraoral tissues, including gastrointestine (Wu et al., Reference Wu, Rozengurt, Yang, Young, Sinnett-Smith and Rozengurt2002), nasal epithelium (Finger et al., Reference Finger, Bottger, Hansen, Anderson, Alimohammadi and Silver2003), airway (Deshpande et al., Reference Deshpande, Wang, McIlmoyle, Robinett, Schillinger, An, Sham and Liggett2010), skin (Shaw et al., Reference Shaw, Mansfield, Colquitt, Lin, Ferreira, Emmetsberger and Reed2018), heart (Foster et al., Reference Foster, Blank, See Hoe, Behrens, Meyerhof, Peart and Thomas2014) and brain (Singh et al., Reference Singh, Vrontakis, Parkinson and Chelikani2011a). It has been reported that TAS2Rs expressed in extraoral tissues are deeply involved in health and disease by contributing to physiological functions, such as sensing potentially harmful molecules and regulating the metabolic/immune system (Lu et al., Reference Lu, Zhang, Lifshitz and ZhuGe2017). To date, a number of studies have reported about TAS2Rs-associated disorders and diseases including food intake (Andreozzi et al., Reference Andreozzi, Sarnelli, Pesce, Zito, Alessandro, Verlezza, Palumbo, Turco, Esposito and Cuomo2015), insulin homeostasis (Dotson et al., Reference Dotson, Zhang, Xu, Shin, Vigues, Ott, Elson, Choi, Shaw, Egan, Mitchell, Li, Steinle and Munger2008), cardiovascular disease (Shiffman et al., Reference Shiffman, O’Meara, Bare, Rowland, Louie, Arellano, Lumley, Rice, Iakoubova, Luke, Young, Malloy, Kane, Ellis, Tracy, Devlin and Psaty2008) and cancer (Martin et al., Reference Martin, Nachtigal, Selman, Nguyen, Sasman, Dellaire and Dupré2019).

Despite crucial roles in physiological functions and related diseases, investigations on TAS2Rs have been hampered by the lack of structural information. Neither X-ray crystal nor cryo-electron microscopy (EM) structures are available for TAS2Rs. Thus the only structural information is from several predicted structures based mostly on homology modelling using Class A GPCRs (Singh et al., Reference Singh, Pydi, Upadhyaya and Chelikani2011b; Di Pizio & Niv, Reference Di Pizio and Niv2015; Pydi et al., Reference Pydi, Jaggupilli, Nelson, Abrams, Bhullar, Loewen and Chelikani2015; Pándy-Szekeres et al., Reference Pándy-Szekeres, Munk, Tsonkov, Mordalski, Harpsøe, Hauser, Bojarski and Gloriam2018). Such homology models for GPCRs are problematic because the seven transmembrane domains (TMD) form nearly parallel α-helices so that their axial orientation is highly sequence dependent. Moreover, although TAS2Rs are generally classified as Class A GPCRs, the sequence similarity of TAS2Rs with other class A GPCRs is much less than 30% (Di Pizio et al., Reference Di Pizio, Levit, Slutzki, Behrens, Karaman and Niv2016) making homology modelling even less reliable. To ameliorate this problem we developed a systematic approach to take a GPCR with a known structure as a template to fix the initial tilts for the seven TMDs and then sampling first all coupled axial rotations (BiHelix step, 35 million combinations) (Abrol et al., Reference Abrol, Bray and Goddard2012) and then various combinations of tilts and rotations (SuperBiHelix, 13 trillion combinations) (Bray et al., Reference Bray, Abrol, Goddard, Trzaskowski and Scott2014) combined with side chain optimization (SCREAM) to finally select the lowest 25 conformation likely to play a role in ligand binding. This methodology, referred to as the GPCR ensemble of structures in membrane bilayer environment (GEnSeMBLE), has been successful for predicting other GPCRs, including Class A [CCR5 (Berro et al., Reference Berro, Yasmeen, Abrol, Trzaskowski, Abi-Habib, Grunbeck, Lascano, Goddard, Klasse, Sakmar and Moore2013; Abrol et al., Reference Abrol, Trzaskowski, Goddard, Nesterov, Olave and Irons2014), AA3 (Goddard et al., Reference Goddard, Kim, Li, Trzaskowski, Griffith and Abrol2010), DP prostaglandin (Shankar et al., Reference Shankar, Goddard, Kim, Abrol and Liu2018) and CB1 (Ahn et al., Reference Ahn, Scott, Abrol, Goddard and Kendall2013; Scott et al., Reference Scott, Abrol, Ahn, Kendall and Goddard2013)], Class B [GLP1 (Kirkpatrick et al., Reference Kirkpatrick, Heo, Abrol and Goddard2012)] and Class C [TAS1R2/1R3 sweet heterodimer (Kim et al., Reference Kim, Chem, Abrol, Goddard and Guthrie2017)], but this is the first application of this methodology to TAS2Rs [an earlier application of some of these tools was used to predict the structure of TAS2R38 (Tan et al., Reference Tan, Abriol, Trzaskowski and Goddard2012)].

Having the structure for the GPCR is just the first step. To describe the signal transduction we must consider how binding of agonist and G protein (GP) to the GPCR leads to activation. We find that binding an agonist to the GPCR-GP complex results in a conformational change of the GPCR that is transmitted to the bound GP, leading to opening the tight GP-GDP complex to form the activated state with an open Gα subunit. In our recent report for κ opioid receptor (κOR), we found that the Gi protein (GiP) makes strong anchors to all three intracellular loops (ICLs) of the κOR, stabilizing the active conformation of GPCR (Mafi et al., Reference Mafi, Kim and Goddard2020a, Reference Mafi, Kim and Goddard2020b). This indicates that both ligand and GP interacting with the GPCR lead to structural changes that stabilize the final activated state. Therefore, to understand the factors responsible for attaining the fully activated GPCR-agonist-GP complex structure, we need to understand the role of these three components. This should help develop a structural basis for developing new more selective and active agonists or antagonists.

In this study, we describe the prediction of the three-dimensional (3D) structure of the fully activated TAS2R4 human bitter taste receptor with several agonists using the GEnSeMBLE (Abrol et al., Reference Abrol, Bray and Goddard2012; Bray et al., Reference Bray, Abrol, Goddard, Trzaskowski and Scott2014) and DarwinDock (Griffith, Reference Griffith2017) complete sampling methods. We find that the agonist binding modifies the hydrogen-bond (HB) network in the binding pocket, which triggers a shift in the conformation of Y6.48, which probably acts as a transmission switch similar to that observed in typical Class A GPCRs. Moreover, we find that the GP, gustducin heterotrimeric protein (responsible for the bitter and basic taste; hereafter referred to as Ggust), binds to the TAS2R4 by forming salt bridge (SB) anchors to each of the three ICLs of TAS2R4, orienting the Gα5 helix to make extensive interactions with residues in the cytoplasmic region of TAS2R4. We also find several unique motifs conserved only across TAS2Rs that result in an activation mechanism distinct from that of typical Class A GPCRs. The structure of this fully activated TAS2R4-agonist-Ggust complex provides information that should be useful in understanding universal and diverse features across other TAS2Rs and for other subfamilies of GPCRs.

Results and discussion

Structural prediction for TAS2R4

Since no crystal structure is available for TAS2Rs, we predicted the 7-helix TMD of TAS2R4 using the GEnSeMBLE complete sampling technique. This method has successfully predicted 3D structures of many GPCRs (Goddard et al., Reference Goddard, Kim, Li, Trzaskowski, Griffith and Abrol2010; Abrol et al., Reference Abrol, Bray and Goddard2012; Kirkpatrick et al., Reference Kirkpatrick, Heo, Abrol and Goddard2012; Tan et al., Reference Tan, Abriol, Trzaskowski and Goddard2012; Ahn et al., Reference Ahn, Scott, Abrol, Goddard and Kendall2013; Berro et al., Reference Berro, Yasmeen, Abrol, Trzaskowski, Abi-Habib, Grunbeck, Lascano, Goddard, Klasse, Sakmar and Moore2013; Scott et al., Reference Scott, Abrol, Ahn, Kendall and Goddard2013; Abrol et al., Reference Abrol, Trzaskowski, Goddard, Nesterov, Olave and Irons2014; Bray et al., Reference Bray, Abrol, Goddard, Trzaskowski and Scott2014; Kim et al., Reference Kim, Chem, Abrol, Goddard and Guthrie2017; Shankar et al., Reference Shankar, Goddard, Kim, Abrol and Liu2018). The first step in GEnSeMBLE is to select a template from known structures to specify the initial tilts of the seven TMDs. Despite its low sequence similarity, most analyses support the classification of TAS2Rs with Class A GPCRs (Di Pizio and Niv, Reference Di Pizio and Niv2015; Di Pizio et al., Reference Di Pizio, Levit, Slutzki, Behrens, Karaman and Niv2016). Therefore, based on the sequence alignment for Class A GPCRs for which active structures are available, we selected four templates with highest homology to TAS2R4 for the GEnSeMBLE predictions (Table S1)); serotonin 2C (5-HT2C), angiotensin 2 (AT2), cannabinoid 1 (CB1) and rhodopsin. For each template we mapped the residues to TAS2R4 and examined coupled 30° rotations about the tilt axis of all seven TMDs. This BiHelix step examines 12**7 = 35 million combinations from which we select 2,000 that we build into 7-helix bundles while reoptimizing side chain conformations. From these 2,000 conformations we examined the 10 lowest energy shown in Table S2. We then selected two structures based on energy and diversity for the SuperBiHelix step that considers changing the tilt polar angle (θ) by ±10°, the tilt azimuthal angle (ϕ) by ±15°, ±30° and each axial angle (χ) by ±15°, ±30° for a total of (3 × 5 × 5)**7 = 13 trillion conformations. For each of these two cases, we select the 2,000 from SuperBiHelix predicted to have the lowest energy that we build into 7-helix bundles while reoptimizing side chain conformations by SCREAM. From this 2 × 2,000 we selected the best 25 by energy and diversity, as shown in Table S3, for docking each agonist. From these 25 we finally selected the energetically most favourable structure obtained from the 5-HT2C template for further study (Fig. S1).

Ligand binding

Ligand binding site

Over 30 compounds are known to activate TAS2R4 (Wiener et al., Reference Wiener, Shudler, Levit and Niv2012). Of these, 10 ligands show binding constants of 100 μM effective concentration or stronger. Many of these ligands are not-specific to TAS2R4 (Table S4). Rubusoside (Rubu), a steviol glycoside isolated from the Chinese sweet tea plant Rubus suavissimus, is known to bind to TAS2R4 with 50 μM effective concentration (and to TAS2R14 with 400 μM effective concentration). Although steviol glycosides are generally known as nonnutritive sweeteners, they elicit a lingering bitter aftertaste that is mediated by TAS2R4 and TAS2R14 (Hellfritsch et al., Reference Hellfritsch, Brockhoff, Stahler, Meyerhof and Hofmann2012). Thus, we focus on Rubu as an agonist of TAS2R4, which we compare to quinine, a well-known bitter compound that binds to several bitter receptors, including TAS2R4.

We used DarwinDock to predict the binding poses of Rubu and quinine to the most favourable 7-TMD conformation of TAS2R4 predicted by GEnSeMBLE (Fig. S2). DarwinDock considers ~50,000 poses of the ligand in the protein binding site and selects the top 100 by energy. In this process the six hydrophobic residues (I, L, V, F, Y and W) are replaced by Ala to allow space for sampling, but after selecting the top 100 poses by energy, we use SCREAM to add back these hydrophobic residues independently to each pose to provide each ligand pose a unique environment (Li et al., Reference Li, Kim, Goddard, Chen and Tan2015). Then we select the lowest energy poses for further analysis.

We find that Rubu forms HBs to one residue in TM3 (D923.36) and two in TM7 (K2627.35 and S2637.36) (the superscript is Ballesteros–Weinstein GPCR numbering (Ballesteros and Weinstein, Reference Ballesteros and Weinstein1995)). Charged (or polar) residues in TM3 of the binding pocket often play a crucial role in the ligand recognition and binding for Class A GPCRs, such as D3.32 for β2 adrenergic receptor (β2AR) and all opioid receptors. Similarly, we find that D923.36 plays an important role in Rubu binding to TAS2R4, as discussed in detail below. This D923.36 in TAS2R4 is unique among TAS2Rs, where most TAS2Rs have asparagine at this position (Fig. S3). Thus D923.36 may play an important role in selectivity for TAS2R4.

Quinine is known to have a protonated nitrogen at physiological conditions (Thompson et al., Reference Thompson, Lochner and Lummis2007), and our docking studies find that it makes a SB with D923.36. However, quinine is known to bind to nine other TAS2Rs that do not have a negatively charged residue at position 3.36, yet they lead to similar or even higher binding affinity (Wiener et al., Reference Wiener, Shudler, Levit and Niv2012). Thus we expect that formation of this SB is not likely to be essential for binding of quinine and may not induce a higher binding affinity. In addition to this interaction with D923.36, quinine forms HBs to residues in TM5 (S1845.46) and TM6 (Y2426.51). These interactions of Rubu and quinine with TAS2R4 obtained from docking evolved during MD simulations after inserting in lipid bilayer and water box, as discussed below.

Steviol glycosides have a steviol as a core structure, containing a carboxyl hydrogen (R1 side) and a hydroxyl hydrogen (R2 side) at each end, with R1 and R2 replaced by various sugars (glucose or rhamnose) as shown in Fig. 1a. Rubu is the smallest among steviol glycosides, with just a glucose ring at each end of the core structure (Table S5). During the MD simulations with full lipid bilayer and solvent for 480 ns, Rubu made interactions with residues in TM3, TM4, TM5, TM6 and extracellular loop 2 (ELC2) (Fig. 1c, S4A and Table S6). Due to the hydrophobic core and the hydrophilic sugar rings of Rubu, residues in the middle of the binding pocket and close to the steviol core contribute to the binding mainly through hydrophobic interactions, whereas those close to the sugar rings are involved in HBs and polar interactions. The MD simulations were carried out with the GP bound (TAS2R4-agonist-Ggust); interactions between GPCR and Ggust are discussed in section ‘TAS2R4 in complex with the full heterotrimeric gustducin G protein.

Fig. 1. Chemical structures of (a) the steviol glycosides and (b) quinine, where sugar (glucose or rhamnose) is attached at R1 and R2 in (a). Binding sites for (c) Rubu and (d) quinine to TAS2R4. Rubu has HBs to residues in TM3 (F883.32, M893.33, D923.36, S933.37), TM4 (Y1474.62), TM5 (L1775.39, L1815.43), TM6 (Y2506.59) and ELC2 (E158, T162, N164). Quinine has a SB to D923.36 and HBs to residues in TM3 (F883.32, M893.33), TM5 (L1815.43, Q1885.50) and TM6 (Y2426.51).

In comparison, quinine is relatively small, interacts with fewer residues in TM3, TM5, and TM6 (Fig. 1b,d, S5A and Table S6). Van der Waals or hydrophobic interactions dominate, except for the SB interaction of the quinine tertiary nitrogen with D923.36.

In addition to Rubu, other bulkier steviol glycosides can also activate TAS2R4 (Hellfritsch et al., Reference Hellfritsch, Brockhoff, Stahler, Meyerhof and Hofmann2012), although experimental result shows that those are not as bitter as Rubu (Table S5). For example, Rebaudioside M (RebM) has six sugar rings (three on each side of the core) and exhibits moderate bitterness. To investigate how such bulkier ligands bind to TAS2R4 and their activity (bitterness), we predicted an initial binding pose of RebM by matching to that of Rubu. This is because its large size complicates getting accurate results directly from docking. Then we performed MD simulation to equilibrate RebM in the binding site. We find that RebM makes more interactions with residues in TM2, TM7 and ECL1, compared to Rubu (Figs S6A, S6B and Table S6). It has been reported that TM1, TM2 and TM7 are more involved in binding of antagonists (Di Pizio et al., Reference Di Pizio, Levit, Slutzki, Behrens, Karaman and Niv2016), which may be a reason for RebM being much less bitter than Rubu. Although the six sugar rings of RebM lead to a number of degrees of freedom with many possible conformations of the sugar rings that might lead to a different binding mode, the TAS2R4-RebM-Ggust complex seems converged after 480 ns of MD simulations (Fig. S6C).

A derivative of RebM, hydRebM, exhibits much less bitterness, despite a very similar structure (Table S5). The only difference with RebM is that an alkene group in the core structure is substituted with hydroxyl and methyl groups. In comparison, isoRebM, in which the double bond of the alkene is shifted to the next carbon in the ring, shows bitterness similar to that of RebM (Table S5). These dramatic differences in activation for similar structures indicate that the hydroxyl group in hydRebM must play a role in suppressing activation. Indeed as described below our MD simulations show that this hydroxyl group makes an internal HB with one of R2 sugars, which may hinder the proper positioning of the ligand for binding and activation (Figs S7 and S8B). This hydroxyl group also makes an HB with Y2506.59, which may lead to a change in the hydRebM binding pose or it may act to restrict the structural change of TM6 for activation, but this interaction was not retained during the dynamics (Fig. S8C). For the Rubu and RebM systems, we found that Y2506.59 contributes to the binding, but mainly through hydrophobic interactions (Table S6). To fully understand the difference in bitterness across steviol glycosides, it may be necessary to investigate the transition in which the GP induces the inactive state to form the fully active state.

The transmission switch of Y2396.48

Among the residues in the binding pocket, D923.36 likely plays a particularly crucial role in ligand binding. It forms two stable HBs with Rubu and a SB with quinine, contributing to almost ¼ and ½ of the total binding, respectively (Fig. 1, S4C and S5C and Table S6). Not only does D923.36 provide a large energy stabilization via polar interactions, but it contributes to transferring the signal for activation by changing the HB network (Fig. 2). Thus in the apo protein structure from GEnSeMBLE, we find that D923.36 makes an HB interaction with Y2396.48. But the presence of agonist causes the side chain of Y2396.48 to break this interaction with D923.36 so that it rotates anti-clockwise. Indeed, it is well-known that the highly conserved W6.48 in Class A GPCRs plays a crucial role in activation, where it is known as a ‘transmission switch’ (Zhou et al., Reference Zhou, Yang, Wu, Guo, Guo, Zhong, Cai, Dai, Jang, Shakhnovich, Liu, Stevens, Lambert, Babu, Wang and Zhao2019). Although Y2396.48 is not as highly conserved in TAS2Rs as is W6.48 in typical Class A GPCRs, this significant movement of Y2396.48 upon ligand binding suggests that this conformational change may contribute to the rotation and structural change in TM6, which is one of most significant changes between inactive and active states observed in Class A GPCRs.

Fig. 2. (a) Top and (b) side views showing D923.36 – Y2396.48 interactions in the absence (cyan) and presence (tan) of the agonist, but the agonist is omitted for clarity.

To validate this hypothesis, we prepared two structural models without an agonist. The first model (model 1) is the one constructed by GEnSeMBLE (Fig. S1), where it was equilibrated with neither agonist nor Ggust, although this model was constructed using an activated GPCR as a template. The second model (model 2) is the pre-activated structure, which contains no agonist but is coupled to the inactive Ggust coupled to GDP. This is motivated by our discovery that GPs form anchors that induce the conformational change of the receptor (Mafi et al., Reference Mafi, Kim and Goddard2020a) discussed in the following section. This led to our G protein-first hypothesis, in which GP binds first to the receptor to initiate activation followed by agonist binding to complete activation. Both models find that Y2396.48 maintains a stable direct or water-mediated HB with D923.36 for 100 ns of MD simulation (Fig. S9A). A further metadynamics simulations (metaMD) based free energy analysis shows an energy minimum corresponding to a water-mediated HB between D923.36 and Y2396.48 (Fig. S9B). Thus, Y2396.48 and D923.36 play crucial roles in the activation process, just as does W6.48 for typical Class A GPCRs.

We attribute the decreased bitterness of RebM and hydRebM relative to Rubu to the interaction of these ligands with Y2396.48. Compared to Rubu, both RebM and hydRebM have two additional sugar rings attached to the first sugar ring at the R1 side, and one of these sugars can reach Y2396.48 to make an HB (Fig. S10). This additional interaction may hinder the conformational change of Y2396.48 needed for activation, resulting in suppressing activation. A similar interaction may also account for the lower bitterness observed in other bulky steviol glycosides (Table S5).

TAS2R4 complexed with the full heterotrimeric gustducin G protein

Although structures for many GPCRs with ligand binding sites have been predicted successfully using GEnSeMBLE and DarwinDock, it remains a challenge to predict a fully activated structure with decisive information for understanding the signal transduction mechanism and for developing new therapeutics. This is because the activation process requires binding of both agonist and the cognate GP. Both components are usually essential for activating GPCRs to induce subsequent signal transduction. In our recent MD and metaMD studies of κOR and μOR, we discovered that these GPCRs binds strongly to its cognate GP via SB and HB interactions at all three ICLs (Mafi et al., Reference Mafi, Kim and Goddard2020a). These interactions were not identified in the cryo-EM structures because of the low resolution and disorder. To obtain this fully activated structure for TAS2R4, we used the lessons from κOR and μOR to predict the complex structure. Here we used the gustducin heterotrimeric protein, which is responsible for the bitter taste.

Indeed, we found essentially the same interactions for the TAS2R4-Ggust complex as for GiP with κOR and μOR. Fig. 3a shows the equilibrated TAS2R4-Rubu-Ggust structure immersed in the lipid bilayer. For the 480 ns MD simulation, the root mean square deviation (RMSD) of the TAS2R4 was within 3 Å while that of the whole protein complex was 4 Å, showing the stability of the complex structure (Fig. S4B).

  • • For anchor 1, we find that two positive residues in ICL1 form SBs with two negative residues in the Gβ subunit, K37ICL1 – D312Gβ and R40ICL1 – D333Gβ (Fig. 3b).

  • • For anchor 2, we find that R123 in ICL2 forms SBs with both E25Gα and E28Gα of the Gα subunit (Fig. 3c).

  • • For anchor 3, we find that two lysine residues in ICL3 interact with both the Gα5 helix and with the Ras-like domain of the Gα subunit, K209(ICL3) – D341(D337)Gα and K212ICL3 – E318(D316)Gα (Fig. 3d).

Fig. 3. (a) The structure of TAS2R4-Rubu-Ggust complex. Important polar interactions at three anchors: (b) anchor 1: ICL1 with Gβ, (c) anchor 2: ICL2 and GαN and (d) anchor 3: ICL3 and Gα. (e) Polar interactions between the Gα5 helix of the Gα subunit with TAS2R4.

These residues involved in anchor formation were well positioned when we superimposed the mouse μOR-GiP structure with TAS2R4-Ggust during the model construction step (Fig. S11). A free energy analysis using metaMD shows that these SBs should form spontaneously with activation of Ggust as it approaches TAS2R4, stabilizing the structural complex (Fig. S12). The three anchors formed between TAS2R4 and Ggust were quite stable over ~0.5 μs MD simulation, although occasional fluctuations and switching of SB pairs were observed (Fig. S4D–F).

We found similar interactions between TAS2R4 and Ggust for the TAS2R4-quinine-Ggust complex structure (Fig. S5), and also for the RebM and hydRebM systems (Figs S6 and S7). Therefore, we conclude that this formation of anchors between Ggust and the ICLs of the GPCR is likely a common feature across all TAS2Rs and Class A GPCRs.

It is well known that activation of the GPCR-agonist-GP complex involves coupling of the Gα5 helix of the Gα subunit with residues deep in the GPCR (Onrust et al., Reference Onrust, Herzmark, Chi, Garcia, Lichtarge, Kingsley and Bourne1997; Oldham et al., Reference Oldham, Van Eps, Preininger, Hubbell and Hamm2006; Oldham & Hamm, Reference Oldham and Hamm2008). Indeed for κOR and μOR we found that Gα5 forms a network of HBs and SBs to the GPCR (Mafi et al., Reference Mafi, Kim and Goddard2020a). For these opioid receptors, a conserved arginine residue in ICL2 (R170 for κOR and R179 for μOR respectively) plays an important role in the interaction with Gα5, forming a SB with D350Gα5. For TAS2R4, we find that the highly conserved K1093.53 over all TAS2Rs plays the same role, forming a SB with D350Gα5 (Fig. 3e).

For κOR and μOR the terminal carboxylate of Gα5 (F354) is also important in proper positioning of the Gα5 helix, making a SB with R6.32 in TM6 in the fully activated structure. For most TAS2Rs, we expect an analogous interaction because most TAS2Rs have a conserved positively charged residue in TM6 (6.32 or 6.36 position) (Fig. S3). However TAS2R4 does not have a positively charged residue at either the 6.32 or 6.36 position. Therefore, we expect that F354Gα5 would make a SB with an alternative neighboring positive residue in TAS2R4. Our predicted structure leads to two candidates for this interaction, K2306.39 and K288H8, both of which are well conserved (Figs 3e and S3). To test these possibilities we first constructed a K2306.39 – F354Gα5 pair and carried out the MD simulation, finding that it changed to the K2306.39 – K288H8 pair (Fig. S4G). For the TAS2R4-quinine-Ggust complex structure, F354Gα5 formed a SB with both of these lysine residues, alternating during the MD (Fig. S5G). For the RebM system F354Gα5 formed a stable SB with K2306.39, whereas hydRebM shows an alternative interaction of F354Gα5 with either K2306.39 or K288H8 (Figs S6H and S7H). Although it is unclear whether this interaction should depend on the ligand, we consider it likely that the terminal carboxylate of Gα5 makes a SB with either or both of K2306.39 and K288H8 in TAS2R4.

We expect that the final position of the Gα5 helix in the activated structure may be achieved in stages. There are reports that there is a stepwise sequential interactions of the Gα5 helix from its first contact toward a deeper insertion that induces opening the α-helical domain and GDP release (Du et al., Reference Du, Duc, Rasmussen, Hilger, Kubiak, Wang, Bohon, Kim, Wegrecki, Asuru, Jeong, Lee, Chance, Lodowski, Kobilka and Chung2019). Indeed, we find that bringing the GDP bound Ggust towards the GPCR leads to an initial SB contact of F354Gα5 with the highly conserved K1093.53 (Fig. S13). As the GP is activated the Ras-like domain and the α-helical domain of the Gα subunit sandwiched around the GDP open up so that GDP is exposed to exchange with GTP. Thus in the activation mechanism, the Gα5 helix first interacts with the receptor through the F354Gα5 – K1093.53 SB interaction, but then the helix gradually inserts more deeply upon activation. During this activation process, D350Gα5 makes a new SB with K1093.53 while F354Gα5 forms a SB with K2306.39 (or K288H8) of TAS2R4. This triggers the opening of the Gα and eventual GDP-GTP exchange signaling. Interestingly, even for the structural model with the inactive Ggust (tightly coupled to the GDP), residues involved in the anchor formation were placed sufficiently close enough to make SBs (Fig. S13C–E). This suggests that the anchors form first, to help to position the Gα5 helix for the insertion. In our previous study of the κOR and μOR in which we discovered the formation of GPCR-GiP anchors that induce conformational changes in the receptor, we suggested the ‘G protein-first’ paradigm, in which GP binds first to the receptor to initiate activation, followed by agonist binding to further open the cytoplasmic region of the receptor to complete the activation with GDP-GTP exchange (Mafi et al., Reference Mafi, Kim and Goddard2020a). We suggest that this applies to the TAS2Rs also.

Conserved motifs in TAS2Rs

The 1-2-7 interactions

Although TAS2Rs are generally considered to belong to Class A, this classification of TAS2Rs is somewhat ambiguous due to the low sequence similarity. Moreover, it is known that TAS2Rs have a quite different pattern of conserved motifs compared to typical Class A GPCRs (Di Pizio et al., Reference Di Pizio, Levit, Slutzki, Behrens, Karaman and Niv2016). For example typical Class A GPCRs have highly conserved N1.50, D2.50 and N7.49 in the NPxxY motif that can coordinate a sodium ion (or protonated H2O) and is known as the sodium binding pocket (Liu et al., Reference Liu, Chun, Thompson, Chubukov, Xu, Katritch, Han, Roth, Heitman, IJzerman, Cherezov and Stevens2012). Based on high-resolution X-ray structures, the sodium ion is coordinated to the highly conserved D2.50 and S3.39 sites plus water molecules, stabilizing the inactive conformation. This sodium ion binding pocket is collapsed by the structural changes in the GPCR accompanying the activation process. For TAS2Rs, however, the aspartate that plays a critical role in coordination of the sodium ion for Class A is replaced with highly conserved arginine (Fig. S3). Thus, a sodium ion is unlikely to bind at this site as observed in typical Class A GPCRs. For TAS2Rs, mutagenesis studies show that the S7.50A mutation induces hyperactivity (Pydi et al., Reference Pydi, Bhullar and Chelikani2012), leading Pydi et al. to propose that S7.50 stabilizes the inactive state by forming an HB with R2.50. However our MD simulations find that in the active state, R552.50 forms an HB network with N241.50, H2767.49 and S2777.50 (Fig. 4a). We found that this HB network is stable and maintained for the full ~0.5 μs MD simulation (Fig. S14A–D). Moreover, our metaMD shows that this HB network is energetically favourable (Fig. S14E). All residues involved in this TM1–2-7 interaction are highly conserved, indicating that this HB network plays an important role in the activation process for TAS2Rs.

Fig. 4. TM1–2-7 interactions (a) in the active state and (b) in the inactive state, respectively.

On the other hand, for the inactive structure constructed using GEnSeMBLE based on the inactive 5-HT2C template and then equilibrated (Table S7 and S8), we finds that R552.50 makes an HB with S953.39, instead of N241.50, H2767.49, or S2777.50 (Fig. 4b). Note that S3.39 is also a highly conserved residue in Class A GPCRs, where it plays a role in binding sodium together with D2.50. In contrast, S3.39 is less conserved in TAS2Rs, where it is often mutated to threonine or asparagine (Fig. S3). Since the long side chain of R552.50 can make a direct HB with S953.39 and there is no role for the sodium ion as in Class A GPCRs, there has presumably been less evolutionary pressure at this site in TAS2Rs, since it need only to make an HB with R552.50. Our metaMD studies also indicate that R552.50 prefers to interact with S953.39 rather than N241.50 (Fig. S15). These results indicate that R2.50 likely plays an important role in activation, just as does D2.50 for typical Class A GPCRs, but differently by stabilizing either the inactive or active state by changing the HB network together with N1.50, S3.39, H7.49 and S7.50.

The DRY motif and E6.30

The D(E)3.49R3.50Y3.51 motif together with E(D)6.30 is another well conserved motif in typical Class A GPCRs. The R3.50 – E(D)6.30 SB interaction is considered to play a role in maintaining the inactive state, where it is known as the ‘ionic lock’ (Zhou et al., Reference Zhou, Yang, Wu, Guo, Guo, Zhong, Cai, Dai, Jang, Shakhnovich, Liu, Stevens, Lambert, Babu, Wang and Zhao2019). For TAS2Rs, however, neither the DRY motif nor E(D)6.30 is conserved, and there is no alternative residue at a nearby position to form a SB for the ionic lock. Some Class A GPCRs, such as the opioid receptors, possess an HB that stabilizes the inactive state, instead of a SB, although they still have the DRY motif in TM3.

For TAS2Rs, there are several candidates for this HB: Y1063.50 or K1093.53 in TM3 and Q2216.30 in TM6 (Fig. S16A). Both residues in TM3 are highly conserved, while the 6.30 position is also well conserved with a polar residue (usually serine) for TAS2Rs. However, our metaMD studies on the inactive structure find that none of these pairs are energetically favourable, suggesting that TAS2Rs may not have a TM3-6 interaction stabilizing the inactive state (Figs S16B and S16C). In fact, this possibility of a less stable inactive state or more stable active state of TAS2Rs has been suggested from the significantly different agonist-to-antagonist ratio of TAS2Rs compared to other GPCRs (Di Pizio et al., Reference Di Pizio, Levit, Slutzki, Behrens, Karaman and Niv2016). Our results support the premise that TAS2Rs have an intrinsically less stable inactive state. This may be because of their roles in protecting animals from harmful compounds, but further experimental studies are needed for validation.

Conclusion

We used advanced computational techniques to predict the 3D structure of the fully activated TAS2R4 human bitter taste receptor coupled to Ggust and to each of several agonists. We found that just as for κOR and μOR, the Ggust forms stable SBs with all three ICLs of TAS2R4 while the Gα5 helix makes extensive interactions with residues in the cytoplasmic region of the GPCR. In the binding pocket, D923.36 provides a large binding energy through polar interactions with the agonist, while triggering rotation of Y2396.48 by changing the HB network. We find that the unique motifs conserved in TAS2Rs play a role in the activation process but differently compared to typical Class A GPCRs. We also find a less stable inactive state, which may be an intrinsic property of TAS2Rs. This structural information about the fully activated TAS2R4 bitter taste receptor provides new insights that may be useful for understanding of variety of activation mechanisms for other TAS2Rs, providing an insight into the diversity across the GPCR family. These insights should provide guidance useful in developing new therapeutic targets.

Methods

Modelling the human TAS2R4-ligand-G protein complex

The GEnSeMBLE technique (Abrol et al., Reference Abrol, Bray and Goddard2012; Bray et al., Reference Bray, Abrol, Goddard, Trzaskowski and Scott2014) was used to predict the TMD of TAS2R4. The loops were constructed with MODELLER (Fiser et al., Reference Fiser, Do and Sali2000), using the mouse μOR structure (PDBID: 6DDF) as a template. DarwinDock (Griffith, Reference Griffith2017) was used to predict the binding site for Rubu and quinine to the TAS2R4 TMD. Initial binding poses of RebM and hydRebM were prepared by matching steviol cores to that of Rubu, followed by 10 cycles of simulated annealing from 50 K to 600 K to optimize the sugar positions. The nucleotide-free gustducin heterotrimeric protein (Guanine nucleotide-binding protein Gt subunit alpha-3: GNAT3, Guanine nucleotide-binding protein Gt subunit beta-1: GNB1 and Guanine nucleotide-binding protein Gt subunit gamma-13: GNG13) was constructed using our optimized GiP complexed with μOR (Mafi et al., Reference Mafi, Kim and Goddard2020a, Reference Mafi, Kim and Goddard2020b) as a template to perform homology modelling using Prime (Schrödinger) (Jacobson et al., Reference Jacobson, Pincus, Rapp, Day, Honig, Shaw and Friesner2004). The GiP was chosen because there is a reasonably high similarity (> ~ 63%) between these two heterotrimeric GP structures. Then, we included the myristoyl-Gly2 in the N-terminus of the GαN helix. Similarly, the inactive Ggust bound GDP structure was constructed using the crystal structure of heterotrimeric GiP bound with GDP (PDBID: 1GOT) as a template for the homology modelling. The initial structure of the TAS2R4-Rubu-Ggust complex was modelled by superimposing the predicted human TAS2R4-Rubu with the cryo-EM structure of mouse μOR-Gi (PDBID: 6DDF).

System preparation and equilibration

The constructed protein was immersed into 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) lipid bilayer, using a pre-equilibrated bilayer structure with 277 molecules (Dong et al., Reference Dong, Goddard and Abrol2016). The protein and lipid membrane were placed in a ~ 100 × 100 × 140 Å3 box, with water molecules and 100 mM concentration of ions (sodium and chloride). The final GPCR-Ggust system contained ∼150,000 atoms. For the inactive structure the system contained 160 POPC with ~70,000 atoms total in a ~ 80 × 80 × 110 Å3 box. We used the Amber14 force field for proteins (Dickson et al., Reference Dickson, Madej, Skjevik, Betz, Teigen, Gould and Walker2014), Ambertools 16 for the lipid (Case et al., Reference Case, Betz, Cerutti, Cheatham, Darden, Duke, Giese, Gohlke, Goetz, Homeyer, Izadi, Janowski, Kaus, Kovalenko, Lee, LeGrand, Li, Lin, Luchko, Luo, Madej, Mermelstein, Merz, Monard, Nguyen, Nguyen, Omelyan, Onufriev, Roe, Roitberg, Sagui, Simmerling, Botello-Smith, Swails, Walker, Wang, Wolf, Wu, Xiao and Kollman2016) and the generalized Amber force field for the ligands (Wang et al., Reference Wang, Wolf, Caldwell, Kollman and Case2004). Water molecules were described using the TIP3P (Jorgensen et al., Reference Jorgensen, Chandrasekhar, Madura, Impey and Klein1983) model. Steepest-descent energy minimization was performed first for the constructed system to relax the whole structure. Then, we equilibrated with 5 ns MD simulation, while the positions of all heavy atoms of proteins and ligands were restrained with a force constant of ~2.4 kcal (mol Å2)−1 for the first 1 ns, which was then gradually reduced every 1 ns to 0.12 kcal (mol Å2)−1. We continued the equilibration with over 200 ns of NPT simulation (and over 100 ns for the inactive structure) without positional restraints to relax the complex structure. During this equilibration step, parts of the helices of TAS2R4, GαN and Gα5 of the Gα subunit, and residues involved in important interactions were often constrained with a force constant of ~1.2 kcal (mol Å2)−1 to retain their secondary structures and interactions as the system was relaxed. For free energy analyses, we applied the well-tempered metaMD (Barducci et al., Reference Barducci, Bussi and Parrinello2008) implemented in PLUMED (Tribello et al., Reference Tribello, Bonomi, Branduardi, Camilloni and Bussi2014). All simulations were carried out with a 2 fs time step at 310 K and 1 bar using the GROMACS (Abraham et al., Reference Abraham, Murtola, Schulz, Páll, Smith, Hess and Lindahl2015) MD software with PLUMED. After the equilibration we carried out 480 ns of MD on Anton2 (Shaw et al., Reference Shaw, Grossman, Bank, Batson, Butts, Chao, Deneroff, Dror, Even, Fenton, Forte, Gagliardo, Gill, Greskamp, Ho, Ierardi, Iserovich, Kuskin, Larson, Layman, Lee, Lerer, Li, Killebrew, Mackenzie, Mok, Moraes, Mueller, Nociolo, Peticolas, Quan, Ramot, Salmon, Scarpazza, Schafer, Siddique, Snyder, Spengler, Tang, Theobald, Toma, Towles, Vitale, Wang and Young2014). VMD (Humphrey et al., Reference Humphrey, Dalke and Schulten1996) and Chimera (Pettersen et al., Reference Pettersen, Goddard, Huang, Couch, Greenblatt, Meng and Ferrin2004) programs were partially used for analysis and visualization.

Human bitterness scores

Pure (~98%) Stevia bitter glycosides were obtained from suppliers (Chromadex Standards, Inc. Los Angeles, CA) or isolated from Stevia leaf according to the procedures of Hellfritsch et al., (Reference Hellfritsch, Brockhoff, Stahler, Meyerhof and Hofmann2012). Samples were made by dissolving standards in pure sensory-grade water at least 24 h before the test to various levels. Pyschophysical measurements of bitterness intensity were performed as described by Hellfritsch et al. (Reference Hellfritsch, Brockhoff, Stahler, Meyerhof and Hofmann2012). The 300 ppm level was considered typical for consumption in products such as beverages.

Acknowledgements

Funding for this project was provided by Cargill Global Food Research. WAG thanks Dr. Fan Liu for early discussions about TAS2R4 and its activation. The computational resources for this research were provided by the Anton2 computer at the Pittsburgh National Supercomputing Center (MCB180091P) and KISTI National Supercomputing Center (KSC-2018-CHA-0049).

Open Peer Review

To view the open peer review materials for this article, please visit http://doi.org/10.1017/qrd.2021.1.

Supplementary Materials

To view supplementary material for this article, please visit http://dx.doi.org/10.1017/qrd.2021.1.

References

Abraham, MJ, Murtola, T, Schulz, R, Páll, S, Smith, JC, Hess, B and Lindahl, E (2015) GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1, 19–25.CrossRefGoogle Scholar
Abrol, R, Bray, JK and Goddard, WA III (2012) Bihelix: towards de novo structure prediction of an ensemble of G-protein coupled receptor conformations. Proteins 80(2), 505–518.CrossRefGoogle ScholarPubMed
Abrol, R, Trzaskowski, B, Goddard, WA III, Nesterov, A, Olave, I and Irons, C (2014) Ligand- and mutation-induced conformational selection in the CCR5 chemokine G protein-coupled receptor. Proceedings of the National Academy of Sciences of the United States of America 111(36):13040–13045.CrossRefGoogle ScholarPubMed
Ahn, KH, Scott, CE, Abrol, R, Goddard, WA III and Kendall, DA (2013) Computationally-predicted CB1 cannabinoid receptor mutants show distinct patterns of salt-bridges that correlate with their level of constitutive activity reflected in G protein coupling levels, thermal stability, and ligand binding. Proteins 81(8), 1304–1317.CrossRefGoogle Scholar
Andreozzi, P, Sarnelli, G, Pesce, M, Zito, FP, Alessandro, AD, Verlezza, V, Palumbo, I, Turco, F, Esposito, K and Cuomo, R (2015) The bitter taste receptor agonist quinine reduces calorie intake and increases the postprandial release of cholecystokinin in healthy subjects. Journal of Neurogastroenterology and Motility 21(4), 511–519.CrossRefGoogle ScholarPubMed
Ballesteros, JA and Weinstein, H (1995) Integrated methods for the construction of three-dimensional models and computational probing of structure-function relations in G protein-coupled receptors. Methods in Neuroscience 25, 366–428.CrossRefGoogle Scholar
Barducci, A, Bussi, G and Parrinello, M (2008) Well-tempered metadynamics: a smoothly converging and tunable free-energy method. Physical Review Letters 100, 020603.CrossRefGoogle ScholarPubMed
Berro, R, Yasmeen, A, Abrol, R, Trzaskowski, B, Abi-Habib, S, Grunbeck, A, Lascano, D, Goddard, WA III, Klasse, PJ, Sakmar, TP and Moore, JP (2013) Use of G-protein-coupled and -uncoupled CCR5 receptors by CCR5 inhibitor-resistant and -sensitive human immunodeficiency virus type 1 variants. Journal of Virology 87(12), 6569–6581.CrossRefGoogle ScholarPubMed
Bray, JK, Abrol, R, Goddard, WA III, Trzaskowski, B and Scott, CE (2014) SuperBiHelix method for predicting the pleiotropic ensemble of G-protein–coupled receptor conformations. Proceedings of the National Academy of Sciences of the United States of America 111(1):E72–E78.CrossRefGoogle ScholarPubMed
Case, DA, Betz, RM, Cerutti, DS, Cheatham, TE III, Darden, TA, Duke, RE, Giese, TJ, Gohlke, H, Goetz, AW, Homeyer, N, Izadi, S, Janowski, P, Kaus, J, Kovalenko, A, Lee, TS, LeGrand, S, Li, P, Lin, C, Luchko, T, Luo, R, Madej, B, Mermelstein, D, Merz, KM, Monard, G, Nguyen, H, Nguyen, HT, Omelyan, I, Onufriev, A, Roe, DR, Roitberg, A, Sagui, C, Simmerling, CL, Botello-Smith, WM, Swails, J, Walker, RC, Wang, J, Wolf, RM, Wu, X, Xiao, L and Kollman, PA (2016) AMBER16. San Francisco: University of California.Google Scholar
Deshpande, DA, Wang, WC, McIlmoyle, EL, Robinett, KS, Schillinger, RM, An, SS, Sham, JSK and Liggett, SB (2010) Bitter taste receptors on airway smooth muscle bronchodilate by localized calcium signaling and reverse obstruction. Nature Medicine 16(11), 1299–1304.CrossRefGoogle ScholarPubMed
Di Pizio, A and Niv, MY (2015) Promiscuity and selectivity of bitter molecules and their receptors. Bioorganic and Medicinal Chemistry 23(14), 4082e4091.CrossRefGoogle ScholarPubMed
Di Pizio, A, Levit, A, Slutzki, M, Behrens, M, Karaman, R and Niv, MY (2016) Comparing Class A GPCRs to bitter taste receptors: structural motifs, ligand interactions and agonist-to-antagonist ratios. Methods in Cell Biology 132, 401–427.CrossRefGoogle Scholar
Dickson, CJ, Madej, BD, Skjevik, AA, Betz, RM, Teigen, K, Gould, IR and Walker, RC (2014) Lipid14: the amber lipid force field. Journal of Chemical Theory and Computation 10(2), 865–879.CrossRefGoogle ScholarPubMed
Dong, SS, Goddard, WA III and Abrol, R (2016) Conformational and thermodynamic landscape of GPCR activation from theory and computation Biophysical Journal 110(12), 2618–2629.CrossRefGoogle ScholarPubMed
Dotson, CD, Zhang, L, Xu, H, Shin, YK, Vigues, S, Ott, SH, Elson, AE, Choi, HJ, Shaw, H, Egan, JM, Mitchell, BD, Li, X, Steinle, NI and Munger, SD (2008) Bitter taste receptors influence glucose homeostasis. PLoS One 3(12), e3974.CrossRefGoogle ScholarPubMed
Du, Y, Duc, NM, Rasmussen, SGF, Hilger, D, Kubiak, X, Wang, L, Bohon, J, Kim, HR, Wegrecki, M, Asuru, A, Jeong, KM, Lee, J, Chance, MR, Lodowski, DT, Kobilka, BK and Chung, KY (2019) Assembly of a GPCR-G protein complex. Cell 177(5), 1–11.CrossRefGoogle ScholarPubMed
Finger, TE, Bottger, B, Hansen, A, Anderson, KT, Alimohammadi, H and Silver, WL (2003) Solitary chemoreceptor cells in the nasal cavity serve as sentinels of respiration. Proceedings of the National Academy of Sciences of the United States of America 100(15), 8981–8986.CrossRefGoogle Scholar
Fiser, A, Do, RK and Sali, A (2000) Modeling of loops in protein structures. Protein Science 9(9):1753–1773.CrossRefGoogle ScholarPubMed
Foster, SR, Blank, K, See Hoe, LE, Behrens, M, Meyerhof, W, Peart, JN, Thomas, WG (2014) Bitter taste receptor agonists elicit G-protein-dependent negative inotropy in the murine heart. The FEBS Journal 28(10):4497–4508.Google ScholarPubMed
Goddard, WA III, Kim, SK, Li, Y, Trzaskowski, B, Griffith, AR and Abrol, R (2010) Predicted 3D structures for adenosine receptors bound to ligands: comparison to the crystal structure. Journal of Structural Biology 170(1), 10–20.CrossRefGoogle ScholarPubMed
Griffith, AR (2017) DarwinDock & GAG-Dock: methods and applications for small molecule docking, PhD thesis, California Institute of Technology.Google Scholar
Hellfritsch, C, Brockhoff, A, Stahler, F, Meyerhof, W and Hofmann, T (2012) Human psychometric and taste receptor responses to steviol glycosides. Journal of Agricultural and Food Chemistry 60(27), 6782–6793.CrossRefGoogle ScholarPubMed
Humphrey, W, Dalke, A and Schulten, K (1996) VMD – visual molecular dynamics. Journal of Molecular Graphics 14, 33–38.CrossRefGoogle ScholarPubMed
Jacobson, MP, Pincus, DL, Rapp, CS, Day, TJ, Honig, B, Shaw, DE and Friesner, RA (2004) A hierarchical approach to all-atom protein loop prediction. Proteins 55(2):351–367.CrossRefGoogle ScholarPubMed
Jorgensen, WL, Chandrasekhar, J, Madura, JD, Impey, RW and Klein, ML (1983) Comparison of simple potential functions for simulating liquid water. The Journal of Chemical Physics 79, 926–935.CrossRefGoogle Scholar
Kim, SK, Chem, Y, Abrol, R, Goddard, WA III and Guthrie, B (2017) Activation mechanism of the G protein-coupled sweet receptor heterodimer with sweeteners and allosteric agonists. Proceedings of the National Academy of Sciences of the United States of America 114(10), 2568–2573.CrossRefGoogle Scholar
Kirkpatrick, A, Heo, J, Abrol, R and Goddard, WA III (2012) Predicted structure of agonist-bound glucagon-like peptide 1 receptor, a class B G protein-coupled receptor. Proceedings of the National Academy of Sciences of the United States of America 109(49), 19988–19993.CrossRefGoogle ScholarPubMed
Li, Q, Kim, SK, Goddard, WA III, Chen, G and Tan, H (2015) Predicted structures for kappa opioid G-protein coupled receptor bound to selective agonists. Journal of Chemical Information and Modeling 55(3), 614–627.CrossRefGoogle ScholarPubMed
Liu, W, Chun, E, Thompson, AA, Chubukov, P, Xu, F, Katritch, V, Han, GW, Roth, CB, Heitman, LH, IJzerman, AP, Cherezov, V and Stevens, RC, Structural basis for allosteric regulation of GPCRs by sodium ions (2012) Science 337(6091):232e236.CrossRefGoogle ScholarPubMed
Lu, P, Zhang, CH, Lifshitz, LM and ZhuGe, R (2017) Extraoral bitter taste receptors in health and disease. The Journal of General Physiology 149(2), 181–197.CrossRefGoogle ScholarPubMed
Mafi, A, Kim, SK and Goddard, WA III (2020a) The atomistic level structure for the activated human κ-opioid receptor bound to the full Gi protein and the MP1104 agonist. Proceedings of the National Academy of Sciences of the United States of America 117(11):5836–5843.CrossRefGoogle Scholar
Mafi, A, Kim, SK and Goddard, WA III (2020b) Mechanism of β-arrestin recruitment by the μ-opioid G protein-coupled receptor. Proceedings of the National Academy of Sciences of the United States of America 117(28), 16346–16355.CrossRefGoogle Scholar
Martin, LTP, Nachtigal, MW, Selman, T, Nguyen, E, Sasman, J, Dellaire, G and Dupré, DJ (2019) Bitter taste receptors are expressed in human epithelial ovarian and prostate cancers cells and noscapine stimulation impacts cell survival. Molecular and Cellular Biochemistry 454(1–2):203–214.CrossRefGoogle ScholarPubMed
Oldham, WM and Hamm, HE (2008) Heterotrimeric G protein activation by G-protein-coupled receptors. Nature Reviews Molecular Cell Biology 9(1), 60–71.CrossRefGoogle ScholarPubMed
Oldham, WM, Van Eps, N, Preininger, AM, Hubbell, WL and Hamm, HE (2006) Mechanism of the receptor-catalyzed activation of heterotrimeric G proteins. Nature Structural & Molecular Biology 13, 772–777.CrossRefGoogle ScholarPubMed
Onrust, R, Herzmark, P, Chi, P, Garcia, PD, Lichtarge, O, Kingsley, C and Bourne, HR (1997) Receptor and betagamma binding sites in the α subunit of the retinal G protein transducin. Science 275(5298), 381–384.CrossRefGoogle ScholarPubMed
Pándy-Szekeres, G, Munk, C, Tsonkov, TM, Mordalski, S, Harpsøe, K, Hauser, AS, Bojarski, AJ and Gloriam, DE (2018) GPCRdb in 2018: adding GPCR structure models and ligands, Nucleic Acids Research 46(D1):D440–D446.CrossRefGoogle ScholarPubMed
Pettersen, EF, Goddard, TD, Huang, CC, Couch, GS, Greenblatt, DM, Meng, EC and Ferrin, TE (2004) UCSF Chimera – a visualization system for exploratory research and analysis. Journal of Computational Chemistry 25(13), 1605–1612.CrossRefGoogle ScholarPubMed
Pydi, SP, Bhullar, RP and Chelikani, P (2012) Constitutively active mutant gives novel insights into the mechanism of bitter taste receptor activation. Journal of Neurochemistry 122(3), 537–544.CrossRefGoogle ScholarPubMed
Pydi, SP, Jaggupilli, A, Nelson, KM, Abrams, SR, Bhullar, RP, Loewen, MC and Chelikani, P (2015) Abscisic acid acts as a blocker of the bitter taste G protein-coupled receptor T2R4. Biochemistry 54(16), 2622–2631.CrossRefGoogle ScholarPubMed
Scott, CE, Abrol, R, Ahn, KH, Kendall, DA and Goddard, WA III (2013) Molecular basis for dramatic changes in cannabinoid CB1 G protein-coupled receptor activation upon single and double point mutations. Protein Science 22(1), 101–113.CrossRefGoogle ScholarPubMed
Shankar, V, Goddard, WA III, Kim, SK, Abrol, R and Liu, F (2018) The 3D structure of human DP prostaglandin G-protein-coupled receptor bound to cyclopentanoindole antagonist, predicted using the DuplexBiHelix modification of the GEnSeMBLE method. Journal of Chemical Theory and Computation 14(3), 1624–1642.CrossRefGoogle ScholarPubMed
Shaw, DE, Grossman, JP, Bank, JA, Batson, B, Butts, JA, Chao, JC, Deneroff, MM, Dror, RO, Even, A, Fenton, CH, Forte, A, Gagliardo, J, Gill, G, Greskamp, B, Ho, CR, Ierardi, DJ, Iserovich, L, Kuskin, JS, Larson, RH, Layman, T, Lee, L, Lerer, AK, Li, C, Killebrew, D, Mackenzie, KM, Mok, SY, Moraes, MA, Mueller, R, Nociolo, LJ, Peticolas, JL, Quan, T, Ramot, D, Salmon, JK, Scarpazza, DP, Schafer, UB, Siddique, N, Snyder, CW, Spengler, J, Tang, PTP, Theobald, M, Toma, H, Towles, B, Vitale, B, Wang, SC and Young, C (2014) In Anton 2: Raising the Bar for Performance and Programmability in a Special-Purpose Molecular Dynamics Supercomputer. SC ‘14. Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, pp. 41–53.CrossRefGoogle Scholar
Shaw, L, Mansfield, C, Colquitt, L, Lin, C, Ferreira, J, Emmetsberger, J and Reed, DR (2018) Personalized expression of bitter taste receptors in human skin. PLoS ONE 13(10), e0205322.CrossRefGoogle ScholarPubMed
Shiffman, D, O’Meara, ES, Bare, LA, Rowland, CM, Louie, JZ, Arellano, AR, Lumley, T, Rice, K, Iakoubova, O, Luke, MM, Young, BA, Malloy, MJ, Kane, JP, Ellis, SG, Tracy, RP, Devlin, JJ and Psaty, BM (2008) Association of gene variants with incident myocardial infarction in the cardiovascular health study. Arteriosclerosis, Thrombosis, and Vascular Biology 28(1):173–179.CrossRefGoogle ScholarPubMed
Singh, N, Vrontakis, M, Parkinson, F and Chelikani, P (2011a) Functional bitter taste receptors are expressed in brain cells. Biochemical and Biophysical Research Communications 406(1), 146–151.CrossRefGoogle Scholar
Singh, N, Pydi, SP, Upadhyaya, J and Chelikani, P (2011b) Structural basis of activation of bitter taste receptor T2R1 and comparison with class A G-protein-coupled receptors (GPCRs). Journal of Biological Chemistry 286(41), 36032–36041.CrossRefGoogle Scholar
Tan, J, Abriol, R, Trzaskowski, B and Goddard, WA III (2012) 3D Structure prediction of TAS2R38 bitter receptors bound to agonists phenylthiocarbamide (PTC) and 6-n-propylthiouracil (PROP). Journal of Chemical Information and Modeling 52(7), 1875–1885.CrossRefGoogle Scholar
Thompson, AJ, Lochner, M and Lummis, SCR (2007) The antimalarial drugs quinine, chloroquine and mefloquine are antagonists at 5-HT3 receptors. British Journal of Pharmacology 151(5), 666–677.CrossRefGoogle ScholarPubMed
Tribello, GA, Bonomi, M, Branduardi, D, Camilloni, C and Bussi, G (2014) PLUMED 2: new feathers for an old bird. Computer Physics Communications 185(2), 604–613.CrossRefGoogle Scholar
Wang, J, Wolf, RM, Caldwell, JW, Kollman, PA and Case, DA (2004) Development and testing of a general amber force field. Journal of Computational Chemistry 25(9), 1157–1174.CrossRefGoogle ScholarPubMed
Wiener, A, Shudler, M, Levit, A and Niv, MY (2012) BitterDB: a database of bitter compounds. Nucleic Acids Research 40, D413–D419.CrossRefGoogle ScholarPubMed
Wu, SV, Rozengurt, N, Yang, M, Young, SH, Sinnett-Smith, J and Rozengurt, E (2002) Expression of bitter taste receptors of the T2R family in the gastrointestinal tract and enteroendocrine STC-1 cells. Proceedings of the National Academy of Sciences of the United States of America 99(4), 2392–2397.CrossRefGoogle ScholarPubMed
Zhou, Q, Yang, D, Wu, M, Guo, Y, Guo, W, Zhong, L, Cai, X, Dai, A, Jang, W, Shakhnovich, EI, Liu, ZJ, Stevens, RC, Lambert, NA, Babu, MM, Wang, MW and Zhao, S (2019) Common activation mechanism of class A GPCRs. eLife 8:e50279.CrossRefGoogle ScholarPubMed
Figure 0

Fig. 1. Chemical structures of (a) the steviol glycosides and (b) quinine, where sugar (glucose or rhamnose) is attached at R1 and R2 in (a). Binding sites for (c) Rubu and (d) quinine to TAS2R4. Rubu has HBs to residues in TM3 (F883.32, M893.33, D923.36, S933.37), TM4 (Y1474.62), TM5 (L1775.39, L1815.43), TM6 (Y2506.59) and ELC2 (E158, T162, N164). Quinine has a SB to D923.36 and HBs to residues in TM3 (F883.32, M893.33), TM5 (L1815.43, Q1885.50) and TM6 (Y2426.51).

Figure 1

Fig. 2. (a) Top and (b) side views showing D923.36 – Y2396.48 interactions in the absence (cyan) and presence (tan) of the agonist, but the agonist is omitted for clarity.

Figure 2

Fig. 3. (a) The structure of TAS2R4-Rubu-Ggust complex. Important polar interactions at three anchors: (b) anchor 1: ICL1 with Gβ, (c) anchor 2: ICL2 and GαN and (d) anchor 3: ICL3 and Gα. (e) Polar interactions between the Gα5 helix of the Gα subunit with TAS2R4.

Figure 3

Fig. 4. TM1–2-7 interactions (a) in the active state and (b) in the inactive state, respectively.

Supplementary material: File

Yang et al. supplementary material

Yang et al. supplementary material

Download Yang et al. supplementary material(File)
File 52.3 MB

Review: Predicted Structure of Fully Activated Human Bitter Taste Receptor TAS2R4 Complexed with G Protein and Agonists — R0/PR1

Published online by Cambridge University Press:  08 April 2021
DOI: https://doi.org/10.1017/qrd.2021.1.pr1 [Opens in a new window]
Lennart Nilsson Stockholms Universitet, Sweden
Date of review:  12 December 2020
Revision round: 0
Role: reviewer
Recommendation/decision: minor-revision

Conflict of interest statement

Reviewer declares none.

Comments

Comments to Author: In this manuscript molecular dynamics simulations using modeled structures of GPCR complexes to describe in atomistic detail interactions that are suggested to be important for the activation of the protein, similarly to what the authors have previously done. In this case the human receptor for bitter taste, TAS2R4, that has been analyzed, Severaltyoes of interactions (mainly hydrogen bonds and salt bridges) that have been identified in other complexes are shown to exist also here. Furthermore the varying degree of bitterness of different compounds is rationalized based on changed interaction patterns.

The simulations are carefully done, although only single simulations have been performed for each system. It is not entirely clear that good equilibration has been obtained fro all systems in the 480 ns MD simulations. For systems of this size (70000-150000) atoms it is only a matter of 1-2 weeks to run a microsecond on a single inexpensive GPU.

Given that there are several variants of the AMBER force fields the authors should provide references for the "Amber 14" and "Ambertools 16" that were used.

Review: Predicted Structure of Fully Activated Human Bitter Taste Receptor TAS2R4 Complexed with G Protein and Agonists — R0/PR2

Published online by Cambridge University Press:  08 April 2021
DOI: https://doi.org/10.1017/qrd.2021.1.pr2 [Opens in a new window]
Libin Ye University of South Florida, United States
Date of review:  14 December 2020
Revision round: 0
Role: reviewer
Recommendation/decision: minor-revision

Conflict of interest statement

Reviewer declares none.

Comments

Comments to Author: The manuscript "Predicted Structure of Fully Activated Human Bitter Taste Receptor TAS2R4 Complexed with G Protein and Agonists" was submitted by Yang et al. The manuscript was well-written following their previous publication (Mafi et al, PNAS, 2020) with a new insight revealed for the TAS2R4 activation because of its low homology to family A GPCR. In particular, the manuscript proposed a possible differential activation mechanism for Taster receptor, which is interesting and should bring interests for readers to continue studying this type of receptors with this computational model. I would suggest publishing this manuscript with some minor questions.

(1) The authors mentioned that the sequence similarity of TAS2Rs with other class A GPCRS is less than 30% on Page 2, line 8. However, they built up the structures based on A family GPCRs on Page 3, line 12. I am wondering if the author made some justification when they selected these four templates or just as usual for building a model for a high-similarity GPCR. If so, how this justification could be to ensure the built structure maximally close to a natural structure of TAS2Rs, though we don’t have one yet.

(2) On page 10, line 4, the authors mentioned that TAS2Rs neither the DRY motif nor E(D) is conserved, and there is no alternative residue at nearby position to form a SB for the ionic lock. As we know that ionic lock between TM3 and TM6 is in favor of forming inactive states. Is this the reason that TAS2Rs have a less inactive conformation observed.

(3) It is interesting that authors mentioned that for TAS2Rs have several candidate residues forming HB such as Y1063.50 or K1093.53 in TM3 and Q2216.30 in TM6 in which residues in TM3 are more highly conserved in comparison to TM6 residues. Can author make an amino acid alignment for all taste receptors and include some other family A GPCRs to highlight these residues incorporated into Fig. S16 or an independent supplementary figure to give readers a visual view how conserved they are.

(4) It was reported that Gai1-3 have a higher affinity than Ggust, did author even try to use any of Gi to do the simulation and what is different from Ggust engagement in terms of fully activated conformation of the receptor.

Decision: Predicted Structure of Fully Activated Human Bitter Taste Receptor TAS2R4 Complexed with G Protein and Agonists — R0/PR3

Published online by Cambridge University Press:  08 April 2021
DOI: https://doi.org/10.1017/qrd.2021.1.pr3 [Opens in a new window]
Bengt J. F. Nordén [Opens in a new window] Chemistry and Chemical Engineering, Chalmers University of Technology, Sweden
Revision round: 0
Role: Chair of the Board of Editors
Recommendation/decision: minor-revision

Comments

Comments to Author: Reviewer #1: The manuscript "Predicted Structure of Fully Activated Human Bitter Taste Receptor TAS2R4 Complexed with G Protein and Agonists" was submitted by Yang et al. The manuscript was well-written following their previous publication (Mafi et al, PNAS, 2020) with a new insight revealed for the TAS2R4 activation because of its low homology to family A GPCR. In particular, the manuscript proposed a possible differential activation mechanism for Taster receptor, which is interesting and should bring interests for readers to continue studying this type of receptors with this computational model. I would suggest publishing this manuscript with some minor questions.

(1) The authors mentioned that the sequence similarity of TAS2Rs with other class A GPCRS is less than 30% on Page 2, line 8. However, they built up the structures based on A family GPCRs on Page 3, line 12. I am wondering if the author made some justification when they selected these four templates or just as usual for building a model for a high-similarity GPCR. If so, how this justification could be to ensure the built structure maximally close to a natural structure of TAS2Rs, though we don’t have one yet.

(2) On page 10, line 4, the authors mentioned that TAS2Rs neither the DRY motif nor E(D) is conserved, and there is no alternative residue at nearby position to form a SB for the ionic lock. As we know that ionic lock between TM3 and TM6 is in favor of forming inactive states. Is this the reason that TAS2Rs have a less inactive conformation observed.

(3) It is interesting that authors mentioned that for TAS2Rs have several candidate residues forming HB such as Y1063.50 or K1093.53 in TM3 and Q2216.30 in TM6 in which residues in TM3 are more highly conserved in comparison to TM6 residues. Can author make an amino acid alignment for all taste receptors and include some other family A GPCRs to highlight these residues incorporated into Fig. S16 or an independent supplementary figure to give readers a visual view how conserved they are.

(4) It was reported that Gai1-3 have a higher affinity than Ggust, did author even try to use any of Gi to do the simulation and what is different from Ggust engagement in terms of fully activated conformation of the receptor.

Reviewer #3: In this manuscript molecular dynamics simulations using modeled structures of GPCR complexes to describe in atomistic detail interactions that are suggested to be important for the activation of the protein, similarly to what the authors have previously done. In this case the human receptor for bitter taste, TAS2R4, that has been analyzed, Severaltyoes of interactions (mainly hydrogen bonds and salt bridges) that have been identified in other complexes are shown to exist also here. Furthermore the varying degree of bitterness of different compounds is rationalized based on changed interaction patterns.

The simulations are carefully done, although only single simulations have been performed for each system. It is not entirely clear that good equilibration has been obtained fro all systems in the 480 ns MD simulations. For systems of this size (70000-150000) atoms it is only a matter of 1-2 weeks to run a microsecond on a single inexpensive GPU.

Given that there are several variants of the AMBER force fields the authors should provide references for the "Amber 14" and "Ambertools 16" that were used.

Decision: Predicted Structure of Fully Activated Human Bitter Taste Receptor TAS2R4 Complexed with G Protein and Agonists — R1/PR4

Published online by Cambridge University Press:  08 April 2021
DOI: https://doi.org/10.1017/qrd.2021.1.pr4 [Opens in a new window]
Bengt J. F. Nordén [Opens in a new window] Chemistry and Chemical Engineering, Chalmers University of Technology, Sweden
Revision round: 1
Role: Chair of the Board of Editors
Recommendation/decision: minor-revision

Comments

No accompanying comment.

Decision: Predicted Structure of Fully Activated Human Bitter Taste Receptor TAS2R4 Complexed with G Protein and Agonists — R2/PR5

Published online by Cambridge University Press:  08 April 2021
DOI: https://doi.org/10.1017/qrd.2021.1.pr5 [Opens in a new window]
Bengt J. F. Nordén [Opens in a new window] Chemistry and Chemical Engineering, Chalmers University of Technology, Sweden
Revision round: 2
Role: Chair of the Board of Editors
Recommendation/decision: accept

Comments

No accompanying comment.