Norepinephrine and Serotonin Transporters

Eric L. Barker and Randy D. Blakely

The monoamine neurotransmitters norepinephrine (NE) and serotonin (5-hydroxytryptamine, 5HT) play important roles in an array of behaviors including sleep, appetite, memory, and mood. At a molecular level, monoamine signaling is dynamically regulated by a diverse set of macromolecules including biosynthetic enzymes, secretory proteins, ion channels, pre- and postsynaptic receptors, and transporters (). NE and 5HT are synthesized from simple amino acid precursors, packaged into synaptic vesicles, and released into the synapse in response to depolarizing stimuli, eliciting pre-and postsynaptic responses through receptor activation. Subsequently, transmitter is efficiently cleared from the extracellular space by transporter proteins localized in the plasma membranes of presynaptic terminals (). Transmitter reuptake is believed to have three important consequences. First, levels of transmitter in the synapse are reduced faster than can be achieved by simple diffusion, allowing for improved temporal discrimination of consecutive release events. Second, the effects of released transmitter are constrained to a smaller area, permitting dense packing of chemically identical but functionally distinct synapses. Finally, transmitter can be recycled for another round of release once it is transported back across the presynaptic membrane and into synaptic vesicles. The neurotransmitter transporter proteins that carry out NE and 5HT clearance are of much interest because they are molecular targets for many antidepressants, such as the tricyclics, fluoxetine, and sertraline as well as substances of abuse such as cocaine and amphetamines (1, 22, 30). Antidepressants and cocaine share the ability to alter neuronal signaling by blocking NE and 5HT transport. Unlike the antidepressants, cocaine also impedes dopamine (DA) clearance and thereby leads to a distinct spectrum of behavioral alterations (see The Dopamine Transporter: Potential Involvement in Neuropsychiatric Disorders). With transporter-mediated clearance pharmacologically inhibited, extracellular levels of transmitter remain elevated longer and can activate receptors at greater distances away from the synapse. Chronic alterations in transporter-mediated clearance may seriously compromise the ability of brain synapses to function properly. Changes in 5HT uptake sites, for example, have long been associated with major affective disorders, particularly depression (49). Thus, neurons may carefully set the level of NE and 5HT transporter expression to match synaptic demands for clearance, with a disruption in transporter expression or regulation providing a molecular liability to mental illness (see The Dopamine Transporter: Potential Involvement in Neuropsychiatric Disorders).

Although the involvement of transporters in NE and 5HT clearance has been clear for several decades (30), progress in understanding transporter structure and regulation has been limited, largely due to difficulties associated with transporter protein purification. Recently, expression and homology-based cloning efforts have (5, 28, 50, 52) revealed that NE transporters (NETs) and 5HT transporters (SERTs) are members of a large gene family comprised of carriers for other neurotransmitters, osmolytes, and nutrients (1). Expression of NETs and SERTs in non-neuronal cells has been achieved (5, 28, 50), establishing model systems for analysis of the structural basis of transporter specificity for neurotransmitters and antagonists. Indeed, it now appears that, unlike previous suggestions, most (if not all) antagonists block uptake by binding directly to the transporter protein. Mechanistic models for how transport is achieved are now more readily testable with overexpression in transfected cells providing opportunities for electrophysiological measurements of transporter turnover as a function of membrane potential, ion gradients, and cytosolic regulators. Furthermore, the availability of transporter-specific antibodies and nucleic acid probes has renewed interest in the endogenous control mechanisms acutely regulating NE and 5HT transport in vivo and whether chronic alterations in NET and SERT genes underlie neuropsychiatric disorders. Below we review our present understanding of NE and 5HT transport and the new insights achieved from molecular cloning and regulation studies.

It is important to understand the ionic and voltage sensitivities for transporters because these properties may dictate uptake rates, thereby controlling the speed by which transporters can clear NE or 5HT from synapses and ultimately the duration of postsynaptic responses. Although the speed of NE and 5HT clearance from central nervous system (CNS) synapses by NETs and SERTs has not been determined, elegant work on a voltage-dependent 5HT uptake process in leech serotonergic synapses (10) provides convincing support for the ability of neurotransmitter transporters to dictate the duration and magnitude of postsynaptic responses. One of the first common properties recognized for NE and 5HT transport was an absolute dependence on extracellular Na+ (), a feature now known to be characteristic of Na+/cotransport processes where energy for inward solute transfer is coupled to the energetically favorable influx of Na+ down its concentration gradient (for a comprehensive review of ion dependence of monoamine transporters, see ref. 58). In addition, extracellular Cl- is absolutely necessary for NE and 5HT transport (58), although the halide specificity appears less rigid than the Na+ requirement: Other anions such as NO-2 and Br- are capable of substituting, at least partially, for Cl-. For SERT, a model for this multisubstrate uptake process has been proposed in which Na+, Cl-, and a protonated 5HT molecule binds to the transporter, forming a quaternary complex which then undergoes a conformational change to release the neurotransmitter and the ions into the cytoplasm (58). Subsequently, intracellular K+ has been proposed to associate with SERT and promote the reorientation of the unloaded carrier for another transport cycle. Thus, intracellular K+ accelerates 5HT influx, presumably by facilitating a conformational change required for external exposure of unoccupied 5HT binding sites on the unloaded transporter (58). Surprisingly, SERT and NET appear to differ in the role of intracellular K+ in transport (); NE influx also is sensitive to intracellular K+, but available data are more consistent with K+ stimulation at a modulatory site rather than direct K+ efflux as observed with SERT. If such a difference exists, the net charge movement per cycle for NETs and SERTs would be distinct and might lead to differences in the voltage sensitivity of transport. Among other transporters in this gene family, sensitivity to intracellular ions such as K+ seems to be far more variant than extracellular ion requirements; the homologous GABA transporter (GAT), for example, appears to lack sensitivity for intracellular K+ altogether (32), suggesting that structural similarities among transporter family members may mask important mechanistic distinctions acquired in the evolutionary divergence from a common ancestral transporter. Recent progress in recording NET and SERT transporter currents by whole-cell patch-clamp techniques (L. DeFelice and R. Blakely, unpublished results) suggests that both carriers move net charge per transport cycle in a voltage-sensitive manner.

Another consequence of neurotransmitter uptake driven by transmembrane ion gradients is transporter-mediated efflux of substrate, as well as influx, if ion gradients are reversed and/or membranes are depolarized (43). For example, reversed NE transport from sympathetic nerves accompanies moderate periods of cardiac ischemia when internal Na+ concentrations rise and the plasma membrane becomes depolarized, possibly contributing to fatal cardiac arrhythmias (43). By analogy, adrenergic receptors might be bathed inappropriately with NE during CNS ischemia accompanying stroke. As transporter-mediated efflux is antagonized by transporter inhibitors, these antagonists may find new uses in the prevention of inappropriate transmitter efflux until normal ion gradients and resting potentials can be restored.

As mentioned previously, SERTs and NETs are the pharmacological targets for a variety of therapeutic antidepressants and abused substances (). Tricyclic antidepressant sensitivity is shared by NETs and SERTs, but not by DA transporters (DATs; see The Dopamine Transporter: Potential Involvement in Neuropsychiatric Disorders) (5, 28, 37, 50). Tertiary amine tricyclics (imipramine, amitriptyline) are more potent at SERTs as compared to the NET-preferring secondary amine tricyclics desipramine and nortriptyline. The steric interactions by which the addition of a single methyl group increases potency of the tertiary amines for SERT are not known; however, mutagenesis of the SERT protein should prove useful in identifying residues important in this effect and allow predictions concerning binding of ligands to the transporter (see below). Other potent NET antagonists include nomifensine, mazindol, and nisoxetine. Highly selective antagonists for SERTs such as paroxetine and fluoxetine have been developed whose chemical structures differ from the tricyclic nucleus, but which are effective antidepressants supporting alterations in serotonin neurons as targets in affective disorders (22). Cocaine is a nonselective, competitive antagonist of NE, 5HT, and DA transport. The addictive potential of cocaine is though to be a consequence of actions on CNS DATs, whereas the life-threatening cardiovascular effects of cocaine may involve blockade of NETs at sympathetic and CNS autonomic synapses. In addition to cocaine, other drugs of abuse including p-chloroamphetamine, fenfluramine, and (3,4-methylenedioxy)methamphetamine (MDMA, "ecstasy") also are inhibitors of 5HT uptake. Interestingly, MDMA and the other amphetamines are neurotoxic substrates for SERTs and additionally cause efflux of 5HT by a transported-mediated exchange process (59).

Like the binding of substrates, antagonist binding to NETs and SERTs also is dependent on extracellular Na+, although ion dependency appears to be complex and varies for different antagonists. For example, the binding of [3H]desipramine (39) and [3H]mazindol (31) to NETs is Na+- and Cl--dependent, suggesting that these antagonists bind to the carrier in the conformation by which NET recognizes substrates. For SERT, the binding of [3H]imipramine appears to require the presence of 2 Na+ ions per imipramine molecule (30), whereas the Na+ dependence of other 5HT transport antagonists varies depending upon the compound (29). Perhaps all ligands share a single Na+-modulated binding site, with additional Na+-linked sites recruited in an antagonist-specific manner. The cocaine analogue 3b-[4-[125I]iodophenyl]tropan-2b-carboxylic acid methyl ester ([125I]RTI-55) labels SERTs and NETs in platelet membranes in a largely Na+-dependent manner (66); however, in contrast to other ligands, RTI-55 binding is independent of Cl-. Unlike imipramine binding, RTI-55 binding to SERTs is markedly pH-sensitive, perhaps indicating the influence of titratable amino acid side chains near the cocaine binding pocket. These data suggest that cocaine and its congeners interact with SERT and NET in a manner distinct from substrates and antidepressants, possibly reflecting contact sites outside of the substrate binding pocket. Given the structural differences between the simple phenylethylamine and tryptamine substrates and the heterocyclic antagonists such as cocaine and antidepressants, it seems naive to assume complete correspondence between substrate and antagonist transporter contact points.

While antidepressants and cocaine are thought to be competitive antagonists of NETs, other ligands which antagonize NE transport and [3H]desipramine binding, such as phencyclidine (PCP) and the sigma opiate ligands, may do so through allosteric mechanisms. At present, it is not clear whether this antagonism is mediated directly by recognition sites on NET or through the aegis of a distinct protein. For example, sites labeled by [3H]desipramine in adrenal medulla are reported to be displaced noncompetitively by the sigma ligands haloperidol and (+)-3-PPP as well as by the PCP1-selective ligand MK-801 (56). The size of PCP/sigma receptor targets (21–33 kD) from photoaffinity-labeling and purification studies (65) appears to be different from that expected for NET (45, 50); however, the rank order of potency for antagonism of [3H]desipramine binding by sigma ligands does not match any characterized sigma or PCP receptor subclass (56). Clarification of the commonality and divergence of these sites using cloned proteins could reveal novel strategies for the therapeutic blockade of NETs, particularly if allosteric modulation of transport is involved. Thus, antagonists which block the actions of cocaine yet permit substrate translocation might be powerful tools in the fight against drug abuse.

High-affinity neuronal NE transport was first described in postganglionic neuronal terminals of sympathetically innervated peripheral tissues including the heart, spleen, and vas deferens (then termed "Uptake 1"; see ref. 30) before being identified in CNS nerve terminal preparations. Neuronal lesions in periphery and brain support a localization of NETs to innervating terminals rather than targets. Kinetically indistinguishable NE transport activities also are present in adrenal chromaffin cells, lung, and placental brush-border membranes. However, NE accumulation alone is not a definitive measure of the noradrenergic character of the preparation under study because NE also is a substrate for the DAT (1). In fact, DA is actually a better substrate (lower Km) than NE for both the NET and DAT, presenting a problem. How do we determine which transporter is responsible for the uptake activity present in a particular tissue? Exploitation of the pharmacological selectivity of transporter antagonists has, until recently, been the principal means of addressing this question. Catecholamine transport into terminals of peripheral sympathetic and CNS noradrenergic neurons is abolished by nanomolar concentrations of the tricyclic antidepressant desipramine; catecholamine uptake in striatal synaptosomes, where DA is the primary neurotransmitter, is largely insensitive to antidepressants yet sensitive to a number of cocaine derivatives (e.g., GBR12909 and GBR12935) which lack potency for NETs (1,50). These pharmacological distinctions, among others, originally helped establish brain NETs and DATs as separate activities, likely to be the products of distinct transporter molecules, and identified NETs as the most prevalent catecholamine transporter in the periphery (30).

Whereas transport measurements provide a sensitive measure of the capacity for catecholamine accumulation, more direct approaches are required to identify and quantitate the distribution of NETs in situ. Historically, the most informative strategies have relied upon the binding of selective radiolabeled antagonists to brain sections and the subsequent anatomic localization of bound ligand by autoradiographic techniques. Initially, [3H]desipramine and [3H]mazindol were exploited as ligands to identify sites of NETs in rodent brain membranes and slices (37); however, [3H]desipramine exhibits high nonspecific binding and [3H]mazindol labels both NE and DA uptake sites. More recently, [3H]nisoxetine has been introduced as a potent and selective agent for labeling NE transport sites (61). Studies by Tejani-Butt (61) demonstrate (a) a high density of [3H]nisoxetine binding sites in rat brain regions containing a high density of noradrenergic soma or terminals, including the locus coeruleus and hypothalamic nuclei, and (b) a low density in regions receiving sparse noradrenergic innervation, such as the striatum (61). A marked loss of [3H]nisoxetine labeled sites occurs following chemical brain lesions with the neurotoxins 6-hydroxydopamine (6-OHDA) and DSP-4, indicating that forebrain labeling is most likely associated with noradrenergic terminals rather than targets or surrounding glia, although a small perisynaptic contribution which disappears with loss of innervation cannot be excluded.

Sites of 5HT uptake have been similarly characterized through the binding of [3H]imipramine, which bears moderately higher affinity for SERTs than for NETs (). However, because of the high nonspecific binding of [3H]imipramine in brain preparations, more recent investigations have utilized more selective ligands such as [3H]paroxetine, [3H]citalopram, and [3H]nitroquipazine. Autoradiographic studies using [3H]citalopram and [3H]imipramine (16) identify the amygdala, thalamus, hypothalamus, CA3 region of the hippocampus, substantia nigra, locus coeruleus, and the raphe nuclei of the midbrain as the rat and human brain regions with the highest level of 5HT uptake sites. As with NET localization, toxininduced lesions of serotonergic neurons in rat brain reduce antidepressant binding sites in raphe projection areas consistent with a presynaptic origin of SERTs in vivo (16). Whereas SERT expression appears to be confined to neurons in the adult CNS, antidepressant-sensitive 5HT uptake has been reported for primary cultures of astrocytes (33). However, until SERT expression in glia can be directly established in vivo, reports of glial 5HT uptake may represent an altered or embryonic phenotype of cells in primary culture.

In addition to neuronal expression, 5HT uptake has been identified in platelets, placenta, pulmonary endothelium, and mast cells (21). In the lung, 5HT transporters efficiently clear plasma-borne 5HT and, with the help of platelets, keep blood levels of free 5HT low. Placental SERTs may protect the heavily vascularized tissue from premature constriction arising from circulating maternal 5HT. Platelet SERTs have become a widely used peripheral index of central serotonergic neuronal systems (for review, see refs. 37, and 49). Although not a universal finding, many studies have found decreased activity and density of platelet SERTs in depressed patients (37, 49). The validity in extrapolating platelet SERT data to disturbances in CNS SERT levels or function assumes identical SERTs in the CNS and periphery with a parallel responsiveness to alterations induced by psychiatric disease. Recent advances in the molecular biology of SERTs has begun to clarify this clinically important issue (see below). Because of the absence of a facile peripheral measure of the NET similar to the SERT-expressing platelet, NET alterations in depression or other affective disorders has received far less attention. Because NET-selective antagonists also have antidepressant activities, further studies are warranted to quantitate potential alterations in NETs accompanying mental illness (see Selective Serotonin Reuptake Inhibitors in the Acute Treatment of Depression and Standard Antidepressant Pharmacotherapy for the Acute Treatment of Mood Disorders).

Protein purification strategies have achieved only limited success in the elucidation of neurotransmitter transporter structure; therefore, to clone transporter cDNAs in the absence of protein-derived sequence information, Pacholczyk et al. (50) turned to expression cloning in COS cells where transport activity induced by cDNA clones could be efficiently screened. These efforts led to the cloning of the first monoamine transporter, the human NET, from the SK-N-SH neuroblastoma cell line. The primary sequence of the human NET cDNA predicts a highly hydrophobic 617-amino-acid polypeptide of ~67 kD with 12 potential transmembrane domains (TMDs) (). Although there is little amino acid sequence homology between NET and the facilitated or Na+-coupled glucose transporters, the 12 TMD structure appears as a common motif for transporter proteins. Indeed, this multiple transmembrane domain motif is common to membrane-associated proteins, particularly those responsible for ion and solute transport. Comparison of the predicted primary amino acid sequence for the human NET with that of a previously cloned GABA transporter (GAT) revealed 46% absolute identity, establishing the existence of an Na+/Cl--dependent neurotransmitter transporter gene family. The sequence similarity evident between NET and GAT cDNAs provided a route to clone SERTs from rat brain (5) and a rat basophilic leukemia (RBL) cell line by homology screening (28). Despite sequence differences in the original reports, both brain- and RBL-cell-derived rat SERT cDNAs now are known to encode a GAT/NET homologue of 630 amino acids (6, 52) with a predicted size of ~68 kD and 50% identity to human NET. Furthermore, the cloning of the identical SERTs from rat brain and a rat mast cell line provides strong support for identical peripheral and CNS SERTs, particularly important in light of the frequent use of human platelet SERTs as a diagnostic tool for psychiatric disease (37, 49). Recently, Ramamoorthy et al. (52) have identified a functional human SERT, which bears 92% sequence homology with the rat SERT, and have verified its functional properties in transfected cells. A murine SERT also has recently been cloned and characterized (13).

The cloning of NET and SERT cDNAs has provided important new tools to define the tissue and cellular distribution of transporter gene expression. Thus, in situ hybridization studies on rat brain sections hybridized with specific NET oligonucleotide (20) and cRNA (44) probes confirm the synthesis of NETs in brainstem nuclei, thereby corroborating radioligand binding data observed using [3H]antidepressants. However, autoradiography of [3H]ligand binding does not possess sufficient spatial resolution to resolve actual membrane sites of NET expression. Thus, the high density of NETs reported in noradrenergic nuclei may arise from carriers expressed on the cell soma, dendrites, or axons, and it also could reflect intracellular access of ligands. Antibodies that specifically recognize NETs have been developed (45) and are currently being used to discern whether these transporters are tightly localized near neurotransmitter release sites on presynaptic membranes or whether they are distributed uniformly across axonal and dendritic membranes. Presumably, for clearance of NE to affect spatial and temporal aspects of synaptic signaling, NETs should be present at or near sites of release and response. A more diffuse membrane distribution on axons or surrounding glia might indicate a cooperative role among NETs on adjacent terminals to provide a more compartmental reduction of extracellular NE levels.

Examination of SERT distribution by Northern blot hybridization of rat and mouse RNAs reveals single SERT mRNA species in brainstem, midbrain, lung, spleen, gut, and adrenal gland (5, 13, 28). In situ localization of SERT mRNA in both rodent and human brain demonstrates prominent expression in the serotonergic neurons of the median and dorsal raphe nuclei (3, 5, 13, 28). Interestingly, Lesch et al. (41), using sensitive polymerase chain reaction techniques, have obtained evidence for low levels of SERT mRNA in forebrain regions lacking serotonergic soma, although the cellular sites for this expression remain unknown. Anti-SERT antibodies, which detect the transporter in vivo (46), provide a means to assess the legitimacy of proposed nonneuronal SERT expression in the CNS. As expected, antibody staining of rat brain sections reveals abundant SERT expression in the raphe neurons and within serotonergic axons and varicosities, consistent with autoradiographic data; however, no SERT immunoreactivity is detected in surrounding glia (R. D. Blakely, unpublished results).

Comparisons of the predicted sequences of the cloned NET and SERTs reveal multiple topological similarities, many of which are shared by other GAT/NET gene family members as well. The NH2 termini of NETs and SERTs lack hydrophobicity characteristics of a signal sequence for membrane insertion and bear no asparagine-linked glycosylation sites; thus, as for many other transport proteins, the NH2 termini are predicted to reside in the cytoplasm. An even number of TMDs following the NH2 terminus places the COOH terminus also in the cytoplasm (). Following TMD3, both SERT and NET contain a large extracellular loop with multiple canonical asparagine-linked glycosylation sites, suggesting an extracellular localization of this domain. Glycosylation of membrane proteins can contribute to their folding, stability, trafficking, or ligand recognition. Human NETs undergo sequential glycosylation in transfected cells, modifications that appear critical for transporter stability and/or trafficking (45), similar to observations for b-adrenergic receptors (36). Analogous studies with recently developed anti-SERT antibodies confirm that both native and transfected SERTs exist as glycoproteins with asparagine-linked sugars. Furthermore, preliminary studies with these antibodies suggest that brain and platelet SERTs may undergo differential post-translational modifications. Thus, while the primary amino acid sequences of CNS and peripheral SERTs appear identical, tissue-specific structural modifications are apparent, suggesting caution in absolute extrapolation of platelet SERT studies to CNS SERTs in psychiatric disease. Additionally, conserved cysteine residues are located within the large extracellular loop between TMDs 3 and 4 and may serve, via potential disulfide bridge formation, to maintain a functional conformation of the protein as has been shown for the nicotinic acetylcholine receptor (7). Protein phosphorylation at serine, threonine, and tyrosine residues is involved in the acute regulation of many proteins. The putative cytoplasmic domains of NETs and SERTs contain multiple, canonical sites for serine/threonine phosphorylation (), and acute regulation of transporter activity by exogenous hormones has been reported (see below); however, validation of the phosphorylation of any of these sites has not been achieved, limited largely by an absence of purification methods suitable for phosphoprotein analysis. Progress on NET and SERT phosphorylation as well as other post-translational modifications (e.g., acylation) should be accelerated by the recent development of antibodies capable of immunoprecipitating NET and SERT proteins (45, 46).

Because the transport activities conferred by cloned transporter cDNAs is inhibited by tricyclic antidepressants, amphetamine, and cocaine, over appropriate concentration ranges, the cloning of NET and SERT defined each transporter's primary structure and established a single protein subunit as competent for drug recognition as well as transport activity; however, still lacking is a knowledge of the native stoichiometry of transporter monomers and whether accessory proteins are organized with the transporters in an active complex. Site-directed mutagenesis of the transporter proteins leading to altered or abolished ligand binding supports the cloned transporters as direct targets for antagonists (35; R. D. Blakely, unpublished results). In support of a larger functional entity, native placental SERTs sized by gel filtration are much larger than the size of a single SERT monomer (54); however, the contributions of detergent and lipid to these estimates cannot be completely discounted. Expression of NETs and SERTs in non-neuronal hosts, yet bearing drug sensitivities similar to those of transporters found in native membranes, suggests that if any additional subunits exist, such accessory proteins are likely to be modulatory rather than essential to function.

The shared structural features visible in a comparison of NET and SERT sequences subsequently has been observed in a large number of related transporter homologues. These include carriers for DA, glycine, taurine, proline, creatine, and betaine transporters (1), among others (). Estimates from genomic hybridization studies suggest as many as 30 members within the GAT/NET gene family; indeed, additional GAT/NET homologues have been cloned by several laboratories that, at present, lack defined substrates. The Na+/L-glutamate and Na+/glucose cotransporters reside in distinct gene families unrelated to the GAT/NET group (1). Interestingly, these latter transporters do not exhibit a requirement for extracellular Cl-, providing one mechanistic feature likely to help in the categorization of as yet uncloned transport activities. Approximately 40% identity is detected in 2 {ewc MVIMG, MVIMAGE,!times.bmp} 2 comparisons of human NET with other GAT/NET homologues; DA, NE, and 5HT transporters are most closely related to each other, defining a small subdivision united by both sequence and pharmacology (). At the primary amino acid level, approximately 20% of NET and SERT residues are absolutely conserved across all GAT/NET family members. Interestingly, this sequence identity is distributed nonuniformly, with particularly high conservation evident in TMDs 1–2 and 5–8 (1, 5). These shared residues most likely represent sites involved in common properties such as Na+ and Cl- binding; alternatively, they may dictate the global architecture required for substrate translocation across the plasma membrane.

One feature which separates the NE, 5HT, and DA transporters from other GAT/NET homologues is high-affinity recognition of the psychoactive agents cocaine and amphetamine, and, for NET and SERT, the tricyclic antidepressants. Although the sites of antagonist recognition have yet to be directly established, the expression of cloned and mutant NETs, SERTs, and DATs in transfected cells is rapidly illuminating functional properties of shared structural features. Sequence comparisons show the NH2 and COOH termini to be poorly conserved and, thus, potentially involved in unique attributes of each carrier. However, chimera studies wherein these domains are swapped between NETs and SERTs reveal no alteration in substrate or antagonist selectivity (6); furthermore, removal by mutagenesis of GAT tails fails to alter transport properties (4), suggesting NH2 and COOH termini to be inconsequential for ligand recognition. The NH2 and COOH tails, which contain multiple sites for protein phosphorylation, may comprise regulatory domains specific for each transporter, whereas the regions comprised of transmembrane domains between these tails are likely to contribute to the formation of the translocation pore and antagonist binding sites.

The simple structure of the neurotransmitters NE and 5HT invites comparisons between modes of monoamine recognition by receptors and transporters. In adrenergic receptors, the protonated NH2 group on NE is believed to ion pair with an intramembrane aspartate residue, while catechol OH groups form hydrogen bonds with serine residues on a nearby TMD (36). Likewise, structural features of catecholamines required for high-affinity recognition by NETs and SERTs confirm the importance of ring hydroxyl groups and a protonated, unsubstituted NH2 group (30). Furthermore, TMD aspartate and serine residues are among the handful of residues specific to the NET, SERT, DAT wing of the GAT/NET gene family, raising possibilities of similar strategies for the recognition of neurotransmitters by transporters and receptors (35). Site-directed mutagenesis of the TMD1 aspartate residue in NET, SERT (R. D. Blakely, unpublished results), and DAT (35) markedly alters substrate and antagonist recognition. For DAT, retention of low-affinity cocaine binding and substrate recognition suggests a selective alteration in the ligand binding pocket by the aspartate mutation rather than a gross destabilization of transporter protein. Antibody studies reveal no difference between wild-type and TMD1 aspartate-mutant NET and SERT proteins, similarly consistent with a direct alteration at the binding site rather a global modification of protein stability. Nonetheless, the idea that the TMD1 aspartate residue directly coordinates the catecholamine NH2 group via salt-bridge formation is problematic in light of the fact that other GAT/NET proteins have an uncharged glycine residue in the position occupied by the NET, DAT, and SERT aspartate, and yet they bind substrates with protonated NH2 groups. Moreover, what is unique about NE, DA, and 5HT as GAT/NET substrates is not the presence of a protonated NH2 group to possibly bind the TMD1 aspartate, but rather the absence of an acidic side chain linked to the substrates' a-carbon. One or more of the residues shared by NET, SERT, and DAT but not by other GAT/NET homologues, such as the TMD1 aspartate, may represent a binding site for the decarboxylated catecholamine and indoleamine substrates. Finally, the differential drug sensitivities observed between species variants of the same transporter may assist in the search for molecular determinants of antagonist binding. For example, the recent identification of a Drosophila SERT that transports 5HT selectively but recognizes NET antagonists like mazindol with high affinity (15) may provide important clues to how antagonists bind and block transport.

Are all endogenous NETs and SERTs identical to the transporters identified by recent cloning, or do multiple subtypes of the transporters exist as has been found for other membrane proteins such as ion channels and receptors? Virtually identical transport activities can arise from more than one gene product, suggesting caution in extrapolating beyond available data. For example, distinct genes are known to encode two functionally similar vesicle transporters responsible for intracellular NE packaging in adrenal gland and neural tissues (17). Multiple genes also encode several pharmacologically distinct GABA transporter isoforms (8) (). Tissue-specific differences in ligand recognition by NETs have been reported (30), although these discrepancies may reflect assay differences more than the properties of NET structural variants. A single human genomic locus for NETs has been identified, spanning a locus of least 10 kB on chromosome 16q12.2 (11) with multiple introns evident. Multiple mRNA species are revealed by human NET cDNA probes in SK-N-SH and PC-12 cells as well as by rat and human primary tissues (44, 50, 55). It is possible that these multiple RNA species reflect differential use of noncoding elements in the NET mRNAs synthesized and have little or no impact on the structure of the expressed protein. Coding region variants also could be derived from the single human NET gene. For example, alternative splicing of RNAs derived from a single glycine transporter gene (represented in ) results in two transporters with different NH2 termini, but indistinguishable transport activities (9). Further studies on NETs cloned from different sources should aid in the evaluation of NET variants.

The identification of a- and b-adrenergic receptor subtypes which show pharmacological selectivity for NE and epinephrine, respectively, raises questions of whether epinephrine-selective transporters exist within the GAT/NET gene family. Epinephrine can be accumulated by sympathetic terminals through NETs, although its affinity as a substrate is lower than that of NE. To date, no data support a unique epinephrine carrier in the mammalian peripheral nervous system. Indeed, the hormonal nature of epinephrine in the peripheral nervous system would not appear to require a specialized transport system because epinephrine's actions are not spatially constrained to the same degree as catecholamine release at synapses. However, studies with antibodies to phenylethanolamine N-methyl transferase (PNMT), the enzyme that converts NE to epinephrine, reveal the presence of putative epinephrine-synthesizing neurons and terminals within the rodent and human brainstem. Unlike other brainstem neurons that secrete NE, these PNMT-positive neurons do not express NET mRNAs (44). Perhaps PNMT-positive neurons synthesize a unique NET homologue specialized to retrieve epinephrine at synaptic sites. Such a transporter has actually been reported at amphibian sympathetic synapses. Sympathetic terminals in the frog release epinephrine rather than NE and express a desipramine-sensitive catecholamine uptake activity (37). The frog transporter, unlike the mammalian NET, prefers epinephrine over NE as a substrate, and can be detected with [3H]desipramine in radioligand binding assays. Could some of the nonspecific effects of antidepressants be due to antagonism of epinephrine transporters on epinephrine terminals innervating brainstem autonomic control centers? Because the physiological relevance of PNMT expression in brainstem neurons is controversial, cloning and characterization of mammalian epinephrine transporters, if indeed these proteins exist, could provide new tools to investigate the role of epinephrine as a transmitter in the mammalian brain.

The search for multiple subtypes of SERTs commands special attention due to the frequent use of platelet SERTs as a peripheral marker of CNS 5HT neurons. Using the polymerase chain reaction, Lesch et al. (42) have amplified identical SERT cDNAs from human platelet and brain mRNAs consistent with the aforementioned single identity of peripheral and CNS rodent SERTs, at least at the amino acid level. While tissue-specific post-translational modifications of SERT may occur, these data support the derivation of human peripheral and CNS SERTs from a common gene. Whether platelet and brain SERTs are regulated equivalently in psychiatric disease, however, remains to be determined, an issue that must be resolved if the diagnostic use of platelet SERTs is to continue as a window into CNS serotonergic function. In the rat, a single RNA species hybridizes to SERT cDNA probes on Northern blots of various tissues, whereas at least three SERT mRNAs are detected on blots of human tissues and cell lines (3, 52). The significance of these multiple human SERT RNAs is presently unknown, but, like the human NET gene, the human SERT gene (located on chromosome 17q11.1–17q12) possesses multiple exons (52). To date, no subtypes of SERT have been identified, but future progress into potential SERT structural variants should be accelerated by the recent cloning of the human SERT genomic locus (R. D. Blakely, unpublished results).

Many aspects of noradrenergic neurotransmission, including biosynthesis and release of transmitter, are tightly regulated. Given that transporters control the temporal aspects of transmitter actions after release, it would not be surprising that NETs also are subject to acute regulation by membrane and cytoplasmic factors. Because transport activity is a function of neurotransmitter concentration, synaptic NET activity is expected to increase as the concentration of extracellular NE increases until NETs reach maximal velocity at saturating NE concentrations. The kinetic properties of NETs in native and transfected cells demonstrate a Km of ~0.5 mM, indicating transporter saturation at low micromolar concentrations. Synaptic concentrations of NE at peripheral synapses, which generally have wide synaptic spaces, may reach high micromolar levels; even higher concentrations may be reached transiently in the more confined synaptic spaces in the CNS. At saturation, NETs must either be converted to a more active state or be joined by other NETs previously held in intracellular compartments; otherwise, synaptic recovery will not keep pace with release, NE may spillover to extrasynaptic sites, and less NE will be recovered for repackaging. Alternatively, an alteration in the activity or number of NETs at noradrenergic synapses could be utilized to alter the level and lifetime of synaptic NE independent of control mechanisms regulating release. Gillis first demonstrated a rapid increase in the retention of NE by cat atrium after stimulation of the heart's sympathetic innervation (25). Rorie et al. (57), utilizing electrical stimulation of adrenergic fibers in the dog saphenous vein and measurement of both NE overflow and metabolite production, detected enhanced NE transport paralleling the increase in impulse flow. Similar findings have been reported by Eisenhofer et al. (18) using pharmacological manipulation of peripheral sympathetic neurons in unanesthetized rabbits in vivo. These studies indicate that NE uptake increases in parallel with increased firing rate and release. Whether the increased NET activity simply reflects an increased turnover of a fixed number of NETs as substrate levels rise or involves an alteration in NET density remains to be determined. Sustained elevation of intracellular Ca2+ following repetitive terminal depolarization could provide an intrinsic signal to move transporters from subcellular sites to the terminal membrane.

NE release is altered by a number of endogenous agents which act on presynaptic terminals; likewise, acute hormonal regulation of NET activity has been reported. Multiple groups (e.g., see refs. 60 and 63) have found NETs in central and peripheral nervous systems to be sensitive to angiotensin peptides. Angiotensin II or III typically reduces transport and increases release, although acute stimulation of transport has been reported for rat brainstem neuronal cultures (60). Atrial natriuretic peptide has been reported to acutely elevate NET activity and reverse the inhibitory effects of angiotensin II and III at dosages subthreshold for its own response (62, 64). Insulin, in rat brain synaptosomes and PC12 cells, produces a rapid (within 1 min), dose-dependent reduction in NE uptake (19, 51). In synaptosomes, insulin reduces NET Vmax with no effect on Km, consistent with either a reduction in surface pools of NETs or an alteration in capacity of a fixed number of transporters (51). Insulin's effects are observed in PC12 cells despite reserpine depletion of vesicular catecholamine stores and thus are not likely to arise from alterations in DA or NE release (19). Although the intracellular effectors of these peptide responses are not known, direct treatment of bovine adrenal chromaffin cells in vitro with agents that elevate or mimic intracellular cAMP is reported to reduce NE uptake (12). Clearly, the presence of serine/threonine phosphorylation sites on human NETs raises questions as to whether any of these effects are mediated by protein phosphorylation. Unfortunately, only a few studies report a kinetic basis for hormone-altered transport, and the presence of an NE release pathway (often modulated in parallel) confounds analysis. Use of purified NET proteins and stably transfected cell systems should help clarify these effects and permit a direct evaluation of NET protein phosphorylation.

Hormones may also modulate NE uptake capacity by altering NET gene expression. For example, Figlewicz et al. (20) have shown that chronic intraventricular administration of insulin to rats in vivo significantly reduces steady-state levels of NET mRNA in the locus coeruleus. In vitro, chronic insulin treatment reduces NE uptake and levels of desipramine-labeled NETs in PC12 cells (19). Pertussis toxin treatment of chromaffin cells also appears to modulate NET expression in a delayed fashion most compatible with reduced gene expression or mRNA stability (12). Regardless, these studies indicate that expression levels of NETs can be modulated by external hormonal influences; these findings are of potential clinical relevance particularly where endocrine dysfunction is suspected. Indeed, sympathetic NETs have been reported to be up-regulated in human diabetic cardiomyopathy (24) as well as in rodents where insulin-secreting b cells have been destroyed with streptozotocin (23). NETs also can be regulated in parallel with tissue NE concentrations: Depletion of brain NE with reserpine causes a decrease in uptake sites, whereas increased synaptic NE availability due to monoamine-oxidase inhibitors results in an increase in sites as labeled by [3H]desipramine (40). An inability of the NET gene to respond to hormonal cues in CNS neurons could result in inappropriate levels of synaptic NE clearance and improper receptor stimulation, precipitating behavioral disturbances. Although this idea is clearly speculative, DNA and antibody probes are now available to examine human NET protein and gene regulation in the context of human neuropsychiatric disorders.

Analysis of the amino acid sequences of the cloned human SERT reveals six potential sites of phosphorylation by protein kinase A and protein kinase C (52); five of these recognition sites also are conserved in rat SERT (5,28). Acute and chronic regulation of SERT by protein kinase C and cAMP have been reported, possibly involving one or more of these potential phosphorylation sites. Activation of protein kinase C with phorbol esters causes a dose-dependent inhibition of SERT activity in bovine pulmonary endothelial cells, platelets, and RBL cells that is blocked by protein kinase C inhibitors (2, 47, 48). However, in the RBL cell line, activation of adenosine receptors coupled to the phosphoinositide hydrolysis signaling pathway also should activate protein kinase C and, in contrast, increase SERT activity, suggesting differential regulation by protein kinase C and/or additional pathways engaged in these cells (47). After cholera toxin and forskolin treatment, human placental choriocarcinoma (JAR) cells display enhanced SERT activity and increased cell-surface transporter density; however, the delayed nature of this effect compared to the rapid rise in intracellular cAMP levels suggests an effect on mRNA stability or gene transcription (14). Indeed, the levels of SERT mRNA are markedly elevated by cholera toxin treatment (53). The effects of cAMP on SERT expression may reflect a cell or species-specific sensitivity of the human SERT gene to second messengers in that increases in cAMP levels have been reported to induce down-regulation of SERT activity in rat PC12 cells and C33-14-B1 mouse fibroblasts (34). Regardless, SERTs and NETs, like other molecules involved in neurotransmitter signaling, appear to be sensitive to chronic changes in intracellular regulatory cascades and could be inappropriately regulated in mental illness.

Blockade of 5HT uptake is an immediate effect of the antidepressant 5HT uptake inhibitors, whereas the therapeutic effects of these drugs are observed only after 2–3 weeks of treatment. Compensatory responses at pre- and/or postsynaptic receptors resulting from prolonged transporter blockade are generally thought to be involved in the therapeutic actions of the antidepressants; thus, potential adaptive responses in SERT expression associated with chronic antidepressant treatment also are of much interest. For example, chronic treatment with the tricyclic antidepressant clomipramine decreases [3H]imipramine binding in platelets from healthy human volunteers. Conversely, amitriptyline treatment reportedly increases SERT sites, whereas imipramine has no effect on platelet [3H]imipramine binding (37). At the molecular level, long-term treatment with selective SERT antagonists, but not monoamine-oxidase inhibitors, decreases the steady-state levels of SERT mRNA in rat brains (41), consistent with differential effects of these agents on rat brain SERT sites measured with radioligand binding (26). Because these studies were performed in animals and healthy subjects, a more pertinent clinical question becomes, How are SERTs regulated in depressed patients? Several postmortem brain studies as well as the previously mentioned platelet SERT data report that [3H]imipramine binding is reduced in patients with affective disorders (37, 49) (Alterations in platelet and brain SERTs observed in depression also are reviewed in Nemeroff et al.). In some (but not all) studies, a delayed increase in platelet [3H]imipramine binding has been observed in depressed patients following clinical improvement resulting from antidepressant treatments, lending some credence to the use of platelet SERTs as a diagnostic marker for depression (37). However, an absolute correlation between platelet SERT levels and severity of clinical depression remains to be established (see Selective Serotonin Reuptake Inhibitors in the Acute Treatment of Depression and Standard Antidepressant Pharmacotherapy for the Acute Treatment of Mood Disorders).

In addition to depression, SERT inhibitors have found clinical usefulness in the treatment of obsessive–compulsive disorder, panic disorder, eating disorders, alcoholism, and premenstrual syndrome (22), although none of these disorders has as yet been directly linked to altered SERT gene expression. The identification of the human genetic locus for SERT and the isolation of human SERT genomic clones will assist in evaluating possible hereditary SERT variations that might underlie a predisposition to psychiatric disease. In addition, greater inspection of SERT expression in humans in vivo should be feasible due to the combination of advanced brain imaging techniques with potent and selective SERT ligands (38). Modeling of SERT-related disorders in rodents may be facilitated by the cloning of the murine SERT, now known to be localized to mouse chromosome 11 (13, 27), and its use in the generation of targeting vectors for transgenic ablation of the SERT locus.

Although uptake of the monoamine neurotransmitters has been studied for almost three decades, our basic knowledge of NET and SERT structure and regulation is rudimentary. The cloning of NET and SERT cDNAs has provided (a) information regarding primary transporter sequence and (b) the necessary tools to more directly examine the many biophysical and pharmacological properties associated with their activities. Although significant progress has been achieved in defining ionic and substrate specificity of native and cloned transporters, we know little about how substrates bind to the transporters and are thereby shuttled across the plasma membrane, or how antagonists bind. The role of protein phosphorylation or other post-translational modifications in regulating transporter function is only beginning to be evaluated, but it promises to be a fertile area for future studies now that NET and SERT proteins can be visualized. Hints of altered NET and SERT gene regulation after hormonal stimulation suggest significant gains to be acquired from systematic analysis of genomic regulatory elements that control transporter expression. For many, the focus has shifted from establishment of the primary structures of NETs and SERTs to exploiting new DNA and antibody tools for an understanding of how these molecules bind and transport substrates and antagonists, how they become localized to synaptic sites, the degree to which they respond to regulatory cues, and whether hereditary genetic variations contribute to neuropsychiatric disorders.

published 2000