Journal of Neurophysiology

AP180 Maintains the Distribution of Synaptic and Vesicle Proteins in the Nerve Terminal and Indirectly Regulates the Efficacy of Ca2+-Triggered Exocytosis

Hong Bao, Richard W. Daniels, Gregory T. MacLeod, Milton P. Charlton, Harold L. Atwood, Bing Zhang

Abstract

AP180 plays an important role in clathrin-mediated endocytosis of synaptic vesicles (SVs) and has also been implicated in retrieving SV proteins. In Drosophila, deletion of its homologue, Like-AP180 (LAP), has been shown to increase the size of SVs but decrease the number of SVs and transmitter release. However, it remains elusive whether a reduction in the total vesicle pool directly affects transmitter release. Further, it is unknown whether the lap mutation also affects vesicle protein retrieval and synaptic protein localization and, if so, how it might affect exocytosis. Using a combination of electrophysiology, optical imaging, electron microscopy, and immunocytochemistry, we have further characterized the lap mutant and hereby show that LAP plays additional roles in maintaining both normal synaptic transmission and protein distribution at synapses. While increasing the rate of spontaneous vesicle fusion, the lap mutation dramatically reduces impulse-evoked transmitter release at steps downstream of calcium entry and vesicle docking. Notably, lap mutations disrupt calcium coupling to exocytosis and reduce calcium cooperativity. These results suggest a primary defect in calcium sensors on the vesicles or on the release machinery. Consistent with this hypothesis, three vesicle proteins critical for calcium-mediated exocytosis, synaptotagmin I, cysteine-string protein, and neuronal synaptobrevin, are all mislocalized to the extrasynaptic axonal regions along with Dap160, an active zone marker (nc82), and glutamate receptors in the mutant. These results suggest that AP180 is required for either recycling vesicle proteins and/or maintaining the distribution of both vesicle and synaptic proteins in the nerve terminal.

INTRODUCTION

Clathrin-mediated endocytosis is thought to play a major role in replenishing the synaptic vesicle (SV) pool in nerve terminals. Along with a large number of accessory proteins, clathrin recycles SVs through several distinct sequential steps, including formation of coated vesicles, fission of endocytotic vesicles from the plasma membrane, and removal of the clathrin coat (reviewed by Brodin et al. 2000; Brodsky et al. 2001; De Camilli and Takei 1996; Zhang and Ramaswami 1999). Unlike the “kiss-and-run” mode of endocytosis, which is believed to recycle SVs that do not intermingle significantly with the plasma membrane during exocytosis (Fesce et al. 1994; Palfrey and Artalejo 2003), clathrin-mediated endocytosis internalizes SVs that collapse into the plasma membrane (Heuser and Reese 1973). Hence, an important step in this pathway is to sort and recruit cargo components unique to SVs to ensure that newly recycled vesicles are fully competent for exocytosis.

AP180 is one of the clathrin accessory and assembly proteins promoting the formation of clathrin-coated vesicles (Ahle and Ungewickell 1986; Murphy et al. 1991; Ye and Lafer 1995) and regulating the size of SVs (Morgan et al. 1999; Nonet et al. 1999; Zhang et al. 1998). AP180, alone with the adaptor AP2 and a putative adaptor Stoned, is also thought to act as an adaptor protein sandwiched between the clathrin coat and the cargo membrane. Because of their strategic locations within coated vesicles, these adaptors are also implicated in assisting the retrieval of SV proteins (De Camilli and Takei 1996; Zhang 2003; Zhang et al. 1999). The results obtained from recent genetic and biochemical studies appear to support this notion. Stoned has adaptor-like and clathrin- and eps15-binding motifs (Stimson et al. 1998) and directly interacts with synaptotagmin I (Phillips et al. 2000). Mutations in the stoned locus mislocalize synaptotagmin I and cysteine-string protein (CSP) to the axonal membrane and impair synaptic transmission regardless of whether or not the number of vesicles remains normal (Fergestad and Broadie 2001; Fergestad et al. 1999; Phillips et al. 2000; Stimson et al. 1998; 2001). Biochemical studies reveal that AP2 interacts directly with synaptotagmin I, suggesting that it may help retrieve synaptotagmin I during endocytosis (Hao et al. 1999; Jarousse et al. 2003; Zhang et al. 1994). AP180 binds clathrin and AP2 (Ahle and Ungewickell 1986; Hao et al. 1999), making it likely that AP180 may act synergistically with AP2 to help retrieve synaptotagmin I into SVs. AP180 mutants in both Drosophila (lap) and Caenorhabditis elegans (unc-11) are uncoordinated in locomotion and defective in synaptic transmission even though only 1/3 of the total vesicle pool is lost in the lap mutant (Zhang et al. 1998) and none of the pool in the unc-11 mutant (Nonet et al. 1999). Unexpectedly, synaptobrevin is mislocalized to axonal membranes in the unc-11 mutant, suggesting that UNC-11 may help recycle synaptobrevin during endocytosis (Nonet et al. 1999). These observations lead to the hypothesis that vesicle recycling can be decoupled from retrieval of SV proteins and that the “quality” of SVs could significantly influence exocytosis. In Alzheimer's patients, the level of AP180 is significantly reduced prior to the loss of synapses, implicating a functional role for AP180 in cognitive integrity (Yao 2004).

However, it remains unclear whether LAP also plays a role in vesicle protein retrieval and how such a defect might affect exocytosis. In addition, recent studies of the Drosophila endophilin (endo) (Verstreken et al. 2002) and synaptojanin (synj) (Verstreken et al. 2003) mutants have questioned whether depletion of SVs readily translates to a reduction in single action-potential-evoked basal transmitter release. In these mutants, the vesicle pool is severely depleted, but the basal transmitter release remains normal, suggesting that factors other than vesicle numbers may play an important role in transmitter release. To address these questions and to advance our understanding of AP180, we have further characterized the lap mutant using electrophysiology, optical imaging, electron microscopy, and immunocytochemistry. Our results reveal a novel role for AP180 in maintaining the efficacy of calcium (Ca2+) coupling to exocytosis at steps downstream of calcium entry and vesicle docking. Furthermore, our studies show that not only neuronal synaptobrevin (n-Syb), but also CSP, synaptotagmin I, Dap160, the active zone marker nc82, and glutamate receptors are mislocalized to extrasynaptic regions along the axon. Finally, the morphology of synapses is changed from the typical “beads-on-a-string” appearance to “rope”-like shapes. These findings suggest that AP180 indirectly regulates the efficacy of Ca2+-triggered exocytosis by either helping recycle SV proteins and/or maintaining the proper subcellular distribution of synaptic and vesicle proteins in the nerve terminal. The redistribution of both pre- and postsynaptic components may in part reflect developmental alterations to compensate for the impaired synaptic transmission in the lap mutant.

METHODS

Fly stocks and genetics

The lap mutant, a deficiency uncovering the lap locus (Df1801), and the Canton Special (CS) wild-type strain were described previously (Zhang et al. 1998) and used in all our studies presented here. These flies were raised at either room temperature (20–23°C) or 19°C on regular cornmeal-based fly foods. The null or severe loss of function mutant of the lap gene was obtained as nontubby transheterozygotes (i.e., lap/Df) by crossing the lap mutant with its deficiency, each of which was balanced with a TM6B balancer (lap/TM6B and Df1801/TM6B). Previously, we have demonstrated that the lap mutant chromosome harboring an insertion of P element only affects the lap locus and that the revertant of this line was essentially identical to the CS wild type (Zhang et al. 1998). We believe that CS can best serve as a wild type for the lap mutant after losing the revertant line. Additional alleles of lap identified by us (unpublished) and by others (Babcock et al. 2003) display similar electrophysiological defects as the initial lap allele (see results). This further confirms the specificity of the lap mutation.

Immunocytochemistry and microscopy

Larval body-wall muscle neuromuscular junction (NMJ) preparations were dissected and fixed in either paraformaldehyde (4%) or Bouin's fixatives. After an hour incubation with a blocker solution (2–5% bovine serum albumin, 0.2% Tween-20 in PBS), synaptic proteins were stained overnight at 4°C using primary antisera, followed by 1–2 h incubation with secondary antibodies at room temperature, and visualized using a Nikon fluorescence microscope. The fluorescence signal in the mutant was generally reduced compared with that in the wild type (not shown), likely due to loss of SVs. The final pictures presented here were taken on laser scanning confocal microscopes with the fluorescence signal adjusted to equal levels in the mutant and the wild type. Images presented in Fig. 8 were obtained by using a Leica 4D TCS microscope with a ×40 and/or ×100 oil objective (1.4 numerical aperture and 0.41 μm Z steps). Images presented in Figs. 9 and 10 were obtained by using a Leica SP2 AOBS microscope with a ×63 oil objective (1.4 numerical aperture and 0.122 μm Z steps). The following primary antibodies were used in our experiments: antibodies to SV proteins (cysteine-string protein, CSP, mAB49, used at 1:50, courtesy of Dr. Konrad Zinsmaier; synaptotagmin I, rabbit polyclonal, used at 1:500, courtesy of Dr. Noreen Reist; n-synaptobrevin, rat polyclonal, used at 1:400, courtesy of Dr. Hugo Bellen; Dap160, rabbit polyclonal, used at 1:200, courtesy of Drs. Jack Roos, Regis Kelly, and Hugo Bellen, and glutamate receptor III, rabbit polyclonal, used at 1:300, courtesy of Dr. Aaron DiAntonio; the active zone mAB nc82, used at 1:50, courtesy of Dr. Erich Buchner) and an FITC-conjugated antibody to HRP (goat polyclonal, used at 1:200, Jackson ImmunoResearch Laboratories). Secondary antibodies used in this study include anti-mouse Texas Red, anti-rat TRITC, and anti-rabbit FITC (Jackson ImmunoResearch Laboratories), all at 1:200.

Electron microscopy

Dissected NMJ preparations were fixed and prepared for transmission electron microscopy according to the methods established by Atwood et al. 1993. Briefly, dissected samples were fixed for 2 h at room temperature in a fixative containing 0.1 M phosphate buffer (PB), 3% glutaraldehyde, 4% sucrose, and 0.5% formaldehyde. Fixed samples were then washed for 60 min with PB and postfixed for 1 h with 1% osmium tetroxide in 0.1 M PB. After postfixation, samples were washed three times in PB. Samples were then dehydrated in an ethanol series and embedded in an Epon-Araldite mixture. Longitudinal muscles 6 and 7 in segment 3 and 4 were sectioned transversely to locate nerve terminals on target muscles. Thin sections (75 nm) were mounted on Formvar-coated single-slot grids, doubled stained with uranyl acetate and lead citrate, and examined by a Hitachi H-7000 transmission electron microscopy at 75 kV. Chemical reagents were purchased from Ted Pella (Redding, CA).

Vesicles near synapses in type I boutons were sampled in serial sections to compare vesicle availability in mutant and wild-type larval nerve terminals. Morphologically docked vesicles were defined as those touching or within 20 nm of the presynaptic membrane beneath or near the electron-dense T-bar, which is thought to be the focal point and putative active zone for vesicle release in fly synapses (Feeney et al. 1998; Koenig et al. 1993). Physically, vesicles further than 20 nm from the presynaptic membrane are probably not releasable by a single nerve impulse (Parsegian 1977). We also sampled two other populations of synaptic vesicles: those within 50 nm of the densely stained presynaptic membrane and those within 500 nm, which includes almost all vesicles within reasonable range of an individual synapse. The number of docked vesicles is excluded from these two vesicle pools. We define a “synapse” as the densely stained, closely apposed pre- and postsynaptic membranes evident in electron micrographs and an “active zone” as the prominent T-bar structure with its attendant cluster of synaptic vesicles (Atwood et al. 1993). Limited freeze-fracture images of fly synapses suggest that the presynaptic calcium channels at active zones responsible for nerve-impulse-evoked release of neurotransmitter are arranged loosely around the T-bar on the presynaptic membrane (Feeney et al. 1998).

Electrophysiology

The third instar body wall muscle-synapse preparation was dissected as described previously (Jan and Jan 1976) and used for both intracellular recording and two-electrode voltage clamping in longitudinal muscles 6 and 7 bathed in HL-3 saline (Stewart et al. 1994; Zhang et al. 1998). Membrane potentials or synaptic currents were amplified through an Axoclamp 2B amplifier (Axon Instruments), filtered at 2 kHz, digitized and recorded on a Dell PC computer equipped with pClamp8 software (Axon Instruments). Microelectrodes (filled with 3 M KCl) with an input resistance of 12–18 MΩ were used for voltage clamping and of 20–32 MΩ for intracellular recording. Muscle input resistance was routinely monitored with a 1-nA current injection and by using the bridge mode on the Axoclamp 2B amplifier to cancel out the series resistance from the microelectrode. Analysis was included only in muscles with an input resistance >5 MΩ. Spontaneous miniature excitatory junction potentials (mEJPs or minis) were obtained by intracellular recording from muscles bathed in HL-3 solutions containing 1 mM Ca2+ and 1 μM tetrodotoxin, which may eliminate spontaneous nerve firing and reduces the probability for multiquantal release (Zhang et al. 1998). Excitatory junction currents (EJCs) were evoked by directly stimulating the innervating motor axons with a suction electrode at 0.2 Hz. The extracellular Ca2+ concentration varied from 0.4 to 10 mM and it is further specified in results.

To obtain the amplitude of EJCs at low [Ca] for the Ca2+ cooperativity plot, we stimulated the motor nerve 20 times at 0.2 Hz and measured EJCs as well as the number of failures. Concerned about the possibility that the high mini frequency in the lap mutant could contaminate the amplitude of these evoked EJCs and thereby reduce the slope of Ca2+ cooperativity plot, we decided to correct the measured EJC amplitude for minis that are by chance coincident with the EJCs. As first suggested by Schwarz and colleagues (Robinson et al. 2002), we arbitrarily chose a point 110 ms after the stimulus artifact and collected all minis for which the peak falls at this point. The average of amplitude of these minis was then subtracted from the average EJC amplitude. The average EJC amplitude for each muscle fiber was obtained by dividing the sum of the adjusted EJC amplitude with the total number of stimuli, including failures. Because mini frequency is typically higher after nerve stimulation, it is likely that we may have overcorrected for the number of spontaneous minis. However, such an over correction would change Ca2+ cooperativity in the opposite direction observed in our studies.

Analysis of the amplitude and frequency of mEJPs or EJCs was conducted using Mini Analysis (Synaptsoft) and Origin (OriginLab). Miniature EJPs with a slow rise time arising from electrically coupled neighboring cells (Gho 1994; Zhang et al. 1998) were excluded from the final analysis. The unpaired Student's t-test was used for data statistics. ANCOVA test was used for treatment of the slope of Ca2+-dependence. Results were considered statistically different when the P value was <0.05.

Calcium imaging

The Ca2+ indicator, dextran-conjugated (10 kDa) Oregon Green 488 bis-(o-aminophenoxy)-N,N,N′,N′-tetraacetic acid 1 (BAPTA-1; OGB-1, Molecular Probes, Eugene, OR), was loaded into motor nerve terminals using the forward filling technique described by Macleod et al. (2002). Briefly, the preparation was scanned through the objective (×40, 0.55 numerical aperture) of a Nikon Optiphot upright microscope fitted with a BioRad 600 scan head (BioRad; Mississauga, Ontario, Canada) using an Argon ion laser (at 0.5% intensity) in combination with a BHS filter set (exciter filter: 488 nm DF10; emission filter: OG 515 nm LP; dichroic reflector: 510 nm LP). Scanned images were acquired to PC and processed using ImageJ software (http://rsb.info.nih.gov/ij/). Fluorescence values (F) were calculated for scan lines through OGB-1-filled boutons by numerically subtracting the average pixel intensity of the background in the scan line, containing no OGB-1, from the average pixel intensity of the entire line. The effect of bleaching was removed by subtracting a trendline fitted to F when no stimulus pulses had been applied to the nerve. ΔF/F is calculated as the change in FF) divided by the resting level of F. The data were fit to a first-order exponential in Sigma Plot (SPSS, Chicago, IL). Unpaired 2-tailed t-tests were used to test differences in means.

Changes in the plateau level of Ca2+ indicator fluorescence in larval motor nerve terminals, in response to nerve stimulation trains, have previously been quantified using a frame scan protocol (Macleod et al. 2002). In this study, we alternated frame scan data and line scan data acquisition from the same type-1b while stimulating at 20 Hz. A regression of the estimates of the plateau maximum ΔF/F attained from line scan data analysis on the estimates attained from frame scan data analysis gave an X variable coefficient of 0.96 and a correlation coefficient of 0.95 (P < 0.001). This analysis demonstrates that estimates derived from line scan data will, on average, be the same as those derived from frame scan data. All Ca2+ imaging was performed in HL6 containing 0.5 mM Ca2+ at room temperature (∼22°C).

RESULTS

Evoked transmitter release is dramatically reduced in the lap mutant

Synaptic terminals spontaneously release transmitter in the absence of the nerve impulse (Fatt and Katz 1952). The frequency of spontaneous release is largely a presynaptic property, although retrograde signaling can influence it. To record the mEJP, we used sharp electrodes to impale third instar larval body-wall muscles bathed in saline containing 1 mM Ca2+ and 1 μM tetrodotoxin (TTX). The average resting potential of muscle fibers in the lap mutant was indistinguishable from that in the wild type (CS, –66.4 ± 0.9 mV, n = 10; lap/Df, –65.9 ± 0.5 mV, n = 11; P > 0.5). In addition to the increased amplitude of minis reported earlier (Zhang et al. 1998), the frequency of mEJPs was 2.5-fold higher in the lap mutant (9.1 ± 0.4 Hz, n = 11) than in the wild type (3.7 ± 0.4 Hz, n = 10) in the presence of Ca2+ (Fig. 1, A and B; P < 0.001).

FIG. 1.

Evoked transmitter release is dramatically reduced, whereas the frequency of spontaneous release is increased in the lap mutant. A and B: frequency of spontaneous miniature excitatory junction potentials (mEJPs or minis) is significantly increased in the lap mutant compared with the CS wild type (***P < 0.001). C and D: excitatory junction currents (EJCs) obtained by 2-electrode voltage clamp from muscles held at –80 mV are also significantly reduced in amplitude in the mutant (***P < 0.001). The extracellular [Ca] was 1 mM calcium for all of the experiments described in this figure.

In sharp contrast with the increase in mini size and frequency, evoked transmitter release is dramatically reduced. In an earlier report, we showed that evoked EJPs were smaller in amplitude in the lap mutant but did not quantify them or measure their Ca2+ dependence (Zhang et al. 1998). Here, we confirmed the reduction of quantal release by measuring the EJC under voltage clamp with the muscle membrane potential clamped to –80 mV to maintain a constant driving force (Fig. 1, C and D). The average amplitude of EJCs was significantly reduced from 67.9 ± 2.7 nA (n = 13) in the wild type to 4.8 ± 0.7 nA (n = 14) in the mutant (P < 0.001). The average mEJC amplitude at the –80 mV holding potential was determined to be 0.61 and 0.98 nA for the wild type and the mutant, respectively (Zhang et al. 1998). Accordingly, the reduction in quantal content is estimated to be 96%, down from ∼111 quanta in the wild type to ∼5 quanta in the lap mutant. It is important to note that this reduction in quantal content is not a result of exhaustion of SVs in the nerve terminal caused by repetitive nerve stimulation. In fact, this small EJP or EJC amplitude was observed when we stimulated the motor nerve after a long rest after dissection with either a single stimulus or five stimuli delivered at a low frequency (≤0.2 Hz). Intracellular recording from a second lap allele, lapSD3 (Babcock et al. 2003), in trans with the original lap mutant or the lap deficiency also showed a similar reduction in quantal release. In lap/lapSD3 and lapSD3/Df larvae, the average amplitude of evoked EJPs obtained in 1 mM [Ca]-containing saline was 10.7 ± 1.6 mV (n = 5) and 6.2 ± 1.1 mV (n = 4), respectively. The amplitude of these EJPs was similar to that in the lap/Df larva (8.8 mV ±1.0 mV, n = 11). However, all these lap alleles showed a dramatic reduction in evoked EJP amplitudes, which was 40.4 ± 1.6 mV (n = 10) in the wild type (CS; P < 0.001). These results provide further genetic evidence that the lap mutation likely deletes the lap locus. In following studies, we hence focused on the original lap allele. We and others have previously shown that quantal size increases concurrently with the enlargement of synaptic vesicles, suggesting that postsynaptic receptors most likely remain unaffected in the mutant (Karunanithi et al. 2002; Zhang et al. 1998). Immunoreactivity of glutamate receptors shows no apparent differences in the overall intensity of receptor clusters for the mutant and the wild type (see Fig. 10). Hence, the great attenuation of evoked response originates primarily at the presynaptic side in the mutant neuromuscular junction (NMJ).

lap mutant displays a strikingly low release probability

Calcium not only triggers vesicle fusion, but also controls the amount of transmitter release (Katz 1969). To probe possible primary defects in presynaptic terminals, we examined the efficiency of Ca2+coupling to transmitter release at the NMJ by measuring the failure rate defined as the fraction of stimuli failing to evoke transmitter release at different [Ca]. At low extracellular [Ca] (≤0.4 mM), we failed to evoke release from the mutant larvae while succeeding 80% times in the wild-type larvae (not shown). Further examination and analysis of failure rates in slightly higher [Ca] are shown in Fig. 2. At 0.6 mM [Ca], the lap mutant showed a high failure rate (32.1 ± 5.2%, n = 7, P < 0.001), whereas the wild type had no failures at all (a 100% success rate, n = 9) after nerve stimulation at 0.2 Hz (Fig. 2A). Most surprisingly, the mutant persistently showed considerable numbers of failures even when the extracellular [Ca] was further raised ≤0.9 mM (Fig. 2B). At these Ca2+ concentrations, the wild-type larvae did not show any failures. The failure rate for the mutant was 21.9 ± 7.2% (n = 8) at 0.7 mM [Ca], 16.9 ± 7.1% (n = 8) at 0.8 mM [Ca] and to 3.1 ± 1.3% (n = 8) at 0.9 mM [Ca]. These unusually high failure rates suggest that the release probability is extremely low in the lap mutant, which may account for the dramatic reduction in quantal content.

FIG. 2.

The lap mutant displays high rates of synaptic failures in response to low-frequency nerve stimulation. A: representative traces of synaptic currents evoked by nerve stimulation in 0.6 mM extracellular [Ca]. While the wild-type larvae faithfully respond to each stimulus, the lap mutant displays a significantly higher rate of failures than the wild type. →, stimulus artifacts. B: histograms of failure rates for the wild type (CS) and the lap mutant (lap/Df) at [Ca] ranging from 0.6 and 0.9 mM. At these [Ca]s, the mutant shows significantly higher failure rates compared with the wild type, which has no failures at all (**P < 0.001; *P < 0.05). Synaptic failures observed at 0.8 and 0.9 mM Ca2+ are striking and have not been reported in other mutant flies.

Calcium sensitivity and cooperativity are both reduced in the lap mutant

The high failure rate observed in the lap mutant indicates that Ca2+-sensitivity may be reduced. To test this possibility, we attempted to “rescue” the defect in exocytosis by increasing the extracellular [Ca]. In the presence of high [Ca] in the saline, the amplitude of EJCs in the wild type increased dramatically in ≤6 mM saline and then declined at 10 mM for unknown reasons. In contrast, the EJC amplitude in the lap mutant was only slightly increased (Fig. 3A). These results and the low release probability together suggest that Ca2+ coupling to exocytosis, Ca2+ entry into the nerve terminal, or the readily releasable pool of SVs is reduced in the mutant.

FIG. 3.

Ca2+ sensitivity and cooperativity are both reduced in the lap mutant. A: dose-dependence of evoked excitatory junction currents (EJCs) on extracellular [Ca]. In the wild type, the EJC amplitude increases significantly when the extracellular [Ca] is increased from 0.4 to 5 mM. At 10 mM Ca2+, EJC amplitude is reduced. In contrast, the lap mutant shows only a slight increase in the amplitude of EJCs when [Ca] is raised. B: logarithmic plot of the EJC amplitude against the extracellular [Ca]. The slope of these lines represents Ca2+ cooperativity, which is 3.9 and 2.1 for the wild type and the mutant, respectively. These slopes are significantly different (P < 0.05; ANCOVA test). In all the experiments shown here, the nerve was stimulated at 0.2 Hz, and data were obtained from ≥5 animals from each genotype.

To further examine Ca2+ sensitivity, we determined Ca2+ cooperativity by measuring the slope of the logarithmic plot of EJC amplitude against [Ca] at low concentration ranges. At low Ca2+ concentrations, it is suggested that at least four Ca2+ ions act in a cooperative fashion to trigger exocytosis (Dodge and Rahamimoff 1967; Smith et al. 1985). As predicted, the logarithmic plot of EJC amplitude against extracellular [Ca] ranging from 0.5 to 1 mM shows a linear relationship for both the wild type and the mutant (Fig. 3B). However, the slope of these two lines is significantly smaller for the lap mutant (slope = 2.1) than for the wild type (slope = 3.9, P < 0.05), demonstrating that Ca2+ cooperativity of transmitter release is significantly reduced in the mutant. As noted in methods, we undertook careful measures to eliminate the potential contamination of EJCs by spontaneous minis that could skew toward a shallow slope for the mutant. At present, it is unknown how this change of the slope comes about. Nonetheless, the low Ca2+ cooperativity is consistent with the possibility that Ca2+ is less effective in triggering transmitter release, perhaps due to a change either in Ca2+ sensors on the vesicle (Littleton et al. 1994) or the release machinery (Stewart et al. 2000).

Exocytotic defect occurs at steps downstream of calcium entry

Our measurements of the failure rate, Ca2+ sensitivity and cooperativity assume no change in Ca2+ entry between the mutant and wild-type nerve terminals. However, with the exception of Ca2+ cooperativity, all other physiological defects could arise from impaired Ca2+ entry to the nerve terminal in the lap mutant. To investigate this possibility, we examined synaptic Ca2+ influx in larval NMJs using optical imaging (Macleod et al. 2002). The Ca2+-sensitive dye dextran-conjugated Oregon Green 488 BAPTA-1 (OGB-1) was loaded from the cut end of the segmental nerve and allowed to diffuse and equilibrate within synaptic boutons for 1 h prior to imaging. We measured Ca2+ influx by line scanning of single boutons after nerve stimulation with single pulses as well as a short train of stimulation at 20 Hz in a total of three wild-type and seven mutant larvae. Our results showed that the relative change in fluorescence intensity and the decay time constant after a single stimulation were indistinguishable between the wild-type control (24.6 ± 5.4%, tau 71.7 ± 9.8 ms, n = 6 cells) and the mutant (27.1 ± 4.2%, n = 14; tau 66.5 ± 6.3 ms, n = 13; Fig. 4, A–D). In response to a repetitive stimulation (20 Hz, 0.5 s), the rise time, average accumulation, and decay kinetics of intraterminal Ca2+ also remained unaffected by the lap mutation (Fig. 4, E–I). The average maximal change of fluorescence was 38.2 ± 4.2 (n = 5) in the wild type and 41.1 ± 5.3% in the mutant (n = 12). The decay time constant was 170 ± 6 ms (n = 4) in the wild type and 203 ± 32 ms (n = 9) in the mutant, but they are not statistically different (P = 0.34), suggesting that the mutant clears its Ca2+ at a normal rate. The normal entry and kinetics of terminal Ca2+ strengthen our argument that Ca2+ coupling to exocytosis is severely disrupted in the lap mutant.

FIG. 4.

Ca2+ influx and decay are both normal at the lap mutant terminal. A–D: calcium influx and decay kinetics within single boutons (type 1b) after nerve stimulation by single pulses. Calcium influx and decay were measured as fluorescence changes of dextran-conjugated Oregon Green 488 bis-(o-aminophenoxy)-N,N,N′,N′-tetraacetic acid 1 (BAPTA-1; OGB-1). A1 and B1: the vertical line across the bouton (filled with OGB-1) shows the position of a laser line scanned at a rate of one line every 4.06 ms by a confocal microscope. A2 and B2: representative traces of calcium fluorescence within single boutons in the wild type and the mutant. These are serial line scans accumulated from left to right over a period of 500 ms. The white solid line (arrowhead) indicates the time at which a single stimulus pulse was delivered to the segmental nerve. A3 and B3: representative plots of relative changes (ΔF/F) of fluorescence within single synaptic boutons after nerve stimulation. The look-up table, representing pixel values 0–255, is shown to the right. Image pixels in A and B have not been processed or smoothed. C: a histogram comparing the average change of fluorescence in synaptic boutons between the wild type and the mutant larvae (P = 0.71). D: a histogram comparing the average decay time constant (tau) of fluorescence in synaptic boutons between the wild type and the mutant larvae (P = 0.67). The number of terminals boutons sampled (separate muscles, from 3 separate wild type and 7 mutant larvae) is shown on each histogram. E–I: summary results of calcium rise and decay within single boutons after a brief train of nerve stimulation (20 Hz, 2 s). E1 and F1: vertical lines mark the position of laser line across single boutons. E2 and F2: representative traces of fluorescence change within single boutons in the wild type and the mutant. The white vertical line indicates the delivery of the 1st pulse in the train, whereas the horizontal bar indicates the duration of the 20-Hz stimulation. G: representative plots of relative changes of fluorescence within single boutons after a short train of stimuli at 20 Hz for 2 s. H: a histogram comparing the average change of fluorescence between boutons in the wild type and the mutant larvae (P = 0.71). I: a histogram comparing the average decay time constant (tau) of fluorescence between boutons in the wild type and the mutant larvae (P = 0.34). Scale bar: 5 μm in all panels. Variances are shown as SE of the mean. The number of terminals boutons sampled (separate muscles, from 3 separate wild-type and 12 mutant larvae) is shown on each histogram. [Ca2+] o = 0.5 mM. All statistical comparisons are unpaired 2-tailed t-test.

Number of morphologically docked SVs is moderately reduced in the lap mutant

One of the factors influencing transmitter release at steps downstream of Ca2+ entry could be the availability of SVs at the active zone (Murthy et al. 2001). Because Ca2+ entry is normal, the small EJC amplitude observed in the lap mutant suggests that the number of SVs in the docked pool is reduced. We have previously attributed the impairment in release to a 1/3 loss of the SV pool in the lap mutant (Zhang et al. 1998). However, we did not quantify docked vesicles. To this end, we investigated the distribution of SVs near the active zone with transmission electron microscopy. We counted vesicles on three serial sections in which the central one always had an electron-dense “T” bar, the putative active zone in Drosophila (see arrows in Fig. 5) (Atwood et al. 1993; Jia et al. 1993; Koenig et al. 1993). Consistent with our early results (Zhang et al. 1998), the total number of SVs, including those within 50 and 500 nm to the plasma membrane but excluding docked vesicles, is significantly reduced in the lap mutant (Fig. 5). We further focused on the morphologically docked vesicles, which were defined as the group of vesicles touching or within 20 nm of the densely stained presynaptic membrane underneath or near the T-bar. These vesicles could potentially be releasable during initial stimulation (Parsegian 1977). The wild-type synapse contained on average 6.5 ± 0.82 morphologically docked vesicles/active zone (n = 14 active zones), whereas this number was 4.2 ± 0.55 (n = 21) in the lap mutant, representing a 35% reduction (Fig. 5C).

FIG. 5.

The number of morphologically docked vesicles at the active zone is mildly reduced at the lap mutant NMJ. A and B: representative electron micrographs of cross-sections of larval NMJs from the wild type (CS, A) and the lap mutant (lap/Df, B). It is apparent that the total number of SVs is significantly reduced in the mutant compared with that in the wild type. “T”-shaped electron-dense materials (“T”-bars) in the wild type (CS) and the lap mutant (lap/Df) are putative active zones (→). C: average numbers of 3 populations of vesicles inside the nerve terminal of the wild type (CS) and the lap mutant (lap/Df). These 3 populations are morphologically “docked vesicles” that are within 20 nm or in contact with the plasma membrane at the active zone, vesicles within 50 nm of the presynaptic active zone membrane, and vesicles within 500 nm of the presynaptic active zone membrane, excluding docked vesicles, respectively. On average, the docked-vesicle pool is mildly but significant statistically (*P < 0.05). The remaining 2 vesicle populations are more dramatically reduced in the lap mutant (** P < 0.01; *** P < 0.001).

We next used the active zone marker nc82 monoclonal antibody to estimate the number of active zones in the wild-type and the mutant larvae (Wucherpfennig et al. 2003; Qin et al. 2005). Figure 6 shows nc82 immunoreactive puncta on muscle 4 in segment 3 in the wild type and the lap mutant. A total of eight muscles from four wild-type and seven muscles from five mutant larvae were examined. The lap mutant had a slightly higher average number (381.4 ± 29.4, n = 7) of nc82-positive puncta per muscle compared with the wild type (299.8 ± 27.4, n = 8). However, this difference was not statistically significant from each other (P = 0.06). This is consistent with our observation that the number of T-bars appeared similar between the mutant and the wild type from our limited sections (data not shown), although serial sections would be better for determining the number of active zones. These studies suggest that the lap mutation does not have a significant effect on the total number of active zones. However, the location of the active zone differed in the mutant (see discussion later).

FIG. 6.

The number of active zones remains unchanged in the lap mutant. A and B: Nc82 antibody-stained NMJs on muscle 4 in segment 3 from the wild type (CS, A) and the lap mutant (lap, B) are shown here. Insets: enlarged areas identified by →. C: histograms showing that the average number of nc82-positive puncta is slightly increased in the lap mutant compared with the wild type. However, this increase is not statistically significant (t-test, P = 0.06).

Paired-pulse induces short-term facilitation instead of synaptic fatigue in the lap mutant

Ultrastructural reconstruction of mammalian central synapses reveals that the number of morphological docked vesicles and the number of readily releasable pool (RRP) of SVs are almost identical (Schikorski and Stevens 2001). The rather mild reduction in the number of morphologically docked vesicles observed in the lap mutant predicts that the RRP at the active zone may not be significantly reduced. We reasoned that if the RRP were limited, a brief and repetitive stimulation would lead to a rapid synaptic fatigue in the lap mutant, as has been shown in endo (Verstreken et al. 2002), synj (Verstreken et al. 2003), and dap160 mutants (Marie et al. 2004; Koh et al. 2004). To test this hypothesis, we stimulated the synapse with paired pulses delivered at 20 and 50 ms apart at different extracellular [Ca]. As with most synapses, the Drosophila NMJ also modifies its synaptic strength following the activity pattern of the motoneuron. Short-term plasticity can be determined experimentally by monitoring the relative change of EJC amplitude in response to two successive pulses of stimuli. Quantitatively, we define this change as the facilitation index (FI), which equals (EJC2-EJC1)/EJC1 X100%. A positive plasticity index represents facilitation, which represents an increased release of transmitter aided by residual Ca2+ following the initial action potential (Xu-Friedman and Regehr 2004; Zucker 1999). In contrast, a negative plasticity index stands for depression, which is often caused by the depletion of SVs by the first action potential. Short-term plasticity depends on both the extracellular calcium concentration and the interval of the two stimuli. In general, the synapse facilitates at low [Ca] and depresses at high [Ca]. Furthermore, synaptic facilitation or depression increases when the interval shortens.

The result of short-term plasticity in the wild type and the lap mutant is summarized in Fig. 7. At 0.8 mM [Ca], both wild-type and the mutant larvae showed mostly facilitation. However, the degree of facilitation differed between these two animals. The mutant synapse facilitated by 45.3 ± 14.8% (n = 10) and 48.7 ± 7.3% (n = 10) at the 50- and 20-ms stimulus intervals, respectively. The wild-type synapse displayed a relatively weak facilitation by 11.1 ± 2.0% (n = 9) and 5.1 ± 5.8% (n = 9) at the same stimulus intervals. These differences are mild but significantly different statistically (P < 0.001 for 20-ms interval, P < 0.05 for 50-ms interval). At this low [Ca], synaptic depletion does not occur in both the wild type and the lap mutant primarily because the release probability is low.

FIG. 7.

Synaptic facilitation instead of depression is evoked in the lap mutant by paired pulses. Paired pulses delivered at 50- and 20-ms intervals are used to induce short-term synaptic plasticity (facilitation or depression) in the lap mutant (lap/Df) and wild type (CS) larvae at 0.8 and 5 mM [Ca]. The change in transmitter release in response to the 2 pulses is quantified by calculating the facilitation index (FI) using the equation (EJC2-EJC1)/EJC1 X100%. A: representative evoked EJCs by paired pulse at 20-ms interval and 0.8 mM [Ca]. Facilitation is observed in both animals under these experimental conditions. A1: plots of average FIs induced by paired pulse at 50- and 20-ms intervals and 0.8 mM [Ca]. Facilitation is more profound in the lap mutant (lap/Df) than in the wild-type larvae (CS, *P < 0.05; ***P < 0.001). B: representative evoked EJCs by paired pulse at 20-ms interval and 5 mM [Ca]. At this [Ca], the wild type shows dramatic depression whereas the lap mutant displays facilitation. B1: plots of average FIs induced by paired-pulse at 50- and 20-ms intervals and 5 mM [Ca]. The FIs between the mutant (lap/Df) and the wild-type (CS) larvae are significantly different (***P < 0.001). The lack of depression by paired pulses at these brief intervals suggest that releasable pool of vesicles is unlikely the major and immediately limiting factor for the dramatic reduction in quantal content observed in the lap mutant.

To increase the release probability in both the wild type and the mutant, we raised extracellular [Ca] to 5 mM. At this [Ca], the wild-type synapses showed a profound depression with FIs ranging from −25.5 ± 1.4% (n = 8) at 50-ms interval to −31.8 ± 2.5% (n = 9) at 20-ms interval. In contrast, the mutant synapse did not show depression. Instead, it displayed 2.7 ± 3.3% (n = 17) and 3.9 ± 3.6% (n = 12) facilitation at 20- and 50-ms paired-pulse intervals, respectively. The differences in FIs are statistically significant (P < 0.001). These results show whether it is at low or high [Ca], the mutant synapse shows persistent facilitation instead of depression when paired pulses are delivered. At these brief intervals (20–50 ms), vesicle recycling or recruitment is not expected to play a role to replenish the RRP. Hence, these results suggest that the limitation of the releasable vesicle pool is unlikely the major and direct factor contributing to the reduction of release probability in the lap mutant. This provides further support to the possibility that the coupling between calcium and secretion is disrupted.

SV proteins are abnormally distributed to extrasynaptic axonal regions

While the mild reduction in docked vesicles may contribute to the reduced quantal content, it is insufficient to account for the change in Ca2+ sensitivity or cooperativity. What could explain these exocytotic defects in the lap mutant downstream of vesicle docking and Ca2+ entry? During vesicle recycling, clathrin-mediated endocytosis accomplishes two major tasks: to retrieve vesicular components (i.e., lipids and proteins) and to assemble new vesicles. Even though these two events are tightly coupled under physiological conditions (Sankaranarayanan and Ryan 2000), a partial failure in recycling SV proteins may occur when clathrin-mediated endocytosis is impaired. This possibility was first suggested by studies of the AP180 mutant (unc-11) in C. elegans, in which synaptobrevin was mislocalized to axonal membranes (Nonet et al. 1999). Should a loss of synaptobrevin and/or other SV proteins occur in the lap mutant, transmitter release could also be impaired. Additionally, perturbation of clathrin-dependent vesicle trafficking may also affect either the localization of synaptic proteins and/or the development of the synapse. This is suggested by the altered distribution of nc82-positive puncta shown in Fig. 6.

To test these hypotheses, we examined the distribution of SV proteins at the NMJ and the morphology of synaptic boutons using immunocytochemistry. The Drosophila NMJ is a well-characterized system for synaptic studies due in part to its large synapse size and stereotyped projection of synaptic boutons on identifiable muscle surfaces (Atwood et al. 1993; Hoang and Chiba 2001; Jia et al. 1993; Keshishian et al. 1993). Here, we used antibodies to three SV proteins (synaptotagmin I, n-Syb, and CSP) to examine the morphology of the mutant and wild-type NMJ with a particular focus on the subcellular localization of these SV proteins in extrasynaptic regions between synaptic boutons. As shown in Fig. 8, synaptotagmin I, n-Syb, and CSP were generally well localized within synaptic boutons at the NMJ in the wild type (n = 10 larvae). Importantly, these immunoreactive boutons were discrete and separated from each other giving a beads-on-a-string appearance. This typical pattern of SV protein distribution is well established for the fly NMJ (e.g., Estes et al. 1996), suggesting that SV proteins are usually absent from extrasynaptic regions that connect two neighboring boutons (Fig. 8, A and B; also see Figs. 9A and 10A). These SV proteins were also found within synaptic boutons in the lap mutant (Fig. 8, C and D; n = 20). However, the morphology of most synapses was altered to a rope-like appearance instead their normal shape. Importantly, these SV proteins were also detected in contiguous and tubular structures that fill the extrasynaptic regions between individual synaptic boutons (compare a, a1, and b1 with c, c1, and d1). These results show that the synaptic morphology is altered to ropes and that these SV proteins or SVs are mislocalized to extrasynaptic axonal regions in the lap mutant.

FIG. 8.

Synaptic vesicle proteins are mislocalized at 3rd instar larval NMJs. Fluorescent images of immunoreactive NMJs on muscle 4 show the localization of 3 SV proteins (Syt, synaptotagmin I; CSP, cysteine string protein; and n-Syb, neuronal synaptobrevin) in synaptic boutons and presynaptic membranes. The muscles on which these synapses are formed are not labeled and thus invisible. A–a1: Syt (green) and CSP (red) are colocalized to synaptic boutons in the wild type (CS) forming a “beads-on-a-string” appearance. Note that these synaptic vesicle proteins are restricted to the bouton area and absent in the extrasynaptic axonal region between boutons (see → in the enlarged images shown in a and a1). B–b1: Syt (green) and n-Syb (red) are also colocalized to synaptic boutons in the wild type (CS) forming the beads-on-a-string appearance. Note that these synaptic vesicle proteins are restricted to the bouton area and absent in the extrasynaptic region between boutons (see enlarged images in b and b1, →). C–c1: Syt (green) and CSP (red) are colocalized to synaptic boutons in the lap mutant (lap/Df). However, they also appear in the axonal region between individual boutons forming a “rope”-like morphology. This mislocalization of vesicle proteins is more apparent in the enlarged images (c and c1, →). D–d1: Syt (green) and n-Syb (red) are colocalized to synaptic boutons in the lap mutant (lap/Df). However, they also appear in the axonal region between individual boutons (see the enlarged images in d and d1, →).

FIG. 9.

Co-localization of cysteine-string protein (CSP) and neuronal synaptobrevin (n-syb) with HRP in the extrasynaptic region. A: localization of CSP (red) and horseradish peroxidase (HRP, green) at the wild type (CS) NMJ. While CSP is localized to synaptic boutons, HRP marks the axonal membrane as well as the bouton membrane. Both CSP and HRP are found in synaptic boutons. Note that CSP is absent from the extrasynaptic axonal region between synaptic boutons (see →). B: localization of CSP (red) and HRP (green) at the lap mutant NMJ. Note that CSP is now mislocalized to the extrasynaptic region between synaptic boutons marked by HRP (see →). C: localization of n-Syb (red) and HRP (green) at the wild-type (CS) NMJ. Similar to CSP, n-Syb is also localized to synaptic boutons. Note that n-Syb is absent from the extrasynaptic region between synaptic boutons (see →). D: localization of n-Syb (red) and HRP (green) at the lap mutant NMJ. Unlike in the wild type, n-Syb is found within synaptic boutons and in the extrasynaptic region between synaptic boutons (see →).

FIG. 10.

Localization of Dap160 and glutamate receptors in the lap mutant. A and B: localization of Dap160 (a presynaptic marker, green) and cysteine-string protein (CSP, red) at the wild type (CS) and the lap mutant NMJ. As with other presynaptic components, both Dap160 and CSP are localized to synaptic boutons in the wild type and in the mutant. However, both Dap160 and CSP are absent in the wild type (see A, →) but found in the lap mutant (see B →), in intersynaptic bouton regions. C and D: localization of glutamate receptor III (a postsynaptic marker, green) and CSP (red) at the wild-type (CS) and the lap mutant NMJ. Glutamate receptor III is primarily found in postsynaptic density closely opposing the presynaptic terminal identified with CSP (C). Small amount of the receptor is also found along the intersynaptic bouton regions (see C2, →). Similar patterns of glutamate receptor III distribution are found in the lap mutant. However, more receptors are found in the intersynaptic bouton regions to match the shape of the rope-like synapse in the mutant (D2, →).

To confirm that these SV proteins are indeed mislocalized to axonal regions, we used an antibody to the neuronal membrane protein Nervana/HRP (Jan and Jan 1982; Sun and Salvaterra 1995) to mark the axonal membrane. In doubly stained NMJs with antibodies to HRP and CSP or n-Syb, we confirmed that CSP and n-Syb were localized primarily to synaptic boutons in the wild type (n = 8). As expected, the HRP antibody labeling is found in the extrasynaptic areas between boutons that are usually devoid of SV proteins (Fig. 9, A and C, see →). A small number of NMJs in the wild type occasionally show a “spillover” of SVs proteins from synaptic boutons to the axonal area. The localization of SV proteins to synaptic boutons remains in the wild type even after extensive stimulation of the motor axon with high potassium-containing saline (not shown), suggesting that SVs proteins are efficiently internalized from the plasma membrane.

In the lap mutant, these SV proteins were also found within synaptic boutons (n = 12). However, the lap mutant differed from the wild type in that the extrasynaptic regions were expanded in 60% of the synaptic branches. Furthermore, the SV proteins were found in the extrasynaptic axonal regions between some boutons regardless whether the morphology of synaptic boutons was altered or not (Fig. 9, B and D, see →). Synaptotagmin I displayed a similar mislocalization pattern to those found for CSP and n-Syb in the lap mutant (n = 10, not shown). In rare cases, the motor axon bundle innervating muscles 12 and 13 was also stained with a polyclonal antibody to n-Syb in the lap mutant (data not shown). The mislocalization of these SV proteins occurred in nearly all synapses examined. Taken together, these results are consistent with the possibility that either these SV proteins or intact SVs are mislocalized to the extrasynaptic regions.

Other synaptic components are also mislocalized to extrasynaptic regions

The alterations in SV protein distribution and synapse morphology are similar to the finding that clathrin heavy chain and dynamin are diffusely distributed at the NMJ (Zhang et al. 1998). This raises the possibility that lap mutations may also affect the distribution of other synaptic proteins due to its effects on either SV recycling or general vesicle trafficking. We therefore examined the subcellular localization of the endocytotic protein Dap160 (Koh et al. 2004; Marie et al. 2004; Roos and Kelly 1998, 1999) in synaptic terminals and type III glutamate receptors (GlutRIII) (Marrus et al. 2004) in the postsynaptic density using specific antibodies (Fig. 10). Dap160 was restricted within synaptic boutons in the wild type, but it was redistributed into regions between synaptic boutons in the mutant (Fig. 10, A and B; n = 8). Although we did not examine all the presynaptic components, our results suggest that presynaptic proteins as well as SV proteins are both present in extrasynaptic regions. As one would expect, glutamate receptors were highly enriched in the muscle membrane opposing synaptic boutons (Marrus et al. 2004; Petersen et al. 1997) (Fig. 10C). Unlike most presynaptic and SV proteins, however, a small number of the receptor is also found in extrasynaptic regions between synaptic boutons in the wild type (Fig. 10C, →; n = 8) (Marrus et al. 2004). Nonetheless, the usual beads-on-a-string morphology of synapses remains in the wild type. In the lap mutant, no apparent change in GlutRIII and DAP160 immunoreactivity was detected (n = 16). As in the wild-type larvae, GlutRIII also remained in the postsynaptic density opposing synaptic boutons. However, they were redistributed to adapt to the rope-like pattern of NMJs in the mutant, indicating that the receptor was also mislocalized at the NMJ. The redistribution of receptors is consistent with the altered distribution of active zones in the lap mutant (Fig. 6).

DISCUSSION

In the present study, we have examined the role of the endocytotic protein AP180 (LAP) in synaptic transmission and synaptic protein localization at the Drosophila NMJ. Our results are consistent with a primary role for AP180 in vesicle recycling revealed in our previous work. In addition, we have uncovered several features of exocytotic defects in the lap mutant that are more perplexing and cannot be explained simply by the reduction in the SV pool alone. Our studies show that the exocytotic defects occur at steps downstream of calcium influx. Contrary to previous assumptions, the reduction in vesicle docking (34%) is rather mild compared with the 96% reduction in quantal content. The exocytotic defect cannot be explained entirely by the change in vesicle docking, as calcium sensitivity and cooperativity and failure rate are also altered in the lap mutant. We have further shown that n-Syb, CSP, and synaptotagmin I, three vesicle proteins critical for calcium-evoked release, are mislocalized to the extrasynaptic axonal regions in the lap mutant. The endocytotic protein DAP160, the active zone maker nc82, and glutamate receptors are also found in the extrasynaptic regions at the mutant NMJ. These results suggest that both vesicle recycling and the subcellular distribution of pre- and postsynaptic proteins are impaired. Hence, possibilities remain that LAP is involved in not only vesicle recycling but also synaptic development.

AP180 is required for calcium-evoked transmitter release at steps downstream of calcium influx and vesicle docking

Our studies have revealed three lines of evidence supporting the hypothesis that LAP/AP180 regulates the efficacy of Ca2+ coupling to transmitter release at the Drosophila NMJ. First lap mutations significantly reduce Ca2+-evoked release but leave spontaneous fusion relatively intact. In fact, the rate for spontaneously release, indicated by the frequency of minis, is actually increased, perhaps due to developmental homeostasis that compensates for the loss of synaptic release evoked by action potentials (Davis and Goodman 1998; Turrigiano and Nelson 2004). Second, Ca2+ sensitivity is reduced in the lap mutant, evidenced by the high failure rate and inability by high [Ca] to rescue transmitter release in the mutant. Third, we show a reduction in Ca2+ cooperativity in the lap mutant, providing further support that Ca2+ is less efficient in triggering exocytosis. We have definitely shown that these exocytotic defects occur at steps downstream of calcium influx in the nerve terminal. A key question remaining is whether these physiological defects can be explained simply by the increase in vesicle size and/or by the reduction in vesicle numbers, two prominent defects revealed in our initial characterization of the lap mutant (Zhang et al. 1998). Quantal size and vesicle size increase concurrently (Karunanithi et al. 2002; Zhang et al. 1998), consistent with a normal density of postsynaptic receptors. Furthermore, spontaneous release appears normal and even occurs at a higher rate. These results indicate that the defect in exocytosis is primarily presynaptic and that the change in vesicle size has minimal effects on vesicle fusion per se. The change in vesicle size, however, may influence its response to calcium should the vesicle protein or lipid compositions be altered.

One might argue, as we previously assumed (Zhang et al. 1998), that these synaptic physiological defects are simply caused by a reduction in the total number of SVs and docked vesicles. However, insights from our current results and from recent genetic studies of other endocytotic mutants now force us to take a fresh look at the relationship between SV pools and synaptic transmission. First, the reduction in calcium sensitivity and cooperativity cannot be easily explained by changes in the number of docked vesicles at the active zone. Second, the number of docked vesicles is only mildly reduced compared with the reduction in quantal content. At the morphological level, each larval muscle is estimated to have on average 50 synaptic boutons and a total of >500 active zones (Atwood et al. 1993; Jia et al. 1993). Assuming some of these active zones are “silent” due to the lack of docked vesicles (Atwood and Wojtowicz 1999), 400–450 active zones are estimated to be functional. At 1 mM [Ca], the typical quantal content is estimated to be 110 (see Fig. 3), suggesting that each active zone releases transmitter from 0.24 to 0.28 SVs. In the lap mutant, the number of active zones does not appear to change based on our limited random EM sections. The average number of docked vesicles (4.2 SVs) per active zone in the mutant should be sufficient to support a normal level of basal exocytosis. Our estimated number of active zones using the nc82 antibody was ∼300 in the wild type and 381 in the mutant. They are slightly below the estimated numbers of active zone by electron microscopy. One likely reason for this could be the difficulty counting and resolving all nc82-positive puncta at light microscopic levels. Nonetheless, our results suggest that there is no statistical difference in the number of active zones per muscle fiber between the mutant and the wild type. Unless some of the active zones in the extrasynaptic region are “inactive,” this finding further supports the notion that the exocytotic defect occurs downstream of vesicle docking. Third, paired-pulse stimulation induces facilitation rather than depression in the lap mutant, suggesting that the RRP of SVs is not a direct and limiting factor for the dramatic reduction in basal release. In comparison, other endocytotic mutations, such as endo (Verstreken et al. 2002), synj (Verstreken et al. 2003), and dap160 (Koh et al. 2004; Marie et al. 2004), show an immediate synaptic depression on repetitive stimulation. Fourth, the C. elegans AP180 mutant unc-11 (Nonet et al. 1999) and the Drosophila stn mutant (Fergestad et al. 1999; Stimson et al. 1998, 2001) are also defective in synaptic transmission even though the number of SVs remains normal in the terminal. In fact, the number of docked SVs is increased in the unc-11 mutant (Nonet et al. 1999). Finally, other endocytotic mutants in Drosophila, such as endo (Verstreken et al. 2002) and synj (Verstreken et al. 2003), do not alter basal transmitter release even though they are all nearly depleted of SVs, displaying far fewer SVs than the lap mutant does. The dap160 mutant also has reduced numbers of SVs similar to that observed in the lap mutant. It reduces quantal content but keeps the amplitude of EJPs normal (Koh et al. 2004; Marie et al. 2004). Based on these observations, we favor the possibility that the moderate change in docked vesicles unlikely accounts substantially for the reduction in quantal release in the lap mutant.

The precise cause for the mild reduction in docked vesicle pool is unknown. One possibility is the high frequency of spontaneous fusion of SVs may deplete the size of docked vesicle pool. Other factors, such as the size of SVs and active zones could influence the quantification of morphologically docked vesicles. We do not have data to quantitatively compare the size of active zones between the wild type and the lap mutant. However, it is known that the average size of SVs is increased in the lap mutant (Karunanithi et al. 2002; Zhang et al. 1998). This may lead to fewer vesicles being able to fit around the T-bar at the active zone. A third possibility is that this reduction in the number of docked vesicles could be related to possible loss of synaptotagmin I on vesicles. Reist and colleagues showed previously that vesicle docking is reduced in synaptotagmin I mutants (Reist et al. 1998).

AP180 indirectly maintains the efficacy of calcium-evoked transmitter release

Although AP180 might directly participate in calcium coupling to exocytosis, we deem it very unlikely because of its biochemical properties as a clathrin-assembly protein and the endocytotic phenotypes revealed in the lap and unc-11 mutant. Therefore we favor the idea that AP180 indirectly regulates synaptic transmission through endocytosis. At present, the change in the distribution of both synaptic and vesicle proteins at the NMJ stands out as the most striking morphological phenotype in the mutant. The questions are then how this subcellular change in protein distribution might come about and whether it contributes to the physiological defects in the mutant. A simple and straightforward interpretation of the aberrant protein localization at the NMJ is that it results naturally from SV depletion in endocytotic mutants. Although an easily acceptable and logical assumption, it may not be supported by current experimental data. In endo and synj mutants, a severe depletion of SV occurs such that the reserve pool is abolished and the remaining SVs are found near the active zone. Consequently, SV proteins are only found in the peripheral areas within the synaptic bouton (Verstreken et al. 2002, 2003). However, SV proteins remain inside synaptic boutons and are absent from the extrasynaptic axonal regions in these mutant animals. An absence of synaptotagmin I and CSP in extrasynaptic regions in endo and synj mutants has been recently confirmed by others (D. Dickman and T. Schwarz, personal communications). Similarly, a partial depletion of SVs in dap160 mutants does not result in an accumulation of SV proteins in extrasynaptic regions (Koh et al. 2004; Marie et al. 2004). In contrast, SV proteins are mislocalized in unc-11 and stn mutant animals despite the fact that the SV pool remains normal in these animals. Our work confirms the observation of synaptobrevin mislocalization first made in the unc-11 mutant (Nonet et al. 1999) and further shows that CSP and synaptotagmin I are also mislocalized to extrasynaptic regions in the lap mutant. We note that the phenotype of SV protein mislocalization in the lap mutant is reminiscent of the mislocalization of CSP and synaptotagmin I in the stn mutant (Fergestad and Broadie 2001; Fergestad et al. 1999; Stimson et al. 1998, 2001). However, the axonal regions between synaptic boutons are expanded in the lap mutant but not in the stn mutant. Furthermore, the lap phenotype differs from the complete translocation of vesicle proteins to the plasma membrane observed in shits1 flies after the depletion of SVs at the restrictive temperature (Estes et al. 1996; van de Goor et al. 1995). In shits1 flies, vesicle proteins are depleted inside the synaptic bouton and reside exclusively in the plasma membrane. In the lap mutant, the vesicle protein is more diffusely and evenly distributed within synaptic boutons as well as the extrasynaptic membrane. We realize that a true co-localization of vesicle proteins and the plasma membrane will be difficult to resolve under light microscopic levels. Nonetheless, the close overlap with the membrane marker HRP suggests that some of the vesicle proteins are present in the plasma membrane. Hence, a reasonable conclusion is that mislocalization of SV proteins in nerve terminals in the lap mutant may not be a general consequence of defective vesicle recycling and depletion of SV pools.

The change in SV protein localization in the lap mutant could be explained by at least two possibilities. One possibility is that these changes result from a general defect in synaptic development. This is supported by the observation that synapse morphology is dramatically altered in the mutant to rope-like boutons. With this change in synapse shape, not only are SV proteins but also other presynaptic proteins, including putative active zones marked by nc82 antibody, as well as postsynaptic receptors redistributed along the altered NMJ in the mutant. This developmental reorganization of synaptic morphology may be a result of reduced endocytosis of surface molecules such as receptors or other signaling molecules. It is noted that synaptic morphology is also altered in other endocytotic mutants, such as endo (Verstreken et al. 2002) and dap160 (Koh et al. 2004; Marie et al. 2004). However, these mutants have different effects on synaptic morphology. For example, endo mutations enlarge synaptic boutons and do not display the rope-like synapse seen in the lap mutant (Verstreken et al. 2002). This suggests that AP180 may have additional roles in synapse development.

Alternatively, the change in synaptic morphology as well as protein localization may be closely linked to the role of AP180 in SV recycling or vesicle trafficking. Consistent with studies of the unc-11 and stn mutants, our results may suggest that SV protein retrieval is partially failed during endocytosis in the lap mutant resulting in the mislocalization of SV proteins in the extrasynaptic region. The “lost” SV components would then diffuse into the extrasynaptic membranes causing them to expand the surface area such that the usual “strings” between synaptic boutons become ropes. To compensate for this change as well as the reduction in synaptic transmission, the synapse may expand both the pre- and postsynaptic regions by translocating pre- and postsynaptic proteins into extrasynaptic regions.

Although both possibilities may co-exist and account for the phenotypes observed in the lap mutant, we favor the second working model because it better explains the electrophysiological defect in the lap mutant. A failure to fully recycle synaptotagmin I, CSP, and n-Syb could have significantly negative consequences on the amount of these proteins on SVs thereby affecting calcium-evoked release. Both synaptotagmin I (Broadie et al. 1994; Fernandez-Chacon et al. 2001; Geppert et al. 1994; Mackler et al. 2002; Sudhof 2004) and CSP (Dawson-Scully et al. 2000) have been shown to play important roles in coupling Ca2+ with release. Interestingly, the lap mutant phenocopies the increase in mini frequency and reduction in evoked release observed in some alleles of the syt I mutant (Broadie et al. 1994; DiAntonio and Schwarz 1994; Littleton et al. 1994; Reist et al. 1998). Although synaptotagmin I may not play a role in calcium cooperativity (Broadie et al. 1994; Fernandez-Chacon et al. 2001; but see Littleton et al. 1994), soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) proteins (Bevan and Wendon 1984; Cull-Candy et al. 1976; Stewart et al. 2000) have been shown to contribute to Ca2+ cooperativity. According to the model proposed by Stewart and colleagues (2000), Ca2+ cooperativity reflects the cooperative formation of ‘tight’ trans-SNARE complexes from “loose” trans-SNARE complexes (Chen et al. 1999). SNARE complexes have been suggested to directly respond to Ca2+ during exocytosis (Sorensen et al. 2002). We have observed that the change in Ca2+ cooperativity in the lap mutant is similar to those reported in n-syb hypomorphic alleles (Stewart et al. 2000). Together with the observation that n-Syb is mislocalized in the lap mutant, our electrophysiological data implies that the reduction in Ca2+ cooperativity might result from a partial loss of n-Syb on SVs.

Potential mechanisms and implications of AP180 action in synapses

A role for AP180 in protein retrieval would be in good agreement with both the biochemical properties of AP180 and its strategic location within coated vesicles. AP180 might help retrieve synaptotagmin I during endocytosis through its interaction with the clathrin-AP2-synaptotagmin complex (De Camilli and Takei 1996; Zhang et al. 1999). Our observation that synaptotagmin I is mislocalized in the lap mutant may support this notion. However, we are puzzled that synaptotagmin I is localized normally in the unc-11 mutant (Nonet et al. 1999). Such a difference could reflect differences in synapse types (GABAergic and cholinergic in C. elegans vs. glutamatergic in Drosophila). It is also known that LAP differs from UNC-11 dramatically in the size and sequence at the C-terminus (Nonet et al. 1999; Zhang et al. 1998), although the sequence responsible for this different function is not apparently clear. In comparison with synaptotagmin I, the molecular mechanism by which CSP and n-Syb are recycled remains elusive. It remains to be determined whether AP180 plays a role (direct or indirect) in facilitating the recycling of CSP and n-Syb. A likely possibility is that CSP and n-Syb may be recruited into SVs through their interaction with other vesicle proteins. For example, CSP may be indirectly recycled into SVs by its direct interaction with synaptotagmin I (Evans and Morgan 2002). Further studies are needed to reach a better understanding of CSP and n-Syb recruitment during endocytosis.

It should be noted that SV proteins traditionally thought to be exocytotic are increasingly intertwined with vesicle recycling. Specific AP2-synaptotagmin I interactions have been demonstrated experimentally important for synaptotagmin I recycling (Jarrousse et al. 2003). Similarly, deletion or perturbation of synaptotagmin I (Fukuda et al. 1995; Jorgensen et al. 1995; Littleton et al. 2001; Poskanzer et al. 2003; but see Reist et al. 1998) and synaptobrevin (Deak et al. 2004) leads to partial depletion of SVs. These observations raise the possibility that defects in protein retrieval may further exacerbate the impairment of endocytosis as well as exocytosis through a negative cycle. It is interesting to note that AP180, synaptobrevin, and synaptotagmin, but not all synaptic proteins, are significantly reduced prior to the loss of synapses in the neocortex of early and mild Alzheimer's patients (Masliah et al. 2001; Sze et al. 2000; Yao 2004; Yao and Coleman 1998; Yao et al. 2003). It remains to be determined whether similar synaptic dysfunctions are also found in these patients. Our studies have advanced the understanding of AP180 in clathrin-mediated endocytosis and its likely indirect roles in exocytosis by retrieving vesicle proteins. Our studies also raise an important question on whether SV composition is indeed altered. A direct demonstration for a change in vesicle proteins would be a definitive proof. However, it has been hampered by the difficulty obtaining sufficient amount of adult brains from the lap mutant, which dies mostly as larvae. A conditional mutation, such as a temperature-sensitive paralytic allele of the lap mutant, would be valuable for such biochemical studies. RNAi-mediated knockdown techniques could also be used to reduce LAP levels in viable adult flies. These flies would be valuable for studying not only the precise mode for AP180 in SV protein retrieval but also the potential role for LAP in synaptic development.

GRANTS

This work was supported by a CARRER Award (IBN-0093170) from the National Science Foundation and a startup fund from UT-Austin to B. Zhang, by an Undergraduate Research Award at UT-Austin to R. W. Daniels, and by grants from the Canadian Institutes of Health Research to H. L. Atwood, M. P. Charlton, and G. T. Macleod.

Acknowledgments

We are grateful to W. Thompson for use of the fluorescence microscope and for constructive comments on the manuscript, to L. Marin (University of Toronto) for conducting the electron microscopic work, to A. Bardo and J. Mendenhall at UT's Imaging Center for assistance with confocal imaging. We thank A. DiAntonio (Glutamate receptor III), H. Bellen (Synaptobrevin, DAP160), E. Buchner (Nc82 mAb), R. Kelly (DAP160), N. Reist (Synaptotagmin I), J. Roos (DAP160), and K. Zinsmaier (CSP) for the gift of antibodies and L. Pallanck and M. Babcock for the SD3 allele of lap. We also thank D. Dickman and T. Schwarz for personal communication of their unpublished work on SV protein localization in endo and synj mutants. We thank two anonymous reviewers for constructive comments.

Present addresses: R. W. Daniels, Dept. of Molecular Biology and Pharmacology, Washington University School of Medicine, St. Louis, MO 63110; and G. T. Macleod, Div. of Neurobiology, University of Arizona, Tucson, AZ 85721.

Footnotes

  • * R. W. Daniels and G. T. Macleod contributed equally to this work.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

REFERENCES

View Abstract