Amphetamines, new psychoactive drugs and the monoamine transporter cycle.

In monoaminergic neurons, the vesicular transporters and the plasma membrane transporters operate in a relay. Amphetamine and its congeners target this relay to elicit their actions: most amphetamines are substrates, which pervert the relay to elicit efflux of monoamines into the synaptic cleft. However, some amphetamines act as transporter inhibitors. Both compound classes elicit profound psychostimulant effects, which render them liable to recreational abuse. Currently, a surge of new psychoactive substances occurs on a global scale. Chemists bypass drug bans by ingenuous structural variations, resulting in a rich pharmacology. A credible transport model must account for their distinct mode of action and link this to subtle differences in activity and undesired, potentially deleterious effects.

Amphetamines – a diverse class of compounds

The amphetamines are a diverse class of chemical compounds. They comprise synthetic compounds and naturally occurring alkaloids such as ephedrine and cathinone (Box 1), which are synthesized in the plant species Ephedra and Catha, respectively. Extracts from these plants have been used for their remedial and psychostimulant effects for millenia (for an excellent historical overview see [1]). Lazăr Edeleanu synthesized the eponymous compound amphetamine in Berlin in 1887 (Box 1); many derivatives were produced within the next two decades [1]. The availability of pure compounds allowed their pharmacological characterization, which was initiated by Sir Henry Dale [2]. Biel and Bopp [3] chemically defined an amphetamine as containing: (i) an unsubstituted phenyl ring, (ii) a two-carbon side chain between the phenyl ring and nitrogen, (iii) an α-methyl group, and (iv) a primary amino group (Box 1). Many compounds meet these criteria. In addition, the targets (i.e., the monoamine transporters) apparently interpret the rules in a flexible manner: for instance, methamphetamine, ephedrine, and methylphenidate violate the primary-amine rule (Box 1). Hence, this large group of compounds is difficult to define on chemical grounds. An operational definition based on the pharmacology (‘amphetamine-like action’) falls short of accounting for the complex mode of action: in fact, there is a continuum ranging from amphetamine-triggered release to methylphenidate-induced blockage of uptake. A clear-cut definition would facilitate imposing legal restrictions on marketing activities. The difficulties arising from a chemistry-based definition are also exemplified by cathinones: the naturally occurring cathinone is a ketone. It serves as a scaffold for numerous substitutions: methylation produces meth(yl)cathinone, which allows regulations restricting the sales of cathinone to be bypassed. However, cathinone derivatives are not only used illicitly: they include the antidepressant drug bupropion and the anorectic agent diethylpropion. Synthetic analogs of methcathinone have recently become popular as ‘designer drugs’ or ‘legal highs’ on the illicit drug market [4]. Often, they are referred to as ‘bath salts’, ‘plant food’, or ‘research chemicals’ to facilitate their distribution and to obviate prosecution. Popular methcathinone derivatives are 4-methylmethcathinone (mephedrone, Box 1) and 3,4-methylenedioxymethcathi-none (methylone). More recently, 3,4-methylenedioxypyrrovalerone (MDPV, Box 1) has reached the market; it is an example of a very potent, psychoactive synthetic cathinone and illustrates the problem of a laissez-faire approach to recreational drug use: MDPV exerts stimulant-like effects at low doses, but life-threatening side effects are seen at high doses or upon chronic use [5]. Accordingly, the authorities have banned amphetamines including the first-generation synthetic cathinones such as mephedrone, methylone, and MDPV in the USA and the EU.

Ignorance is blatant with respect to new psychoactive substances (NPS) which flood the drug market at present (Box 2). Chemical substitutions may render a given drug compatible with current legislation, but they may also introduce dramatic changes in the activity of the compound and in its mode of action. Widespread consumption of NPS may be associated with health risks such as neurodegeneration. These are linked – at least in part – to the mode of action. Accordingly, it is of relevance to elucidate the different modes of action of amphetamine, its congeners, and in particular of stimulant NPS. The main targets of amphetamines (and presumably of many stimulant NPS) are the neurotransmitter:sodium symporters (NSS) for the monoamines dopamine (DAT/solute carrier protein SLC6A3), serotonin (SERT/SLC6A4), and noradrenaline/norepinephrine (NET/SLC6A2). Amphetamines are exogenous substrates of and generate a current through these transporters [6]. The action of amphetamine has been studied for more than 100 years [2]. Nevertheless, there are still many open questions: (i) how do these compounds differ from physiological substrates to act as efficient releasers? (ii) What is the role of protein kinases in triggering efflux? (iii) Does the oligomeric arrangement of monoamine transporters affect the action of amphetamine? (iv) Why does amphetamine elicit a blocking (paradoxical) action on disease-associated mutants of, for example, the dopamine transporter/SLC6A3? (v) Does the lipid composition of the plasma membrane affect the action of amphetamine? In this review we address these issues and argue that their resolution is important to understanding the differences in the actions of individual compounds.

The principle targets of amphetamines: monoamine transporters

In monoaminergic neurons, retrieval of neurotransmitters into the synaptic vesicles is accomplished by a relay that is made up of several components (Figure 1): (i) vesicles are tethered to the plasmalemmal monoamine transporter (DAT, SERT, or NET). The resulting spatial proximity promotes efficient (re)filling of the docked vesicle at the reuptake site [7]. (ii) The turnover number of the reserpine-sensitive, proton-driven vesicular monoamine transporter VMAT2 (400 min−1) [8] is higher than that of the plasmalemmal transporters (120–180 min−1) [9,10]. This creates a sink and limits diffusion of monoamines within the cytosol of the presynaptic bouton. (iii) This is further enhanced by the action of mitochondrial monoamine oxidases (MAO-A and B) and catechol O-methyltransferase (COMT) [11] which degrade cytosolic monoamines. Amphetamines interact with several targets within this relay; in other words, they are exogenous substrates of the plasmalemmal transporters (DAT, SERT, NET), and of the vesicular transporters VMAT1 and VMAT2, and inhibit MAO. The concerted action of amphetamines on these three targets is the core tenet of the ‘weak base hypothesis’ [12,13]: amphetamines enter monoaminergic terminals via DAT, SERT, or NET and subsequently accumulate in the synaptic vesicles by the action of VMATs. Therein, by their physicochemical nature, they dissipate the proton gradient and preclude inward transport of monoamines. The resulting elevation in cytosolic monoamine neurotransmitters is further aggravated by MAO inhibition. Thus, they rise to concentrations that allow occupation of the internal substrate binding site [14] and their subsequent outward transport by plasmalemmal transporters. Amphetamines also bind to the trace amine receptors TAR1, a Gs/Gq-coupled receptor, and to LGC (ligand-gated channel)-55, an amine-gated chloride channel related to the pentameric ligand-gated channels (e.g., GABAA or glycine receptors) and which was originally identified in C. elegans [15]. Finally, it is also likely there are also additional unidentified targets, for example a cytosolic target that accounts for the ability of amphetamine to downregulate surface levels of the glutamate transporter EAAT3 [16]. Several arguments suggest that the monoamine transporters are the principal site of action, which must be addressed by amphetamines to elicit their biological responses, in particular their psychostimulant effects. While blockage of VMAT allows the accumulation of cytosolic dopamine, it fails to induce dopamine release in DAT-deficient neurons [17]. Furthermore, TAR1 is not activated by all psychoactive amphetamines (p-chloroamphetamine, for instance, is inactive); stimulation of TAR1 actually reduces dopamine release and thus decreases sensitivity to amphetamine [18,19]. The preferred ligands for the amine-gated chloride channel LGC-55 are tyramine and phenylethylamine; amphetamine engages the channel only at high concentrations (EC50 around 150 μM) [20]. In addition, the internalizing action on EAAT3 explains long-term changes in neuronal excitability but cannot account for the immediate action of amphetamines [16].

Figure 1

Schematic of the transport cycle. A kinetic scheme of substrate (S) interaction with either the outward- (o) or inward- (i) facing transporter (T) and with cotransported substrates. Substrates can be either physiological substrates such as monoamines or amphetamines; co-substrates can be sodium (Na+) or chloride (Cl−). There are several additional reactions: in other words, sequential binding of Na+ ions, chloride, and substrate to the outward-facing conformation, and the corresponding release steps from the inward-facing conformation ([27] for a more detailed model). These have been omitted for the sake of clarity. The transporter switches from the outward- to the inward-facing conformation via an occluded state (occ). In the serotonin transporter (SERT) the return through the occluded empty state is contingent on binding of K+ (symbolized by K+ in brackets). The conducting state (Cond) is achieved via an inward-facing conformation [27]. In the presence of amphetamine, amply supplied internal substrate and elevated internal sodium, the transporter releases substrate by running backwards through the cycle (reaction pathway indicated by red arrows). Burst-like dopamine (DA) effluxing events have been recorded by Kahlig and colleagues [47], but their relation to the transport cycle is unclear (indicated by the dashed arrow on the left-hand side).

The psychostimulant action of amphetamines relies on reverse transport

Transport by NSS is by definition coupled to the downhill movement of ions. Accordingly, NSS rely on Na+/K+-ATPase which generates gradients for sodium (outside>>inside) and potassium (inside>>outside). In most cells, chloride moves passively along its electrochemical gradient driven by the membrane potential (outside>>in->inside). It is generally accepted that sodium and chloride serve as transported co-substrates for NSS [21]. In the transport cycle, two sodium ions and one chloride ion and the substrate are sequentially loaded into the outward-facing conformation of the transporter, are transiently locked into an occluded state transport and, upon opening of an inner gate, are finally released into the cytosol. The ability of the monoamine transporters to concentrate substrate substantially more than 10-fold by inward transport suggests that two Na+ ions are cotransported. It is worth pointing out that there is no formal proof for cotransport of Cl− and of the second Na+ ion. Thus, it is not known in which state the transporter completes the cycle (returning in the empty apo-state or in a Na+- and/or Cl−-bound state). It is clear though that SERT differs from its close relatives by relying on the potassium gradient for the return step [22]. Given that monoamines are thought to be transported as charged species (i.e., with their amine nitrogen protonated) [23], it is possible to deduce the tentative net charge movement across the membrane during the transport cycle based on the stoichiometries: for SERT, where flux measurements suggest only a single Na+ ion moves with the substrate, the balance predicts no charge movement, while for DAT and NET positive charge moves into the cell [22]. Surprisingly, all three monoamine transporters generate transport-associated currents. The transport cycle is based on the alternating access model, which was originally conceived as a Gedanken experiment [24]. It is the most plausible description of substrate translocation; in fact, the predicted intermediates (e.g., outward-facing, occluded substrate-bound; inward-facing, occluded empty stage) have been visualized in the crystal structures of several bacterial transporters, including the homolog for mammalian neurotransmitter transporters, the leucine transporter LeuTAa [25]. A detailed kinetic analysis indicates several additional intermediate states in SERT [26,27]. It is gratifying to note that nine different conformations have already been identified in crystals of the bacterial betaine transporter BetP [28,29]. Although BetP is unrelated to SLC6 transporters, it uses the same fold of pseudosymmetric inverted repeats for the translocation process [30]. There is one major set of observations, however, that have been difficult to reconcile with the alternative access model. These are the transport-associated currents seen in all monoamine transporters [31,32]. Their magnitude suggests that charge is moved in considerable excess to the translocated substrate, indicating that the stoichiometries outlined above cannot account for the current. Accordingly, it was proposed that transporters function in a channel mode, in which the outer and inner gates open simultaneously, and substrate and ions permeate in single-file mode [33,34]; this may be true for SERT of Drosophila melanogaster [35]. However, detailed kinetic analysis shows that the conducting state of human SERT is reached via an inward-facing conformation, which allows influx of sodium ions; the magnitude of the uncoupled conductance is limited because the probability that the transporter assumes this particular conformation is low [27]. It therefore appears unlikely that the channel mode is the preferred transport mode of mammalian transporters.

The principal mechanism by which amphetamine elicits its biological responses, in particular the psychostimulant effects, is accounted for by its ability to induce efflux of monoamines by reverse transport. This effect is large in magnitude, it does not require any neuronal activity, and has hence been termed non-exocytotic release. Before dwelling on the mechanistic details, it is worth pointing out that amphetamines are taken up by plasma-membrane monoamine transporters as exogenous substrates [31]. Accordingly, they inhibit the physiological monoamine reuptake in a competitive manner [36]. As a consequence of both amphetamine-induced reverse transport and inhibition of reuptake, the synaptic monoamine concentration increases, which in turn activates post- and presynaptic receptors [37]. The activation of postsynaptic receptors propagates the signal and contributes to the biological response. Stimulation of presynaptic autoreceptors decreases the quantal release of monoamines upon excitatory inputs; this is further supported by the observed depletion of vesicular monoamine storage by amphetamines acting at the vesicular monoamine transporters (see above).

How to account for the reverse transport mode

Amphetamines generate a current which exceeds the stoichiometrically coupled ion movement that accompanies the transport process: an uncoupled conductance [31,32]. Hence, the exogenous substrate amphetamine will compete with the endogenous substrates on the intracellular side for binding to the transporter primary binding site. In addition to the coupled conductance accompanying the electrogenic NSS such as DAT and NET, the uncoupled conductance, in particular in SERT, raises the intracellular sodium levels and increases the affinity of the transporter for all substrates available for outward transport. Therefore, any rise in intracellular sodium (e.g., by blockade of the Na+/K+-ATPase, or dissipation of the sodium gradient by the sodium–proton ionophore monensin), triggers efflux of endogenous substrates – a conjecture that has been experimentally verified [14,38]. In addition, if transmembrane ion gradients are changed, reversal of transport is initiated either upon lowering of extracellular sodium [39] or raising extracellular potassium concentrations [40]. Thus, there are two conceptually important types of reverse transport: (i) the trigger is the binding to and transport of releasers, and (ii) changes in the ion composition of the extra- or intracellular fluid. In either case, there must be a releasable pool of substrate accessible on the cytoplasmic side, which is kept low under physiological conditions by the relay outlined above.

Several models have been proposed to account for the reverse transport. Originally, the explanation focused on the ‘alternating access hypothesis’ as a starting point. Accordingly, the transporter was metaphorically proposed to operate as a ‘revolving door’: the outward transport of the physiological monoamine was initiated by the inward transport of amphetamine. This model assumed that (i) the trigger (and thus the driving force) was provided by the binding and inward movement of the amphetamine molecule, and (ii) forward and reverse transport were accomplished by the same transporter molecule. This was inferred from the observed substrate and inhibitor selectivity; for example, cocaine blocks substrate uptake and amphetamine-induced substrate efflux [41]. This ‘revolving door’ model is also referred to as ‘facilitated exchange diffusion model’ [42].

The conceptual problem with the revolving door metaphor is evident from an inspection of the transport cycle (Figure 1): the physiological return step is the apo-version of the transporter, in other words the empty transporter (in the case of SERT, this state is thought to be the K+-loaded version of the transporter, see above). Amphetamines can per se drive the full transport cycle because they elicit transport-associated currents, which are carried by the empty inward facing conformation [27]. At present, it is unclear if the transporter can accommodate K+ and substrate at the same time during its return step. Nevertheless, it is unclear if the K+ hypothesis holds true for SERT alone or if it could even be extended to the other members of the SLC6 family. Thus, simple ‘exchange diffusion’ requires the transporter to run in reverse, similarly to an enzymatic back-reaction; in other words, it would mean that the transporters move through the steps from the outside to the inside and back (Figure 1). The revolving door metaphor for the explanation of reverse transport is at the very least misleading, because the transporter only moves through a half-cycle: the revolving door is in fact jammed. This interpretation is supported by the observation that high internal, physiological substrate levels preclude the appearance of the channel mode [26]. In other words, in the presence of high internal substrates, the transporter is less likely to visit conformations in the left-hand part of the scheme in Figure 1: it seesaws through the right-hand part of the cycle. The semantics may be ignored, but there are additional issues that must be addressed.

First, if exchange diffusion was so readily achievable, the monoamine transporters would be prone to futile cycling when transporting their cognate substrates. It may be argued that rapid removal by the vesicular monoamine transporters may prevent futile cycling, but there is little evidence of futile cycling of cognate substrates in the absence of vesicular transporters: it is not seen in transfected cells which express monoamine transporters. Amphetamines are prone to futile cycling because they diffuse back through the membrane [26].

Second, it is questionable that the very same transporter moiety, which supports amphetamine influx, is the moiety that acts as the outward carrier for cognate substrate. SLC6 transporters form oligomers [43,44]. These are formed in the endoplasmic reticulum (ER) but, at the cell surface [45], the oligomers are kinetically trapped (i.e., they do not readily exchange [46]) (Figure 2). These oligomers may account for amphetamine-induced release spikes or bursts, which are detected by amperometry when DAT-expressing cells are challenged with amphetamine [47]. The burst-like release was originally attributed to a channel mode which supported spikes of dopamine efflux [47]. The assumption underlying this interpretation is to posit an internal binding site for amphetamine which is not recognized by the cognate substrate. Occupation of this site by amphetamine is thought to promote long-lasting open states, and this action has been metaphorically referred to as the ‘molecular stent hypothesis’ [48]. There are many arguments that question this model [26,49]. Suffice it to say that it is also doubtful on chemical grounds that DAT can form a continuous aqueous pore which is selective for dopamine but does not allow the permeation of amphetamine. Thus, we propose that the burst-like release of dopamine reflects (more or less) synchronous outward transport by higher-order oligomers. Because these are less frequent than monomers and dimers (e.g., hexamers are <10% [46]), spikes of release are detected at low frequency over amperometric noise. It is also worth noting that amphetamines can trigger release of GABA through a concatemer of SERT and GAT1 [50]. There are additional findings which support the conjecture that the oligomeric nature of monoamine transporters is important for the action of amphetamines ([51] for review). This may be difficult to reconcile with the observation that amphetamine dissociates oligomers [52,53]. However, oligomer formation was assessed by co-immunoprecipitation from detergent extracts and by crosslinking. The amphetamine-induced changes may reflect a conformational change and a difference in lipid environment (see below).

Figure 2

The molecular effects of amphetamines. Schematic illustration of the effects of amphetamines (AMPH, light-green circles) on the reverse operation of neurotransmitter:sodium symporters (NSS). NSS are present in the plasma membrane either as monomers or oligomers [46]. They are physically linked to the vesicles, and this allows their efficient refilling with monoamines (MA, yellow circles) [7]. The oligomer-based counter-transport model [50] is shown on the left side of the figure and illustrates that the effect of amphetamine relies, at least in part, on an intact oligomer. Amphetamines target the vesicular monoamine transporter (VMAT) and lead to either inhibition and/or reversal of the transport direction to increase the cytosolic concentration of MA, thereby enabling reverse transport. Furthermore, amphetamines inhibit enzymes such as monoamine oxidases A and B (MAO) and thereby prevent the degradation of MA. Abbreviation: TAR1, trace amine receptor 1.

Third, membrane lipids modify the transport activity of SERT and DAT. This includes cholesterol, which promotes inward transport [54,55]. In fact, cholesterol has been visualized in the crystal structure of Drosophila DAT, wedged in a groove formed by transmembrane helices 1a, 5, and 7 [56]. This helps to explain why cholesterol promotes the outward-facing conformation of DAT [54]. The second lipid known to affect monoamine transporter is the phosphoinositide phosphatidylinositol-4,5-bisphosphate (PIP2). PIP2 is a low-abundance constituent of the inner leaflet, serves as a precursor of signaling molecules (i.e., inositoltriphosphate, IP3; and diacylglycerol, DAG), and is a signaling molecule in its own right: it regulates the functions of several transmembrane proteins, in particular of transporters (e.g., Na+/Ca+-exchanger [57]) and ion channels (e.g., voltage-gated K+ and Ca2+ channels [34]). SERT and DAT are also regulated by PIP2 [58,59]. Manipulations that deplete PIP2 reduced amphetamine-induced currents through, and reverse transport by, SERT without affecting inward transport or surface levels [58]. These observations show that PIP2 can specifically affect the conformational equilibrium in the transport cycle. The very binding site of PIP2 in DAT appears to affect the membrane trafficking of DAT [59]. Importantly, the functional impact of the PIP2–DAT interaction is mediated by the N-terminus: abolishing binding of PIP2 to DAT by introducing the appropriate mutations also specifically eliminated amphetamine-induced efflux [59]. Most importantly, Drosophila melanogaster knock-in flies which express this mutant (human) DAT rather than their endogenous version are phenotypically normal unless challenged with amphetamine: basal locomotion (and circadian rhythm) is identical to that of wild type flies but the amphetamine-induced hyperactivity is greatly reduced [59]. This experiment formally proves that the interaction of PIP2 with DAT is relevant to the action of amphetamine in vivo. A further observation suggests that the membrane lipid environment is important: DAT is targeted to lipid microdomains (detergent-resistant membranes or ‘lipid rafts’) by flotillin-1: depletion of flotillin-1 blunts amphetamine-induced substrate efflux in neurons [60] and amphetamine-induced hyperlocomotion in Drosophila melanogaster [61]. It is also worth noting that there is an obvious link to protein kinase-dependent regulation of monoamine transporters (see also below): disruption of the interaction with flotillin-1 also impairs the regulation of DAT by protein kinase C (PKC) [60] and Ca2+/calmodulin-dependent protein kinase II (CaMKII) [62]. Finally, the interaction between DAT and flotillin is relevant to understanding attention deficit/hyperactivity disorder (ADHD): DAT-R615C, a heterozygous mutation associated with ADHD, fails to associate with flotillin but is constitutively associated in a complex with CaMKII [63].

Fourth, there is ample evidence that the phosphorylation in DAT and also SERT contribute to the action of amphetamines (reviewed in [64]): amphetamine-induced efflux is blunted by elimination of the phosphorylation sites in the N-terminus of DAT [41] or by inhibition of either PKC or αCaMKII [50,65-69]. Most importantly, there is formal proof that CaMKII interacts with DAT [66-68] and modifies the amphetamine-induced dopamine efflux in vivo in mice [69]. Furthermore, mice deficient in αCaMKII have a reduced locomotor response to amphetamine; in addition, behavioral sensitization is also impaired [69]. Thus, αCaMKII not only supports acute amphetamine-induced dopamine efflux but is also important in shaping the chronic response to amphetamine. A mechanistic explanation must take into account the interplay of candidate phosphorylation sites, the lipid environment, and the oligomeric arrangement (see above). The most plausible explanation is to posit that (i) C- and N-termini are in close vicinity (for which there is direct evidence [70,71]) and (ii) are engaged and modified by kinases at the cognate sites, which (iii) introduces negative charges, and thus may alter the interaction of the transporter with phospholipids, and (iv) the resulting conformational change is transmitted by the N-terminus, which acts as a lever: restricting the mobility or truncation of the N-terminus eliminates amphetamine-triggered efflux [10]. DAT is palmitoylated at C580 in the proximal segment of its C-terminus [72]. Palmitoylation counteracts the action of PKC (i.e., to promote downregulation; Box 3). At the very least, these observations provide circumstantial evidence for a relation between phosphorylation and the lipid environment of the membrane (palmitoylation affects the association of protein with lipid rafts).

Concluding remarks

The transport cycle is governed by a series of conformational equilibria. Rate constants have been inferred from electrophysiological measurements [27], but the thermodynamics underlying the transition are not understood. It is conceivable that the conformational equilibria in the cycle may depend on the oligomeric state of the transporter and the lipid environment and that it can be shifted by regulatory modifications (phosphorylation, palmitoylation) state and/or by mutations. A mechanistic interpretation of the action of amphetamines must take the transport cycle into account. We have argued that a seesaw mode is the most plausible way to account for the kinetics of reverse transport. This exchange diffusion mode is facilitated by conformational switches, which are brought about by the concerted action of the N-terminus of the transporter and the lipid environment in which the transporter is embedded. Thus, transporter may function as a signal integrator and coincidence detector. In the presence of elevated intracellular Na+, the appropriate lipids, and an activated kinase, monoamine transporters may also function as a device to allow physiological monoamine release [73]. Similarly, mutations may shift monoamine transporters into this release mode: a prominent example is DAT-A559V, which was identified in individuals affected by ADHD [74]. Under these conditions, amphetamines may predominantly act as inhibitors, a conjecture which was verified for DAT-A559V. At the very least, this hypothesis is worth exploring because it may explain the paradoxical beneficial actions of amphetamines in ADHD (Box 4).