ABSTRACT
Skeletal muscle (SKM) differentiation is a highly regulated process leading to the formation of specialised cells with reorganised compartments and organelles, such as those of the early secretory pathway. During SKM differentiation the Golgi complex (GC) redistributes close to the nuclear envelope and in small distinct peripheral structures distributed throughout the myotube. Concurrently, GC elements closely associate with endoplasmic reticulum-exit sites (ERES). The mechanisms underlying this reorganisation and its relevance for SKM differentiation are poorly understood. Here, we show, by time-lapse imaging studies, that the changes in GC organisation involve GC fragmentation and redistribution of ERES with the formation of tightly associated GC–ERES units. We show that knockdown of GM130 (also known as GOLGA2) or p115 (also known as USO1), two regulators of the early secretory pathway, impairs GC and ERES reorganisation. This in turn results in inhibition of myotube fusion and M-cadherin (also known as CDH15) transport to the sarcolemma. Taken together, our data suggest that the correct reorganisation of the early secretory pathway components plays an important role in SKM differentiation and, thus, associated pathologies.
INTRODUCTION
Myogenesis is a complex and highly regulated process that results in the formation of multinucleated cells named myotubes (Braun and Gautel, 2011; Buckingham and Rigby, 2014), which further differentiate in mature skeletal muscle (SKM) cells called myofibres. After the initiation of the myogenic differentiation programme (Braun and Gautel, 2011; Buckingham and Rigby, 2014), the subsequent phases of myogenesis are characterised by cell fusion accompanied by the reorganisation of pre-existing compartments and organelles (Franzini-Armstrong, 2004; Ralston, 1993), including vesicular trafficking components (Kaisto and Metsikko, 2003; Rossi et al., 2008; Volpe et al., 1992).
The formation of multinucleated myotubes is a multistep process entailing initial recognition and adhesion between myoblasts, their alignment, and finally membrane breakdown and fusion (Doberstein et al., 1997). This process requires the interplay of several factors, among them the protein M-cadherin (also known as CDH15) has been proposed to exert a pivotal role by mediating cell–cell adhesion (Donalies et al., 1991). M-cadherin is member of the cadherin family, a group of transmembrane glycoproteins, that mediate Ca2+-dependent homophilic cell–cell adhesion and play a crucial role during SKM differentiation (Donalies et al., 1991). M-cadherin expression is upregulated in SKM cell lines during myotube formation, and declines after completion of this process. Conversely, M-cadherin inhibition impairs myoblast fusion (Pouliot et al., 1994; Zeschnigk et al., 1995).
In most eukaryotic cells, three main endoplasmic reticulum (ER) subcompartments can easily be recognised: rough ER, smooth ER and ER-exit sites (ERES). However, in adult SKM cells the distinction of the three subcompartments is more complicated. During early SKM cell differentiation smooth ER undergoes a series of modifications that lead to the formation of the sarcoplasmic reticulum (SR), responsible for Ca2+ homeostasis and present in close association with transverse tubules for muscle contraction (Rossi et al., 2008). Interestingly, in this process, the early secretory pathway components also change their distribution. The Golgi complex (GC) undergoes architectural modifications that result in the formation of a perinuclear rim and dispersed elements along the myotube (Kaisto and Metsikko, 2003; Rossi et al., 2008; Volpe et al., 1992). Later, in muscle fibres, the GC perinuclear rim is lost and the GC appears as very small elements, localised nearby the nucleus and in the fibre body in a fibre-type dependent manner (Ralston et al., 1999). These GC elements are very small compared to other cell types; however, each element maintains the typical stacked organisation (Ralston et al., 1999). The GC reorganisation coincides with the remodelling of the microtubule network and changes of the microtubule-organising centre from being a classic centrosome into a combination of a perinuclear belt and centrosomal remnants in the cell body (Tassin et al., 1985a, b; Zaal et al., 2011). Interestingly, GC reorganisation involves also its close localisation with ERES that, in differentiating muscle cells, progressively cluster in proximity of the GC (Ralston et al., 1999; Lu et al., 2001). The organisation of GC in peripheral small elements has been suggested to occur because a relocation of the ERES near to the microtubule minus end induces the de novo assembly of a GC structure at this location (Lu et al., 2001). However, more recently it has been hypothesised that GC reorganisation could entail more complex mechanisms, such as the progressive modification of existing structures (Zaal et al., 2011). The mechanism underlying this transition, and the influence of such unique morphology on SKM differentiation, has not yet been elucidated.
The growing opinion that GC and ERES are capable of a reciprocal fine regulation (Glick, 2014; Ronchi et al., 2014), and the increasing evidence of a strong correlation between human diseases and the shape and function of the GC (Bexiga and Simpson, 2013; Percival and Froehner, 2007) oriented our work at further understanding the mechanisms responsible of the close correlation between GC and ERES and its importance for SKM differentiation and fusion.
Our results show that the changes in GC organisation in differentiating SKM cells involve GC fragmentation and redistribution of ERES with the formation of GC–ERES units. Knockdown experiments showed that GM130 (also known as GOLGA2) and p115 (also known as USO1) are involved in GC and ERES localisation in differentiating cells, and can inhibit myotube fusion by interfering with M-cadherin transport to the plasma membrane. Taken together, our data suggest that the correct reorganisation of the early secretory pathway components plays an important role in SKM differentiation.
RESULTS
In myotubes, the GC and ERES form dynamic units by redistributing predominantly pre-existing GC membranes
Aiming to further elucidate the mechanisms underlying the reorganisation of GC peripheral structures in differentiating SKM cells, and to understand to what extent they form de novo or from pre-existing membranes, we analysed GC and ERES distribution in differentiating C2C12 cells by immunostaining and live-cell imaging. As previously reported, C2C12 cells undergo differentiation upon medium exchange, giving rise to a cell population with different degrees of differentiation, where the GC and ERES reorganise into a specific pattern (Lu et al., 2001). We first analysed the reciprocal localisation of ERES and GC in cells exposed for different times to differentiation medium by quantifying the colocalisation levels of Sec13 and GM130, markers for ERES and GC respectively (Fig. 1A–C). The mean Pearson coefficient of the Sec13 and GM130 labelling, which was used as measure for colocalisation, increased with the time when the cells were exposed to differentiation medium (Fig. 1B) confirming that ERES and GC become intimately associated during muscle differentiation. Consistent with this, by analysing the relative frequency of the Pearson coefficient, we could observe a shift of the cell population towards higher Pearson coefficients after 24 and even more after 48 h of exposure to differentiation medium, compared to undifferentiated cells (Fig. 1C). We qualitatively scored GC shape modifications and changes in ERES localisation during differentiation (Fig. 1A). We subdivided undifferentiated and differentiating cells in four distinct phases based on their GC and ERES distribution. In phase 0, where cells have not yet started to differentiate based on the absence of the differentiation marker α-actinin, the GC forms a ribbon located at one side of the nucleus (Fig. 1A, undifferentiated cells are marked with a white circle). In phase 1, cells start to express visibly α-actinin and to redistribute the GC around the nucleus (Fig. 1A, phase 1 cells are marked with a white square, arrowheads point towards the perinuclear rim). In both phase 0 and phase 1, ERES localise throughout the cell as discrete structures, the majority of which are not localised near to GC structures as highlighted in the line scans in Fig. 1A (Sec13 profile in green and GM130 in red). In phase 2 (Fig. 1A, cells marked with an asterisk), cells show a more complete redistribution of the GC around the nuclear envelope (Fig. 1A, arrowheads) and some emerging peripheral structures (Fig. 1A, double arrowheads point towards the peripheral structures). At the same time ERES concentrate close to the GC, and the number of structures diffused in the rest of the cell appears reduced as shown by the line scan of the fluorescent signal of the GC and ERES. In phase 3 cells form elongated myotubes (Fig. 1A, cell with double asterisk) with several GC peripheral elements throughout the cell (Fig. 1A, cell with double asterisk, see double arrowheads) and ERES are predominantly localised close to these GC structures.
During SKM differentiation, the ER undergoes a series of modifications such as the expression of muscle-specific SR proteins and the formation of specialised domains (Ralston, 1993; Rossi et al., 2008). Consistent with data previously reported, the detection of the ER marker protein disulfide-isomerase (PDI, also known as P4HB) in Fig. 1D further shows that the transformation from undifferentiated myoblasts to differentiated myotubes involves the transition from a reticular dense ER to a thinner ER, which is distributed throughout the entire cell, mainly aligned along the myotube axis (Ralston, 1993)
Live-cell imaging of differentiating cells transfected with the GC marker Golgi–CFP and the ERES marker Sec23a–EYFP allowed us to monitor the dynamics of ERES and GC over time (Fig. S1). GC and ERES fluorescence signals were segmented, and the ratio between ERES fluorescence in the GC and the total ERES fluorescence were analysed over time. Quantifying the ERES fluorescence and superimposing with the GC fluorescence value showed that this value remained almost constant over time in phase 0 cells (Fig. S1, cell A). In contrast, in a cell passing from phase 0 to 2 (Fig. S1, cell B) the proximity of the two markers increased steadily, approaching a plateau level within 5 h. Cells in phase 3 (Fig. S1, cell C) steadily maintain the plateau level.
The transition from phase 1 to phase 2 was characterised by the formation of several new peripheral GC structures. To address the question as to what extent these GC peripheral structures form de novo or from pre-existing GC membranes, we imaged phase 1–2 differentiating C2C12 cells for 15 min at 10 s time intervals (Fig. 2A–D; Movies 1–4). The analysis revealed a variety of events (Fig. 2A–E). We observed 2.4±1.2 GC splitting events per cell, where a GC element detaches from the pre-existing GC (Fig. 2A,B,E). Fig. 2A,B and Movies 1 and 2 show such an event with a peripheral GC element forming by detaching from the perinuclear region (Fig. 2A), or the splitting of a GC element into two separated elements (Fig. 2B). In addition 1.6±0.9 merging events per cell, consisting of the fusion of two already separated GC structures, were also observed (Fig. 2B,E). Only rarely (0.25 events/cell) did new GC elements appear without any relation to pre-existing GC structures, which is possibly attributable to de novo biogenesis of these GC structures (Fig. 2D,E). Taken together, these data support the hypothesis that the changes in GC distribution during the early phases of SKM differentiation predominantly involve the reorganisation of pre-existing GC elements. Interestingly, most GC elements observed were accompanied by a group of ERES close by (arrowheads in Fig. 2A,B and Movies 1 and 2 point towards a Golgi structure surrounded by ERES) that displaced in a syntonic fashion, indicating that these two entities move as a unit once it has formed (see arrowheads in Fig. 2A–D and Movies 1–4).
The data described above, together with previous work showing that GC and ERES organisation persist despite the lack of normal microtubule tracks in differentiating SKM cells (Zaal et al., 2011), and the growing evidence of a reciprocal regulation of GC and ERES (Glick, 2014; Ronchi et al., 2014), introduces the hypothesis that the GC also has a leading role in ERES positioning in differentiating C2C12 cells. To test this hypothesis, we perturbed the GC organisation by treating cells with the fungal metabolite brefeldin A (BFA) (Lippincott-Schwartz et al., 1989) and followed ERES behaviour in differentiating SKM cells. As shown in Fig. 3A,B and Movie 5, BFA treatment of phase 2 or 3 C2C12 cells resulted in the disappearance of GC structures. Most of the perinuclear GC structures close to the nuclear envelope (Fig. 3A, see arrows and fluorescence line scans) and most peripheral GC elements disappeared, indicative of the well-described re-absorption of GC enzymes to the ER after BFA treatment (Lippincott-Schwartz et al., 1989). Coincident with this, the perinuclear localisation of ERES changed significantly (Fig. 3A, fluorescence line scans), while bigger fluorescent dots appeared in the perinuclear region and throughout the cell (see arrowheads in Fig. 3A,B). These bigger Sec23a–YFP-positive structures showed a close proximity with the immunofluorescence signal for the GC matrix protein GM130 (Fig. 3B, see arrowheads and fluorescence line scan).
GM130 and p115 are required for the formation of ERES–GC units during C2C12 cell differentiation
The evidence described above and previous data showing that GM130 is involved in a membrane-tethering mechanism (Nakamura et al., 2010; Seemann et al., 2000) raised the question of whether GM130 could play a role in the organisation of GC–ERES units during SKM differentiation. Based on previous data from the literature demonstrating a direct interaction between GM130, p115 (Nakamura et al., 1997) and GRASP65 (also known as GORASP1) (Barr et al., 1998; Barr et al., 1997), we transfected C2C12 cells with siRNAs targeting the three proteins and analysed the effects on GC and ERES distribution both in undifferentiated and differentiated cells. In undifferentiated C2C12 cells, knockdown of GM130, p115 or GRASP65 resulted in no apparent changes in Golgi morphology (Fig. 4A). However, we observed that the number of ERES in proximity of the GC was reduced upon single knockdown of all the three proteins (Fig. S2C). In parallel, electron microscopy (EM) analyses showed minor differences in the ultrastructure of the GC upon depletion of the different proteins (Fig. S2A). With all treatments, the GC appeared stacked and the morphology was consistent with the observed GC structure in control cells. Quantitative analysis of the number of ERES surrounding the GC, measured by counting the number of ERES and normalising for the length of the ER sheets within ∼1 µm of the surface of the nearest GC, showed a reduction for all siRNAs (Fig. S2B), consistent with the immunofluorescence data (Fig. S2C). However, as previously reported for GM130 (Tangemo et al., 2011), none of the siRNA treatments showed an effect on ER to plasma membrane transport of the well-established transport marker, the viral protein VSVG as shown in Fig. 4B, suggesting that, in undifferentiated cells, knockdown of these proteins and the resulting morphological changes in ERES are not playing a major role for VSVG traffic.
When cells were exposed to differentiation medium, differentiating C2C12 cells started to express the muscle differentiation markers α-actinin, M-cadherin and junctophilin 2 (JP2, also known as JPH2) similar to cells treated with the negative control (Neg9) siRNA (Fig. 4D,E; see also Fig. S3).
Interestingly, once induced to differentiate, C2C12 cells treated with GM130 and p115 siRNAs showed profound effects in the organisation of the early secretory pathway components compared to control and GRASP65-depleted cells (Fig. 5A–D). In GM130-knockdown cells, the distribution of GC structures was clearly different from control cells (Fig. 5A). The GC appeared more compact with fewer peripheral elements and did not surround the perinuclear region. Quantification showed that the percentage of nuclei with perinuclear GC staining was significantly reduced from 46.8±3.5% in Neg9 siRNA treated cells to 24.4±5.4% in GM130 knockdown cells (mean±s.d.; graph in Fig. 5B). Interestingly, upon staining with the ERES-specific antibody anti-Sec13, we could observe that, in GM130-knockdown C2C12 cells, ERES were rarely concentrated at the perinuclear region or clustered in peripheral elements as occurs in Neg9-transfected cells (Fig. 5D), but rather were present throughout the cell. Depletion of p115 resulted in a phenotype very similar to the one observed in GM130-knockdown cells (see Fig. 5A–D). Colocalisation analyses with the nucleoporin p62 (also known as NUP62) (Fig. S4A), which localises at the nuclear envelope, revealed that GM130 and p115 depletion prevents the accumulation of ERES at the perinuclear rim (Fig. 5C, Fig. S4A). On the contrary, GRASP65 depletion did not cause major changes in GC or ERES distribution compared to control cells (Fig. 5A–D; Fig. S4A).
In order to understand whether the loss of clustering of ERES was specific or involved other vesicular systems, as COPI vesicles, we stained GM130- and p115-knockdown cells with EAGE antibody (Pepperkok et al., 1993). Interestingly, in contrast to what was observed for the distribution of ERES, both in GM130- and p115-knockdown cells, the vesicular protein complex COPI was accumulating in close proximity to GC structures (see Fig. S5) as in control cells.
As further support of our observations, we analysed GC and ERES morphology and distribution upon GM130, p115 and GRASP65 knockdown by means of electron microscopy. Consistent with the immunofluorescence analysis (Fig. 1A), electron tomography of the perinuclear region revealed morphological differences of the early secretory pathway between undifferentiated and differentiated C2C12 cells (Fig. 6A). In undifferentiated myoblasts, the perinuclear ER regions are normally flat (in Fig. 6A, left panel), while the perinuclear ER in differentiated cells is rough (Fig. 6A, right panel) with several invaginations facing the nearby GC structures (Fig. 6A, right panel, black arrows), which are attributable to ERES as previously demonstrated by Lu and co-workers (Lu et al., 2001). Indeed, when observed on thin sections, the evaginations or buddings of the perinuclear ER pointing towards the GC, are decorated with an electron-dense coat that can be ascribed to the COPII coat (see Fig. 6B, Neg9, black arrow) (Lu et al., 2001).
Interestingly, as shown in Fig. 6B, the analysis of thin sections obtained from siRNA-treated differentiated C2C12 cells revealed that GM130 and p115 knockdown induced the formation of disordered, partially un-stacked GC structures with only a few GC elements adjacent to the perinuclear ER. Moreover, consistent with the data obtained by ERES immunofluorescence labelling (Fig. 5D), the number of ERES on the perinuclear ER in close proximity to GC elements was considerably reduced in GM130- and p115-knockdown cells, compared to GRASP65 knockdown or control cells (Fig. 6C).
GM130 and p115 knockdown inhibits M-cadherin transport to the plasma membrane and cell fusion
Although the expression levels of muscle differentiation markers α-actinin, JP2 and M-cadherin appeared unaffected in GM130 or p115 knockdown cells (Fig. 4D,E), cell morphological features were significantly altered compared to GRASP65 knockdown or control cells (Fig. 7A,B). At 48 h after induction of differentiation, GM130- or p115-knockdown cells showed a less-elongated morphology, were mainly mono-nucleated and showed a reduced fusion index compared to control cells (Fig. 7B). In addition while M-cadherin strongly accumulated at the plasma membrane in control cells, GM130 or p115 knockdown drastically inhibited the plasma membrane transport of this muscle-specific protein involved in myoblast fusion (Fig. 7C). In differentiating C2C12 cells transfected with Neg9 or GRASP6-targeting siRNA, the M-cadherin-specific fluorescence intensity line plots showed two peaks, one close to the nuclear periphery (the nuclear envelope) and one at the cell periphery (the plasma membrane), with the plasma membrane peak exceeding the nuclear envelope one by almost factor of 2, consistent with an efficient ER-to-plasma membrane transport of M-cadherin (Fig. 7A, white arrowheads, see also line scans). In contrast, in cells treated with siRNAs targeting GM130 or p115, the plasma membrane-associated peak was drastically reduced or could not even be detected, and only the nuclear envelope-associated peak presented high fluorescence levels (Fig. 7A, white arrowheads, see also line scans). These data suggest that transport of M-cadherin from the ER to the plasma membrane in differentiating cells is strongly inhibited, with an accumulation of the marker in the nuclear envelope, thus indicating that ER exit of M-cadherin is most likely inhibited upon GM130 or p115 knockdown, and suggesting that these two proteins can affect the fusion capability by interfering with M-cadherin localisation at the sarcolemma.
Parallel experiments showed a significant reduction of the transport of the transport marker VSVG to the plasma membrane upon GM130 or p115 knockdown (Fig. 7D). Together with the observation that in undifferentiated cells GM130 or p115 knockdown had little to no effect on VSVG plasma membrane transport (Fig. 4B), this result suggests that while these two proteins may not be essential for transport of VSVG to the plasma membrane in undifferentiated C2C12 cells, they become crucial for the formation of the ERES–GC units to facilitate transport during SKM differentiation. Indeed, when cells are differentiating, they need to sustain a more-intense and/or specialised trafficking of molecules responsible for further cell differentiation, such as M-cadherin.
DISCUSSION
The present work has been aimed at a better understanding of the importance and the mechanisms underlying the reorganisation of ERES and GC during SKM differentiation.
Our data suggest that during SKM differentiation, ERES and GC associate as dynamic units, which form the typical peripheral GC–ERES elements previously described (Lu et al., 2001). Time-lapse analyses of this reorganisation indicates that the respective GC elements form from pre-existing Golgi membranes, rather than being synthesised de novo. Whether ERES are forming mainly around GC elements, or whether the GC attracts an increasing number of ERES while moving through the cell body could not be resolved by our time-lapse analyses and remains to be elucidated.
The organisation of GC–ERES that we have described here is comparable to the one previously described in plants (Brandizzi and Barlowe, 2013; Brandizzi and Barlowe, 2014; daSilva et al., 2004), and Drosophila melanogaster S2 cells (Kondylis and Rabouille, 2009), where the GC and ERES form intimately associated units.
Our data here show that knockdown of GM130 and p115 inhibits the formation of these units in differentiating cells and inhibits cell fusion by reducing M-cadherin transport to the plasma membrane. Interestingly, in undifferentiated cells, the depletion of GM130, p115 and GRASP65 has no apparent effect on the morphology of the GC and no major functional effects on the ER-to-plasma membrane transport efficiency of the established transport marker VSVG, but reduces the number of ERES surrounding it by a factor of almost 2. Also, GM130- or p115-knockdown cells are able to start their differentiation programme by expressing early differentiation markers such as α-actinin, JP2 and M-cadherin, similar to control cells. Therefore, we conclude that the p115 or GM130 knockdown-related reduction of ERES or any possible alterations of ERES–Golgi communication, which we have not so far observed in our experiments, at the undifferentiated stage, do not alter the ability of the cells to enter the early SKM differentiation programme. However, when differentiation progresses, and the ERES and GC undergo a more profound reorganisation in control or GRASP65-knockdown cells, we do see that GM130 and p115 depletion impairs this reorganisation, that VSVG transport to the membrane significantly reduced and that M-cadherin transport to the plasma membrane is strongly inhibited at the ER level. This suggests that the reorganisation of ERES and GC into stably associated units plays a crucial role for the efficient transport of differentiation-specific cargo during terminal SKM differentiation. Consistent with this is the fact that the degree of transport inhibition of VSVG appears rather marginal compared to the inhibition of M-cadherin, which may be explained by the possibility that the viral protein VSVG, in contrast to endogenous M-cadherin, may also be able to use p115- and GM130-independent routes to leave the ER.
Our results showing that GM130, p115 and GRASP65 depletion reduce the number of ERES in undifferentiated cells, while in differentiated cells GRASP65-knockdown cells behave like control cells may appear contradictory. However, in our view these data reinforce the muscle differentiation-specific role(s) of GM130 and p115. Moreover, these data underline the importance of the reorganisation of the early secretory pathway for the SKM differentiation process. Organising GC and ERES in units throughout the cell body may have the advantage that transport carriers, forming at ERES, do not need to travel long distances along microtubules to reach their target, as previously shown in fibroblast-like cells (Scales et al., 1997). Myoblasts are characterised by a packed cytoplasm where movement of larger long-range carriers (VTCs; Scales et al., 1997) may be difficult or even impossible, and thus the ERES–GC units can guarantee efficient ER-to-GC transport. Interference with their function, as found upon knockdown of GM130 or p115, thus results in transport inhibition to the plasma membrane and therefore lack of essential factors at that location, which are required for cell fusion. We can speculate that this could reflect the need for either an increased efficiency of transport in general or for specific transport routes that need to be activated during the differentiation process.
How GM130 or p115 are involved in the reorganisation of GC and ERES is still an open question. The two proteins could participate in tethering mechanisms linking ERES with GC structures through the interaction of GM130 and p115 (Seemann et al., 2000). Alternatively, GC structures could generate a signalling microenvironment that leads to ERES formation in the proximity of GC structures, a process that might be regulated by GM130 and/or p115. In fact, data obtained by BFA treatment and siRNA-mediated depletion suggest that, in differentiating SKM cells, ERES preferentially organise around a GC remnant structure associated with GM130. This is in agreement with earlier data (Ronchi et al., 2014) proposing that GC can induce ERES biogenesis. Our data therefore support the hypothesis that the GC exerts a hub function in the regulation of transport via the modulation of vesicle distribution, fusion, ERES assembly and localisation (Luini and Parashuraman, 2016; Ronchi et al., 2014) and thus confer, to the GC, a central role in the organisation of ERES in a physiological environment such as SKM differentiating cells.
Studies on skeletal muscles obtained from dystrophic models showed that the sorting and trafficking of correctly processed proteins and lipids from the GC to the sarcolemma are required for the maintenance of sarcolemmal integrity during the repeated contraction-relaxation cycles, preventing injury (Percival and Froehner, 2007). Until now, it has not been directly determined whether and how changes to GC morphology could interfere with SKM function.
Our data showing that GM130 and p115 knockdown can interfere with M-cadherin transport and affecting the index of fusion, demonstrate a functional link between GC–ERES organisation and the early phases of SKM differentiation and allocates to GM130 and p115 a central role in this process. Different from the observation that the reorganisation of GC and ERES can occur despite the lack of myoblast fusion (Ralston, 1993), our data suggest that GC and ERES organisation have an essential role in this process by participating in the targeting of M-cadherin to the plasma membrane. In turn, the lack of M-cadherin at the plasma membrane could itself trigger a cascade of effects on SKM differentiation. In fact, M-cadherin targeting at cell–cell contacts during myoblast fusion has been shown to be responsible of the Rac1 GTPase cascade activation (Charrasse et al., 2007), providing to M-cadherin a role in the activation of the promyogenic cascade.
Interestingly, recent work (Shamseldin et al., 2016) has attributed an important role to GM130 during muscle development by finding and characterising a homozygous frame-shift mutation in this protein. The mutation was found in a patient with a neuromuscular disorder who presented with developmental delay, seizures, progressive microcephaly and muscular dystrophy. The authors showed that the truncated protein is either not produced or is very unstable, suggesting that this is close to a null mutation where GM130 functions are largely lost. These results are consistent with our finding that GM130 and p115 are essential for the regulation of SKM differentiation in C2C12 cells.
In summary, our work shows for the first time that a coordinated reorganisation of the GC and ERES is required for the correct transport of M-cadherin during SKM differentiation, and that loss of the integrity of their association leads to defective differentiation. GM130 and p115 play a fundamental role in the formation and maintenance of GC–ERES units, therefore ensuring the correct delivery to the plasma membrane of proteins that are fundamental for SKM differentiation, such as M-cadherin.
MATERIALS AND METHODS
Plasmids and siRNA
The Golgi–CFP plasmid encoding for 81 amino acids of the precursor to the human β-1,4-galactosyltransferase (Clontech Laboratories), and Sec23a–EYFP plasmid (Forster et al., 2006) were used in live imaging experiments. ts-O45-G–CFP was obtained from Kai Simons (MPI-CBG, Dresden, Germany). The following Silencer Select siRNAs (Ambion, Thermo Fisher) were used: GM130 (s97432), p115 (s80014), GRASP65 (s92890), and Silencer Select Negative Control 9 siRNA (Neg9; s444246).
Antibodies
The following antibodies were used in immunofluorescence (IF) and western blotting (WB) according the manufacturer's instructions at the dilutions indicated. Primary antibodies were: mouse anti-α-actinin (clone EA-53, Sigma-Aldrich; IF, 1:1000; WB, 1:1000), mouse anti-PDI (clone 1D3, Enzo Life Sciences; IF, 1:200), mouse anti-M-cadherin (clone 12-G4, Merck-Millipore; IF, 1:200; WB, 1:500), goat anti-JP2 (Y-15, Santa Cruz Biotechnology; WB, 1:1000), mouse anti-GM130 (clone 35GM130, BD Biosciences; IF, 1:500), rabbit anti-GM130 (Ab52649, Abcam; WB, 1:500), mouse anti-giantin (clone G1/33, Enzo Life Sciences; IF, 1:1000), rabbit anti-giantin (Ab24586, Abcam, IF, 1:2000), rabbit anti-GRASP65 (Ab30315, Abcam; IF, 1:200; WB, 1:500), rabbit anti-p115 (13509-1-AP, Proteintech; IF, 1:200; WB, 1:500), rabbit anti-Sec13 (1:100 for IF; Verissimo et al., 2015), rabbit-anti Lamin B1 (Ab16048, Abcam; WB, 1:2000), mouse anti p62-FITC conjugated (clone 53, BD Biosciences; IF, 1:100), rabbit anti-EAGE (1:2000 for IF; Pepperkok et al., 1993), and mouse anti-VSVG (1:100 for IF; Kreis, 1986). The secondary antibodies used in immunofluorescence were purchased from Thermo Fisher Scientific; horseradish peroxidase (HRP)-conjugated secondary antibodies for western blot analysis were purchased from Sigma-Aldrich.
Cell culture and transfection
The mouse myogenic cell line C2C12 was obtained from the American Type Culture Collection (http://www.lgcstandards-atcc.org/). Cells were maintained in DMEM high-glucose culture medium (Gibco) supplemented with 10% (v/v) heat inactivated fetal calf serum (Gibco), 2 mM glutamine (Gibco), 1 mM pyruvate (Gibco), according to ATCC guidelines. For plasmid transfection, 100,000 cells were subcultured on gelatin-coated 35 mm glass bottom microwell dishes (MatTek), and, the following day, transfected with Lipofectamine Plus reagent (LifeTechnologies) and switched into differentiation medium (high-glucose DMEM containing 2% horse serum). Live-cell imaging was performed in phenol-free differentiation medium. For siRNA experiments, cells were plated on 35 mm dishes or eight-well Lab-Teks (Nunc) gelatin-coated plates (100,000 and 8000 cells per well, respectively) and transfected with siRNAs with RNAiMax reagent (LifeTechnologies) according to the manufacturer's instructions. Cells were allowed to grow in culture medium and switched to differentiation medium after 48 h. Samples were processed for western blot, immunofluorescence labelling or live imaging experiments at selected time points as indicated in the results section. The efficiency of knockdown of GM130, p115 and GRASP65, was routinely verified by western blot (Fig. 4C) and immunostaining.
VSVG–CFP transport assay
Cells transfected with siRNA were infected with a VSVG–CFP-encoding adenovirus, ts-O45-G–CFP, for 1 h at 37°C and incubated for 16 h at 39°C (Keller et al., 2001). To release the temperature-dependent transport block, cells were incubated at 32°C for 45 min and then fixed. Fixed cells were stained with an anti-VSVG antibody and imagined with a multi-position wide-field microscope Scan R (Olympus).
Immunofluorescence staining
Cells were fixed with 3% paraformaldehyde, in PBS (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4, 1.4 mM KH2PO4) for 20 min at room temperature. After extensive washing, cells were permeabilised with 0.5% Triton X-100 in PBS, washed, and incubated for 60 min in PBS with 5% goat serum in order to reduce unspecific binding. Cells were then incubated with primary antibody, extensively washed and then incubated with secondary antibodies following the manufacturer's instructions. Hoechst 33342 (Sigma-Aldrich) was added to the secondary antibody solution to stain nuclei.
EM analysis
Cells growing on MatTek dishes were fixed by addition of 2.5% glutaraldehyde (GA; Electron Microscopy Sciences) in 0.1 M cacodylate buffer. All the subsequent EM processing steps (OSO4, Uranyl Acetate, dehydration, Epon embedding) were performed using a PELCO Biowave Pro microwave processor (Ted Pella) as described in Schieber et al., 2010. The cells were flat-embedded. After polymerisation of the resin at 60°C for 48 h, the coverslips were removed, therefore exposing the cell monolayers at the block surface. The blocks were then sectioned with a Leica UC7 microtome. 70 nm or 300 nm thick sections were collected on Formvar-coated slot grids and imaged with a Philips Biotwin CM120 (for 2D imaging of thin sections) or a FEI Tecnai F30 (for tomography) electron microscope. For the image shown in Fig. 6A, tilt series were acquired and then the tomograms were reconstructed and segmented with the IMOD software package (Kremer et al., 1996). The quantification of ERES from EM micrographs reported in Fig. S2C was performed by counting the number of ERES and normalising for the length of the ER sheets within ∼1 μm of the surface of the nearest GC. Data are reported as number of ERES per µm ER in proximity to the GC structure. The quantification of ERES per µm for ER in the perinuclear region reported in Fig. 6 was determined by measuring the length of the perinuclear ER underlying a GC structure and counting the number of electron-dense budding structures in the segment.
Cell imaging and image analysis
Live imaging of differentiating C2C12 cells was carried out with a multi-position time-lapse spinning disc confocal microscope VOX (Perkin Elmer) equipped with an incubation chamber at 37°C with 5% CO2. Living cells were imaged in high-glucose DMEM without Phenol Red containing 2% horse serum. Videos obtained with the VOX microscope were processed and analysed with Volocity software.
3D images were processed with Volocity software to calculate the proximity of ERES to the GC (Fig. S1). GC and ERES fluorescence signals were segmented, and the ratio between ERES fluorescence in GC and the total ERES fluorescence was plotted along the time in the graph for several cells.
Specimens obtained by immunofluorescence staining, were imaged with the VOX spinning disc confocal microscope (Perkin Elmer), the LSM510-Meta (Zeiss) and the SP5 laser scanning confocal microscopes (Leica Microsystems) or with the multi-position time-lapse wide field microscope Scan R (Olympus) depending on the experiment. Colocalisation analyses reported in Fig. 1 were performed on images obtained with the VOX spinning disc confocal microscope (Perkin Elmer), deconvolved with Huygens software (Scientific Volume Imaging) and further analysed for colocalisation with the Volocity software.
Image analyses for the VSVG assay were manually performed with Fiji software by calculating the ratio between fluorescence at the membrane and total VSVG fluorescence.
Immunofluorescence determination of ERES colocalising with the GC in undifferentiated cells was performed with Volocity software. ERES and GC signals were segmented and the relative number of ERES superimposing with the GC was calculated by dividing the ERES colocalising with GC by the GC surface area.
ERES localisation at the perinuclear region of differentiated myotubes was calculated manually with Fiji software by drawing a ring around each nucleus at the level of p62 signal and an outer ring, adjacent to the first, that were both 4 pixels wide (resolution of the images 8.4 pixels/µm). The mean fluorescence was measured for each ring in the Sec13 channel. Results are expressed as ratio of the florescence signal between the inner and the outer ring (see Fig. S4B).
Western blots
For western blot analysis, C2C12 cells were transfected in 35 mm Petri dishes, after 48 h differentiation were washed in PBS and, whole-cell extracts were prepared by directly adding 200 μl of 2× Laemmli buffer to the dishes, and lysates were scraped, collected in tubes and Benzonase (Merck Millipore) and MgCl2 were added for 10 min at room temperature. Heat-denatured samples were run on NuPage Bis-Tris Gels (4–12%, Thermo Fisher Scientific) and transferred onto pre-activated PVDF membranes. The membranes were saturated in PBS containing 0.2% Tween 80 (PBS-T) and 5% dry milk, incubated with the primary antibodies, extensively washed with PBS-T, incubated with the appropriated HRP-conjugated antibodies, washed extensively and revealed with a Pierce western blotting detection kit (Thermo Fisher Scientific). Fiji was utilised to quantify the intensities of immunoreactive bands from three independent experiments; sample loading was normalised to anti-Lamin B1 immunoreactive bands.
Statistical analyses
Student's two tailed t-test analyses were performed using GraphPad Prism version 7.00 for Windows (GraphPad) and Microsoft Excel. Data are expressed as means±s.d.
Acknowledgements
Support of the EMCF and ALMF core facilities at EMBL-Heidelberg is acknowledged. We thank Fatima Verissimo and Nurlanbek Duishuev for helpful discussions, Christian Tischer and Yury Belyaev for invaluable help in image analysis. The authors also thank the members of the Rainer Pepperkok team for discussion and support.
Footnotes
Author contributions
Conceptualization: E.G., R.P.; Methodology: E.G., P.R.; Investigation: E.G.; Resources: R.P.; Data curation: E.G., P.R.; Writing - original draft: E.G., R.P.; Writing - review & editing: E.G., R.P., P.R.; Funding acquisition: E.G., R.P.
Funding
This work was supported by the European Molecular Biology Laboratory (EMBL); E.G. was funded by a Marie Curie IEF fellowship from the Seventh Framework Programme (grant 326691-TRAFFIC IN SKM) and iNEXT project 2435, funded by the Horizon 2020 programme of the European Commission (grant number 653706).
References
Competing interests
The authors declare no competing or financial interests.