|Home | About | Journals | Submit | Contact Us | Français|
The ubiquitin ligase anaphase-promoting complex (APC) recruits the coactivator Cdc20 to drive mitosis in cycling cells. However, the nonmitotic functions of Cdc20-APC have remained unexplored. We report that Cdc20-APC plays an essential role in dendrite morphogenesis in postmitotic neurons. Knockdown of Cdc20 in cerebellar slices and in postnatal rats in vivo profoundly impairs the formation of granule neuron dendrite arbors in the cerebellar cortex. Remarkably, Cdc20 is enriched at the centrosome in neurons, and the centrosomal localization is critical for Cdc20-dependent dendrite development. We also find that the centrosome-associated protein histone deacetylase 6 (HDAC6) promotes the polyubiquitination of Cdc20, stimulates the activity of centrosomal Cdc20-APC, and drives the differentiation of dendrites. These findings define a novel postmitotic function for Cdc20-APC in the morphogenesis of dendrites in the mammalian brain. The identification of a centrosomal Cdc20-APC ubiquitin signaling pathway holds important implications for diverse biological processes including neuronal connectivity and plasticity.
The proper development and patterning of dendrites is essential for the establishment of neuronal connectivity. Dendrites represent the critical receptive limb of neurons, and accordingly the morphology of the dendritic arbor influences the processing of synaptic information within a neural circuit (Parrish et al., 2007). Perturbations in dendrite morphology are thought to play critical roles in the pathogenesis of diverse neurological disorders including Down’s, Fragile X, and Rett syndromes as well as adult neurodegenerative diseases (Kaufmann and Moser, 2000; Knobloch and Mansuy, 2008). Therefore, elucidation of the molecular mechanisms governing dendrite differentiation not only deepens our understanding of neuronal circuitry but also potentially provides insight into diseases that affect the human brain.
Neurons are postmitotic cells that exit the cell cycle permanently. However, components of the cell cycle machinery are expressed in neurons (Becker and Bonni, 2005; Greene et al., 2004). In proliferating cells, protein ubiquitination plays an essential role in coordinating the events of the cell cycle. The anaphase-promoting complex (APC) is an evolutionarily conserved, multisubunit ubiquitin ligase that is critical for normal cell cycle transitions (King et al., 1995; Zachariae et al., 1996). The APC governs both mitotic progression and G1 maintenance via the binding of APC coactivators, Cdc20 and Cdh1, thereby allowing the timely degradation of cell cycle substrates (Peters, 2006). Remarkably, the core APC subunits are highly expressed in the mammalian brain, and Cdh1-APC operates in the nucleus of postmitotic neurons to regulate axon growth (Konishi et al., 2004; Lasorella et al., 2006; Stegmuller et al., 2006). However, the function of Cdc20-APC outside of the cell cycle in postmitotic cells has remained unexplored.
We report that Cdc20-APC is required for dendrite morphogenesis in mammalian neurons. Cdc20 is enriched at centrosomes in neurons, and this subcellular locus is critical for the ability of Cdc20-APC to drive dendrite development. We also forge an intimate biochemical and functional link between the centrosome-associated protein histone deacetylase 6 (HDAC6) and Cdc20-APC in neurons. Finally, we identify the centrosomally localized protein Id1 as a substrate of Cdc20-APC in neurons and demonstrate that Cdc20-APC-induced degradation of Id1 plays a critical role in dendrite development. These findings identify Cdc20-APC as the first ubiquitin ligase that specifies dendrite morphogenesis in the mammalian brain.
To investigate the role of Cdc20-APC in postmitotic neurons, we characterized the function of the key regulatory subunit Cdc20 in granule neurons of the rat cerebellar cortex. Granule neurons represent an ideal model for the study of neuronal morphogenesis and connectivity in the brain (Altman, 1972; Hatten and Heintz, 1995). Using four different antibodies, we found that Cdc20 protein is expressed in primary granule neurons (Figures 1A and S1A). Cdc20 protein levels increased with maturation, and introduction of small interfering RNAs targeting Cdc20 in granule neurons confirmed the specificity of the commercial antibody used throughout this study (Figures 1A and S1B). Finally, in situ hybridization of mouse cerebellar sections at P14 and P21 revealed Cdc20 mRNA expression in the internal granule layer (IGL) and Purkinje cell layer (Figure S1C). Together, these data indicate that Cdc20 is expressed in postmitotic neurons of the developing rodent cerebellar cortex. The time course of Cdc20 expression in primary neurons also suggested a possible role for Cdc20 in dendrite morphogenesis.
To interrogate Cdc20 function in neurons, we used a plasmid-based method of RNA interference (RNAi) to knockdown Cdc20 expression acutely. Expression of short hairpin RNAs (shRNAs) targeting two distinct regions of Cdc20 efficiently reduced the levels of exogenous and endogenous Cdc20 in heterologous cells and granule neurons, respectively (Figures 1B and S1D). Cdc20 knockdown in granule neurons led to a striking dendritic phenotype characterized by short primary dendrites (Figure 1C). Total dendrite length was reduced by approximately 65% relative to control transfected neurons (Figure 1D). Cdc20 knockdown did not affect cell survival, ruling out cell death as a cause of the Cdc20 RNAi-induced phenotype (Figure 1E). In addition, Cdc20 RNAi had little or no effect on axon length (Figure 1F). These results suggest that Cdc20 RNAi impairs the growth and development of dendrites.
We next performed a rescue experiment using an expression plasmid encoding an RNAi-resistant form of Cdc20 (Cdc20-RES) (Figure 1G). Expression of Cdc20 encoded by wild type cDNA had little effect on the Cdc20 RNAi-induced dendritic phenotype (Figures 1H and 1I). In contrast, expression of Cdc20-RES in the background of Cdc20 RNAi restored dendrite arbor formation, increasing total dendrite length to 82% that of control transfected neurons (Figures 1H and 1I). These data indicate that the Cdc20 RNAi-induced loss of dendrites is the specific result of Cdc20 knockdown.
To determine if Cdc20’s dendrite-promoting function is generalized in postmitotic neurons, we assessed the role of Cdc20 in cerebral cortical and hippocampal neurons. As with granule neurons, Cdc20 protein increased with maturation in cerebral cortical and hippocampal neurons (Figures S2A and S2B). Remarkably, in both populations of neurons, Cdc20 RNAi simplified the dendritic arbor, reducing the number of primary branches as well as secondary and tertiary dendritic branches, leading to dramatically reduced total dendrite length (Figures S2C-S2F). These results suggest that Cdc20 represents a generalized cell-intrinsic mechanism driving dendrite morphogenesis in mammalian neurons.
We next characterized the temporal dynamics of Cdc20 function in the development of dendrites. In cohort analyses of granule neurons transfected at a time prior to the generation of dendrites, Cdc20 RNAi flattened the dendrite growth curve (Figure S3A). In analyses in which neurons were transfected at a time when they already bear dendrites, Cdc20 knockdown triggered progressive shortening of dendrites (Figure S3B). Thus, Cdc20 promotes both the growth and maintenance of dendrites.
Time-lapse analyses of individual granule neurons revealed that dendrites display dynamic elongation and retraction during 48 hours of analysis in control neurons, which cumulatively led to increased total dendrite length (Figures S3C-S3G). However, dendrites in Cdc20 knockdown neurons often displayed retraction with little evidence of elongation, which cumulatively led to decreased total dendrite length (Figures S3C-S3G). Analyses of individual dendrites revealed similar results (Figures S3C and S3H-S3K). Collectively, these data demonstrate that Cdc20 supports the growth and maintenance of dendrites by promoting dendrite elongation and preventing dendrite retraction.
Having demonstrated a dendritogenic function for Cdc20 in primary neurons, we next examined if Cdc20 is required for dendrite morphogenesis in the context of intact developing brain tissue. Using a biolistics method to trigger RNAi in rat organotypic cerebellar slices, we found that Cdc20 knockdown markedly inhibited the growth of primary dendrites and impaired dendritic arbor elaboration in granule neurons, leading to substantially reduced total dendrite length (Figures 2A and 2B). To determine the function of Cdc20 in the cerebellar cortex in vivo, we electroporated P3 rat pups with a Cdc20 RNAi plasmid that also encodes GFP (U6-cdc20i/CMV-GFP) or the corresponding control plasmid (U6/CMV-GFP). Animals were sacrificed at P8, and cerebella were subjected to immunohistochemical analyses using the GFP antibody. GFP-positive granule neurons were easily identifiable by their somal size and characteristic “T”-shaped parallel fiber (Figure 2C, left). Knockdown of Cdc20 by two distinct shRNAs in rat pups caused a marked dendritic deficit, characterized by shortened primary dendrites and a reduction in dendritic arbor complexity in IGL granule neurons (Figure 2C, right). Quantification of dendrites in IGL granule neurons revealed that Cdc20 RNAi substantially decreased total dendrite length to 40% that of granule neurons in control transfected animals (Figure 2D). Cdc20 knockdown had little or no effect on the health of granule neurons, parallel fiber patterning, or the number of parallel fibers associated with IGL granule neurons (Figure 2E and data not shown). Together, these findings suggest that Cdc20 is required for dendrite development in the cerebellar cortex in vivo.
We next examined whether Cdc20’s function in neurons is APC-dependent by assessing the role of the core APC subunit APC2 in dendrite morphogenesis (Peters, 2006). APC2 RNAi in granule neurons phenocopied the dendrite deficiency observed with Cdc20 RNAi (Figures 3A and S4A). As reported, Cdh1 knockdown did not affect dendrite growth (Figure 3A) (Konishi et al., 2004). These data suggest that Cdc20, but not Cdh1, functions in concert with the APC ubiquitin ligase to promote dendritic differentiation.
To gain further insight into the mechanism of Cdc20-APC function in dendrite development, we ascertained the subcellular location of Cdc20-APC in neurons. Subcellular fractionation of granule neurons revealed that Cdc20 resides predominantly in the cytoplasm, where a fraction of the APC core subunit Cdc27 is also present (Figure 3B). Immunofluorescence analyses confirmed that Cdc20 is expressed in granule neurons by co-staining for the neuron-specific class III β-tubulin Tuj1 (Figure 3C, right). Cdc20 exhibited diffuse, low intensity expression in the cytosol with marked concentration in a single perinuclear punctum, which was knocked down by Cdc20 RNAi, indicating that the punctate signal represents Cdc20-specific immunoreactivity (Figure 3C and S1D). Strikingly, Cdc20 tightly co-localized with the centrosomal markers γtubulin and pericentrin (Figures 3D and 3F). In contrast, neither an endoplasmic reticulum (protein disulphide isomerase (PDI)) nor a Golgi marker (GM130) co-localized with Cdc20 (Figure 3D). In corroborating biochemical analyses, centrosomal fractions isolated from granule neurons revealed that Cdc20 cofractionates with the centrosomal proteins γtubulin and 14-3-3ε (Figure 3E) (Pietromonaco et al., 1996). These data demonstrate that Cdc20 is enriched specifically at the centrosome in neurons.
We next evaluated the function of centrosomal Cdc20 in neurons. To localize Cdc20 at the centrosome, a sequence encoding 92 amino acids of the PACT domain from AKAP450 was inserted N-terminal to the RNAi-resistant Cdc20-RES cDNA (PACT-Cdc20-RES) (Gillingham and Munro, 2000). Since a pool of Cdc20 was observed in the nucleus of neurons (Figure 3B), a nuclearly targeted Cdc20-RES expression plasmid was also generated (NLS-Cdc20-RES). We confirmed the proper nuclear and centrosomal localization of NLS-Cdc20-RES and PACT-Cdc20-RES, respectively, in cells and neurons (Figures S5A and 3F). Nuclear Cdc20-RES failed to rescue the Cdc20 RNAi dendrite phenotype in granule neurons in primary dissociated and organotypic cerebellar cultures (Figures 3G, 3H, and S6A). In contrast, centrosomal Cdc20-RES completely reversed the dendritic impairment induced by Cdc20 RNAi in both settings (Figures 3G, 3H, and S6A). Neither the PACT domain alone nor a PACT domain-Cdh1 fusion protein restored dendrite length in the background of Cdc20 RNAi (Figure 3G), indicating that centrosomal Cdc20, and not the PACT domain itself, promotes dendritic morphogenesis. Together, these data suggest that localizing neuronal Cdc20-APC to the centrosome drives dendrite morphogenesis.
We next determined if the centrosomal localization of Cdc20 is required for its ability to promote dendrite development. Deletion of the N-terminal domain of Cdc20 (amino acids 1–167) disrupted the centrosomal concentration of Cdc20 in neurons, consistent with results in nonneural cells (Figures 4A and 4B) (Kallio et al., 2002). Further analysis of the N-terminal domain of Cdc20 revealed a requirement for a 57 amino acid region (111–167) in concentrating Cdc20 at the centrosome (Figures 4A, 4B, and S5B). A Cdc20-RES mutant in which the 57 amino acid region is deleted (Cdc20Δ111-167-RES) failed to reverse the Cdc20 RNAi-induced dendrite loss (Figures 4C and 4D). Fusion of the PACT domain to Cdc20Δ111-167-RES restored the ability of the mutant Cdc20-RES to stimulate dendrite development in the background of Cdc20 RNAi, suggesting that this 57 amino acid region of Cdc20 harbors a centrosomal localization signal (CLS) (Figures 4E and S5C). Finer deletions of the 57 amino acid region revealed a requirement for a short 19 amino acid motif (111–129) for both the centrosomal localization of Cdc20 and Cdc20-induced dendrite elaboration (Figures 4F-4H, S5D, and S5E). Fusion of the PACT domain to Cdc20Δ111-129-RES restored the ability of the mutant Cdc20-RES to stimulate dendrite development, suggesting that the function of amino acids 111–129 is to provide a CLS (data not shown). Collectively, our results establish that centrosomal localization of Cdc20 is essential for the ability of Cdc20 to promote dendrite morphogenesis.
We next investigated how Cdc20-APC is controlled at the centrosome in neurons. The class IIb histone deacetylase HDAC6 localizes to the basal body, a centrosome-derived organelle, and may regulate primary ciliary morphogenesis (Pugacheva et al., 2007). In addition to its two N-terminal deacetylase domains, HDAC6 harbors a C-terminal zinc finger ubiquitin-binding domain, ZnF-UBP (Seigneurin-Berny et al., 2001). Because of its localization, involvement in ubiquitin signaling, and function in cellular morphogenesis, we asked whether HDAC6 plays a role in the control of Cdc20-APC function in dendrite development.
We first assessed whether HDAC6 interacts with Cdc20 in cells. HDAC6 associated with Cdc20 in transfected cells (Figure 5A). Biochemical and immunofluorescence studies verified that HDAC6 resides in neuronal centrosomes (Figure 5B and data not shown). Accordingly, HDAC6 interacted with Cdc20 endogenously in neurons (Figure 5C). In other analyses, we found that the C-terminal region of Cdc20 (amino acids 311–499) interacted with HDAC6 (Figure S7A). A Cdc20-RES mutant lacking the HDAC6-interacting region (Cdc20-1-310-RES) failed to restore dendrite length in the background of Cdc20 RNAi, suggesting that the physical interaction of Cdc20 with HDAC6 is critical for Cdc20-dependent dendrite development (Figures S7A and S7B).
To determine if HDAC6 influences Cdc20-APC activity, we devised a cell-based degradation assay to monitor changes in centrosomal Cdc20-APC activity. We designed an expression plasmid encoding the first 88 amino acids of the mitotic Cdc20-APC substrate securin fused to firefly luciferase. This segment of securin contains the destruction box (D-box), an APC degron (Zur and Brandeis, 2001). We used the PACT domain to localize the degradation reporter to the centrosome (Figure 5D). A mutation in the securin D-box was introduced into the reporter to serve as a control. Cdc20 knockdown caused an increase in relative levels of the securin-luciferase reporter in neurons, demonstrating that this assay represents a valid readout for centrosomal Cdc20-APC activity (Figure 5D). Importantly, HDAC6 RNAi in granule neurons also increased the relative abundance of the securin-luciferase reporter (Figure 5D). In control experiments, expression of HDAC6 shRNA induced knockdown of exogenous as well as endogenous HDAC6 in cells (Figure 5E). These data suggest that endogenous HDAC6 promotes Cdc20-APC activity at the centrosome in neurons.
We next determined whether endogenous HDAC6 is relevant to dendrite morphogenesis. HDAC6 knockdown using two different shRNAs in granule neurons markedly reduced dendrite elaboration and decreased total dendrite length (Figure 5F and data not shown). HDAC6 RNAi also disrupted the formation of the granule neuron dendritic arbor in the cerebellar cortex in slices and in rat pups in vivo (Figures S6B and 5G). Further, knockdown of HDAC6 in hippocampal neurons decreased total dendrite length and reduced dendritic branching, suggesting that HDAC6 has a generalized dendritogenic function in mammalian neurons (Figures S8A and S8B). Together, these data indicate that HDAC6 knockdown phenocopies the effect of Cdc20 RNAi in neurons. In epistasis experiments in granule neurons, we found that overexpression of HDAC6 stimulated dendrite elaboration, an effect suppressed by Cdc20 RNAi, suggesting that HDAC6 promotes dendrite elaboration in a Cdc20-dependent manner (Figure 5H). Collectively, our findings suggest that HDAC6 promotes the activity of centrosomal Cdc20-APC and thereby drives dendrite morphogenesis.
We next determined the molecular basis underlying the ability of HDAC6 to stimulate neuronal Cdc20-APC activity and function. HDAC6 knockdown did not appear to alter the centrosomal localization of Cdc20 or centrosomal structure in neurons (Figures S9A and S9B). Because HDAC6 harbors deacetylase activity, we examined the effect of HDAC6 on the acetylation status of Cdc20 but detected no appreciable change in Cdc20 acetylation upon manipulation of HDAC6 levels (data not shown). Recent evidence suggests that autoubiquitination of Cdc20 leads to increased Cdc20-APC activity in cycling cells (Reddy et al., 2007; Stegmeier et al., 2007). In addition, HDAC6 may stabilize polyubiquitinated proteins via its C-terminal ZnF-UBP domain (Boyault et al., 2006). These observations raised the question of whether HDAC6, through its ZnF-UBP domain, might promote the polyubiquitination of Cdc20, thereby activating Cdc20-APC.
Expression of wild type HDAC6 robustly increased the polyubiquitination of co-expressed Cdc20 in cells (Figure 6A). Likewise, expression of an HDAC6 mutant deficient in deacetylase activity (HDAC6(DM)) increased Cdc20 ubiquitination (Figure 6A). Importantly, an HDAC6 mutant lacking the ZnF-UBP domain (HDAC(ΔC)) did not augment Cdc20 ubiquitination, suggesting that the ZnF-UBP domain but not the deacetylase activity of HDAC6 is required to increase Cdc20 polyubiquitination (Figure 6A). In in vitro ubiquitination experiments, addition of full-length HDAC6 but not HDAC6(ΔC) markedly promoted Cdc20 ubiquitination (Figure S10A). However, the HDAC6 ZnF-UBP domain was dispensable for Cdc20 binding in vitro (Figure S10B). These results suggest that HDAC6 directly binds Cdc20 and subsequently stimulates Cdc20-APC activity via ZnF-UBP domain-dependent stabilization of Cdc20 polyubiquitination. In agreement with the results in heterologous cells and in vitro, we found that the ubiquitination of neuronal Cdc20 was robustly reduced by HDAC6 knockdown (Figure 6B).
The requirement of the HDAC6 ZnF-UBP domain in promoting Cdc20 polyubiquitination would predict that the ZnF-UBP domain is critical for the ability of HDAC6 to promote dendrite development. To test this prediction, we performed structure-function analyses of an RNAi-resistant form of HDAC6 (HDAC6-RES) in the background of HDAC6 RNAi in neurons (Figure 6C). Expression of HDAC6-RES or deacetylase-deficient HDAC6-RES effectively reversed the HDAC6 RNAi-induced dendrite deficit (Figures 6D and 6E). However, an HDAC6-RES mutant lacking the ZnF-UBP domain failed to rescue the HDAC6 RNAi-induced dendrite phenotype, suggesting that HDAC6 promotes the elaboration of dendrites through the ZnF-UBP domain (Figures 6D and 6E). Together, our results suggest that HDAC6 via its ZnF-UBP domain promotes Cdc20 polyubiquitination, stimulates Cdc20-APC activity, and drives dendrite morphogenesis.
Recent studies have employed the Cdc20-specific deubiquitinase USP44 to demonstrate that Cdc20 polyubiquitination stimulates Cdc20-APC activity in dividing cells (Stegmeier et al., 2007). The USP44 gene is expressed in neurons, including granule neurons of the rodent cerebellar cortex (http://mouse.brain-map.org and data not shown). Knockdown of USP44 in neurons dramatically increased the polyubiquitination of Cdc20 and centrosomal Cdc20-APC activity, confirming that endogenous USP44 deubiquitinates Cdc20 in neurons (Figures 6F, 6G, and 6I). USP44 knockdown substantially increased dendrite length and branching in cerebellar granule neurons and hippocampal neurons, suggesting that Cdc20 deubiquitination restrains dendrite development (Figures 6H, S8C and S8D). In epistasis experiments, Cdc20 RNAi suppressed the USP44 RNAi-induced increase in dendrite length, suggesting that USP44 knockdown stimulates dendrite development in a Cdc20-dependent manner (Figures 6J). Together, these results suggest that the polyubiquitination of endogenous Cdc20 plays a critical role in Cdc20-dependent dendrite morphogenesis.
A major question that remained to be addressed is how centrosomal Cdc20-APC promotes dendrite development. We reasoned that Cdc20-APC might stimulate the ubiquitination and consequent degradation of a centrosomally localized protein that inhibits dendrite growth. Interrogation of the literature for proteins that are localized at the centrosome and harbor a conserved Cdc20-recognition D-box motif revealed that the helix-loop-helix protein Id1 fulfills both criteria (Figure S11A, top) (Hasskarl et al., 2004). We confirmed that Id1 resides at the centrosome in granule neurons (Figures 7A and 7B). Exposure of neurons to the proteasome inhibitor MG132 substantially increased Id1 levels, suggesting that Id1 is regulated by the ubiquitin-proteasome system in neurons (Figure 7C). Expression of Id1 declined in granule neurons with maturation, inversely correlating with increasing Cdc20 protein expression, suggesting that Cdc20 might regulate Id1 abundance in neurons (Figure 7D). Consistent with this possibility, Cdc20 and Id1 formed a physical complex in a D-box-dependent manner in cells (Figure S11A, bottom). Cdc20-APC directly catalyzed the ubiquitination of Id1 in vitro, suggesting that Id1 is a substrate of Cdc20-APC (Figure 7E). Importantly, Cdc20 knockdown triggered the accumulation of endogenous Id1 protein in neurons (Figure 7F). Collectively, these results identify the centrosomally localized protein Id1 as a novel substrate of neuronal Cdc20-APC.
We next characterized the function of Id1 in neurons. Id1 knockdown using two different shRNAs markedly increased dendrite length in granule neurons (Figures 7G, S11B, and S11C). Importantly, Id1 RNAi-induced dendrite growth was specifically reversed by expression of an RNAi-resistant form of Id1 (Figures 7H, S11D, and S11E). We also found that Id1 knockdown in the cerebellar cortex in vivo led to more elaborate dendritic arbors with increased total dendrite length in IGL granule neurons (Figure 7I). Knockdown of Id1 in primary hippocampal neurons also stimulated dendrite growth and branching, suggesting a generalized inhibitory role for Id1 in dendrite morphogenesis in mammalian neurons (Figures S8E and S8F).
In epistasis experiments, the combination of Id1 RNAi and Cdc20 RNAi led to increased total dendrite length that was indistinguishable from that of Id1 RNAi alone, suggesting that Id1 lies functionally downstream of Cdc20 (Figure 7J). In gain-of-function analyses of Id1, expression of a mutant Id1 protein in which the D-box motif was mutated substantially reduced dendrite length (Figure S11F). These results suggest that Cdc20-APC-induced degradation of the centrosomal protein Id1 plays a critical role in promoting dendrite growth and branching. Collectively, we have elucidated a novel centrosomal Cdc20-APC ubiquitin ligase pathway that governs dendrite morphogenesis in the mammalian brain (see model in Figure 7K).
In this study, we have discovered a centrosomal Cdc20-APC ubiquitin signaling pathway that plays a fundamental role in dendrite morphogenesis in the mammalian brain. The ubiquitin ligase Cdc20-APC has received wide attention for its essential function in driving the metaphase-anaphase transition in cycling cells (Peters, 2006). Our study reveals a novel postmitotic function for this mitotic ubiquitin ligase and suggests that the mitotic machinery may play widespread roles in unique aspects of neural development.
The dendrite-promoting effect of Cdc20-APC provides an interesting counterpoint to the axon-specific regulatory effect of Cdh1-APC (Konishi et al., 2004). Whereas Cdh1-APC functions in the nucleus in neurons to control axon growth and patterning (Konishi et al., 2004; Lasorella et al., 2006; Stegmuller et al., 2006), Cdc20-APC operates at the centrosome to drive dendrite morphogenesis through the promotion of both dendrite growth and maintenance. Thus, the APC may act as a pivotal regulator of neuronal morphogenesis through specific recruitment of the coactivators Cdc20 and Cdh1 in distinct subcellular compartments.
Elucidation of the centrosomal HDAC6/Cdc20-APC signaling cassette implies that both HDAC6 and Cdc20-APC may exert novel functions beyond the control of dendrite morphogenesis. HDAC6 localized to the centrosome-derived basal body has been implicated in the control of primary cilium stability in epithelial cells (Pugacheva et al., 2007). Although the role of the ZnF-UBP domain of HDAC6 in ciliary biology remains undefined, identification of the intimate centrosomal HDAC6/Cdc20-APC link raises the possibility that centrosomal Cdc20-APC also operates outside the cell cycle in epithelial cells to control the morphology of the primary cilium. This scenario raises intriguing parallels between the biology of primary cilia in epithelial cells and dendrite development in neurons.
The identification of Cdc20-APC as the first E3 ubiquitin ligase that specifically regulates the growth and elaboration of dendrites in mammalian neurons has important ramifications for our understanding of the cell-intrinsic control of neuronal morphology and circuitry. Genetic studies in fly and mammalian neurons have highlighted the importance of transcription factors in diverse aspects of the cell-intrinsic control of dendrite morphogenesis, including growth, branching, and cell type-specific arborization (Parrish et al., 2007). In future studies, it will be important to determine how the activities of centrosomal Cdc20-APC and transcription factors are integrated in the specification and elaboration of dendrites. Another line of inquiry should address how Cdc20-APC controls the effects of extrinsic cues on dendrite morphogenesis in the developing brain. Beyond dendrite growth and elaboration, it will be interesting to determine if the centrosomal Cdc20-APC pathway might contribute to dendritic refinement and plasticity.
The identification of Id1 as a substrate of neuronal Cdc20-APC provides important clues as to how Cdc20-APC regulates dendrite morphogenesis. Since Id1 localizes at the centrosome but does not likely represent an integral component of the centrosome, our findings suggest that the centrosome acts as a signaling platform for the Cdc20-APC ubiquitin ligase pathway in the control of dendrite development. In agreement with this conclusion, Cdc20 knockdown does not appear to disrupt the structure of the centrosome as assessed by expression of the centrosomal proteins pericentrin and γtubulin (Figure S12). How the Id1 signal from its centrosomal locale is coupled to the regulation of dendrites is an important question for future studies. Whether additional centrosomal substrates of neuronal Cdc20-APC regulate dendrite morphogenesis also remains to be addressed.
During brain development, centrosomes have been demonstrated to function in neuronal migration as well as axonal specification and growth (Higginbotham and Gleeson, 2007). Our finding that Cdc20-APC operates from a centrosomal locus to specify the growth and elaboration of dendrites highlights a connection between centrosomal biology and dendrite morphogenesis. The centrosomal proteins LIS1 and DISC1, implicated in lissencephaly and schizophrenia respectively, play roles in dendritic development (Duan et al., 2007; Liu et al., 2000). Thus, the centrosome in postmitotic neurons may represent a critical signaling platform for dendritogenic pathways, and the deregulation of these pathways may contribute to the pathogenesis of neurological and psychiatric disorders.
Primary cerebellar granule neurons were prepared from P6 rat pups and transfected using a modified calcium phosphate method as described (Konishi et al., 2004). Granule neurons are postmitotic at the time of transfection as they fail to incorporate bromo-deoxyuridine (Konishi and Bonni, 2003). To avoid the possibility that the morphological effects of RNAi or protein expression were due to any effect of these manipulations on cell survival, a plasmid encoding the anti-apoptotic protein Bcl-XL was included in all neuronal transfections except those in which survival was assessed. Expression of Bcl-XL had little or no effect on dendritic or axonal morphology as reported (Gaudilliere et al., 2004; Konishi et al., 2004). In additional experiments, granule neurons were transfected with the Cdc20, HDAC6, USP44, Id1, or control U6 RNAi plasmid along with a GFP expression plasmid in the absence of Bcl-XL (Figure S13). In all cases, the dendrite phenotypes observed in the presence of Bcl-XL paralleled exactly the dendrite effects observed without Bcl-XL (Figure S13).
Hippocampal and cerebral cortical neuron cultures were prepared from rat embryos (embryonic day 18) as described (Goslin et al., 1998). Neurons were plated on poly-L-lysine-coated in Neurobasal medium plus B-27 (Invitrogen). Transfections were performed on DIV 1.
P6-P9 rat cerebella were dissected for organotypic slice cultures and transfected as described (Gaudilliere et al., 2004; Shalizi et al., 2006). In vivo electroporation of P3 Sprague-Dawley rat pups was performed as described (Konishi et al., 2004).
For visualization of centrosomal proteins, cells were fixed in methanol for 10 min at −20° and subjected to immunofluorescence analysis with the indicated antibodies using standard protocols. For other immunocytochemistry experiments, cells were fixed in 4% paraformaldehyde for 20 minutes at room temperature.
Ubiquitination of in vitro translated substrate by Cdc20-APC was performed as described (Konishi et al., 2004; Rape and Kirschner, 2004). Lysates of HeLa cells arrested in early mitosis were prepared. After addition of (1:1) APC lysis buffer (1% Triton X-100, 150 mM NaCl, 50 mM Tris pH 7.2, and 2mM EDTA, and 1 mM DTT plus protease inhibitor cocktail and NaF), the APC was immunoprecipitated using a mouse Cdc27 antibody (Santa Cruz) followed by protein G-sepharose. Immunoprecipitates were washed once with the APC lysis buffer followed by two washes with hypotonic lysis buffer (SB). In vitro ubiquitination was performed by incubation of APC and eluted FLAG-Cdc20 with 35S-Met-labeled, in vitro translated myc-Id1 for 1 hr at 37 °C in a reaction mix consisting of 5 mM Tris pH 7.5, 1 mM MgCl2, 2 mM ATP, 0.1 mM EGTA, 30 U/ml creatine phosphokinase, 7.5 mM phosphocreatine, and 2 μM ubiquitin aldehyde together with 12.5 μg ubiquitin, 400 ng ubiquitin-activating enzyme (E1), and 500 ng UbcH10 (Boston Biochem). Reaction products were resolved by SDS-PAGE followed by nitrocellulose transfer. Radioactivity was visualized by Phosphorimager analysis.
The axonal and dendritic morphology of granule neurons in culture, in cerebellar slices, and in vivo were analyzed in a blinded fashion using a Nikon Eclipse TE2000 epifluorescence microscope to capture images. SPOT software was used to measure process length.
Analyses were performed using Statview 5.0.1 software. All histogram data are presented as mean + SEM unless otherwise specified. The t-test was used for comparisons in experiments with only two groups. In experiments with more than two groups, analysis of variance (ANOVA) was performed followed by Fisher’s Least Significant Difference or the Bonferroni test for pairwise comparisons among three and greater than three groups, respectively. For analysis of more than two groups of non-parametric data, the Kruskal-Wallis test was used.
Supported by NIH grants NS051255 and NS41021 (AB), a Ruth L. Kirschstein NRSA research fellowship, NCI, and Brain Science Foundation grant (AHK), the Albert J Ryan Foundation (PB), and a Human Frontier Science Program grant (YI). We thank members of the Bonni laboratory for helpful discussions and critical reading of the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.