Interaction of Stoned and Synaptotagmin in Synaptic Vesicle Endocytosis

Stoned protein colocalization in endocytotic network domains

Our previous studies suggested that STNA and STNB may act cooperatively to regulate synaptic vesicle recycling events inDrosophila (Fergestad et al., 1999). Both proteins localize to the presynaptic compartment and occupy common subsynaptic domains. Recent work from a variety of laboratories has precisely defined spatial and functional domains within synaptic boutons at theDrosophila NMJ. These domains include the active zone, periactive zone, membrane-associated and internal vesicular pools, and a well defined “network” or “lattice” domain that exclusively localizes endocytotic proteins (Estes et al., 1996; Gonzalez-Gaitan and Jackle, 1997; Roos and Kelly, 1999). The endocytotic domain is of particular interest because of our hypothesis that the Stoned proteins mediate vesicular recycling. Previous studies have shown that α-Adaptin, a subunit of the endocytotic AP2 Clathrin-associated adapter complex, and Dynamin, the GTPase “pinchase” mediating endocytosis, both localize to the highly characteristic lattice occupying the area surrounding the active zone domains (Gonzalez-Gaitan and Jackle, 1997).

We first determined the localization of STNA and STNB relative to these well defined presynaptic domains. Both STNA and STNB proteins colocalize tightly with the endocytotic proteins α-Adaptin and Dynamin (Fig. 1A). All four proteins lie within the endocytotic lattice that surrounds but excludes the exocytotic active zones (Fig. 1A) (Gonzalez-Gaitan and Jackle, 1997; Roos and Kelly, 1999; Teng et al., 1999). Like Dynamin, both STNA and STNB are tightly associated with the plasma membrane and do not occupy cytosolic domains in the bouton interior (Fig. 1A). Markers of the active zone and vesicular pools, such as the SV-associated Cysteine String Protein (CSP), do not colocalize with the Stoned proteins but rather occupy the domains within the endocytotic lattice (Fig. 1B). This confocal analysis supports the localization of both STNA and STNB proteins with the endocytotic network and not with SV pools and areas of exocytosis.

The shibire^TS1 mutation disrupts Dynamin function and provides a temperature-dependent block in the vesicle-budding step of endocytosis (Kosaka and Ikeda, 1983). Stimulation of the Drosophila NMJ inshibire^TS1 mutants at the restrictive temperature (30°C) depletes the SV population because SVs are driven into the plasma membrane in the absence of endocytosis (Koenig and Ikeda, 1989). Immunological staining of these vesicle-depleted shibire^TS1terminals shows that SV markers, such as CSP, become associated exclusively with the plasma membrane (data not shown) (Estes et al., 1996). We labeled shibire^TS1SV-depleted terminals with antibodies against STNA, STNB, and α-Adaptin. No alteration in the endocytotic network, including the distribution of the Stoned proteins, was observed (data not shown). These studies confirm that both STNA and STNB are associated with the plasma membrane and do not associate with internal vesicles. Returning SV-depleted shibire ^TS1 terminals to the nonrestrictive temperature (22°C) allows endocytosis to resume, resulting in mass membrane retrieval from the plasma membrane (Koenig and Ikeda, 1989; Kuromi and Kidokoro, 1998). After SV depletion (30°C for 10 min) and brief recovery to permit massed endocytosis (22°C for 10 min), we again observed no detectable alteration in the expression pattern of the endocytotic proteins, including both STNA and STNB (data not shown). These data support the conclusion that the Stoned proteins occupy only the endocytotic domain within synaptic boutons and are tightly associated with the plasma membrane.

In previous studies, we documented a prominent mislocalization of Synaptotagmin I protein in stoned mutants (Fergestad et al., 1999), showing that stoned function is required to maintain Synaptotagmin I in tight synaptic bouton domains and to prevent its loss throughout the arbor and proximal regions of the axons and its eventual degradation. We wanted to determine whether this relationship was reciprocal by testing whether the Stoned proteins are mislocalized and/or degraded in the absence of Synaptotagmin I. Immunohistochemical studies in syt^AD4, a null allele of synaptotagmin (Broadie et al., 1994; DiAntonio and Schwarz, 1994), revealed no detectable alteration in either STNA or STNB expression at the embryonic NMJ. Both Stoned proteins were maintained in tight bouton puncta in the complete absence of Synaptotagmin I (data not shown). These data suggest that the Stoned proteins are specifically required for the recycling of Synaptotagmin I but do not require Synaptotagmin I for their localization within the endocytotic domains. Furthermore, these data suggest that the phenotypes observed insynaptotagmin mutants do not result from aberrant localization of the Stoned proteins.

Synapses in stoned mutants display decreased rates of plasma membrane endocytosis

Our previous studies of stoned mutants documented severely impaired synaptic transmission and a reduced number of morphologically abnormal SVs, suggesting a defect in vesicular recycling at the synapse (Fergestad et al., 1999). To assay for SV recycling defects directly, we used the fluorescent lipophilic dye FM1–43 in membrane retrieval and SV-recycling assays at the NMJ. Extracellularly applied FM1–43 dye is incorporated into SVs after endocytosis, reliably maintained in SVs and vesicular intermediates, and released during stimulated exocytosis (Betz and Bewick, 1992). Incubating Drosophila larval NMJ preparations with FM1–43 in a depolarizing solution of high K⁺ (90 mm) results in specific dye incorporation in SVs of the presynaptic boutons that can be released after subsequent depolarization and fusion (Ramaswami et al., 1994; Kuromi and Kidokoro, 1998). We used this analysis, in various experimental paradigms, to assay endocytosis and SV recycling in stoned mutant NMJs.

Wild-type and stoned mutant third instar animals were dissected in the same chamber and stimulated identically with 5 min of high-K⁺ saline in the presence of FM1–43. Control boutons revealed robust endocytosis and strongly incorporated dye, whereas stoned mutant boutons loaded dye very poorly under the same conditions (Fig.2A). The mean density of FM1–43 incorporation in larval NMJ boutons (>5 μm) was quantified and normalized to that of the control. The viablestn^C mutant animals display a significant impairment (53 ± 5% of control; p < 0.0001) in dye uptake (Fig. 2B). Examination of two lethal stoned alleles,stn^13–120 andstn^PH1, revealed an even more profound defect in endocytosis. Because severe stonedmutations are all embryonic lethal (Fergestad et al., 1999), normalized comparisons of FM1–43 dye uptake to control were done at the embryonic NMJ (Fig. 2C). Wild-type embryonic NMJs can be loaded with FM1–43 by high-K⁺ depolarization, and all synaptic boutons appear to label, generating a signal comparable with antibody staining against SV proteins at the same stage (21–23 hr AF). In sharp contrast, both lethal stoned alleles showed a >90% reduction in dye incorporation, making measurable endocytosis essentially undetectable (Fig. 2D). These studies show that three different alleles of stoned all show severe defects in exo–endo SV recycling at the NMJ synapse and that the severity of the recycling defect correlates with the severity of the mutant allele examined.

Although such a dye uptake impairment implies a direct defect in membrane retrieval, we cannot dismiss the possibility that the reduced exocytosis observed in stoned mutants (Stimson et al., 1998;Fergestad et al., 1999) causes a coupled reduction in endocytosis. To address this possibility, we examined spaced durations of depolarization stimulation and FM1–43 dye loading. Brief dye loading with high K⁺ for <1 min (30 sec and 1 min intervals assayed) resulted in a similar defect in stoneddye loading (data not shown; compare with Fig. 2B). Dye loading with 5 min of stimulation provided a slight increase in bouton fluorescence intensity in both control and mutant NMJ terminals, but the stoned-specific defect in endocytosis remained unaltered (data not shown). Furthermore, 10 min of high-K⁺ application and dye labeling resulted in no further increase in synaptic dye incorporation of either control or stoned mutant animals, suggesting that the cycling SV pool is maximally saturated after <5 min of high-K⁺ stimulation. Similarly, to determine whether the striking defects in endocytosis in the embryonic lethal mutants were caused by delayed endocytosis, we applied dye for 5 min in calcium-free saline after 5 min of high-K⁺ stimulation with dye. Longer periods of dye application did not improve the FM1–43 bouton labeling in either wild-type controls or stoned mutants (data not shown). These studies suggest that the recycling SV pool is saturated at these loading times and that stoned mutants have a specific and severe reduction in SV endocytosis. These findings show that the defect in dye uptake is independent of the stimulation duration and probably results from a smaller recycling pool of SVs instoned mutants.

Synapses lacking Stoned show spatially altered patterns of SV recycling

Large NMJ boutons (>3 μm, typically 3–5 μm) in normalDrosophila third instar larvae show characteristic patterns of SV pools (Ramaswami et al., 1994). Wild-type boutons loaded with high K⁺ always incorporate FM1–43 dye in a circular pattern with the fluorescence restricted to cortical regions underlying the plasma membrane and an absence of signal in the central regions of the bouton (Fig.3A). This dye incorporation shows that the recycling SV pool filled via high-K⁺ stimulation is spatially restricted in a characteristic peripheral ring. In contrast, the bouton interior contains a reserve pool of SVs that are accessed only under conditions of intense transmission demand (Kuromi and Kidokoro, 1998). Previous studies have shown that the reserve pool can only be loaded with FM1–43 after high-frequency (>30 Hz) stimulation or after complete elimination and mass renewal of the SV population with the Dynamin mutant shibire^TS1(Kuromi and Kidokoro, 1998, 2000). We wanted to determine whether the ready/reserve SV pool boundary is maintained in stonedmutants and whether Stoned proteins may play a role in the spatial dynamics of SV recycling.

In clear contrast to the normal condition, standard high-K⁺ FM1–43 labeling instn^C boutons resulted in the dye filling the entire bouton including the center of the bouton where reserve pool vesicles are normally located (Fig. 3, compareA, B). Recycled vesicles in stonedmutants lack the normal spatial restriction defining the readily releasable and reserve SV pools. This spatial distribution pattern is reminiscent of the FM1–43 signal after dye uptake in the Dynamin mutant shibire^TS1 (Kuromi and Kidokoro, 1998). After temperature-dependent depletion of all SVs,shibire^TS1 animals returned to the permissive temperature undergo mass membrane retrieval coinciding with the formation of early endosomes/cisternae and repopulation of the entire bouton with SVs (Koenig and Ikeda, 1989). We assayedshibire^TS1 SV dynamics by first loading the NMJ terminal at the permissive temperature (22°C; Fig.3C) and then reloading the same terminal after temperature-dependent (30°C) depletion of all SVs (Fig.3D). Before unloading, theshibire^TS1 boutons display a labeled circular pool of SVs corresponding to the readily releasable SV pool identical to that of wild-type controls (Fig. 3, compare A,C). However, after total SV depletion,shibire^TS1 terminals load both ready and internal reserve pools comparable with the pattern observed instoned mutants (Fig. 3, compare B, D). Thus, stoned mutants display aberrant trafficking of newly endocytosed membrane, which may be inappropriately targeted into sorting endosomes in the bouton interior. This conclusion is consistent with the significantly increased incidence of enlarged vesicles and multivesicular bodies we observed previously in stonedmutants at the EM level (Fergestad et al., 1999). This conclusion also supports our hypothesis that loss of stoned function results in increased segregation of membrane and/or protein to the sorting and degradation pathways, at the expense of the recycling SV pool.

Synapses in stoned mutants exhibit slower, but complete, release of FM1–43

Application of high-K⁺ saline to FM1–43-labeled synaptic boutons results in a second round of exocytosis that releases the dye contained within the SVs (unloading). Terminals loaded with FM1–43 for 5 min release most of this dye via the fast-cycling SV pool after a comparable 5 min unloading period (Fig. 4A). Under conditions of equal loading and unloading periods, the majority of dye in both wild-type (85.5 ± 1.0%) andstn^C (83.7 ± 5.3%) NMJ synapses is released via Ca²⁺-dependent exocytosis (Fig. 4A). In contrast, shortening unloading times to 1 min of high-K⁺ saline application was still sufficient to unload the majority of dye in wild-type terminals (88.1 ± 1.8%), but the amount of dye released from stoned boutons is significantly reduced (71.8 ± 3.4%; p = 0.0006; Fig.4B). These findings confirm that the readily releasable pool is smaller in stoned mutants, and although these vesicles are competent to fuse, they do so in a slower time course. The impairment of dye release is consistent with the defect in exocytosis observed in stoned mutants (Stimson et al., 1998;Fergestad et al., 1999) and may result from the aberrant vesicle trafficking observed in stoned terminals. These data further suggest that the aberrantly distributed SVs instn^C mutants are releasable and do not have the “barrier” thought to spatially separate the reserve and ready SV pools (Kuromi and Kidokoro, 1998).

Previously characterized defects in exocytosis and endocytosis suggested that SV maturation in stoned mutants may be impaired (Fergestad et al., 1999). Elegant studies on rat hippocampal cultures have recently estimated the time course for SV maturation (“repriming”) to be from 5 to 40 sec (Ryan et al., 1993; Ryan and Smith, 1995; Klingauf et al., 1998; Stevens and Wesseling, 1999). To test whether the delay period from endocytosis to exocytosis was increased in stoned mutants, we examined the lapsed time required before loaded dye could be released from NMJ boutons. FM1–43 dye was loaded with high-K⁺ saline (30 sec and 1 and 5 min), and then the preparation was washed in calcium-free saline for a variable period (Δt = 0 and 30 sec and 1, 5, 10, and 15 min) before K⁺-evoked unloading (Δt = 0 and 30 sec and 1 and 5 min). No significant change between controls and stoned mutant boutons in the amounts of FM1–43 release was detected in these assays (data not shown). These studies suggest that although fewer SVs are recycled via the endo–exo pool in stoned mutants, no difference in the rate of SV maturation was detectable.

Plasma membrane recycling in stoned mutants is delayed

Time-lapse studies using FM1–43 dye uptake assays indicate the rates of membrane retrieval after the fusion event to be τ_1/2 ∼20 sec (Ryan and Smith, 1995) or even faster (Kavalali et al., 1999). To determine whether endocytosis is delayed after exocytosis in stoned mutants we applied FM1–43 either during a 30 sec high-K⁺stimulus or for 30 sec immediately after the stimulus (Fig.5) (Ryan, 1999). In wild-type NMJ terminals, a 30 sec application of high-K⁺saline with FM1–43 loads synaptic terminals to levels similar to those of longer loading times (Fig. 5A), indicating that endocytosis is tightly temporally coupled to exocytosis. FM1–43 application for 30 sec immediately after the stimulation results in much lower levels of dye uptake, indicating that reduced endocytosis continues after the stimulation period (Fig. 5B). In contrast, stoned mutant boutons display greatly reduced endocytosis during the initial time period, when exocytosis and endocytosis levels are normally tightly coupled, and substantial levels of dye uptake only after the depolarizing stimulation. The striking dye uptake difference normally seen between stoned mutants and control animals is no longer present when the dye is added to the preparation after a 30 sec delay (wt = 104 ± 7;stn^C = 111 ± 11;p = 0.72; Fig. 5B). Because longer dye application times do not allow complete loading, the delay in loading the stoned SV pool cannot alone account for the decreases in overall dye uptake (see Fig. 2). Thus, stoned mutants show both a significantly delayed onset of endocytosis and a significantly smaller recycling SV pool.

Overexpression of Synaptotagmin I rescues thestoned phenotype

The Stoned proteins and Synaptotagmin I specifically interact in the presynaptic terminal. Both STNA and STNB have been shown to bind Synaptotagmin I directly (Phillips et al., 2000), and Synaptotagmin I is specifically mislocalized and subsequently degraded instoned mutants (Fergestad et al., 1999). Moreover, thestoned and synaptotagmin mutant phenotypes are strikingly similar; both show comparably decreased and nonsynchronous synaptic transmission, decreased synaptic vesicle density, and aberrant, enlarged synaptic vesicles (Reist et al., 1998; Fergestad et al., 1999). One hypothesis to explain these diverse findings is that the Stoned proteins and Synaptotagmin I mediate the same endocytotic function and that Stoned is required to recruit and/or maintain Synaptotagmin I during plasma membrane endocytosis.

A key prediction of this hypothesis is that elevated levels of Synaptotagmin I should alleviate the severe phenotypes observed instoned mutants. To test this hypothesis, we used neurally expressing GAL4 drivers to mediate expression of a UAS–Synaptotagmin I transgene construct, thus elevating Synaptotagmin I levels in synaptic boutons (Brand and Perrimon, 1993; Littleton et al., 1999). We first tested whether overexpression of Synaptotagmin I in the embryonic lethal stoned mutant background would rescue viability. Homozygous lethal stn^13–120animals, containing the UAS–Synaptotagmin I construct alone, remained embryonic lethal in the absence of a GAL4 driver. However, two temporally different neural GAL4 drivers both rescued the embryonic lethality of stn^13–120 in a manner consistent with the onset of their expression. First, the 1407-GAL4 driver expresses throughout the nervous system during embryogenesis (Sweeney et al., 1995) but ceases expression after hatching. When the 1407 driver was crossed to UAS–Synaptotagmin I;stn^13–120 animals, mutant animals now hatched (∼98%) at normal times (21 ± 1 hr AF) but then proceed to die as L1 stage larva, consistent with the termination of the 1407 expression. Second, the 4G-GAL4 driver is expressed during later stages of embryogenesis but then remains expressed in the nervous system throughout the life of the animal (K. Bodily and K. Broadie, unpublished observations). When driving Synaptotagmin I expression instn^13–120 mutants with the 4G driver, embryos also now hatch (∼96%), although this hatching is delayed (mutant animals now hatch between 21 and 35 hr AF), consistent with the later onset of expression. Furthermore, maintained Synaptotagmin I overexpression with 4G now rescues the embryonic lethal allele stn^13–120 to adult viability. These results show that elevated Synaptotagmin I can rescue the stoned lethality and that persistent elevated Synaptotagmin I is required to compensate for the loss of Stoned during maintained synaptic function.

The prediction from these studies is that elevated levels of Synaptotagmin I can alleviate the endocytosis defects caused bystoned mutation. To test this prediction, we assayed FM1–43 dye uptake at the larval NMJ. Overexpression of Synaptotagmin I instn^C mutants rescues the endocytotic functional defects observed in stn^C(Fig. 6A). UAS–Synaptotagmin I; stn^C larvae without a GAL4 driver are similar tostn^C mutants alone (stn^C = 52 ± 5%; UAS–Synaptotagmin, stn^C = 52 ± 6%) and show no rescue of the dye uptake defect. Strikingly, however, the introduction of the 4G-GAL4 driver almost completely rescues the defects in dye uptake (90 ± 7%; no longer significantly different from control; Fig. 6B). Interestingly, the overexpression of Synaptotagmin I alone in a control background (4G/UAS–Synaptotagmin I) shows a striking increase in the amount of dye loaded (147 ± 7.0%; p < 0.0001) compared with that of controls (Fig. 6). These findings show that thestoned mutant phenotypes can be directly rescued by elevation of Synaptotagmin I levels in the presynaptic terminal. The similarity of the stoned and synaptotagmin mutant phenotypes and the data presented here suggest that the sole role for the Stoned proteins may be to maintain the presynaptic function of Synaptotagmin I.

The Stoned proteins are required for SV recycling

The Stoned proteins were originally proposed to act as regulators of SV exocytosis (Stimson et al., 1998). More recently, ultrastructural studies in stoned mutants documented a decreased SV density throughout presynaptic boutons, abnormally large SVs, and an aberrant abundance and distribution of other vesicular intermediates (Fergestad et al., 1999), suggesting a role for the Stoned proteins in SV recycling. The subsynaptic distribution of both STNA and STNB, colocalizing the Stoned proteins with the endocytosis lattice network (Gonzalez-Gaitan and Jackle, 1997; Roos and Kelly, 1998), supports the hypothesis that the Stoned proteins are specifically involved in the endocytosis of SVs from the plasma membrane. In agreement with this hypothesis, the STNA protein contains five DPF motifs (α-Adaptin-interacting motifs; a subunit of AP2), and STNB has a region with strong homology to AP50 (also a subunit of AP2), contains a proline-rich N terminal (SH3 binding), and has seven NPF motifs (recognizing EH domains) (Andrews et al., 1996; de Beer et al., 1998;Stimson et al., 1998; Owen et al., 1999). STNB is not theDrosophila AP50 homolog, however, because a tightly defined AP50 gene has been characterized by us elsewhere in theDrosophila genome (Zhang and Broadie, 1999). Furthermore, STNA also contains the Clathrin β-propeller binding motif LL(D/E/N)φ(D/E) and four Yxxφ AP2 binding motifs, and STNB contains 12 Yxxφ AP2 binding motifs and a (D/E)xxxLL AP2 binding motif, all of which are thought to be involved in organizing the endocytotic machinery (Sengar et al., 1999; Pearse et al., 2000). Together, the molecular nature of the Stoned proteins, their subsynaptic localization, and the stoned mutant phenotypes strongly suggest that the Stoned proteins interact with the endocytotic machinery to mediate SV endocytosis.

SV endocytosis from the plasma membrane depends on the Stoned proteins

The FM1–43 dye uptake studies presented here demonstrate a specific endocytotic defect in stoned mutants. Both the viable and embryonic lethal stoned alleles have a smaller endo–exo-recycling SV pool, which is reduced in parallel to the severity of a stoned allelic series. These findings extend and support our ultrastructural analyses at the embryonic NMJ of both embryonic lethal and viable stoned alleles, showing depleted and aberrant presynaptic SV pools (Fergestad et al., 1999), but contrast with an ultrastructural study at the third instar NMJ on the viable stn^C allele that reported an increase in SV density (Stimson et al., 1998). Because of these paradoxical results, the data presented here suggest that the majority of SVs in the viable allele are not part of the cycling pool and/or are not fusion competent. The spatial localization of endocytosed dye within stoned terminals was found throughout the bouton, rather than restricted to the readily releasable SV domain (Kuromi and Kidokoro, 1998). This deficit may be caused by a lack of competent SVs near the active sites, resulting in a mobilization of SVs from the reserve pool near the center of the bouton (Kuromi and Kidokoro, 1998). A shunt of this type could result in recycled SVs being sent to the reserve pool domain in stoned mutants. Similarly, an accumulation of nonfusion competent SVs at the active site might occlude immediate transport of SVs from the endo–exo-cycling pool to the periphery of the bouton. This hypothesis might explain the significant delay in exocytosis in stoned mutants. However, the stoned mutants are capable of eventually exocytosing the FM1–43 marker at levels identical to that of wild-type controls. The release of ∼85% of the loaded dye in stoned mutants suggests that the majority of labeled SVs are capable of subsequent exocytosis. Thus, although exocytosis is delayed in stonedanimals and the recycling SV pool is reduced, the limited exo–endo pool can complete its cycle.

Retrieval of SV components from the plasma membrane is tightly temporally coupled to exocytosis (t_1/2< 20 sec) (Kavalali et al., 1999). Mutation of stoneddisrupts this temporal coupling and delays the onset and rate of endocytosis after exocytosis. Delayed application of FM1–43 tostoned synapses after stimulation reveals that the delayed component of endocytosis is comparable with wild-type levels. There are several possible explanations for these altered endocytosis kinetics instoned mutants. A likely rationale is that global membrane retrieval is delayed in stoned mutants, suggesting that Stoned proteins may play a role in either initiating endocytosis or facilitating the speed of endocytosis. An alternative scenario may be that there are multiple pathways for SV endocytosis, which differ in temporal kinetics, and that the rapid endocytosis mechanism is specifically impaired in stoned mutants. Although it is formally possible that delays in exocytosis might also explain the delayed membrane retrieval, our previous recordings of synaptic transmission show that exocytosis speed is only very minimally (<100 msec) impaired in stoned mutants (Fergestad et al., 1999).

A specific loss of Synaptotagmin I in stoned mutant synapses, as well as a global degradation and/or loss of Synaptotagmin I, was documented by us previously (Fergestad et al., 1999). Here we show that overexpression of Synaptotagmin I in severalstoned alleles rescues lethality and restores synaptic endocytosis function to wild-type levels. This finding is very persuasive for a specific role for the Stoned proteins in recycling and/or maintaining Synaptotagmin I at the synapse. Alternatively, because overexpressing Synaptotagmin I in control animals causes a striking increase in the size of the exo–endo pool, perhaps increasing the levels of Synaptotagmin I results in a general increase in synaptic function that may compensate in some nonspecific way for the depression of transmission in stoned. Curiously, synaptic potential amplitudes were not significantly increased in recordings ofDrosophila larvae overexpressing Synaptotagmin I (Littleton et al., 1999). Because increases in the exo–endo pool have been shown to correlate with increases in the quantal content of evoked transmission (Stevens and Sullivan, 1998; Kuromi and Kidokoro, 1999,2000), this discrepancy may be caused by a technical limitation of the voltage-recording method or saturation of the postsynaptic receptors by increased neurotransmitter release. The fact that both Stoned proteins bind specifically to Synaptotagmin I (Phillips et al., 2000), that Synaptotagmin I is specifically lost from the synapse instoned mutants (Fergestad et al., 1999), that the ultrastructural and electrophysiological phenotypes ofstoned and synaptotagmin mutant animals are strikingly similar (Reist et al., 1998; Fergestad et al., 1999), and that the interaction of synaptotagmin and stonedmutants is lethal (Phillips et al., 2000) strongly argues for the specificity of the Stoned–Synaptotagmin I interaction.

Recycling of Synaptotagmin I requires the Stoned proteins

How are the numerous different components of the SV recognized and recombined in the precise stoichiometric ratios required for synaptic function? There is an increasing body of evidence that a cast of specific recycling proteins is required to recognize and retrieve specific components of the SV during plasma membrane endocytosis. At center stage, the AP2 complex plays a prominent role in clathrin-mediated SV endocytosis (Cremona and De Camilli, 1997; Hannah et al., 1999). Genetic removal of the α-Adaptin subunit inDrosophila shows that the AP2 complex is absolutely required for the SV endocytotic process in this system (Gonzalez-Gaitan and Jackle, 1997). Moreover, the AP2 complex has been shown to bind Synaptotagmin I (Zhang et al., 1994), and it thus seems likely that this complex mediates the endocytotic recovery of Synaptotagmin, likely in addition to other integral SV proteins.

We propose that the role of the Stoned proteins may be to recycle Synaptotagmin I by mediating the association with the AP2 complex. It has been shown recently that both STNA and STNB bind with high specificity to Synaptotagmin I (Phillips et al., 2000) and therefore likely act in a cooperative manner in Synaptotagmin I retrieval. Studies by Haucke and De Camilli (1999) have shown that the AP2–Synaptotagmin I interaction can be stimulated by the presence of Yxxφ-containing peptides, which also enhance the recruitment of AP2 to the plasma membrane. Because both Stoned proteins contain multiple copies of these tyrosine-based motifs, as well as other AP2 and Clathrin binding domains (see above), we hypothesize that the Stoned proteins specifically promote the retrieval of Synaptotagmin I from the plasma membrane by mediating the AP2–Synaptotagmin I binding. Because the Stoned proteins are encoded by a dicistronic locus, polarity constraints have to date prevented an independent dissection of the roles of STNA and STNB in Synaptotagmin I endocytosis. Why does the process require two proteins, and what does each contribute to the recycling mechanism? Targeted homologous knock-out techniques are being used to explore these questions (Rong and Golic, 2000).

As the proposed calcium sensor for SV fusion, Synaptotagmin is likely to function as a key regulator of transmission strength. The number of Synaptotagmin proteins in a SV membrane may play an important role in regulating the response of the presynaptic terminal to depolarizing stimuli (Littleton et al., 1999). We have shown here that overexpression of Synaptotagmin I alone is capable of substantially increasing the size of the endo–exo SV pool. Therefore, the Stoned proteins, by regulating Synaptotagmin I recycling, also act as key regulators of neurotransmission strength. Future experiments are focusing on the specific regulation of Synaptotagmin I levels by each of the Stoned proteins.