We identified the transposon insertion mutant P{SUPor-P}CG2095^KG02723 in simultaneous unbiased forward screens for Drosophila mutants with alterations in NMJ presynaptic growth and glutamate receptor cluster formation 56. We selected this mutant for further study on the basis of a dramatic semi-penetrant presynaptic overgrowth phenotype and reproducible loss of glutamate receptor immunoreactivity (see results, below). The KG02723 P-element insertion was mapped by the Berkeley Drosophila Genome Project using inverse PCR to the first exon of predicted gene CG2095 17. We confirmed the KG02723 P-element insertion site in CG2095 by site-specific PCR primers.

We cloned and sequenced full-length CG2095 cDNA from the wild-type Drosophila strain Oregon R (Accession number AY905551). A BLAST search using this Oregon R CG2095 amino acid sequence showed 33% identity to rat and human Sec8 (Fig. 1A). Conversely, BLAST searches using mammalian Sec8 against the translated Drosophila genome showed close matches only to CG2095. Thus, CG2095 represents the sole Drosophila sec8 homolog. Hereafter, we refer to CG2095 as sec8.

Because the P{SUPor-P}CG2095^KG02723 P element insertion mutant represents a mutation in the Drosophila sec8 gene, we refer to this allele as 'sec8P1'. We verified that the P-element insertion was responsible for the sec8P1 phenotype (see below) by complementation analysis and remobilization of the P-element 18 to create precise excisions ('revertants'). These revertants are phenotypically indistinguishable from other control (w¹¹¹⁸) animals. Furthermore, sec8P1 mutants retain their phenotype in trans to a deficiency, Df(3R)Tpl10, that completely eliminates the CG2095 gene. These results show that the sec8P1 phenotype is specifically due to the P-element insertion in sec8.

Many transposon insertion mutants, including the sec8P1 mutants, represent hypomorphic alleles. Therefore, we generated deletions of the sec8 gene via imprecise excision of the KG02723 P-element in sec8P1, using standard methods 18. Among the excision mutants isolated was a CG2095-specific deletion that we refer to as 'sec8Δ1'. To confirm that sec8Δ1 only disrupts the sec8 gene and provide an accurate molecular description of this mutation, we sequenced the sec8 genomic region in sec8Δ1 mutants. Sequencing showed that the sec8Δ1 deletion begins before the ATG, 37 bp into the first 5' exon, and ends 73 bp prior to the 3' end of the last exon. Neighboring genes are completely intact. sec8Δ1 therefore represents a zygotic null. sec8Δ1 homozygous mutants hatch at the same time as controls (approximately 24 h after egg laying, AEL), but do not undergo their first molt and die as first instar larvae (~ 46 h AEL). The molecular nature of the sec8P1 and sec8Δ1 mutations is summarized in Fig. 1B.

Homozygous sec8P1 and sec8Δ1 mutants both hatched at approximately normal time (22–24 h AEL) and appeared grossly normal by light microscopy (appropriate size, segmentation, mouthooks, etc.). There was no obvious difference in mutant animal size, behavior or viability immediately after hatching. However, homozygous sec8Δ1 mutants did not undergo their first molt and did not live past first instar larval stage (~ 48 h AEL). Homozygous sec8P1 mutants appeared to molt and pupate normally, and therefore lived through larval development, but did not eclose and therefore died as pupae. In view of the viability and the fact that sec8 is almost completely deleted in sec8Δ1 mutants, we conclude that sec8P1 represents a hypomorphic allele, while sec8Δ1 represents a zygotic null allele.

The Drosophila embryonic/larval neuromuscular junction (NMJ) is a well-established model glutamatergic synapse that can be used to examine the effects of Sec8 loss on synaptic development and function. To test whether Sec8 is expressed at the NMJ, we raised rabbit polyclonal antibodies against Sec8 using a synthetic peptide composed of Sec8 amino acids 440–460, and affinity purified these antibodies using the original peptide epitope. Immunoreactivity of this antibody was widely distributed throughout many tissues and was particularly enriched in synapses, including the NMJ (Fig. 2A). Because the pre and postsynaptic membranes are separated by only approximately 15 nm 1920, it is impossible to distinguish definitively by light microscopy whether Sec8 immunoreactivity directly at the NMJ is pre or postsynaptic. Nevertheless, several indications suggest that Sec8 is both pre and postsynaptic. First, there is clearly Sec8 immunoreactivity throughout the muscle cell outside the area delimited by presynaptic HRP staining; therefore, sec8 must be expressed in the postsynaptic muscle cell. Postsynaptic localization of Sec8 is also suggested by the fact that Sec8 immunoreactivity overlaps immunoreactivity for the postsynaptic glutamate receptor subunit GluRIIA. Three-dimensional reconstructions and rotations (not shown) show Sec8 immunoreactivity extending above the interior face of the muscle, consistent with presynaptic localization. We did not observe any Sec8 immunoreactivity in segmental nerve axons, suggesting that presynaptic Sec8, if present, must be localized specifically at the terminals, but there is little or no immunoreactivity in the centers of presynaptic boutons. Thus, presynaptic Sec8 must be localized specifically near the terminal membrane. We obtained essentially identical results using a recently described and independently generated antibody 21 (data not shown).

Sec8 antibody immunoreactivity in the embryonic/first instar (L1) neuromusculature was dramatically reduced in homozygous sec8P1 mutants, and was almost undetectable in sec8Δ1 mutants (Fig. 2B). On immunoblots using whole-animal extracts, our Sec8 antibody recognized a single 111 kDa band (not shown). The intensity of this band was reduced, on average, to approximately one-half normal in both sec8P1 and sec8Δ1 L1 mutants (Fig. 2C). This suggests that despite the deletion of the sec8 gene in sec8Δ1 mutants, and almost complete elimination of Sec8 in the NMJ of homozygous mutants, there is substantial maternal contribution of Sec8 in other tissues, as previously shown for other sec proteins 78212223.

We conclude from our immunohistochemical results that Sec8 is probably expressed both pre and postsynaptically at the Drosophila NMJ, and that sec8 mutant NMJs have almost no remaining Sec8 but substantial maternal Sec8 persists in other parts of the sec8Δ1 mutant animals.

Sec proteins have been implicated in the membrane addition required for neurite outgrowth 78910. Formation of Drosophila larval NMJs involves extensive neurite outgrowth and presynaptic arborization. Examination of motor nerve terminals in the ventral body wall neuromusculature of Drosophila sec8 mutants therefore provides an excellent opportunity to test whether Sec8 plays a role in nerve terminal growth in vivo. Approximately halfway through Drosophila embryonic development (12 h After Egg Laying, AEL), motor neuron growth cones exit the CNS and travel along the developing segmental nerve toward their body wall muscle targets. Approximately 13 h AEL, development of ventral NMJs begins when motor neuron growth cone filopodia contact the target muscles. Over the next several hours, filopodia-ringed growth cones in contact with appropriate targets collapse to form nascent presynaptic processes of indistinct shape, termed 'prevaricosities'. By 17–22 h AEL, parts of the prevaricosities begin to constrict and/or swell such that presynaptic terminals (known as synaptic boutons) are formed 24252627282930. These presynaptic boutons are functional very early; electrophysiological analysis shows that NMJ transmission occurs within minutes of contact between motor nerve axons and muscles 3132. At the time of hatching (22–24 h AEL), presynaptic boutons are still relatively indistinct, morphologically, but nonetheless are clearly highly functional since neuromuscular transmission is required for hatching and subsequent crawling of the newly hatched first instar (L1) larva. During larval development (24–120 h AEL), the presynaptic arborization grows dramatically in order to accommodate rapidly growing larval muscles. This growth involves additions in NMJ length, branches and bouton number (from about 8–10 in at hatching to 50–100 in the third instar 6/7 NMJ). Boutons also become more distinct during NMJ development, such that the third instar NMJ typically appears as 'beads on a string' 28293033.

To examine motor nerve terminal morphology, we used fluorescently-conjugated anti-HRP antibodies (which allow visualization of all neuronal membrane), and confocal microscopy (Fig. 3). In sec8P1 and sec8Δ1 mutants, NMJ morphology appeared normal in first instar (L1) larvae (Fig. 3A). L1 mutant motor nerve insertion sites appeared normal, presynaptic swelling size was appropriate for the animal age, and the arborizations were similar in number and shape to those from age-matched control animals. The number of boutons and branches in mutant L1 ventral longitudinal muscle 6/7 NMJs were statistically unchanged compared to control animals (Fig. 3D; Control boutons per 6/7 NMJ = 13.36 ± 0.62, n = 14; sec8P1 = 11.85 ± 0.75, n = 13, p = 0.130; sec8Δ1 = 13.14 ± 0.67, n = 14, p = 0.816; control branches per 6/7 NMJ = 1.62 ± 0.22, n = 21; sec8P1 = 1.43 ± 0.25, n = 21, p = 0.577; sec8Δ1 = 1.82 ± 0.28, n = 22, p = 0.587). However, in sec8P1 third instar larvae (sec8Δ1 mutants did not survive to this stage of development), muscle 6/7 NMJs showed consistent and qualitatively obvious changes in morphology and terminal number (Fig. 3B). With less penetrance, sec8P1 mutant third instar larvae showed dramatically abnormal neurite growth (e.g. Fig. 3C) that was never observed in control NMJs. The average number of presynaptic boutons and branches was significantly increased in sec8P1 mutant larvae compared to controls (Fig. 3E; Control boutons per 6/7 NMJ = 38.45 ± 1.65, n = 22; sec8P1 = 47.50 ± 3.36 n = 22, p = 0.020; control branches per 6/7 NMJ = 3.14 ± 0.30, n = 22; sec8P1 = 5.41 ± 0.71, n = 22, p = 0.006). Homozygous sec8P1 and sec8Δ1 L1 mutants did not show any obvious defects in body wall muscle shape or size. Muscle insertions were correct and muscles did not detach abnormally during manual dissection or subsequent treatment. We conclude from these results that loss of Sec8 does not significantly impair growth of motor nerve terminals or presynaptic arborization. Instead, surprisingly, loss of Sec8 triggered an increase of NMJ size during larval development.

Sec8 has been suggested to play a role in calcium-dependent neurotransmitter secretion 11. To determine whether transmission in sec8 mutants is functionally abnormal, we used voltage clamp synaptic electrophysiology on NMJs. First we examined sec8Δ1 mutants, since Sec8 was almost completely eliminated from the NMJ in these animals (Fig. 2B). Whole-cell patch clamp recordings from homozygous sec8Δ1 mutants showed that sec8Δ1 mutant NMJs were capable of robust endogenous synaptic transmission (Fig. 4A). This result is consistent with the fact that newly hatched sec8Δ1 mutant larvae appeared to crawl and feed efficiently. The frequency of spontaneous synaptic currents (aka 'spontaneous excitatory junction currents', or 'sEJCs') was normal in sec8Δ1 mutants (Control = 13.30 ± 1.25 Hz, n = 11; sec8Δ1 = 12.34 ± 2.78 Hz, n = 10, p = 0.750). These data suggest no major disruption of presynaptic function after loss of Sec8. However, sEJC amplitude in sec8Δ1 mutant first instar NMJs was mildly but significantly reduced (K-S statistic = 0.379, p < 0.0001). Reduced sEJC amplitude suggests loss or mislocalization of postsynaptic glutamate receptors.

To test glutamate receptor function directly, we pressure ejected 1 mM glutamate on to patch-clamped (-60 mV) postsynaptic muscle. Pressure ejection of glutamate provides a means of measuring postsynaptic receptor function independent of presynaptic glutamate release. Drosophila NMJ receptors have low affinity for glutamate compared to mammalian glutamate receptors 34; 1 mM is barely saturating. Currents triggered by pressure ejection of glutamate were significantly reduced in sec8Δ1 mutants compared to controls (Fig. 4B–C; Control = 1858 ± 222.4 pA, n = 13; sec8Δ1 = 1181 ± 105.5 pA, n = 11, p = 0.016). This reduction in glutamate-gated current amplitude could be due to a decrease in the number of functional postsynaptic glutamate receptors, or a change in the biophysical properties of individual receptors. In embryonic/L1 Drosophila, these two possibilities can be distinguished by measuring the single channel current size of synaptic receptors directly. The input resistance of embryonic/L1 Drosophila muscle cells, coupled with the large conductance of insect muscle glutamate receptors, allows whole-cell patch clamp mode discrimination of delayed single channel closing on the falling phase of some sEJCs 35363738. The amplitude of these delayed synaptic receptor closings was not significantly different between sec8Δ1 mutants and controls (Fig. 4D–E; control = 12.02 ± 1.15 pA, n = 11; sec8Δ1 = 10.70 ± 0.94 pA, n = 10, p = 0.392), suggesting that the decrease in glutamate-gated current size measured in sec8 mutants is due to a loss of functional postsynaptic receptors, rather than a change in receptor properties. These latter experiments could not be repeated in sec8P1 mutant third instar larvae because synaptic receptors are not accessible for rapid agonist application because presynaptic terminals tend to be embedded deep within a subsynaptic reticulum, SSR. Decreased muscle input resistance in L3 animals also makes it impossible to discriminate single channel current amplitudes. However, two-electrode voltage clamp recordings revealed that, as in sec8Δ1 mutant L1 animals, sEJC amplitude was significantly reduced in sec8P1 mutant third instar larvae (Fig. 4F–G; K-S statistic = 0.815, p < 0.0001), consistent with loss of postsynaptic glutamate receptors in both mutant alleles throughout larval development.

To confirm the electrophysiological results and determine whether loss of receptors in sec8 mutants might be due to loss of a specific receptor subtype, we performed immunocytochemical experiments on the same synapses examined electrophysiologically (muscle 6 NMJs). Drosophila NMJs contain two spatially, pharmacologically and biophysically distinct subtypes of postsynaptic glutamate receptor 3839404142, referred to as 'A-type' and 'B-type' receptors. A and B-type receptors appear to be molecularly identical heterotetramers that differ by one subunit: A-type receptors contain the subunit GluRIIA, but not GluRIIB, and B-type receptors contain the subunit GluRIIB, but not GluRIIA. Both receptor subtypes also contain the subunits GluRIIC (aka GluRIII), GluRIID and GluRIIE 394041.

We visualized A-type glutamate receptors using an antibody against the receptor subunit GluRIIA. We visualized B-type receptors using an antibody against the receptor subunit GluRIIB. We also used an antibody against the subunit GluRIIC, which is shared by both glutamate receptor subtypes. At Drosophila embryonic/L1 NMJs, glutamate receptor subunit immunoreactivity is visible as small puncta clustered under and around presynaptic terminals 3738404142. Figure 5A shows embryonic NMJs double-stained using anti-HRP antibodies to visualize presynaptic nerve terminals (magenta) and anti-GluRIIA antibodies to visualize A-type postsynaptic glutamate receptors (green). In sec8P1 and sec8Δ1 mutant first instar larvae, the cluster size of both A- and B-type glutamate receptors was significantly reduced (Fig. 5A–B; GluRIIA Control cluster sizes = 0.45 ± 0.02 μm², n = 115 clusters in 9 animals; sec8P1 = 0.22 ± 0.01 μm², n = 124 clusters in 11 animals, p < 0.0001; sec8Δ1 = 0.24 ± 0.01 μm², n = 142 clusters in 11 animals; GluRIIB Control cluster sizes = 0.32 ± 0.01 μm², n = 117 clusters in 9 animals; sec8P1 = 0.24 ± 0.01 μm², n = 126 clusters in 11 animals, p < 0.0001; sec8Δ1 = 0.25 ± 0.01 μm², n = 142 clusters in 11 animals; GluRIIC control cluster sizes = 0.32 ± 0.01 μm², n = 140 clusters in 9 animals; sec8P1 = 0.26 ± 0.01 μm², n = 139 clusters in 11 animals, p < 0.0001; sec8Δ1 = 0.22 ± 0.01 μm², n = 141 clusters in 11 animals). Glutamate receptor cluster sizes were also significantly reduced in sec8P1 mutant third instar larvae (Fig. 5B, bottom; Control GluRIIA cluster sizes in third instar larvae = 1.42 ± 0.06 μm², n = 120 clusters in 9 animals; sec8P1 = 0.44 ± 0.03 μm², n = 120 clusters in 9 animals; p < 0.0001; Control GluRIIB cluster sizes = 0.82 ± 0.05 μm², n = 160 clusters in 11 animals; sec8P1 GluRIIB cluster sizes = 0.47 ± 0.02 μm², n = 150 clusters in 11 animals; p < 0.0001; Control GluRIIC cluster sizes = 1.52 ± 0.06 μm², n = 90 clusters in 6 animals; sec8P1 GluRIIC cluster sizes = 0.98 ± 0.05 μm², n = 90 clusters in 6 animals; p < 0.0001). Revertant third instar larvae showed no change in glutamate receptor cluster size (Fig. 5B, bottom; Control GluRIIA cluster sizes = 1.42 ± 0.06 μm², n = 120 clusters in 9 animals; sec 8rev5 = 1.33 ± 0.06 μm², n = 140 clusters in 11 animals, p = 0.303).

We also quantified the immunoreactivity of the presynaptic proteins cysteine string protein (CSP) and synaptotagmin (SYT), as well as the MAGUK protein discs-large (DLG), which is primarily localized postsynaptically. These antigens do not form distinct immunoreactive clusters. Therefore, we quantified CSP, SYT and DLG immunoreactivity by measuring average fluorescence intensity in the NMJ minus average non-NMJ muscle background intensity over an identical area of neighboring muscle membrane, normalized to control genotype (e.g.: immunoreactivity = (f_NMJ - f_muscle)_mutant / (f_NMJ-f_muscle)_control). This approach avoids use of fluorescence intensity from a different wavelength as a control, which would be inappropriate because fluorescence of different fluorophores and detection of different emission spectra can vary independently from that of the 'target' fluorophore. Immunoreactivity for CSP, DLG, and SYT was normal in both sec8P1 and sec8Δ1 (Fig. 5C–D; CSP fluorescence control = 1.0 ± 0.13, n = 14; sec8P1 = 1.19 ± 0.15, n = 14, p = 0.351; sec8Δ1 = 0.88 ± 0.11, n = 14, p = 0.519; DLG fluorescence control = 1.0 ± 0.11, n = 14; sec8P1 = 0.79 ± 0.07, n = 14, p = 0.132; sec8Δ1 = 1.06 ± 0.14, n = 14, p = 0.732; SYT fluorescence control = 1.0 ± 0.08, n = 14; sec8P1 = 0.84 ± 0.09, n = 14, p = 0.223; sec8Δ1 = 0.91 ± 0.11, n = 14, p = 0.571). Since CSP and SYT are critical for normal synaptic transmission 43, and DLG is required for proper localization of GluRIIB 3844, normal immunoreactivity for CSP, SYT and DLG is consistent with relatively normal neurotransmission and GluRIIB clustering that we observed.

Taken together, our electrophysiological and immunocytochemical results suggest that sec8 mutant NMJs are capable of relatively normal neurotransmission, except for a slight loss of postsynaptic glutamate receptors.

Although the loss of postsynaptic glutamate receptors that we observed in sec8 mutants was slight, we felt this phenotype deserved further investigation since Sec8 has been implicated in glutamate receptor trafficking, but Sec8-dependent trafficking of non-NMDA receptors has not been demonstrated 121314. Drosophila NMJ glutamate receptors are most similar in amino acid sequence to mammalian non-NMDA kainate receptors 4041. The Drosophila genome also encodes NMDA receptor subunits, which are expressed in the CNS but not in muscle 4546.

To determine whether the loss of postsynaptic receptors might be due to changes in receptor subunit expression, we measured glutamate receptor subunit mRNA levels using quantitative real-time RT-PCR. In embryonic/L1 sec8 mutants, the relative levels of GluRIIA, GluRIIB and GluRIIC mRNA were not reduced compared to controls (Fig. 6A; GluRIIA: Control = 1.0 ± 0.03 arbitrary units (a.u.), n = 7; sec8P1 = 1.21 ± 0.12 a.u., n = 8, p = 0.153; sec8Δ1 = 1.13 ± 0.06 a.u., n = 10, p = 0.128; GluRIIB: Control = 1.0 ± 0.02 a.u., n = 8; sec8P1 = 1.18 ± 0.05 a.u., n = 9, p = 0.016; sec8Δ1 = 1.13 ± 0.05 a.u., n = 10, p = 0.067; GluRIIC: Control = 1.0 ± 0.03 a.u., n = 8; sec8P1 = 1.21 ± 0.07 a.u., n = 9, p = 0.018; sec8Δ1 = 1.13 ± 0.06 a.u., n = 10, p = 0.129).

To determine whether the loss of postsynaptic receptors involved decreased glutamate receptor protein production or increased degradation, we measured total GluRIIA and GluRIIB subunit protein using immunoblots. GluRIIA and GluRIIB are thought to be expressed only in body wall muscle 404748. Immunoblot analysis suggested a reduction in neither total GluRIIA nor total GluRIIB protein in sec8 mutants (Fig. 6B–C; GluRIIA Control = 1.0 ± 0.11 a.u., n = 6; sec8P1 = 1.24 ± 0.07 a.u., n = 6, p = 0.114; sec8Δ1 = 1.29 ± 0.09 a.u., n = 6, p = 0.085; GluRIIB Control = 1.0 ± 0.12 a.u., n = 5; sec8P1 = 1.07 ± 0.19 a.u., n = 6, p = 0.754; sec8Δ1 = 1.24 ± 0.15 a.u., n = 5, p = 0.254). This result suggests that, in sec8 mutants, glutamate receptor protein may be dispersed throughout the cell instead of appropriately localized in the postsynaptic membrane, where it would have been detectable immunocytochemically and electrophysiologically. These results, in combination with our electrophysiology and immunocytochemistry, are consistent with a minor role for Sec8 in non-NMDA glutamate receptor trafficking.

Previous work has demonstrated that exocyst proteins associate biochemically with microtubule proteins 10, and that this association destabilizes microtubules 15. To test whether Sec8 regulates microtubules at the Drosophila NMJ, we examined pre and postsynaptic NMJ microtubules using an antibody against acetylated tubulin, which preferentially recognizes polymerized tubulin. Microtubules are abundant in presynaptic motor axon terminals and throughout postsynaptic muscle cells (Fig. 7A–B and 49505152). Both pre- and postsynaptically, microtubules are intimately associated with the NMJ (Fig. 7A–B and 49505152). Within the Drosophila presynaptic motor terminal, microtubules form distinct loops that appear to be critical for presynaptic arborization and regulation of bouton growth 49505354. Postsynaptic muscle microtubules also regulate NMJ growth and development 5152. In L1 animals overexpressing sec8 cDNA in muscles, microtubule immunoreactivity was noticeably decreased (Fig. 7E), but this decrease was not statistically significant (Fig. 7C–D; Control MT density = 1.00 ± 0.17, N = 11; 24B Gal4;UAS sec-8 (muscle overexpression of sec8) = 0.72 ± 0.09, N = 13, p = 0.14).

In contrast, microtubule immunofluorescence at NMJs was approximately doubled in sec8P1 and sec8Δ1 L1 mutant larvae (Fig. 7C,E; Control synaptic MT density = 1.00 ± 0.12, N = 11; sec8P1 = 2.57 ± 0.31, N = 11, p = 0.0002; sec8Δ1 = 1.95 ± 0.28, N = 11, p = 0.005). Very little Sec8 is localized extrasynaptically in wildtype animals. Therefore, a change in synaptic microtubule immunofluorescence (where Sec8 is normally found), but not in extrasynaptic microtubule immunofluorescence (where only very little Sec8 is found), is most consistent with synapse-specific regulation of microtubules by Sec8. Extrasynaptic microtubule immunofluorescence was not significantly increased in either sec8P1 or sec8Δ1 L1 mutant larvae (Fig. 7D,E; Control extrasynaptic MT density = 1.00 ± 0.17, N = 11; sec8P1= 1.39 ± 0.33, N = 11, p = 0.31; sec8Δ1 = 1.33 ± 0.25, N = 11, p = 0.28). We conclude from these data that Sec8 inhibits microtubules in vivo, as predicted by previous biochemical data showing inhibition of microtubule formation by sec proteins 1015.

Our data suggest that Drosophila Sec8 plays a role in glutamate receptor localization and microtubule regulation. These two phenotypes could represent separate functions for Sec8, or could be due to the same root cause. Mammalian glutamate receptors are trafficked to the synapse via the microtubule network 555657. Therefore, it is reasonable to surmise that the defects in postsynaptic glutamate receptor localization measured in sec8 mutants might be due to misregulation of microtubules. To address this, we mimicked the increase in microtubules observed in sec8 mutants by overexpressing alpha tubulin. Overexpression of an alpha tubulin transgene in postsynaptic muscles using the gal4-UAS system doubles microtubule density (Images not shown; Control MT density = 1.00 ± 0.17, N = 11; UAS-α tub/B24 gal4 tubulin = 2.21 ± 0.24 a.u., n = 12, p = 0.001), presumably due to a shift in the dynamic equilibrium between disassembled and assembled tubulin subunits. This shift toward an increase in microtubules caused a significant decrease in the size and number of postsynaptic glutamate receptor clusters, similar to that observed in sec8 mutants (Fig. 8; Control GluRIIA cluster size = 0.45 ± 0.02 μm², n = 115 clusters in 8 animals; UAS-α tub/B24 gal4 = 0.28 ± 0.01 μm², n = 104 clusters in 9 animals, p < 0.0001; Number of control clusters = 0.68 ± 0.06 clusters/μm², n = 13 NMJs; UAS-α tub/B24 gal4 = 0.47 ± 0.05 clusters/μm², n = 14 NMJs, p = 0.014). These results suggest that the loss of postsynaptic glutamate receptors observed in sec8 mutants may be secondary to disruption of microtubule networks.

To identify genes required for glutamatergic synapse development, we screened GT1 and SuPor-P transposon insertions from the Berkeley Drosophila Gene Disruption Project 17. Specifically, we first identified homozygous mutants using the FlyBase Insertions Query Form http://flybase.bio.indiana.edu/. Approximately 10% (220 of 2185 stocks) of the insertion lines available in March 2003 contained lethal inserts. We examined dechorionated embryos from each of the 220 stocks to ensure that homozygous mutants developed to at least late stage embryogenesis (16–17 h after egg laying, AEL), which is when NMJs begin to form. Of the 220 mutants, 202 developed into morphologically mature embryos (condensed CNS, clear segmentation, trachea, mouth hooks, etc.). These 202 mutants were rebalanced using a chromosome-specific GFP balancer for unambiguous identification of homozygous P-element mutants and then screened immunocytochemically using anti-HRP and anti-GluRIIA antibodies to visualize presynaptic nerve terminals and postsynaptic glutamate receptors, respectively. The remaining 18 stocks were discarded from future consideration. P{SUPor-P}CG2095^KG02723 mutants (herein referred to as 'sec8P1') were identified as a member of the subgroup of mutants with defects in both presynaptic morphology and postsynaptic glutamate receptor expression. A more complete description of this screen, and the mutants identified therein, can be found elsewhere 56.

sec8Δ1 mutants were created by imprecise excision of P{SUPor-P}CG2095^KG02723 from sec8P1, using standard methods for P-element mobilization 18. Four hundred and forty-five independent mutant lines of flies were created; approximately 8% of these (35 of 445) were homozygous lethal. The lethal lines were then screened using a PCR-based assay with three pairs of primers. One pair of primers was designed to amplify a portion of sec8 near the P-element insertion. A second set of primers was designed to amplify a portion of the gene CG2082, located immediately 5' to CG2095, and the third set of primers was designed to amplify a portion of gene CG2091, located immediately 3' to sec8. Mutants in which the first (sec8 specific) primer pair did not produce a product but the second and third primer sets did suggested that sec8 was specifically deleted but the neighboring genes were left intact. Eight mutants of this type were identified. Lethal lines containing partial deletions of sec8 were further characterized using several sets of PCR primers. Both forward and reverse primers were designed to every 1 kb of genomic DNA beginning 500 bp upstream and downstream of sec8. Using these primers, one deletion mutant (sec8Δ1) was isolated in which the first forward primer (positioned approximately 500 bp 5' of sec8) and the last reverse primer (positioned approximately 500 bp 3' of sec8) produced a product of approximately 1 kb. This PCR product was sequenced to determine the exact location and size of this deletion (sec8Δ1).

Precise excisions of P{SUPor-P}CG2095^KG02723 ('revertants') were produced by the same P-element mobilization that produced sec8Δ1. Revertants were initially identified based on eye color and reversion of the lethal insertion phenotype, then confirmed by PCR. Measurements from revertants, in which the sec8 P-element was precisely excised, are indicated in the text and figures. For all experiments, homozygous mutant larvae were identified by use of an appropriate GFP-tagged balancer chromosome. 'Control' genotypes in all experiments are w¹¹¹⁸. Animals were age-matched for all experiments using timed laying plates and morphological staging of late-stage embryos.

Wild-type CG2095 (sec8) cDNA was cloned by RT-PCR amplification of embryonic RNA. Total RNA was isolated using Trizol extraction and reverse transcribed using cDNA-specific primers and cloned. The 3256 bp sec8 cDNA was ligated into the pUAST vector using the KpnI and XbaI sites, and then sequenced for confirmation. Our results are consistent with the predicted gene structure shown in FlyBase, although we noted several polymorphisms. We used our cDNA clone sequence to predict the CG2095 amino acid sequence. Our predicted Oregon R CG2095 amino acid sequence contains three differences compared to the predicted CG2095 translation product as annotated in FlyBase: Oregon R CG2095 encodes an M at amino acid 119, a D at amino acid 328, and a Y at amino acid 669.

Fly transformation was performed by DNA microinjection of embryos using standard methods (Genetic Services, Inc, Cambridge MA). Ten independent UAS-sec8 transgenic strains were created.

Quantitative real-time RT-PCR of Drosophila glutamate receptor subunits was performed as previously described 41. Briefly, total RNA was isolated from 22–24 h AEL embryos using Trizol extraction and reverse transcribed using sequence specific primers. Real-time RT-PCR was then performed using subunit cDNA-specific primers in an Opticon 2 real-time cycler (MJ Research, Waltham MA), using SYBR green (Molecular Probes, Eugene, OR) for fluorescent measurement of amplicon quantity. Nonspecific fluorescence due to primers was eliminated by measuring fluorescence after a short 'holding step' at a temperature sufficient to melt primer dimers but not desired product (MJ Research Technical Note # 004). As a loading control, actin 5C RNA levels were also measured for every extraction using actin 5C-specific primers, and GluRII RNA levels normalized using this measurement. Each measurement represents the actin 5C c(t) divided by a GluRII c(t), where both measurements were made from the same RNA isolation and RT reaction.

To generate Sec8 antibodies, rabbit polyclonal antisera was raised against a synthesized peptide composed of Sec8 amino acids 440–460 (GTSNNSDAFKEHRRNASDASV). The antiserum was affinity purified and used at 1:500. For staining and microscopy, animals were manually dissected and fixed for 30–60 min in either Bouin's fixative (when Sec8 or GluRII antibodies were used), or 4% paraformaldehyde (for all other staining). Note that the Sec8 antibody recently described in 21 as not showing immunoreactivity at the fly NMJ, actually does so (and shows staining very similar to our antibody) when preparations are fixed in Bouin's fixative. Mouse monoclonal anti-GluRIIA ('8B4D2', Iowa Developmental Studies Hybridoma Bank, Iowa City, IA) was used at 1:100. Rabbit polyclonal anti-GluRIIB and anti-GluRIIC 39 were gifts from Dr. Aaron DiAntonio (Washington University, St. Louis, MO) and used at 1:2000 and 1:3000, respectively. Fluorescently conjugated anti-HRP (Jackson Immunoresearch Labs, West Grove, PA) was used at 1:100. Mouse monoclonal anti-CSP was used at 1:200 65. Mouse monoclonal anti-Discs Large (DLG) ('4F3', Iowa Developmental Studies Hybridoma Bank, Iowa City, IA) was used at 1:1000. Mouse monoclonal anti-synaptotagmin (SYT) was used at 1:500 66. Mouse monoclonal anti-acetylated tubulin (Sigma) was used at 1:1000. Goat anti-rabbit or goat anti-mouse fluorescent (FITC or TRITC) secondary antibodies (Jackson Immunoresearch Labs, West Grove, PA) were used at 1:400. Confocal images were obtained using an Olympus FV500 laser-scanning confocal microscope. All images represent maximum intensity projections of confocal Z-stacks. Image analysis and quantification was performed on maximum intensity projections of Z-stacks using ImageJ software. Three-dimensional surface reconstructions were generated from confocal Z-stacks using Amira 3.1 (Mercury Computer Systems, Chelmsford, MA).

Quantification of glutamate receptor cluster size was performed as previously described 37384267. Every cluster on a 6/7 NMJ from each animal was measured. Drosophila NMJ glutamate receptor clusters form over the course of hours and then stabilize to form distinct clusters with relatively invariant diameter 3767. These diameters stay relatively constant, even throughout larval development 30. Because changes in the number of postsynaptic receptors within each cluster are linearly proportional to changes in cluster area; cluster area is a quantitative and sensitive optical measure of receptor number in the Drosophila NMJ 303767. Cluster area was quantified and measured from maximum intensity Z-projections of confocal image stacks, using automated edge finding and area measurement in ImageJ software (NIH, Bethesda MD). This method agrees well with manual measurements 37, is highly reproducible 384267, and involves no experimenter bias for cluster selection because the software selects and measures every bright spot regardless of shape or size. An alternative measurement of glutamate receptor abundance is fluorescence intensity. Although fluorescence intensity measurements gave qualitatively identical results for this study (data not shown), we do not prefer them for embryonic/L1 receptor abundance measurements because intensity measurements assume changes in protein density within individual PSDs (for which we have no evidence) or underestimate differences in receptor abundance due to fluorescence measurement from pixels between individual GluR clusters. Because immunoreactivity for microtubules and the synaptic proteins DLG, CSP and SYT did not show distinct clusters, we quantified their abundance using fluorescence intensity. However, all of these measurements are relative to 'background' fluorescence (measured from non-muscle areas for microtubules, or nonsynaptic areas for DLG, CSP and SYT) in the same image and filter channel 68. We feel that this is an important wavelength-matched control for variation due to possible differences in staining, mounting of the preparation, or microscope settings.

After protein isolation for immunoblotting, protein abundance was quantified using Bradford assays, and the same amount of protein was loaded per lane on 6% SDS-polyacrylamide gels. The amount loaded was 20 μg for GluRIIB blots, 50 μg for GluRIIA blots and 20 μg for Sec8 blots. After electrophoresis, protein was transferred to nitrocellulose membranes. Nitrocellulose membranes were blocked with 5% milk in TBST (10 mM Tris, pH 8.0, 150 mM NaCl and 0.1% Tween 20) for 1–2 h at room temperature. Anti-Sec8 was used at 1:500 and anti-GluRIIA ('8B4D2' concentrate, Iowa Developmental Studies Hybridoma Bank, Iowa City, IA) was used at 1:200. GluRIIB antibodies usable for immunoblots have not previously been described. To generate Anti-GluRIIB antibodies for use in immunoblots, rabbit polyclonal antisera was raised against synthetic peptides containing C-terminal sequence of GluRIIB: RQSRDSTSTGYSSLEQITSASSAKKKK. For immunoblotting, GluRIIB antibodies were used at 1:400. Note that these GluRIIB antibodies are not the same as used for in situ staining; as described above, the GluRIIB antibodies used for in situ staining are described in Marrus et al. 39. After incubating at 4°C overnight, the membranes were washed with TBS-T. Anti-mouse HRP-conjugated secondary antibodies were used at 1:5000 for GluRIIA blots and anti-rabbit AP-conjugated secondaries were used at 1:1000 for GluRIIB and Sec8 blots. Protein bands on blots were visualized using enhanced chemiluminescence (GluRIIA) or via a colorimetric reaction (GluRIIB and Sec8). Both GluRIIA and GluRIIB bands are completely eliminated in appropriate mutants (Df(2L) [AD9] and Df(2L) [SP22]) 40 and Liebl, Ng, Sheng, Karr and Featherstone, unpublished).

All electrophysiology was performed on the ventral body wall muscle 6 at 19°C. For electrophysiology on first instar (L1) larvae we used a whole-cell patch clamp technique. Third instar larval muscles are too large for whole cell patch clamp. Therefore, we used two-electrode voltage clamp technique for third instar (L3) larval recordings. Both techniques were performed as previously described 363868. The sEJCs analyzed in this study, in either first or third instar larvae, are unlikely to include events resulting from endogenous action potentials, based on two pieces of evidence: (1) Endogenous action potential evoked activity is much larger, typically patterned, and thus easily recognizable, and (2) Drosophila sEJCs recorded in 0 mM calcium with 5 μm TTX occur with less frequency but are not larger, in both embryos/first instars and third instar larvae [68 and unpublished observations]. Therefore, although we do not methodologically exclude any particular type of synaptic event and therefore properly use the conservative term 'sEJC' to describe what we are measuring, all of the events shown and analyzed in this study are likely to represent only true spontaneous monovesicular 'mEJCs', or 'minis'.

For patch clamp electrophysiology on L1 (22–24 h AEL) larvae, animals were manually dissected by gluing (Histoacryl Blue, Braun Germany) them to sylgard-coated coverslips under standard Drosophila saline (135 mM NaCl, 5 mM KCl, 4 mM MgCl₂, 1.8 mM CaCl₂, 5 mM TES, 72 mM sucrose). A slit was then made along the dorsal midline using a glass capillary pulled to a sharp point, and the body walls glued flat to the coverslip. After dissection, the exposed muscle sheath was enzymatically removed by 30–90s exposure to 1 mg/ml collagenase (type IV, Sigma). Muscle 6 was then whole-cell patch clamped (-60 mV) in standard Drosophila saline (135 mM NaCl, 5 mM KCl, 4 mM MgCl, 1.8 mM CaCl, 5 mM TES, 72 mM sucrose) using standard patch-clamp techniques. Pressure ejection of 1 mM glutamate dissolved in extracellular saline utilized a Picospritzer III (General Valve/Parker Hannifin, Fairfield NJ). Pressure ejection pipettes were approximately the same size and tip diameter as unpolished patch pipettes. Glutamate leak was minimized by continuous back-pressure; no evidence for glutamate leak and subsequent desensitization of receptors was ever observed, and would be identical between genotypes. Data were acquired using an Axopatch 1 D amplifier and a PC running pClamp9 software (Axon Instruments, Union City, CA). Data were subsequently analyzed using Clampfit9 (Axon Instruments).

For two-electrode electrophysiology on third instar (L3) larvae (110–120 hr AEL), Dissections were made as with first instar larvae, except that collagenase was not used. Larval dissections and recordings were also performed under standard Drosophila saline, as described above. Standard two-electrode voltage clamp techniques were used (holding potential -60 mV), as previously described 36. Electrodes for TEVC were filled with 3 M KCl. Data was acquired using an Axon GeneClamp 500B and a PC running Axoscope software. Data were subsequently analyzed using Clampfit9 (Axon Instruments)

In all cases (data from both L1 and L3 animals) sEJCs were detected and analyzed using Clampfit9's template-matching method 6970, which identifies synaptic events based on shape matching to a data-based ideal template. In Drosophila embryos and L1 larvae, small sEJCs as recorded and analyzed for this study represent a mixture of spontaneous calcium-dependent and calcium-independent monovesicular events. Changing the calcium concentration to 0 mM in the recording saline decreases the frequency of embryonic/L1 sEJCs, but not their amplitude 68.

All statistical comparisons were made using unpaired T-tests or, for distributions, Kolmogorov-Smirnov tests. Statistical significance in figures is represented as follows: * = p < 0.05, ** = p < 0.01, and *** = p < 0.001. All error bars represent S.E.M.