Disconnection between membrane signalling and actin networks can have catastrophic effects depending on cell size and polarity. The survival motor neuron (SMN) protein is ubiquitously involved in assembly of spliceosomal small nuclear ribonucleoprotein particles. Other SMN functions could, however, affect cellular activities driving asymmetrical cell surface expansions. Genes able to mitigate SMN deficiency operate within pathways in which SMN can act, such as mRNA translation, actin network and endocytosis. Here, we found that SMN accumulates at membrane protrusions during the dynamic rearrangement of the actin filaments. In addition to localization data, we show that SMN interacts with caveolin-1, which mediates anchoring of translation machinery components. Importantly, SMN deficiency depletes the plasma membrane of ribosomes, and this correlates with the failure of fibroblasts to extend membrane protrusions. These findings strongly support a relationship between SMN and membrane dynamics. We propose that SMN could assembly translational platforms associated with and governed by the plasma membrane. This activity could be crucial in cells that have an exacerbated interdependence of membrane remodelling and local protein synthesis.
The establishment of cell polarity depends on the ability to dissect functional subcellular domains in response to intrinsic developmental programmes or external stimuli (Wodarz, 2002; Bryant and Mostov, 2008). An advantageous mechanism to select the protein profile both in space and in time is the asymmetrical distribution or translation of mRNAs (Besse and Ephrussi, 2008; Medioni et al., 2012; Xing and Bassell, 2013; Jung et al., 2014). At least two factors contribute to restricting protein synthesis to discrete compartments: the highly oriented cytoskeletal network, and the packaging of mRNAs into dedicated ribonucleoprotein particles (mRNPs) (Holt and Bullock, 2009; Xing and Bassell, 2013). Furthermore, membrane receptor signalling provides essential inputs to ensure a fine-tuning of the cytoplasmic control of gene expression (Latham et al., 1994; Willis et al., 2007; Ridley, 2011). Signal transduction pathways, such as the mammalian target of rapamycin (mTOR) signalling cascade, switch receptor activity into localized translational outputs by activating components of translation machinery (Berven et al., 2004; Hoeffer and Klann, 2010). Notably, an elegant work has highlighted the ability of the plasma membrane to govern local protein synthesis not only as mediator of extracellular cues (Tcherkezian et al., 2010). The authors showed that the plasma membrane of neural cells sequesters inactive components of translation machinery, including the 40S and 60S ribosomal subunits. In an activity-dependent manner, ribosomes are released from the plasma membrane and are recruited into translating polyribosomes, to promote spatially restricted protein production. Furthermore, a great deal of progress has been made in identifying mechanisms that regulate the local translation, revealing RNA-binding proteins involved in the sorting or translation of mRNAs (Xing and Bassell, 2013). Some of these, such as the zipcode-binding protein 1 (ZBP1) and the KH domain-containing splicing regulatory protein (KSRP, also known as KHSRP), contact the mRNA upon transcription or splicing (Pan et al., 2007; Holt and Bullock, 2009). Other RNA-binding proteins, such as the fragile X mental retardation protein (FMRP, also known as FMR1) and the fused in sarcoma protein (FUS), affect their target transcripts preferentially at the translational level (Napoli et al., 2008; Darnell et al., 2011; Yasuda et al., 2013). Consistent with the crucial role of the local protein synthesis in the generation of cell polarity, the genetic alterations of RNA-related proteins have been linked to severe neurological disorders (Bassell and Warren, 2008; Kwiatkowski et al., 2009).
The survival motor neuron (SMN) protein is known to be involved in important tasks of RNA metabolism (Li et al., 2014). In humans, two almost identical genes, SMN1 and SMN2, encode SMN protein. Although SMN1 produces full-length transcripts, SMN2 mainly produces an alternatively spliced mRNA, whose product is rapidly degraded. As a consequence, SMN2 can not compensate for SMN1 defects, unless SMN2 is present in multiple copies. Mutations of SMN1 result in spinal muscular atrophy (SMA), the leading genetic cause of infant mortality (Hamilton and Gillingwater, 2013). The complete loss of SMN is not compatible with the cell viability in all tissue types, whereas reduced SMN levels affect cells differentially. The large α-motoneurons in the spinal cord display the highest susceptibility to SMN deficiency, but the molecular mechanism underlying this selectivity remains obscure. To date, the best-characterized function of SMN is the cytoplasmic assembly of the small nuclear RNPs (snRNPs), the core components of the pre-mRNA splicing machinery (Pellizzoni et al., 2002). However, an emerging idea is that the loss of additional function(s) of SMN unrelated to RNA splicing could co-contribute to SMA pathogenesis. In particular, it has been strongly suggested that SMN is involved in the trafficking and/or translation of target transcripts (Rossoll et al., 2003; Tadesse et al., 2008; Glinka et al., 2010; Peter et al., 2011; Hubers et al., 2011; Fallini et al., 2011, 2014; Yamazaki et al., 2012; Rathod et al., 2012; Sanchez et al., 2013). In this context, it is noteworthy to mention cross-species conserved genes able to mitigate the SMN loss-of-function defects (Dimitriadi et al., 2010). Among others, the modifier gene plastin 3 (PLS3) has been shown to protect fully against deficiency in SMA (Oprea et al., 2008). PLS3 is an F-actin-bundling protein, primarily involved in the regulation of the actin cytoskeleton (Delanote et al., 2005). Interestingly, zebrafish SMN mutants display reduced PLS3 protein production, without changes at the transcriptional level (Hao et al., 2012). Importantly, proteins encoded by SMA modifier genes operate within distinct cellular networks, specifically, actin dynamics, endocytosis and mRNA translational control (Dimitriadi et al., 2010). These apparently dissimilar pathways might act in concert, and their perfect coupling might converge on the generation and maintenance of cell polarity (Gibbings et al., 2009; Lee et al., 2009; De Rubeis et al., 2013). Collectively, these findings prompted us to investigate a possible involvement of SMN at the functional interplay between actin dynamics and translational control networks.
In this study, we provide evidence for an intimate link between SMN and plasma membrane dynamics. We found that SMN interacts with caveolin-1, a key regulator of plasma membrane composition and expansion. We demonstrate that SMN coexists with translation machinery components in caveolin-rich membrane domains. Importantly, SMN deficiency depletes the plasma membrane of ribosomal proteins, and this correlates with the impairment of actin and membrane remodelling. In conclusion, our findings reveal a possible involvement of SMN in local protein production underlying plasma membrane-actin network axis.
Subcellular distribution of SMN during actin remodelling
All membrane protrusions known to occur at the leading edge of the cell, such as lamellipodia and filopodia, are dynamically orchestrated by changes in the polymerization and assembly of the actin cytoskeleton (Ridley, 2011). In order to explore a functional link between SMN and actin dynamics, we subjected fibroblasts to an ATP depletion and recovery assay. This approach stimulates the extension of actin-based membrane protrusions, and provides a useful tool to address key determinants of the actin network in living cells (Svitkina et al., 1986; Bear et al., 2002). In our system, after 30 min of ATP recovery, ∼70% of the cells exhibited different types of membrane protrusions. At this time point, an immunofluorescence assay was performed to visualize the subcellular distribution of SMN. As expected, in unstimulated fibroblasts, SMN displayed a punctuate pattern distributed throughout the whole cell body, and a typical localization in nuclear bodies, named gems (Fig. 1A, unstimulated). Surprisingly, in ATP-recovering cells, we observed a strong redistribution of SMN in vesicle-like structures within the protrusive areas of the plasma membrane. Interestingly, SMN-positive dots were also detectable along the cell perimeter of stimulated cells (Fig. 1A,B, stimulated). In order to correlate SMN distribution to the actin filament meshwork, we probed cells with phalloidin (Fig. 1C). Stimulated fibroblasts displayed a well-organized cortical actin network underneath the plasma membrane, as well as several filopodial extensions. In these cells, SMN accumulated at the tips of both the cortical actin arc and filopodia. Notably, in similar membrane protrusions observed in unstimulated fibroblasts, SMN was not detectable (see detail in Fig. 1C), suggesting that the peripheral recruitment or localization of SMN is mainly governed in an activity-dependent context. In support of this, and to exclude exacerbated effects due to the ATP depletion and recovery treatment, we tested the localization of SMN upon more physiological conditions known to induce actin filament rearrangements (Ridley, 2011). In particular, we performed immunofluorescence assays of starved fibroblasts stimulated with fetal bovine serum (FBS) for 30 min, or with basic fibroblast growth factor (bFGF, 50 ng/ml) for 15 min. Both conditions led to extension of cellular protrusions, although with a lesser frequency compared to the ATP depletion and recovery assay. However, in cells exhibiting membrane remodelling, we observed a peripheral accumulation of the SMN protein (see Fig. S1A), consistent with the idea that a fraction of SMN moves to the periphery in an activity-dependent manner. Given that accumulation of SMN at the raffling membrane could occur for a thickness effect, we examined the SMN distribution in fibroblasts expressing a freely diffusible green fluorescent protein (GFP). Following ATP depletion and recovery, we observed that GFP was present at the membrane protrusions, but it was not able to compensate for SMN recruitment at the same subregions (Fig. S1B). Finally, we also verified the subcellular distribution of another RNA-related protein. We checked the localization of KSRP, given that this RNA-binding protein interacts with SMN (Tadesse et al., 2008), and shuttles between nuclear and cytoplasmic compartments. In contrast to SMN, KSRP was not enriched in membrane protrusions, and its localization was almost unaffected in stimulated cells (see Fig. S1C).
Interestingly, these findings reveal a peripheral recruitment of SMN in structures underlying actin filament remodelling.
A fraction of SMN colocalizes with translation machinery at membrane protrusions
We asked whether molecules related to translational control were present in the same subregions where SMN was shown. In particular, we imaged cells for both RNA and components of translation machinery (Fig. 2A). Cellular RNA, including mRNPs, was labelled by using an RNA-selective fluorescent probe. In parallel, protein synthesis machinery was visualized by the immunostaining of ribosomal protein S6 (S6, also known as RPS6). In the case of RNA, in unstimulated cells, maximal fluorescence was detectable in the nucleoli and throughout the nucleus. A diffuse and faint signal was also present in the cytoplasmic compartment. Under the same conditions, the ribosomal protein S6 displayed a typical pattern throughout the cytoplasm (Fig. 2A, unstimulated). In stimulated cells, we found an asymmetrical redistribution of both RNA and S6, with a robust detection of the respective signals at the membrane protrusions (Fig. 2A, stimulated). Finally, to verify the presence of initiation sites of protein synthesis, we checked the phosphorylation status of S6 by using an antibody against S6 phosphorylated at Ser235 or Ser236 (phospho-S6). Phospho-S6 immunostaining suggested that there was active translational hotspots at the leading edge of stimulated cells (Fig. 2A, stimulated). To validate this result, we carried out the surface sensing of translation technology (SUnSET) (Schmidt et al., 2009). Newly synthesized proteins were traced by puromycin incorporation and revealed by immunostaining with anti-puromycin antibody. As shown in Fig. 2B, in stimulated fibroblasts, newly synthesized proteins were enriched at membrane protrusions, this pattern clearly overlapped with phospho-S6 as well as SMN immunostaining. Notably, the total fluorescence intensity of phospho-S6 was significantly stronger in fibroblasts extending protrusions (Fig. 2A,B, stimulated). This was consistent with a switching on of the mTOR pathway. We further monitored the mTOR pathway by western blot analysis. As shown in Fig. 2C, the p70S6 kinase, a downstream target of mTOR, shifted to an active status in fibroblasts extending protrusions, and this correlated with Ser235 or Ser236 phosphorylation in the S6 protein. This result was not surprising, because a direct link between mTOR activity and actin dynamics is a well-accepted notion (Berven et al., 2004; Hoeffer and Klann, 2010). Our main interest was to verify the colocalization between SMN and translation machinery during the actin remodelling. To this end, we subjected stimulated cells to dual immunofluorescence for SMN and S6, or SMN and phospho-S6, and then analysed slides by confocal laser scanning microscopy (Fig. 2D). Immunostaining of both SMN and the S6 proteins appeared to be overlapping in several subcellular regions. By performing z-stack imaging and orthogonal analysis, we identified several yellow dots that showed colocalization between SMN and both S6 and phospho-S6. Interestingly, the colocalization spots were preferentially located near the cell surface of stimulated cells (see graphs in Fig. 2D).
These data show that SMN partially compartmentalizes with translation machinery in cellular structures in which dynamic changes of the actin cytoskeleton are coupled with spatially restricted protein synthesis.
SMN associates with the plasma membrane
Taking into account our images displaying SMN dots along the cell perimeter of ATP-recovering cells (Fig. 1B), we suspected a physical contact of SMN with the plasma membrane. In order to support this hypothesis, we prepared plasma-membrane-enriched fractions from both unstimulated and ATP-recovering fibroblasts, and performed western blot analysis to check for SMN presence. As shown in Fig. 3A, SMN was detectable in plasma membrane fractions of both control (unstimulated) and ATP-recovering fibroblasts (stimulated), without significant differences. This result was apparently in disagreement with images showing the increased peripheral localization of SMN under stimulating conditions. We suppose that during actin remodelling, cytoplasmic SMN granules accumulate within membrane protrusions that are in close proximity, but not physically anchored, to the plasma membrane. In addition, a membrane-bound fraction of SMN, as revealed by the biochemical approach, is usually present, even if apparently it is less detectable by immunofluorescence assays. When we probed blots with an antibody specific for the core-splicing factor protein SmB (also known as SNRPB), only the whole-cell extracts displayed immunoreactive bands (Fig. 3A). This is consistent with the fact that the canonical SMN complex assembles Sm proteins into the cytoplasm (Li et al., 2014). We also tested components of the translation machinery. We used antibodies against S6 and L7 (also known as RPL7), markers of 40S and 60S ribosomal subunits, respectively. Both S6 and L7 were present in plasma membrane fractions from unstimulated fibroblasts (Fig. 3A). In contrast to SMN, no, or almost no, ribosomal proteins were detectable in plasma membrane fractions from stimulated cells. Interestingly, this result was in agreement with the previously published work (Tcherkezian et al., 2010) reporting that inactive ribosomes, anchored to plasma membrane of unstimulated neurons, are released in an activity-dependent manner. Notably, the endoplasmic reticulum (ER) marker, calreticulin, was present but not enriched in our plasma membrane preparations (Fig. 3A). In addition to fibroblasts, we also achieved similar results in HeLa cells. In western blot analyses, we observed a decrease of S6 in the plasma membrane fractions from both ATP-recovering and EGF-stimulated cells (see Fig. S2). This event correlated with Thr389 phosphorylation in p70S6 kinase, confirming that ribosomes are released from the plasma membrane in combination with mTOR switch-on. Likewise in fibroblasts, the amount of SMN anchored to plasma membrane fractions of HeLa cells appeared to be almost unchanged. This result not only excluded the possibility of a cell-type-specific event, but pointed to the idea that a translational platform anchored to the plasma membrane might occur commonly in many cell types. Furthermore, we carried out a more stringent fractionation approach to dissect membranes in different subdomains (Fig. 3B). Detergent-resistant membranes (DRMs), containing lipid rafts, can be separated from heavy non-raft domains and analysed by western blot assay. In both unstimulated and ATP-recovering fibroblasts (stimulated), the majority of SMN protein was detectable in the non-raft fractions of membranes. However, SMN was also present in lipid-raft-containing light fractions. Interestingly, the amount of raft-localized SMN protein increased following the cellular ATP-recovery treatment (Fig. 3B, stimulated). Apparently, L7 was not recovered in DRMs. Given that detergent extraction affects lipid–protein interactions, so that only the proteins strongly associated with lipids are detectable in DRMs (Schuck et al., 2003), we could not exclude the presence of ribosomal proteins also in lipid rafts subdomains.
Collectively, we provide biochemical evidence of a physical association of SMN with plasma membrane. In addition, these data suggest that SMN protein relocalizes at specific plasma membrane subdomains during actin or cell surface remodelling.
SMN interacts with caveolin-1
To further define the SMN–membrane association, we verified a possible interaction between SMN and caveolin-1. Caveolin-1 is the main component of plasma membrane caveolae, which are implicated in cell membrane composition and expansion, and cytoskeleton remodelling (Grande-García et al., 2007; Dubroca et al., 2007). Furthermore, we were very interested to explore a possible link between caveolin-1 and SMN, because caveolin-1 unexpectedly appears in the interaction map of modifier genes of SMN defects (Dimitriadi et al., 2010). To this end, we prepared plasma membrane fractions from unstimulated fibroblasts, and carried out co-immunoprecipitation assays by using anti-caveolin-1 antibody. SMN co-precipitated with caveolin-1 from the plasma membrane fractions, which also displayed immunoreactivity to the anti-L7 antibody (Fig. 4A, left). This result was confirmed by reciprocal co-immunoprecipitation using a monoclonal antibody against SMN (Fig. 4A, right). These findings provided biochemical evidence that SMN coexists with components of translation machinery in caveolin-rich membrane domains. In parallel, we imaged cells subjected to dual immunostaining for SMN and caveolin-1. Interestingly, colocalization spots were seen in both intracellular and peripheral subregions (see Fig. S3A). SMN–caveolin-1 association was also explored by an in situ proximity ligation assay (PLA), which reveals endogenous protein–protein interactions in fixed cells. Both unstimulated and ATP-recovering fibroblasts were subjected to PLA using antibodies against SMN and caveolin-1, and then processed for confocal microscopy (Fig. 4B,C). In unstimulated cells, several PLA dots were present, and some of these dots were clearly detectable at the cellular leading edge (Fig. 4B). Interestingly, PLA occurrence was significantly increased in ATP-recovering fibroblasts (Fig. 4C). PLA specificity was validated by testing different combinations of primary antibodies against SMN and caveolin-1 (see Fig. S3B). By performing in situ PLA, we confirmed that the interaction between SMN and caveolin-1 also occurs in cultured neurons from rat spinal cord (see Fig. S3C). PLA was performed in untreated or brain-derived neurotrophic factor (BDNF)-stimulated spinal cord neurons. Notably, BDNF treatment led to higher PLA signals, compared to the unstimulated cultures. These data prompted us to investigate whether caveolin-1 could affect SMN and/or ribosome linkage at the plasma membrane. We assessed this aspect by preparing plasma membrane fractions from caveolin-1-depleted fibroblasts (Fig. 4D). Fibroblasts transfected with control small interfering RNA (siRNA) (siC) or siRNA against caveolin-1 (siCav-1) were subjected to the ATP depletion and recovery assay, and then processed to obtain plasma membrane fractions. Western blot analysis showed that caveolin-1 deficiency did not change the SMN abundance at the plasma membrane of unstimulated cells. More substantial differences were observed in plasma membrane fractions from ATP-recovering cells (stimulated), which retained more SMN protein following caveolin-1 knockdown. Interestingly, caveolin-1 deficiency depleted plasma membrane compartments of ribosomal proteins (Fig. 4D).
These results suggest that SMN could contact the plasma membrane in a caveolin-1-independent manner. However, caveolin-1 could affect turn over of membrane SMN. In addition, we provide evidence that caveolin-1 could mediate ribosome anchoring.
SMN depletion impairs actin dynamics
By performing loss of function studies, we evaluated the effects of SMN function in actin dynamics. We reduced the expression of endogenous SMN by the transient transfection of fibroblasts with SMN1-selective siRNAs (siSMN). Scrambled siRNAs were used as control (siC). At 48 h post transfection, we probed actin filaments and imaged cells by using fluorescence microscopy (Fig. 5A). Unstimulated cells displayed mild perturbations of the actin meshwork in the absence of SMN (Fig. 5A, unstimulated). Major differences were observed under stimulating conditions (Fig. 5A, stimulated). In particular, stimulated control cells were able to properly reassemble actin filaments, exhibiting robust actin arcs, as well as several filopodial extensions. Under the same condition, SMN-depleted cells displayed an aberrant and unorganized actin network. In the absence of SMN, the amount of actin filaments in the growing tips of filopodia appeared to be low compared to similar structures observed in control cells (Fig. 5A′). Overall, the ability to extend membrane protrusions was significantly reduced in SMN-deficient fibroblasts (Fig. 5B). Moreover, owing to the non-oriented cytoskeleton, we thought that SMN-depleted cells were unable to deliver mRNPs for local translation. To verify this idea, we assessed the subcellular localization of β-actin mRNA. We focused on the β-actin transcript for two main reasons: (1) the β-actin mRNA asymmetrically relocalizes in regions where actin protein production supports cell polarity, and (2) its peripheral distribution has been linked to SMN function. To visualize both β-actin transcript and the SMN protein, fibroblasts extending protrusions were subjected to combined fluorescent in situ hybridization (FISH) and immunostaining. As showed in Fig. 5C, in the absence of SMN, β-actin mRNA appeared diffuse throughout the cytoplasm, without asymmetrical accumulation to the cell periphery. The quantification of the FISH signal confirmed the reduced peripheral recruitment of β-actin transcript in SMN-depleted cells (Fig. 5D). These results strongly confirm the notion that SMN loss of function is deleterious in actin-based membrane remodelling.
SMN loss of function depletes the plasma membrane of ribosomes
It has been reported that dysregulated mTOR signalling occurs in the SMA disease context (Kye et al., 2014). This prompted us to verify whether, in our SMN-depleted cells, the perturbation of actin network combines with a reduced mTOR activity. Western blot analysis of fibroblasts stimulated to extend protrusions revealed that the p70S6 kinase was activated in both control and SMN-depleted cells (Fig. 6A). Phospho-S6 immunostaining confirmed that mTOR and p70S6 activity was independent of SMN expression levels (see Fig. S4A). Notably, in stimulated SMN-knockdown cells, phospho-S6 immunostaining appeared diffuse throughout the cytoplasm, without peripheral accumulations. Apparently, our findings are in disagreement with the notion that SMN deficiency dysregulates mTOR activity (Kye et al., 2014). However, we do not exclude perturbed translational events downstream of mTOR. To assess this question, we evaluated in vivo the protein synthesis rate of SMN-depleted cells by performing a SUnSET assay. Both siC- and siSMN-transfected fibroblasts were subjected to ATP depletion and recovery assay, in which the ATP recovery step was carried out in the presence or absence of the mTOR inhibitor rapamycin. Newly synthesized proteins were traced by puromycin incorporation and detected by western blot analysis with anti-puromycin antibody (Fig. 6B). In non-stimulating conditions, no significant differences were observed in the protein synthesis efficiency following SMN depletion. In stimulated cells, SMN knockdown resulted in an ∼25% reduction of translation. This difference was completely abolished by rapamycin treatment, suggesting that a subset of protein synthesis, downstream of mTOR, could depend on SMN function. As shown above, the plasma membrane of unstimulated fibroblasts exhibited immunoreactivity to SMN as well as ribosomal components of translation machinery (Fig. 3A). Next, therefore, we asked whether SMN could be implicated in the ribosome–membrane association. To this end, we performed western blot analysis of plasma membrane fractions from transiently SMN-depleted fibroblasts (Fig. 6C). In the absence of SMN, the ribosomal proteins were not detectable in plasma membrane compartments, although their total expression remained unchanged (Fig. 6A). This could be explained by an exacerbated activation of mTOR pathway. We excluded this because, in our system, SMN depletion was not sufficient to activate the signalling cascade downstream of mTOR (Fig. 6A). Moreover, we isolated DRMs fractions from unstimulated siC- and siSMN-transfected fibroblasts. By using western blot analyses, we showed that SMN deficiency changed the membrane distribution of both L7 and caveolin-1. In particular, the abundance of caveolin-1 was mainly reduced at the DRMs (Fig. 6D). These results suggest that SMN can affect the protein composition of the plasma membrane and highlight a possible involvement of SMN in assembly of ribosomal proteins into caveolin-1-rich membranes. In order to exclude a cell culture artefact, we also addressed the main crucial aspects of this study in a SMA pathological background. We prepared plasma membrane fractions from primary fibroblasts of both a severe type I SMA patient and an unaffected individual. By western blot analysis, we found that the SMA condition reduces the amount of ribosomes in plasma membrane compartments (Fig. 7A). In these cells, we also tested DRM fractions, confirming similar results obtained in transfected fibroblasts (Fig. S4B). Furthermore, by performing a western blot analysis, we showed that the ATP recovery treatment switched on mTOR activity to a similar degree in unaffected and SMA-affected fibroblasts (Fig. 7B). However, protein synthesis rate downstream of mTOR activation (stimulated) was reduced in SMA fibroblasts by ∼25%, as evaluated by the SUnSET assay (Fig. 7C). No significant differences were observed in unstimulated conditions. Finally, we also subjected the SMA fibroblasts to an ATP depletion and recovery assay in order to test their ability to undergo actin cytoskeleton remodelling. By probing the actin network with phalloidin, we observed that SMA fibroblasts failed to rearrange actin filaments in stimulating conditions (Fig. 7D, stimulated). This was clearly evident by defective lamellipodia formation and irregular filopodial protrusions. As expected, in non-stimulating conditions, SMN deficiency impacted on cytoskeleton organization to a lesser extent (Fig. 7D, unstimulated). The results obtained in SMA patient fibroblasts confirm the idea that SMN is involved in membrane sequestering of ribosomes, and recapitulate main crucial aspects observed in transfected SMN-depleted cells.
It is still unclear why motoneurons, more than any other cell type, need high levels of SMN protein for their survival. Taking into account the housekeeping role of SMN in RNA metabolism, the idea is that SMN expression levels could become crucial depending on the phenotypic context in which this protein operates. Two peculiarities of motoneurons are of particular interest: (1) the large extension and polarity, and (2) the neuromuscular junction (NMJ), a specialized membrane domain, which physically and functionally interconnects the motoneuron to muscle. Both these features are structurally governed by dynamic rearrangements of actin filament, which in turn require local protein synthesis of key regulatory factors. Consistent with the above hypothesis, the interaction map of SMA modifier genes identifies three specific pathways in which SMN can act: actin remodelling, local translational control and endocytosis (Dimitriadi et al., 2010). Of note, membrane signalling dictates each of these pathways, and their crosstalk drives the establishment of cell polarity. This scenario suggests a new context in which SMN can work, and raises some important questions: can SMN mediate the functional interplay of cellular pathways specified by SMA modifier genes? Can an exacerbated cell polarity represent the ‘Achilles' heel’ for SMN-depleted cells? Moreover, and most importantly: how could SMN, an RNA regulatory protein, interconnect membrane signalling and the actin network? To investigate in vitro some of these aspects, we took advantage of an ATP depletion and recovery assay. This method, based on the action of metabolic inhibitors on the actin cytoskeleton, stimulates and recapitulates typical changes of actin filament underlying the establishment of cell polarity (Svitkina et al., 1986). For this specific issue and approach, the fibroblast is the best-performing cell.
In our experience, checking for subcellular distributions of a protein during a specific event can help researchers to understand its function. Previous attempts to map the intracellular localization of SMN have revealed that this protein is mainly present in the nucleus and perinuclear cytoplasm of all cell types. This localization has been linked to the canonical role of SMN in snRNPs biogenesis. However, the presence of SMN granules in subcellular regions, such as the axonal growth cone, filopodia-like structures and the NMJ (Fan and Simard, 2002; Fallini et al., 2012; Dombert et al., 2014), implies an involvement of SMN beyond snRNP assembly. Here, we provide evidence in line with this assumption. First, when we stimulate fibroblasts to reassemble the actin cytoskeleton, SMN appears to be strongly redistributed at the cell protrusions, where the biogenesis of actin network begins (Svitkina et al., 1986). Both components of translation machinery and newly synthesized proteins are present at the same substructures. Indeed, it has been already reported that foci of active protein synthesis are present at the leading edge of fibroblasts during migration (Willett et al., 2013). This suggests that the molecular context in which the SMN function takes place might be related to local translational control. In more detail, colocalization studies show that SMN contacts the translational hotspots underlying membrane remodelling. A colocalization between SMN and polyribosomes has been previously reported (Béchade et al., 1999). However, in that study a direct link to actin dynamics was not proposed. Next, SMN relocalizes to membrane protrusions in a cellular context in which the mTOR–p70S6K signalling cascade couples to actin remodelling. Regarding SMN localization, we also obtained similar results in fibroblasts stimulated with serum or bFGF. These findings highlight two important unique functions of SMN: (1) that cell surface remodelling combines with changes in cytoplasmic distributions of SMN, and (ii) that SMN accumulates at the cell periphery in an activity-dependent manner. In this framework, receptor tyrosine kinases (RTKs) could be a triggering component for SMN trafficking, given that actin dynamics require RTK activation. Further studies will focus on this aspect.
Surprisingly, more than a signalling-dependence for SMN distributions, here we found a physical link between SMN and the plasma membrane. Immunofluorescence images support SMN–membrane linkage mostly in stimulated cells, in which the cytoskeleton organization unmasks SMN dots along the cell perimeter. Biochemical approaches reveal that SMN is always detectable in plasma-membrane-enriched fractions from fibroblasts, independently of their activity to extend membrane protrusions. In more detail, by performing DRM isolation, we observed that SMN is partially associated with lipid-raft-containing fractions, which retain more SMN protein following ATP recovery treatment. The notion that SMN associates with lipid rafts could be coherent with published works suggesting a link between lipid rafts and translational control. In particular, the netrin receptor DCC, which anchors protein synthesis machinery in neurons (Tcherkezian et al., 2010), partially localizes at lipid rafts and this association is required for axon outgrowth (Hérincs et al., 2005). Importantly, deficiency of the RNA-binding protein FMRP changes lipid rafts properties (Kalinowska et al., 2015). In contrast to SMN, ribosomal subunits are present only in plasma membrane preparations from unstimulated fibroblasts. In agreement with Flanagan and co-workers, we confirm that ribosomes are released from the membrane compartment in an activity-dependent manner (Tcherkezian et al., 2010). However, further studies are necessary to elucidate the connection between the SMN fraction that accumulates in close proximity to the plasma membrane and the SMN fraction found in direct association with it. In this regard, the demonstration of a bidirectional movement of fluorescently tagged SMN granules in the axon could be relevant (Zhang et al., 2003). Notably, these mobile SMN granules do not include core-splicing factor proteins. It will be interesting to assess whether the trafficking of these granules to and from the plasma membrane contributes to the rapid turnover of the membrane SMN. In this context, an important novelty of our study is that SMN associates with caveolin-1. Interestingly, cellular events triggering membrane expansions increase intracellular contacts between SMN and caveolin-1. It is of note that an intimate connection between SMN and caveolin-1 has already been suggested (Dimitriadi et al., 2010). Hart and colleagues observed that proteins related to endocytosis, including caveolin-1, unexpectedly predominate in the interaction map of SMN modifier genes. Consistent with this, defects in vesicle endocytosis and recycling have been observed at the NMJ in mice severely deficient for SMN (Kong et al., 2009). More investigations are needed to understand the functional and physical relationship between SMN and caveolin-1, as well as other factors, pertinent to endocytosis. Indeed, the interdependence of the mRNA translational control with the endocytosis pathway is still enigmatic. In this context, two of our results become attractive: (1) ribosomal proteins are sequestered like SMN in caveolin-1 enriched domains; and (2) caveolin-1 mediates anchoring of ribosomal proteins at the plasma membrane. By performing loss-of-function experiments, we found that SMN-depleted fibroblasts fail to rearrange F-actin and, consequently, to generate membrane protrusions. The non-oriented actin cytoskeleton, occurring in the absence of SMN, impairs the recruitment of β-actin transcripts at the leading edge of the cell. In this regard, we do not exclude that perturbations of β-actin-mRNA-binding proteins linked to the SMN fate might produce cumulative effects (Fallini et al., 2012). Next, in our system, SMN knockdown cells retain the ability to activate the mTOR pathway. Despite the mTOR pathway switch on, SMN loss of function reduces the protein synthesis rate by ∼25%. Of note, in non-stimulating conditions, the translational efficiency is almost unaffected by SMN loss of function. This result suggest a possible functional role for SMN in a subset of protein synthesis downstream of mTOR. Finally, a recent work has shown that localized mTOR-dependent protein synthesis is a prerequisite for the membrane expansion during axon outgrowth (Gracias et al., 2014). Given that SMN controls different tasks within RNA metabolism, the effect of its reduction could not only depend on SMN localization. However, this does not negate the role of SMN at the cell protrusions. Importantly, we show that SMN deficiency depletes the plasma membrane of ribosomes, and this correlates with the altered distribution of caveolin-1 at DRMs. Of note, this result was also confirmed in SMA-affected fibroblasts.
Collectively, our findings are compatible with a model in which SMN arranges the translational platform anchored to the plasma membrane, whose components are engaged for local protein synthesis underlying membrane remodelling (Fig. 7E). Future works should be planned to understand how SMN contributes to the packaging and/or delivery of membrane proteins competent for local translation. In addition, our study shows that a cytoplasmic pool of SMN colocalizes with S6 near the plasma membrane. It will be interesting to explore in what direction, from and/or to the plasma membrane, it occurs. This study opens a novel and exciting scenario in SMN investigations providing the opportunity not only to advance in functional characterization of the ubiquitous SMN protein, but also to understand how this new aspect of SMN might be crucial for motoneuron activity.
MATERIALS AND METHODS
Antibodies and reagents
The following antibodies were used: anti-SMN mouse monoclonal antibody (cat. no. 610647, BD Transduction Laboratories; work dilution for western blotting, 1:10,000; for immunofluorescence, 1:150); anti-SMN rabbit polyclonal antibody (cat. no. sc-15320, Santa Cruz Biotechnology; work dilution for immunofluorescence, 1:150); anti-S6 rabbit polyclonal antibody (cat. no. 2217, Cell Signaling Technology; work dilution for western blotting 1:1000; for immunofluorescence, 1:100); anti-phospho-S6 (Ser235 or Ser236) rabbit monoclonal antibody (cat. no. 4858, Cell Signaling Technology; work dilution for western blotting 1:1000; for immunofluorescence 1:100); anti-phospho-p70S6K (Thr389) rabbit polyclonal antibody (cat. no. 9234, Cell Signaling Technology; work dilution for western blotting 1:1000; for immunofluorescence, 1:100); anti-KSRP rabbit polyclonal antibody (cat. no. 5398, Cell Signaling Technology; work dilution for immunofluorescence, 1:100); anti-α-tubulin mouse monoclonal antibody (cat. no. T6074, Sigma-Aldrich; work dilution for western blotting, 1:2000); anti-flotillin-1 rabbit polyclonal antibody (cat. no. F1180, Sigma-Aldrich; work dilution for western blotting, 1:1000); anti-Caveolin-1 rabbit polyclonal antibody (cat. no. sc-894, Santa Cruz Biotechnology; work dilution for western blotting, 1:5000; for immunofluorescence, 1:200); anti-Caveolin-1 mouse monoclonal antibody (cat. no. MAB5736, R&D Systems; work dilution for immunofluorescence, 1:200); anti-Caveolin-1 monoclonal antibody (cat. no. NB100-615, Novus Biologicals; work dilution for immunofluorescence 1:200); anti-L7 rabbit monoclonal antibody (cat. no. ab 172478, Abcam; work dilution for western blotting, 1:1000); anti-Rab5 rabbit polyclonal antibody (cat. no. ab 18211, Abcam; work dilution for western blotting, 1:500); anti-digoxigenin sheep polyclonal antibody (cat. no. ab 64509, Abcam; work dilution for immunofluorescence 1:100); anti-puromycin mouse monoclonal antibody (cat. no. MABE343, Millipore; work dilution for western blotting, 1:25,000; for immunofluorescence, 1:10,000); and anti-calreticulin rabbit polyclonal antibody (cat. no. ADI-SPA-600, Stressgen Biotechnologies Corporation; work dilution for western blotting, 1:3000). The secondary antibodies conjugated to horseradish peroxidase were purchased from Jackson Immuno Research Laboratories and used at a dilution of 1:10,000. The Alexa-Fluor-488 or Alexa-Fluor-594-conjugated secondary antibodies were purchased from Life Technologies and used at a dilution of 1:250. The anti-Sm mouse monoclonal antibody Y12, was a gift from Mauro Cozzolino (CNR, Institute of Translational Pharmacology, Italy). Anti-β1-integrin rabbit polyclonal antibody, was a gift from Rita Falcioni (Regina Elena Cancer Institute. Italy). The SytoRNASelect probe and Alexa-Fluor-594-conjugated phalloidin were from Life Technologies.
Cell cultures, treatments and transfections
hTert-immortalized human fibroblasts, obtained from Silvia Soddu (Regina Elena Cancer Institute. Italy) (Rinaldo et al., 2012), were cultured in Dulbecco's modified Eagle's medium (DMEM, Gibco) supplemented with heat inactivated 10% FBS (Gibco), penicillin-streptomycin (Gibco) and GlutaMAX (Gibco), in a 5% CO2 humidified atmosphere, at 37°C. For serum stimulation, fibroblasts were cultured for 16 h in free-serum medium and then were incubated in presence of 15% fetal bovine serum for 30 min. For bFGF stimulation (Prepotech), starved fibroblasts were treated with bFGF (50 ng/ml) for 15 min. Human fibroblasts from the SMA type I patient (GM00232) and healthy control (GM08333) were obtained from the Coriell Institute for Medical Research (Camden, NJ, USA), and cultured in DMEM medium supplemented with 10% FBS, penicillin-streptomycin, and GlutaMAX, in 5% CO2 humidified atmosphere, at 37°C. For EGF stimulation (Prepotech), starved HeLa cells were treated with EGF (50 ng/ml) for 5 min. For knockdown experiments, cells were transfected with a combination of three siRNA-27 duplexes targeting the human SMN1 gene or the human caveolin-1 gene. Universal scrambled siRNA duplex was used as negative control (OriGene). For GFP transfection experiments, hTert-immortalized human fibroblasts were transfected with pEGFP plasmid (Clontech). Lipofectamine 2000 (Life Technologies) was used as the transfection reagent, according to the manufacturer's instructions. Cells were harvested after 48 h post transfection. Informed consent was obtained for all tissue donors, and all clinical investigation have been conducted according to the principles expressed in the Declaration of Helsinki.
ATP depletion and recovery assay
The ATP depletion and recovery assay was performed as described previously (Svitkina et al., 1986). Cells were incubated in phosphate-buffered saline (PBS) supplemented with 1 mM CaCl2, 1 mM MgCl2 and 20 mM NaN3 for 1 h. This treatment causes a rapid depletion of the cellular ATP, inducing a reversible disassembly of actin filaments. NaN3-containing buffer was then replaced with fresh medium supplemented with heat inactivated 10% FBS for 30 min to allow ATP recovery, leading to a rapid restoration of actin cytoskeleton, which occurs as a synchronous burst of membrane protrusions.
Immunofluorescence and confocal laser scanning microscopy
Cells were fixed with 4% formaldehyde in PBS, permeabilized in 0.2% Nonidet P40 (Boehringer Mannheim) for 10 min, and blocked with 1% BSA in PBS at room temperature. Samples were incubated sequentially with the appropriate primary and secondary antibodies. Slides were mounted with ProLong with Dapi (Life Technologies), and examined by conventional epifluorescence microscope (Olympus BX51; Milano, Italy). Images were captured by a SPOT RT3 camera and elaborated by IAS software. For colocalization analysis, slides were examined by acquiring a z-series of 25 optical sections at 0.8 μm intervals, with a ×63/1.2 NA water-corrected oil immersion CApochromat lens, and a digital zoom of 3, with the confocal system TCS-SP5 (Leica Microsystems GmbH, Wetzlar, Germany). Orthogonal projections (of an x- or z-plane corresponding to a point on the y-axis) were calculated from a 20 μm z-stack. The quantification of SMN and S6, and SMN and phospho-S6 colocalization over membrane protrusions and cytoplasm was performed as previously described (Tcherkezian et al., 2010).
Western blot analysis
Protein extracts were loaded on pre-cast NuPAGE 4–12% gels (Life Technologies) and transferred onto nitrocellulose membranes (GE Healthcare; Milano, Italy). Immunodetection of the reactive bands was revealed by chemiluminescence (ECL kit, GE Healthcare).
Preparation of plasma membrane-enriched fractions
The preparation of plasma-membrane-enriched fraction was performed as previously described (Vachon et al., 1987) with little modification. Cells were lysed in buffer A (5 mM Tris-HCl pH 7.4, 1 mM EGTA, 1 mM DTT and 320 mM sucrose). Extracts were passed five times through a 26G needle and centrifuged at 1000 g for 10 min at 4°C. The supernatant was kept and the pellet was quickly vortexed in the presence of the original volume of lysis buffer and centrifugated at 1000 g for 10 min at 4°C. The two supernatants were pooled and centrifuged at 24,000 g for 20 min at 4°C in a Beckman SW41 rotor. The supernatant was discarded and the pellet was resuspended in 12 ml of buffer B (5 mM Tris-HCl pH 7.4, 1 mM EGTA and 1 mM DTT), and centrifuged at 24,000 g for 30 min at 4°C at in a Beckman SW41 rotor. The supernatant was discarded and the pellet was processed in lysis buffer (100 mM Tris-HCl pH 8.0, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100). The protein extract was processed for western blot analysis.
The isolation of DRM fractions, containing lipid rafts, was performed as previously described (Clement et al., 2010).
For co-immunoprecipitation experiments, plasma-membrane-enriched fractions were resuspended in immunoprecipitation buffer [50 mM Tris-HCl pH 7.4, 250 mM NaCl, 5 mM EDTA, 0.1% Triton X-100, 5% glycerol and complete protease inhibitor cocktail (Roche)]. Extracts were passed five times through a 25G needle, incubated on ice for 15 min, and clarified at 9000 g for 15 min at 4°C. 250 µg of extract was immunoprecipitated in immunoprecipitation buffer with 1 µg of caveolin-1 polyclonal antibody or SMN monoclonal antibody, pre-adsorbed with anti-rabbit-IgG- or anti-mouse-IgG-conjugated magnetic beads (New England BioLabs) for 3 hours at 4°C. The same amount of non-specific IgG was used as control. After five washes with immunoprecipitation buffer, complexes were eluted by boiling in Laemmli's buffer and analysed by SDS-PAGE on 10% polyacrylamide gels followed by immunoblotting.
In situ proximity ligation assay
Cultures, treated as indicated were immediately fixed in 4% PFA for 15 min and thereafter subject to in situ PLA using Duolink In Situ Detection Reagents Red kit (DUO92008, Sigma-Aldrich), according to the manufacturer's instructions. Different combinations of primary antibodies to SMN and caveolin-1 were used. Then, PLA signals were detected by a TCS SP5 confocal laser scanning microscope (Leica Mycrosystems) using an 63×1.35 NA oil immersion objective. High-resolution images were acquired as a z-stack with a 0.5 μm z-interval. Two-dimensional projections at maximum intensity of each z-series were generated with the LAS AF software platform (Leica Microsystems). High-resolution (taken using a 63×, 1.35 NA objective) images from single scans were analysed by ImageJ (NIH) to calculate the density of PLA puncta.
Primary neuron cultures and treatments
Spinal cords were obtained from embryonic day 13 fetal rats. After removal of the dorsal root ganglia, spinal cords were dissected, washed with 5 ml of Earl's Balanced Salt Solution (Gibco), and centrifuged for 2 min at 150 g. The tissue was resuspended and incubated for 15 min at 37°C with 0.02% trypsin followed by addition of DNase I (80 μg/ml) and trypsin inhibitor (0.52 mg/ml). Digested tissues were mechanically dissociated and centrifuged at 150 g for 10 min. The dissociated cells were plated at a density of 12×104 cells/cm2 in neurobasal media supplemented with B27 (Gibco). For BDNF stimulation, neuron cultures at 3 days in vitro (DIV) were incubated for 30 min in neurobasal media, and then stimulated with 10 ng/ml BDNF (Prepotech) for 15 min. Fixed neurons were processed for the in situ PLA assay.
Preparation of nick-translated probe
One digoxigenin-labelled oligonucleotide probe was used to detect human β-actin mRNA. The probe was obtained by PCR amplification of human cDNA using the following oligonucleotides: forward, 5′-ACACCCGCCGCCAGCTCACCATGGA-3′; reverse, 5′-CCAGGTCCAGACGCAGGATGGCATG-3′. Digoxigenin (DIG)-labelled DNA probe was generated using a DIG-Nick Translation Mix Kit (Roche) according to the manufacturer's instruction.
Fluorescent in situ hybridization
Fibroblasts, subjected to ATP depletion and recovery assay, were processed for immunofluorescence, as above reported. FISH was performed as previously described, with little modification (Welshhans and Bassell, 2011). Briefly, cells were fixed in 4% formaldehyde, rinsed three times in PBS containing MgCl2, permeabilized in 0.2% Nonidet P40 for 5 min and equilibrated in 1× SSC and pre-hybridized with hybridization buffer (40% formamide, 10% dextran sulfate, 4 mg/ml bovine serum albumin, 2× SSC, 1× PBS) for 1 h at 37°C. Then, cells were incubated overnight with DIG-labelled oligonucleotide probe in hybridization buffer at 37°C, rinsed with 40% formamide and 1×SSC (two times, 20 min each) at 37°C, five times with 1× SSC (two rinses for 15 min each and three rinses for 5 min each) at room temperature. Cells were washed three times with PBS containing MgCl2 and incubated with anti-digoxigenin antibody for 2 h at room temperature, followed by Alexa-Fluor-594-conjugated secondary antibody incubation, for 45 min at room temperature, and mounted with ProLong with Dapi (Life Technologies). Stacks of images of membrane protrusions were acquired, and images were analysed using ImageJ.
The SUnSET assay was performed as previously described (Schmidt et al., 2009). Briefly, cells were subjected to the ATP depletion and recovery assay, in the presence or absence of 10 nM rapamycin. 10 μM of puromycin was added for an additional 10 min following the ATP recovery. Cells were harvested and processed for western blot analysis by using anti-puromycin antibody. Immunoblotting images were quantified by ImageJ. For experiments in which an immunofluorescence step was combined to the SUnSET assay, after ATP depletion and recovery assay, cells were incubated with 10 μM puromycin for 5 min. Samples were washed twice in ice-cold PBS supplemented with 355 μM cycloheximide, and then incubated on ice in PBS containing 0.002% Nonidet P-40 (Boehringer Mannheim), for 30 s, followed by rapid washing in ice-cold PBS. Cells were then fixed and processed for immunofluorescence as described above.
Data are presented as mean±s.d. or s.e.m. Statistical significance between groups was determined using the GraphPad Prism software. The unpaired t-test was performed to assess the significance; P<0.05 was considered statistically significant.
We thank Dr T. Costa and Dr M. Caruso for helpful discussion and critical reading of the manuscript. We thank Dr I. Pagano for technical assistance. We are also grateful to Dr R. Butler for the careful editing of the manuscript.
The authors declare no competing or financial interests.
M.G.D.C. designed experiments and wrote the manuscript. F.G. performed ATP depletion and recovery assays, plasma membrane preparations, transfection, western blotting, and co-immunoprecipitation and immunofluorescence experiments. C. Pisani performed FISH analysis, western blotting. M.T.C. performed primary neuron cultures and treatments. A.B., A.O. and T.I. performed data collections. S.F.-V. performed confocal microscopy. N. Canu performed the PLA analysis. L.M., N. Corbi and M.A.-T. provided reagents, materials and tools. C. Passananti supervised and provided guidance and crucial input into the study. All authors read and approved the final version of the manuscript.
This research was supported by FARMM onlus; Filas-Regione Lazio funding ‘Sviluppo della Ricerca sul Cervello’; and Association Française contre les Myopathies [grant number 15586 to L.M.].
Supplementary information available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.176750/-/DC1
- Received June 30, 2015.
- Accepted December 29, 2015.
- © 2016. Published by The Company of Biologists Ltd