Trafficking of Vesicular Neurotransmitter Transporters

Abstract

Vesicular neurotransmitter transporters are required for the storage of all classical and amino acid neurotransmitters in secretory vesicles. Transporter expression can influence neurotransmitter storage and release, and trafficking targets the transporters to different types of secretory vesicles. Vesicular transporters traffic to synaptic vesicles as well as large dense core vesicles, and are recycled to synaptic vesicles at the nerve terminal. Some of the intrinsic signals for these trafficking events have been defined and include a dileucine motif present in multiple transporter subtypes, an acidic cluster in the neural isoform of the vesicular monoamine transporter (VMAT2) and a polyproline motif in the vesicular glutamate transporter VGLUT1. The sorting of VMAT2 and the vesicular acetylcholine transporter (VAChT) to secretory vesicles is regulated by phosphorylation. In addition, VGLUT1 uses alternative endocytic pathways for recycling back to synaptic vesicles following exocytosis. Regulation of these sorting events has the potential to influence synaptic transmission and behavior.

Introduction: Neurotransmitter Transporters Subtypes

Synaptic transmission requires two types of neurotransmitter transporters. Following exocytosis, plasma membrane transporters remove neurotransmitters from the synaptic cleft, to terminate signaling and recycle the neurotransmitters for another round of exocytosis (1, 2). Vesicular neurotransmitter transporters package neurotransmitters into the lumen of secretory vesicles to allow exocytotic release (3, 4). Four types of vesicular transporters have thus far been identified^* and include: 1) the vesicular acetylcholine transporter or VAChT (5); 2) vesicular monoamine transporters or VMATs (5); 3) the vesicular GABA and glycine transporter, VGAT, also known as the vesicular inhibitory amino acid transporter or VIAAT (6); and 4) the vesicular glutamate transporters or VGLUTs (7). Mammals express two VMAT genes, VMAT1 and 2 (5), and three VGLUTs, VGLUT1, 2 and 3 (7, 8). VMAT2 is expressed in all aminergic neurons in the brain, mast cells and some neurons in the gut (9) whereas VMAT1 is expressed in peripheral neuroendocrine cells including adrenal chromaffin cells (9). VGLUT1 and 2 are expressed in complementary subsets of glutamatergic neurons in the CNS (10). In contrast, the most recently identified isoform, VGLUT3, is co-expressed with VAChT and VMAT2 in a number of cholinergic and aminergic cell types (11–13).

1) The Functional Impact of Vesicular Transporter Trafficking

Vesicular transporter expression can influence neurotransmitter release

The response of a post-synaptic cell to a single exocytotic event (one secretory vesicle) is defined as quantal size. Over the past ten years, it has become increasing clear that vesicular transporter expression can regulate the amount of neurotransmitter contained in secretory vesicles, and thus may influence quantal size (4, 14). In vitro overexpression of mammalian VAChT (15) VGLUT1 (16, 17) and VMAT2 (18), as well as Drosophila VGLUT increases quantal size (19). Conversely, a decrease in quantal size has been observed in VAChT knockdown mice (20), VGAT knockout heterozygotes (21) and VGLUT heterozygotes (22). These data suggest that altered transporter expression may increase or decrease the number of vesicular transporters that localize to each synaptic vesicle; it should be noted, however, this has not yet been demonstrated directly. Indeed, the number of transporters that actually reside on a synaptic vesicle in any system remains unclear, and estimates range from 1 to 14 (23, 24).

Overexpression of Drosophila VMAT increases amine dependent behavior (25) and a number of mouse knockout models indicate that decreasing transporter expression can affect a number of complex behavior in vivo (20, 22, 26–31). Similarly, it is possible that regulated changes in transporter localization have synaptic and behavioral sequelae, underscoring the potential importance of vesicular transporter trafficking in vivo (32–36).

Vesicular transporters traffic to least two distinct types of secretory vesicles

Transporter trafficking may influence synaptic transmission by directing the transporters to different types of secretory vesicles. In addition to their storage in synaptic vesicles, monoamines, and perhaps other neurotransmitters, are stored in large dense core vesicles (LDCVs) in neurons (37, 38). (LDCVs are analogous to the large dense core granules (LDCGs) found in neuroendocrine cells (37, 38)). Synaptic vesicles and LDCVs differ in several important ways, including their subcellular sites of release (see Fig. 1A for additional details). Importantly, VMAT2 localizes to both synaptic vesicles and LDCVs in neurons (39–41). Therefore, regulated changes in VMAT2 trafficking have the potential to alter the relative distribution of the transporter to synaptic vesicles versus LDCVs, and the subcellular site of monoamine release.

Fig. 1.

Trafficking Pathways Proposed for Vesicular Neurotransmitter Transporters. A) Secretory vesicle subtypes. Vesicular transporters localize to different types of secretory vesicles in neurons. These include synaptic vesicles (labeled SV in the figure), and large dense core vesicles (LDCVs, labeled LV in the figure). LVs, but not synaptic vesicles, co-release neuromodulatory peptides, Both SVs and LVs undergo calcium-mediated exocytosis (red arrows) but differ in the site and mode of release. SVs undergo regulated release in nerve terminals and more specifically, at the active zone, a discrete presynaptic region located opposite to post-synaptic receptors. Unlike SVs, LVs are found in both the nerve terminal as well as the cell body and dendrites, and do not closely appose the plasma membrane. The release of neurotransmitter from LVs activates receptors outside of the synapse (labeled “Extra synaptic receptors” in panel A) rather than postsynaptic receptors. The release of neurotransmitter from LVs can also activate somatodendritic autoreceptors, which are found on both the nerve terminal and somatodendritic membranes of most if not all neurons, and provide negative feedback to limit excess neurotransmitter release.

B) Trafficking to secretory vesicle subtypes. Vesicular transporters traffic to SVs and LVs via different routes. For sorting to SVs (purple arrows), proteins are thought the exit the trans Golgi network (TGN) in constitutive secretory (CS) vesicles, and are likely to undergo an endocytic step to generate a mature SV. In contrast, proteins destined for LVs sort into the regulated secretory pathway at the TGN (orange arrows). Immature LVs (im. LVs) are thought to mature via AP1- and PACS-1 dependent removal of some proteins, shown here as a budding event.

C) Synaptic vesicle trafficking at the synapse. The generation of mature SVs as well as SV recycling at the nerve terminal may occur via several mechanisms. Following fusion with the plasma membrane, transporters on constitutive secretory (CS) vesicles may be sorted to a mature SV directly from the plasma membrane using clathrin and the adaptor complex AP2. Alternatively, they may sort to SVs via a two-step pathway involving AP3 and budding from an early endosomal intermediate (EE). Following exocytosis, SVs and vesicular transporters may recycle via either of these pathways. It is also possible that SVs recycle using a kiss-and-run mechanism (not shown).

D) Alternative models of transporter trafficking. Alternative models have been proposed for the sorting of some vesicular transporters to SLMVs in neuroendocrine cells (64, 65) (pathway labeled “PC12?”), and possibly to SVs in neurons (pathway labeled “Neurons?”). In one alternative model, proteins such as VAChT that are removed from immature LVs are sorted to an early endosome where they are sorted into SLMVs. Vesicles that leave immature LVs may represent a “constitutive-like” secretory pathway as originally proposed for secretory granules in pancreatic islet cells (63, 150). Trafficking of constitutive-like vesicles to the nerve terminal might allow sorting to mature SVs via endocytosis.

It addition to monoamines, it is possible that other types of neurotransmitters are released from vesicle types that are similar to LDCVs (42–47) albeit without a dense proteinaceaous core. In glutamatergic, hippocampal neurons, heterologously expressed VMAT2 is sorted to clear vesicles that can be released at somatodendritic sites (44). Similarly, somatodendritic release in dopaminergic neurons is mediated by clear, pleiomorphic vesicles known as tubulovesicular bodies (39, 48, 49). At least one VGLUT isoform, VGLUT3, localizes to dendrites (8, 11) and glutamate release from dendrites regulates signaling in cortical pyramidal cells (45, 46). In addition, GABA release from dendrites has been reported to mediate retrograde signaling in the cortex (46). Since synaptic vesicles are thought to be restricted to the nerve terminal, these data suggest that both VGLUT and VGAT localize to other types of secretory vesicles, and the possibility that taransporter trafficking regulates the subcellular site of GABA and glutamate release.

2) Sorting Vesicular Transporters to Synaptic Vesicles and LDCVs

Synaptic vesicle biogenesis and the possible use of alternative endocytic pathways

During synaptic vesicle biogenesis, proteins bound for synaptic vesicles at the nerve terminal are thought to first traffic to the plasma membrane (50, 51) and vesicular transporters are likely to follow a similar pathway (52). In Fig. 1B, we show exit from the TGN via constitutive secretory vesicles (50), although it remains possible that some synaptic vesicle proteins exit the TGN on more specialized organelles (53). It is generally accepted that vesicles exiting the TGN do not contain the complete complement of proteins found on mature synaptic vesicles, and require an additional trafficking step(s) to fully mature (51), most likely at the nerve terminal (Figs. 1B and C) (51, 54) (but possibly at the soma, see (55)).

To complete the maturation of synaptic vesicles via endocytosis, they may either bud directly from invaginations at the plasma membrane (56) or pass through an endosomal intermediate (57) (Fig. 1C). Both processes may use the clathrin adaptor complex AP2, but the second model requires an additional AP3 dependent step (57). The AP3 dependent pathway is clearly active in the biogenesis of SLMVs in neuroendocrine cells (57–59). AP3 is unlikely to be required for the biogenesis of most synaptic vesicles, or the sorting of vesicular transporters onto mature synaptic vesicles in most neurons (60, 61). However, a mouse knockout of the mu3B subunit of the neuron-specific AP3B complex shows lowers number of synaptic vesicles in inhibitory terminals, and a modest change in the amount of VGAT imunoreactivity in synaptosomes (62); thus, it remains possible that AP3 plays a major role in de novo synaptic vesicle biogenesis in some neurons.

It is also possible that sorting into the regulated secretory pathway is the first step for targeting some cargo to synaptic vesicles. As previously shown in pancreatic islet cells, proteins may be removed from immature LVs and sort into a “constitutive-like” secretory pathway (63). A similar pathway has been suggested to sort VAChT to SLMVs in PC12 cells (see Fig. 1D) (64, 65).

SVs recycle at the nerve terminal via alternative endocytic pathways and localize to at least two functional pools

Endocytosis is required for recycling synaptic vesicles at the nerve terminal as well as their biogenesis (66). Importantly, both fast and slow endocytic pathways have been identified (67, 68).^* The slow endocytic pathway, sometimes referred to as bulk, or compensatory endocytosis is particularly important for vesicle recycling under conditions of sustained release. Similar to the AP3 dependent mode of synaptic vesicle biogenesis (57, 71), the slower endocytic path is likely to require a cisternal, or endosomal intermediate and possibly AP3 (67, 68, 72, 73).

Interestingly, trafficking through fast versus slow endocytic pathways may determine whether a synaptic vesicle will go respectively to either a “readily releasable” or “reserve” pool of synaptic vesicles at the nerve terminal (74–77)^#. Thus, it is possible that proteins such as VGLUT1 (see below) that use slow and fast modes of endocytosis could show differential sorting to different pools of vesicles.

LDCV biogenesis

LDCVs (LVs in Figs. 1 and 2) are generated via a different route than synaptic vesicles and do not require endocytosis for their biogenesis (79). LDCV proteins are sorted into the regulated secretory pathway as they exit the TGN (80). The first step in the regulated secretory pathway is the generation of an immature large dense core vesicle (“im. LV” in Fig. 1B) (81); maturation requires the removal of some soluble and membrane proteins, represented in Fig. 1 as a budding event. Exit from the TGN and maturation of the LV has been suggested to require the clathrin adaptor complex AP1 (82) and the adaptor protein PACS-1 (83–85) (however see (86)).

Fig. 2.

Sorting signals have been identified in VMAT2 (A) VAChT (B) and VGLUT1 (C). All three proteins are likely to contain twelve hydrophobic domains (5, 112, 113, 151), but only ten may completely cross the membrane in VGLUT1 (152). As in Fig. 1, signals and pathways for sorting to SVs and SLMVs are colored purple; those relevant to LVs are orange. Sorting signals in VGAT/VIAAT are not known and the prediction of ten transmembrane domains (153, 154) has not been experimentally tested. A) In VMAT2 (and VAChT) a large lumenal loop between transmembrane domains 1 and 2 is glycosylated (orange “Y”). Glycosylation at this site may facilitate sorting of VMAT2 to LVs (see text). All other known sorting signals are encoded in the C-terminal, cytoplasmic domains of the vesicular transporters. For VMAT2, this includes a dileucine motif (IL) that is required for endocytosis and sorting to SLMVs in neuroendocrine cells (95). The endocytosis motif is embedded in a larger, extended dileucine motif that includes upstream glutamate residues (EE). The upstream glutamate residues are required for sorting into the regulated secretory pathway and onto LVs (44, 102). The initial step for sorting onto LVs is likely to occur at the TGN and involve sorting away from constitutive secretory vesicles (CS) and onto immature LVs (im. LV). A phosphorylated acidic cluster at the extreme C-terminus of VMAT2 (DDEE[P]SE[P]SD) helps retain VMAT2 in the LV as it matures. B) VAChT also contains a dileucine motif (LL) required for endocytosis and sorting to SLMVs (95, 96, 110). However, a novel tyrosine based motif YNYY may direct these trafficking events under some circumstances (111). A serine upstream of the dileucine motif ([P]S) in VAChT undergoes phosphorylation by PKC (102, 129). Substitution of an acidic residue to mimic phosphorylation drives a portion of VAChT into the regulated secretory pathway and on to LVs in PC12 cells (102). D) Endocytosis and recycling of VGLUT1 to SVs in hippocampal neurons requires a dileucine-like motif (FV). Under some circumstances, a polyproline motif that binds endophilin also contributes to VGLUT1 recycling to SVs (112). A distinct pathway that does not require the polyproline motif allows compensatory recycling of VGLUT, and requires the AP3 adaptor complex (112)(not shown).

3) Vesicular Transporter Sorting Signals

Vesicular transporters use several motifs involved in trafficking other membrane proteins (for a review see (87)). Previously defined motifs include the tyrosine-based motifs NPXY and YXXphe, where phe is a bulky hydrophobic residue. Dileucine-based signals contain either two leucines (LL) or a combination of other hydrophobic residues such as isoleucine plus leucine (IL) (88, 89). For most, if not all dileucine motifs, additional upstream acidic residues are required for the function of the motif as a whole and an acidic reside at position −4 and −5 (EXXXLL) facilitates binding to clathrin adaptor proteins (APs) (90–93). Acidic residues are also required in the acidic cluster that allows binding or the endopeptidase furin and other proteins to the PACS-1 adaptor (85, 94).

VMATs

Since both the biogenesis and recycling of SVs are thought to require an endocytic step, endocytosis signals are critical for the trafficking of all vesicular transporters. For VMAT2, a dileucine motif (IL) encoded in the C-terminus is required for its efficient endocytosis in PC12 cells as well as in hippocampal neurons (44, 95) (Fig. 2A). Moreover, the dileucine motif in VMAT2 may be sufficient for sorting to Synaptic-Like Microvesicles (SLMVs) in neuroendocrine cells; however, demonstrating this effect requires that trafficking to LDCVs is blocked (96, 97). This caveat makes it difficult to predict whether the dileucine motif in VMAT2 will be sufficient for sorting to synaptic vesicles in neurons, despite its demonstrated importance for endocytosis in neurons (44). Multiple forms of endocytosis may occur in neurons (98, 99), and proteins may sort to synaptic vesicles via alternate mechanisms (100, 101). Thus, it is possible that signals other than, or in addition to, the dileucine motif could help sort VMAT2 to synaptic vesicles in neurons.

Additional signals are required for the localization of VMAT2 to LDCVs. These include acidic residues upstream of the dileucine motif EEXXXLL (Fig. 2A) (102), that conserved in VMAT1 and DVMAT (103). In PC12 cells, VMAT2 mutants lacking the upstream acidic residues show dramatic defects in sorting to LDCVs (104). Furthermore, in PC12 cells as well as hippocampal neurons, these mutants traffic directly to the to the plasma membrane rather than entering the regulated secretory pathway (44). These data are consistent with the idea that the acidic residues in the dileucine motif of VMAT2 sort the transporter away from constitutive secretory vesicles and into the regulated secretory pathway at the TGN.

In most cases, both the hydrophobic and the acidic residues in the dileucine motif are thought to function as a single unit (87). Thus, it is likely that the hydrophobic residues in the VMAT2 dileucine motif (IL) help sort VMAT2 to LDCVs. However, it has not been possible to experimentally test this possibility since the IL in VMAT is also required for endocytosis and dileucine mutants are trapped at the plasma membrane (44).

An additional acidic cluster or patch at the end of the C-terminal domain (DDEESESD) is required for the localization of VMAT2 to LDCVs in PC12 cells (104), and it is possible that this motif performs a similar function in neurons. The acidic cluster motif was originally characterized in the protease furin, and for both VMAT2 and furin, it is thought to determine whether the protein is retained in LDCVs as they mature (85, 94, 104, 105). The serine/threonine directed kinase CKII phosphorylates both furin (105) and VMAT2 (106). For VMAT2, sorting to LDCVs is dramatically reduced either by deletion of the acid patch or by mutation of the phosphorylated serines to acidic residues (104).

Additional signals for localizing VMAT2 to secretory vesicles may lie outside the C-terminus (97, 107). Glycosylation has been implicated in the trafficking of synaptotagmin (108), and for VMAT2, a decrease in glycosylation may correlate with a decrease in its localization to synaptic vesicles in vivo (107). In addition, a recently described signal for localizing VMATs to LDCVs in PC12 cells is encoded in the N-linked glycosyl groups of the lumenal loop (97). The glycosylated loop is not itself sufficient to target VMAT2 to LDCVs; however, the C-terminal domain of VMAT2 is only competent for sorting to LDCVS if the lumenal loop is present and glycosylated (97).

VAChT

In non-neuronal cells and the cholinergic cell line SN56, mutation of the dileucine motif in VACHT (LL) inhibits endocytosis (55, 95, 109, 110). In addition mutation of novel tyrosine based motif (YNYY) in the C-terminal trafficking domain (111) of VAChT was shown to cause retention at the plasma membrane in at least one study (111) (See however (110)). These data suggest that VAChT may contain two distinct endocytosis motifs (111). The C-terminus of VAChT, which contains both motifs, is both necessary and sufficient for sorting to SLMVs (65, 110, 111). Furthermore, mutation of the LL motif disrupts the steady state localization of VAChT to SLMVs in SN56 cells (110) and also blocks sorting to SLMVs in PC12 cells (96). However, it is not known how the LL motif affects the trafficking of VAChT at the synapse. It remains possible that both the LL and YNYY motifs could contribute to VAChTs localization to synaptic vesicles in neurons, either during de novo biosynthesis and/or recycling at the nerve terminal.

VGLUTs

As is the case for VMAT2 and VAChT, a dileucine-like motif in VGLUT1 (FV) appears to play an important role for the endocytosis of VGLUT1 (112). However, unlike other vesicular transporters, VGLUT1 also contains a polyproline domain (PRPPPP) that works in concert with the dileucine motif (112–114). The polyproline motif binds to endophilin, and previous studies have demonstrated an important role for endophilin in both endocytosis and the remodeling of lipid membranes (115–119). Deletion of the polyproline motif in VGLUT1 has no effect on endocytosis when neurons are briefly stimulated (1 minute). In contrast, the same deletion decreases the rate of VGLUT1 endocytosis during longer periods of exocytosis and synaptic vesicle recycling (5 minutes) (112). Since neither VGLUT2 nor VGLUT3 contain the polyproline motif, they may recycle less efficiently under similar conditions (120).

Interestingly, the polyproline mutant in VGLUT1 is required for efficient endocytosis only during extended periods of active synaptic vesicle release; the polyproline deletion mutant undergoes endocytosis at a rate equivalent to wild type after exocytosis has ceased (112). Conversely, a distinct pathway is required for VGLUT1 endocytosis following prolonged periods of exocytosis (112). This pathway is likely to depend on the AP3 adaptor complex, since 1) it is sensitive to brefeldin A, and AP3-dependent budding requires the brefeldin sensitive small GTPase ARF1 (112); and 2) endocytosis of VGLUT1 through this pathway is blocked in mocha mice, which lack the AP3 delta subunit (61, 112).

These results are important for two reasons. First, they show that AP3 may function to recycle VGLUT1 to synaptic vesicle, regardless of the controversial role for AP3 in synaptic vesicle biogenesis (58, 60). Second, they establish at least two alternate modes of VGLUT1 recycling, one relatively fast and the other slow. (We exclude here the ultrafast mode of kiss and run recycling).

Other studies suggest that fast and slow endocytic pathways at the synapse may traffic respectively to the readily released versus the reserve pool of synaptic vesicles (74–76) and a recent study in Drosophila suggests that some vesicles in the reserve pool have a larger quantal size (36). It is tempting to speculate that regulated trafficking through the slow versus fast pathways could determine how many VGLUT molecules localize to synaptic vesicles that reside in different, functional pools.

Binding Partners

Trafficking of other membrane proteins requires a complex assembly of cytosolic factors and it is likely that vesicular transporters use many of the same components. Dominant negative constructs that inhibit the function of dynamin I (K44A), and clathrin (AP180-C) also block the internalization of VAChT, supporting the idea that vesicular transporters use clathrin and dynamin-based endocytic mechanisms (109, 110). In addition, in a few cases, vesicular transporters have been shown to directly bind adaptor proteins and other elements of the trafficking machinery. As noted above, VGLUT1 binds endophilin via a polyproline motif (112–114), and VMAT2 binds the adaptor PACS-1, possibly regulating exit from immature LDCVs (104). In addition, VAChT has been suggested to bind AP1 and AP2 (109, 111) via either the dileucine motif (109) or the novel YNYY motif (111).

In addition to these well-known adaptor proteins, it is possible that vesicular transporters require other proteins to interact with the sorting machinery. Two recent reports highlight the use of invertebrate genetic systems to isolate novel binding partners. In C. elegans, genetic experiments support a direct interaction between the mutant alleles of VAChT and synaptobrevin (121). Another genetic screen in C. elegans has shown that sorting of VGAT to synaptic vesicles requires the novel interacting protein unc-46 (122). In unc-46 mutants, VGAT labeling is increased at the cell body, and the remaining protein that does enter the axon is spread diffusely over the plasma membrane rather that localizing to synaptic vesicles (122). The mechanism by which unc-46 influences VGAT trafficking is not yet clear, but has been suggested to help recruit VGAT to synaptic vesicles either at the cell body or the nerve terminal (122).

4) The Regulation of Transporter Trafficking

Phosphorylation of VAChT and VMAT2 may regulate trafficking

Vesicular transporters may undergo multiple forms of regulation. These include regulated changes in transcription (123, 124) and the inactivation of transport by heterotrimeric G proteins (125, 126). Here we focus on mechanisms that may more directly affect trafficking.

In PC12 cells, VAChT localizes primarily to SLMVs, whereas VMATs localize primarily to LDCVs (127, 128). Similar to VMATs, the VAChT dileucine motif contains an upstream acidic residue at the −4 position, but unlike VMATs, VAChT shows a serine at −5 site (RSERDVLL), which can be phosphorylated by PKC (R[Phospho-S]ERDVLL) (102, 129). Phosphoserine is functionally an acidic residue, and thus similar to acidic glutamate in the VMAT2 dileucine motif (KEEKMAIL) (102, 129). Moreover, the substitution of an acidic amino acid at the phosphorylated serine (REERDVLL) increases the localization of VAChT to LDCVs three-fold in PC12 cells, partially mimicking the trafficking pattern of VMAT2 (102). These observations raise the possibility that VAChT trafficking may be regulated by phosphorylation, presumably at the level of the TGN and the sorting of VAChT into the constitutive versus the regulated secretory pathway (Fig. 2B).

VAChT resides primary on SLMVs in PC12 cells, and on synaptic vesicles in bona fide neurons (130, 131). However, as noted above, it is possible that VAChT may localize to other types of vesicles. Indeed, acetylcholine can be released from somatodendritic sites in vivo, and may localize to LDCV fractions derived from neuronal tissue (42, 43). It is therefore conceivable that in neurons as well as PC12 cells, phosphorylation by PKC could regulate the fraction of VAChT that is targeted to the regulated secretory pathway and to LDCV-like vesicles.

It is also possible that phosphorylation of VAChT regulates other trafficking events. Blockade of VAChT phosphorylation may disrupt trafficking to SLMVs in PC12 cells (129) and phosphorylation of the extended dileucine motif has been shown to regulate the endocytosis of the CD3 gamma subunit of the T cell receptor (132). Mutation of sites upstream of the dileucine motif does not appear to effect vesicular transporter endocytosis in non-neuronal or neuroendocrine cell lines (95). However, as noted above, trafficking in neuroendocrine cells and neurons may differ, and it is possible that VAChT phosphorylation could regulate endocytosis in neurons. In hippocampal slices, activation of PKC blocks the ability of the VAChT inhibitor vesamicol to inhibit ACh release, and this effect was correlated with an increase in VAChT phosphorylation (133). This effect would seem more likely to occur during endocytosis than on exit from the TGN.

For VMAT2, phosphorylation by CKII of the acid patch motif at the extreme C-terminus may regulate its localization to LDCVs (104). Substitution of acidic residues at the phosphorylation sites increases the localization of VMAT2 to immature LDCVS (104). Thus, similar to furin, the phosphorylation state of VMAT2 may determine whether it is removed from LDCVs as they mature.

Since CKII is constitutively active, phosphatases are more likely to regulate the function of the acid patch in VMAT2. For furin, the removal of phosphate from serines in the acidic cluster regulates exit from immature LDCVs (85). In addition, furin that reaches the plasma membrane can be recycled, and dephosphorylation may regulate its return to the TGN from the plasma membrane (134, 135). Interestingly, phosphorylated furin that is endocytosed from the plasma membrane traffics to the TGN indirectly through late endosomes, whereas dephosphorylated furin is sorted directly back to the TGN (135). For VMAT2, regulation of dephosphorylation could potentially determine the amount of VMAT2 that is removed from LDCVs as they mature (Fig. 2A). Inhibition of dephosphorylation could increase sorting into either the constitutive or constitutive-like secretory pathway, and potentially increase VMATs localization to synaptic vesicles.

VGAT also undergoes phosphorylation, and interestingly, phosphorylation of VGAT has been observed in neurons but not other non-neuronal cells (136). This difference suggests the possibility of a specific neuronal function for VGAT phosphorylation, but this remains untested.

Phosphorylation may also indirectly regulate vesicular transporter trafficking. PKA does not directly phosphorylate VMATs, but is required for the localization of VMAT1 and 2 to LDCVs in PC12 cells (137). Additional phosphorylation events associated with VAChT are suggested by a series of physiological studies on cholinergic signaling at the neuromuscular junction (see (138, 139)). However, further studies will be needed to elucidate the mechanisms underlying these phenomena and their potential contribution of VAChT trafficking.

Additional Regulatory Mechanisms

Additional regulatory mechanisms relevant to transporter trafficking include alternative mRNA splicing of exons representing trafficking domains, demonstrated previously for plasma membrane transporters (140–142). The Drosophila VMAT ortholog shows variable use of a 3′ splice site in the last exon of the gene, which leads to two divergent carboxy-terminal domains (103). As a result of this difference, only one version undergoes efficient endocytosis in vitro (103).

An extensive series of studies has shown that blockade of amine uptake at the plasma membrane increases the localization of mammalian VMAT2 to synaptic vesicles (143–145). In contrast, drugs that promote efflux such as amphetamines decrease the localization of VMAT to synaptic vesicles (146). These events may involve the upstream activation of dopamine autoreceptors (147, 148). However, the mechanism by which activation of autoreceptors regulates VMAT2 trafficking remains to be determined. Finally, recent data show that the expression of VGLUT1, and possibly its localization to synaptic vesicles, follows a circadian pattern (35). Additional mechanisms are no doubt involved in the regulation of vesicular transporter trafficking, and identifying these mechanisms remains one the current challenges for the field.

Future Directions

Many other questions about vesicular transporter trafficking remain unanswered. Although some motifs have been have tentatively assigned to particular trafficking events, they cannot fully account for all of the known pathways required for sorting to secretory vesicles. Thus, it is likely that additional signals remain to be identified. For example, the potential contribution of ubiquitination to vesicular transporter trafficking is not known. It is also unlikely that all transporter trafficking pathways have been identified. In neuroendocrine cells, trafficking to SLMVs has been suggested to occur in some cases via a late endosome intermediate (100), but the extent to which neurons use similar, alternative pathways is not known. We note that at least a portion of VGLUT1 would appear to localize to synaptic vesicles in the absence of known endocytosis motifs (112), supporting the notion that alternative targeting mechanisms are active in at least some neurons.

It is unknown whether vesicular transporters use specific signals for trafficking to the axon versus the somatodendritic compartment. Moreover, there is no information on the potential mechanisms by which vesicular transporters are degraded. This extreme length of many axons highlights the consequences of degrading a transporter at the nerve terminal. To be replaced, it must be synthesized and sorted into a precursor vesicle at a Golgi stack that might be a meter away.

For all of these known and potential trafficking events it remains unclear how variations between neuronal subtypes may influence transporter trafficking. Glutamatergic hippocampal neurons have proven to be a very useful model, but it is possible that variations in neurochemical identity will have dramatic effects on transporter trafficking. This possibility is underscored by the recent finding that loss of dynamin I differentially affects endocytosis at inhibitory versus excitatory synapses (149). New mammalian model systems representing a variety of different synaptic subtypes are likely to be required to fully understand these effects. Invertebrate systems such as C. elegans and Drosophila may also be useful in this regard, although the extent to which neuronal trafficking mechanisms are conserved is somewhat controversial.

Surprisingly, the number of vesicular transporters that reside on each secretory vesicle is not known. It has been suggested that as few as one in Drosophila (23) or as many as fourteen VGLUT molecules in mammalian preparations (24) may localize to an individual synaptic vesicle. One of the most important topics for future studies will be determining this number and how it may be regulated. Despite observed variations in neurotransmitter content, it is difficult to imagine that the synapse could tolerate large, random fluctuations in quantal size.

Finally, we believe that it will be critical to determine how changes in vesicular transporter trafficking may affect the nervous system as a whole. In vitro studies have provided tantalizing hints about the potential impact of trafficking on quantal size and synaptic transmission. However, we will not be able to determine how changes in transporter trafficking may affect behavior until we can study trafficking mutants in vivo.

Table 1.

Summary of Vesicular Transporter Types and Their Trafficking Signals

Transporter	Substrates	Established Trafficking motifs	Proposed Functions
VMAT1	Dopamine, Serotonin, Norepinephrine
VMAT2	Dopamine, Serotonin, Norepinephrine, Histamine	IL	Endocytosis in PC12 cells and in neurons.
		EExxx(IL)	Sort to immature LDCVS in PC12 cells. Sort to regulated secretory pathway in neurons.
VAChT	Acetylcholine	LL	Endocytosis. Sort to SLMVs in PC12 cells.
		YNYY	Endocytosis?
		[Phospho-S]xxxLL	Induce sorting to LDCVs in PC12 cells.
VGAT	GABA, Glycine	?
VGLUT1	Glutamate	FV (dileucine-like)	Endocytosis in neurons.
		PPPP	Endocytosis in neurons.
VGLUT2	Glutamate	(no PRPPPP)
VGLUT3	Glutamate	(no PRPPPP)