Human macrophages rescue myoblasts and myotubes from apoptosis through a set of adhesion molecular systems

Stromal cells are microenvironmental cells defined by their ability to support the development, maintenance, proliferation and differentiation of tissue-specific cell types. For example, supportive stromal cells are crucially involved in haematopoiesis. The supportive stromal cell compartment is established before the formation of the haematopoietic system; it remains dynamic, instead of being static once established, reacting to extrinsic signals that can either damage or enhance stromal cell function and number (Muller-Sieburg and Deryugina, 1995). In non-haematopoietic tissues, organogenic processes are similarly driven by stromal cells that control parenchymal cell functions such as migration, proliferation, differentiation and programmed cell death (Charbord et al., 2000; Lapidot and Petit, 2002).

The mechanisms underlying supportive functions of stromal cells are incompletely understood. Stromal cells probably provide a complex molecular milieu that influences the behaviour of local stem cells, which have the choice of many fates, including quiescence, self-renewal, differentiation and apoptosis (Charbord and Moore, 2005). The molecules in this milieu are not well defined but probably include a mixture of cytokines, extracellular matrix components and cell adhesion molecules (Charbord and Moore, 2005).

The spectrum of stromal cells is debated and remains ill defined (Muller-Sieburg and Deryugina, 1995). Nevertheless, macrophages (MPs) constitute one definite cell type in this spectrum and were previously recognised to play a major role in tissue repair and homeostasis maintenance (Gordon, 1995). In addition to their classical functions, including microbicidal activity, phagocytosis and antigen presentation, these multifaceted cells efficiently support growth and differentiation of other cell types (Gordon, 1995; Laskin and Laskin, 2001). Their supportive effect was documented with respect to erythroblasts, hepatocytes, neurons, oligodendrocytes and myogenic cells (Blasi et al., 1987; Cantini et al., 1994; Sadahira and Mori, 1999; Takeishi et al., 1999; Polazzi et al., 2001; Gras et al., 2003; Chazaud et al., 2003b).

In contrast to bone marrow, where stromal cells are in place to support an ever-changing haematopoietic compartment, skeletal muscle is normally a stable tissue that uses newly recruited MPs to support post-injury muscle regeneration (McLennan, 1996; Pimorady-Esfahani et al., 1997; Lescaudron et al., 1999). In a previous study, we found that, upon activation, a small myogenic stem cell population residing beneath the basal lamina of each adult myofibre, the so-called muscle satellite cells (Mauro, 1961), can attract circulating monocytes and interplay with MPs to enhance their growth (Chazaud et al., 2003b). In vitro studies suggested that MPs can support myogenic precursor cell (mpc) growth by stimulating their proliferation through soluble mitogenic factors, and by preventing their apoptosis through direct cell-cell contacts involving unknown molecular systems (Chazaud et al., 2003b).

MP-derived soluble factors inducing mpc proliferation have long been reported (Cantini et al., 1994; Cantini and Carraro, 1995; Massimino et al., 1997), and the literature on myogenic cell growth factors is extensive (reviewed by Hawke and Garry, 2001). By contrast, the significance of direct contacts between MPs and mpcs has not been previously explored in the setting of muscle regeneration. In fact, relatively little is known regarding the relevance of apoptosis to skeletal muscle homeostasis and repair, although evidence exists indicating that enhanced apoptosis plays a role during muscle aging, muscular dystrophy, muscle denervation and unloading (reviewed by Jejurikar and Kuzon, 2003). Normal adult myofibres are somewhat resistant to apoptosis. Their sarcoplasm is refractory to mitochondrial cytochrome c-dependent activation of type II caspases (Burgess et al., 1999). Caspase-3 protein, which acts on the execution of cell death, is absent in normal myofibres (Ruest et al., 2002). Upstream protective mechanisms against apoptosis include blockage of the two caspase-3 activation pathways, as the caspase-8 inhibitor ARC is expressed (Koseki et al., 1998), and caspase-9 activator Apaf-1 is absent (Burgess et al., 1999), from skeletal muscle. The only physiological circumstance in which caspase-3 protein appears in adults is in regenerating muscle (Ruest et al., 2002). Such an expression of caspase-3 protein in regenerating muscle is transient and might allow muscle to get rid of excess replicating satellite cells or to delete improperly innervated, newly formed myofibres (Ruest et al., 2002). In addition to its role in apoptosis, caspase-3 also participates to myofibrillar proteolysis (Du et al., 2004). Once regeneration is complete, caspase-3 mRNA remains detectable in the repaired muscle whereas caspase-3 protein becomes undetectable (Ruest et al., 2002). Finally, from an evolutionary perspective, it seems important for skeletal muscle tissue to be protected from pro-apoptotic signals linked to exercise-associated mitochondrial stress (Burgess et al., 1999) and, consequently, mechanisms promoting restoration of the protected status of myogenic cells after muscle damage must exist and could implicate stromal cells.

We examined if and how MPs could play a significant role in regulation of myogenic cell death during regeneration. We first extended our previous observations by analysing MP protective effects against spontaneous and staurosporine (STS)-induced apoptosis of human mononucleated myoblasts and multinucleated myotubes. Then, we selected candidate anti-apoptotic effector-counterligand molecular systems using DNA macroarray analysis, with confirmatory RT-PCR and immunodetection in human MPs and mpcs. Four systems previously implicated in cell-contact-mediated survival of other cell types were identified and shown to mediate in vitro MP anti-apoptotic effects on mpcs by functional studies: vascular cell adhesion molecule 1 (VCAM-1; CD106) binding to very late antigen 4 (VLA-4); intercellular cell adhesion molecule 1 (ICAM-1; CD54) binding to leukocyte function associated molecule 1 (LFA-1); chemokine CX3CL1 binding to CX3CR1; and platelet-endothelial cell adhesion molecule 1 (PECAM-1; CD31) homophilic binding to another PECAM-1. Finally, we used a mouse model of post-injury muscle regeneration to demonstrate spatiotemporal correlation between MP influx and fading of injury-induced mpc apoptosis.

As assessed by annexin V labelling, the addition of human monocyte-derived MPs to primary mpc cultures inhibited spontaneous apoptosis of mononucleated myoblasts (Fig. 1A). The protective effect of MPs was dose dependent (P<0.05), apoptosis being inhibited by 75% at the 1:5 (mpc:MP) ratio (P<0.001) (Fig. 1A). Because the rate of spontaneous mpc apoptosis was low (7.1±1.3% of the cells), further experiments were performed after induction of mpc apoptosis by STS (Dominov et al., 1998; Columbaro et al., 2001). The protective effect of MPs was strong enough to reduce STS-induced mpc apoptosis (Fig. 1B). At the 1:10 (mpc:MP) ratio, apoptosis was inhibited by 74% as assessed by annexin V labelling (P<0.001) (Fig. 1B) and by 60% as assessed by DIOC-6 staining (P<0.01). MPs inhibited STS-induced myoblast apoptosis in a dose-dependent and saturable way (P<0.001) (Fig. 1B).

In culture, mpcs proliferate and give rise to mononucleated myoblasts that subsequently fuse with each other to form multinucleated myotubes. It is well established that myoblast apoptosis occurs at times of serum deprivation used to boost myogenic differentiation. In addition, it has been shown that myotubes are at particular risk of undergoing apoptosis upon stimulation with extrinsic stimuli as a result of their poor Bcl-2 expression (Dominov et al., 1998; McArdle et al., 1999; Ruest et al., 2002). Consistently, in our experiments, myotubes were 1.6-fold more sensitive to STS than myoblasts (P<0.05) (Fig. 1C,D). Because the large size of myotubes precluded flow cytometry analysis, determination of caspase-3 activity was used to compare the anti-apoptotic effect of MPs on STS-treated myoblasts and myotubes (Dominov et al., 1998). MPs more efficiently rescued myotubes than myoblasts from STS-induced apoptosis, as they decreased caspase-3 activity by 21% in myoblasts and by 39% in myotubes [values at 5 hours, 1:2 (mpc:MP) ratio, P<0.005] (Fig. 1C,D).

Expression of the Bcl-2 anti-apoptotic protein is important for survival of expanding myogenic cells (Dominov et al., 1998). As compared with mpc and MP cultures, co-cultures of mpcs with MPs showed enhanced expression of Bcl-2 (Fig. 2A). Pro-survival signalling pathways in myogenic cells include the mitogen-activated protein kinase and extracellular signal-regulated kinase (MAPK-ERK1/2) cascade, and the phosphatidylinositol 3-kinase (PI 3-kinase) and serine/threonine protein kinase Akt/PKB pathway (Ostrovsky and Bengal, 2003; Reuveny et al., 2004). These pathways operate through sequential phosphorylation events. Both pathways were activated in co-cultures, as assessed by increased phosphorylation of both ERK1/2 and Akt (Fig. 2A).

To evaluate to what extent MP phagocytosis of damaged cells (Geske et al., 2002) could have participated in the decreased number of apoptotic cells in co-cultures, we used potent inhibitors of MP phagocytic activity, including H₂O₂ at low concentrations and cytochalasin D (Elliott and Winn, 1986; Rubartelli et al., 1997; Anderson et al., 2002). The addition of phagocytosis inhibitors to co-cultures did not significantly modify the decreased rate of apoptotic mpcs observed in the presence of MPs (Fig. 2B). Consistently, the total number of mpcs did not significantly vary during the 6 hour time of co-culture, as assessed both by cell count [35,400±600 cells/cm² in mpc culture versus 35,750±1300 cells/cm² in co-cultures of mpcs with MPs (1:2)] and creatine phosphokinase level determination [9.04±4.4 UI/ml in mpc culture versus 7.3±3.8 UI/ml in co-cultures of mpcs with MPs (1:2)]. Altogether, these results substantiate the view that MPs induce pro-survival signalling and decrease mpc apoptosis.

To select the cell-cell molecular systems at work in the transduction of anti-apoptotic signals in mpcs, we used an mRNA profiling technique that allows analysis of a huge number of genes at once. Among the 375 genes represented on the DNA macroarray membrane we used, 12 (Table 1) had products known to be involved in anti-apoptotic signals mediated by cell-cell contacts.

Four of these products were constitutively expressed by human mpcs and had their counterligands expressed by human MPs (Table 1), as follows: (1) VCAM-1 binding to VLA-4 (α4β1 integrin); this adhesion system mediates the protective effects of MPs to erythroblasts (Hanspal and Hanspal, 1994; Sadahira and Mori, 1999), of stromal cells to haematopoietic stem cells, B cells and plasma cells (Koopman et al., 1994; Oostendorp et al., 1995; Wang et al., 1998; Hayashida et al., 2000; Minges Wols et al., 2002; Hall et al., 2004) and of endothelial cells to mast cells (Mierke et al., 2000). It is also involved in protection of T cells and retinal ganglion cells (Rose et al., 2000; Leussink et al., 2002; Leu et al., 2004). (2) ICAM-1 binding to LFA-1 (αLβ2 integrin); this system mediates protective effects of endothelial cells to transmigrating lymphocytes (Borthwick et al., 2003). Its implication in support of bone marrow stromal cells to T cells (Winter et al., 2001), and of follicular dendritic cells to B cells (Koopman et al., 1994) has been also reported, but in vitro experiments have yielded somewhat discrepant results (Zen et al., 1996; Wang and Lenardo, 1997). (3) CX3CL1 (fractalkine) binding to CX3CR1; CX3CR1 uniquely binds membrane-anchored and shed soluble forms of CX3CL1 (Bazan et al., 1997), which are respectively involved in firm cell-to-cell adhesion (Fong et al., 1998; Umehara et al., 2001) and in chemotaxis (Chapman et al., 2000). MPs and neural cells reciprocally signal through this system to suppress apoptotic cell death (Harrison et al., 1998; Boehme et al., 2000; Meucci et al., 2000; Mizuno et al., 2003; Deiva et al., 2004). This system also prevents apoptosis in intestinal epithelium (Brand et al., 2002). (4) Homophilic PECAM-1 interactions; endothelial cell PECAM-1 prevents apoptosis of both neighbouring endothelial cells (Bird et al., 1999; Evans et al., 2001; Gao et al., 2003) and transmigrating leucocytes (Ferrero et al., 2003) through homophilic PECAM-1 interactions.

Exposure to MP-conditioned medium reinforced mRNA expression of all counterreceptors by mpcs (Table 1). Results of DNA macroarray were confirmed by RT-PCR. VCAM-1, ICAM-1, CX3CL1 and PECAM-1 mRNAs were detected in MPs (Fig. 3A). It was shown that mpcs, which are already known to express the counterreceptor β1 integrin (Vachon et al., 1997), expressed α4, αL and β2 integrins, and CX3CR1 and PECAM-1 mRNAs (Fig. 3A). The corresponding proteins were immunodetected at the cell surface of MPs and mpcs, except PECAM-1, which could not be visualised in mpcs despite positive detection by immunoblotting in differentiated mpcs (Fig. 3B). Expression of all four receptors was much stronger in myotubes than in myoblasts (Fig. 3B).

Adhesion assays were used to assess functional availability of the candidate molecular systems at the MP-mpc interface. It was found that mpcs adhered on a MP monolayer in a dose-dependent, time-dependent and saturable fashion (Fig. 4A,B), allowing the involvement of specific receptor-ligand interactions to be assessed. Blocking antibodies against both CX3CL1-CX3CR1, VCAM-1-VLA-4, ICAM-1-LFA-1 or PECAM-1 failed to inhibit mpc adhesion on MPs significantly (data not shown), suggesting robust redundancy of cell adhesion systems involved in mpc-MP contacts. Therefore, adhesion assays using mpcs deposited on coats of immobilised ligands were used to test the functionality of each receptor. It was shown that mpcs adhered in a dose-dependent and saturable way to VCAM-1 (Fig. 4C), CX3CL1 (Fig. 4D), ICAM-1 (Fig. 4E) and PECAM-1 (Fig. 4F), as previously shown for other cell types (Meyer et al., 1995; Imai et al., 1997; Bird et al., 1999; Goda et al., 2000). These results strongly suggested that VLA-4, CX3CR1, LFA-1 and PECAM-1 expressed at the mpc surface were used to bind VCAM-1, membrane-bound CX3CL1, ICAM-1 and PECAM-1, respectively.

Co-cultures of mpcs with MPs were incubated with blocking antibodies to assess anti-apoptotic effects of the selected molecules in our model. Blockage of each molecular system inhibited the beneficial effects of MPs on mpc survival (P<0.05) (Fig. 5), as assessed by increased annexin V labelling and caspase-3 activity, and decreased DIOC-6 staining following exposure to blocking antibodies. Consistently, caspase-3 activity was increased in myotubes in the presence of the different blocking antibodies (Fig. 5C).

To examine the relevance of in vitro results, we injected snake venom notexin in tibialis anterior muscle of adult mice (Lefaucheur and Sebille, 1995). At 3 hours post-injury, TUNEL-positive nuclei were abundant in damaged areas (Fig. 6A) (15.6 apoptotic cells/field). It was found that 60% (9.5 cells/field) of TUNEL-positive cells were desmin negative and presumably corresponded to non-muscle cells; 40% (6.1 cells/field) of apoptotic cells were desmin positive and probably corresponded to activated satellite cells, interstitial myoblasts or myotubes (Fig. 6A). As MP influx proceeded in the regenerating muscle, the number of both the overall apoptotic cell population and of TUNEL-desmin double-positive cells dramatically decreased (Fig. 6A). At 48 hours post-injection, MP infiltration was massive and apoptotic mpcs were extremely rare, accounting for no more than 20% (0.12 cells/field) of the remaining apoptotic cells (0.62 cells/field) at this time (Fig. 6A). At this time point, both anti-apoptotic effectors VCAM-1, ICAM-1, CX3CL1 and PECAM-1, and their counterreceptors VLA-4, LFA-1, CX3CR1 and PECAM-1, were expressed by MPs and mpcs, respectively, in the regenerating areas (Fig. 6B).

In the present study, we demonstrate that: (1) MP influx is temporally correlated with fading of mpc apoptosis during in vivo post-injury muscle regeneration; (2) MPs rescue differentiating mpcs more than cycling mpcs from apoptotic cell death; (3) MPs deliver anti-apoptotic signals to mpcs through direct cell-cell contacts involving VCAM-1-VLA-4, ICAM-1-LFA-1, PECAM-1-PECAM-1 and CX3CL1-CX3CR1 interactions.

Muscle damage is known to induce massive MP infiltration of the injury site (McLennan, 1996; Pimorady-Esfahani et al., 1997). Initially, the role of these blood-borne cells was believed to be limited to clearance of necrotic fibres (McLennan, 1996; Pimorady-Esfahani et al., 1997). However, several in vivo and in vitro studies have shown that MPs are essential in orchestration of the muscle repair process (Grounds, 1987; Lescaudron et al., 1999).

The decision of a cell to proliferate, differentiate or undergo apoptosis is an integrated response to its growth factors and adhesive environment (Schwartz and Ingber, 1994). At the population level, mpc growth depends on both cell-cycling activity and cell survival. Previous studies on mpc-supporting cues have focused on MP-released soluble growth factors (Robertson et al., 1993; Cantini and Carraro, 1995; Merly et al., 1999). Some particular growth factors, such as insulin growth factor I (IGF-I), in addition to being a potent myogenic differentiation factor (Tureckova et al., 2001), can both stimulate mpc proliferation in the presence of other soluble factors (Napier et al., 1999) and promote mpc survival (Lawlor and Rotwein, 2000). However, our previous (Chazaud et al., 2003b) and present studies indicate that MP-derived soluble factors globally stimulate mpc proliferation, as assessed by thymidine incorporation, whereas direct MP cell contacts confer protection against apoptosis to mpcs.

Apoptotic cell death is a normal developmental event involving both proliferating myoblasts and postmitotic myofibres (Garcia-Martinez et al., 1993; Tidball et al., 1995; McClearn et al., 1995). As shown in our in vivo study, apoptosis of myogenic cells also occurs during regeneration of postnatal muscle. In this setting, TUNEL-positive myogenic cells disappear as MP infiltration proceeds. Obviously, this might reflect both MP phagocytosis of dead cells and the delivery of an MP pro-survival signal to living mpcs.

In vitro, the protective effects of MPs were twofold stronger towards post-mitotic differentiating mpcs than towards cycling myoblasts. The physiological significance of this finding remains elusive. Furthermore, mpcs are at risk of undergoing apoptosis for different reasons during the proliferation and differentiation process. Fast-cycling myoblasts must face difficulties in maintaining adequate DNA repair that might constitute an intrinsic signal for apoptosis (Wang and Walsh, 1996). Then, as they withdraw from the cell cycle and begin to differentiate, mpcs are at particular risk of myoblast-fusion-associated apoptosis, induced by endoplasmic reticulum stress (Nakanishi et al., 2005). Finally, myotubes elongate through additional myoblast fusion and must progressively stabilise their structure by establishing close association with the extracellular matrix (ECM) (Huppertz et al., 2001). Myogenic cell adhesion to the microenvironment seems to be crucial for their survival, as demonstrated by increased muscle cell apoptosis associated with deficiencies in ECM-binding proteins such as α5 and α7β1 integrins, and ECM proteins such as laminins (Vachon et al., 1996; Vachon et al., 1997; Miyagoe et al., 1997; Taverna et al., 1998; Montanaro et al., 1999). It is possible that MP-supportive cues help myotubes to achieve their adhesion-induced stabilisation safely. In line with this view, myotubes, which poorly express Bcl-2 and are therefore more sensitive than myoblasts to STS-induced apoptosis (Dominov et al., 1998), are endowed with stronger expression of the four receptors involved in adhesion-induced pro-survival signalling.

VLA-4, but not LFA-1, PECAM-1 and CX3CR1, was previously reported to be expressed by mpcs and to increase with myogenic differentiation (Rosen et al., 1992). The authors evaluated VLA-4 as an ECM receptor (Rosen et al., 1992), although this integrin, like LFA-1, is also involved in cell-cell adhesion and signalling. All four receptors expressed by mpcs, upon binding of their respective ligands VCAM-1, ICAM-1, PECAM-1 and CX3CL1, were previously shown to mediate anti-apoptotic signalling in a variety of non-muscle cell types (see Results section). VLA-4, PECAM-1 and CX3CR1 mediate activation of the PI 3-kinase/Akt survival pathway (Meucci et al., 2000; Gao et al., 2003; Ferrero et al., 2003; Deiva et al., 2004) and, in addition, CX3CR1 activates ERK1/2 (Brand et al., 2002; Deiva et al., 2004). These signalling pathways are both involved in mpc survival (Lawlor and Rotwein, 2000). Among many pro-survival effects, Akt phosphorylates BAD, causing its release from the complex it forms with Bcl-2, allowing Bcl-2 to exert its anti-apoptotic activity freely (Song et al., 2005). Both Akt and ERK1/2 pathways lead to inhibition of caspase-3 (Allan et al., 2003; Song et al., 2005), the major effector of the last step of muscle cell apoptosis (Tews, 2002). Consistently, co-cultures of MPs with mpcs showed increased Akt and ERK1/2 phosphorylation, increased Bcl-2 expression and decreased caspase-3 activity.

We previously showed that, early after activation, mpcs secrete a set of chemoattractants to initiate recruitment of circulating monocytes into damaged muscle (Chazaud et al., 2003b). Once recruited, monocytes differentiate into MPs, which are monocyte-derived MPs expressing VCAM-1, ICAM-1, PECAM-1 and CX3CL1, as shown herein. The newly recruited MPs release soluble factors that both amplify recruitment of MPs and stimulate mpc proliferation (Chazaud et al., 2003b). In addition, according to our DNA array, soluble factors produced by MPs reinforce mpc expression of VLA-4, LFA-1, PECAM-1 and CX3CR1 by 20-80%. Thus, when MPs enter into contact with mpcs, both cell types appropriately express anti-apoptotic ligands and counterreceptors.

In conclusion, the present study highlights the complex network of intercellular signalling and communication involved in the organisation of the stromal support of myogenesis. Our data indicating that inflammatory cells, i.e. macrophages, are beneficial for muscle regeneration are in accordance with in vivo studies showing that blocking inflammation with anti-inflammatory drugs might be deleterious for muscle regeneration and repair (Mishra et al., 1995; Shen et al., 2005). Moreover, evidence that a set of adhesion molecules rescue mpcs from apoptosis might open the possibility of improving myoblast transfer therapy. A strong limitation of this therapeutic approach consists of early massive cell death of non-mechanical origin (Chazaud et al., 2003a), affecting >95% of transplanted mpcs (Skuk and Tremblay, 2000). Moreover, mpcs induced to proliferate actively ex vivo to obtain a huge number of cells for transplantation was shown to increase their susceptibility to undergo apoptosis upon deprivation of extrinsic supportive cues (Rehfeldt et al., 2004). It seems likely that the use of anti-apoptotic cells or molecules could limit massive transplanted cell death, thus allowing appropriated mpc proliferation, differentiation and striated muscle repair.

Unless indicated, culture media components were from Invitrogen (Gibco) and culture plastics from TPP AG (Trasadingen). Human mpcs were cultured from muscle samples as previously described (Chazaud et al., 2003b). Only cultures presenting over 95% CD56⁺ cells (immunolabelling using anti-CD56 antibodies, diluted 1/20; Sanbio/Monosan) were used. Growing medium [HAM-F12 medium containing 15% fetal calf serum (FCS)] was used for culturing mpcs. To obtain myotubes, medium was replaced by HAM-F12 medium containing 5% FCS (differentiating medium) at time of subconfluence and cells were further cultured during 10 days (Lafuste et al., 2005).

MPs were obtained from monocytes isolated from human blood as previously described (Chazaud et al., 2003b). Briefly, monocytes were seeded at 0.5×10⁶ cell/ml in Teflon bags (AFC) in RPMI medium containing 15% human AB serum for 8 days.

In each series of experiments, the number of mpcs remained constant whereas the number of MPs varied. Undifferentiated mpcs were seeded at 10,000 cells/cm². Differentiated myotubes were counted in order to seed the appropriate number of MPs, from 1:10 ratio. Co-cultures were incubated in growing or differentiating medium for 6 or 24 hours at 37°C. In some experiments, mpc apoptosis was first induced by staurosporine (STS) treatment (1 μM for 6 hours). In some experiments, blocking antibodies were added in co-cultures of mpcs with MPs at saturating concentrations (calculated from IC50 or from previous studies): anti-CX3CL1 (3 μg/ml, 51637.1 clone; R&D Systems) (Chazaud et al., 2003b), anti-CX3CR1 (15 μg/ml, TP502; Torrey Pines Biolabs) (Chapman et al., 2000), anti-VCAM-1 (5 μg/ml, 1G11 clone; Immunotech) (Minges Wols et al., 2002), anti-VLA-4 (5 μg/ml, HP2/1 clone; Immunotech) (Hayashida et al., 2000), anti-LFA-1 (5 μg/ml, TS1/22 clone; Endogen) (Hayashida et al., 2000), anti-ICAM-1 (5 μg/ml, 84H10 clone; Immunotech) (Winter et al., 2001), anti-PECAM-1 (5 μg/ml, VM64 clone, Biodesign International). In other experiments, co-cultures were performed in the presence of hydrogen peroxide (0.2 mM; Sigma) (Anderson et al., 2002) or cytochalasin D (1 μg/ml, Sigma) (Elliott and Winn, 1986; Rubartelli et al., 1997). Controls included addition of whole IgGs from mouse and rabbit (3 μg/ml; Vector Laboratories).

Trypsin was used to detach mpcs and detection of apoptotic cells was performed using annexin V plus CD14 labelling and DIOC-6 plus CD14 staining. CD14 labelling was used to exclude MPs detached by the trypsinisation procedure (Fig. 7A). Cells were resuspended in 100 μl buffer (140 mM NaCl, 2.5 mM CaCl₂, 10 mM HEPES pH 7.4) containing either 2 μl of annexin V (Roche Diagnostics) or 70 nM of DIOC-6 (Molecular Probes) and 10 μl of TRITC-conjugated anti-CD14 antibodies (RMO52; Immunotech) for 30 minutes. Cells were washed before analysis by flow cytometry on a FACSCalibur (BD Biosciences). Apoptosis of mpcs was significantly increased by STS treatment, reaching 30±13% of the cells (annexin V detection) and 44±13% of the cells (DIOC-6 detection) (P<0.01) (Fig. 7B). As the range of apoptotic mpcs was 19-60% (annexin V detection) and 30-70% (DIOC-6 detection) of the cells, mpc apoptosis was expressed in percentage of apoptosis evaluated in STS-treated mpc cultures (without MPs). In co-cultures of MPs with untreated mpcs, CD14 expression was not affected (Fig. 7C); in co-cultures of MPs with STS-treated mpcs, we observed no more than 4-5% of CD14^- cells among CD45⁺ cells (Fig. 7C), indicating that gating allowed exclusion of >95% of MPs.

Proteins from mpcs, MP cultures and mpc-MP co-cultures were extracted in lysis buffer (50 mM Hepes pH 7.4, 100 mM NaCl, 1% Nonidet P-40, 40 mM EGTA pH 8.0, 100 mM DTT, 2 μg/ml leupeptin, 2 μg/ml aprotinin, 1 μg/ml pepstatin) and recovered after centrifugation at 4000 g for 10 minutes at 4°C. Protein concentration was determined using the BCA protein assay kit from Pierce. Aliquots corresponding to 30 μg of proteins were diluted in caspase-3 reaction buffer (1 M Hepes pH 7.4, 40 mM EDTA pH 8.0, 100 mM DTT, 25% sucrose) and incubated during 8 hours at 37°C in a microplate with caspase-3 substrate (Ac-DEVD-AFC fluorogenic substrate, Biomol Research Laboratories). Enzymatic activity was measured every 30 minutes with a fluorescence plate reader FL600 (Bio-Tek) and was expressed in arbitrary units.

Before being allowed to adhere on a confluent monolayer of MPs at various densities (5000 to 50,000 mpcs per well) and for various times (30 to 120 minutes), mpcs were labelled with 5-bromo-2-deoxyuridine (BrdU) for 72 hours. BrdU was then quantified using a colorimetric assay (Cell proliferation ELISA BrdU kit; R&D Systems).

Flat-bottomed 96-well sterile plates were coated with human recombinant VCAM-1, CX3CL1, ICAM-1 or PECAM-1 (0.001 to 100 nM) (R&D Systems) in phosphate-buffered saline (PBS). Non-specific binding sites were blocked with 1% bovine serum albumin for 30 minutes at room temperature. 30,000 mpcs per well were allowed to adhere for 2 hours at 37°C. Non-adherent cells were removed by gentle PBS washes. Cells were fixed with acetone and methanol for 15 minutes and stained with 0.5% Violet Crystal for 15 minutes. The number of adherent cells was evaluated by reading the OD at 540 nm.

Total RNA was prepared from human mpcs and MPs using the RNeasy mini kit (Qiagen). All further steps were performed according to the manufacturer's instructions in the human cytokine array GA001 kit (R&D Systems). For mpc and MP samples, 5 and 7 μg of total RNA gave labelled cDNA of 600,000 and 800,000 cpm, respectively, which was deposited on membranes. Results were read using a Phosphorimager (Amersham) after 72 hours exposure time. Analysis was performed using Image Quant software (Amersham), which allows background noise subtraction, correction for the variation of density for housekeeping genes and, finally, for comparison of densitometric signals. Results were expressed in arbitrary units.

Total mpc or MP RNA (1.5 μg) was reverse transcribed and amplified using OneStep RTPCR (Qiagen) and specific primers. For CX3CL1 [primers described in Lucas et al. (Lucas et al., 2001)], amplification was performed at 94, 64 and 72°C for 30 seconds, 30 seconds and 1 minute, respectively, for 38 cycles. For CX3CR1 [primers described in Muehlhoefer et al. (Muehlhoefer et al., 2000)], amplification was performed at 94, 55 and 72°C for 30 seconds, 30 seconds and 45 seconds, respectively. For VCAM-1 [primers described in Serradell et al. (Serradell et al., 2002)], amplification was performed at 94, 53 and 72°C for 30 seconds, 30 seconds and 45 seconds, respectively, for 38 cycles. For α4 integrin (GenBank # NM_000885), the sense primer used was 5′-CGA ACC GAT GGC TCC TA-3′ and the antisense primer was 5′-AGT ATG CTG GCT CCG AAA AT-3′, amplification was performed at 94, 55 and 72°C for 30 seconds, 30 seconds and 45 seconds, respectively, for 40 cycles. For ICAM-1 [primers described in Besch et al. (Besch et al., 2002)], amplification was performed at 94, 65 and 72°C for 30 seconds, 30 seconds and 45 seconds, respectively, for 50 cycles. For αL integrin (GenBank # BC008777), the sense primer used was 5′-TTT GAG AAG AAC TGT GGG GAG GAC-3′ and the antisense primer was 5′-GGT GGG CGA GAT GGA AGG T-3′, both amplification was performed at 94, 60 and 72°C for 30 seconds, 45 seconds and 2 minutes, respectively, for 40 cycles. Amplification products (10 μl) were subjected to electrophoresis on 2% agarose gel containing ethidium bromide for visualisation.

Total proteins from mpc and MP cultures, and co-cultures of mpcs with MPs, were extracted as described in Davaille et al. (Davaille et al., 2002). Protein concentration was determined using the BCA protein assay kit. Aliquots corresponding to 15 μg of proteins were subjected to western blot. Anti-phosphorylated Akt (1/1000; Cell Signalling Technology), anti-phosphorylated ERK1/2 (1/1000; Promega), anti-Bcl-2 (1/500; Santa Cruz Biotechnology), anti-PECAM-1 (1/500, Dakocytomation) or anti-β-actin (1/1000; Santa Cruz Biotechnology) antibodies were added overnight and revealed using peroxidase-conjugated anti-mouse, anti-rabbit or anti-goat antibodies (1/4000; Santa Cruz), which was detected using a chemiluminescence kit (Amersham Biosciences).

Human cells cultured on coverslips were labelled with primary antibodies (same references as above) for 2 hours: anti-CX3CL1 (50 μg/ml), anti-CX3CR1 (15 μg/ml), anti-VCAM-1 (15 μg/ml), anti-VLA-4 (15 μg/ml), anti-ICAM-1 (15 μg/ml), anti-LFA-1 (15 μg/ml), anti-PECAM-1 (15 μg/ml), revealed using biotinylated antibody (1/200), HRP-streptavidine (1/200) and DAB substrate kit for peroxidase (Vector Laboratories). Controls included incubation with whole IgGs from the species of the primary antibody (50 μg/ml; Vector Laboratories).

Notexin (10 μl of 25 μg/ml in PBS; Sigma) was injected into the tibialis anterior of adult C57/B6 mice. At various times after injection, muscles were removed, snap frozen in nitrogen-chilled isopentane (-160°C) and kept at -80°C until use. 7 μm-thick cryosections were treated for immunolabelling.

Muscle cryosections were incubated with rabbit polyclonal desmin antibodies (60 μg/ml; Abcam) and further treated to detect apoptotic nuclei (Apoptag Red; Qbiogen). Slides were examined under an Axioplan 2 Zeiss microscope (Carl Zeiss) and images were captured with an Orca ER digital camera (Hamamatsu Photonics KK) using Simple PCI software (C-Imaging Compix). Apoptotic desmin^- and desmin⁺ cells were counted in at least 20 randomly chosen fields within the injured area (×20 objective).

Muscle cryosections were double labelled with either desmin antibodies (60 μg/ml; Abcam) and anti-VLA-4 (15 μg/ml, Chemicon International) or anti-LFA-1 (10 μg/ml; Abcam) or anti-PECAM-1 (10 μg/ml, Santa Cruz) or anti-CX3CR1 (10 μg/ml; R&D Systems, using the MOM kit from Vector Laboratories) antibodies to detect mpc expression. Slides were treated with anti-CD11b antibodies (10 μg/ml; BD Biosciences) and anti-CX3CL1 (15 μg/ml; Abcam) or anti-VCAM-1 (15 μg/ml; R&D Systems) or anti-ICAM-1 (50 μg/ml; Chemicon International) or anti-PECAM-1 (10 μg/ml; Santa Cruz) antibodies to detect MP expression. To evaluate MP infiltration after injury, slides were treated with anti-CD11b as above or anti-F4/80 antibodies (20 μg/ml; Abcam). Primary antibodies were detected with either cy3-labelled or FITC-labelled secondary antibodies (Jackson ImmunoResearch Laboratories). Controls included incubation with whole IgGs from species of the primary antibody (Vector Laboratories). Slides were examined as described above.

Except DNA array, all experiments were performed using at least three different cultures or animals. The Student's t-test and ANOVA analysis were used for statistical analyses. P<0.05 was considered significant.

This work was supported by the Association Française contre les Myopathies and Fondation de France. We thank S. Lotersztajn for helpful discussions.