α-COP binding to the survival motor neuron protein SMN is required for neuronal process outgrowth

Abstract

Spinal muscular atrophy (SMA), a heritable neurodegenerative disease, results from insufficient levels of the survival motor neuron (SMN) protein. α-COP binds to SMN, linking the COPI vesicular transport pathway to SMA. Reduced levels of α-COP restricted development of neuronal processes in NSC-34 cells and primary cortical neurons. Remarkably, heterologous expression of human α-COP restored normal neurite length and morphology in SMN-depleted NSC-34 cells in vitro and zebrafish motor neurons in vivo. We identified single amino acid mutants of α-COP that selectively abrogate SMN binding, retain COPI-mediated Golgi–ER trafficking functionality, but were unable to support neurite outgrowth in cellular and zebrafish models of SMA. Taken together, these demonstrate the functional role of COPI association with the SMN protein in neuronal development.

Introduction

Insufficient levels of the survival motor neuron (SMN) protein result in motor neuron degeneration and subsequent muscle weakness that is the hallmark of spinal muscular atrophy (SMA). There are two predominant hypotheses on how low levels of SMN cause SMA. SMN is essential for assembly of the snRNP spliceosome, and SMN-depletion results in anomalous RNA splicing (1–6). Based on observations of the SMN protein in motor neuron axons and dendrites, a second hypothesis is that ineffective axonal transport induces synaptic defects (7–13). Both scenarios may converge on the roles of SMN in the regulation and transport of specific post-transcriptionally processed mRNAs.

In severe mouse models of SMA, progressive weakness leads to death, implying that motor units rapidly become dysfunctional (7,14–18). Disease appears to be cell autonomous as depletion of SMN in muscle did not induce the SMA phenotype (19–22). Zebrafish smn mutants display defects in motoneuronal development and motor axon outgrowth (23), which are rescued by motoneuron autonomous expression of human SMN (24). Importantly neurite extension defects resulting from SMN-depletion are replicated in cell culture models, providing a reliable biological marker (9,25–27). These models implicate the axon and synapse as a point of failure for the SMA motor neuron. These structures depend heavily on successful intracellular trafficking for development, maintenance and survival, and a decrease in synaptic vesicles has been observed in SMA model mice (28). We reported that the α-COP subunit of the COPI vesicle (29,30) directly binds SMN and moves with SMN in neurites of cultured cells (31). COPI positive puncta have clearly been visualized in axons (31,32), COPI and other Golgi derived proteins are necessary for successful maturation of neural processes (33).

NSC-34 cells are a hybrid of motor neuron enriched spinal cord and mouse neuroblastoma (34). Depletion of SMN has been shown to result in decreased neurite outgrowth in NSC-34 cells (27,35,36). We recently established a NSC-34 cell clone that contains an inducible murine SMN shRNA to model the cellular pathology of SMA. Mutation of the α-COP binding dilysine motif found in exon 2b eliminated the ability of SMN to restore neurite outgrowth in these cells (35). To test this whether COPI and its interaction SMN is necessary for formation of normal neurites, here we describe the successful identification of point mutations of highly conserved amino acids of α-COP that eliminate SMN binding while retaining normal COPI-mediated Golgi–ER trafficking functionality. We then demonstrated SMN binding defective α-COP mutants were unable to support neurite outgrowth in both NSC-34 cell culture and zebrafish models of SMA. These observations led us to hypothesize that the COPI vesicle selects and transports the SMN protein and other associated cargoes necessary for development of the axon and/or dendrite complex, a process dependent on α-COP binding to SMN.

Results

α-COP is necessary for neurite formation in NSC-34 cells and mouse cortical neurons

Under low serum conditions, rodent NSC-34 cells project neurites and express many components of motor neurons (37,38). In our initial study, α-COP depletion resulted in short neurites in SH-SY5Y cells (31). To gain further insight into the role of α-COP in neurite outgrowth, we established NSC-34 cells stably expressing a murine-specific α-COP shRNA, which produced ∼85% reduction in α-COP protein expression as measured by western blot (Fig. 1A). When cultured under conditions that promote neuronal differentiation [DMEM/F12, 1% fetal bovine serum (FBS)], the α-COP-depleted NSC-34 cells were dramatically unable to form neurites even after 72 h, compared with control cells, which produced long primarily bipolar processes under these conditions (Fig. 1A). This was repeated with two additional shRNA to confirm that this was not merely an off-target effect (data not shown). Similar effects were observed with the same α-COP shRNA lentiviral infection of primary cortical murine neurons, in which both Map2 positive dendrites and Tau positive axons were significantly shorter in α-COP-depleted cells compared with those infected with a scrambled shRNA control virus (Fig. 1B).

Figure 1.

α-COP is required for neuronal process extension. (A) Comparison of wild-type α-COP-depleted NSC-34 cell morphology as seen with immunofluorescence for α-tubulin. Right panel demonstrates western blot levels of α-COP protein in the stable shRNA expressing cultures. Scale bar, 12.5 μm. (B) Primary cortical neuron cultures isolated from neonatal mice were infected with lentiviral α-COP shRNA on the day of plating and cultured for 5 days. Immunofluorescent staining of Map2 and Tau showed that α-COP knockdown significantly decreased both dendritic and axonal length (P < 0.001 by Students t-test). Scale bar, 50 μm.

The highly conserved C-terminus of α-COP mediates SMN binding

To understand the relevance of the α-COP interaction with SMN for neuronal biology, we sought to identify point mutations in α-COP that would selectively inactivate SMN binding while maintaining normal COPI vesicle function. Our original yeast two-hybrid screen and subsequent in vitro binding assays of the α-COP and SMN protein interaction revealed that the C-terminal half of the protein [amino acids (aa) 751–1224) was sufficient for SMN binding (31). To map a minimal domain for SMN binding, we constructed a series of progressive truncations of α-COP consisting of deletions beginning at amino acid 1224 in a background of FLAG-tagged α-COP amino acids 751–1224, and GST-tagged deletions where amino acids were sequentially removed beginning at 751 (Fig. 2A). Immunoprecipitations of the α-COP truncations were carried out using antibodies to the FLAG or GST tags followed by western blot analysis. Unexpectedly, all of the α-COP deletion mutants were unable to bind SMN (Fig. 2B and C). We believe that due to its highly structured C-terminal domain, these truncations resulted in misfolding of α-COP and therefore failed to co-precipitate SMN. A different approach was therefore required to map the α-COP-SMN interaction.

Figure 2.

The highly conserved C-terminus of α-COP mediates SMN binding. (A) Map of N-terminal and C-terminal deletions of α-COP (751–1224). (B and C) HEK293TT cell were transfected with FLAG-tagged N-terminal deletions (B) or GST fused C-terminal deletions (N-term-GST) (C) of α-COP (751–1224) and FLAG immunoprecipitations and GST pull-downs immunoblotted for endogenous SMN. FLAG-bacterial alkaline phosphatase (BAP) was included as a control. (D and E) Wild-type and mutants R965A, Y1090H, R1178A, L1064A, R1083A and E1086A of α-COP were immunoprecipitated with FLAG antibodies and complexes immunoblotted for SMN and α-COP. Asterisks denote a non-specific band recognized by the anti-SMN antibody. (F) Wild-type and mutant FLAG α-COP were expressed in HEK293TT cells and lysates immunoprecipitated with FLAG antibody. The bound protein complexes were immunoblotted for FLAG-α-COP, endogenous COPI subunits β-COP, γ-COP and ε-COP and for endogenous SMN.

Site-directed mutagenesis of SMN binding proteins has commonly been utilized to map interaction domains, including of arginine–glycine (RG) peptide motifs, which have previously been shown to mediate SMN binding to other partners such as Gemin2 (39–41). We noted the presence of two minimal RG motifs (RG965-6 and RG1178-9) in the C-terminus of α-COP as potential SMN binding sites. However, site-directed mutagenesis of these motifs to alkaline (R965A and R1178A) was insufficient to prevent co-immunoprecipitation with SMN (Fig. 2D). We then opted to target highly conserved amino acids for site-directed mutagenesis. Notably, mutation of tyrosine 1090 to histidine (Y1090H) in α-COP completely abrogated SMN binding (Fig. 2D). Similarly, individual mutation to alanine (A) of highly conserved nearby amino acids leucine 1064 (L1064A), arginine 1083 (R1083A) and glutamate 1086 (E1086A) also crippled binding to SMN (Fig. 2E). To determine whether these α-COP mutant proteins retained physical integrity beyond SMN binding, we tested for maintenance of interactions with other components of the COPI heptamer. We performed Co-IP/western blotting for endogenous β-COP, γ-COP and ε-COP from human HEK293TT cells transfected with full-length FLAG-α-COP constructs. As shown in Figure 2F, wild-type α-COP and R1178A co-immunoprecipitated β-COP, γ-COP and ε-COP. Mutant Y1090H also co-immunoprecipitated these COPI complex subunits, implying that these mutants preserve the ability to assemble the full COPI coatomer complex.

α-COP mutants that do not bind SMN are unable to support neurite outgrowth

Given that α-COP is required for neurite elongation and that α-COP binds SMN, we sought to determine whether these two functions are co-dependent. Using the α-COP-depleted NSC-34 cells as shown in Figure 1, we tested the SMN binding defective α-COP mutants for ability to restore neurite length. Following transfection with human α-COP, western blotting showed that human α-COP protein was expressed in the presence of the murine-specific shRNA (Fig. 3A). Co-immunoprecipitation of human α-COP with murine SMN as well as ε-COP demonstrated normal COPI assembly in the murine cells (Fig. 3B). Immunofluorescence microscopy revealed that neurite outgrowth was restored in α-COP-depleted cultures upon transfection with either wild-type human FLAG-α-COP or mutant R1178A, as measured by α-tubulin staining. In contrast, the Y1090H mutant was unable to restore normal neurite architecture (Fig. 3C and D). These results imply that SMN association is critical for the ability of α-COP to support the formation and maintenance of neuronal processes. Interestingly, expression of SMN does not restore neurite outgrowth in this model, which supports our theory that in models of SMA, α-COP is able to ameliorate disease phenotype by distributing the remaining SMN protein to the neuritic compartment. When α-COP has been depleted by shRNA, it is not available to meaningfully distribute additional SMN is support of neurite outgrowth.

Figure 3.

SMN binding is required for restoration of neurite outgrowth in α-COP-depleted NSC-34 cells. (A) Western blotting detects FLAG-tagged human α-COP protein in NSC-34 cell line expressing the murine-specific α-COP shRNA. (B) FLAG-tagged human α-COP co-immunoprecipitated endogenous murine COPI subunit ε-COP. FLAG-BAP was included as a negative control. (C) Quantification of neurite length with 50 cells per three separate experiments. Cells were co-transfected with α-COP expression vectors and eGFP to identify transfected cells. Wild-type and R1178A significantly restored neurite length (P < 0.01 by one-way ANOVA, followed by post hoc Tukey HSD test), but Y1090H did not. Right panel demonstrates expression levels of exogenous FLAG-α-COP wt or mutants in α-COP-depleted NSC34 cells by western blotting (D) Representative micrographs of the cells stained with α-tubulin (red) mouse antibody and nuclei visualized with blue DAPI stain. Scale bar, 12.5 μm.

Rescue of a cell culture model of SMA by α-COP requires SMN binding

We have previously reported that heterologous expression of α-COP restored neurite outgrowth in the doxycycline-inducible NSC-34 cell model of SMA, and that mutations in the α-COP binding motif of SMN exon 2b rendered it unable to rescue neurite outgrowth (35). To determine whether the ability of α-COP to ameliorate the SMA phenotype in these cells was dependent on ability to bind SMN, we compared the effects of transfection with wild-type or mutant α-COP. Doxycycline-induced SMN-depleted cultures in neurite outgrowth media have significantly shorter neurites than control cultures following 72 h of elongation (P < 0.01), as we have previously described (35). While transfection of either wild-type human α-COP or the R1178A mutant, which retains SMN binding, restored neurite length in SMN-depleted NSC-34 cells (P < 0.01 compared with enhanced green fluorescent protein (eGFP) alone), the SMN binding defective α-COP mutants Y1090H, R1083A and E1086A each lost the capacity to restore normal neurite outgrowth in the SMN-depleted cells (Fig. 4A and B). Western blotting revealed that over-expression of the various FLAG-α-COP constructs did not increase SMN protein levels in doxycycline-treated cultures, indicating that this rescue is not the result of α-COP increasing endogenous SMN expression or reducing the degree of doxycycline-induced knockdown (Fig. 4C).

Figure 4.

SMN binding is required for α-COP rescue of neurite outgrowth in a cell culture model of SMA. NSC-34 cell clone 4 #56 that contains a doxycyline-inducible shRNA to murine SMN was co-transfected with wild-type or mtuant α-COP along with eGFP. (A) Immunofluorescence for neurites using antibody to α-tubulin (red) with eGFP indicating co-transfected cells compared with untransfected neighbors. Scale bar 25 μm. (B) Quantification of neurite length with 50 cells analyzed in each of three replicate experiments shows (P < 0.01 by post hoc Tukey HSD test following one-way ANOVA). Statistically significant changes are marked with an asterisk. Expression of wild-type and R1178A, which bind SMN, restored neurite length to normal levels. Y1090H, R1083A and E1086A mutants, which do not bind SMN, provided no significant rescue compared with eGFP alone. (C) Expression of exogenous FLAG-α-COP, and SMN was confirmed by western blotting using ß-actin as a control for protein loading.

The α-COP Y1090H mutant retains Golgi–ER retrograde transport function

Although the Y1090 mutant that is unable to bind to SMN retained ability to bind to other COPI complex proteins, it remained possible that this α-COP Y1090H/COPI complex was not fully functional and that this might explain the inability of the Y1090 mutant to restore neurite outgrowth in either α-COP or SMN-depleted cell cultures. We chose to assay α-COP function by measuring retrograde transport from Golgi to ER using the well-characterized temperature-sensitive YFP-VSVGts045-KDEL receptor (42). Unlike membrane-bound ER residents that use direct COPI binding to concentrate in COPI vesicles, soluble ER-resident proteins use C-terminal KDEL peptide sequences to bind to the KDEL receptor (KDELR), which resides in the Golgi and mediates their COPI-dependent return to the ER (43–46). The YFP-VSVGts045-KDELR chimera cycles between the ER and the cis-Golgi at permissive temperatures (32°C). At the non-permissive (40°C), protein that has undergone retrograde trafficking becomes trapped in the ER due to misfolding, providing an assay to examine COPI-dependent retrograde traffic (46,47).

We used the NSC-34 cell model with stable α-COP knockdown described above to test the impact of α-COP depletion on COPI trafficking. If COPI-dependent retrograde trafficking was altered by α-COP depletion, we would expect to see increased YFP-VSVGts045-KDELR remaining associated with the Golgi as indicated by its resident GM130 protein, following the switch to restrictive temperatures. As shown in Figure 5A, following YFP-VSVGts045-KDELR transfection of control NSC-34 cells, VSVGts045-KDELR localized predominantly to the GM130 positive Golgi complex at 32°C. When shifted from 32 to 40°C, VSVGts045-KDELR redistributed to the ER. In contrast, cells expressing α-COP shRNA demonstrated a defect in retrograde trafficking as VSVGts045-KDELR remained predominantly at the Golgi at the non-permissive temperature (40°C), demonstrating that we have successfully impinged upon normal Golgi–ER trafficking by depleting α-COP.

Figure 5.

α-COP Y1090H retains Golgi–ER retrograde transport function. (A) Transfection of NSC-34 cultures with the temperature-sensitive COPI cargo YFP-VSVGts45-KDELR shows that α-COP-depleted cultures have reduced Golgi–ER retrograde transport resulting in retention of the YFP fluorescence (green, arrow) in the Golgi (GM130, red). Quantification of the percentage of cells with an ER pattern of YFP fluorescence at the permissive (32°C) and restrictive (40°C) temperatures shows that α-COP-depleted cultures have significantly less (P = 0.007) retrograde transport compared with control cultures. (B) α-COP knockdown NSC-34 cells were transfected with YFP-VSVGts45-KDELR alone or in combination with FLAG-α-COP-wt, FLAG-α-COP-R1178A or FLAG-α-COP-Y1090H. Quantification of the number of cells with an ER pattern of YFP fluorescence reveals that all three constructs significantly restore Golgi–ER transport at the restrictive temperature (P = 0.005, 0.014 and 0.02 respectively). YFP fluorescence is green (arrow). Golgi are GM130 positive (red). Scale bar 12.5 μm. Although the percentage of cells with ER localization of the YFP in the Y1090H α-COP cells was reduced compared with wild-type, the difference was not significant (P = 0.18).

These results allowed investigation of the functionality of the SMN binding defective α-COP mutations. When the α-COP-depleted NSC-34 cells were transfected with wild-type human FLAG-α-COP, mutants R1178 or Y1090H, the VSVGts045-KDELR successfully redistributed to the ER at the non-permissive 40°C from predominant localization to the Golgi complex at 32°C (Fig. 5B). These data indicate that the α-COP mutant Y1090H retains the ability to support canonical COPI-dependent retrograde trafficking despite its inability to bind SMN.

Motor axon outgrowth rescue by human SMN and α-COP RNAs in mz-smn zebrafish

Zebrafish mz-smn mutants are a genetic model that results from depletion of smn from the earliest stages of development by removing both maternal and zygotic smn. A low level of human SMN is expressed thus generating a valid genetic model of SMA (23). Maternal-zygotic SMN (mz-smn) mutants were generated as described previously (23). These mutants show defects in motor axon outgrowth and motoneuron development. We therefore set out to test whether α-COP could rescue these smn defects in vivo as these did in our in vitro model. We injected RNAs of human SMN or α-COP into the mz-smn mutants. Animals were scored as having severe, moderate, mild or no defects of their motor axons 28 h post fertilization (hpf) as previously described (23). As shown in and quantified in Figure 6A and B, mz-smn mutants have abnormal motor axon morphologies. Wild-type human FLAG-SMN rescued normal axon morphology as previously reported (24). The two SMN dilysine mutants (KK76/77 and KK82/83), which are defective for α-COP binding and were unable to restore neurite outgrowth in our NSC-34 model of SMA (35), failed to completely rescue motor neurons and displayed a severe phenotype in between wild-type SMN RNA and mz-smn mutants. These data imply that α-COP binding is important for SMN function in neuronal development in a vertebrate system. Similarly, mz-smn zebrafish were injected with either wild-type human α-COP or the mutant Y1090H RNAs. Strikingly, wild-type α-COP completely rescued motor neuron morphology at 28 hpf (hours post fertilization) to a degree comparable to that observed with wild-type SMN. However, injection of RNA coding for the α-COP Y1090H mutant, which is unable to bind SMN, showed no significant rescue. Expression of all five transgenes was confirmed by western blot analysis with antibodies against human SMN or α-COP (Fig. 6C). Expression of α-COP did not increase the minimal RFP-SMN protein levels. These data show in a reciprocal fashion that the interdependent interaction of SMN with α-COP is critical for neuronal development in both cell culture and vertebrate models of SMA.

Figure 6.

α-COP interaction with SMN is required for motor axon development and outgrowth in a zebrafish model of SMA. Motor axon outgrowth was rescued by human SMN and α-COP RNA in mz-smn zebra fish model. (A) 250 pg of wild-type or mutant human SMN or α-COP (wt or Y1090H) RNA was injected into one cell stage mz-smn mutants zebrafish embryos. Three independent experiments were performed with 14–21 animals per experiment. (B) Animals were scored as having severe, moderate, mild or no defects in their motor axons at 28 h post fertilization (hpf). (C) Western blots were performed on protein from 28 hpf embryos. mz-smn have the Tg (hsp70: RFP-SMN) background that supplies low levels of RFP-SMN. Blots confirm expression of all SMN and α-COP proteins.

Discussion

The role of α-COP in eukaryotic cell biology has focused on its characterization as the major component of the Golgi–ER COPI transport vesicle. The presence of α-COP throughout developing neuronal processes and its co-localization and movement with SMN in varied neuronal cell culture types suggested its repertoire extends beyond the canonical Golgi–ER pathway (31). We previously reported that α-COP is capable of binding both SMN protein as well as a specific subset of mRNAs, inferring it may serve to deliver essential cargoes that enable maturation of axons and dendrites (48). It was recently reported SMN and the core snRNP protein SmB co-localize with COPI-trafficking vesicles in neural cells (13), and SMN co-localizes with hnRNP R in presynaptic mouse motoneuron axons (49), implicating functions of SMN beyond snRNP assembly in the recruitment and trafficking of RNA particles into axons and axon terminals. Here we demonstrate that α-COP depletion reduces neurite formation in both cultured NSC-34 cells and in primary cortical neurons, which show dramatic shortening of both Map2-positive dendrites and Tau-positive axons. In several motor degenerative diseases, including SMA, it is unclear whether loss of axon function is due to a defect in the proximal soma or dysfunction of the axon and muscle unit (50). Failure to properly form and maintain the synapse has been implicated in murine models of SMA (21,51,52). While motor neurons and axons are present at normal numbers during embryogenesis, there is reduced occupancy of motor synapses (53,54).

What does the interaction of SMN with α-COP instruct about the pathogenesis of SMA? SMN has been reported to have 20 or more binding partner proteins (reviewed in 55,56). The majority are components of the snRNP or so-called Gemins, accounting for SMN's best-characterized activity of snRNP assembly (3,57). SMN exon 2a mediates Gemin2 association while exon 2b is sufficient for α-COP binding (31,35). Mutations of the dilysine motifs at amino acid 76/77 or 82/83 abolished the ability of GST-tagged SMN exon 2b to bind α-COP, yet these retained the ability to co-immunoprecipitate Gemins 2 and 3 (31,35). While these mutants have not been tested in snRNP assembly assays, it is reasonable to expect based on maintenance of Gemin interactions that this functionality is preserved. In our NSC-34 cell model of SMA, wild-type SMN restored proper motor neuron morphology, but SMN with either the KK76/77AA mutation or KK82/83QQ mutation lost the ability to compensate for SMN depletion and rescue neurite outgrowth (35). A spate of recent reports has identified splicing deficits in the absence of SMN, but there is little agreement among these (4–6,58). We have also detected a limited number of alternatively spliced mRNAs following shRNA induced SMN depletion in the NSC-34 cell clone (H. Li, S. Custer, T. Gilson, A. Todd, H. Lin, Y. Liu and E. Androphy, submitted for publication). Thus, there is insufficient conclusive evidence that any major splicing errors are responsible for the motor dysfunction in SMA.

As we sought to identify the residues in α-COP required for binding to SMN, we were unexpectedly unable to map a short domain responsible for their association. In the yeast α-COP protein, the C-terminal 70 amino acids form an elongated β strand that loops back across multiple helices of the C-terminal domain (59). ε-COP binds to amino acids 899–993 and to the C-terminus of the β-sheet of α-COP. Comparison of human and yeast is readily possible due to the high degree of conservation. Mutations of Y1090H, L1064A, R1083A and E1086A, which are clustered in juxtaposed α helices, selectively destroyed the interaction with SMN but not with ε-COP. The Y1090H point mutation retains normal COPI transport activity as demonstrated by the ability to complex with the other COPI components and restores trafficking of VSVGts-KDELR in α-COP-depleted NSC-34 cells. Thus, we have been able to separate SMN-binding from traditional Golgi–ER COPI functions.

One of the more remarkable findings we describe here is that heterologous expression of α-COP restored neurite outgrowth in SMN deleted cultured cells and rescued motor neuron pathfinding in smn deficient zebrafish embryos. The supposition is that this occurs by directing the remaining SMN protein to maturing processes. It is challenging to envision how additional α-COP might influence snRNP assembly or alternative mRNA splicing. Conceivably this complementation outcome might reflect another pathway, however, the fact that the α-COP Y1090H, R1083A and E1086A SMN-binding defective mutants lost this capacity implies that neurite formation dependent on the full function of both α-COP and SMN. It has been demonstrated by our laboratory and others that there is a redistribution of SMN following α-COP depletion, with a portion of the cytoplasmic SMN appearing to be sequestered in the trans-Golgi network (31,60), and we have observed other effects of SMN depletion on Golgi organization (S. Custer and E. Androphy, unpublished data). An additional consideration is that COPI is involved in autophagosome maturation (61). SMN-depleted cells exhibit reduced autophagy and autophagic flux, which can be somewhat compensated for by over-expression of α-COP (62). How this might relate to intracellular trafficking within neurons and vitality in SMA is not known.

Mounting evidence from spontaneous and transgenic mouse models supports the impact of COPI function in neuronal health. Mice with point mutations in δ-COP, otherwise known as archain1, exhibit cerebellar ataxia due to degeneration of the Purkinje neurons (63). Similar to α motor neurons in the spinal cord, Purkinje cells are extremely large and must maintain an enormous dendritic arbor as well as lengthy axons, thus their cytoarchitecture may make them sensitive to small changes in intracellular trafficking. Mice with point mutations in the COPI-associated protein Scyl1 display cerebellar atrophy as well as motor neuron disease, and a recently reported Scyl1 knockout mouse develops a strong motor neuron disease phenotype reminiscent of amyotrophic lateral sclerosis (ALS) (64). Severe and progressive Golgi fragmentation, an early hallmark of many neurodegenerative diseases, was demonstrated in motor neurons of progressive motor neuronpathy (pmn) mice due to loss of Golgi-localized tubulin-binding cofactor E (TBCE). ARF1/TBCE-mediated cross-talk coordinates COPI formation and tubulin polymerization at the Golgi (65). These mouse models intersect with COPI function and underscore its importance in motor neuron biology and normal function. The zebrafish experiments shown here provide evidence in a non-mammalian, vertebrate model of SMA that the basic biological importance of the α-COP and SMN interaction in neuronal cells persists. Whether COPI vesicles are essential for the proper distribution of SMN protein in neuronal cells or acts in concert with SMN to deliver mRNA or other essential cargoes to the dendritic and axonal compartments remains to be determined. Our findings further the viewpoint that SMN has neurologically relevant functions outside the spliceosome. The cytosolic functions of SMN are under-appreciated and need to be investigated more closely as we further our search for meaningful therapies in this disease. The data presented here demonstrate for the first time the importance of α-COP and its interaction with SMN to the growth and maintenance of neuronal cells.

Materials and Methods

Plasmids

C-terminal deletions of the α-COP751-1224 construct (31) were cloned into pExchange 4a (NotI/XhoI) N-terminal deletions of the α-COP751-1224 were fused with GST by cloning into NotI of the mammalian expression vector pCN-GST-Myc (66). Site-directed mutations R965A, R1178A, Y1090H, L1064A, R1083A and E1086A were introduced into human α-COP C-term amino acids 751–1224 and full-length expression constructs using QuickChange (Invitrogen, CA) (31). SMN wild-type, mutant constructs and the YFP-VSVGts045-KDELR plasmid were previously described (35,42).

Cell culture and shRNA knockdown

HEK293TT and NSC-34 cells were cultured in DMEM (Gibco) with 10% fetal bovine serum (FBS; Atlas Biologicals, Fort Colins, CO) and 1% penicillin and streptomycin (Gibco) at 37°C. NSC-34 cells were transfected with α-COP shRNA construct or non-silencing control (Sigma, TRCN0000313321 and SHC-002) using Lipofectamine 2000 (Invitrogen, CA). Cells stably expressing the shRNA were isolated after 6 weeks selection in 2.5 µg/ml puromycin (Gibco). Clones with decreased α-COP protein levels were identified by western blotting analysis using α-COP antibody (Aviva System Biology, San Diego, CA). The murine-specific α-COP shRNA lentiviral vector along with the non-silencing control (TRCN0000313321 and SHC-002) were purchased from Sigma. Virus was produced in HEK293TT cells by transfections with α-COP shRNA vector, Pax2 and MD2.G and isolated at 48 and 72 h post-transfection, concentrated by centrifugation over a 20% sucrose cushion, and resuspended in NeurobasalA/B27/Glutamax. Primary mouse cortical cells were isolated from FVB/NJ neonates and cultured in NeurobasalA with B27 supplement and grown on Poly-d-Lysine (PDL)-coated glass coverslips at 30 000 cells per ml. To determine the impact of α-COP knockdown on neuronal survival and morphology, cells were adhered to coverslips for 6 h before being infected with α-COP shRNA lentivirus. For neurite extension assays, NSC-34 cells were grown on PDL-coated coverslips in DMEM/F12 (Gibco) with 1% tet-system FBS (Clontech, Mountain View, CA) as previously described (35).

Co-immunoprecipitation assays

Human α-COP expression vectors were transfected into HEK293TT cells using Lipofectamine 2000 (Invitrogen, CA). 48 h after transfection, cells were washed once with cold PBS and lysed in 50 mm Tris–HCl, pH 8.0, 150 mm NaCl, 1% NP-40, and ethylenediaminetetraacetic acid (EDTA)-free protease inhibitor cocktail for 20 min on ice and clarified by centrifugation at 14 000g for 15 min at 4°C. Lysates were incubated with anti-FLAG M2 beads (Sigma) for 4 h at 4°C and washed four times with wash buffer (50 mm Tris–HCl, pH 8.0, 150 mm NaCl, 0.2% NP-40). Beads were heated in 5 × sample buffer and bound proteins separated by SDS–PAGE. Proteins were probed by western blotting with an anti-FLAG antibody (M2, 1:7500, Sigma) or antibodies against SMN (MANSMA 6, 1:1000).

GST pull-down assays

For GST pull-down experiments, human HEK293TT cells were transfected with GST fused α-COP C-terminal deletions using Lipofectamine 2000. After 36 h, cells were washed with PBS and lysed in 500 µl lysis buffer (50 mm Tris–HCl, pH 8.0, 350 mm NaCl, 1% NP-40, 0.25% sodium deoxycholate and protease inhibitor cocktail). The lysate was centrifuged at 14 000g for 15 min at 4°C and supernatant incubated with glutathione beads (GE Healthcare) for 4 h at 4°C. The beads were washed four times with wash buffer (50 mM Tris–HCl, pH 8.0, 100 mm KCl, 100 µm EDTA, 0.2% NP-40 and 2.5% glycerol). Beads were heated in sample buffer and proteins were detected by western blotting with rabbit anti-GST antibody (1:500) or anti-SMN antibody MANSMA 6 (1:1000).

Immunofluorescence

Cells were fixed in 4% paraformaldehyde for 10 min, and incubated in blocking buffer (PBS with 5% normal goat serum (Clontech) and 0.25% Triton X-100) for 30 min. Fixed cells were incubated with primary antibody for 1–2 h and rinsed three times with PBS. Secondary antibodies conjugated with Alexa Fluor 488 or 594 were added for 45 min and washed three times with PBS. Cover slips were mounted onto microscope slides using Prolong Gold with DAPI and imaged using an Olympus FV1000 MPE with a Spectra-Physics 2-photon MaiTai DeepSee laser.

The KDELR-tag trap assay

NSC-34 cells (wild-type or stable α-COP knockdown) were grown on glass coverslips, transfected with YFP-VSVGts045-KDELR and either empty pExchange4a, FLAG-αCOP-wt, FLAG-αCOP-R1178A or FLAG-αCOP-Y1090H, and immediately transferred to permissive temperatures (32°C) overnight. One set of coverslips for each condition was fixed to document the YFP-KDELR pattern at the permissive temperature and the remaining coverslips were transferred to the restrictive temperature (40°C) for 2 h and fixed. After fixation in 4% paraformaldehyde, all coverslips were blocked in 5% normal goat serum with 0.1% Triton X-100 for 30 min. Golgi were labeled with antibody against GM130 (Cell Signaling, D6B1) diluted 1:250) and anti-rabbit Alexa Fluor 594. Samples were imaged using a fluorescence microscope Olympus FV1000 MPE and the YFP signal in transfected cells was scored as predominantly Golgi, or predominantly ER. The experiment was performed three times and scored by two separate observers. Quantification was expressed as percentage of cells with predominantly ER localized YFP signal at the various temperatures. Statistical significance between groups was determined by two-tailed Student's t-test.

Neurite outgrowth assays

Untreated or α-COP-depleted NSC-34 cells were seeded on a PDL-coated 18 mm coverslips in a 12-well plate. After overnight culture, cells were transfected with human wt or mutant α-COP expression plasmids along with eGFP-C1 using Lipofectamine 2000. 24 h post-transfection, the media was replaced with neurite outgrowth media (DMEM/F12, 1% FBS). Following 72 h of neurite extension, cells were fixed in 4% paraformaldehyde for 10 min and incubated in blocking buffer (PBS with 5% normal goat serum and 0.25% Triton X-100) for 1 h. Neurites were stained with a monoclonal antibody against α-tubulin (1:5000, Sigma) followed by goat anti-mouse conjugated Alexa Fluor 594 (1:1000, Life Technologies). Neurites were measured as described previously (35). For rescue of SMN-depleted NSC-34 cells, cultures were maintained in doxycycline (2 µg/µl) for 48 h to induce SMN knockdown and then reverse transfected with wild-type or mutant α-COP at the time of plating onto PDL-coated coverslips. Neurite extension was carried out for 72 h post-transfection in neurite outgrowth media (DMEM/F12, 1% tet-system FBS) with or without doxycycline, and neurite length was measured by α-tubulin immunofluorescence. All images were observed with a ×20 objective. Neurite measurements were performed on at least 50 cells from three independent experiments by two independent observers. Statistical significance was determined by one-way ANOVA followed by post hoc Tukey HSD test for significance between samples.

Primary cortical neuron culture

Primary cortical neurons were isolated from newborn FVB/NJ pups. Briefly, mice were washed with 70% ethanol then quickly decapitated. Brains were removed and set in a dish of Neurobasal-A medium. Using fine forceps, meninges were removed, the cortices isolated and minced. Following DNase digestion, the tissue was passed through a series of pipette tips until a single cell suspension was achieved, and then plated at 3 × 10⁴ cells/ml in Neurobasal-A with B27 and 200 mm Glutamax. Cells were allowed to adhere to the dish for 6 h before being infected with lentivirus expressing the murine-specific α-COP shRNA or non-silencing control. Cultures were maintained for 5 days in vitro and then fixed in 4% paraformaldehyde and immunolabeled with either Map2 to visualize dendrites or Tau to visualize axons.

RNA injection to zebrafish embryos

Wild-type and maternal:zygotic (mz) smn mutants were used in these studies (23). mz-smn mutants lack all zebrafish Smn but carry a transgene Tg (hsp70:RFPSMN) that expresses very low levels of human RFP-SMN. These animals, therefore, have no fish smn and low human SMN thus constituting a legitimate genetic model of SMA. All fish were grown and maintained in the Ohio State University (OSU) zebrafish facility following established protocols and OSU animal welfare guidelines. mRNA was made from linearized DNA plasmids. Wild-type human SMN, SMN-K76/77A and SMN-K82/83Q in the pcDNA vector were linearized with SmaI. Wild-type and mutant α-COP in the pcDNA vector were linearized with MluI. Capped RNA was generated using the T7 mMESSAGE mMACHINE kit (Ambion, Austin, TX) following the protocol of the manufacturer.

One–two cell stage mz-smn embryos were injected with 250 pg of each mRNA separately. At 28 hpf, injected and control uninjected embryos were anesthetized with tricaine (250 µg/ml, Sigma A-5040) and fixed overnight at 4°C in 4% formaldehyde/PBS. Embryos were mounted on glass coverslips for imaging on a Leica confocal microscope. Immunohistochemistry was performed with the znp1 mouse monoclonal antibody (Hybridoma bank) (67). Motor axons innervating the mid-trunk (myotomes 6–15) on both sides of the fish were scored as described (68). Three experiments were performed for each RNA with an average of 16 embryos scored per experiment. Samples of zebrafish embryos injected with mRNA at the one- to two-cell stage and uninjected wild-type embryos were collected for western blot at 28 hpf. Samples were generated by boiling 25 uninjected and injected embryos in 75 µl blending buffer (63 mm Tris, pH 6.8, 5 mm EDTA, 10% SDS). 10 µl (equivalent to 3 embryos or 75 µg of protein) were added to 10 µl of sample buffer (100 mm Tris, pH 6.8, 0.2% bromophenol blue, 20% glycerol, 200 mm dithiothreitol) and run on a 7% polyacrylamide gel, blotted to nitrocellulose, probed with mouse anti-SMN (1:500; MANSMA12) or anti α-COP (1:200) (31) and detected by chemiluminescence of bound HRP-conjugated mouse antibody.

Funding

This research was supported by the National Institute of Neurologic Disease/National Institutes of Health under R01 NS082284 to E.J.A. and R01NS050414 to C.E.F.