Trace Amines and Their Receptors

A. β-Phenylethylamine, p-Tyramine, and Related Compounds

1. Synthesis

The archetypal trace amines are synthesized after initial decarboxylation of precursor amino acids (Fig. 4). This pathway is directly analogous to the synthesis of the monoamine neurotransmitters, with the trace amines being the end product if the tyrosine hydroxylase (EC 1.14.16.2) or tryptophan hydroxylase (EC 1.14.16.4) steps of neurotransmitter synthesis are omitted. As such, PEA, TYR, and TRP can be formed directly by the action of aromatic l-amino acid decarboxylase (AADC; EC 4.1.1.28) on l-phenylalanine, l-tyrosine, and l-tryptophan, respectively (Boulton and Wu, 1972, 1973; Saavedra, 1974; Snodgrass and Iversen, 1974; Silkaitis and Mosnaim, 1976; Dyck et al., 1983). Both m- and o-isoforms of tyramine have also been identified but have rarely been studied, and they are present in even smaller quantities than the p-isoforms (Boulton, 1976; Davis, 1989). OCT and p-synephrine can subsequently be formed by the sequential action of dopamine-β-hydroxylase (EC 1.14.17.1) (Boulton and Wu, 1972, 1973) and phenylethanolamine-N-methyl transferase (PNMT; EC 2.1.1.28). The trace amines can also undergo N-methylation by action of the enzymes PNMT or indolethylamine N-methyltransferase (EC 2.1.1.49) to generate additional TAAR ligands, N-methylphenylethylamine, N-methyltyramine, N-methyltryptamine, and, at least in some species, DMT (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Endogenous synthetic and metabolic routes for trace amines. DBH, dopamine-β-hydroxylase; INMT, indolethylamine N-methyltransferase; PAH, phenylalanine hydroxylase; TH, tyrosine hydroxylase.

The similarity to monoamine neurotransmitter synthesis has often led to the synthesis of trace amines being reported as neuronal. However, it should be borne in mind that AADC expression is not restricted to neuronal cells. AADC is present in a number of other cell types, including glia (Li et al., 1992b; Juorio et al., 1993), blood vessels (Li et al., 2014), and cells of the gastrointestinal tract (Lauweryns and Van Ranst, 1988; Vieira-Coelho and Soares-da-Silva, 1993), kidney (Christenson et al., 1970; Lancaster and Sourkes, 1972; Aperia et al., 1990; Hayashi et al., 1990), liver (Bouchard and Roberge, 1979; Ando-Yamamoto et al., 1987; Dominici et al., 1987), lungs (Lauweryns and Van Ranst, 1988; Linnoila et al., 1993), pancreas (Lindström and Sehlin, 1983; Furuzawa et al., 1994; Rorsman et al., 1995), and stomach (Lichtenberger et al., 1982). In such cells, it can reasonably be expected that AADC will convert any precursor amino acids present into the corresponding trace amine(s). The physiologic function of AADC in non-neuronal tissue is generally poorly understood but it does provide a mechanism for the local production of ligands for TAARs that are localized to non-neuronal tissue, and investigation of possible colocalization of AADC with TAARs is an area for future studies. Furthermore, a distinct group of neurons that contain AADC, but not tyrosine hydroxylase or serotonin, are present in the mammalian central nervous system (Jaeger et al., 1983, 1984; Fetissov et al., 2009; Kitahama et al., 2009). These D-neurons offer a potential trace aminergic neuronal system.

Although AADC is widely accepted as the vertebrate synthetic enzyme for PEA, TYR, and TRP, the precursor amino acids are in fact rather poor substrates for AADC. Indeed, the K_m values for decarboxylation of l-phenylalanine, l-tyrosine, and l-tryptophan approach the limits of solubility of each in aqueous media (Christenson et al., 1970; Juorio and Yu, 1985a). Although this is markedly, and selectively, improved both in vitro and in vivo by the presence of organic solvents (Lovenberg et al., 1962; Juorio and Yu, 1985a,b), this does raise questions about how PEA, TYR, and TRP are being synthesized in vivo. Substrate-selective regulation of AADC has been reported (Bender and Coulson, 1972; Sims and Bloom, 1973; Sims et al., 1973; Rahman et al., 1981; Siow and Dakshinamurti, 1985), along with a number of splice variants of the enzyme (O’Malley et al., 1995; Rorsman et al., 1995; Chang et al., 1996; Vassilacopoulou et al., 2004). Whether one or more of these exhibits enhanced selectivity for the production of PEA, TYR, and/or TRP requires systematic investigation. Furthermore, l-phenylalanine, l-tyrosine, and l-tryptophan are all also substrates for additional amino acid decarboxylase enzymes (Table 1), although the role of these putative additional sources of PEA, TYR, and TRP synthesis has not yet been investigated.

View this table:

• View inline
• View popup

TABLE 1

Decarboxylated products of common amino acids, their known TAAR ligand status, and putative physiologic effects

Putative enzymes were obtained from the Brenda-enzymes.org database. Only those known or predicted to be present in eukaryotes are shown. See the text for citations, with the exception of entries with footnotes.

a. Regulation of Aromatic l-Amino Acid Decarboxylase

As described above, AADC is found in both neuronal and non-neuronal cells and alternative splicing of exons 1 and 2 within the 5′-untranslated region has been established as allowing for distinct control of neuronal and non-neuronal expression (Albert et al., 1992; Ichinose et al., 1992; Hahn et al., 1993; Sumi-Ichinose et al., 1995). A variety of transcription factor binding sites have been identified within both the neuronal (Chireux et al., 1994; Aguanno et al., 1995) and non-neuronal (Aguanno et al., 1996) promoter regions through which tissue-selective expression could occur. A variant on this alternative splicing, in which the non-neuronal variant was spliced to the neuronal acceptor site, has also been suggested in G cells of the rat stomach antrum (Djali et al., 1998), which may indicate cell type–selective plasticity in the control of AADC expression. Recently, a number of cis-acting polymorphisms of AADC have been identified with putative clinical relevance (Li and Meltzer, 2014; Eisenberg et al., 2016), and disease-associated AADC coding variants are also known (Graziano et al., 2015; Kojima et al., 2016; Montioli et al., 2016).

The functional significance of alternative splicing within the coding region of AADC is poorly defined. A splice variant lacking exon 3 has been reported to be widely expressed in both neuronal and non-neuronal tissue (O’Malley et al., 1995; Chang et al., 1996), with this shorter variant suggested to be devoid of both l-DOPA and l-5-hydroxytryptophan decarboxylating activities (O’Malley et al., 1995). An even shorter variant, lacking exons 11–15, has also been reported to be expressed in non-neuronal tissues (Vassilacopoulou et al., 2004), although enzyme activity of this variant was not examined. Additional coding splice variants were also reported to be present in pancreatic β cells (Rorsman et al., 1995), although again the functionality of these putative alternative forms does not appear to have been further investigated. Whether activity toward substrates other than l-DOPA and l-5-hydroxytryptophan is lost, or even enhanced, is unknown, but the apparent widespread expression of an ostensibly nonfunctional variant seems unlikely, and a role of one or more splice variants in selective trace amine synthesis could provide an answer to this paradox.

Although it is not a rate-limiting step in the synthesis of catecholamine and indoleamine neurotransmitters under physiologic conditions, AADC activity is regulated in a biphasic manner. Both early changes, consistent with alterations in phosphorylation status, and delayed, longer-lasting changes in enzyme expression have been reported (Buckland et al., 1992, 1996, 1997; Hadjiconstantinou et al., 1993; Berry et al., 1996). Multiple consensus phosphorylation sites for protein kinase A (PKA) (Young et al., 1993; Duchemin et al., 2000), protein kinase C (PKC) (Young et al., 1994; Zhu et al., 1994), protein kinase G (Duchemin et al., 2010), calmodulin-dependent kinase II (Hadjiconstantinou et al., 2010), and proline-directed kinase (Hadjiconstantinou et al., 2010) are present, and both site-directed mutagenesis of individual phosphorylation sites (Hadjiconstantinou et al., 2010) or selective activation/inhibition of individual protein kinases (Young et al., 1993, 1994; Zhu et al., 1994; Duchemin et al., 2000, 2010) alters AADC activity. Direct evidence for phosphorylation of AADC by PKA (Duchemin et al., 2000) and protein kinase G (Duchemin et al., 2010) has been provided, although PKC does not appear to directly increase phosphorylation despite the presence of consensus recognition sites (Duchemin et al., 2000).

In the retina, AADC activity is increased in response to light (Hadjiconstantinou et al., 1988) or selective antagonism of α₂-adrenergic receptors (Rossetti et al., 1989) or D₁-like dopamine receptors (D1Rs) (Rossetti et al., 1990). Consistent with this, light stress was recently reported to increase retinal PEA levels (de la Barca et al., 2017). In contrast, D1R agonists decrease both basal and light-induced AADC activity (Rossetti et al., 1990). Similar responses to dopamine receptor ligands have also been observed in various rodent brain regions (Zhu et al., 1992, 1993, 1994; Hadjiconstantinou et al., 1993; Cho et al., 1997, 1999; Neff et al., 2006). Regulation of AADC by serotonergic receptors (Neff et al., 2006) and N-methyl-d-aspartate (NMDA) glutamatergic receptors (Hadjiconstantinou et al., 1995; Fisher et al., 1998) has also been reported. There is also some evidence for AADC regulation associated with systemic lupus erythematosus (Bengtsson et al., 2016), Parkinson’ disease (Gjedde et al., 1993), and schizophrenia (Reith et al., 1994). In experimental animals, regulation of AADC after spinal cord injury has also been reported (Li et al., 2014; Wienecke et al., 2014; Azam et al., 2015).

In each of the above cases, the reported changes in AADC activity are normally rather modest (approximately 30%) and insufficient to change endogenous dopamine levels (Berry et al., 1994; Cho et al., 1999). Such treatments have, however, been shown to change both PEA and TYR levels (Juorio, 1979; Juorio et al., 1991; Berry, 2004) and can generally be summarized as follows: treatments that increase monoamine neurotransmitter receptor activation decrease PEA/TYR synthesis, whereas treatments that decrease receptor activation increase PEA/TYR synthesis. In this respect, it is important to note that reports of changes in AADC activity have almost exclusively used only l-DOPA as a substrate. Whether greater changes in the activity of AADC toward other substrates (particularly l-phenylalanine, l-tyrosine, and/or l-tryptophan) occurs is unknown, although as described above there is evidence that substrate-dependent regulation of AADC is possible (Rahman et al., 1981; Juorio and Yu, 1985a,b; Siow and Dakshinamurti, 1985). Kinetic studies have reported alterations in AADC V_max in response to pharmacological agents (Zhu et al., 1992; Young et al., 1994; Duchemin et al., 2010) and both V_max and K_m in response to site-directed mutagenesis of consensus phosphorylation sites (Hadjiconstantinou et al., 2010).

b. Other Sources of β-Phenylethylamine, p-Tyramine, and Related Compounds

i. Microbiota-Derived Trace Amines

There is a growing recognition of the role of the commensal microbiota in health and disease, including neurologic and psychiatric disorders (Dinan and Cryan, 2015; Fung et al., 2017), cancer and its chemotherapy (Alexander et al., 2017; Roy and Trinchieri, 2017), metabolic disorders (Sonnenburg and Bäckhed, 2016; Brunkwall and Orho-Melander, 2017), and immune disorders (Honda and Littman, 2016; Thaiss et al., 2016; Fung et al., 2017; Luo et al., 2017). Prokaryotes contain a large array of decarboxylase enzymes, many of which include l-amino acids in their substrate profile (Zheng et al., 2011; Nelson et al., 2015). Indeed, the production of PEA, TYR, and TRP by commensal prokaryotes has been established (Marcobal et al., 2006; Irsfeld et al., 2013; Williams et al., 2014; Yang et al., 2015a), and bacterial production of these compounds was the original basis of Nencki’s studies on putrefaction and fermentation (see Grandy, 2007). With TAARs established to be present throughout the body, it is expected that the role of trace amines and their receptors in mediating host-microbiota interactions will become a growing area of interest.

ii. Food-Derived Trace Amines

PEA, TYR, and TRP, as well as other endogenous trace amines, have long been recognized as being abundant in commonly ingested foods (Coutts et al., 1986). Although this presence is often viewed from the perspective of bacterial-induced food spoilage (Naila et al., 2010), foods in which anaerobic fermentation is part of their production are enriched in trace amines (Toro-Funes et al., 2015; Gardini et al., 2016; Pessione and Cirrincione, 2016). Aged cheeses, fermented meats, red wine, soy products, and chocolate are well established as being enriched in one or more of PEA, TYR, and TRP (Coutts et al., 1986).

Another rich source of food-derived trace amines is seafood, from molluscs and crustaceans to fish, where high concentrations of agmatine, cadaverine, OCT, PEA, putrescine, spermidine, spermine, TRP, and TYR are found (An et al., 2015; Biji et al., 2016). In fact, the name octopamine originates from octopus from where it was first isolated (Erspamer, 1952). Although under normal conditions the concentrations of food-derived trace amines in the body are unlikely to reach sufficient levels to activate TAARs other than those expressed in the stomach (e.g., TAAR1 and TAAR9; Ohta et al., 2017) and certainly do not exert indirect sympathomimetic actions, there are conditions in which this may occur. Individuals who adversely react to trace amine–rich nutrients such as seafood, cheese, and wine are known (Finberg and Gillman, 2011; Biji et al., 2016), although the molecular basis of the sensitivities are not established in all cases. Genetic deficiency in monoamine oxidases (MAO)-A/B, which are key trace amine–metabolizing enzymes, are known to significantly increase urinary trace amine concentrations and pressor responses to TYR (Lenders et al., 1996). Furthermore, the well known “cheese effect” in patients treated with MAO-A inhibitors is explained by pronounced accumulation of TYR, sufficient to indirectly elevate blood norepinephrine concentrations to the point of inducing hypertensive crisis, severe migraine, and even death (Finberg and Gillman, 2011). With the growing acceptance of TAARs by the wider scientific community, the role of nutrient-derived TAAR ligands in health and disease is likely to become a growing area of nutritional biochemistry and food science interest and to overlap with studies of the molecular basis of host-microbiota interactions. Indeed, the possibility that food-derived trace amines can activate host TAARs has now begun to be recognized (Ohta et al., 2017).

A variety of herbal extracts and brews that are (or have been) used in various cultures as medicines, for religious ceremonies, or for their psychotropic effects are also reported to either contain high levels of trace amines or to alter their concentrations within the body, with such changes implicated in the subsequent physiologic responses. For example, N-methyltyramine has been identified in Ginkgo biloba preparations (Könczöl et al., 2016). Various trace amine compounds have also been suggested to be present in Chinese herbal medicines (Zhang et al., 2016) and in a variety of religious herbal brews (Leonti and Casu, 2014), including ayahuasca where DMT is thought to be one of the major active constituents (Riba et al., 2003).

3. Storage and Passage across Membranes

Unlike their neurotransmitter structural analogs, PEA, TYR, and TRP do not appear to be stored. All three readily diffuse across synthetic lipid bilayers in the absence of transporters, with diffusion half-lives of 15 seconds or less (Berry et al., 2013). Consistent with this, PEA passage across the blood-brain barrier occurs by a mechanism consistent with passive diffusion (Mosnaim et al., 2013), whereas TYR passage across intestinal epithelial cells also appears to be diffusion mediated (Tchercansky et al., 1994). Release of PEA from the whole brain (Henry et al., 1988), striatal slices (Dyck, 1989), or synaptosomes (Berry et al., 2013) is not increased by K⁺-induced depolarization. Similar results were also obtained with TYR (Henry et al., 1988; Berry et al., 2013), with only a small increase in release in response to K⁺-induced depolarization observed in brain slices (Dyck, 1989). Likewise, no effect of depolarization on TRP release from striatal slices was present (Dyck, 1989). The lack of effect of K⁺-induced depolarization strongly suggests that these compounds are neither stored in synaptic vesicles nor released by exocytosis, presumably indicating “release” by simple diffusion across the membrane, consistent with previous reports of a lack of reserpine-sensitive vesicular storage (Juorio et al., 1988). In contrast, veratridine-induced depolarization does increase the release of TYR from striatal slices (Dyck, 1989), although whether similar effects occur with PEA and/or TRP was not studied. Whether this represents a subset of synaptic vesicles that TYR (and possibly other trace amines) can access requires further investigation. Consistent with a lack of vesicular storage, however, the half-life of the endogenous pool of PEA and TRP has been estimated to be less than 30 seconds (Durden and Philips, 1980). OCT has also been reported to have a markedly higher turnover rate than either norepinephrine or dopamine (Molinoff and Axelrod, 1972).

As recently discussed, the above-noted lack of effect of depolarization on PEA, TRP, and, to a less clear extent, TYR release from neuronal preparations is difficult to explain thermodynamically (Berry et al., 2016). Even if “release” is by diffusion, membrane depolarization should increase this due to the subsequent increase in the electrochemical gradient. The lack of such an effect may indicate the presence of a membrane transporter regulating the synaptic levels of PEA, TYR, and TRP. Consistent with such a conclusion, PEA transport into rabbit erythrocytes was previously reported to involve an unknown transporter (Blakeley and Nicol, 1978), as was transport across Caco-2 cells (Fischer et al., 2010). Through the use of pharmacological probes, organic cation transporter 2 (OCT2; Slc22A2) was recently identified as a likely transporter of TYR in rat brain synaptosomes, exhibiting a K_t approximately equal to 100 nM (Berry et al., 2016), in reasonable agreement with estimates of endogenous TYR concentrations (Berry, 2004). OCT2 is widely regarded as a polyspecific transporter (Couroussé and Gautron, 2015) that interacts with a wide variety of therapeutics (Kido et al., 2011; Hacker et al., 2015), not dissimilar to the broad substrate tuning reported for TAAR1 (see IV.B.1.c. Trace Amine–Associated Receptor 1 Ligands). Should a role for OCT2 in regulating trace amine membrane passage at physiologically relevant concentrations be validated, this would implicate the trace amine/TAAR system as a new molecular target to be considered with respect to adverse drug interactions for therapeutics that interact with the OCT2 transporter.

In combination, the above studies suggest that the archetypal trace amines readily cross cell membranes by diffusion but that their levels are further regulated by one or more transporters. The current understanding of the synthesis, metabolism, and transport of PEA and TYR in neurons is summarized in Fig. 5.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Current understanding of the interplay between the TAAR1 and dopamine systems. β-PEA and TYR are synthesized in dopaminergic terminals. Unlike dopamine, they are not stored within synaptic vesicles and readily diffuse across plasma membranes. Reuptake into presynaptic terminals appears to be aided by OCT2. TAAR1 has a predominantly intracellular location, in which both pre- and postsynaptic effects are possible. Whether a transporter contributes to postsynaptic uptake of trace amines is as yet unknown. TAAR1 may translocate to the cell surface after heterodimerization with D2R, an effect that promotes preferential D2R signaling through the G_i signal transduction cascade rather than β-arrestin 2 pathway. TAAR1 interactions with D1R do not occur. A complex system of crosstalk between the dopamine and TAAR1 system is present. Presynaptic activation of D2R, via the G_i signal transduction cascade, results in inhibition of PEA/TYR synthesis. Extracellular metabolism of dopamine by COMT generates 3-MT, an agonist at TAAR1 that like extracellular PEA and TYR can activate the TAAR1/D2R heteromer complex both at pre- and postsynaptic membranes. Dotted blue arrows indicate molecular movement, solid green arrows indicate receptor-mediated stimulation, and solid red lines indicate receptor-mediated inhibition. DOPAC, 3,4-dihydroxyphenylacetic acid; L-Phe, l-phenylalanine; l-Tyr, l-tyrosine; PAA, phenylacetic acid/4-hydroxyphenylacetic acid; TH, tyrosine hydroxylase; VMAT2, vesicular monoamine transporter 2.

B. Isoamylamine

Although isoamylamine was identified as a biologically active component of both decaying meat and tissue samples at the same time as PEA, TYR, and TRP (Barger and Dale, 1910), it has been far less studied in subsequent years, presumably due to less structural similarity to subsequently established neurotransmitters (c.f. Figs. 1 and 3). Isoamylamine can be produced by the decarboxylation of leucine (Table 1) and is a component of commonly consumed fermented foodstuffs (Cunha et al., 2011; Bach et al., 2012; Coton et al., 2012). Although a variety of leucine decarboxylase activities have been described in prokaryotes (Coton et al., 2012), including some commensal gut microbiota (Haughton and King, 1961), there is no known leucine decarboxylase enzyme in vertebrates (Table 1). Isoamylamine production could occur, however, via methionine decarboxylase (EC 4.1.1.57) or valine decarboxylase (EC 4.1.1.14), both of which include l-leucine in their substrate profile (Table 1).

The initial descriptions of MAO activity identified isoamylamine as a substrate (Blaschko et al., 1937; Pugh and Quastel, 1937) and further characterization has suggested that isoamylamine exhibits selectivity toward MAO-B, similar to that of PEA (Peers et al., 1980). In addition, isoamylamine is also a substrate for SSAO/VAP-1/AOC3 (Elliott et al., 1989). Isoamylamine is an agonist at TAAR3 (Liberles and Buck, 2006), and TAAR3 and its responses are described in section IV.E.2 Trace Amine–Associated Receptor 3. Isoamylamine is excreted in male mouse urine (Nishimura et al., 1989) and may induce puberty in female mice (Nishimura et al., 1989), suggesting a possible role as a pheromone, which would most likely require that a nonmicrobial synthetic pathway is present. Others, however, have failed to replicate the puberty-accelerating effect of isoamylamine (Price and Vandenbergh, 1992).

Previously, isoamylamine was reported to share a gastrin secretion-stimulating effect with PEA and TYR (Dial et al., 1991) and it has recently been reported to regulate gastrointestinal motility in a tetrodotoxin-sensitive manner (Sánchez et al., 2017), possibly indicating a neuron-mediated mechanism. In this regard, it is interesting to note that isoamylamine was suggested to be an endogenous neurally active compound at least 40 years ago (Tashiro et al., 1974). More recently, cardiac levels of isoamylamine were reported to negatively correlate with septal wall thickness and to be increased in patients suffering cardiac arrest (Meana et al., 2016). Isoamylamine was also suggested to be beneficial in an animal model of endotoxemia (Yen et al., 2016). The mechanism and importance of these putative effects requires further study.

C. Trimethylamine

Although repeatedly identified as a component of various human bodily fluids between the mid-1850s and early 1910s (see Mitchell and Smith, 2016), a physiologic relevance of trimethylamine was not suggested until the 1930s when an increase in its excretion was reported to occur during menstruation (Ashley-Montagu, 1938), an effect that has subsequently been confirmed (Shimizu et al., 2007). Although trimethylamine levels have been associated with various disease conditions (Chhibber-Goel et al., 2016), are sexually dimorphic (Gavaghan McKee et al., 2006), and a proposed male pheromone in mice (males having a genetic deficiency in trimethylamine metabolism) (Li et al., 2013), there is no known endogenous synthetic pathway. Rather, trimethylamine production is thought to occur as a by-product of prokaryotic degradation of dietary choline, betaine, phosphatidylcholine, and l-carnitine (Zhang et al., 1999; Craciun and Balskus, 2012; Zhu et al., 2014; Kalnins et al., 2015; Fennema et al., 2016). Such well documented production by the commensal microbiota, along with an established host receptor (TAAR5, see subsequent sections), makes trimethylamine an attractive candidate molecule for investigating the growing appreciation of the role of the microbiota in health and disease. The apparent lack of nonmicrobial production of trimethylamine raises questions as to whether it is the true endogenous ligand for TAAR5 and offers an interesting parallel to the archetypal trace amines TYR, PEA, and TRP, which were previously often consigned to the role of metabolic by-products (Berry, 2004). With a receptor that is selectively activated by trimethylamine now identified, a re-examination of whether trimethylamine production is solely dependent on the commensal microbiota is warranted.

Trimethylamine has been confirmed to be a high-affinity agonist at TAAR5 from multiple species (Liberles and Buck, 2006; Ferrero et al., 2012; Li et al., 2013; Wallrabenstein et al., 2013; Zhang et al., 2013). Effects clearly demonstrated to be TAAR5 mediated are described in sections IV.C.4 Trace Amine-Associated Receptor 5 and IV.E.4 Trace Amine–Associated Receptor 5, although these are rather few in number at present. The main interest in trimethylamine thus far has been with respect to its degradation. This primarily occurs via the enzyme flavin monooxygenase 3 (FMO3; EC 1.14.13.8), which is prevalent in hepatic tissue, including in humans (Fennema et al., 2016). This generates trimethylamine-N-oxide (TMAO), a compound that has been increasingly implicated as playing a role in both cardiovascular and metabolic disorders (Fennema et al., 2016; Zhang and Davies, 2016). Interestingly, the K_m for trimethylamine metabolism by FMO3 is rather high (28 µM; Lang et al., 1998) and is approximately 100-fold in excess of the EC₅₀ values of trimethylamine at TAAR5 from various species (Berry et al., 2017). This raises the possibility of previously unsuspected physiologic effects of nanomolar trimethylamine levels and that TAAR5 may be under tonic activation. As described in later sections, at least two receptor targets for trimethylamine are likely to be present, since TAAR5 KO prevents low-dose but not high-dose trimethylamine effects (Li et al., 2013).

An evolutionary deletion of FMO3 in male mice underlies their markedly increased urinary trimethylamine levels (Li et al., 2013). In humans, a genetic deficiency in FMO3 leads to the condition of trimethylaminuria (Humbert et al., 1970), in which large quantities of trimethylamine are excreted in sweat, urine, and exhaled air (Fennema et al., 2016). Kidney damage (Bain et al., 2006; Chhibber-Goel et al., 2016) and gut dysbiosis (Fennema et al., 2016) can also lead to trimethylaminuria, as can invasion by pathologic microflora associated with bacterial vaginosis or infections of the oral cavity (Fennema et al., 2016; Zhang and Davies, 2016). The increased trimethylamine urinary excretion associated with menstruation (Mitchell and Smith, 2010) may reflect regulation of FMO3 by female sex hormones (Coecke et al., 1998; Shimizu et al., 2007). Consistent with this, FMO3 activity is altered during pregnancy (Hukkanen et al., 2005). Interestingly, trimethylamine urinary levels were found to be very stable in a recent study in children, showing the least intraindividual variability during repeated sampling, whereas those of its metabolite TMAO were highly variable (Maitre et al., 2017), suggesting that trimethylamine levels are more tightly controlled than those of its primary metabolite. This could indicate a physiologic relevance of the parent compound. Trimethylamine has also been detected in the feces of at least one avian species (the black-bellied whistling duck) (Robacker et al., 2000), where it may play a role in either conspecific (Mueller et al., 2008) or heterospecific (Robacker et al., 2000) communication.

Trimethylaminuria is generally classified as a rather benign condition, except for the socially stigmatizing strong fish odor associated with afflicted individuals, and this is the basis of the more common name for the disorder—fish malodor syndrome. The pronounced malodor associated with trimethylaminuria does, however, result in a severe loss of quality of life for affected persons (Fennema et al., 2016).

TMAO can also be produced by bacterial action on trimethylamine through various oxygenase enzymes or by the nonenzymatic action of reactive oxygen species (Brown and Hazen, 2017). In addition, prokaryotes can convert TMAO back to trimethylamine (Zhang et al., 1999). Alternatively, dehydrogenase enzymes from at least some bacteria can convert trimethylamine into dimethylamine (Kim et al., 2001; Shi et al., 2005) with subsequent conversions to methylamine and ammonia (Kim et al., 2001), although whether this ability is shared by commensal bacteria is unknown.

At high concentrations trimethylamine has been reported to inhibit macromolecule synthesis, leading to teratogenesis in mouse embryos due to inhibition of receptor-mediated precursor uptake (Guest and Varma, 1992; Guest et al., 1994). This effect has been suggested to be more pronounced in male offspring due to trimethylamine-mediated inhibition of testosterone synthesis (Guest and Varma, 1993). FMO3 expression is regulated by both glucagon and insulin, with insulin decreasing FMO3 levels in a sexually dimorphic manner (Miao et al., 2015), and dysregulated trimethylamine metabolism has been implicated in the pathology of both obesity and diabetes (Dumas et al., 2006; Toye et al., 2007), including in humans (Elliott et al., 2015). Trimethylamine levels have also been correlated to the onset of symptoms in the interleukin-10 KO mouse model of inflammatory bowel disease (Murdoch et al., 2008), although the basis of this relationship is unclear. As previously indicated, there is a growing body of work that implicates the trimethylamine metabolite TMAO in cardiovascular disease (Koeth et al., 2013; Miao et al., 2015). Although few studies have examined for a role of the parent compound, one study has linked trimethylamine to atherosclerosis in patients with human immunodeficiency virus infection (Srinivasa et al., 2015). Elevated trimethylamine levels have also been suggested in patients with a malignant peripheral nerve sheath tumor (Fayad et al., 2014) and glomerulosclerosis (Hao et al., 2013), although the relevance of these observations to disease processes has not been determined. High-altitude pulmonary edema has also been reported to be associated with a decrease in plasma trimethylamine levels (Luo et al., 2012), although the relevance of this beyond serving as a potential biomarker is unknown.

Trimethylaminuria has sporadically been reported to be associated with epilepsy (McConnell et al., 1997; Pellicciari et al., 2011), and a lowering of the epileptogenic threshold has been reported in experimental animals in response to trimethylamine (Gajda et al., 2003; Leniger et al., 2004; Sayyah et al., 2007; Nassiri-Asl et al., 2008), possibly as a consequence of the opening of gap junctions (Gajda et al., 2003; Bocian et al., 2011; Chang et al., 2013). That epileptic activity could be controlled by a diet restricted in trimethylamine precursors (McConnell et al., 1997; Pellicciari et al., 2011) raises the possibility of a direct causal link between elevated trimethylamine and a lowered epileptogenic threshold. Possibly related to this, trimethylamine has also been reported to increase the amplitude of theta wave oscillations (Bocian et al., 2011). A variety of psychiatric and behavioral disorders have been suggested to coexist with trimethylaminuria, although how many of these are either simply coincidental (McConnell et al., 1997) or secondary to the social stigma associated with the pronounced halitosis and body odor (Todd, 1979; Mitchell and Smith, 2001) is unclear. Some behavioral abnormalities have, however, been reported to be ameliorated with diets low in trimethylamine precursors (McConnell et al., 1997), indicating a possible link between trimethylamine metabolism and central neuronal function.

Lower maternal urinary trimethylamine levels have been reported to be associated with fetal growth retardation (Maitre et al., 2014), whereas maternal plasma trimethylamine levels may be a marker for discrimination of trisomy 18 from trisomy 21 (Bahado-Singh et al., 2013). Interestingly, FMO3 expression is not turned on until after birth in humans and continues to increase until adulthood (Fennema et al., 2016), suggesting a developmental component to the trimethylamine/TMAO axis. Trimethylamine is present in breast milk and its concentration in amniotic fluid increases markedly during gestation (Lichtenberger et al., 1991). Together, these studies raise the possibility of an involvement of trimethylamine in programming fetal and neonatal metabolism.

D. O-Methyl and N-Methyl Derivatives

The O-methyl metabolites of the catecholamine neurotransmitters (3-MT, 4-methoxytyramine, metanephrine, and normetanephrine) were identified as agonists at TAAR1 during the initial characterization of the receptor family (Bunzow et al., 2001), with the effect of 3-MT being subsequently confirmed by others (Sotnikova et al., 2010). Although very few studies have examined the physiologic effects of these compounds, a complex spectrum of locomotor behaviors and the corresponding striatal intracellular signaling events are induced by intracerebroventricular 3-MT administration, and these are partially ameliorated in TAAR1-KO animals (Sotnikova et al., 2010). 3-MT levels are increased in brains with Parkinson disease during the development of dyskinesias after chronic l-DOPA administration (Rajput et al., 2004). As such, TAAR1 may be a target for improving the on-off phenomenon associated with long-term l-DOPA therapy. Such effects certainly point to at least 3-MT having physiologic roles beyond being a mere marker of extracellular dopamine levels, and it is expected that the identification of TAAR1 as a receptor target for the O-methyl neurotransmitter metabolites will lead to further studies in these areas.

N-methyl metabolites of trace amines have also been shown to be ligands for TAARs (Lindemann and Hoener, 2005); however, with at least some of these compounds, this is species dependent (Simmler et al., 2016; Berry et al., 2017). N-methyltyramine is widely distributed in plant species (Smith, 1977; Stohs and Hartman, 2015), where its presence has been known since 1950 (Kirkwood and Marion, 1950). In addition, N-methylphenylethylamine has been identified in human urine (Reynolds and Gray, 1978). Like the parent compounds, at high concentrations, N-methyl trace amines have long been known to cause pronounced pressor effects (Hjort, 1934), most likely due to either direct or indirect sympathomimetic effects (Stohs and Hartman, 2015). In addition, and at much lower concentrations, N-methyltyramine has been reported to stimulate gastric (Yokoo et al., 1999) and pancreatic (Tsutsumi et al., 2010) secretions. DMT is perhaps the most widely studied of the N-methyl derivatives due to its potent hallucinogenic properties, which likely have a polypharmacological basis (Carbonaro and Gatch, 2016). In addition to neural effects, DMT also appears to possess immunomodulatory properties (Szabo, 2015).

E. 3-Iodothyronamine

Like the archetypal trace amines, 3IT is an endogenous compound found at nanomolar levels throughout the body in both rodents and humans (Scanlan et al., 2004; Zucchi et al., 2014). Although it is often reported as a derivative/metabolite of thyroid hormones (Hoefig et al., 2016), whether 3IT is indeed derived from either 3,3′,5-triiodothyronine or thyroxine after decarboxylation and deiodination is a matter of some conjecture (Ackermans et al., 2010; Hoefig et al., 2011, 2015; Hackenmueller et al., 2012). The available evidence indicates that 3IT is not formed by the action of AADC (Hackenmueller et al., 2012; Hoefig et al., 2012) but may be formed by ornithine decarboxylase (EC 4.1.1.17) (Hoefig et al., 2015). Once formed, 3IT is a substrate for MAOs, SSAO/VAP-1/AOC3, and deiodinases, resulting in the production of 3-iodothyroacetic acid and thyronamine (Piehl et al., 2008; Wood et al., 2009; Saba et al., 2010; Hackenmueller and Scanlan, 2012; Laurino et al., 2015). In addition, N-acetyl, O-sulfonate, and glucuronide conjugates of 3IT have been reported to be widely distributed throughout the body (Pietsch et al., 2007; Hackenmueller and Scanlan, 2012).

In 2004, 3IT was identified as a high-affinity agonist at TAAR1 (Scanlan et al., 2004). This has subsequently been verified by numerous laboratories (Mühlhaus et al., 2014; Cöster et al., 2015; Chiellini et al., 2017), with 3IT also shown to act as an agonist at TAAR2 (Babusyte et al., 2013; Cichero and Tonelli, 2017) and as an inverse agonist at TAAR5 (Dinter et al., 2015c). 3IT is promiscuous, however, and also interacts with high affinity at α₂-adrenoceptors (Regard et al., 2007; Dinter et al., 2015b), β-adrenergic receptors (Meyer and Hesch, 1983; Kleinau et al., 2011; Dinter et al., 2015a), muscarinic acetylcholine receptors (Laurino et al., 2016), transient receptor potential cation channel subfamily M member 8 ion channels (Khajavi et al., 2015; Lucius et al., 2016), various monoamine and organic anion transporters (Snead et al., 2007; Panas et al., 2010), and molecular target(s) within mitochondria, including the F₁-F₀ATP synthase (Cumero et al., 2012) and possibly complex III (Venditti et al., 2011). 3IT also tightly binds to the plasma protein apolipoprotein B-100 (Roy et al., 2012).

The physiologic effects of 3IT are often, although not exclusively, opposite to those of the thyroid hormones. A variety of physiologic effects of 3IT have been reported and a number of reviews of such effects have previously been published to which the reader is referred for further details (Ianculescu and Scanlan, 2010; Zucchi et al., 2010, 2014; Piehl et al., 2011; Hoefig et al., 2016). Here, we will provide a brief overview of such effects with the following caveat: the promiscuous nature of 3IT makes it highly unlikely that all, or even most, of the physiologic effects described are mediated via one or more TAARs. Where there is unequivocal evidence of TAAR-mediated 3IT responses, these are discussed in the subsequent sections focused on those individual receptor subtypes. Furthermore, it was recently suggested that at least some of the effects previously ascribed to 3IT may in fact be a function of one or more of its metabolites (Hoefig et al., 2016; Laurino and Raimondi, 2017), adding a further level of complexity to the elucidation of the physiologic relevance and pharmacological effects of 3IT.

F. Polyamines

Like the archetypal trace amines, polyamines are found in all species (Miller-Fleming et al., 2015). They are traditionally viewed as consisting of cadaverine, putrescine, spermidine, and spermine. Polyamine metabolism is complex, not least because of their ability to interconvert, and this has been detailed fully in numerous review articles (Miller-Fleming et al., 2015; Pegg, 2016). Putrescine is primarily derived from l-ornithine by the action of ornithine decarboxylase (EC 4.1.1.17; Table 1). Putrescine itself can be converted to spermidine by spermidine synthase (EC 2.5.1.16), which in turn can be converted to spermine by spermine synthase (EC 2.5.1.22). Both spermidine synthase and spermine synthase are dependent on the action of S-adenosylmethionine decarboxylase (EC 4.1.1.50) to generate decarboxylated S-adenosylmethionine as a donor of aminopropyl groups. Back conversion of spermine to spermidine, and spermidine to putrescine can occur after N¹-acetylation by spermidine/spermine N¹-acetyltransferase, and subsequent oxidation by polyamine oxidase (EC 1.5.3.11). A distinct spermine/spermidine N⁸-acetylation pathway is also present, which may generate substrates of MAO (Youdim et al., 1991). Cadaverine is produced by direct decarboxylation of lysine through the action of lysine decarboxylase (EC 4.1.1.18; Table 1).

Although polyamines are usually considered distinct from monoaminergic signaling systems, a putative diamine binding site was reported to be present in both human TAAR6 and TAAR8 (Li et al., 2015). Furthermore, at least in fish, olfactory detection of cadaverine was shown to be dependent on TAAR13c (Hussain et al., 2013), with other teleost-specific TAAR isoforms also recognizing diamines (Table 2). Although these TAAR isoforms are not found in terrestrial vertebrates, pan-TAAR-KO prevents the innate avoidance behavior shown by mice in response to the presence of both putrescine and cadaverine (Dewan et al., 2013). Which mouse TAAR isoform(s) is responsible, however, remains unknown. Furthermore, whether polyamines other than putrescine and cadaverine also act as ligands at tetrapod TAARs is an area for systematic study, but it is perhaps quite likely that spermine and/or spermidine will also interact with one or more TAARs, given their close structural homology to putrescine and cadaverine and similar ecological context to other TAAR ligands such as trimethylamine and isoamylamine.

The physiology of cadaverine and putrescine remains largely a mystery, beyond their association with decaying carcasses due to bacterial degradation of proteinaceous material and innate repulsive properties in most species. The classic polyamines as a whole have been implicated in a variety of mammalian cellular processes, most notably in the regulation of cell growth and proliferation, cellular stress responses, ion channel function, and as allosteric regulators of glutamatergic NMDA and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors (Miller-Fleming et al., 2015; Pegg, 2016). One other polyamine perhaps deserves mention in these regards. Agmatine is formed by decarboxylation of arginine (Table 1), primarily via arginine decarboxylase (EC 4.1.1.19), and can itself be subsequently converted into putrescine after elimination of urea by the enzyme agmatinase (EC 3.5.3.11). A variety of physiologic and pathologic functions have been ascribed to agmatine (Table 1) (Piletz et al., 2013), many of which bear a striking overlap with the reported effects of TAARs and/or trace amines. For example, agmatine has been shown to interact either directly or indirectly with the function of a variety of neurotransmitter receptors, including imidazoline receptors (Li et al., 1994; Raasch et al., 2001), NMDA receptors (Yang and Reis, 1999), and serotonin 5-HT_2a receptors (Taksande et al., 2009). Agmatine also modulates the functions of K_ATP channels (Shepherd et al., 1996) and ASICs (Li et al., 2010) and interacts with a variety of transporter proteins, including OCT2 (Gründemann et al., 2003), OCT3 (Gründemann et al., 2003), and mitochondrial transport systems (Grillo et al., 2007). Like the known TAAR ligands PEA, TYR, and isoamylamine, regulation of gastrin secretion by agmatine has also been reported (Dial et al., 1991). This is far from an exhaustive list; agmatine has been reported to have a wide variety of molecular targets and physiologic effects (Molderings and Haenisch, 2012; Piletz et al., 2013; Laube and Bernstein, 2017), and binding to one or more TAAR may just be one of many mechanisms by which agmatine can exert its effects.

Although the molecular basis of its effects remains largely unknown, agmatine dysregulation has been implicated in affective disorders, cognition, drug abuse/addiction, and metabolic disorders (Piletz et al., 2013). Although agmatine is known not to interact with TAAR1 (Hu et al., 2009), it does interact with multiple zebrafish TAARs, including TAAR13c (Table 2) (Hussain et al., 2013). Putative interaction at mammalian TAAR subtypes is an area for future study, particularly in light of the presence of a diamine binding pocket in TAAR6 and TAAR8 that is conserved with that present in TAAR13c (Li et al., 2015).

G. Putative Other Trace Amines

The majority of TAARs remain orphan receptors with no known cognate ligand. Indeed, as described in subsequent sections, this is problematic with respect to following current IUPHAR conventions for naming the receptor family. From the preceding sections, however, it is readily apparent that a large number of TAAR ligands are the direct products of amino acid decarboxylation (Table 1), and this is perhaps a good place to start the search for endogenous ligands of currently orphaned TAARs. With this in mind, Table 1 provides, for reference, the decarboxylated products of common amino acids and, where known, their physiologic effects and TAAR binding activity.

A. Synthesis and Degradation

Although invertebrate synthesis and degradation of TYR and OCT occurs by analogous pathways to those of vertebrates, distinct enzymes are present (Blenau and Baumann, 2001; Roeder, 2005; Verlinden et al., 2010a). Dedicated tyrosine decarboxylase enzymes are found in various invertebrate species (Livingstone and Tempel, 1983; Ishida and Ozaki, 2012; McCoole et al., 2012; Christie et al., 2014) and, at least in Drosophila, distinct neuronal and non-neuronal isozymes exist (Cole et al., 2005). Although an AADC-like, DOPA decarboxylase enzyme may also be present, this does not appear to act on l-tyrosine (Han et al., 2010). In addition to being active in its own right, TYR can be further acted upon by tyramine-β-hydroxylase enzymes to generate OCT (Monastirioti et al., 1996; Alkema et al., 2005; Nishimura et al., 2008; Châtel et al., 2013). Again, multiple isozymic forms are present in some species (Châtel et al., 2013). Formation of TYR from dopamine via a dopamine dehydroxylase pathway is also possible in some species (Walker and Kerkut, 1978; Roeder, 2005).

Whereas vertebrate degradation of TYR and OCT occurs primarily by MAOs, the primary invertebrate degradative route is thought to be via N-acetylation and N-methylation (Dewhurst et al., 1972; Roeder, 2005; Verlinden et al., 2010a), with sulfate, β-alanyl, glycosyl, and ɣ-glutamyl conjugations also possible (Maxwell et al., 1980; Wright, 1987; Sloley, 2004; Roeder, 2005).

B. Storage and Release

OCT release has been demonstrated to occur from insect neuronal preparations in response to both electrical- and K⁺-induced depolarization (Morton and Evans, 1984; Orchard and Lange, 1987; Verlinden et al., 2010a), with typical neurotransmitter-like exocytotic release from synaptic vesicles (Consoulas et al., 1999). Released TYR and OCT are primarily removed from the synaptic cleft by dedicated transporter systems (Roeder and Gewecke, 1989; McClung and Hirsh, 1998; Gallant et al., 2003; Donly and Caveney, 2005), with some evidence for species specificity in the transporters present (Roeder, 2005).

C. Octopamine Receptors

The first invertebrate OCT receptor to be identified was found in the mushroom bodies of the D. melanogaster brain and was named OAMB (Han et al., 1998). More recently, it has been suggested that this receptor be renamed OctαR, reflective of its homology to vertebrate α-adrenergic receptors (Evans and Maqueira, 2005; Bayliss et al., 2013). The OAMB receptor is coupled to Ca²⁺ accumulation (Zeng et al., 1996; Han et al., 1998; Grohmann et al., 2003; Balfanz et al., 2005; Evans and Maqueira, 2005). Orthologs of OAMB have subsequently been identified in Apis (Blenau and Baumann, 2001; Grohmann et al., 2003; Balfanz et al., 2014), Anopheles (Kastner et al., 2014), Schistocerca (Verlinden et al., 2010b), and Locusta (Hiripi et al., 1994; Ma et al., 2015), among other invertebrate species. Whether the OAMB receptor is a true OCT receptor or a mixed OCT/TYR receptor has been a matter of some recent conjecture (El-Kholy et al., 2015). Very recently, a putative novel receptor with homology to α₂-adrenoceptors was reported in Drosophila, coupled to G_i-mediated inhibition of adenylyl cyclase, and activated by both OCT and TYR (Qi et al., 2017).

Subsequent to the OAMB receptor discovery, three further invertebrate OCT receptors were identified (Oct1βR, Oct2βR, and Oct3βR; Evans and Maqueira, 2005; El-Kholy et al., 2015), with strong homology to the vertebrate β_1–3 adrenergic receptors, respectively (Maqueira et al., 2005). Like their vertebrate counterparts, all three β-receptors are coupled to the cAMP signal transduction cascade through activation of adenylyl cyclase (Chang et al., 2000; Maqueira et al., 2005; Balfanz et al., 2014). A fourth β-like receptor has also been suggested to be present in honeybees (Hauser et al., 2006; Balfanz et al., 2014). For a more detailed review of species-specific expression of the various isoforms, the reader is referred to a number of excellent dedicated reviews on invertebrate OCT/TYR receptors (Hauser et al., 2006; Verlinden et al., 2010a).

D. Tyramine Receptors

Dedicated invertebrate TYR receptors were first suggested in 1990 (Arakawa et al., 1990; Saudou et al., 1990), with three GPCRs (TyrRI, TyrRII, and TyrRIII) subsequently identified. TyrRI is the least selective of the three, recognizing both OCT and TYR (Robb et al., 1994; Reale et al., 1997), although in most species there is a higher affinity for TYR (Vanden Broeck et al., 1995; Evans and Maqueira, 2005; Bayliss et al., 2013). TyrRI is homologous to vertebrate α₂-adrenoceptors (Evans and Maqueira, 2005) and may exhibit agonist-dependent coupling, since TYR was reported to induce adenylyl cyclase inhibition (Vanden Broeck et al., 1995; Blenau and Baumann, 2001), whereas OCT induces an increase in cytosolic Ca²⁺ (Robb et al., 1994; Evans and Maqueira, 2005; Beggs et al., 2011). A second TyrRI isoform was very recently reported in the cockroach (Blenau et al., 2017).

TyrRII and TyrRIII show much greater selectivity toward TYR over OCT (Cazzamali et al., 2005; Bayliss et al., 2013), although TyrRIII also responds to a variety of other amines (Bayliss et al., 2013). Orthologs of TyrRII have been identified in multiple species (Cazzamali et al., 2005; Hauser et al., 2006; Huang et al., 2009; Bayliss et al., 2013; Wu et al., 2015), although TyrRIII has thus far only been identified in Drosophila. TyrRII activation is associated with a release of Ca²⁺ from intracellular stores (Cazzamali et al., 2005; Huang et al., 2009), whereas TyrRIII is coupled to an inhibition of adenylyl cyclase (Bayliss et al., 2013).

In addition to GPCRs, one ionotropic TYR receptor has been reported in C. elegans (Pirri et al., 2009; Ringstad et al., 2009). The ligand-gated ion channel LGC-55 receptor is a TYR-gated Cl⁻ channel that can also be activated by PEA (Safratowich et al., 2014) and is involved in the control of locomotor activity. Orthologs of LGC-55 have also been identified in other nematode species (Pirri et al., 2009).

E. Physiologic Responses

TYR and more particularly OCT have been implicated in a variety of physiologic responses in invertebrates. Their receptors are not restricted to the nervous system but rather are distributed throughout the body (Saraswati et al., 2004; Roeder, 2005; Vierk et al., 2009; Ma et al., 2015). Although responses to TYR have generally received less attention, when both have been studied, TYR generally induces responses opposite to those of OCT (Saraswati et al., 2004; Ma et al., 2015). Although some of the responses (in particular those toward OCT) are similar to those that have been proposed to be mediated by TAARs in vertebrates, it is important to emphasize again that vertebrate TAARs and the invertebrate OCT/TYR receptors are evolutionarily very distant from each other (Gloriam et al., 2005; Lindemann et al., 2005; Hashiguchi and Nishida, 2007).

A. Evolution of Trace Amine–Associated Receptors

At present, the majority of evidence suggests that an ancestral TAAR-like protein first emerged in lamprey (Gloriam et al., 2005; Hashiguchi and Nishida, 2007; Libants et al., 2009; Eyun et al., 2016), with a conserved TAAR signature motif appearing later, after the divergence of jawed vertebrates from jawless fish (Fig. 6). Others have, however, suggested that TAAR ancestral emergence only occurred after the divergence from lamprey (Hussain et al., 2009; Tessarolo et al., 2014) largely because of the absence of the conserved TAAR motif from the lamprey receptors. Lamprey do innately avoid PEA sources (Imre et al., 2014), a response known to be TAAR mediated in jawed vertebrates; although the molecular basis of this lamprey avoidance behavior is unknown, it is potentially consistent with ancestral TAARs being present.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Phylogenetic tree of TAARs. Sequences from the publications by Azzouzi et al. (2015), Gao et al. (2017), Eyun et al. (2016), Tessarolo et al. (2014), and Vallender et al. (2010) were used. Using the jackhmmer program from the hmmer-3.1b1 software package (http://hmmer.org, Howard Hughes Medical Institute) and chimpanzee TAAR1 as a seed, we iteratively built a hidden Markov model (HMM) representing TAARs. All of the sequences were scored and aligned against this HMM (which has 339 match positions). Next, all sequences between 250 and 450 amino acids long, with at least 280 match positions to the HMM, and with a starting methionine were selected. We made the selection nonredundant, correcting apparent sequencing errors in a few cases where the different sources above disagreed. Based on the alignment of the selected sequences to the TAAR-HMM, protein distances were calculated using the protdist program from the phylip software package version 3.69 (with default parameters, http://evolution.genetics.washington.edu/phylip.html, University of Washington). A tree was then generated using the neighbor program from the same package, with default parameters, and using human histamine receptor H2 as an outgroup.

Among mammals, the highest number of TAARs identified thus far is in the flying fox, with 26 functional genes, whereas the bottlenose dolphin is the only vertebrate species known to contain no functional TAAR genes (Eyun et al., 2016). Like in other tetrapods, mammalian TAARs generally belong to nine subfamilies (TAAR1–TAAR9). Within these, repeated species-specific expansion, duplication, and pseudogenization events have occurred. This has resulted in great variability in the total number of functional TAARs present between species (Table 5), including the appearance of species-specific isoforms (Lindemann et al., 2005; Vallender et al., 2010; Eyun et al., 2016). For example, four putative new subgroups of TAARs were recently described: TAAR E1, which is present in the common shrew, hedgehog, lesser hedgehog tenrec, and African elephant; and TAARs M1–M3 found in two marsupial species, tammar wallaby and opossum (Eyun et al., 2016), consistent with previous suggestions of marsupial-specific TAARs (Vallender et al., 2010). These new families appear to belong to the clade II, tertiary amine-recognizing group of TAARs, and as such are most closely related to TAAR5–TAAR9 (Eyun et al., 2016). Of these putative new TAARs, species-specific variants of TAAR E1, M2, and M3 were reported.

Two species appear rather unique (Table 5); the bottlenose dolphin, as previously described, has no functional TAAR genes, only possessing three pseudogenes (Eyun et al., 2016). Meanwhile, the dog is the only known species to express functional TAARs but no TAAR1, and its genome being represented with functional TAAR4 and TAAR5 plus a further two pseudogenes (Vallender et al., 2010).

Interestingly, the number of TAARs appears to be correlated with the number of olfactory receptors in most species, with the exception of Canis, which has more than 800 olfactory receptors but only two functional TAARs, with the olfactory receptors likely compensating for the majority of functions served by TAARs in other species (Eyun et al., 2016). There are up to seven functional genes in Xenopus (Mueller et al., 2008; Eyun et al., 2016), although the TAAR complement of the true frogs (Rana sp.) has not been determined. This is a notable knowledge gap for the determination of the evolutionary relationships of olfactory functions, with the amphibian olfactory system regarded as an evolutionary midpoint between teleosts and tetrapods (Duchamp-Viret and Duchamp, 1997; Gliem et al., 2013).

The mammalian TAAR1–TAAR9 family seems to have two distinct evolutionary patterns. Primary amine-detecting clade I TAARs (TAAR1–TAAR4) (Ferrero et al., 2012) appear to be evolving under strong negative (purifying) selection criteria, whereas the tertiary amine-detecting clade II TAARs (TAAR5–TAAR9) (Ferrero et al., 2012) show significant variations in gene numbers and appear to evolve under the influence of positive selection pressures (Eyun et al., 2016). Similarly, clade III zebrafish TAARs have undergone strong positive selection (Hussain et al., 2009). Thus, the profiles of TAAR expression appear to be evolutionarily determined by adaptive responses; indeed, there is now evidence that TAAR expression can either be influenced by habitat changes or determine habitat choice (Churcher et al., 2015; Fatsini et al., 2016), consistent with a role in the detection and response to migratory cues.

As expected from the above, the primary amine-detecting TAAR1–TAAR4 appear to be evolutionarily older, more conserved, and are generally represented by a single isoform in the majority of genomes, with the exception of TAAR4 (in which a low level of species-specific expansion has occurred) (Hussain et al., 2009; Eyun et al., 2016). In contrast, the tertiary amine-detecting mammalian TAARs (TAAR5–TAAR9) have arisen more recently with multiple species-specific isoforms of each subtype present, with the exception of TAAR5 (the oldest of the clade II genes) (Eyun et al., 2016). The only TAAR that is not expressed in the olfactory system (TAAR1) is the oldest member of the family, with TAAR4 the second oldest, and these two family members share considerable overlap in their ligand selectivity (Lindemann et al., 2005). Other evolutionarily ancient members of the family (TAAR2, TAAR3, and TAAR5) are also thought to predate the origin of amniotes (Eyun et al., 2016). All other mammalian TAAR subfamilies are currently thought to derive from a single copy of one of these ancestral TAARs via gene duplication (Eyun et al., 2016). The teleost-specific clade III TAARs are thought to be the most recent family members to have emerged (Hussain et al., 2009).

B. Trace Amine–Associated Receptor 1

TAAR1 is the best characterized of the TAAR family and is currently regarded as the main target for PEA and TYR, although in this respect it should be noted that TAAR4, a pseudogene in humans, is also a target but has rarely been studied. TAAR1 is expressed in brain structures associated with psychiatric disorders, in particular in key areas where modulation of dopamine [ventral tegmental area (VTA)] and serotonin [dorsal raphe nuclei (DRN)] occurs (Table 4), and the presence of TAAR1 variants was recently reported to be associated with schizophrenia (John et al., 2017). In the periphery, TAAR1 expression suggests putative roles in regulating immune system responses and controlling energy metabolism. The current understanding of the cellular distribution and pharmacological profile of TAAR1 in validated disease models is summarized in Tables 4 and 6 and described in detail in the following sections.

Characterization of TAAR1-KO and TAAR1-overexpressing (OE) mouse and rat lines has greatly contributed to a better understanding of TAAR1 function (Wolinsky et al., 2007; Lindemann et al., 2008; Di Cara et al., 2011; Revel et al., 2011, 2012a; Espinoza et al., 2015a; Harmeier et al., 2015; Black et al., 2017; Schwartz et al., 2017). TAAR1-KO mice exhibit an enhanced response to amphetamine, resulting in an increased hyperlocomotion that is correlated with more pronounced increases of dopamine, norepinephrine, and serotonin (Wolinsky et al., 2007; Lindemann et al., 2008). TAAR1-OE mice show phenotypically opposite effects with lowered sensitivity to amphetamine, whereas their wild-type littermates maintain a normal hyperlocomotion response (Revel et al., 2012a). Furthermore, the spontaneous firing rate of dopaminergic neurons in the VTA and serotonergic neurons in the DRN is increased in TAAR1-KO mice (Bradaia et al., 2009; Revel et al., 2011). Thus, TAAR1 plays an active role in the expression of behaviors traditionally associated with classic neurotransmitters, supporting its potentially important role in the modulation of neurotransmitter functions and, hence, in mental illness.

2. Central Nervous System Effects

3. Effects in the Periphery

1. Trace Amine–Associated Receptor 2

Whereas all other TAAR genes have a single exon, TAAR2 (previously known as GPR58 or G protein–coupled receptor 58; Table 3) contains two exons, encoding a functional protein of 351 amino acids in humans (Lindemann et al., 2005). TAAR2 shares closest homology with TAAR5 (previously known as the putative neurotransmitter receptor, PNR; Table 3) at 42% identity, and with the 5-HT₄ serotonin receptor at 34% identity (Lindemann et al., 2005). Ligands for TAAR2 have not been identified thus far, either among known endogenous trace amines or other volatile amines. Evolutionary mapping, however, suggests that the receptor will be tuned toward activation by primary amines (Ferrero et al., 2012). Like all TAARs, with the exception of TAAR1, TAAR2 is found within the olfactory epithelium where it is coupled to G_olf stimulation of cAMP accumulation (Liberles and Buck, 2006). The role of TAAR2 and other TAARs in olfactory sensory function is discussed in detail in section IV.E.

Beyond the olfactory system, TAAR2 mRNA is found in various leukocyte populations in humans and mice, including B cells, granulocytes, monocytes, natural killer cells, and T cells, a pattern that mirrors the expression of TAAR1 (Nelson et al., 2007; Babusyte et al., 2013; Table 4). The presence of TAAR2 at the protein level has been confirmed in granulocytes (Babusyte et al., 2013), although it should be noted that validation of the selectivity of the antibody used was not provided. It has been noted previously that leukocyte stimulation results in an increase in both TAAR1 and TAAR2 mRNA levels (Nelson et al., 2007). Furthermore, it has been shown that PEA and TYR can be released from activated platelets (D’Andrea et al., 2003). Babusyte et al. (2013) also showed that these TAAR1 ligands, as well as 3IT, can act as chemoattractants for neutrophils, with subnanomolar EC₅₀ values. Such a chemoattractant action requires the presence of both TAAR1 and TAAR2, as evidenced by the selective small interfering RNA knockdown of individual receptors eliminating the chemotactic responses (Babusyte et al., 2013). Since neither PEA nor TYR is an agonist at TAAR2, a heterodimerization between TAAR1 and TAAR2 to initiate neutrophil migration has been hypothesized (Babusyte et al., 2013). Indeed, direct coimmunoprecipitation experiments indicated that such a heterodimerization is involved in the regulation of chemotactic responses to PEA (Babusyte et al., 2013). Taken together, the above studies suggest that TAAR2 is required for the expression of TAAR1-mediated recruitment of leukocytes to the sites of injury, and that this involves heterodimerization with TAAR1.

TAAR2 mRNA transcripts have also been reported in nonsensory cells in other mammalian species, in duodenal mucosal cells of the gastrointestinal system in mice (Ito et al., 2009), and in the rat heart and testis (Chiellini et al., 2012). Whether a similar pattern of expression is observed in human tissues requires further study. In a study with a small sample size, defunctionalizing single nucleotide polymorphisms of TAAR2 were reported in up to 10% of humans, with a further minor increase in patients with schizophrenia (Bly, 2005). These studies have not been replicated thus far and their clinical relevance remains unclear, especially given the lack of reports of TAAR2 presence in the brain.

Recent attempts have been made to apply structure-based in silico computational protocols to the development of TAAR2 homology models for the prediction of TAAR2 ligands (Cichero and Tonelli, 2017). With a consensus TAAR defining motif now identified, further development of such models is expected to provide new leads and the identification of TAAR2-selective ligands, which will be a major boost to elucidating its physiologic roles.

2. Trace Amine–Associated Receptor 3

TAAR3 (previously known as GPR57 or G protein–coupled receptor 57; Table 3) is a pseudogene in humans (Lindemann et al., 2005), although a functional protein with defined physiologic roles in olfaction is encoded in other species (discussed in detail in section IV.E.2).

3. Trace Amine–Associated Receptor 4

TAAR4 (previously known as trace amine receptor 2 or TA2, 5-HT₄P; Table 3) is also a pseudogene in humans (Lindemann et al., 2005) but encodes a functional protein in other species, where an overlapping ligand selectivity with TAAR1 is present (discussed in section IV.E.3).

4. Trace Amine–Associated Receptor 5

The TAAR5 gene encodes a functional protein of 337 amino acids in humans. The most prominent expression of TAAR5 is in the olfactory epithelium, where it plays a role in the detection of socially relevant odor cues (Li et al., 2013; Wallrabenstein et al., 2013; Zhang et al., 2013) (discussed in detail in section IV.E.4). In the olfactory system, TAAR5 appears to be coexpressed with G_olf and stimulates cAMP accumulation (Liberles and Buck, 2006). It has also been shown that human TAAR5 can couple to the G_s cascade (Wallrabenstein et al., 2013), the G_q/11 cascade (Dinter et al., 2015c), and G_12/13-dependent mitogen-activated protein kinase pathways (Dinter et al., 2015c), suggesting that in different cell groups TAAR5 might demonstrate functional selectivity by coupling to different signaling modalities. Whether such an effect is due to receptor heterodimerization, as seen with other TAAR isoforms, with different partners present in different cell types, or due to ligand bias, requires systematic study.

Beyond the olfactory system, low levels of TAAR5 mRNA have been reported in various leukocyte populations, with the greatest expression seen in B cells (Babusyte et al., 2013). Others, however, have reported that TAAR5 is only found in mouse, and not in human, leukocytes (Nelson et al., 2007). Expression of TAAR5 mRNA has also been reported in several mouse brain regions such as the amygdala, arcuate nucleus, and ventromedial hypothalamus, with an overlapping localization of TAAR1 and TAAR5 in the amygdala and ventromedial hypothalamus (Dinter et al., 2015c; Table 4). Spinal cord (Gozal et al., 2014), testis (Chiellini et al., 2012), and intestinal (Kubo et al., 2015) expression has also been reported in rats, although it should be noted that, at least with TAAR1, such expression patterns could not be validated with higher-quality reagents. Whether similar patterns of TAAR5 expression occur in humans requires further detailed studies.

As previously described, TAAR5 is predicted to be tuned to activation by tertiary amines (Ferrero et al., 2012), and trimethylamine has consistently been found to be the most active and selective agonist at human TAAR5, with dimethylethylamine a less potent partial agonist (Liberles and Buck, 2006; Wallrabenstein et al., 2013; Zhang et al., 2013). One group, however, found activity of trimethylamine only at rodent, and not human, TAAR5 (Stäubert et al., 2010). Several synthetic ligands and one additional natural ligand for TAAR5 have also been identified. Interestingly, the putative thyroid hormone metabolite and TAAR1 agonist 3IT has been reported to be a TAAR5 inverse agonist (Dinter et al., 2015c). The activity of 3IT, trimethylamine, and dimethylethylamine at TAAR5 has been confirmed by recent docking studies using homology models (Cichero et al., 2016). Furthermore, these studies identified two TAAR5 antagonists (Fig. 7), each of which was active at micromolar concentrations. The synthetic chemical 1-(2,2,6-trimethylcyclohexyl)hexan-3-ol (Timberol; Symrise AG, Holzminden, Germany) (Fig. 7) has also been identified as an antagonist of TAAR5 (Wallrabenstein et al., 2015). Although none of these antagonists have to date been tested in animals to better evaluate TAAR5 functionality, this is a considerable advance given the lack of good-quality antagonists at all other TAAR isoforms, making this is an exciting area for future development. A synthetic TAAR5-selective agonist, 2-(α-naphthoyl)ethyltrimethylammonium iodide (alpha-NETA) (Fig. 7), has also recently been described (Aleksandrov et al., 2018a; Aleksandrov et al., 2018b). Using the brain event-related potentials in the paired-click paradigm, a model that directly recapitulates sensory gating deficits seen in schizophrenia, alpha-NETA was shown to decrease sensory gating in rats (Aleksandrov et al., 2018a), suggesting that TAAR5 may be a novel molecular locus for sensory gating deficits seen in schizophrenia. Furthermore, alpha-NETA affected mismatch negativity-like response in rats, a cognitive paradigm known to be reflective of schizophrenia-related cognitive deficits described in experimental animal models and humans (Aleksandrov et al., 2018b). It is also worth noting that a protocol for the generation of large (milligram) quantities of human TAAR5 has been described (Wang et al., 2011), providing an additional approach to further understanding TAAR5 structure and functions.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Recently described synthetic TAAR5 ligands.

5. Trace Amine–Associated Receptor 6

The TAAR6 gene (previously known as TRAR4, TA4, or trace amine receptor 4; Table 3) encodes a functional human protein of 345 amino acids. Similar to TAAR2, no ligands for TAAR6 have been found thus far, although it is predicted to be activated by tertiary amines (Ferrero et al., 2012) and/or diamines (Li et al., 2015). The signal transduction events occurring at TAAR6 have also not been investigated but it is expected that like other olfactory TAARs, TAAR6 will be coupled to G_olf and subsequent cAMP accumulation. Beyond the olfactory system, TAAR6 mRNA has been found in mouse duodenal mucosal cells (Ito et al., 2009), in the rat spinal cord (Gozal et al., 2014) but not in the human spinal cord (Duan et al., 2004), and in the rat testis (Chiellini et al., 2012) (Table 4). TAAR6 transcripts have been reported in several human brain regions, including the amygdala and hippocampus (Borowsky et al., 2001; Duan et al., 2004), basal ganglia, and frontal cortex and substantia nigra (Duan et al., 2004), and these levels may exceed those of the most thoroughly investigated member of the TAAR family, TAAR1 (Duan et al., 2004). In the periphery, TAAR6 mRNA has been reported to be present in the human kidney (Borowsky et al., 2001) and in various human leukocyte populations (D’Andrea et al., 2003; Babusyte et al., 2013). In all cases, the functional consequences of TAAR6 activation are unknown. The flying fox TAAR gene family expansion (26 genes and 10 pseudogenes) is in part due to an expansion of its TAAR6 (four functional, six pseudogenes) complement (Eyun et al., 2016).

Despite minimal knowledge on the biology and physiologic role(s) of TAAR6, several studies have identified single nucleotide polymorphisms (Duan et al., 2004; Chang et al., 2015) and other genetic variants (Pae et al., 2010) in patients with schizophrenia (Duan et al., 2004; Pae et al., 2008a) and affective disorders (Abou Jamra et al., 2005; Pae et al., 2008b, 2010). Particularly interesting are the investigations of the potential role of TAAR6 single nucleotide polymorphisms in schizophrenia etiology and treatment, even though results of these studies are somewhat conflicting. Although several studies indicated a link in patients of Korean, European, and Afro-American origin (Duan et al., 2004; Vladimirov et al., 2007; Pae et al., 2008a,b), later studies failed to confirm associations in Japanese, Chinese, and European populations (Ikeda et al., 2005; Duan et al., 2006; Ludewick et al., 2008; Sanders et al., 2008; Vladimirov et al., 2009). TAAR6 polymorphisms have also been reported to be connected with therapeutic responses to the antipsychotic aripiprazole (Serretti et al., 2009), whereas a more complex epistatic relationship between TAAR6 and heat shock protein 70 polymorphisms has been associated with both the development of schizophrenia and treatment outcomes (Pae et al., 2009). TAAR6 variants have also been suggested to influence asthma patient responsivity to corticosterone (Chang et al., 2015).

6. Trace Amine–Associated Receptor 7

TAAR7 is a pseudogene in humans (Lindemann et al., 2005) but is a major site of mammalian species variation in other species. The increased TAAR repertoire in standard laboratory rodents—the mouse genome contains 15 functional receptors and one pseudogene, whereas the rat genome contains 17 functional TAAR genes and two pseudogenes (Borowsky et al., 2001; Lindemann et al., 2005; Lindemann and Hoener, 2005; Eyun et al., 2016)—is primarily due to an expansion of the TAAR7 subfamily. Likewise, the flying fox possesses a pronounced TAAR7 expansion with 16 functional variants (Eyun et al., 2016).

7. Trace Amine–Associated Receptor 8

The TAAR8 gene (previously known as GPR102, TRAR5, TAR5, TA5, or trace amine receptor 5; Table 3) encodes a functional protein of 342 amino acids in humans. At present, only tertiary volatile amine ligands, N-methylpiperidine and N,N-dimethylcyclohexylamine, for nonhuman TAAR8 isoforms have been identified (Ferrero et al., 2012; Li et al., 2015). TAAR8, including the human isoform, does contain a putative diamine binding domain (Li et al., 2015), suggesting the existence of diamine ligands as well and this is described in more detail in IV.E. Trace Amine–Associated Receptors in Olfaction.

Like other olfactory TAARs, TAAR8 is coupled to G_olf, stimulating cAMP accumulation (Li et al., 2015), but has also been reported to be coupled to G_i (Mühlhaus et al., 2014). Beyond the olfactory system, TAAR8 mRNA has been reported in the human amygdala (Borowsky et al., 2001) and leukocytes (D’Andrea et al., 2003; Babusyte et al., 2013; Table 4). Multiple isoforms of TAAR8 were found in various tissues from other species, including the rat cortex and cerebellum (Chiellini et al., 2012) and spinal cord (Gozal et al., 2014). In astroglia, expression was increased by lipopolysaccharide activation (D’Andrea et al., 2012). Low-level expression of the TAAR8b isoform was reported in various mouse brain regions (Mühlhaus et al., 2014), leukocytes (Nelson et al., 2007), kidney (Borowsky et al., 2001; Chiellini et al., 2012), heart, intestines, lung, skeletal muscle, spleen, stomach, and testis (Chiellini et al., 2012).

8. Trace Amine–Associated Receptor 9

The TAAR9 gene (previously known as TRAR3, TAR3, or trace amine receptor 3; Table 3) encodes a functional human protein of 348 amino acids. TAAR9 is thought to be tuned toward activation by tertiary amines, and like the case of TAAR8c, rat TAAR9 can be activated by N-methylpiperidine and N,N-dimethylcyclohexylamine, although only at micromolar concentrations (Liberles and Buck, 2006; Ferrero et al., 2012). It is notable that rat TAAR9 can also be activated by a currently unknown component(s) of both carnivore and noncarnivore urine (Ferrero et al., 2011), and this provides a good starting point for identifying putative endogenous ligands. As with other olfactory TAARs, TAAR9 is coupled to G_olf-mediated cAMP accumulation (Ferrero et al., 2011). Beyond its role in the mammalian olfactory system, TAAR9 mRNA has been found in the full range of human leukocytes (D’Andrea et al., 2003; Babusyte et al., 2013), in addition to the pituitary gland and skeletal muscle (Vanti et al., 2003) (Table 4). TAAR9 mRNA has also been reported to be present in the mouse gastrointestinal tract, where it is preferentially localized to duodenal mucosal cells (Ito et al., 2009), as well as the spleen (Regard et al., 2008) and spinal cord (Gozal et al., 2014). A loss-of-function mutation in the TAAR9 gene with unknown clinical relevance has been reported to be present in up to 20% of the human population (Vanti et al., 2003; Müller et al., 2010).

1. Trace Amine–Associated Receptor 2

Potent ligands for TAAR2 are yet to be identified, although several compounds that can activate this receptor at high concentrations were recently reported (Saraiva et al., 2016).

2. Trace Amine–Associated Receptor 3

Isoamylamine, a biogenic amine produced by leucine decarboxylation and known to be innately aversive to mice, was identified as an agonist of mouse TAAR3 (Liberles and Buck, 2006; Liberles, 2015), a receptor that is a pseudogene in humans (Lindemann and Hoener, 2005). Isoamylamine has been proposed to act as a pheromone in mice, inducing puberty in females (Nishimura et al., 1989), although this has also been questioned (Price and Vandenbergh, 1992). Intriguingly, strong class I major histocompatibility complex (MHC)–dependent female choice for genetically diverse and dissimilar males in the greater sac-winged bat (Saccopteryx bilineata) was recently correlated with the female TAAR3 genotype (Santos et al., 2016), indicating that TAARs and olfactory cues may be key mediators in mammalian MHC and immune system-based mate choice. The MHC-based ligand responsible for this response is currently unknown.

3. Trace Amine–Associated Receptor 4

TAAR4, which is a functional protein in many mammalian species but not in humans, has an overlapping pharmacological profile with TAAR1, being potently activated by the archetypal trace amine PEA (Liberles and Buck, 2006; Dewan et al., 2013). PEA is a component of urine, and analysis of 38 mammalian species indicated significantly higher PEA concentrations in carnivore urine, with some producing in excess of 1000-fold more than herbivores (Dewan et al., 2013). Mice, as well as rats, are known to innately avoid a PEA odor source (Ferrero et al., 2011) and enzymatic depletion of PEA from carnivore urine reduced the repellant properties of the urine (Dewan et al., 2013). Furthermore, TAAR4-KO mice show no avoidance response to either PEA or carnivore urine (Dewan et al., 2013). Thus, at least in mice, PEA activation of TAAR4 underlies innate avoidance responses to predator urine. Consistent with this, PEA olfactory exposure sufficient to induce an innate avoidance response has been reported to result in activation (as measured by increased c-fos expression) of the posteroventral region of the medial amygdala and the dorsomedial region of the ventromedial hypothalamus (Pérez-Gómez et al., 2015), brain areas implicated in fear, anxiety, panic, and defensive behaviors in rodents. Interestingly, PEA urinary levels have also been reported to be increased in response to stress (Grimsby et al., 1997; Paulos and Tessel, 1982); as such, PEA may serve as both a conspecific and heterospecific urinary warning cue. In tigers, PEA has also been proposed to serve a pheromone function (Brahmachary and Dutta, 1979), although the TAAR4 expression profile of tigers is currently unknown.

4. Trace Amine–Associated Receptor 5

Human, macaque, rat, and mouse TAAR5 are potently activated by trimethylamine (Liberles and Buck, 2006; Horowitz et al., 2014), which is most often regarded as a product of choline and/or l-carnitine metabolism in the body by the microflora of the gastrointestinal tract, oral cavity, and vagina, as described above (Fennema et al., 2016; Zhang and Davies, 2016). TAAR5 is also activated by spoiled fish, likely due to the presence of trimethylamine (Horowitz et al., 2014), and might serve as a mechanism for the innate avoidance of foods that could harbor pathogenic microorganisms and thus pose a danger to health.

In mice, trimethylamine is a sexually dimorphic odor that can evoke sex-specific, concentration-dependent behaviors (Liberles and Buck, 2006; Li et al., 2013), and this has led to suggestions that trimethylamine is a murine pheromone (Liberles, 2014). Trimethylamine is present in male mouse urine from where it is detected by TAAR5 (Liberles and Buck, 2006); enzymatic elimination of urinary trimethylamine disrupts behavioral responses to male mouse urine (Li et al., 2013). Male mice have particularly high concentrations of trimethylamine in urine (up to 5 mM; >1000-fold higher than in female mice or rats) due to a genetic deficiency in FMO3 (Li et al., 2013; Liberles, 2015) preventing its metabolism. Whereas rats (like humans) find trimethylamine innately aversive, mice demonstrate attraction to lower (urinary) concentrations, with aversion occurring to higher levels (Li et al., 2013; Liberles, 2015). Mice lacking TAAR5 demonstrate no attraction to trimethylamine but still avoid high concentrations (Li et al., 2013), suggesting that TAAR5 mediates the attractive responses due to high-affinity detection of trimethylamine, whereas a second, low-affinity, unknown receptor is responsible for the aversive behavior. Interestingly, specific anosmia for trimethylamine in humans was reported to be unrelated to polymorphisms in the coding sequence of the TAAR5 gene (Wallrabenstein et al., 2013).

5. Trace Amine–Associated Receptors 7, 8, and 9

It has been shown that unknown components of carnivore and noncarnivore urine can activate rat TAAR7f, TAAR8c, and TAAR9 (Ferrero et al., 2011). TAAR8c and TAAR9 can also be activated by N-methylpiperidine and N,N-dimethylcyclohexylamine, respectively (Liberles and Buck, 2006; Ferrero et al., 2012).

6. Teleost Olfactory Responses

The archetypal diamine cadaverine is a major component of rotting flesh that potently activates zebrafish TAAR13c present in olfactory sensory neurons that project to dorsolateral glomeruli of the olfactory bulb, evoking innate avoidance behavior (Hussain et al., 2013; Dieris et al., 2017). Structure-activity analysis indicated that TAAR13c has a preference for medium-sized diamines containing an odd number of carbons and contains a nonclassic monoamine recognition pocket with two distinct cation recognition sites (Hussain et al., 2013). Intriguingly, cadaverine also induces avoidance responses in mice, and this behavior is lost in mice with a cluster TAAR2-TAAR9-KO, suggesting that at least one of these receptors is responsible for the detection (Dewan et al., 2013). Interestingly, human TAAR6 and TAAR8, mouse TAAR6 and TAAR8b, and rat TAAR6 and TAAR8a also have TAAR13c-like diamine binding pockets and thus are likely candidates for cadaverine-detecting TAARs (Li et al., 2015). Ecologically relevant ligands for a number of other teleost TAARs have been identified (Table 2), although the physiologic responses to these ligands of other teleost TAARs have not yet been investigated (Li et al., 2015). As previously described, lamprey have also been reported to innately avoid a PEA source (Imre et al., 2014), although whether this is TAAR mediated is unknown.

A. Better Understanding of the Physiologic Role(s) of Trace Amine–Associated Receptors and Their Endogenous Ligands

B. Development of Selective Trace Amine–Associated Receptor 1 Ligands as Therapeutics

C. Trace Amine–Associated Receptors in Ecology

The demonstration that olfactory TAARs are involved in the initiation of innate behavioral responses to diverse environmental cues has generated interest in TAARs within the olfaction research community. Interest is also beginning to be shown by those studying ecology and there have been the first suggestions that TAARs may be involved in migration and mate choice of individual species. It is expected that these initial results will generate further interest that will allow further advances to be made in identifying innate behaviors that are controlled by olfactory TAARs. Such advances will almost certainly also lead to the identification of new TAAR ligands, which it is hoped will allow for the deorphaning of additional TAAR family members. With a select number of compounds that are associated with diverse ecological contexts and selective toward individual TAARs already identified, the olfactory TAARs offer an amenable system to probing the neural basis of survival behaviors associated with different ecological contexts. Currently, only one neural pathway involved in the innate behaviors initiated in response to olfactory TAAR activation has begun to be mapped. It is expected that this will be resolved in the coming years, providing a major breakthrough in understanding the cellular basis of the drives for various vertebrate behaviors.

With large variations in the complement of TAARs between species and activation of individual TAARs demonstrated in response to both conspecific and heterospecific urines, it seems clear that TAARs are involved in species-specific adaptation responses. As such, the evolution of TAARs appears to be a good option for studying the evolution of olfactory systems, at least with respect to innate survival mechanisms. From this perspective, amphibians are of particular interest because their olfactory systems have been suggested to represent an evolutionary midpoint between teleosts and mammals. There are few complete genome sequences available for amphibia and the TAAR complement of two Xenopus species are the only ones to have been reported. Unfortunately, Xenopus are evolutionarily rather distant from the true frogs (Ranitidae sp.). Determining the TAAR complement of representative true frog species would be advantageous and provide a starting point for addressing evolutionary questions about the origins of vertebrate olfactory systems.

One interesting aspect that the TAAR system may offer is the ability of a single cue to directly activate both centrally and peripherally mediated responses. Certainly, centrally mediated behaviors are induced by activation of TAARs in the olfactory passages. In addition, however, at least some TAAR ligands can readily cross cell membranes; with TAARs present throughout the body, direct activation of systemic responses may also be possible. In this regard, it is particularly interesting that the strongest evidence for peripheral TAAR responses so far is in cells controlling energy metabolism and those of the immune system. These are two systems that depending on the precise context would, on the surface, appear to be advantageous to activate in response to the possible presence of predators, pathogenic microbes (spoiled food), or in preparation for mating. Such a situation could provide an explanation for why the evolutionary origin of the entire family (TAAR1) is not involved in the detection of olfactory cues.

D. Trace Amine–Associated Receptors, Nutrition, the Microbiome, and Health

The roles of nutrition and the constitutive microbiota in health and disease are a growing area of research. The presence within various foodstuffs of compounds that are now known to be selective ligands for individual TAARs has long been known. Within the food science community, the presence of these compounds has been almost exclusively studied from the point of view of food spoilage and microbial contamination. With the growing recognition and acceptance of TAAR functions throughout the body, but particularly in the gastrointestinal tract and immune system, it is expected that over the next few years there will be a growing interest in whether there is a role to play for TAARs in both the beneficial and adverse effects of food in some individuals.

The search for metabolites that can link the constitutive microbiota to health outcomes has become an active area considerably aided by the development of powerful metabolomic techniques. With the levels of known TAAR ligands readily modified by the microbiota and a defined family of receptors present, it is expected that the trace amine system will develop into a locus for such studies. Besides the potential of the trace amine system providing a molecular pathway for the putative link between the microbiota, gastrointestinal tract, and central nervous system disorders, particularly intriguing is the molecular link that the trace amine system could provide between nutrient intake, the microbiota, and the immune system.