Synaptic Assembly of the Brain in the Absence of Neurotransmitter Secretion

See allHide authors and affiliations

Science  04 Feb 2000:
Vol. 287, Issue 5454, pp. 864-869
DOI: 10.1126/science.287.5454.864


Brain function requires precisely orchestrated connectivity between neurons. Establishment of these connections is believed to require signals secreted from outgrowing axons, followed by synapse formation between selected neurons. Deletion of a single protein, Munc18-1, in mice leads to a complete loss of neurotransmitter secretion from synaptic vesicles throughout development. However, this does not prevent normal brain assembly, including formation of layered structures, fiber pathways, and morphologically defined synapses. After assembly is completed, neurons undergo apoptosis, leading to widespread neurodegeneration. Thus, synaptic connectivity does not depend on neurotransmitter secretion, but its maintenance does. Neurotransmitter secretion probably functions to validate already established synaptic connections.

Synapses are focal points of communication between nerve cells. Together, billions of synapses account for the unique connectivity of the brain (1). To establish the synaptic network, outgrowing axons are precisely directed to their targets using a variety of guidance cues and recognition signals on the outgrowing axon and the target cell (2). Fusion of neurotransmitter vesicles at the axon tip is believed to supply the membrane for axonal outgrowth (3). The concomitant neurotransmitter release is thought to have a trophic role and to provide essential signals for the correct targeting of axons and synapse formation: Components of the presynaptic secretion machinery are already expressed in immature neurons before they differentiate (4). These components are targeted to the axon tip, where they are thought to be essential for axonal outgrowth in vitro (3, 5). Outgrowing neurons have an active synaptic vesicle cycle, secrete neurotransmitters before synapse formation, and up-regulate this secretion once the growth cone comes close to its target (6). After synapses have formed, the secretion capacity of nerve terminals is up-regulated, and the Ca2+affinity and tetanus toxin sensitivity increase.

Many genes have been identified that function in neurotransmitter secretion at mature synapses (7), and drastic phenotypes have been observed upon deleting such genes in mice (8). However, all currently known gene deletions do not impair normal brain development. This can be explained by the fact that only certain aspects of the presynaptic function are abolished in these mutants. In particular, spontaneous, quantal transmitter release is retained even in the most severely affected mutants (9). We have now identified the munc18-1 gene as an essential gene for all components of neurotransmitter release throughout the brain. Munc18-1/nSec1 (10) is a neuron-specific protein of the SEC1-family of membrane-trafficking proteins. Munc18-1 is expressed throughout the brain and interacts with at least three classes of proteins, which suggests that it may regulate cell polarization as well as focal secretion at synapses (10, 11). We abolishedmunc18-1 expression in mice by homologous recombination (Fig. 1) (12). This resulted in a completely paralyzed organism. Null mutant embryos are alive until birth, but die immediately after birth, probably because they cannot breathe (13).

Figure 1

Generation of mice lackingmunc18-1. (A) Gene structure of wild-typemunc18-1 (wt) and after homologous recombination (null). Boxes indicate exons with encoding amino acid numbers above. Five exons are replaced with a neomycin resistance gene (NEOR). Arrow pairs indicate PCR primers for genotyping. The top arrow indicates the Northern probe, and the grey boxes indicate Southern probes. H, Hind III; E, Eco RI; B, Bam HI; K, Kpn I; and S, Sma I. (B) Southern blot ofmunc18-1 mutants. Eco RI–digested genomic DNA from wild types (+/+), heterozygotes (+/−), and homozygotes (−/−) was hybridized with probe 1. Two independent mouse lines (called 18 and 23 after the ES cell clone) contained identical deletions. (C) RNA blot of munc18-1 mRNA in wild types and homozygotes around birth. (D) Immunoblot analysis of munc18-1 at E18 in all genotypes. Blots were probed with a munc18-1–specific monoclonal antibody (left) or with a general munc18 polyclonal antibody (right) to detect possible compensatory changes in other munc18 isoforms. RBH, rat brain homogenate.

Synaptic transmission can normally be detected as soon as a synapse is formed. In the mammalian neocortex, the first synapses are observed at embryonic day 16 (E16) (14). Electron microscopy of the marginal zone in control embryos (13) confirmed the presence of a few synapses at E16 (15), although we did not detect synaptic secretion events in neocortical slices at E17 (16). However, at E18, synaptic events were readily observed (Fig. 2A). In contrast, null mutants lacked synaptic events (Fig. 2A). Nevertheless, in these mutants postsynaptic receptors were functional (Fig. 2B), spontaneous action potentials were occasionally observed (Fig. 2C), and ion channels showed normal properties (Fig. 2D). To study neurotransmitter secretion in a different, earlier developing synapse, we investigated the diaphragm neuromuscular junction between E15 and E18 (16). In these synapses too, synaptic events were detected in controls, but null mutants lacked synaptic events at E15, E16, and E18 (Fig. 2E). Application of α-latrotoxin or electrical nerve stimulation strongly stimulated synaptic transmission in control littermates but caused no response in null mutants, although their postsynaptic receptors were functional (Fig. 2, E and F).

Figure 2

munc18-1–deficient mice lack synaptic neurotransmitter release. (A) Whole-cell voltage clamp recordings from neocortical neurons in slices at E18. Wild-type neurons showed frequent spontaneous synaptic events (5.4 ± 2.4 min−1; mean ± SEM, n = three animals and four recordings), whereas neurons in null mutant slices were completely devoid of such activity (n = three animals and five recordings; total recording time, 30.8 min). (B) γ-Aminobutyric acid (GABA) iontophoresis (1 M, 200 to 400 nA, 10 to 20 pulses for 0.1 s at 4 Hz) in mutant brain slices caused a normal postsynaptic response, sensitive to bicuculline (20 μM in bath). (C) Spontaneous action potentials in the cell-attached mode. (D) Voltage-gated inward ion currents in whole-cell mode in mutant cells. (C) and (D) indicate that we recorded from neurons, not glial cells. (E) Intracellular recordings in diaphragm muscle fibers from control and null mutant littermates. Control embryos exhibited miniature endplate potentials (MEPPs): 5.4 ± 0.6 min−1 at E15 (one animal, nine fibers ± SEM); 1.3 ± 0.3 min−1 at E16 (two animals, five fibers per animal ± SEM); and 2.8 ± 0.6 min−1at E18 (two animals, five fibers per animal ± SEM). No MEPPs were detected in munc18-1–deficient mice at E15 (nine fibers; total recording time, 18.8 min), E16 (five fibers, 21.5 min recording), or E18 (11 fibers, 35.5 min recording). α-Latrotoxin (4 nM) induced massive neurotransmitter release in controls but was completely ineffective in mutant muscle fibers. (F) Carbachol (1 mM) elicited a strong postsynaptic potential and contraction in E18 mutant muscle fibers.

Despite the general, complete, and permanent loss of synaptic transmission in the knockout mice, their brains were assembled correctly (Fig. 3). Neuronal proliferation, migration, and differentiation into specific brain areas were unaffected. At E12, brains from null mutant and control littermates were morphologically indistinguishable (Fig. 3, A and B) (17). At birth, late-forming brain areas such as the neocortex appeared identical in null mutant and control littermates, including a distinctive segregation of neurons into cortical layers (Fig. 3, C and D). Furthermore, fiber pathways were targeted correctly in null mutants: The growth cone marker GAP-43 showed a normal distribution, including marked staining of fiber bundles and synaptic layers (Fig. 3, G and H). Immunolabeling patterns for presynaptic markers were similar in mutant and control littermates (Fig. 3, I and J). The levels of several synaptic proteins were also normal in the mutants, suggesting that neuronal differentiation and synthesis of synaptic components proceeded normally (18). Moreover, synapses were readily formed in null mutants, being already present in the neocortex at E16 at the onset of synaptogenesis (14). These synapses exhibited all the signs of synaptic complexes (Fig. 3I). The synapses in controls and knockouts contained the same numbers of total and docked synaptic vesicles (19). Other aspects of the ultrastructure were also indistinguishable between mutants and controls (20).

Figure 3

Correct assembly of the brain in the absence of neurotransmitter secretion. (A and B) Coronal sections of developing brains at E12 from control [(A) and (B)] and null mutant [(A′) and (B′)] littermates stained with hematoxylin and eosin. (A) is located anterior to (B). C, cortex; CB, cerebellar anlage; T, tectum; L and M, lateral and medial ganglionic eminence; B, brainstem. (C and D) Normal architecture of the neocortex in null mutant mice at E18. Scale bars, 100 μm in (C), 20 μm in (D). (E) Clustering of acetylcholine receptors in the diaphragm neuromuscular junction visualized with fluorescent α-bungarotoxin at E18. Scale bars, 15 μm. (F) Accumulation of acetylcholinesterase at synaptic sites in the diaphragm at E18. Scale bars, 250 μm. (G and H) GAP-43 staining of the developing cortex at E18 (H) and the giant fiber pathway above the pontine nucleus (PN). M, marginal zone (synaptic layer); F, fimbria fornix fiber bundle. Scale bars, 200 μm in (G), 20 μm in (H). (I andJ) Synapsin I (I) and synaptobrevin/VAMP II (J) staining at E18, showing that synaptic vesicle markers are transported to the synaptic layer (M). Scale bars, 200 μm. (K) Synapses in the neocortex marginal zone at E16 showing apparently normal synaptic structures in the null mutants (K′) with clustered and docked synaptic vesicles near or at presynaptic active zones and postsynaptic densities. Scale bars, 50 nm.

Formation of neuromuscular synapses requires precise navigation of nerve terminals over large distances. At the neuromuscular junction, nerve terminals secrete signals such as agrin that induce clustering of postsynaptic acetylcholine receptors and acetylcholinesterase (21, 22). Thus, assembly of receptor and enzyme clusters provides a sensitive measure of pathfinding and synaptogenesis. α-Bungarotoxin and acetylcholinesterase staining of diaphragm muscles revealed clear receptor and enzyme clusters in null mutant mice (Fig. 3, E and F) (23), which suggests that long-range axonal pathfinding and initial synapse formation must have occurred.

After initial brain assembly, extensive cell death of mature neurons was observed in the null mutants (Fig. 4), occurring first in lower brain areas that mature and form synapses relatively early (Fig. 4, D through F). For example, at E12 the brainstems of null mutant and control littermates were morphologically similar (Fig. 3B). Although this area expanded further in controls, the neurons disappeared in null mutants until, at E18, the lower brainstem was almost completely lost. Degeneration occurred later in the midbrain and basal forebrain (Fig. 4, A through C). Brain areas that develop last, especially the neocortex, were indistinguishable from those of controls at birth. The degeneration in the mutant brains exhibited all characteristics of apoptosis: Apoptotic bodies were readily observed with standard histology (Fig. 5, A and B). These were positive in TUNEL staining (Fig. 5, C and D) (24). Apoptotic bodies contained condensed chromatin (Fig. 5G). A final phase of these degenerations was accompanied by abundant staining for activated macrophages (Fig. 5, E, F, and H) (25) and disintegration of lower brain areas around birth (Fig. 4).

Figure 4

Massive neurodegeneration after assembly of the neuronal networks. Coronal brain sections of control [(A) through (F)] and null mutant [(A′) through (F′)] littermates between E14 and birth. The asterisks identify locations where degeneration starts. Scale bars, 1 mm. B, brainstem; H, hypothalamus; P, putamen; T, thalamus; Te, tectum. In (A) to (C), C indicates cortex; in (D) to (F), C indicates cerebellar anlage.

Figure 5

Neurons undergo massive apoptosis after initial synaptogenesis. (A and B) Cell death exhibiting dark indented nuclei in the thalamus of null mutant mice at E18 stained with hematoxylin and eosin (A) or methylene blue (B) is shown. Scale bars, 20 μm. (C and D) TUNEL staining of the null mutant hippocampus. Scale bars, 200 μm in (C), 20 μm in (D). (E, F, and H) Activated macrophage staining in the hippocampus and thalamus at E18 with F4/80 antibody (25). In controls, activated macrophages were largely restricted to the ventricles (E). In the null mutants, dense staining was observed throughout the tissue and colocalized with apoptotic cell bodies (F and H) that were probably engulfed by macrophages (H). H, hippocampus; FIM, fimbria; Th, thalamus. Scale bars, 200 μm in (E) and (F), 20 μm in (H). (G) Electron micrograph of an affected neuron in the null mutant thalamus, showing compacted chromatin inside the nucleus. Scale bar, 3 μm.

Thus, a gene deletion that abolishes presynaptic secretion and therewith synaptic transmission allows an apparently normal initial assembly of the brain. This is quite unlike gene deletions for developmental factors, which generally result in some defect in the assembly of the brain (agenesis) (26). Ablation ofmunc18-1 renders the brain synaptically silent, identifyingmunc18-1 as the currently most upstream essential protein in neurotransmitter release. Axons in the munc18-1 mutants extend normally and form precise connections between widely separated brain areas, and synapses develop that look morphologically normal. It has been proposed that early spontaneous activity of neurons results in Ca2+ transients and may be critical for proper differentiation and pathfinding (27). However, this spontaneous activity must act by a mechanism that does not require synaptic neurotransmitter release, becausemunc18-1–deficient neurons differentiate normally, exhibit spontaneous action potentials, and perform apparently normal pathfinding. After synaptic assembly of the brain, activity-dependent selection is thought to maintain certain synaptic connections for adult life, whereas others are discarded (1, 2). Our data indicate that the neuronal networks are synaptically assembled and reach the selection stage without synaptic transmission, but cannot persist without it. When synaptic transmission is absent in newly established synaptic connections, these synapses degenerate and neurons go into apoptosis. Finally, our data indicate that the distribution of membrane to the growing axon tip and the release of signals that allow correct axon targeting depend on different molecular mechanisms than neurotransmitter release by regulated exocytosis of synaptic vesicles.

  • * To whom correspondence should be addressed. E-mail: m.verhage{at} (M.V.); tsudho{at}

  • Present address: Center for Physiology and Pathophysiology, University of Göttingen, Germany.


Stay Connected to Science

Navigate This Article