Binding to Rab3A-interacting Molecule RIM Regulates the Presynaptic Recruitment of Munc13-1 and ubMunc13-2*

Transmitter release at synapses between nerve cells is spatially restricted to active zones, where synaptic vesicle docking, priming, and Ca2+-dependent fusion take place in a temporally highly coordinated manner. Munc13s are essential for priming synaptic vesicles to a fusion competent state, and their specific active zone localization contributes to the active zone restriction of transmitter release and the speed of excitation-secretion coupling. However, the molecular mechanism of the active zone recruitment of Munc13s is not known. We show here that the active zone recruitment of Munc13 isoforms Munc13-1 and ubMunc13-2 is regulated by their binding to the Rab3A-interacting molecule RIM1α, a key determinant of long term potentiation of synaptic transmission at mossy fiber synapses in the hippocampus. We identify a single point mutation in Munc13-1 and ubMunc13-2 (I121N) that, depending on the type of assay used, strongly perturbs or abolishes RIM1α binding in vitro and in cultured fibroblasts, and we demonstrate that RIM1α binding-deficient ubMunc13-2I121 is not efficiently recruited to synapses. Moreover, the levels of Munc13-1 and ubMunc13-2 levels are decreased in RIM1α-deficient brain, and Munc13-1 is not properly enriched at active zones of mossy fiber terminals of the mouse hippocampus if RIM1α is absent. We conclude that one function of the Munc13/RIM1α interaction is the active zone recruitment of Munc13-1 and ubMunc13-2.

Transmitter release at synapses between nerve cells is spatially restricted to active zones, where synaptic vesicle docking, priming, and Ca 2؉ -dependent fusion take place in a temporally highly coordinated manner. Munc13s are essential for priming synaptic vesicles to a fusion competent state, and their specific active zone localization contributes to the active zone restriction of transmitter release and the speed of excitation-secretion coupling. However, the molecular mechanism of the active zone recruitment of Munc13s is not known. We show here that the active zone recruitment of Munc13 isoforms Munc13-1 and ubMunc13-2 is regulated by their binding to the Rab3A-interacting molecule RIM1␣, a key determinant of long term potentiation of synaptic transmission at mossy fiber synapses in the hippocampus. We identify a single point mutation in Munc13-1 and ubMunc13-2 (I121N) that, depending on the type of assay used, strongly perturbs or abolishes RIM1␣ binding in vitro and in cultured fibroblasts, and we demonstrate that RIM1␣ binding-deficient ubMunc13-2 I121 is not efficiently recruited to synapses. Moreover, the levels of Munc13-1 and ubMunc13-2 levels are decreased in RIM1␣-deficient brain, and Munc13-1 is not properly enriched at active zones of mossy fiber terminals of the mouse hippocampus if RIM1␣ is absent. We conclude that one function of the Munc13/RIM1␣ interaction is the active zone recruitment of Munc13-1 and ubMunc13-2.
The release of neurotransmitters at chemical synapses between nerve cells is spatially restricted to specialized compartments of the axonal plasma membrane. These so called active zones, at which exocytosis takes place in a temporally highly coordinated manner, are typically localized in axon terminals, face the postsynaptic signal reception apparatus, and consist of electron-dense, insoluble proteinaceous material (1)(2)(3).
Most presynaptic axon terminals contain hundreds of vesicles that are clustered in close proximity of the active zone. Usually, only a fraction of vesicles docked at the active zone are in a fusion-competent primed state, and only these primed vesicles, which are referred to as the readily releasable pool, can fuse with the plasma membrane in response to an elevation of the intracellular Ca 2ϩ concentration. The readily releasable vesicle pool represents a reservoir that allows a synapse to repetitively release neurotransmitters during bursts of action potentials, and its size determines synaptic release probability and signaling capacity (1,4).
All known active zone-specific proteins interact with each other in multiple combinations (9,15,17,22,27,30) and thus form a complex proteinaceous network. These interactions between active zone components are likely to serve functions beyond mere recruitment and anchoring of the active zone protein complement. Rather, the active zone protein network is thought to be responsible for the temporal coordination of synaptic vesicle docking, priming, and fusion, and for the spatial restriction of these processes to the active zone subdomain of the presynaptic plasma membrane (1).
The molecular processes that are mediated by individual active zone-specific proteins are in most cases still unknown. An exception are the members of the Munc13 family of essential synaptic vesicle priming proteins, Munc13-1, ubMunc13-2, bMunc13-2, and Munc13-3. Studies on deletion mutant mice showed that in the absence of Munc13s, synaptic vesicle priming is abolished, resulting in a complete block of spontaneous and evoked synaptic transmission (31,32). The same phenotype is observed in Caenorhabditis elegans and Drosophila mutants lacking the corresponding invertebrate homologues, Unc-13 and Dunc-13 (33,34). Munc13s and their invertebrate homologues are thought to exert their priming function by binding syntaxin 1 and stabilizing it in an open conformation that is able to form SNARE (soluble N-ethylmaleimide-sensitive fusion factor attachment protein receptor) complexes (35)(36)(37). In addition, Syntaxin binding by Munc13-like proteins may regulate subsequent steps in the transmitter release process (38). Given that Munc13-mediated synaptic vesicle priming is absolutely essential for synaptic transmission, the active zonespecific localization of Munc13s is a key contributor to the spatial restriction of transmitter release to active zones and to the speed of excitation secretion coupling at synapses (1).
The Munc13 isoforms Munc13-1 and ubMunc13-2 bind to the Zn 2ϩ finger region of ␣RIMs via their evolutionarily conserved N-terminal region (30,39), allowing the formation of tripartite Munc13-␣RIM-Rab3 complexes (39). RIM1␣ plays a crucial role in the control of synaptic transmitter release and presynaptic forms of synaptic long term potentiation. Hippocampal neurons lacking RIM1␣ exhibit reduced synaptic vesicle priming, altered short term synaptic plasticity, and changes in evoked asynchronous transmitter release (25), a phenotype that is similar to that of Munc13-deficient neurons but less severe (29,32). Specific interference with the interaction between Munc13-1 and ␣RIMs inhibits synaptic vesicle priming (30,39), indicating that Munc13/␣RIM interaction has a regulatory function in synaptic vesicle priming in vivo, and that the two proteins act in the same presynaptic regulatory pathway. This notion is supported by the findings that Munc13-1 levels are dramatically reduced in the brains of RIM1␣-deficient mice (22), and that in C. elegans the phenotypes of loss of function mutations in both, the unc-13 and unc-10 (RIM) genes, are partially rescued by the overexpression of a mutant Syntaxin that preferentially adopts an open conformation (36,40). In addition to its role in basal synaptic transmitter release, RIM1␣ is essential for long term potentiation (LTP) 2 of synaptic transmission at mossy fiber synapses in the hippocampus and at parallel fiber synapses in the cerebellum (23,24). Both types of LTP are expressed presynaptically and require the activity of protein kinase A, and protein kinase A-mediated phosphorylation of RIM1␣ is necessary for the expression of mossy fiber LTP (24). This role of RIM1␣ in mossy fiber LTP appears to be critical for cognitive functions as RIM1␣-deficient mice exhibit severely impaired learning and memory (41).
Whereas the evidence for a functional interaction of Munc13s and ␣RIMs in vivo is extensive, the exact role of this interaction is not known. The currently available data are compatible with a role of ␣RIMs in both, anchoring and direct regulation of Munc13-1 and ubMunc13-2. In the present study, we determined the role of RIM1␣ in the synaptic recruitment of Munc13-1 and ubMunc13-2. Our data show that RIM1␣ regulates the active zone recruitment of Munc13-1 and ubMunc13-2.
Cosedimentation Assays-For cosedimentation experiments using synaptosomal extracts, crude synaptosomes from rat brain were solubilized at a protein concentration of 2 mg/ml in 50 mM Tris (pH 8), 150 mM NaCl, 1 mM EGTA, 1% sodium cholate, 0.2 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, and 0.5 g/ml leupeptin. After stirring on ice, insoluble material was removed by centrifugation (60 min at 250,000 ϫ g max ). The equivalent of 4 mg of total extracted protein was then incubated overnight at 4°C with 20 g of GST-Munc13 fusion protein, immobilized on glutathione-agarose beads. Beads were then washed three times with solubilization buffer containing 0.1% sodium cholate. Bound proteins were eluted with 100 l of SDS-PAGE sample buffer and analyzed by SDS-PAGE and immunoblotting. Immunoblots for native ␣RIMs were performed using a monoclonal anti-␣RIM antibody (Transduction Laboratories). For cosedimentation assays using purely recombinant His-RIM1␣(131-214), the His-tagged proteins were purified on nickel-nitrilotriacetic acid beads. The assay procedure was as described above for native RIM1␣ from crude synaptosomes, with the exception that 1% Triton X-100 was used for solubilizing bacterially produced His-RIM1␣(131-214) and a monoclonal anti-His 6 antibody (Qiagen) was used for immunoblotting.
Expression of Munc13s and RIM1␣ and Assay of Phorbol Esterdependent Protein Translocation in HEK293 Cells-HEK293 cells grown on glass coverslips coated with 0.5% gelatin in 6-cm dishes were transfected with pEGFP-Munc13-1, pEGFP-Munc13-1 I121N , pEGFP-ubMunc13-2, pEGFP-ubMunc13-2 I121N , or pCMV5-RIM1␣, either using Lipofectamine 2000 (Invitrogen) or the calcium phosphate coprecipitation method (45). Cells were fixed with 4% paraformaldehyde in phosphate buffer 24 h after transfection. The Munc13 fusion proteins were visualized directly by the EGFP signal, and RIM1␣ was detected using indirect immunofluorescence with a monoclonal anti-RIM1␣ antibody and a secondary Alexa 594-conjugated antimouse IgG antibody (Molecular Probes). For the analysis of phorbol ester-induced protein translocation, HEK293 cells were transfected with Munc13 and RIM1␣ expression constructs 2 days prior to the experiment, as described above. The medium was changed before the experiment, and 4␤-12-O-tetradecanoylphorbol-13-acetate (Sigma) was added from an acetone stock solution directly to the cell culture (100 nM final concentration). Cells were then incubated for 60 min at 37°C, washed twice with phosphate-buffered saline, and fixed with 4% paraformaldehyde in phosphate buffer. Munc13s and RIM1␣ were detected as described above. Coverslips were mounted onto slides with Mowiol (Calbiochem) and observed with an Axiovert 200 LSM 510 confocal laser scanning (Zeiss) microscope.
Western Blots-For Western blot sample preparation, cortices from adult mouse brains were removed and homogenized in 25 mM Hepes/KOH (pH 7.4), 100 mM NaCl, 1 mM EGTA (pH 8.0), 0.2 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, and 0.5 g/ml leupeptin. After determination of protein concentrations, samples were dissolved in SDS-PAGE sample buffer and analyzed according to standard procedures (46,47).
Immunohistochemistry-Animals were deeply anesthetized with tribromoethanol and perfused transcardially with 200 ml of NaCl solution (9 g/liter) containing sodium nitrite (1 g/liter) for 10 min. The brains were quickly and carefully removed and frozen in isopentane at Ϫ40°C. Sagittal 12-m frozen cryostat sections were collected and post-fixed with either cold 4% paraformaldehyde in phosphate-buffered saline (pH 7.4) or methanol (Ϫ20°C). The slices were blocked in phosphate-buffered saline containing 2% normal goat serum and 0.25% Triton X-100 for 30 min at room temperature. Thereafter, sections were incubated with polyclonal antibodies to Munc13-1 (1:10,000) (49) and monoclonal antibodies to synaptophysin (Synaptic Systems; 1:1,000) at 4°C for 16 h. The antibodies were diluted in phosphate-buffered saline that contained 5% normal goat serum and 0.1% Triton X-100. After three rinses in phosphate-buffered saline, the sections were incubated with Alexa 488-or Alexa 568-labeled goat anti-rabbit or anti-mouse IgG secondary antibodies (Molecular Probes) at room temperature. Coverslips were mounted using Mowiol (Calbiochem) and viewed using an Olympus BX61 microscope and AnalySIS imaging program (Olympus Corp.). Single plane images of cultured neurons were acquired with an Axiovert 200 LSM 510 confocal laser scanning microscope (Zeiss). Relative fluorescence intensity levels were analyzed with a web-based image processing and analysis program in Java, Image J 1.32.
Protein Analysis in RIM1 Mutant Mice-Munc13-1 and ubMunc13-2 protein levels were assessed by Western blotting using polyclonal antibodies to Munc13-1 (49) and ubMunc13-2. Levels of NMDAR1 and GDP dissociation inhibitor were determined as loading controls using monoclonal anti-NMDAR1 (50) and polyclonal rabbit anti-GDP dissociation inhibitor antibodies (provided by Dr. Reinhard Jahn, Göttingen). Secondary antibodies were horseradish peroxidasecoupled goat anti-rabbit antibodies (Bio-Rad) or Alexa Fluor 680-coupled goat anti-rabbit antibodies for Odyssey (Molecular Probes). The respective stainings were visualized by enhanced chemiluminescence (Amersham Biosciences) and quantified with the Odyssey imaging system (LI-COR).

Identification of RIM1␣ Binding-deficient Munc13-1 and ubMunc13-2 Mutants in Reverse Yeast Two-hybrid Assays-An
N-terminal truncated Munc13-1 variant that lacks its entire L region including the ␣RIM binding domain (Munc13-1(520 -1736)-EGFP) exhibits reduced priming activity in neurons (30). This observation indicates that ␣RIM binding regulates active zone trafficking or priming activity of Munc13-1 and most likely also of ubMunc13-2. However, the elimination of the entire L region of Munc13-1 may also have functional consequences that are not related to the ␣RIM/Munc13-1 interaction.
To develop more suitable molecular tools for the functional analysis of Munc13/␣RIM interactions, we used a reverse yeast two-hybrid approach for the identification of missense mutations in Munc13-1 and ubMunc13-2 that disrupt their interaction with ␣RIM while leaving other interactions unaffected. The N terminus of ubMunc13-2 (LexA-ubMunc13-2(1-181), which is highly homologous to the N-terminal region of Munc13-1 and contains the minimal ␣RIM binding region (Fig. 1, A and B), was used as a model protein in this screen in combination with VP16-RIM1␣(1-344).
To analyze the effect of the I121N mutation on ␣RIM binding in the context of living cells, eukaryotic expression vectors encoding full-length recombinant versions of wild-type and mutant Munc13-1 with a C-terminal attached EGFP tag were generated in pEGFP-N1 (Munc13-1-EGFP and Munc13-1 I121N -EGFP). These EGFP-tagged Munc13-1 variants and fulllength RIM1␣ were then overexpressed in HEK293 cells either alone or in combination. When expressed singly in HEK293 cells, Munc13-1-EGFP and Munc13-1 I121N -EGFP formed large aggregates in addition to a diffuse cytoplasmic pool of protein (Fig. 3). In contrast, overexpressed RIM1␣ was found in the nucleus as well as diffusely distributed in the cytoplasm (Fig. 3). When Munc13-1-EGFP and RIM1␣ were coexpressed, Munc13-1 again formed large aggregates and RIM1␣ was invariably coenriched with Munc13-1 in these aggregates (Fig.  3A). On the other hand, when Munc13-1 I121N -EGFP and RIM1␣ were coexpressed in HEK293 cells, RIM1␣ was never recruited into Munc13-1 I121N -EGFP aggregates. Rather, the cellular RIM1␣ distribution remained diffuse and partially nuclear as was the case when it was expressed in the absence of Munc13-1-EGFP (Fig. 3B). Due to the presence of a C 1 domain, Munc13-1 as well as Munc13-2 and Munc13-3 bind diacylglycerol and phorbol esters and translocate to the plasma membrane of overexpressing fibroblasts and chromaffin cells in response to phorbol ester binding (51,52). We took advantage of this phorbol ester-dependent translocation of Munc13-1 and tested if the binding to Munc13-1 causes the cotranslocation of RIM1␣ in coexpressing cells. We found that Munc13-1-EGFP and RIM1␣ cotranslocated to the plasma membrane of coexpressing HEK293 cells upon treatment with 100 nM 4␤-12-Otetradecanolyphorbol-13-acetate (Fig. 3C). In contrast, RIM1␣ expressed alone or in combination with Munc13-1 I121N -EGFP did not translocate to the plasma membrane in response to phorbol ester treatment (Fig. 3D). Taken together, our coexpression studies in HEK293 cells show that Munc13-1 binds RIM1␣ in a cellular environment and that the I121N mutation is sufficient to abolish the Munc13-1/␣RIM interaction, without altering phorbol ester sensitivity and translocation properties of the Munc13-1 protein. Given the high sequence homology and very similar RIM1␣ binding characteristics of Munc13-1 and ubMunc13-2 ( Figs. 1 and 2), ubMunc13-2 is likely to behave identically.
Analysis of the Subcellular Distribution of Munc13-1 and ubMunc13-2 in RIM1␣ Knock-out Mice-To further analyze the role of ␣RIMs in the presynaptic targeting and retention of Munc13, RIM1␣ knock-out mice (22) were analyzed. Previously published studies on these mutants showed that Munc13-1 expression levels are strongly reduced in the absence of RIM1␣ (22). We quantified the levels of Munc13-1 and both splice variants of Munc13-2, i.e. bMunc13-2, which does not bind ␣RIMs, and ubMunc13-2, which has been shown to bind RIM1␣ (30,39), in wild-type and littermate RIM1␣-deficient adult mice using the levels of NMDAR1 as a loading control. We found that the expression levels of both, Munc13-1 and ubMunc13-2, to be reduced by ϳ60% in RIM1␣ knock-out animals (Fig. 5), whereas the levels of bMunc13-2 were not altered (data not shown). Lentivirus-infected Munc13-1/2 double deficient hippocampal neurons expressing the indicated ubMunc13-2 constructs were immunostained at days in vitro 18 for Bassoon (red) to mark presynaptic active zones, and analyzed by fluorescence microscopy. A, ubMunc13-2-EYFP expressed in Munc13-1/2 double deficient neurons was concentrated at presynaptic structures and colocalized with Bassoon. B, ubMunc13-2 I121N -EYFP expressed in Munc13-1/2 double deficient neurons was diffusely distributed along the axon and did not colocalize with Bassoon. Data are from representative neurons. In total, 10 or more neurons per condition were analyzed, always yielding identical results. Scale bars, 10 m (low magnification, large images); 3 m (high magnification, small images). C, quantitative analysis of presynaptic targeting of ubMunc13-2-EYFP and ubMunc13-2 I121N -EYFP. In transfected neurons whose axons could be traced for longer distances under the microscope, the degree of colocalization of the respective overexpressed ubMunc13-1-EYFP variant with the active zone marker Bassoon was determined. The bar diagram represents the average fraction (in %) of Bassoon containing active zones along axons of individual transfected neurons that also contained ubMunc13-2-EYFP or ubMunc13-2 I121N -EYFP (n ϭ 3 neurons each for ubMunc13-2-EYFP and ubMunc13-2 I121N -EYFP transfections). Error bars represent S.E. The asterisk indicates a statistically significant difference (p Ͻ 0.05) from the wild-type control. FIGURE 5. Analysis of Munc13-1 and ubMunc13-2 protein levels in brains of RIM1␣ deletion mutant mice. A, representative immunoblot analysis of the levels of Munc13-1, ubMunc13-2, and NMDAR1 in brains from littermate wild-type and homozygous RIM1␣ deletion mutant mice. The levels of NMDAR1 were probed on the same blot as loading control. B, quantification of the levels of ubMunc13-2 and Munc13-1 in brains from littermate wild-type and RIM1␣-deficient mice. Proteins were quantified by immunoblotting with isoform-specific primary and Alexa Infra Red-labeled secondary antibodies and the Odyssey detection system. All signals were normalized using the levels of NMDAR1, which were probed on the same blot, as loading control. Error bars represent S.E. (n ϭ 4 mice for each genotype).
We determined whether the Munc13-1 and ubMunc13-2 fractions that remain in the absence of RIM1␣ are distributed abnormally. For that purpose, we performed subcellular fractionation experiments on cerebral cortex homogenates from wild-type and littermate RIM1␣-deficient mice and examined the distribution of Munc13-1 and ubMunc13-2 in the various subcellular fractions obtained. In agreement with previously published data (29), we found Munc13-1 to be enriched in synaptosomal (P2) and synaptic membrane fractions (LP1), but also in soluble fractions (S1, S3, and LS1). A similar subcellular distribution was observed for ub-Munc13-2 (Fig. 6A). This subcellular distribution of Munc13-1 and ubMunc13-2 was not altered in RIM1␣-deficient brains, with the notable exception of ubMunc13-2 levels in fraction S1, which arises after removal of crude synaptosomes from total homogenates. The levels of ubMunc13-2 in this fraction appeared to be increased in the case of RIM1␣-deficient brains as compared with wild-type controls (Fig.  6B), indicating that ubMunc13-2 is less efficiently anchored within the presynaptic protein network in the absence of RIM1␣.
To analyze the effect of RIM1␣ deficiency on the subcellular distribution of Munc13-1 and ub-Munc13-2 in more detail, soluble and membrane-bound protein fractions were separated by ultra-thorax homogenization and high-speed centrifugation. The levels of membrane-bound and soluble Munc13 proteins were determined by quantitative Western blotting using NMDAR1 and GDP dissociation inhibitor levels as loading controls, and the ratio between insoluble and soluble protein levels were calculated. We found that in the absence of RIM1␣, the relative amounts of soluble Munc13-1 and ubMunc13-2 are significantly increased as compared with the wild-type situation (Fig. 6C). Taken together, our analyses of Munc13 protein levels in subcellular fractions from wild-type FIGURE 6. Analysis of Munc13-1 and ubMunc13-2 proteins in subcellular fractions of cortex homogenates from RIM1␣ deletion mutant mice. Blots of subcellular fractions of cortex homogenates from littermate wild-type and RIM1␣ deletion mutant mice were analyzed with antibodies against Munc13-1, ubMunc13-2, and NMDAR1. A, subcellular distribution of Munc13-1, ubMunc13-2, and NMDAR1 in subcellular fractions of cortex homogenates from wild-type mice. B, subcellular distribution of Munc13-1, ubMunc13-2, and NMDAR1 in subcellular fractions of cortex homogenates from RIM1␣ deletion mutant mice. Subcellular fractions were designated as follows: H, homogenate; P1, nuclear pellet; S1, supernatant after synaptosome sedimentation; P3, light membrane pellet; S3, cytosolic fraction; P2, crude synaptosomal pellet; LP1, lysed synaptosomal membranes; LS1, supernatant after LP1 sedimentation. Data in panels A and B are representative of at least three independent experiments with identical results. C, quantification of soluble and insoluble fractions of ubMunc13-2 and Munc13-1 in brain homogenates from littermate wild-type and RIM1␣-deficient mice. The levels of ubMunc13-2 and Munc13-1 were quantified by immunoblotting with isoform-specific primary and Alexa Infra Red-labeled secondary antibodies and the Odyssey detection system. All signals were normalized using the levels of NMDAR1 (membrane-bound fractions) and GDP dissociation inhibitor (soluble fractions), which were probed on the same blots, as loading control. Values are expressed as % of wild-type levels (n ϭ 4 mice for each genotype). Error bars represent S.E.
␣RIMs Recruit Munc13-1 and ubMunc13-2 to Active Zones and RIM1␣-deficient brains show that both RIM1␣ binding partners, Munc13-1 and ubMunc13-2, are less efficiently anchored to synaptic membranes and, presumably as a consequence, exhibit a higher turnover rate and lower total expression levels in the absence of RIM1␣. These findings further support the notion that binding of Munc13-1 and ubMunc13-2 to ␣RIMs is critical for their proper recruitment to and anchoring at active zones.
Distribution of Munc13-1 in the Hippocampus of Wild-type and RIM1␣ Knock-out Mice-So far, all our experimental data indicated that ␣RIMs are involved in the presynaptic targeting and/or anchoring of Munc13-1 and ubMunc13-2. To further analyze such a role of ␣RIMs in the intact brain, we studied the distribution of Munc13-1 in wild-type and RIM1␣-deficient hippocampus by immunofluorescence staining. Similar experiments on ubMunc13-2 were not possible because of the lack of suitable antibodies. In wild-type hippocampus, Munc13-1 was found to be strongly enriched in the granule cell mossy fiber terminals, which terminate within the stratum lucidum of the CA3 region, and much less abundant in the surrounding neuropil and cell body layers (Fig. 7A). In contrast, hippocampal sections from RIM1␣-deficient brains showed no evidence of a relative accumulation of Munc13-1 in mossy fiber terminals of the stratum lucidum (Fig. 7B), although the presynaptic marker synaptophysin was distributed normally. This misdistribution of Munc13-1 in RIM1␣-deficient hippocampus was most apparent upon analysis of the Munc13-1 immunofluorescence intensity across the neuropil and cell body layers of the CA3 region. In wild-type sections, Munc13-1 immunofluorescence intensity increased sharply at the interface between the cell body layer and stratum lucidum and decreased sharply at the interface between stratum lucidum and the adjacent neuropil layer of the stratum radiatum. In RIM1␣-deficient hippocampal sections, on the other hand, Munc13-1 immunofluorescence intensity increased only moderately at the interface between the cell body layer and stratum lucidum and remained constant at the interface between stratum lucidum and stratum radiatum. We quantified the degree of enrichment of Munc13-1 and synaptophysin in mossy fiber terminals of wildtype and RIM1␣-deficient hippocampal sections by determining the Munc13-1 and synaptophysin immunofluorescence intensity in the stratum lucidum and stratum radiatum of individual sections and calculating the percent fluorescence intensity increase from stratum radiatum to stratum lucidum (Fig.  7C). The average immunofluorescence intensity increase for synaptophysin in stratum lucidum over stratum radiatum was similar for wild-type and RIM1␣-deficient hippocampal sections (13 Ϯ 6%, wild-type, 20 Ϯ 8%, RIM1␣-deficient, n ϭ 4), which indicates that synaptophysin is enriched in mossy fiber terminals irrespective of the presence of RIM1␣. In contrast, an immunofluorescence intensity increase for Munc13-1 in stratum lucidum over stratum radiatum was only detectable in wild-type hippocampal sections (27 Ϯ 6%, n ϭ 4) but not in RIM1␣-deficient sections (1 Ϯ 4%, n ϭ 4), indicating that mossy fiber terminal enrichment of Munc13-1 depends on the presence of RIM1␣. Taken together, these findings provide in vivo evidence for a role of ␣RIMs in presynaptic targeting and/or anchoring of Munc13-1 at active zones. Left panels, hippocampal sections were double immunostained for Munc13-1 (green) and synaptophysin 1 (red ) and viewed using an Olympus BX61 microscope. Right panels, relative fluorescence intensity levels. Data are representative of four independent experiments that yielded identical results (see C). Arrows indicate the transition between the respective strata of the hippocampal CA3 region. Scale bars, 80 m; or, stratum oriens; py, stratum pyramidale; lu, stratum lucidum; rad, stratum radiatum. C, quantitative analysis of Munc13-1 and synaptophysin immunofluorescence intensity in stratum lucidum (lu) as compared with stratum radiatum (rad), as a measure for Munc13-1 and synaptophysin enrichment in mossy fiber terminals. For each section and each antigen, the average fluorescence intensity was calculated for the entire stratum lucidum (lu) and for the first 20 m of the adjacent stratum radiatum (rad ), using fluorescence intensity scans like the ones shown in A and B. The bar diagram expresses the mean percent increase of immunofluorescence intensity in stratum lucidum (lud ) as compared with the adjacent stratum radiatum (rad, set to 100%) (n ϭ 4 mice per genotype and antigen). Error bars represent S.D. The asterisk indicates a statistically significant difference ( p Ͻ 0.05) from the wild-type control.

to Active Zones
In contrast to the effects of RIM1␣ deficiency on the expression and localization of Munc13-1 and ubMunc13-2, the expression level (32) and distribution (Fig. 8) of RIM1␣ are not altered in the absence of Munc13-1 and ubMunc13-2. Rather RIM1␣ and Bassoon were found to be perfectly colocalized in both, wild-type and Munc13-1/2 double deficient hippocampal neurons (Fig. 8).

DISCUSSION
Munc13-like proteins contain two modules that have evolved differentially. The mammalian Unc-13 homologs, Munc13-1, bMunc13-2, ubMunc13-2, and Munc13-3, and the corresponding C. elegans N-terminal alternative splice variants, Unc-13-MR and Unc-13-LR are very similar in their C-terminal R-region, which contains a tandem C 1 /C 2 domain, two Munc13 homology domains and a C-terminal C 2 domain (30). The Unc-13-LR splice variant contains an N terminus that is highly homologous to that of Munc13-1 and ubMunc13-2, whereas the N-terminal regions of bMunc13-2, Munc13-3, and Unc-13-MR are completely unrelated. These differentially evolved Munc13 modules have distinct functional roles. The conserved C-terminal R-region is necessary and sufficient for secretory vesicle priming (30,37,38,53,54). The N-terminal ␣RIM binding L-region of Munc13-1 and ubMunc13-2, on the other hand, was shown to convey a synapse-specific function that is not directly involved in the priming reaction but influences it indirectly (30,39). ␣RIM binding was postulated to either mediate active zone recruitment or anchoring of Munc13-1 and ubMunc13-2 or to cause conformational changes in these Munc13s that affect their priming activity.
The present data indicate that RIM1␣ is not the only active zone anchoring protein for Munc13-1 and ubMunc13-2. The Munc13-1 proteins that remain in RIM1␣-deficient brains are still mostly localized in the synaptic neuropil (Fig. 7) and large fractions of both, Munc13-1 and ubMunc13-2, remain associated with membrane fractions in the absence of RIM1␣ (Fig. 6), indicating that RIM2␣ or other active zone components contribute to the anchoring of Munc13-1 and ubMunc13-2 to the active zone protein network. However, RIM1␣ appears to be the main active zone recruitment protein for Munc13-1 and ubMunc13-2 because overall Munc13-1 and ubMunc13-2 levels are reduced by 60% in RIM1␣-deficient brains (Fig. 5) and the synaptic recruitment of the ␣RIM binding-deficient ubMunc13-2 I121N -EYFP is severely perturbed (Fig. 4). The reduced expression levels of Munc13-1 and ubMunc13-2 in RIM1␣-deficient brains are most likely due to increased turnover and degradation of the proteins in the cytosol as compared with the active zone-associated fraction.
All currently available functional data on the Munc13/␣RIM interaction are compatible with a role of ␣RIMs in the accumulation of Munc13-1 and ubMunc13-2 at active zones. In hippocampal neurons (30) and in the calyx of Held (39), selective disruption of the Munc13/␣RIM interaction leads to decreased synaptic vesicle priming. These effects can be explained by inefficient active zone recruitment of the essential priming proteins Munc13-1 and ubMunc13-2 upon perturbation of their interaction with ␣RIMs.
Apart from their relevance for our understanding of the regulation of basal synaptic transmission, the present data are also of importance with regard to the interpretation of the role of FIGURE 8. Analysis of the synaptic targeting of RIM1␣ in Munc13-1/2 double deficient neurons. Hippocampal neurons prepared from a control mouse (A) and a Munc13-1/2 double deficient mouse (B) were fixed at 18 days in vitro and immunostained with antibodies to RIM1␣ (green) and Bassoon (red ). In both genotypes, RIM1␣ is colocalized with Bassoon at active zones. Data are from representative neurons. In total, 10 or more neurons per condition were analyzed, always yielding identical results. Scale bars, 10 m (low magnification, large images); 2 m (high magnification, small images).
␣RIMs Recruit Munc13-1 and ubMunc13-2 to Active Zones ␣RIMs in mossy fiber LTP. ␣RIMs contain binding sites for Rab3s and Munc13s (39), and both, RIM1␣ and Rab3A, are essential for mossy fiber LTP (23,55). Given the central and essential role of Munc13s in determining the transmitter release efficacy of synapses, it is possible that Munc13-1 and ubMunc13-2 are involved in the final execution of Rab3/␣RIMmediated mossy fiber LTP. At first glance, this hypothesis is contradicted by the finding that the heterozygous Munc13-1 deletion mutant mice, whose Munc13-1 are reduced by about 50%, show normal mossy fiber LTP, whereas in RIM1␣ deletion mutants, in which Munc13-1 levels are reduced by 60%, mossy fiber LTP is abolished (23). However, the absence of RIM1␣ does not only cause reduced Munc13-1 expression but also a comparable reduction of ubMunc13-2 levels (Fig. 5) and a misdistribution of Munc13-1 with strongly impaired synaptic accumulation in mossy fiber terminals (Fig. 7). Thus, the loss of RIM1␣ is likely to have more deleterious effects on the Munc13 levels in mossy fiber terminals than the elimination of a single Munc13-1 allele, where active zone recruitment of Munc13s is not perturbed. The deficiency in Munc13 accumulation at mossy fiber terminals in RIM1␣ deletion mutant mice may then well be one cause of the observed mossy fiber LTP deficiency.
Clearly, additional studies are needed to determine as to whether Munc13-1 and ubMunc13-2 are really effectors of Rab3/␣RIM-dependent mossy fiber LTP. For example, future experiments will have to determine whether phosphorylation of RIM1␣, which is essential for mossy fiber LTP (24), regulates the Rab3/RIM1␣/Munc13 interaction and increases the levels or activities of Munc13-1 and ubMunc13-2 in mossy fiber terminals. Ultimately, mouse genetic experiments involving knock-in mutants that express ␣RIM binding-deficient Munc13-1 and ubMunc13-2 or Munc13 binding-deficient RIM1␣ instead of the respective wild-type proteins are necessary to determine the functional significance of Munc13-␣RIM complexes in mossy fiber LTP. In this regard, the present study makes an important contribution. Given that only limited structural information on the Munc13-1-␣RIM complex is currently available (39), we had to resort to random mutagenesis and reverse yeast two-hybrid screening to identify subtle mutations in Munc13-1 and ubMunc13-2 that interfere with ␣RIM binding (Fig. 1). This approach led to the identification of a point mutation, I121N, that depending on the assay used, strongly perturbs or abolishes ␣RIM binding of Munc13-1 and ubMunc13-2 (Figs. 1-3) without affecting their expression or phorbol ester sensitivity (Fig. 3), indicating that the mutation does not interfere with overall protein folding. Knock-in mutant mice expressing Munc13-1 I121N and ubMunc13-2 I121N instead of the corresponding wild-type proteins could serve as useful tools for the dissection of the molecular processes that mediate mossy fiber LTP.