Computational Methods for the Discovery and Optimization of TAAR1 and TAAR5 Ligands

Abstract

G-protein-coupled receptors (GPCRs) represent a family of druggable targets when treating several diseases and continue to be a leading part of the drug discovery process. Trace amine-associated receptors (TAARs) are GPCRs involved in many physiological functions with TAAR1 having important roles within the central nervous system (CNS). By using homology modeling methods, the responsiveness of TAAR1 to endogenous and synthetic ligands has been explored. In addition, the discovery of different chemo-types as selective murine and/or human TAAR1 ligands has helped in the understanding of the species-specificity preferences. The availability of TAAR1–ligand complexes sheds light on how different ligands bind TAAR1. TAAR5 is considered an olfactory receptor but has specific involvement in some brain functions. In this case, the drug discovery effort has been limited. Here, we review the successful computational efforts developed in the search for novel TAAR1 and TAAR5 ligands. A specific focus on applying structure-based and/or ligand-based methods has been done. We also give a perspective of the experimental data available to guide the future drug design of new ligands, probing species-specificity preferences towards more selective ligands. Hints for applying repositioning approaches are also discussed.

1. Introduction

Trace amines (TA), a group of endogenous compounds found at low levels in both peripheral and brain tissues of vertebrates, notably mammals, encompass classic examples like β-phenylethylamine (β-PEA), p-tyramine, tryptamine, and octopamine [1]. Initially considered inert byproducts of endogenous monoamines, such as dopamine and serotonin, their significance was reevaluated with the discovery of the trace amine-associated receptor (TAAR) family [2,3]. TAARs comprise nine subfamilies encoded by distinct genes and pseudogenes across species, including six genes (TAAR1, TAAR2, TAAR5, TAAR6, TAAR8, and TAAR9) and three pseudogenes (TAAR3, TAAR4, and TAAR7) in humans, and fifteen functional genes in mice [1,4]. Except for TAAR1, other TAARs are predominantly expressed in the olfactory epithelium, forming a unique class of olfactory receptors sensitive to volatile amines linked to innate behaviors [5,6]. However, recent evidence has demonstrated the expression of different TAARs outside the olfactory systems, including specific brain regions [7,8,9].

Among TAARs, TAAR1 has received the most attention, responding not only to trace amines but also to amphetamines and other psychotropic compounds [2,3]. It is expressed at low levels in the brain and periphery. In the central nervous system, TAAR1 is present in regions that are important for the regulation of monoamine systems, such as the ventral tegmental area, the substantia nigra, the dorsal raphe, the prefrontal cortex, and the amygdala [3,10,11]. It regulates the dopamine system, impacting D2 dopamine receptor activity and dopaminergic neuron firing [10,12,13,14,15]. TAAR1 knockout (TAAR1-KO) mice display heightened behavioral and neurochemical responses to dopaminergic compounds, presenting TAAR1 as a promising pharmacotherapeutic target for psychiatric disorders [16]. Recent clinical trials indicate the potential use of TAAR1 agonists for schizophrenia treatment, offering a novel mechanism independent of D2 dopamine receptor blockade [17].

TAAR5, another receptor of the TAAR family, shares a similar brain expression profile with TAAR1. Found in limbic regions such as the amygdala, entorhinal cortex, and nucleus accumbens, TAAR5 modulates emotional behavior and serotonin system function [7]. TAAR5 knockout (TAAR5-KO) mice exhibit anxiolytic and antidepressant-like behaviors, along with alterations in serotonin levels and enhanced 5-HT_1A serotonin receptor function (5-HT_1AR). Furthermore, TAAR5 influences dopamine levels and adult neurogenesis and is involved in sensorimotor functions and cognitive processes like attention and motivation [18,19,20,21]. This evidence positions TAAR5 as a promising drug target for mood disorders and cognitive impairment.

The development of compounds targeting TAAR1 has been extensive in the last 15 years, with Hoffman-La Roche acting as a pioneer in characterizing the first potent and selective TAAR1 full, partial agonists, and antagonists. Another company, Sunovion Pharmaceuticals (now Sumitomo Pharma), developed a TAAR1/5-HT_1AR agonist, SEP-363856 (Ulotaront) [22], which evidenced promising results in a phase II clinical trial for schizophrenia [23,24,25,26]. These data boosted the research on discovering new TAAR1 ligands and the effort to understand the mechanism of how endogenous and synthetic TAAR1 agonists bind to the receptor. The pharmacology of TAAR5, less studied than TAAR1, is still in its infancy and only a few ligands have been described. This review describes the work conducted so far into the computational methods used to discover ligands for these two members of the TAAR family, by giving an update of the comprehension of the mechanisms of ligand–receptor interactions. A perspective of the experimental data available and of the viability of repositioning strategies has also been detailed.

2. Molecular Modeling Studies in the Discovery and Optimization of TAAR1/5 Ligands

2.1. Computational Methods Exploring mTAAR1 Ligands

In search for novel agonists active on the murine orthologue of TAAR1, Chiellini et al. rationally designed a small series of thyronamine analogs, which were tested in vitro [27]. The final aim of their study was to understand the molecular basis of TAAR1 activation by designing thyronamine derivatives (1) as synthetic analogues of the endogenous TAAR1 agonist T1AM (Figure 1).

In detail, the authors replaced the oxygen atom tethered to the two aromatic rings of the endogenous ligand, with an isosteric methylene linkage. The OH group was maintained or replaced with the NH₂ group, which retains the same H-bonding donor and acceptor capabilities of the OH group. Finally, the amine-ethyl portion was considered or changed in a terminal amine-ethoxy function (see Figure 1).

Following biological assays, the compounds 1a–d were highlighted, the most potent of them, 1c, exhibiting a comparable EC₅₀ to T1AM (EC₅₀ = 240 nM; T1AM EC₅₀ = 189 nM) (

Table 1. Drug discovery studies focused on mTAAR1 agonists involving modelling techniques. The corresponding references (Ref.) are shown; hTAAR1 agonism ability was not determined.

Entry	Year	Method of Discovery	Use of the Derived Computational-Based Structural Information	Proposed Hit(s)	mTAAR1 EC50	Ref.
1	2015	Rational design (synthesis)	SAR rationalization	1c	240 nM	[27]
2	2016	Rational design (synthesis) combined with previously reported docking analysis	Hit-to-lead optimization	2b	98 nM	[28]

, entry 1).

To investigate the binding mode of the newly synthetized compounds at the receptor binding site, a docking procedure was performed in a homology model (HM) of the mTAAR1, built according to a ligand-based homology modeling procedure [29]. In particular, the HM was developed using the X-Ray of β₂-adrenoreceptor (β₂-ADR) in complex with an irreversible agonist as a template (PDB ID = 3PDS) [30], and the reference compound included in the ligand-based HM calculations was T1AM. The construction of the receptor HM employed the MOE software (MOE2013) [31]. The procedure included the alignment of the target sequence to the template by means of the BLOSUM62 matrix, followed by a loop search to rebuild the missing portions. The most promising model was selected according to the best packing quality function. After a minimization step, the quality of the obtained model was evaluated through comparison with the Ramachandran plot. The docking of the candidates was performed via the Surflex docking module implemented in Sybyl-X1.0 [32], and the best-scored poses were selected for the ligand/receptor energy minimization. Such poses were further submitted to ten runs of docking with the MOE-Dock genetic algorithm. The poses with best scores and lowest RMSD with respect to the output of the minimization were selected as the most stable poses. The position of the binding site was derived through a comparison with the template complex.

The docking analysis provided important clues as to the interaction mode of the tested compounds. In detail, the most potent compound, 1c, exhibited two H-bonds to D102 and Y291, in addition to two cation–π interactions (the first between the ligand protonated amine and residue Y287, the second between residue R86 and the ligand aniline ring). Moreover, a π–π stacking interaction was observed between the ligand aminoethyloxy-phenyl moiety and residue Y287. The study also pointed out some structural modifications tolerated on the thyronamine scaffold: the replacement of the phenol hydroxyl with an amino group, the increase in the distance between the charged amine and the aromatic ring by inserting an oxygen bridge, and the replacement of the 3-iodo substituent with an alkyl group.

On this basis, the lead compound (1c) was further optimized in a following study [28] involving the docking-based drug design, synthesis, and in vitro evaluation of fourteen analogs (2, 3). The introduced modifications were intended to restore a H-bond with R82 present for T1AM and not for compound 1c, with the introduction in 2 of the ethylamine chain in place of oxo-ethylamino featured by the previous hit 1c (Figure 1). In addition, the introduction of small alkyl substituents (Me, i-Pr) on both the outer and inner rings, or the removal of the methylene bridge, have been taken into account. This kind of approach has been managed both in the 2 series and in compounds 3, as highly related 1c analogues (Figure 1).

In vitro tests of compounds 2, 3 were performed, and most of them exhibited ameliorated activity at mTAAR1 (up to EC₅₀ = 98 nM, for compound 2b) (Table 1, entry 2). In particular, the replacement of the oxy-ethylamino sidechain with the ethylamino one proved to be advantageous, as well as the concomitant replacement of the amino group of the outer ring with the hydroxyl moiety, as shown by 2b (Figure 1). In addition, the presence of small alkyl substituents onto the phenyl ring tethered to the terminal chain was effective, making most of the derivatives of 2, 3 more potent than compounds 1 previously.

The docking of the newly synthesized analogues allowed for the rationalization of the obtained results. Surprisingly, the orientation of the two most potent compounds (2a and 2b) was reversed with respect to the first series of thyronamine analogs; however, they did exhibit a favorable network of interaction. In particular, compounds 2a and 2b formed a H-bond with D102 with the aniline moiety, while their protonated amine interacted via H-bonds with T83 and D284. However, both of them were selected for further in vivo investigation to ascertain their ability to modulate plasma glucose level. The docking procedure employed the previously obtained HM of mTAAR1 and was performed with the Surflex docking module implemented in Sybyl-X1.0. The top-scored poses were submitted to ligand/protein energy minimization by means of the MOE software.

2.2. Computational Methods Exploring hTAAR1 Ligands

The first study devoted to the search for hTAAR1 ligands involved a computer-aided drug discovery campaign applying an in silico virtual screening (VS) strategy [33]. In detail, a few hundred compounds previously reported as 5-HT_1AR and/or α₁-adrenoreceptor (α₁-ADR) agonists were evaluated via molecular docking calculations [34,35,36,37,38,39]. The exploited hTAAR1 structural model was built via homology modeling taking as its template the X-Ray of the human β₂-ADR in complex with a known agonist (PDB code: 3PDS) [30]. This calculation was achieved by applying MOE software [31]. Following this, docking studies were performed using the Surflex docking tool in the SybylX1.0 software [32]. The binding site in the hTAAR1 receptor was defined considering a range of 9Å around the key residue D103. The putative docking mode of RO5166017, β-PEA, T1AM (taken as reference TAAR1 agonists), and EPPTB (taken as reference TAAR1 antagonist) [40] was explored and compared with those of the aforementioned GPCR ligands.

The scouted library comprehended thirty different scaffolds combined with various substitutions and was screened against a previously published HM of the target [40]. The compounds included a series of aryloxyalchylamines and N1-arylpiperazines featuring 1,3-dioxolane-, 1,3-oxathiolane-, 1,3-dithiolane-, spiro-dioxolane-, 1,4-dioxane-, tetrahydrofuran-, cyclopentanone-, and cyclopentanol-based substituents. Among them, compounds 4–6 (Figure 2) have been deeply investigated in silico and then evaluated via biological assays.

The corresponding docking analysis revealed the presence of a common interaction pattern, constituted by a H-bond to a D103 sidechain, and π–π stacking with residues W264, F267, and F268 (see Figure 2). This information, together with a similar analysis carried out on T1AM and EPPTB as reference hTAAR1 agonist and antagonist, guided the compound selection for in vitro tests. In the initial screening phase, seven compounds displayed some activity as a TAAR1 agonist with a maximum effect (Emax), compared with the standard TAAR1 agonist β-PEA (EC₅₀ = 138 nM), spanning from 40 to 83%. For the most promising compounds, dose response has been calculated revealing 4a as the more potent in mediating cAMP production by TAAR1. This piece of information allowed for a preliminary exploration of the structure–activity relationship (SAR) within this series of compounds.

The dioxolane-based compounds 4, bearing a flexible amino group tethered to the terminal phenoxy, was more effective than those featuring the piperazine substituent. On the contrary, the pentanone-(5) and the pentanol-(6) based compounds were mildly active or inactive, respectively.

While compounds 4a,b and 4e,f were characterized by hTAAR1 agonist activity, compound 4c proved to be antagonist. As a result, six molecules bearing a dioxolane/cyclopentanone scaffold displayed bioactivity towards the target: in particular, five agonists (4a,b, 4e,f, and 5a, hTAAR1 EC₅₀ = 2.4–15.7 μM) and one antagonist compound (4c, hTAAR1 EC₅₀ = EC₅₀ = 9 μM) were individuated. The most interesting agonist, 4a, and antagonist, 4c, proposed have been reported in

Table 2. Drug discovery studies focused on hTAAR1 involving modeling techniques. The related references are reported (Ref.).

Entry	Year	Method of Discovery	Use of Structural Information	HIT Compound Example	hTAAR1 EC50 (IC50)	Ref.
1	2014	VS on hTAAR1 HM	Prospective drug discovery, SAR rationalization	4a	2.4 μM	[33]
				4c	9 μM	[33]
2	2015	VS on hTAAR1 HM	Prospective drug discovery	8b	0.09 μM	[41]
3	2017	In silico aided-drug design	Prospective drug design, SAR rationalization, selectivity/specificity rationalization	9a	1 μM	[42]
4	2018	In silico-aided drug design (QSAR, docking)	QSAR: prospective drug design, docking: selectivity rationalization	11h	11.4 μM	[43]
5	2020	In silico aided drug design (pharmacophore model, docking)	Prospective drug design (pharmacophore model), SAR rationalization (docking)	12q	20 nM	[44]
6	2022	Hit expansion (synthesis of Ulotaront analogs) *	Study of the mechanism of action of Ulotaront, SAR rationalization.	13e	3.5 nM	[45]
7	2022	HTS+hit expansion	SAR rationalization	14o	112 nM	[46]
8	2022	HTS+hit expansion	Interaction mode investigation	16e	4 nM	[47]
9	2023	Similarity search+VS+MD	Drug design process	17b	0.405 μM	[48]
10	2023	Comparative docking+QSAR	Drug design process, selectivity profile elucidation	Guanfacine	20 nM	[49]
11	2024	In silico aided-drug design	SAR rationalization, drug design process	18a	526 nM	[50]

* Ulotaront was discovered through an in vivo phenotypic approach.

(entry 1).

One year later, Lam et al. published a similar study, performing a large-scale VS of more than 3 billion compounds, comprehending both fragment-like and lead-like compounds and referring to known TAAR1 ligands such as I and II (Figure 3) [41].

A set of 63 known TAAR1 ligands together with 161,000 commercially available compounds were docked to 200 HMs using the DOCK3.6 software [51]. It should be noted that II was also identified as a partial agonist, representing an interesting scaffold for the development of agonist and antagonist series. In detail, the VS was carried out based on an HM of hTAAR1 built on the X-Ray structure of the human β₂ adrenergic receptor (β₂-ADR), in the presence of the partial inverse agonist carazolol (PBD code = 2RH1) [52]: several HMs were obtained; their screening performance was evaluated via the Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) measure. The best-performing system was selected for the aforementioned VS of two ZINC libraries (a fragment-like library of 0.357 million compounds, and a lead-like library containing 2.7 million molecules). Among the top-scored compounds, forty-two molecules were selected for in vitro tests. Based on the structure-based studies, all the selected compounds were predicted to share important pharmacophore features with known active compounds, such as the capability to form a salt bridge with D103 and the presence of an aromatic moiety protruding towards TM5.

Following in vitro tests, nine TAAR1 agonists were identified, three of them being active in the low μM, such as compounds 7, 8a, and 8b (Figure 3). In Table 2 (entry 2), the chemical structure of 8b (Guanabenz) as a reference-screened compound is reported.

In 2017, Cichero et al. reported a computationally driven study on hTAAR1, in which several HMs were compared to guide the design of modulators exploring species-specificity profiles [42]. The most potent hTAAR1 agonist identified in this study is reported in Table 2 (9a; entry 3).

In detail, the previously published HMs of hTAAR1 [40], mTAAR1 [27], and h/mTAAR5 [42] were analyzed, as built on the same template, namely the X-Ray of β₂-ADR in complex with a covalently bound agonist (PDB ID = 3PDS) [30]. The putative docking mode of the endogenous ligand T1AM was calculated, relying on flexible docking studies using the Surflex docking module implemented in Sybyl-X1.0. According to this analysis, a H-bond to D103 was confirmed to be key for the TAAR1 agonist activity. Scaffold rigidity together with suitable H-bond features emerged as important properties for the design of selective TAAR1 ligands over TAAR5. Moreover, it was noticed that the presence of the phenol moiety in T1AM promoted promiscuity between TAAR1 and TAAR5, through the formation of an additional H-bond. Based on the above, compounds 9 and 10 were designed (Figure 3), including a biguanide moiety to meet the aforementioned rigidity criteria, maintaining a key basic moiety, and at the same time, the phenol group was removed to achieve TAAR1 selectivity over TAAR5 [42].

Following in vitro tests at h/mTAAR1 and mTAAR5, the low-μM to nM activity at mTAAR1 in eleven compounds out of twenty-seven was highlighted. Some of them also exhibited activity in hTAAR1, but with a 2-fold to 13-fold preference for mTAAR1 with respect to hTAAR1. All the compounds were inactive as mTAAR5 ligands. Subsequent SAR analysis allowed them to individuate features for the design of more potent TAAR1 agonists. In particular, proper hindrance at the ortho-para positions of the benzyl moiety was observed to increase selectivity over hTAAR1 and potency at hTAAR1, respectively.

In the following years, such a study was pursued to achieve a better understanding of m/hTAAR1 selectivity, and at the same time, further optimize the biguanide scaffold [43]. To this aim, two ligand-based QSAR models were developed based on a set of compounds with known activity and species-specificity profiles towards the murine and human orthologues [43], guiding the design of more selective ligands (11) (Figure 4). Among them, compound 11h has been reported as a modest and selective hTAAR1 agonist (Table 2, entry 4).

To develop the QSAR models, the collected dataset included thyronamine analogues [27,28], Guanabenz congeners [41], biguanides [42], imidazoles [53], and oxazolines [54].

For each model, the molecules were assigned to the training and the test set manually, based on representative criteria of the overall TAAR1 biological activity trend and structural variations. Any compound was explored in terms of geometry and conformation energy by means of the systematic conformational search module included in MOE software [31]. Chemoinformatic and QSAR packages of the same software MOE have been exploited, including molecular descriptors calculation. Afterwards, 302 molecular descriptors (2D and 3D) were obtained, and the resulting matrix was evaluated: the QuaSAR-Contingency and Principal Component Analysis (PCA) tools of MOE were employed for pruning molecular descriptors. The results proposed a few key descriptors to discriminate between murine and human orthologues. In particular, flexibility, as well as the number of polarizable H and positively charged groups, were related to hTAAR1 activation, while more rigid and electron-rich groups were predicted to enhance the possibility of activating mTAAR1. This information, together with the SAR obtained in the previous study, allowed for the design of the previously cited piperazine-biguanides 11, which were evaluated in in vitro tests of h/mTAAR1 and mTAAR5.

Two selective mTAAR1 ligands, such as 11c (Figure 4), were obtained, and one hTAAR1 selective ligand was discovered (11h, Figure 4). A docking simulation of these most promising compounds and of the m/hTAAR1 promiscuous agonist 11a explained the observed selectivity. For this analysis, the previously described HMs of mTAAR1 [27] and TAAR1 [40] were used.

The choice of a lipophilic electron-withdrawing moiety at the ortho position of the aromatic core turns in selective mTAAR1 agonists, as shown by 11c, while the only species-specific hTAAR1 agonist, 11h, exhibited an electron-donor group at the para position of the same ring (see Figure 4). The compound 11h’s docking pose highlighted one H-bond with S183, thanks to the methoxy substituent, and an additional H-bond with Y294 thanks to the basic moiety. Conversely, the ligand positioning was quite far from the key residue D103. Accordingly, 11h was a modest but selective hTAAR1 agonist.

For the design of dual-acting m/hTAAR1 agonists, the introduction of small functions endowed with electron-withdrawing properties at the ortho position of the main phenyl, or maintaining the same ring as unsubstituted, is preferred (see 11a, Figure 4).

On the other hand, the mTAAR1 selectivity of compound 11c (Ar = 2-Cl phenyl; mTAAR1 = 2.66 μM) and 11d (Ar = 2-pyrimidinyl; mTAAR1 = 9 μM), bearing an electron-rich moiety as the 2-Cl-phenyl substituent or pyrimidine group, seemed to be achieved through polar contacts involving the T83 sidechain.

The same research group further elaborated on the biguanide scaffold in light of a novel pharmacophore model developed on a set of potent oxazoline discovered by Roche [54], guiding the design of the novel, and more potent, TAAR1 agonists 12 [44]. Initially, the previously mentioned oxazolines were explored in terms of geometry and conformation energy by means of the systematic Conformational Search tool of the MOE software in order to develop the following pharmacophore analysis. Then, a pharmacophore model was calculated using the pharmacophore search module implemented in the MOE software, starting from the alignment of the aforementioned oxazolines onto the most potent one, taken as reference compound. Based on this information, a set of putative TAAR1 agonists (12) were designed. The most interesting analogue developed (12q) is reported in Table 2, entry 5.

Briefly, the biguanide scaffold was simplified to the amidino group, while the aryl-piperazine ring was maintained and decorated with several substituents (Figure 4). In vitro tests on hTAAR1 revealed the bioactivity of most of them, several of which displayed nanomolar activity (up to 20 nM). The docking of 12q in the previously mentioned hTAAR1 HM [40] was performed to support the results of the in vitro tests: the replacement of the biguanide with an amidino group was shown to be advantageous. Indeed, the most promising derivative 12q moved the amidine moiety into the proximity of the hTAAR1 H99 and D103 residues, detecting H-bond contacts (Figure 4). In addition, the folded piperazine, in tandem with the presence of substituents at the phenyl ring, were projected towards I104, F185, S198, W264, F267, F268, and I290 featuring π–π stacking and Van der Waals contacts.

More recently (2022), Heffernan et al. reported a retrospective study on Uloratont (Figure 5) to investigate its interaction mode with the target and to explore the SAR of this successful chemo-type [45].

Ulotaront was, in fact, discovered via the in vivo phenotypic approach [22], whereas the molecular target was revealed at a later stage [22]. The reference compound Ulotaront was docked in an hTAAR1 HM retrieved from the GPCRdb website [55] and built using the β₂-ADR (PDB ID = 3SN6) [56], with the loops modeled on D₂ dopamine receptor (D₂R) (PDB ID = 6CM4) [57], 5-HT_1B (PDB ID: 4IAQ) [58], and adenosine A_2A receptor (PDB ID: 4UHR) [59].

The obtained TAAR1–Ulotaront complex was submitted to simulated annealing molecular dynamics (MD), using a hybrid QM/MM model to enhance the accuracy in the binding site description. In particular, the starting pose for Ulotaront was determined by docking with the program FRED43 (v4.0) [60] while MD calculations were performed using the AMBER44 (v20) [61] simulation package. Based on the reported studies, the importance of a salt bridge interaction involving D103 was confirmed, as the Ulotaront bicyclic core projected towards V184, F195, F267, and F268 (Figure 5).

Some Ulotaront analogs were designed and tested in vitro based on in silico screening (Figure 5). Interestingly, one of the tested compounds exhibited an increased EC₅₀ value (up to 3.5 nM) (Table 2, entry 6), compared with Ulotaront (38 nM). In terms of SAR analysis, the results pointed out the effective role played by the choice of a primary amine group tethered to the main Ulotaront scaffold, as experienced by 13e (hTAAR1 EC₅₀ = 3.5 nM; Figure 5). This property should be accompanied by the S configuration, rather than to the R one. On the contrary, the expansion of the dihydropyran ring of Ulotaront to the tetrahydrooxepin ring impaired the ligand potency, compared with racemic Ulotaront. Finally, moving the sulfur position in the five-membered ring also proved to be disadvantageous to achieve TAAR1 activation.

In the same year, Krasavin et al. reported the discovery of a series of urea derivatives (14; Figure 5) via high throughput screening (HTS) and the subsequent hit expansion of 14o (Table 2, entry 7) [46]. The SAR of the novel series was investigated by the in vitro testing of further analogs, of which the most potent (14n; Figure 5) displayed an EC₅₀ of 33 nM. Moreover, a subset of the tested compounds was evaluated in silico to aid SAR rationalization. In the present case, the structural information was obtained by downloading the AlphaFold model for hTAAR1 (structure id: Q96RJ0) [62,63]. The protein model thus obtained was preprocessed with the use of the protein preparation wizard, included in the Schrödinger Suite (NY, USA, version 2021-4). The compounds were docked in the predicted structure by means of Glide [64], and, in addition to the docking score, an MM/GBSA procedure for the estimation of the free energy of the binding was applied. Both the examination of the docking poses with respect to a reference compound (Ralmitaront) and the free energy calculation allowed for the discrimination between active and inactive compounds. The most promising compounds were also evaluated in vivo, revealing the 3,5-dimethyl-phenyl-substituted analogue (14o) as featuring a statistically significant and dose-dependent reduction in hyperlocomotion in DAT-KO rats.

A similar study was performed by the same group starting with the triazole scaffold featured by 15, 16 (Figure 5), with 16e being the most promising individuated from HTS (see Table 2, entry 8) [47]. Compound 16e exhibited an EC₅₀ of 4 nM, being 30-fold more potent than Ulotaront. Among compounds 15, 16, those bearing the biaryl moiety proved to be more effective than the phenoxy substituted ones. The most promising, 16e, was investigated in silico, to deepen the knowledge of its interaction mode. Again, the hTAAR1 AlphaFold-predicted structure was utilized (ID: Q96RJ0) [62,63]. For all ligands, possible protonation states were calculated with the use of the Epik module of Schrodinger Suite [65]. Ligand docking with the prepared TAAR1 protein model was performed with the use of a Glide induced-fit docking (IFD) method [66].

Compound 16e was submitted to the IFD docking step, and the resulting complex was submitted to MetaDynamics to verify the persistence of the interaction network hypothesized via IF docking. According to the docking pose, the aromatic-rich structure of the ligand forms several lipophilic contacts with F185, F186 and F195, F267, and F268, as well as a π–π stacking interaction with the F267 aromatic ring. Additionally, a salt bridge is observed with the backbones of D274 and I281, and the sidechain of D274 itself. Compound 16e was also evaluated in vivo, displaying pronounced effects on the locomotor activity of MK-801-treated Wistar rats.

Very recently (2023), Wang et al. used the AF model of hTAAR1 for a prospective VS campaign [48]. More than one thousand low molecular weight molecules displaying similarity to Ulotaront [22] (Tanimoto index > 0.5) were retrieved and docked in the hTAAR1 model.

These calculations were conducted using the LibDock module in Discovery Studio 2018 [67]. The active site in TAAR1 was defined based on the known key residue D103 [68]. All compounds were prepared for docking simulation to consider appropriate protonation states, charges, and energy minimization. Among the top-scored molecules, two candidates (17a and 17b; Figure 5) were selected for MD evaluation (Table 2, entry 9). In silico analysis revealed a favorable interaction pattern for compound 17b, involving the formation of two H-bonds to D103, as well as favorable π–π interactions between the thiophene moiety and several aromatic residues (F195, F268, and W264). The two compounds were submitted to in vitro analysis, revealing EC₅₀ values in the low-μM to sub-μM ranges (17a = 6.249 μM, 17b = 0.405 μM). Moreover, they were evaluated against 5-HT and dopamine D2-like receptors, responsible for important off-target activities of traditional antipsychotic drugs. Compound 17b exhibited a desirable selectivity profile, and its evaluation was pursued with an in vivo efficacy and pharmacokinetics study.

In the same year, Cichero et al. performed a combined structure-based and ligand-based study to investigate the differences between the hTAAR1 and α₂-ADR in a drug design perspective [49]. Comparative docking calculations of the dual agonist S18616 [69], as well as of a series of imidazoline/imidazole-based compounds [53,54] with various activities and selectivity profiles, were performed. The X-Ray data of the α₂-ADR receptor (PDB code = 6KUY) [70] and the AlphaFold model of hTAAR1 (AF-Q96RJ0-F1) [71] were exploited.

Molecular docking simulations at the α₂-ADR receptor were performed by means of the DOCK module implemented in MOE software (2019.01 version), applying the template-based approach. The co-crystallized α₂-ADR ligand was taken as a reference compound. As regards the hTAAR1 AF model, the corresponding binding site was selected based on superimposition to the α₂-ADR protein, via Blosum62 (MOE software, 2019.01 version) [31].

In addition, the mentioned collection of agonists was utilized to produce two QSAR models, considering the response towards hTAAR1 and α₂-ADR. The two final models were derived applying the chemoinformatic and QSAR packages of MOE. The calculated 302 molecular descriptors were managed using the chemometric package PARVUS [72] for checking the constant predictors, splitting the data into training and test sets, and selecting the most informative molecular descriptors.

According to the obtained data, Guanfacine (Table 2, entry 10) was reported as a potent dual TAAR1/α₂-ADR agonist. Interestingly, the compound bioactivity was maintained in vivo.

In 2024, the same group explored an SAR rationalization of a series of amino-oxazoline TAAR1 agonists produced by Roche [50] by docking in the AlphaFold predicted structure of hTAAR1 [62,63]. All the molecular docking simulations at the hTAAR1 AF protein model were performed by means of the DOCK tool included in the MOE software [31], via a template-based approach using the previously described S18616-TAAR1 complex. According to the observed information, key requirements were determined for the hTAAR1 ligands, guiding the discovery of a novel chemo-type for the design of new agonists (see Table 2, entry 11). Consequently, a small set of pyrimidinone-benzimidazoles (18, Figure 5) was evaluated via ligand-based methods (FLAP2.2.1. software ligand-based module) [73,74]: molecular interaction fields’ (MIFs) compatibility with the reference compound S18616 was estimated. In addition, molecular docking calculations of compound 18 in the AlphaFold predicted structure were conducted. The results highlighted the low- to sub-micromolar activity of chemically novel compounds such as 18a–c derivatives.

In particular, the choice of the piperazine basic ring in the presence of a small hydrophobic chain in N (10) led to the most promising analogues 18a, 18b (hTAAR1 EC₅₀ = 526–657 nM) exhibiting beneficial Van der Waals contacts and π–π stacking with the hTAAR1 binding site. Removing the piperazine ring impaired the potency of the congeners lack of promising TAAR1 agonist ability.

3. Structural Information of TAAR1 as Druggable Target

3.1. Theoretical Models of TAAR1: An Overview

Several in silico-produced models of TAAR1 were used in drug discovery campaigns: two HMs were built for hTAAR1 [40,41], and one for mTAAR1 [27].

In addition, the AlphaFold-predicted structure of hTAAR1 and one hTAAR1 model by GPCRdb [75] were used as well. Most of the models employed a single X-Ray structure to model the target, more frequently choosing an agonist-bound receptor as a protein template. The most utilized is an X-Ray of the human β_2-ADR, covalently bound to an irreversible agonist, (3PDS) [30] or in complex with Carazolol (PDB ID = 2RH1) [52], or with the high-affinity agonist BI-167107 (PDB ID = 3SN6) [56].

In particular, two computational studies [33,41] highlighted the possibility of retrieving both agonists and antagonists by VS on models built with template structures containing exclusively an agonist [33] or a partial inverse agonist [41], applying a combined ligand- and structure-based approach. Accordingly, structurally significant explanations for agonist-bound and antagonist-bound conformations of TAAR1 are thought to be limited. Notably, this information was then supported by crystallographic evidence for aminergic receptors [76] and discussed by Laeremans et al. [77]. Among the reported examples, Costanzi and Vilar [78] carried out a retrospective VS study on the α₂-ADR, to verify the capability of different conformations of the receptor to enrich the ranking of agonists and antagonists over each other and with respect to decoys. It has been shown that the α₂-ADR inactive conformations in complex with an inverse agonist (2RH1) or antagonists (3NYA) were able to discriminate antagonists from agonists. The active conformation (3P0G), conversely, was able to enrich agonists over antagonists. The inactive state associated with an irreversibly bound agonist (3PDS), however, did not discriminate between agonists and antagonists. Noticeably, the 3PDS structure was used in most of the HMs built for hTAAR1, and the related VS approaches based on this conformation retrieved both agonists and antagonists. The 2RH1 PDB was also used to model TAAR1, and the subsequent VS again retrieved mixed agonists/antagonists, in contrast with the mentioned conformation evaluation. More recently (2019), Scharf et al. [79] reported the possibility of using multiple active conformations of the receptor to favor the discovery of an agonist, always considering the α₂-ADR.

A perspective of the developed m/hTAAR1 theoretical models is reported in

Table 3. List of the HMs produced in the context of drug discovery campaign towards h/mTAAR1 (hTAAR1 in green, mTAAR1 in grey). References (Ref.), resolution (R), release date (R.D.), and percentage of identity (% Id) are reported.

Model n.	TAAR Models	First Published in Ref	Utilized in Ref (s)	Template	R (Å)	R.D.	Presence of Small Molecules/Ligand-Based HM	% Id. (BLAST)
1	hTAAR1	[40]	[33,40,42,43,44,86]	3PDS	3.50	2011	Irreversible agonist (co-crystallized)	31.60%
2	hTAAR1	[41]	[41]	2RH1	2.40	2007	Carazolol (partial inverse agonist) + ligand guided HM	31.60%
3	hTAAR1	[45]	[45]	HM by GPCRdb9 web site. Backbone of the TM helices: 3SN6 (prevalent template) loop coordinates: 6CM4, 4IAQ, 4UHR	3SN6: 3.2 6CM4: 2.87 4IAQ:2.80 4UHR: 2.60	3SN6: 2011 6CM4: 2018 4IAQ:2013 4UHR: 2015	3SN6: high affinity agonist (BI-167107) 6CM4: Risperidone (inverse agonist) 4IAQ: Dihydroergotamine (agonist) 4UHR: selective agonist CGS21680	31.60% (3SN6)
4	hTAAR1	Online source (AlphaFold, Structure ID Q96RJ0) [62,63] last updated in AlphaFold DB version 2022-11-01	[46,47,48]	Structure ID Q96RJ0	NA	NA	no	NA
5	mTAAR1	[27]	[27,28,42,43,86]	3PDS	3.50	2011	Irreversible agonist + ligand-based HM (T1AM)	31.48%

. The percentage of identities with respect to the exploited protein template was calculated by aligning the two sequences with the BLAST-p algorithm [80,81,82]. The sequences were retrieved by the proper Uniprot entries [83]. The BLOSUM62 matrix [84] was used for the alignment, with a gap existence penalty of 11, and a gap extension penalty of 1. The conditional compositional score matrix was used to consider the different amino acid compositions of the query with respect to the frequencies used for the calculation of the substitution matrices [85]. The word-size was set to 3.

Since the Cryogenic Electron Microscopy (Cryo-EM) data of the hTAAR1 and mTAAR1 were recently solved, it is possible to compare the experimental structures of TAAR1 with the templates utilized for TAAR1 HM building and ligand design. Considering hTAAR1, the templates employed for homology modeling mainly included the β₂-ADR (PDB IDs: 3PDS, 2RH1, 3SN6) [30,52,56]. In addition, the m/hTAAR1 AlphaFold modelled structures (AF) were analyzed as well [62,63].

As regards hTAAR1, the PDB IDs 3PDS [30] and 2RH1 [52] contain the coordinates of the β₂-ADR protein- Tequatrovirus T4 lysozyme (LYZ), in complex with an irreversible agonist and a partial inverse agonist, respectively (Figure 6A, template in yellow). The 3SN6 PDB [56] also reports the T4 LYZ, but in this case, the receptor is associated to a G-protein (Figure 6A, template in yellow). The 3SN6 structure is in complex with the BI-167107 agonist [56]. The hTAAR1 AlphaFold structure only involves the receptor in the apo-form.

In the Cryo-EM structures of hTAAR1, such as in PDB code 8W8A [87], the protein target is associated to the three subunits of a G-protein (Figure 6A, experimental hTAAR1 protein in green). By observing Figure 6B, it is possible to highlight an overall agreement in the receptor folding between the experimental structure of the target and the templates as transmembrane domains (TMs), but still, some helices (Hs) and/or extracellular loops (ECLs) significantly deviate from the hTAAR1 conformation.

For the 3PDS/TAAR1 systems, ECL2, H8, TM6, TM1, and TM2 exhibit larger discrepancies (Figure 6B). A very similar situation is observed for 2RH1. Conversely, 3SN6 displays a more adherent conformation to TAAR1 with discrepancies concentrated prevalently at ECL2 and H8. The hTAAR1 AF structure exhibits the best fit to the experimental TAAR1 structure, exhibiting the lowest α-carbon atom RMSD value among the analyzed templates. TM6, however, deviates importantly from the template structure. Such a helix is considered the hallmark of class A GPCR activation; in particular, its shift outwards is an indication of an activated state, while a straighter positioning is indicative of an inactive state [88,89].

In the 3PDS and 2RH1 data, the template protein has been reported in the inactive state, and the AF-predicted structure also presents an inactive-like conformation. 3SN6, on the other hand, exhibits the active conformation. The fact that an agonist-bound form, such as 3PDS, can assume the inactive conformation is coherent with experimental data reporting that agonist binding alone is not sufficient for GPCR complete activation, as the intracellular binding to G-proteins or stabilizing peptides is also required for complete receptor activation [30,76]. Regarding the binding site, it is possible to verify that a good correspondence is reached in terms of the superimposition of conserved amino acids. However, some differences can be highlighted as shown in Figure 7.

For 3PDS, many non-conserved residues are present (Figure 7A). The template/8W8A S203/T194, V114/I104, T118/S108, N312/I290, and W109/H99 couples are shown to exhibit a certain agreement in terms of spatial positioning. For other residues, important differences in terms of steric and/or electrostatic properties, and/or backbone position, arise (Y199/S190, T195/F186, D192/S183, N293/T271, F193/V184, V117/S107). As can be expected, backbone displacements are more easily observed for residues towards the extra-cellular region with respect to more buried residues. In this PDB, residue H93 was mutated to cysteine to anchor the β₂-ADR irreversible agonist. Apart from this difference, a similar situation can be observed for 2RH1 and 3SN6, as the superimposed protein is the same. In the case of 3SN6, several residues composing the binding site were not completely solved. As the AlphaFold structure of TAAR1 was predicted based on the TAAR1 sequence, this analysis is not extendable to this case. However, although conserved, many amino acids display different orientations with respect to the experimental structure. The most relevant cases are related to residues F195 and R179 (ECL2).

Regarding mTAAR1, the utilized template is again β₂-ADR (PDB ID: 3PDS) [30]. When superimposed to mTAAR1, the template shows an overall accordance with respect to the target conformation (Figure 8A). A few discrepancies can be observed between the template and mTAAR1 ECL2, TM1, TM2, TM3, and TM6, with the latter highlighting again the different activation states of the template (inactive) and the mTAAR1 (active).

In terms of residue conservation, the situation is highly superimposable onto the hTAAR1, since most of the binding site residues are conserved in the murine orthologue. However, h/mTAAR1 binding sites differ in four amino acids (A193, Y153, P183, and Y287). These residues are also non-conserved in the β₂-ADR (PDB ID: 3PDS) in which the following substitutions are observed: A193 to S203, Y153 to T164, P183 to F193, and Y287 to N312. Further information on residue conservation between the human and mouse orthologues are reported in Section 3.3.

3.2. TAAR1 Experimental Data and Mutagenesis Information

To date, thirteen Cryo-EM structures containing hTAAR1 are available, as well as ten mTAAR1 structures (

Table 4. List of the available PDB entries (PDB ID) containing hTAAR1 (in cyan) or mTAAR1 (in grey) data. This piece of information has been listed based on the resolution parameter (Å). The corresponding number of non-hydrogen atoms (n. of non-H atoms) and corresponding references (Ref.) are reported.

PDB ID	Ligand Name	Resolution (Å)	n. of Non-H Atoms	Ref.
8W88	Ulotaront	2.60	8307	[87]
8W87	METH	2.80	8228	[87]
8W8A	RO5256390	2.80	8260	[87]
8JLQ	Fenoldopam	2.84	9033	[90]
8WC8	ZH8651	2.90	8062	[91]
8W89	β-PEA	3.00	8230	[87]
8JLR	A77636 (adamantane derivative)	3.00	8934	[90]
8JLP	Ralmitaront	3.23	7343	[90]
8JLN	T1AM	3.24	9117	[90]
8UHB	RO5256390	3.35	8417	[92]
8JSO	D-AMPH	3.40	9045	[90]
8WCA	β-PEA	3.48	8160	[91]
8JLO	Ulotaront	3.52	7343	[90]
8WC3	Ulotaront	3.00	8418	[91]
8WCC	CHA	3.04	2056	[91]
8JLJ	T1AM	3.10	8918	[90]
8WCB	CHA	3.10	7784	[91]
8WC7	ZH8667	3.10	8532	[91]
8WC4	ZH8651	3.10	8394	[91]
8WC9	ZH8651	3.20	8394	[91]
8WC6	β-PEA	3.20	9183	[91]
8JLK	Ulotaront	3.22	8824	[90]
8WC5	TMA	3.30	8511	[91]

). The resolution range varies from 3.52 Å (8JLO) [90] to 2.6 Å (8W88) [87]. All the reported structures are associated with an agonist (involving several chemo-types) and with various G-proteins. They all exhibit activated conformation.

By superimposing all thirteen hTAAR1 available structures (Figure 9A), a good agreement of the reported protein conformations can be noticed. As previously reported by Xu [90], the binding site residues themselves display an overall rigidity in response to the binding of different chemo-types. In particular, the deeper portion of the binding cavity is particularly fixed (residues W264, Y294, F267), and other amino acids display small (S107, S108, F186, F268, I290, I104) or important (T194, S189, R83) differences in sidechain conformations but a superimposed position of the backbone.

Conversely, residues belonging to ECLs (V184, S183) exhibit larger displacement in the backbone positioning. A similar analysis can be extended to mTAAR1, (Figure 9B), by comparing the receptor conformations and residue positioning when in complex with different ligands. Again, a close agreement is observed in the overall conformation of the protein. Moreover, the residues composing the binding sites exhibit very similar orientations, except for S197 and a few moderate displacements of Y287, F264, and P183 residues. However, when comparing the binding site rigidity of hTAAR1 and mTAAR1, it is necessary to consider the larger chemical diversity of the ligands in the case of hTAAR1 with respect to mTAAR1, as well as the number of superimposed structures (thirteen vs. ten).

In addition to the abundance of structural information, extensive mutagenesis experiments have been carried out to clarify the role of the binding site residues in h/mTAAR1 activation by several chemo-types. For hTAAR1 agonism, the flexible phenylethylamine-based derivates, methamphetamine (METH), β-PEA, and amphetamine (S-AMPH), have been evaluated, as well as the bulkier and more rigid Ulotaront, RO5256390, and T1AM [87]. The results for hTAAR1 are reported in

Table 5. Effect of residue mutation in hTAAR1 in terms of activity impairment with respect to the maximum ligand-induced activation. NO (white): none. YES (dark green): the mutation strongly impairs or eliminates activity. PARTIAL (orange): the mutation partially diminishes activation. POOR (dark pink): the activation diminishment is poor. Blue: the mutation augments the ligand-induced activation. Grey cells: no data available.

	hTAAR1 Agonists
Protein Mutants	METH [87]	β-PEA [87]	(S)-AMPH [87]	Ulotaront [87]	RO5256390 [87]	T1AM [87]
D103A	YES	YES	YES	YES	YES	YES
I104A	YES	YES	YES	YES	YES	YES
S107A	YES	YES	YES	YES	YES	YES
F186A	YES	YES	POOR	YES	YES	YES
T194A	YES	YES	-	YES	YES	YES
W264A	YES	YES	YES	YES	YES	YES
F267A	PARTIAL * (70%)	YES	POOR	PARTIAL * (56%)	YES	YES
Y294A	YES	YES	YES	YES	YES	YES
I290T	YES	YES	-	YES	YES	-
I290N	YES	YES	-	YES	YES	-
S80A	POOR * (84%)	PARTIAL (47%)	-	PARTIAL (53%)	PARTIAL (52%)	YES
R83A	YES	YES	-	YES	YES	YES
F185A	NO * (107%)	PARTIAL (52%)	-	PARTIAL (66%)	PARTIAL * (58%)	-
H99A	YES	YES	-	YES	YES	YES
S107C	YES	YES	-	YES	YES	-
S198A	ACTIVATION (150%)	ACTIVATION (129%)	-	NO (99%)	ACTIVATION (113%)	PARTIAL
S108A	-	-	-	-	-	YES
I290A	-	-	YES	-	-	-
F268A	-	-	PARTIAL (~50%)	-	-	YES
V184A	-	-	PARTIAL (~50%)	-	-	YES

* Such data exhibit a large standard deviation value.

[87].

The measurements come from different experiments and the percentages are related to different maximum activation values; some data exhibit significant experimental uncertainty (star-labelled in Table 5). However, it is possible to qualitatively compare the importance of the mutated residues with respect to various chemo-types.

As an example, it is possible to highlight that for all the analyzed ligands, the mutation to alanine of residues D103, I104, S107, W264, and Y294 is detrimental for hTAAR1 activation. This turns in key contacts, guaranteed by aromatic and protonable moieties in the hTAAR1 ligand, as required features. Residues F186, T194, R83, and H99 also affect the agonist binding, except for (S)-AMPH. This could be explained based on the possibility of displaying cation–π contacts or additional H-bonds, thanks to the agonist-protonated nitrogen atom. In the case of (S)-AMPH, the basic group is quite hindered by the methyl group if compared with the other agonists.

The I290 mutation to T or N strongly impairs activity for METH, β-PEA, Ulotaront, and RO5256390. In this case, no data are available for T1AM and (S)-AMPH. However, the latter is affected by the I290A mutation. The S80A mutation has a strong effect on T1AM, as well as various ranges of activation for the other ligands (no data for (S)-AMPH), probably as a key H-bonding feature. The F267A mutation has a strong impact on β-PEA, RO5256390, and T1AM activation, and a lesser (but still significant) impact on METH and Ulotaront. On the contrary, a limited effect is observed for (S)-AMPH.

The F268A mutant was produced to assess only the agonism ability of T1AM and (S)-AMPH, leading to a strong and partial reduction in the protein activation, respectively. A similar situation is observed for V184A. Moreover, the mutation of S107 to C was evaluated for METH, β-PEA, SEP-363856, and RO5256390, leading to a strong decrease in activity. This information confirmed the relevant role played by exhibiting H-bonds and π–π stacking between the protein cavity and the ligand, which has to be endowed with limited dimensions and steric hindrance to fit the protein crevice.

Accordingly, S108A mutation was shown to decrease T1AM-induced activation of TAAR1, whereas a partial reduction was observed in response to the F185A mutation on β-PEA, Ulotaront, and RO5256390. Interestingly, the S198A mutation leads to a complex effect for all the ligands excluding (S)-AMPH (no data), exhibiting no effect or an increased activation in response to ligand stimulation. Mutational effects regarding METH, β-PEA, Ulotaront, and RO5256390 were evaluated by means of miniGs’ recruitment tests in [87], whereas data regarding T1AM and (S)-AMPH were evaluated by CAMYEL assays [90]. In addition to the reported results, mutational data are available regarding the effect of three mutations (S80A, R83A, and H99A) on hTAAR1 activation by Fenoldopam, A77636, and Ralmitaront. In all three cases, the mutations appear to have had a moderate to strong influence on ligand-induced activation [90].

Extensive mutagenesis was also performed for a series of key compounds for mTAAR1 activation [90,91], referring to the protein agonists T1AM, cyclohexylamine (CHA), trimethylamine (TMA), β-PEA, and Ulotaront. The corresponding results for mTAAR1 are reported in

Table 6. Effect of residue mutation in mTAAR1 in terms of activity impairment with respect to the maximum ligand-induced activation. NO (white): none. YES (dark green): the mutation strongly impairs or eliminates activity. PARTIAL (orange): the mutation partially diminishes activation. POOR (dark pink): the activation diminishment is poor. Grey cells: no data available.

mTAAR1 Agonists
Protein Mutants	Ulotaront [90]	T1AM [90]	TMA [91]	CHA [91]	β-PEA [91]	Ulotaront [91]
D102A	YES	YES	YES	YES	YES	YES
S106A	YES	YES	PARTIAL	NO	YES	PARTIAL
I103A	PARTIAL (~60%)	YES	POOR	PARTIAL	PARTIAL	PARTIAL
F185A	POOR *	YES	POOR	PARTIAL	PARTIAL	YES
W261A	YES	YES	YES	YES	YES	YES
F264A	YES	PARTIAL (~50%)	PARTIAL	PARTIAL	PARTIAL	PARTIAL
F265A	YES	YES	NO	PARTIAL	YES	PARTIAL
Y291A	-	YES	YES	NO *	YES	YES
S107A	-	POOR	-	-	-	-
P183A	-	POOR	POOR	NO	PARTIAL	YES
A193T	-	POOR	-	-	-	-
Y153A	-	-	NO	NO	YES	PARTIAL
S197A	-	PARTIAL (~50%)	NO	NO	NO	PARTIAL
Y287A	-	-	PARTIAL	PARTIAL	YES	PARTIAL

* Such data exhibit a large standard deviation value.

[90,91]. Data regarding Ulotaront and T1AM were obtained by GloSensor™ assays [90], while the remaining mutational data and also Ulotaront were obtained by G protein dissociation assays (BRET) [91].

Regarding mTAAR1 agonism ability, for each ligand explored in the mutagenesis experiments, the most required key contacts include the conserved D102 residue. Accordingly, very small agonists such as TMA or CHA are mTAAR1 agonists, featuring a poor potency trend towards the human orthologue [93]. In addition, the W261A mutation deeply affected the ability of all the compounds to activate mTAAR1, as previously observed for the corresponding W264 in hTAAR1. It should be noticed that the majority of the mTAAR1 agonists herein cited exert their agonist role thanks to aromatic residues, such as F264, F265, and Y287, which are reported as affecting the TAAR1 activation. On the contrary, most of the non-aromatic residues analyzed (S107, P183, and A193) poorly affect the ligand binding. On the contrary, interacting with H-bonding and non-aromatic residues in hTAAR1 has been previously reported as key to achieving hTAAR1 agonism (see previous Table 5). This, in turn, suggests less planar but folded hTAAR1 agonists. This information has been previously proposed in the literature via QSAR studies, which have pointed out the effectiveness of more flexible chemo-types for the design of hTAAR1 agonists, while more extended and rigid cores should be preferred for the murine orthologue [43]. In addition, the presence of electron-donor groups is thought to improve hTAAR1 binding ability via H-bonds with the aforementioned key residues S107, P183, and A193.

3.3. Comparison of the Binding Pockets of hTAAR1 and mTAAR1

The recently published structural information involving both h/mTAAR1 sheds light on the species-specificity issue, a critical problem in TAAR1 ligand design [93,94,95,96]. The presence of several structures of h/mTAAR1 in complex with the same ligand allows a strict comparison of the binding site, possibly leading to species-specific effect rationalization. The murine orthologue mTAAR1 was solved in complex with Ulotaront (PDB ID: 8JLK) [90] and T1AM (PDB ID: 8JLJ) [90], as well as β-PEA (PDB ID: 8WC6) [91] and ZH8651 (PDB ID: 8WC4) [91].

The corresponding complexes with hTAAR1 are also available with the following PDB IDs: 8JLO including Ulotaront [90], 8JLN in presence of T1AM [90], 8W89 [87] or 8WCA [76] including β-PEA, 8WC8, and ZH8651 [91].

The superposition between the three-dimensional structures of the h/mTAAR1 in the presence of different chemo-types allows to individuate four non-conserved residues which would help in explaining species-specificity at the protein orthosteric binding site: Y153(m)/F154(h), P183(m)/V184(h), A193(m)/T194(h), and Y287(m)/I290(h) (Figure 10A–C).

As shown in Figure 10A, T1AM, which exhibits higher potency values towards mTAAR1 compared with hTAAR1, seems to be better stabilized at the murine orthologue as endowed with more aromatic (Y287) or hydrophobic residues (A193) than hTAAR1 (I290 and T194 at the same positions).

Xu et al. provided the structural basis for the selectivity of A77636, a catechol derivative reported to be active on hTAAR1 and inactive towards mTAAR1 [90]. As shown in Figure 10D, the Cryo-EM structure of the hTAAR1:A77636 complex highlights the protruding of the ligand towards residues V184 and I290. These two amino acids are non-conserved in the murine orthologue, being mutated to bulkier residues P183 and Y287, respectively. The larger hindrance introduced in the murine case is reflected in the narrowed shape of the binding pocket in the corresponding area, not allowing the adamantane moiety to be accommodated. These data are further (partially) supported by mutagenesis experiments, as both the hV184A and the hV184P mutants exhibited impaired A77636-induced activation [90]. Such information is critical for the design of novel h/mTAAR1 ligands, to control the species-specific aspect. In the same article [90], a rationalization for T1AM preference for mTAAR1 over hTAAR1 was proposed. According to this hypothesis, the non-conserved couple of residues mA193/hT194 would be responsible for the loss of activity at the hTAAR1 with respect to mTAAR1. The key role of such residues in species-specificity was previously proposed by computational studies [95,97].

In the search for m/hTAAR1 antagonists, computational techniques were also utilized to explore species-specificity issues [97]. Indeed, no experimental structures of TAAR1 revealing an antagonist binding mode are available, possibly due to the limited availability of such compounds. Thus, the putative binding modes of EPPTB and the antagonist 4c (Figure 11A), previously reported, were analyzed by docking/MD and docking, respectively [33,97]. The proposed interaction pattern has been reported in Figure 11A–C.

On the other hand, HTS approaches [100] followed by structure-activity optimization allowed for the discovery of the hTAAR1 antagonist RTI-7470-44, endowed with a species-specificity preference over mTAAR1 (Figure 11A) [99]. RTI-7470-44 displayed good blood–brain barrier permeability, moderate metabolic stability, and a favorable preliminary off-target profile. In addition, RTI-7470-44 increased the spontaneous firing rate of mouse ventral tegmental area (VTA) dopaminergic neurons and blocked the effects of the known TAAR1 agonist RO5166017.

Beyond the design of compounds selectively binding to the TAAR1 orthosteric site, allosteric modulation is gaining attention in the field of GPCR [101,102]. Several allosteric modulators were reported for class A GPCRs, some of them being associated with structural information [101,102]. As an example, three allosteric modulators were co-crystallized in complex with the β₂-ADR receptor: compound-15PA (PDB ID: 5X7D) [103], compound-6FA (PDB ID: 6N48) [104], and AS408 (PDB ID: 6OBA) [105] (Figure 12A). Two of them, the phenylalaninamide compound-15PA (Figure 12) and the phenylquinazoline AS408 (Figure 12), exert a negative allosteric modulation, while the aryl-sulphonamide derivative compound-6FA (Figure 12) is a β₂-ADR positive allosteric modulator.

The identified allosteric site for the β₂-ADR is formed by TM 1, 2, 6, 7, ICL1, and H8 (cytoplasmatic end of the receptor, 5X7D), by TM 3, 4 and ICL2 (cytoplasmatic end, 6N48), and TM3, 5 (membrane facing surface, 6OBA). However, the known allosteric sites for class A GPCRs exhibit large location variability, taking into account the lipidic interface, the extracellular interface, and the cytoplasmic interface (Figure 12B).

In addition, ligands that bind both to the orthosteric and protrude towards an allosteric site (bitopic ligands) were reported [102,116]. An example of a bitopic ligand is reported in Figure 12C. A complete discussion of GPCR allosteric sites is beyond the scope of this review, but several valuable reviews are available on the topic [101,102,117,118]. Regarding TAAR1, no allosteric modulators have been reported to our knowledge. However, a putative TAAR1 allosteric binding pocket was hypothesized by Glyakina et al. through a bioinformatic approach [119].

4. Computational Methods Guiding the Discovery of TAAR5 Ligands

4.1. In Silico Screening of Novel TAAR5 Ligands

Concerning the design of mTAAR5 ligands, Cichero et al. reported a VS study [86] of a series of 5-HT_1A receptor ligands [34,35,36,37,38,39,120], formerly screened against TAAR1 [33]. Following the same procedure applied for hTAAR1 [42], the h/mTAAR5 receptors were modelled on the basis of the β₂-ADR (PDB ID:3PDS), using the BLOSUM62 matrix for the target/template alignment. The structures were minimized, and several parameters were considered for quality evaluation, including the Ramachandran plot analysis, the evaluation of proper distribution of hydrophobic/hydrophilic residues in different areas of the protein, and the rotamer strain energy, among others. The T1AM reference compound was docked in the binding site of the obtained model(s) with the use of Sybyl-X1.0, and the best poses were refined by post-docking minimization. Furthermore, the residues around the ligand were submitted to rotamer analysis to explore better conformations. The virtual screening was performed with Sybyl-X1.0. A structural analysis of four HMs (h/mTAAR5 and h/mTAAR1) was reported to guide the design of isoform-selective and species-specific compounds prior to synthesis. In particular, the binding sites were compared in terms of residues conservation and putative interactions with the reference compound T1AM, taking into consideration the different behavior of such compounds towards the considered TAARs. Indeed, T1AM has been reported as a TAAR1 agonists, also featuring hTAAR5 inverse agonist ability. The screening results were selected according to pharmacophore features based on the previously mentioned structure-based study, and subsequent in vitro tests to individuate two novel mTAAR5 antagonists (IC₅₀ = 4.8 ± 1.1 μM and 29 ± 1.4 μM).

The two candidates bear diphenyl-dioxolane (19) or tetrahydrofuran (20) scaffolds, respectively, and exhibit selectivity with respect to mTAAR1 (

Table 7. Drug discovery studies focused on mTAAR5 antagonists, involving modelling techniques. The related references (Ref.) have been reported.

Entry	Year	Method of Discovery	Use of Structural Information	Proposed Hits	mTAAR5 IC50	Ref.
1	2016	VS on mTAAR5 HM	Prospective VS, selectivity rationalization	19	29 μM	[86]
				20	4.8 μM
2	2022	VS on mTAAR HM	Prospective VS	21, 22 Chemical structure not shown	1.1 μM	[121]
3	2023	VS on mTAAR HM	Prospective VS	23	21 μM	[122]
				24	3.5 μM
				25	2.8 μM

, entry 1).

Interestingly, the docking analysis highlighted the presence of specific contacts, such as a H-bond to T115, resulting in the key to TAAR5 selectivity over TAAR1. These data confirmed the key role of T115 in TAAR5, being non-conserved in the hTAAR1 orthologue. In addition, the dioxolane and the tetrahydrofuran derivatives 19 and 20 displayed a switched binding mode, maintaining, in any case, a key salt-bridge with D114, through the compound basic moiety. Several π–π stacking contacts with W265, F287, and Y295 were also reported (Figure 13A).

More recently [121], Bon et al. performed a large-scale VS on a homology model of mTAAR5. In detail, several HMs were built using as templates protein structures with sequence identity superior to 32%, and a resolution of at least 3 Å. On the basis of several evaluation parameters such as the alignment coverage, the backbone RMSD with respect to the template structure, and others, the best HM was built on the Meleagris gallopavo β₁-ADR in complex with the agonist formoterol (PDB ID: 6IBL) [123]. The screening was carried out with AtomNet^® (Atomwise), a structure-based deep convolutional neural network trained for VS purposes, able to predict the affinity of a set of molecules. The binding site was defined on the basis of the ligand position at the template structure. The Enamine In-Stock (https://enamine.net accessed on 23 November 2023) HTS library of around 2 million small molecules was prepared and screened. The top section of the VS results was further submitted to filtering according to various descriptors and clustering with a Tanimoto similarity cutoff of 0.35. Ninety-six compounds were selected among the results, covering different chemical structures. Among the tested compounds, two hits, 21 and 22 (chemical structure not available), were retrieved, exhibiting antagonist behavior [121]. The two compounds experienced IC₅₀ = 2.8 μM and 1.1 μM, respectively (Table 7, entry 2).

In 2023, Nicoli et al. performed a large-scale VS towards a HM of mTAAR5 built starting from the crystal structures of the human β₂-ADR (PDB ID: 4GBR) [124] and the wild turkey β₁-ADR (PDB ID: 2Y03) [125], exhibiting a structural similarity of 28% and 31%, respectively [122].

The ECL2 sequence was modeled based on the ECL2 of neuropeptide receptor Y1 (PDB ID: 5ZBH) [126] (15% sequence identity). MODELLER version 9.25 [127] was employed to generate one hundred possible HMs, and the structure with the best DOPE (discrete optimized protein energy) score was selected. Additionally, the HM was submitted to intra-molecular H-bond optimization at physiological pH with Maestro [128]. The quality of the model was evaluated by various metrics such as Ramachandran plot and steric clashes presence. The model was further optimized through structural and sequence comparison with serotoninergic receptors. Two models (model A and B) were obtained. The SPECS library of approximately 200,000 compounds (https://www.specs.net accessed on 23 November 2023) was prefiltered according to drug-like and pharmacophoric features via Phase by Schrödinger [129]. The filtered compounds were then submitted to docking in the TAAR5 HM(s) with Glide standard protocol [130].

Three compounds (23–25) (Table 7, entry 3) were found to exhibit antagonistic activity at mTAAR5. The corresponding mTAAR5 IC₅₀ values were 21 ± 0.18 μM, 3.5 ± 0.15 μM, and 2.8 ± 0.16 μM, respectively. Such compounds were further investigated by in-membrane molecular dynamics (MD). For each complex, three replicas of 200 ns in the NVT ensemble were carried out and analyzed.

Compound 23 displayed an ionic interaction with D114 through the charged aliphatic tertiary amine, while the hydroxyl group was H-bonded to D114 too.

The aromatic moiety was engaged in π−π interactions with F268 and in Van der Waals contacts with L203. The two most potent compounds, 24, 25, moved the two aromatic rings towards W265 and F268, as the protonated basic moiety involved in a salt bridge with the key residue D114 (see 25 in Figure 13B).

Based on the above, up to now, three HMs of mTAAR5 and one for hTAAR5 have been exploited, guiding the search for novel chemo-types acting as TAAR5 ligands. In

Table 8. List of the HMs produced in the context of drug discovery campaigns towards h/mTAAR5 (hTAAR5 in green, mTAAR5 in grey). References (Ref.), resolution (R), release date (R.D.), and percentage of identity (% Id) are reported.

Model n.	Modelled TAAR	Generated in Ref.	Utilized in Ref (s)	Template	R (Å)	R.D.	Presence of Small Molecules/Ligand-Based HM	% Id. (BLAST)
1	hTAAR5	[86]	[42,86]	3PDS	3.50	2011	Irreversible agonist	34.26%
2	mTAAR5	[86]	[42,86]	3PDS	3.50	2011	Irreversible agonist	33.89%
3	mTAAR5	[121]	[121]	6IBL	2.70	2019	Formoterol (agonist)	33.33%
4	mTAAR5	[122]	[122]	Main templates: 4GBR, 2Y03 ECL2: 5ZBH	4GBR: 3.99, 2Y03: 2.85, 5ZBH: 3.00	4GBR: 2012 2Y03: 2010 5ZBH: 2018	4GBR: S-Carazolol (inverse agonist) 2Y03: Isoprenaline (agonist) 5ZBH: BMS-193885 (antagonist)	4GBR: 33.89% 2Y03: 33.33% 5ZBH: 24.44%

, a perspective of the developed HMs in comparison with the selected protein template is reported.

The percentage of identity was calculated by aligning the two sequences with the BLAST-p algorithm [80,81,82]. The sequences were retrieved using the proper Uniprot entries [83]. The BLOSUM62 matrix was used for the alignment, with a gap existence penalty of 11, and a gap extension penalty of 1. The conditional compositional score matrix was used to consider the different amino acid compositions of the query with respect to the frequencies used for the calculation of the substitution matrices. The word-size was set to 3.

4.2. TAAR5: Better Templates for New HMs

To date, no experimental structure have been reported for m/hTAAR5. In this case, HMs remain a possible strategy to perform drug discovery campaigns toward the target. In search for novel templates, we performed a sequence search towards hTAAR5 with BLAST-P [80,81,82], restricting the query to proteins included in the protein data bank [131]. The results are reported in

Table 9. BLAST-p alignment results according to the BLOSUM62 matrix, using hTAAR5 as query. The proposed template is indicated via its PDB ID, name of the macromolecule, organism, BLAST total score, coverage of the sequence, and percentage of identity (% Id.) between the query and the template.

PDB ID	Description	Scientific Name	Total Score	Query Cover	% Id.
8ITF	Trace amine-associated receptor 9	Mus musculus	299	97%	46.45%
8PM2	Trace amine-associated receptor 7f	Mus musculus	321	96%	45.65%
8W87	Trace amine-associated receptor 1	Homo sapiens	251	95%	38.74%
8JLN	Trace amine-associated receptor 1	Homo sapiens	253	99%	38.33%
8JLJ	Trace amine-associated receptor 1	Mus musculus	247	99%	37.90%
6H7J	β-1 adrenergic receptor	Meleagris gallopavo	158	84%	37.72%
6IBL	β-1 adrenergic receptor	Meleagris gallopavo	158	84%	37.72%
2VT4	β-1 adrenergic receptor	Meleagris gallopavo	155	82%	37.41%
6TKO	β-1 adrenergic receptor	Meleagris gallopavo	159	82%	37.23%
2Y00	β-1 adrenergic receptor	Meleagris gallopavo	153	82%	37.14%
7JJO	β-1 adrenergic receptor	Meleagris gallopavo	152	81%	36.62%
5A8E	β-1 adrenergic receptor	Meleagris gallopavo	150	82%	36.43%
4BVN	β-1 adrenergic receptor	Meleagris gallopavo	149	82%	36.43%
7EJ0	α-2A adrenergic receptor	Homo sapiens	160	82%	36.14%
7XT8	5-hydroxytryptamine receptor 4	Homo sapiens	157	80%	35.45%
4LDE	β-2 adrenergic receptor	Homo sapiens	139	77%	34.96%
5JQH	β-2 adrenergic receptor	Homo sapiens	139	77%	34.96%
4QKX	β-2 adrenergic receptor	Homo sapiens	137	77%	34.96%
8HN1	α-1A adrenergic receptor	Homo sapiens	146	89%	34.85%
7BTS	β-1 adrenergic receptor	Homo sapiens	144	81%	34.78%
6MXT	β-2 adrenergic receptor	Homo sapiens	137	77%	34.59%
8THK	α-1A adrenergic receptor	Homo sapiens	150	86%	34.54%
6KUY	α-2A adrenergic receptor	Homo sapiens	161	85%	34.51%
2R4S	β-2 adrenergic receptor	Homo sapiens	132	77%	34.27%
2R4R	β-2 adrenergic receptor	Homo sapiens	131	77%	34.27%
4GBR	β-2 adrenergic receptor	Homo sapiens	136	77%	34.21%
7YS6	5-hydroxytryptamine receptor 6	Homo sapiens	134	83%	34.12%
6KUX	α-2A adrenergic receptor	Homo sapiens	160	85%	34.07%
6WGT	5-hydroxytryptamine receptor 2A	Homo sapiens	134	82%	33.78%
3P0G	β-2 adrenergic receptor	Homo sapiens	147	78%	33.68%
2RH1	β-2 adrenergic receptor	Homo sapiens	147	78%	33.68%
7BZ2	β-2 adrenergic receptor	Homo sapiens	133	77%	33.46%
7DHI	β-2 adrenergic receptor	Homo sapiens	133	77%	33.46%
3SN6	β-2 adrenergic receptor	Homo sapiens	133	77%	33.45%
6NI3	β-2 adrenergic receptor	Homo sapiens	133	77%	33.45%
3D4S	β-2 adrenergic receptor	Homo sapiens	148	78%	33.16%
6PRZ	β-2 adrenergic receptor	Homo sapiens	148	77%	33.16%
3PDS	β-2 adrenergic receptor	Homo sapiens	146	77%	33.16%

, listing the putative template featuring higher similarity to hTAAR5 than the most exploited 3PDS code. The results are ordered according to the percentage of identity (% Id) (calculated on 23 November 2023), with each putative template being colored based on the receptor family as follows: TAARs in cyan, β-ADRs in white, α-ADR in orange, and 5-HT receptors in green. The most utilized PDB in the literature to develop TAAR5 models (3PDS) is highlighted in violet and reported in bold. Other alignment metrics are reported, such as the total score (the sum of alignment scores of all segments from the query sequence), the query coverage (a measure of the percentage of the query sequence that has a corresponding residue on the aligned sequence, the closer to 100% the better).

Notably, the recent release of more experimental structures more closely related to TAAR5 opens a promising scenario for the development of improved hTAAR5 HMs.

As expected, the top section of the alignment results is occupied by the available TAARs PDB structures (mTAAR9, mTAAR7f, hTAAR1, and mTAAR1). In all cases, the coverage of the sequence is over 95%. 8ITF represents the most promising template for producing an hTAAR5 HM according to the percentage identity value (46.45%). However, the sub-optimal resolution of such Cryo-EM structures must be carefully evaluated (3.46 Å). A possible alternative is the second scored result (8PM2, resolution of 2.92 Å). After the TAARs, the β₁-ADR from various organisms and in combination with various fusion proteins is reported as a promising template, with percentages of identities around 36–37%, and a query coverage of more than 80%. Immediately after, the α-_2A ADR and the 5-HT₄R are proposed, with %ids of 36.14% and 35.45%, respectively. Below this value, several isoforms of the mentioned receptors are proposed, namely the β₂-ADR, the α_1A-ADR, and 5-HT₆R, and the 5HT_2AR. At a %Id of 33.16%, it is possible to find the previously utilized 3PDS PDB. A few examples of promising templates for the modeling of hTAAR5 are represented in Figure 14A.

Concerning mTAAR5, an identical search was performed to individuate novel putative templates for ameliorated homology modelling approaches (

Table 10. BLAST-p alignment results according to the BLOSUM62 matrix, using mTAAR5 as query. The proposed template is indicated via its PDB ID, name of the macromolecule, organism, BLAST total score, coverage of the sequence, and percentage of identity (% Id.) between the query and the template. TAARs as templates are highlighted in cyan.

PDB ID	Description	Organism	Total Score	Query Coverage	% Id.
8ITF	Trace amine-associated receptor 9	Mus musculus	305	97%	46.45%
8PM2	Trace amine-associated receptor 7f	Mus musculus	310	95%	44.24%
8WC3	Trace amine-associated receptor 1	Mus musculus	258	89%	42.44%
8JLJ	Trace amine-associated receptor 1	Mus musculus	260	89%	42.44%
8W87	Trace amine-associated receptor 1	Homo sapiens	259	94%	39.27%
8JLN	Trace amine-associated receptor 1	Homo sapiens	260	94%	39.27%
8UHB	Trace amine-associated receptor 1	Homo sapiens	258	96%	38.39%
6TKO	Beta-1 adrenergic receptor	Meleagris gallopavo	158	83%	37.46%
2Y00	Beta-1 adrenergic receptor	Meleagris gallopavo	155	83%	36.90%
7JJO	Beta-1 adrenergic receptor	Meleagris gallopavo	154	83%	36.90%
6H7J	Beta-1 adrenergic receptor	Meleagris gallopavo	159	84%	36.82%
2VT4	Beta-1 adrenergic receptor	Meleagris gallopavo	158	83%	36.81%
6WGT	5-hydroxytryptamine receptor 2A	Homo sapiens	108	86%	36.70%
6IBL	Beta-1 adrenergic receptor	Meleagris gallopavo	159	86%	36.67%

).

Again, the first positions are occupied by the TAARs (mTAAR9, mTAAR7f, mTAAR1, and hTAAR1). Following the TAARs, the β₁-ADR from meleagris gallopavo is proposed. Noticeably, such a template was already utilized by Bon et al. [121] to build a mTAAR5 HM, with the last reported PDB ID (6IBL) [123]. The 5HT_2AR is also considered, with a percentage of identity of 36.70%. Given the much higher similarity and query coverage of the TAARs templates, such molecules can be considered as the best templates for future mTAAR5 HMs. Once more, the resolution must be taken into consideration.

The use of AlphaFold may also be considered for both the two m/hTAAR5 proteins (ID: Q5QD14 and O14804, respectively) [62,63]. However, in the previously cited VS campaign [121] directed towards the discovery of mTAAR5 antagonists, this possibility was excluded, as the AlphaFold-predicted target structure reported a region with high uncertainty in the proximity of the binding site. Through a comparison of the mentioned mTAAR5-predicted structure with its corresponding human orthologue (Figure 14B), it is possible to assert that this aspect may also be relevant for hTAAR5, further supporting the development of HMs for the human orthologue.

5. TAAR1/5 and Other GPCRs: A Repositioning Perspective

5.1. Comparison of hTAAR1 and Druggable GPCRs

In the last few years, several studies have highlighted the potential of repositioning as a valuable strategy for drug discovery [132]. A few attempts involving TAAR1 were also proposed, such as the previously cited 5HT_1AR and/or α₁-ADR ligands repositioning [33]. Here, we propose and compare some possible druggable targets to develop the repurposing strategy for the search of novel hTAAR1 ligands. This comparison has been conducted based on the percentage of identity (% Id) between the putative template and the hTAAR1 protein, choosing % id. values > 30% as the cut-off for the analysis. The search has been performed referring to proteins whose structural information is experimentally available (

Table 11. BLAST-p alignment results for the hTAAR1 sequence considering only human GPCRs according to the BLOSUM62 matrix. The proposed reference GPCR is indicated via its PDB ID, name of the macromolecule, BLAST total score, coverage of the sequenceand percentage of identity (% Id.) between the query and the template.

PDB ID	Description	Total Score	Query Cover	% Id.
8W87	Trace amine-associated receptor 1	697	100%	100.00%
7WC4	5-hydroxytryptamine receptor 2A	67.8	23%	44.58%
7VOD	5-hydroxytryptamine receptor 2A	67.8	23%	44.58%
7XT8	5-hydroxytryptamine receptor 4	199	86%	37.42%
7CKY	D(1A) dopamine receptor	158	82%	33.55%
7CKX	D(1A) dopamine receptor	158	82%	33.55%
7CKW	D(1A) dopamine receptor	158	82%	33.55%
7F0T	D(1A) dopamine receptor	158	82%	33.55%
7JV5	D(1A) dopamine receptor	157	82%	33.55%
7B6W	Alpha-1B adrenergic receptor	154	87%	32.78%
4GBR	Beta-2 adrenergic receptor	158	85%	32.65%
7C61	5-hydroxytryptamine receptor 1B	174	85%	32.44%
7BZ2	Beta-2 adrenergic receptor	167	85%	31.97%
7DHI	Beta-2 adrenergic receptor	167	85%	31.97%
7XTC	5-hydroxytryptamine receptor 7	138	82%	31.75%
6KR8	Beta-2 adrenergic receptor	170	86%	31.61%
2R4S	Beta-2 adrenergic receptor	164	85%	31.60%
2R4R	Beta-2 adrenergic receptor	163	85%	31.60%
3KJ6	Beta-2 adrenergic receptor	162	85%	31.60%
5V54	5-hydroxytryptamine receptor 1B	175	85%	31.58%
7YM8	alpha1A adrenergic receptor	165	78%	31.58%
7RAN	5-hydroxytryptamine receptor 2A	125	88%	31.29%
6LUQ	Chimera of D(2) dopamine receptor and Endolysin	169	84%	31.16%
7UL3	Histamine H2 receptor	139	89%	31.05%
6K42	Alpha-2A adrenergic receptor	145	84%	31.03%
7EJ0	Alpha-2A adrenergic receptor	167	88%	30.77%
5D5A	Beta-2 adrenergic receptor	175	78%	30.77%
7YS6	5-hydroxytryptamine receptor 6	144	85%	30.58%
8E9W	Muscarinic acetylcholine receptor M3	132	86%	30.57%
8E9Z	Muscarinic acetylcholine receptor M3	131	86%	30.57%
8E9Y	Muscarinic acetylcholine receptor M3	131	86%	30.57%
6G79	5-hydroxytryptamine receptor 1B	151	85%	30.43%
8JLZ	5-hydroxytryptamine receptor 6	147	85%	30.42%
7XTB	5-hydroxytryptamine receptor 6	146	85%	30.42%
5CXV	Muscarinic acetylcholine receptor M1	140	76%	30.27%
6KUW	Alpha-2C adrenergic receptor	156	85%	30.26%
7YMJ	Alpha-1A adrenergic receptor	127	87%	30.00%

). Bearing in mind the previously attempted repurposing methods [54,133,134], we selected the reference proteins shown in Table 11 (in green) in detail: the 5-HT_2AR (PDB ID: 7WC4, % id: 44.58%, query coverage: 23%) [135], the D(1A) dopamine receptor (PDB ID: 7CKY, %id: 33.55%, query coverage: 82%) [136], the α_1B-ADR (PDB ID: 7B6W, %id: 32.78%, query coverage: 87%) [137], the β₂-ADR (PDB ID: 4GBR, %id: 32.65%, query coverage: 85%) [124], the α_1A-ADR (PDB ID: 7YM8, %id: 31.58%, query coverage: 78%) [138], the Histamine H₂R (PDB ID: 7UL3, %id: 31.05%, query coverage: 89%) [139], the 5-HT₆R (PDB ID: 7YS6, %id: 30.58%, query coverage: 85%) [140], and the Muscarinic acetylcholine receptor M3R (PDB ID: 8E9W, %id: 30.57% query coverage: 86%) [141].

The mentioned PDBs were superimposed to the hTAAR1 structure (PDB ID: 8W8A) [87]: the results are presented in Figure 15A–H and point out a comparable overall fold of the two receptors. A larger variability is observed for the extracellular region and H8 (except for 5HT6).

Regarding the TM helices, TM1 exhibits a certain shift with respect to the hTAAR1 structure, especially for the adrenergic, 5-HT₆R, and M₃R. Minor shifts were observed for TM7 and TM2. Among the analyzed receptors, the dopamine receptor D1 (DRD1), the α_1A-adrenergic one (α_1A-ADR), and the muscarinic M₃R exhibit the lowest Cα RMSD with respect to hTAAR1. On the contrary, 5-HT_2AR represents the highest Cα RMSD structure.

Apart from the overall conformation similarity, the residue conservation at the orthosteric pocket was investigated. The hTAAR1 binding site was defined at 5 Å from the reported ligand (PDB ID: 8W8A), including residues T100, I104, V184, F186, D103, R83, V76, M77, S297, L72, Y294, W291, G293, I290, W264, F267, S107, S108, T271, F268, S198, and T194. These residues were compared with the corresponding amino acids at the candidate protein. In addition, residues H99, S80, and S190 were also considered, as they were indicated as key residues for the TAAR binding site [90]. The obtained (non-)conserved residues between hTAAR1 and the proposed reference protein are listed in

Table 12. Conservation (†) of the binding-site residues of a set of hGPCR with respect to hTAAR1 are indicated in yellow. Otherwise, the mutated amino acids are listed. The dopamine receptor D₁R, the α_1a-ADR, α_1b-ADR, the β₂-ADR, the hystidine receptor type 2 (H2R), and muscarinic one type 3 have been reported.

Reference Proteins	hTAAR1 Binding Site
	T100	I104	V184	F186	D103	R83	V76	M77	S297	L72	Y294	W291	G293	I290	W264	F267	S107	S108	T271	F268	S198	T194	H99	S80	S190
5-HT2AR	I	V	L	D	†	T	†	†	†	†	†	†	†	V	†	†	†	T	N	†	†	G	W	†	†
D1R	V	†	S	L	†	A	†	†	†	V	†	†	†	V	†	†	†	T	N	†	†	S	W	K	†
α1B-ADR	A	V	†	E	†	L	†	L	†	†	†	†	†	L	†	†	C	T	L	†	†	S	W	†	Y
β2-ADR	†	V	F	T	†	H	†	V	†	M	†	†	†	N	†	†	V	T	N	†	†	S	W	G	Y
α1A-ADR	A	V	I	E	†	F	†	L	†	†	†	†	†	F	†	†	C	T	M	†	†	S	W	†	Y
H2R	†	V	†	V	†	Y	†	L	†	†	†	†	†	L	†	Y	C	T	F	†	T	D	Y	†	†
5-HT6R	†	V	L	A	†	N	†	†	†	V	†	†	†	T	†	†	C	†	N	†	T	A	W	A	F
M3R	L	C	I	L	†	Y	S	†	Y	I	†	†	†	Y	†	Y	†	N	V	N	G	†	W	F	I

.

The analysis of the 5-HT_2AR receptor highlighted a rather good conservation rate in the binding site, with few non-conserved residues, such as F186/D231, V184/L229, T100/I152, H99/W151, R83/T134, I290/V366, S108/T160, T271/N343, T194/G238, and I104/V156 (Figure 16A).

However, most of these structural variations maintain comparable polarity and steric properties with respect to the hTAAR1 residues. On the other hand, several sidechains of conserved residues significantly differ in their conformations, especially the two phenylalanine residues (F199,TAAR1/F243,5HT_2AR, F268,TAAR1/F340,5HT_2AR). A similar situation can be observed for the D₁R (Figure 16B): T271 is replaced by a N residue, S108 to a T, I290 to a valine, and H99 to a W. Other substitutions are F186/L190, L72/V73, R83/A84, T100/V100, T194/S198, Y294/W321, and S80/K81. Intriguingly, K81′s introduction in place of S81 re-introduces a positive charge in the area previously occupied by R83. In the case of α_1B-ADR, we can observe the same substitution pattern of residues S108(T), H99(W), and T194(S) (Figure 16C) previously mentioned for 5HT_2AR and D₁R receptors. Other non-conserved residues include the hTAAR1 R83 which is substituted with a leucine residue and the following ones: T100/A122, I104/V126, F186/E199, T271/L314, and I290/L334. Additionally, M77 is mutated to L99, S107 to C129, and S190 to Y203, introducing a phenyl moiety partially invading the binding pocket. The β₂-ADR exhibits a larger number of differences in the binding site residues if compared with hTAAR1, such as S107/V117, M77/V87, S80/G90, R83/H93, T271/N266, V184/F193, G186/T195, L72/M82, and I290/N284 (Figure 16D). While most of the polarity properties are maintained, the corresponding residue dimensions increase in the β₂-ADR with respect to the related hTAAR1 amino acid. Further mutated residues such as S190(Y), T194(S), S108(T), and I104(V) are in accordance with the reported residues featured by the other reference GPCRs, D₁R and α_1B-ADR.

In the case of α_1A-ADR, it is possible to observe a large number of substitutions, often coherent with the previously analyzed cases (S190/Y184, T100/A104, M77/L80, S108/T111, H99/W102, I104/V107, T194/S188) (Figure 17A).

Differences can be observed for residues R83 (substituted with a F residue), I290 (changed to F), S107(C), V184(I), T271(M), and F186(E). The HRH2 receptor exhibits a lower number of previously observed substitutions (S190/Y182, M77/L72, S107/C102, I104/V99, S108/T103), whereas several novel residues are introduced (Figure 17B). H99, in this case, is replaced by a Y residue (Y94), I290 is replaced by L274, T194 by D186, and T271 by F254. R83 is again substituted by an aromatic residue (Y78). In addition, S198 is replaced by a threonine residue.

The 5-HT6R exhibits a higher similarity of the binding site to hTAAR1 with respect to β₂-ADR, α_1A-ADR, and H₂R (Figure 17C). Several residues are substituted with similar amino acids (E.g., I104/V107, S107/C110, S198/T196, V184/L182). Again, it is possible to observe the H99/W102 replacement, as well as the aromatic substitution of S190 (to F188). Other replacements are S80/A83, R83/N86, F186/A184, T194/A192, T271/N288, I290/T306, and L72/V75. The M₃R exhibits poor residue conservation at the orthosteric site, with respect to hTAAR1. Apart from previously observed substitutions (R83(Y), F267(Y), H99(W), F186/L, and V184(I), newly introduced replacements are highlighted (S190/I232, T100/L145, I104/C149, I290/Y530, F268/N508, S198/G239, L72/I117, S80/F125, T271/V511, S108/N153, F267/Y507). Consequently, this receptor would be difficult to target in a repositioning perspective, whereas 5HT_2AR, D1R, α_1B-ADR, and 5-HT₆R exhibit a higher potential in this regard.

5.2. Comparison of hTAAR5 and Druggable GPCRs

Regarding TAAR5, the lack of structural information does not allow for the identification of the protein binding site and to perform a three-dimensional comparison with putative reference GPCRs. However, it is possible to perform a sequence alignment between hTAAR1 and hTAAR5, to individuate the putative key residues of the hTAAR5 binding site. Sequence alignment was performed with T-Coffee [142,143], using the PSI-BLAST algorithm [144,145]. As shown in Figure 18, the hTAAR5 binding site may be constituted by residues L83, V87, L88, S91, R94, H110, T111, D114, T115, C118, L119, L194, L196, W200, N204, L207, W265, F268, T269, T272, I291, W292, A294, Y295, and S298.

Out of the twenty-five considered residues, the majority are conserved with respect to hTAAR1. Only residues M77(L), I104(T), S107(C), S108(L), V184(L), F186(L), S190(W), T194(N), S198(L), F268(T), and G293(A) are changed. The reported analysis is coherent with the one reported by Xu [90]. The individuated residues also constitute the binding site in the AlphaFold-modelled hTAAR5 (ID: O14804) [62,63].

Based on the above, it is possible to compare the proposed hTAAR5 binding residues with those of reference GPCRs, paving the way for future drug repositioning strategies. In this case, we performed a search for the most closely related receptors taking hTAAR5 sequence as a reference. The search was performed considering sequences associated with a PDB entry, to allow structure-based drug design. The alignment was performed with T-Coffee PSI-TM algorithm, with the slow/accurate option. The obtained (non-)conserved residues between hTAAR5 and the proposed reference protein are listed in

Table 13. BLAST-p alignment results to the hTAAR5 sequence considering only human GPCRs according to the BLOSUM62 matrix. The proposed reference GPCR is indicated via its PDB ID, name of the macromolecule, BLAST total score, coverage of the sequence and percentage of identity (% Id.) between the query and the template.

PDB ID	Description	Total Score	Query Cover	% Id.
8W87	Trace amine-associated receptor 1	251	95%	38.74%
7EJ0	Alpha-2A adrenergic receptor	160	82%	36.14%
7XT8	5-hydroxytryptamine receptor 4	157	80%	35.45%
6KUY	Alpha-2A adrenergic receptor	161	85%	34.51%
2R4S	Beta-2 adrenergic receptor	132	77%	34.27%
2R4R	Beta-2 adrenergic receptor	131	77%	34.27%
4GBR	Beta-2 adrenergic receptor	136	77%	34.21%
7YS6	5-hydroxytryptamine receptor 6	134	83%	34.12%
6KUX	Alpha-2A adrenergic receptor	160	85%	34.07%
5D5A	Beta-2 adrenergic receptor	147	78%	33.68%
7BZ2	Beta-2 adrenergic receptor	133	77%	33.46%
7DHI	Beta-2 adrenergic receptor	133	77%	33.46%
3KJ6	Beta-2 adrenergic receptor	132	77%	33.10%
6KR8	Beta-2 adrenergic receptor	136	82%	32.46%
7SRQ	5-hydroxytryptamine receptor 2B	114	78%	32.26%
7SRS	5-hydroxytryptamine receptor 2B	112	78%	32.26%
8JLZ	5-hydroxytryptamine receptor 6	128	83%	32.17%
7XTB	5-hydroxytryptamine receptor 6	128	83%	32.17%
6K42	Alpha-2B adrenergic receptor	133	81%	31.97%
7B6W	Alpha-1B adrenergic receptor	148	91%	31.78%
7UL3	Histamine H2 receptor	108	83%	31.58%
7YMJ	Alpha-1A adrenergic receptor	123	89%	31.27%
8HDO	Adenosine A2b receptor	105	81%	31.14%
7YM8	Alpha-1A adrenergic receptor	164	86%	30.77%
6KUW	Alpha-2C adrenergic receptor	146	82%	30.60%
7C61	5-hydroxytryptamine receptor 1B	155	80%	30.33%
6LUQ	Chimera of D(2) dopamine receptor and Endolysin	150	79%	30.11%
8JSP	5-hydroxytryptamine receptor 1A	149	79%	30.00%

.

Among the results, we selected eight hGPCRs of interest (

Table 14. Conservation (†) of the binding-site residues of a set of hGPCR with respect to hTAAR5 is reported in yellow. Otherwise, the mutated amino acids are listed.

Reference Proteins	hTAAR5 Binding Site
	L83	V87	L88	S91	R94	H110	T111	D114	T115	C118	L119	L194	L196	W200	N204	L207	W265	F268	T269	T272	I291	W292	A294	Y295	S298
α2A-ADR	V	†	I	†	N	Y	L	†	V	†	T	I	gap	†	S	S	†	†	F	Y	F	†	G	†	†
β2-ADR	M	†	V	G	H	W	†	†	V	V	T	F	T	Y	S	S	†	†	F	N	N	†	G	†	†
5-HT6R	V	†	M	A	N	W	†	†	V	†	S	†	A	F	S	T	†	†	F	N	T	†	G	†	†
H2R	†	†	†	†	Y	Y	†	†	V	†	T	V	V	Y	D	T	†	Y	F	F	L	†	G	†	†
α1A-ADR	†	†	†	†	F	W	A	†	V	†	T	I	E	Y	A	S	†	†	F	M	F	†	G	†	†
5-HT1BR	V	†	M	†	Y	W	L	†	I	†	T	I	Y	gap	T	A	†	†	F	S	T	†	G	†	†
5-HT1AR	V	†	†	A	Y	F	I	†	V	†	T	I	K	Y	T	A	†	†	F	A	N	†	G	†	†
D1R	V	†	M	K	A	W	V	†	I	S	T	S	†	Y	S	S	†	†	F	N	V	†	G	W	†

, green), as these have been extensively studied as druggable targets in medicinal chemistry [146,147,148,149,150,151,152,153], exhibiting identity percentage values (% Id.) > 30% with respect to hTAAR5: the α_2A-ADR (% Id: 36.14%), the β₂-ADR (% Id: 34.27%), the 5-HT₆ receptor (% Id: 34.12%), the H₂ receptor (% Id: 31.58%), the α_1A-ADR (% Id: 31.27%), the 5-HT_1B (% Id: 30.33%), the 5-HT_1A (% Id: 30.00%), and the D(1A) dopamine receptor (% Id: 29.09%).

All the entries presented an acceptable query coverage (over 75%). The sequence alignment of the best ranked couple of hTAAR5 and reference GPCRs are reported in Figure 19 and Figure 20 [154]. One kind of adrenergic and serotoninergic subfamily of receptors have been reported, as representative of the corresponding GPCR family.

As shown in Figure 19A, by analyzing the α_2A-ADR, it is possible to notice a good conservation rate with respect to the hTAAR5 binding site.

Residues V87, S91, D114, C118, W200, W265, F268, W292, Y295, and S298 are conserved between the candidate protein and TAAR5. Some other residues exhibit substitution with the same type of amino acids: L83, L88, L194, and A294 are substituted with other hydrophobic residues, H110 with another aromatic residue, and N204 to the polar amino acid serine. Other replacements introduce larger differences, such as hindrance (T269F, T272Y, I291F), different polarity (T111 to L, T115 to V, L119 to T, L207 to S), or protonation state (R94 to N). The sequence alignment did not assign a specific residue corresponding to L196 due to the introduction of a gap.

The case of β₂-ADR is almost like the α_2A-ADR one (Figure 19B), with several residues that are still conserved (V87, T111, D114, W265, F268, W292, Y295, and S298). Among the non-conserved amino acids, aromatic residues are changed to other aromatics (H110 to W, W200 to Y). Concerning the hydrophobic residues, they are substituted with amino acids of the same type (L83 to M, L88 to V, A294 to G) or with polar amino acids (L119 to T, L196 to T, L207 to S, I291 to N). Conversely, some polar amino acids are changed to hydrophobic (T115 to V, S91 to G, C118 to V), whereas some others retain the polar feature (N204/S, T272/N). The basic R84 is changed to one H residue. Finally, a significant hindrance is introduced as the L194 is changed to F, and in the case of T269, again replaced with a F residue.

Concerning the 5-HT₆R, residues V87, T111, D114, C118, L194, W265, F268, W292, Y295, and S298 are predicted as conserved (Figure 20A).

Again, aromatic residues are substituted with other aromatics (H110 to W, W200 to F). The hydrophobic features of residues L83, L88, L196, and A294 are retained in the 5-HT₆R (they were substituted by V, M, A, and G, respectively). On the contrary, L119, L207, and I291 are replaced by polar residues (S, T, and T). N204 and T272 are replaced with other polar residues (S and N), while S91 and T115 are changed to non-polar residues (A and V, respectively). The basic R94 is replaced by an N residue. T269 is instead replaced by an F residue, introducing aromaticity and hydrophobicity. Several residues of the H₂R binding site are predicted to be conserved with respect to hTAAR5 (L83, V87, L88, S91, T111, D114, C118, W265, W292, Y295, S298) (Figure 20B). Moreover, most of the introduced changes are more or less compatible with the reference residues of TAAR5. The aromatic residues H110, W200, and F268 are substituted by other aromatic residues (W, Y, and Y, respectively). Many hydrophobic residues are replaced with other hydrophobic amino acid of comparable dimension (L194 to V, L196 to V, I291 to L, and A 294 to G). A shift from hydrophobic to polar is observed for L207(T), and L119(T). Conversely, polar residues T269 and T272 are both changed to a F, and T115 to a V. Polarity is instead retained in the cases of N204(D), while the basic R94 is substituted with a Y residue. Although it is an aromatic residue, the Y still maintains the hydrogen-bond donor moiety due to its R residue, representing an advantage for repurposing.

In the case of the α_1A-ADR, the conserved residues are L83, V87, L88, S91, D114, C118, W265, F268, W292, Y295, and S298. The aromaticity is preserved for residues H110 (W) and W200 (Y), and it was introduced at the T269 (F), I291 (F), and R94 (F). Regarding the hydrophobic residues, they are substituted by residues of the same type in the cases of L194(I), and A294(G), while they are substituted with polar residues in the cases of L119(T), L196(E), and L207(S). Conversely, polar residues T111(A), T115(V), N204(A), and T272(M) are changed to non-polar amino acids.

Regarding the 5-HT_1BR, the number of conserved residues in the binding site decreases to nine (V87, S91, D114, C118, W265, F268, W292, Y295, and S298). Among the non-conserved residues, five hydrophobic amino acids are maintained as hydrophobic (L83 to V, L88 to M, L194 to I, L207 to A, A294 to G), whereas two of them are converted to polar residues (L119 and I291, both converted to T). Aromatic residue H110 is replaced by a tryptophan. Aromatic residues are introduced in place of R94 (changed to Y, again retaining the HBD feature), L196(Y), and T269 (F). Polar residues T111 and T115 are changed to non-polar residues L and I, while polarity is maintained in the case of residues N204(T) and T272(S). A gap is introduced in place of residue W200. In the case of 5-HT_1AR, the situation is very similar. A few differences can, however, be observed, such as the conservation of residue L88. On the contrary, the previously conserved residue S91 is now non-conserved (changed to A). H110 is again changed to another aromatic residue (F), while W200 is changed to Y. Polar residues T111 and T115 are again substituted by other hydrophobic residues (I and V, respectively), whereas T272 is changed to an A. Finally, non-polar residues L196 and I291 are changed to K, and N, respectively. All the other residues are the same as in the previous case. The D₁R exhibits a very low conservation rate in the binding site residues, as only 7/25 are conserved (V87, D114, L196, W265, F268, W292, S298), according to our analysis. The majority of the replacements had already been observed for other receptors among the considered ones (L83 to V, L88 to M, H110 to W, T115 to I, L119 to T, W200 to Y, N204 to S, L 207 to S, T269 to F, T272 to N, A294 to G). In the cases of R94 (A), T111(V), C118(S), L194(S), I291 (V), and Y295 (W), new residues were introduced.

In summary, receptors H₂R and α_1A-ADR exhibit the highest rate of conserved residues within the binding site with respect to hTAAR5 (11/24). However, H₂R might represent a better starting point than α_1A-ADR, as the type of introduced residues exhibits a closer resemblance to the original residues found at TAAR5. On the contrary, in α_1A-ADR, more polarity variations are present. Moreover, the substitution of R94 with a Y residue (H₂R) might be more compatible with the original R residue with respect to the F substitution (α_1A-ADR). In addition to these two GPCRs, α_2A-ADR, 5-HT₆R and 5-HT₁R can also be considered. The conservation rate is in these cases of 10/25, 10/25, and 9/24, respectively. The worst predicted match is between the target and D₁R, which have a conservation rate of only 7/24. Residues V87, D114, W265, and W292 are conserved in all the considered receptors. Y295 is conserved in any of the analyzed cases excluding D₁R. Conversely, some residues are always mutated (R94, H110, always mutated to other aromatics, T115, L119, L194, L196, N204, L207, T269, consistently mutated to a F, T272, I291, A294, always mutated to G).

6. Conclusions

The present review collects and analyzes the applications of computer-aided drug design tools leading to the discovery of novel ligands targeting h/mTAAR1 and/or h/mTAAR5. Additionally, when available, the SARs of the discovered compounds are reported. Most of the studies were performed trying to predict the three-dimensional structure of the target by means of homology modeling techniques. Alternatively, the AlphaFold-predicted structures were also utilized. In most of the cases, the target was modeled on the inactive conformation of template GPCRs, such as the β₂-ADR, or referring to AlphaFold models which were in an inactive-like conformation.

As the h/mTAAR1 experimental structures were solved, we were able to compare the structural information utilized for TAAR1 modeling with the novel structural data. In addition to this retrospective study, we also reported some information of interest for SBDD towards h/mTAAR1, comparing the novel m/hTAAR1 Cryo-EM structures and the related mutagenesis data, and reporting a possible elucidation of the species-specificity issue from a structural perspective [90].

To date, no structure has been solved for TAAR5 and no structural comparison with the previously exploited homology modeling templates has been possible. To improve the quality of HMs of this target, we proposed novel protein templates according to the coupling of protein identity. Along with this, the choice of the TAAR5 AlphaFold alternative has been reported. We have also proposed possible reference targets to guide future drug repurposing for both hTAAR1 and hTAAR5. Regarding hTAAR1, the binding site conservation was discussed based on the structure superposition of the hTAAR1 experimental data and the three-dimensional structure of the selected templates. With experimental information on hTAAR5 still lacking, protein sequence alignment approaches have been applied, referring to proper template GPCRs. The results are expected to give new hints for the future design of novel, more effective, TAAR1/5 ligands.