RIM-Binding Protein, a Central Part of the Active Zone, Is Essential for Neurotransmitter Release

In response to Ca²⁺ influx, synaptic vesicles (SVs) fuse with the active zone (AZ) membrane. An elaborate protein-based cytomatrix covering the AZ membrane is meant to facilitate the release process, but its components and operational principles are poorly understood (1–4). We previously found Bruchpilot to be an essential building block of the Drosophila AZ cytomatrix (the T bar) (5). Here, we used antibodies to screen for additional cytomatrix components.

Rab3-interacting molecule (RIM)–binding proteins (RBPs) are enriched at presynaptic terminals (6–8) and interact with voltage-gated Ca²⁺ channels (VGCCs). Mammals harbor three rbp family loci; Drosophila has a single rbp gene (drbp) (9). We generated antibodies against Drosophila RBP (DRBP) (epitopes in Fig. 1A), which stained presynapses at neuromuscular junctions (NMJs) (Fig. 1B, left). The size of AZ cytomatrices is below the diffraction limit of light microscopy. To overcome this, we used a two-color stimulated emission depletion (STED) (10) microscope providing 50-nm resolution in the focal plane (Fig. 1B, right) (11). Relative to the Bruchpilot C-terminal label, DRBP C-terminal immunoreactivity localized toward the AZ center (Fig. 1C, left; model: Fig. 1F). Vertically, the DRBP label localized ~100 nm closer to the AZ membrane than the Bruchpilot C-terminal label (Fig. 1C, right; quantification in Fig. 1F). Postembedding immunogold labeling against DRBP C terminus yielded similar results (fig. S1). DRBP C- and N-terminal labels are similarly distributed in planar views (compare Fig. 1C and 1D, left), and in vertical views the centers of the DRBP C- and N-terminal signals are on average ~30 nm apart (compare Fig. 1C and 1D, right). The Cacophony (Cac) VGCC (12, 13) localizes beneath the scaffold formed by Bruchpilot in the AZ center (14). DRBP C terminus encircled a small (50 to 100 nm) AZ-central field of Cac^GFP (with green fluorescent protein) (12, 15) (Fig. 1E, left).

• Download high-res image
• Open in new tab
• Download Powerpoint

We subjected the drbp locus to genetic analysis (fig. S2). A strain with a transposon inserted between exons 6 and 7 (MB02027, drbp^Minos) was positioned in trans over a deficiency including drbp, as well as a neighboring locus [Df(3R)S201, short Df, see fig. S2]. In these larvae, DRBP levels at NMJs were reduced to one-third of control levels (Fig. 2A) (drbp^Minos; arbitrary fluorescence intensity 36% ± 2, n = 27; control 100% ± 2, n = 21; P > 0.0001, Mann-Whitney U test). These drbp hypomorphic flies hatched below expected Mendelian ratios, and mutant larvae showed reduced locomotion (fig. S3). Chemical mutagenesis screening provided alleles with premature stop codons (drbp^STOP1, drbp^STOP2, and drbp^STOP3) (Fig. 1A and fig. S2). Animals carrying these alleles over Df only rarely reached adulthood, and mutant larvae barely moved (fig. S3). DRBP immunoreactivity at mutant larval NMJs was completely absent when we used either N- (fig. S4) or C-terminal antibodies (Fig. 2A). Thus, we consider these alleles to be null. Using pacman technology (16), we produced a genomic transgene encompassing the entire drbp locus (Rescue) (fig. S2). One copy of this construct partially restored NMJ staining (Fig. 2A) and partially rescued drbp^STOP1-3 adult vitality. Mutant larval NMJ terminals reached normal morphological size (Fig. 2B) (control: 516 ± 51 μm², n = 20; drbp^STOP1: 554 ± 41 μm², n = 20; P = 0.4903, Mann-Whitney U test), and had normal synapse numbers (Fig. 2B) (control: 454 ± 39, n = 20; drbp^STOP1: 489 ± 34, n = 20; P = 0.36, Mann-Whitney U test), as also seen in transmission electron microscopy (TEM) (fig. S5A, arrows). Postsynaptic glutamate receptor (GluR) fields appeared enlarged (Fig. 2B).

• Download high-res image
• Open in new tab
• Download Powerpoint

Bruchpilot spots in drbp mutants appeared largely unaltered at confocal resolution (Fig. 2B). However, with STED, Bruchpilot signals emerged as atypically organized, no longer forming regular donuts but adopting irregular shapes (compare Fig. 2C, arrows left to right). Bruchpilot signals extended into the cytoplasm unusually far (Fig. 2C, asterisks), not matching postsynaptic GluR fields in either planar (Fig. 2C, arrows) or vertical views (Fig. 2C, arrowheads). Organization of the AZ cytomatrix in drbp mutants was clearly affected in electron microscopy (EM) of both conventionally embedded samples (fig. S5B) and high-pressure frozen- and freeze-substituted (HPF) (14, 17, 18) samples (Fig. 2D). The drbp hypomorph (drbp^Minos/Df) still formed structures resembling T-bar platforms (Fig. 2D, arrows), but they resided on thinner pedestals (Fig. 2D, arrowheads) (T-bar width: controls = 85.5 ± 10, drbp^Minos = 56.3 ± 5.5, P < 0.05; T-bar height: controls = 55.8 ± 1.4, drbp^Minos = 48.5 ± 2.2, P < 0.01, Student’s t test). At AZ membranes of drbp nulls, abnormally shaped electron-dense material was found, but no regular T bars (Fig. 2D). In three-dimensional electron tomography reconstructions, it became obvious that the entire cytomatrix was severely misshapen (Fig. 2E). We occasionally observed free-floating electron-dense material detached from the AZ plasma membrane in drbp nulls (fig. S5, C and D). These atypical electron densities still tethered SVs (fig. S5B, arrows; insets in fig. S5, C and D), a function mediated by the C-terminal end of Bruchpilot (19). With EM, overall SV density in NMJ boutons appeared unaltered (fig. S5A). The number of membrane-proximal SVs (up to a 5-nm distance) counted over the whole AZ, however, were reduced in drpb^STOP1 animals [control: 3.2 ± 0.2, n of AZs = 26; drbp^Minos: 3.1 ± 0.3, n of AZs = 17, P = 0.79; drbp^STOP1: 2.1 ± 0.3, n of AZs = 16; P < 0.005, Student’s t test; scored with HPF, as standard aldehyde fixation is imprecise for this purpose (20)]. For functional analysis, two-electrode voltage-clamp recordings were performed at larval NMJs. In drbp hypomorphs, evoked excitatory junctional current (eEJCs) amplitudes were reduced by about half (Fig. 3A). In drbp nulls, synaptic transmission was practically abolished (Fig. 3A) [control: –52.4 ± 4.9 nA, n = 10; drbp^STOP1: –2.6 ± 0.2 nA, n = 4; drbp^Minos: –23.0 ± 5.3 nA, n = 8; P < 0.001; one-way analysis of variance (ANOVA) Tukey’s post test]. We facilitated synaptic release by recording in elevated extracellular Ca²⁺. In 2 mM Ca²⁺, eEJC amplitudes were reduced to ~10% of control levels in all three null alleles. One copy of the genomic drbp transgene (fig. S2) completely rescued this phenotype (Fig. 3B) (control: –113.7 ± 7.2 nA, n = 12; drbp^STOP1: –11.7 ± 2.8 nA, n = 9; drbp^STOP2: –9.8 ± 3.1 nA, n = 6; drbp^STOP3: –9.6 ± 1.6 nA, n = 8; Rescue: –119.1 ± 12.7 nA, n = 7; P < 0.001; one-way ANOVA, Tukey’s post test). Frequency and amplitude of miniature excitatory junctional currents (mEJCs) were unchanged in drbp^STOP1 (Fig. 3C), despite enlarged GluR fields. Thus, the number of quanta (i.e., SVs) released per individual action potential (AP) (quantal content, Fig. 3D) was dramatically reduced in the absence of DRBP.

• Download high-res image
• Open in new tab
• Download Powerpoint

At paired-pulse stimulation, drbp^STOP1 synapses exhibited an unusually strong facilitation (Fig. 4A). During a 10-Hz stimulation train, a protocol that leads to marked depression of eEJCs in controls, drbp^STOP1 mutants showed sustained facilitation and reached a steady-state level more than double the initial amplitude (Fig. 4B). When stimulated with five consecutive pulses at 100 Hz, drbp^STOP1 eEJCs exhibited substantial recovery and almost reached absolute control levels on the fifth pulse (Fig. 4C). Altogether, these results suggest that drbp null mutants suffer from a severely reduced release probability. However, as SV release can be substantially elevated under high-frequency stimulation, we conclude that the core fusion machinery is still operational in drbp mutants, and we hypothesize that the reduced SV release evoked by single APs is mainly due to defects upstream of the SV fusion process.

• Download high-res image
• Open in new tab
• Download Powerpoint

The eEJC rise times were slightly but significantly increased (Fig. 4D) (control 1.00 ± 0.05 ms, n = 12; drbp^STOP1 1.34 ± 0.09 ms, n = 9; P < 0.01; Student’s t test), whereas mEJC rise times were unchanged (control 1.20 ± 0.06 ms; n = 10; drbp^STOP1 1.21 ± 0.07 ms, n = 10; P = 0.9; Student’s t test). Thus, evoked SV fusion events in drbp mutants appeared desynchronized with the invasion of the presynaptic terminal by an AP. Because this synchronization depends on the spatiotemporal pattern of AP-triggered Ca²⁺ influx into the nerve terminal (21), changes in the abundance of Ca²⁺ channels might contribute to the observed defect. Indeed, Ca²⁺ channel signals, as evaluated by expression of Cac^GFP (5, 22), were slightly but consistently reduced by ~25% measured over the whole NMJ in all three drbp null alleles (Fig. 4F) (control 100% ± 2, n = 136; drbp^STOP1 74% ± 3, n = 52; drbp^STOP2 80% ± 3, n = 19; drbp^STOP3 70% ± 3, n = 45). For the drbp hypomorph, Ca²⁺ channel signal was reduced by about 25% as well (Fig. 4F) (drbp^Minos 71% ± 1, n = 27; P < 0.001, Mann-Whitney U test), whereas the evoked response was only reduced to 44% (Fig. 3A). Measuring Cac^GFP intensity per postsynaptic density (PSD) yielded a slightly stronger 36% decrease in drbp^STOP1 animals compared with controls (control 100% ± 4, n = 48 NMJs; drbp^STOP1 64% ± 3, n = 52; P < 0.001, Mann-Whitney U test). This decrease in Ca²⁺ channel signal intensity was reflected by a ~30% decrease in calcium influx in response to single APs as determined by the ΔF/F amplitude of spatially averaged Ca²⁺ signals recorded from drbp^STOP1 mutant boutons (Fig. 4E) (control: 0.85 ± 0.06, n = 20 boutons; drbp^STOP1 0.57 ± 0.03, n = 44 boutons; P < 0.001, Student’s t test).

Consistent with previous findings in mammals (6, 7, 23) we found a biochemical interaction between DRBP and both Drosophila Ca²⁺ channel α1 subunit Cac (yeast two-hybrid, human embryonic kidney (HEK) cell culture membrane recruitment assay) (fig. S6) and the Drosophila homolog of AZ protein RIM (yeast two-hybrid) (fig. S6, A and B). Interactions were specifically mediated by highly homologous PXXP motifs that bind to the third DRBP SH3 domain (fig. S6, A and B). Recently, RIM has been suggested to tether Ca²⁺ channels to the AZ membrane by means of (i) a direct interaction with the Ca²⁺ channel α1 subunit and (ii) an interaction with RBP (23). Mouse rim-1 and rim-2 double knockouts (DKOs) (23, 24), however, showed a rather moderate reduction in SV release in comparison with the dramatic drbp null phenotype. Thus, DRBP might bundle multiple interactions that are not necessarily downstream of RIM. Mechanistically, loss of DRBP leads to defects in Ca²⁺ channel abundance, Ca²⁺ influx, SV docking, and cytomatrix organization that probably all contribute to the severity of the drbp-release deficit.

Of note, in Bruchpilot mutants, cytomatrix and Ca²⁺ channel clustering defects are more pronounced than in drbp nulls (5, 14). Functionally, drbp and bruchpilot phenotypes appear similar: Both demonstrate decreased and desynchronized evoked SV release with atypical short-term facilitation. However, the deficits in evoked SV release are much more severe in drbp nulls than in bruchpilot nulls [i.e., release occurs at 5% versus 30% (5) of the respective wild-type level]. DRBP levels were clearly reduced in bruchpilot mutants (fig. S7), whereas gross Bruchpilot levels were not altered in drbp mutants (Fig. 2B). Given that even a partial loss of DRBP causes marked reduction in SV release (Fig. 3A), deficits in bruchpilot mutants might be explained, at least in part, by a concomitant loss of DRBP, and DRBP probably serves functions beyond the structural and Ca²⁺ channel–clustering roles of Bruchpilot.

Taken together, we identified DRBP as a central part of the AZ cytomatrix. How, in detail, DRBP functionally integrates into this protein network is subject to future analyses. Notably, the short-term plasticity phenotype of drbp mutants is reminiscent of mammalian munc13-1 KO and caps-1 and caps-2 DKO mutants (25, 26), which implicates functional links between priming factors and DRBP. Consistent with the functional importance of the DRBP protein family suggested by our study, human genetics recently identified two rbp loci associated with autism with high confidence (27, 28).