The establishment of neuronal circuitry during brain development requires the formation of synapses between neurons, and an essential part of this is presynaptic axon differentiation. Synapse development is regulated by the ubiquitin-proteasome pathway (1
). However, an E3 ubiquitin ligase that orchestrates presynaptic differentiation in the mammalian brain remained to be identified.
The cell cycle-regulated ubiquitin ligase, the anaphase-promoting complex (APC), is highly expressed in postmitotic mammalian neurons (fig. S1A
). The APC associates with the critical coactivator Cdc20 or Cdh1 (3
). Although Cdh1-APC controls axon growth and patterning (3
), it does not appear to regulate the number of presynaptic sites in mammalian brain neurons (fig. S13
). We thus asked whether Cdc20-APC plays a role in presynaptic differentiation.
To study presynaptic development, we first characterized clustering of the synaptic vesicle protein synapsin in primary granule neurons isolated from the rat cerebellar cortex. To visualize synapsin clustering, we transfected granule neurons with an expression plasmid encoding synapsin fused to green fluorescent protein (GFP-synapsin) (5
). We observed GFP-synapsin clusters preferentially in the distal portion of axons, and these clusters overlapped with the endogenous synaptic vesicle proteins VAMP2 and VGlut1 (; fig. S1B–D
Cdc20-APC induces presynaptic differentiation in postmitotic neurons
To determine the role of Cdc20-APC in synapsin clustering, we induced knockdown of Cdc20 in neurons using two independent shRNAs (fig. S2A
). Synapsin clustering was robustly reduced in Cdc20 knockdown neurons (; fig. S2B
). Expression of a rescue form of Cdc20 encoded by an RNAi-resistant cDNA (Cdc20-RES), but not Cdc20 encoded by wild type cDNA (Cdc20-WT), reversed the Cdc20 RNAi-induced loss of synapsin clusters in neurons, suggesting that the Cdc20 RNAi-induced phenotype is the result of specific knockdown of Cdc20 in neurons (; fig. S2C
). Cdc20 knockdown in cerebellar slices also reduced synapsin cluster density in granule neuron parallel fiber axons (fig. S3
). In other experiments, knockdown of the core APC subunit APC2 in granule neurons reduced the density of synapsin clusters (; fig. S2, A and B
). Together, these results suggest that Cdc20-APC drives synapsin clustering in postmitotic neurons.
We asked whether Cdc20-APC regulates other aspects of presynaptic differentiation. GFP-synapsin clusters co-localized with clusters of endogenous active zone proteins, including Munc13, Bassoon, Rim1, and Liprin-alpha (fig. S4A
). Accordingly, knockdown of Cdc20 triggered the loss of GFP-Munc13 clusters along the axon in granule neurons (; fig. S4B
). Thus, Cdc20-APC promotes the coordinate clustering of synaptic vesicle and active zone proteins.
We next assessed whether Cdc20-APC promotes the differentiation of functional presynaptic sites that form synapses. Synapsin clusters were apposed to clusters of the postsynaptic protein PSD-95, and Cdc20 knockdown reduced the density of synapsin/PSD-95 co-clusters in granule neurons (). In other experiments in which we measured the ability of axons to undergo synaptic vesicle recycling, synapsin clusters were found to be co-localized with sites of uptake of the dye FM4-64 or a luminal Syt1 antibody (6
), and Cdc20 knockdown reduced the density of synapsin co-clusters with FM4-64 or Syt1 uptake sites (). Collectively, our results suggest that Cdc20-APC drives the differentiation of presynaptic sites that are functionally active and that form synapses.
As the ubiquitin ligase activity of Cdc20-APC is critical for its function, we reasoned that neuronal Cdc20-APC promotes presynaptic differentiation by inducing the degradation of a protein that inhibits presynaptic development. The brain-enriched transcription factor NeuroD2 harbors the Cdc20 recognition motif, the destruction box (D-box) (8
), suggesting that NeuroD2 might be a target of Cdc20-APC in the control of presynaptic development. NeuroD2 protein levels decreased in the developing rat cerebellum and in primary granule neurons with maturation, inversely correlating with the increasing levels of Cdc20 (; fig. S5, A and B
). Endogenous NeuroD2 levels increased in granule neurons treated with the proteasome inhibitor MG132 (). In other experiments, NeuroD2 was found to be conjugated with ubiquitin in neurons (). Together, these results suggest that NeuroD2 is regulated by the ubiquitin-proteasome system.
Cdc20-APC-induced degradation of NeuroD2 drives presynaptic differentiation
We assessed the role of Cdc20-APC in the regulation of NeuroD2 protein abundance in neurons. NeuroD2 formed a physical complex with Cdc20 in cells, but a NeuroD2 protein in which the D-box was mutated (ND2-DBM) failed to associate with Cdc20 (; fig. S5C
). Importantly, endogenous NeuroD2 levels increased upon Cdc20 knockdown in neurons (). Thus, Cdc20-APC controls the abundance of NeuroD2 protein in neurons.
The identification of NeuroD2 as a target of Cdc20-APC in neurons would predict that NeuroD2 suppresses synapsin clustering. Consistent with this prediction, NeuroD2 knockdown increased the density of synapsin clusters in granule neurons by over two fold (; fig. S5, D and E
). Expression of an RNAi-resistant rescue form of NeuroD2 (NeuroD2-RES) but not NeuroD2-WT reversed the NeuroD2 RNAi-induced increase in synapsin cluster density in neurons (; fig. S5F
). NeuroD2 knockdown also increased synapsin cluster density in parallel fiber axons in cerebellar slices (fig. S6A
). In other experiments, NeuroD2 knockdown in granule neurons increased by more than two fold the density of synapsin/FM4-64 and synapsin/Syt1 co-clusters (; fig. S6B–D
). Further, the NeuroD2 knockdown-induced synapsin clusters were apposed to PSD-95 (; fig. S6E
). NeuroD2 knockdown also increased clustering of the active zone protein Munc13 in granule neurons (; fig. S6F
). Thus, NeuroD2 knockdown increases the density of functionally active sites of presynaptic differentiation.
In gain-of-function analyses, expression of the mutant NeuroD2 protein ND2-DBM in neurons reduced the density of presynaptic sites (; fig. S7A–C
). In epistasis analyses, NeuroD2 knockdown suppressed the effect of Cdc20 knockdown on presynaptic differentiation (; fig. S7, D and E
). These results suggest that Cdc20-APC promotes the formation of functional presynaptic axonal sites via NeuroD2 degradation in neurons.
Among the reported NeuroD2 targets is the gene encoding Complexin II (Cplx2), a protein that modulates synaptic vesicle fusion to the presynaptic membrane in invertebrates (9
). We asked if Cplx2 might mediate NeuroD2-dependent control of presynaptic differentiation. Knockdown of Cplx2 robustly increased the density of synapsin clusters in granule neurons, and this phenotype was reversed by expression of Cplx2 encoded by an RNAi-resistant cDNA (Cplx2-RES) but not wild type cDNA (Cplx2-WT) (; fig. S8, A and B
). Cplx2 knockdown also increased the density of functionally active presynaptic axonal sites in granule neurons (; fig. S8C–F
). Finally, Cplx2 knockdown increased synapsin cluster density in granule neuron parallel fiber axons in the cerebellar cortex in slices (fig. S8G
). Thus, Cplx2 knockdown phenocopies the effect of NeuroD2 knockdown on presynaptic differentiation.
The NeuroD2 target gene, Complexin II, inhibits presynaptic differentiation
In epistasis analyses, expression of exogenous Cplx2 in granule neurons reduced the density of functionally active presynaptic sites induced by NeuroD2 RNAi (; fig. S9, A and B
). Further, although knockdown of NeuroD2 or Cplx2 each stimulated presynaptic differentiation, the combination of NeuroD2 and Cplx2 knockdown together did not additively increase the density of presynaptic sites (; fig. S9, C and D
). In other experiments, knockdown of Cplx2 reversed Cdc20 RNAi-induced suppression of presynaptic differentiation (; fig. S9, E and F
). Collectively, these results suggest that Cplx2 mediates the ability of NeuroD2 to suppress presynaptic differentiation and acts as a downstream component of the Cdc20-APC pathway in the control of presynaptic development.
We next determined the role of the Cdc20-APC ubiquitin signaling pathway in presynaptic differentiation in the developing organism. To visualize the morphogenesis of presynaptic axonal differentiation in vivo, we expressed GFP in the cerebellar cortex in rat pups using an electroporation method (3
). These analyses revealed granule neuron cell bodies in the internal granule layer (IGL) and their parallel fiber axons in the molecular layer (ML) (). Parallel fiber axons displayed varicosities along the axon (), which co-localized with punctate synapsin immunoreactivity and were apposed to punctate PSD-95, a marker of postsynaptic structures (). These observations indicate that parallel fiber axon varicosities represent sites of presynaptic axonal differentiation in vivo.
The Cdc20-APC ubiquitin signaling pathway regulates presynaptic axonal differentiation in vivo
Using electroporation, we next induced knockdown of Cdc20, NeuroD2, or Cplx2 in rat pups. Cdc20 knockdown reduced the density of presynaptic parallel fiber varicosities in the cerebellar cortex in vivo. In contrast, knockdown of NeuroD2 or Cplx2 in rat pups increased the density of presynaptic varicosities in the cerebellar cortex (; fig. S10
). These results suggest that the Cdc20-APC ubiquitin signaling pathway cell-autonomously regulates presynaptic axonal differentiation in the developing brain in vivo.
We have discovered a Cdc20-APC ubiquitin ligase signaling pathway that orchestrates presynaptic development and hence the establishment of neuronal circuitry in the brain (see model in fig. S11
). APC function in cell cycle control has been the subject of intense scrutiny (8
). While Cdh1-APC operates in late mitosis and G1 phase of the cell cycle, Cdc20-APC controls the transition of the cell cycle through early mitosis (8
). In the control of axon morphogenesis in postmitotic neurons, Cdh1-APC appears to control axon growth and patterning (3
), while Cdc20-APC operates at a later developmental stage to promote presynaptic axonal differentiation. Thus, in an analogous manner to the temporal control of the cell cycle, the APC in concert with its two different coactivators, Cdh1 and Cdc20, appears to govern distinct temporal phases of axon differentiation in postmitotic neurons in the brain.