Author contributions: M.S. and H.Y.-Y.: collection and/or assembly of data and data analysis and interpretation; A.L.M.: collection and/or assembly of data and manuscript writing; L.A., H.A., and M.M.: collection and/or assembly of data; F.S. and J.C.: provision of study material or patients; G.K.P.: conception and design and manuscript writing; R.M.: data analysis and interpretation and manuscript writing; M.K. and B.C.: conception and design, data analysis and interpretation, and manuscript writing. M.S. and H.Y.-Y. are first authors.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS November 21, 2012.
Macrophages (MPs) exert either beneficial or deleterious effects on tissue repair, depending on their activation/polarization state. They are crucial for adult skeletal muscle repair, notably by acting on myogenic precursor cells. However, these interactions have not been fully characterized. Here, we explored both in vitro and in vivo, in human, the interactions of differentially activated MPs with myogenic precursor cells (MPCs) during adult myogenesis and skeletal muscle regeneration. We showed in vitro that through the differential secretion of cytokines and growth factors, proinflammatory MPs inhibited MPC fusion while anti-inflammatory MPs strongly promoted MPC differentiation by increasing their commitment into differentiated myocytes and the formation of mature myotubes. Furthermore, the in vivo time course of expression of myogenic and MP markers was studied in regenerating human healthy muscle after damage. We observed that regenerating areas containing proliferating MPCs were preferentially associated with MPs expressing proinflammatory markers. In the same muscle, regenerating areas containing differentiating myogenin-positive MPCs were preferentially coupled to MPs harboring anti-inflammatory markers. These data demonstrate for the first time in human that MPs sequentially orchestrate adult myogenesis during regeneration of damaged skeletal muscle. These results support the emerging concept that inflammation, through MP activation, controls stem cell fate and coordinates tissue repair. STEM CELLS2013;31:384–396
Macrophages (MPs) are one of the main agents of innate immunity. They contribute to the host defense and phagocytosis of undesirable pathogens or tissue debris. As a consequence, MPs are mainly considered as “inflammatory” cells that are deleterious for the tissue and efforts are focused on decreasing or preventing their presence in the tissue. Indeed, several studies have shown that MPs enhance tissue damage. For example, MP depletion prevents or reduces the clinical signs of experimental autoimmune neuritis  or attenuates blood-nerve barrier disintegration during chronic nerve injury . MP depletion is also beneficial in the case of pulmonary and hepatic injury induced by toxicants . As the effects of MPs can be either beneficial or deleterious [4–6], depending on the concentration of MP supernatant used (for in vitro studies) or the time point when MP depletion is induced (in vivo models), it is clear that the actions of MPs are complex. In support of this, several studies have demonstrated beneficial roles for MPs in the context of tissue homeostasis and repair. For example, MP-secreted factors have been reported to promote the survival, regeneration and neurite extension of neurons [7–9], proliferation of hepatocytes , and skeletal muscle regeneration [11–13].
The multiplicity of MP effects is likely due to their inflammatory profile or state of activation. Indeed, MPs adopt various inflammatory profiles depending on their environment. To help in understanding these various phenotypes, MPs have been classified into four main subtypes [14, 15]. Proinflammatory or classically activated or M1 MPs are induced by TH1 cytokines. They secrete proinflammatory cytokines and reactive oxygen species and exhibit microbicidal and tumoricidal activities. Alternatively activated or M2a MPs are triggered by TH2 cytokines such a interleukin (IL)-4. These cells secrete a battery of specific markers (YM1, AMCase, and Arginase 1 [Arg1]), are associated with extracellular matrix synthesis, angiogenesis, and parasite containment, and are observed during chronic fibrosis. Regulatory or M2b MPs are induced by a combination of Toll-like receptor ligands and immune complexes and dampen inflammation. Anti-inflammatory or deactivated or M2c MPs are induced by anti-inflammatory compounds (e.g., IL-10 and glucocorticoids). They secrete anti-inflammatory cytokines such as IL-10 and transforming growth factor (TGF)β. Harboring different phenotypes, it is likely that the activated/polarized MPs exhibit various functions during innate immunity. Very few studies have explored their functions on progenitor cells. Interferon (IFN)γ- and tumor necrosis factor (TNF)α-stimulated microglia blocks differentiation of adult neural progenitor cells while microglia activated by IL-4 or low levels of IFNγ stimulates these neural progenitor cells toward oligodendrogenesis and neurogenesis, respectively . In skeletal muscle, we have shown that lipopolysaccharide (LPS)/IFNγ-activated MPs stimulate myogenic cell proliferation while IL-4 and IL-10-activated MPs support their differentiation . Although MP activation states have been clearly defined in vitro, the link between the activation states of MPs and their functions remains unknown for most in vivo situations.
Adult skeletal muscle regeneration is an excellent paradigm of sterile inflammation with complete resolution. While the exact time course of human skeletal muscle regeneration has not been determined due to the restrictions of being able to sample human muscle a limited number of times using the biopsy technique, the time it takes for regeneration to be completed is known to vary depending on the nature of the experimental model used. For example, it has been shown that models of acute muscle damage using neuromuscular electrical stimulation, in contrast to voluntary contractions, are required if the goal is to induce damage and regeneration, clearly observable by signs such as infiltration of MPs into the myofibers, activation and proliferation of satellite cells, and the presence of myofibers positive for embryonic myosin [17–20]. With regard to the time course of human muscle regeneration, studies aiming to examine the endpoint are sparse, with one recent study demonstrating ongoing regeneration of the myofibers and remodeling of the extracellular matrix as late as 30 days after a single bout of electrical stimulation . Several studies have demonstrated the presence of damaged fibers from a range of 2–7 days post-trauma [19–21], confirming the complexity of the sequence of events required for successful and complete regeneration of damaged skeletal muscle. From the few studies investigating muscle regeneration in humans, thus, it appears that biopsy sampling approximately 7 and 30 days post-trauma would be suitable time points for studying processes of inflammation and satellite cell activity during skeletal muscle tissue repair.
In vivo experiments in mouse have shown that MPs are very important for muscle repair . A sequential involvement of proinflammatory and alternatively activated/anti-inflammatory MPs has been delineated in this process. The first appearing proinflammatory MPs enter into the damaged muscle, and a few days later, they convert into anti-inflammatory MPs . Meanwhile, the main muscle stem cells, satellite cells, undergo adult myogenesis through several sequential steps, each of which must be correctly completed to accommodate successful achievement of the subsequent steps. This process begins with the activation of the quiescent satellite cells on the myofiber by a damage signal. Activated myogenic precursor cells (MPCs) proliferate and it is believed that most of them enter into terminal myogenic differentiation, while a subset returns to quiescence to self-renew the satellite cell pool . Adult myogenesis encompasses: (a) activation of quiescent satellite cells and expansion of MPCs; (b) the commitment of MPCs into terminal differentiation. This requires cell-cycle withdrawal and expression of the myogenic program including the expression of the myogenic regulatory factors myogenin and MRF4 and contractile proteins such as myosins; (c) the migration of the cells toward each other to establish cell/cell contacts; and (d) the fusion of the cells into plurinucleated and contractile structures that form myotubes in vitro and myofibers in vivo.
In this work, we studied for the first time in human the relationship between the inflammatory profile of MPs and their regulatory role on the homeostasis of MPCs. We explored using human coculture in vitro which steps of adult myogenesis are controlled by differentially activated MPs and analyzed the involvement of a series of MP-secreted effectors. In vivo, we investigated in parallel the kinetics of adult myogenesis and the presence of MP subtypes during human skeletal muscle regeneration.
MATERIALS AND METHODS
Human MPC Culture
Human MPCs were isolated from normal adult skeletal muscle sample according to the French legislation (protocol registered at the Agence de la Biomédecine in 2007 “Interrelations entre les cellules souches adultes du muscle strié squelettique et les macrophages” and Cochin Hospital Cell Bank, Paris, agreement n° DC-2009-944). Cells were obtained and cultured as previously described  in HAMF12 medium (Gibco, Life Technologies, Grand Island, NY http://www.lifetechnologies.com) containing 15% fetal bovine serum (FBS) and were sorted before the first passage with anti-CD56 magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany, http://www.miltenyibiotec.com) according to the manufacturer instructions. The purity of the cells was assessed for CD56 expression (555518 BD Pharmingen, Franklin Lakes, NJ http://www.bdbiosciences.com) by flow cytometry (FC500 Beckman Coulter cytometer, Villepinte, France, http://www.beckmancoulter.com).
Human MP Culture
MPs were obtained from human Buffy-coat (purchased from Etablissement Francais du Sang according to INSERM-EFS agreement number C CPSL UNT n° 09/EFS/024) as previously described [11, 25]. MPs seeded in 12-well plates (36,000 cells per square centimeter) were activated/polarized for 2–3 days into M1 MPs with 1 μg/ml LPS and 10 ng/ml IFNγ, into M2c MPs with 10 ng/ml IL-10 and 80 ng/ml dexamethasone, or into M2a MPs with 10 ng/ml IL-4. Cells were then extensively washed (four times) with serum-free medium to remove the effectors. Conditioned media were obtained by incubating MPs for 24 hours in advanced RPMI 1640 medium (Gibco) containing 0.5% FBS. Analysis of surface antigen expression was performed by flow cytometry. Activated MPs were detached by nonenzymatic dissociation solution (Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) and gently scraped before being incubated with anti-CD16, HLA-DR, CD80, CD206, and CD163 antibodies (BD Pharmingen).
In Vitro Myogenesis
MPCs were seeded on glass coverslips at 12,000 cells per square centimeter and incubated with or without MP-conditioned medium for 3 days. In some experiments (Fig. 1), MPCs were seeded on activated MPs. In other experiments (Fig. 5), blocking antibodies were added at the time of addition of MP-conditioned medium. Antibody concentrations were chosen based on the ID50 according to the manufacturer and the concentration of secreted effectors by MPs, if known, from the literature: anti-TNFα 1/100 and 1/1,000, anti-IL-1β 1/200, anti-TGFβ 1/200, anti-IL-13 1/100 (Abcam, Cambridge, UK, http://www.abcam.com/), anti-Vascular Endothelial Growth Factor (VEGF) 1/200, anti-IL-6 1/100, anti-TWEAK 1/100 (R&D, Minneapolis, MN, http://www.rndsystems.com/) [11, 26-28].
Commitment into Myocytes
MPCs were seeded on glass coverslips at low density (500 cells per square centimeter). In this condition, contact between MPCs was minimal, so they could not fuse. MP-conditioned medium was then added and the cells were cultured for 72 hours.
MPCs were differentiated into myocytes for 3 days. These differentiated cells were seeded at high density: 50,000 cells per square centimeter, in 96-well plates. In this condition, the differentiation process and migration events were prevented. MP-conditioned medium was added and cells were further cultured for 72 hours.
Cells were labeled with anti-desmin (Ab32362 Abcam) and anti-myogenin (556358 BD Pharmingen) primary antibodies, revealed by Cy3- and biotinylated-conjugated secondary antibodies and streptavidin-DTAF (Jackson Immunoresearch, West Grove, PA, http://www.jacksonimmuno.com/, Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com/). Each experiment was repeated at least three times with different primary cultures and performed in triplicate. Approximately five to seven pictures (×20 magnification) were taken from the triplicates, in order that approximately 200–400 (low cell density) or 1,000 (high cell density) cells were counted per condition per experiment. Differentiation was measured as the number of myogenin-positive nuclei divided by the total number of nuclei. The number of myotubes (cells containing two or more nuclei) was expressed as a percentage of the total cell number (Figs. 1, 5). Fusion index was the number of nuclei within myotubes divided by the total number of nuclei (Fig. 4). Myotubes were also classified according to the number of myonuclei they contained and their distribution was expressed as a percentage of the total number of myotubes (Fig. 4).
MPCs were seeded in 24-well plates at 1,500 cells per square centimeter, MP-conditioned medium was added and cells were imaged every 30 minutes for 16 hours using a Zeiss Axiovert 200 (Oberkochen, Germany, https://www.zeiss.com/). Cell velocity was calculated using Image J software with the Manual Tracking plug-in. Approximately 50 cells were analyzed per condition per experiment.
Migration was performed using two chambers Ibidi inserts (Ibidi GmbH, Martinsried, Germany, http://ibidi.com/). One chamber was filled with 4,500 MPCs in growth medium. The second chamber was seeded with 25,000 MPs. MPs were activated for 3 days. The silicone walls of the device were removed and cells were cultured in advanced RPMI medium containing 0.5% FBS. The gap between the two cell types was imaged at time 0 and 24 hours. The distance that each cell had moved into the gap was analyzed using Image J software. Approximately 100 cells were analyzed per condition per experiment.
Human Muscle Samples
The human muscle biopsy sampling study was approved by the Regional Scientific Ethical Committees of Copenhagen in Denmark (Ref: HD-2008-074) and conformed to the standards set by the Declaration of Helsinki. Volunteers gave written informed consent before inclusion. Three young healthy untrained men (mean ± SD, age 25 ± 2 years; height 182 ± 6 cm; weight 80 ± 4 kg) participated in the study. Briefly, the exercise protocol used 10 sets of 20 maximal voluntary lengthening contractions of the quadriceps muscles, performed on an isokinetic dynamometer (KinCom KC125AP, Chattanooga Group Inc., Chattanooga TN, http://www.chattmed.com). Maximal percutaneous electrical stimulation was delivered to the vastus lateralis muscle during the contraction by a handheld stimulator (Elpha II 3000, Danmeter, Odense, Denmark, http://www.danmeter.dk/). Muscle biopsies were obtained from the stimulated vastus lateralis on days 7 and 30 after the exercise protocol. A biopsy was also collected from the nonexercised leg on day 0 as a control. Muscle biopsies were collected under local anesthesia (1% lidocaine) with a Bergstrom biopsy needle. After collection of the specimen, the fibers were aligned, embedded in Tissue-Tek (Sakura Finetek Europe, AJ Alphen aan den Rijn The Netherlands, http://www.sakuraeu.com/), and frozen by immersion in isopentane, precooled by liquid nitrogen. Samples were stored at −80°C. For each biopsy, 10 serial transverse sections (10 μm) were cut at −24°C using a cryostat and transferred onto SuperFrost Plus glass slides (Menzel-Gläser, Braunschweig, Germany, http://www.menzel.de/).
Human Muscle Immunolabelings
Serial sections of human muscle were treated for hematoxylin and eosin (H&E) staining or stained with antibodies raised against the following molecules: CD56 1/100 (347740; Becton Dickinson), ki67 1/200 (CP249; Biocare Medical, Pike Lane Concord, CA, http://www.biocare.net/), myogenin 1/50 (F5D; Developmental Studies Hybridoma Bank, Iowa City, Iowa, http://dshb.biology.uiowa.edu/), Pax7 1/100 (MO15020; Neuromics, Edina, MN, http://www.neuromics.com/), CD68 1/100 (ab955 Abcam), iNOS 1/25 (ab15323 Abcam), CD206 1/50 (ab8918 Abcam), CD163 1/500 (sc-20066 Santa Cruz, CA, http://www.scbt.com/), Arg 1 1/25 (sc-18355 Santa Cruz), and Cyclooxygenase (Cox2) 1/25 (ab2372 Abcam), revealed by FITC- or Cy3-conjugated secondary antibodies (Jackson Immunoresearch). Pictures were taken with a Zeiss Axio observer.Z1. The slide was scanned with a tile scan camera that automatically collects overlapping focal stacks and processes them to reconstitute the entire muscle section. To confirm coexpression of CD68 labeling with proinflammatory or anti-inflammatory markers, pictures were taken with an sp2 confocal microscope Leica TCS SP2 AOBS with a 1-μm depth.
Scoring Inflammation and Myogenesis in Human Regenerating Muscle
Calculations were made at day 7 postelectrostimulation. Postinjury regeneration was visible at the myofiber level. Discrete “regenerating areas”—corresponding to necrotizing/regenerating myofibers—which contained either CD56poski67pos or myogeninpos cells, were quantified and evaluated for all the stainings. Zones containing numerous neighboring regenerating myofibers were excluded to avoid confusion along the 10 serial sections. A total of 250 of regenerating areas was counted from three different biopsies. For each regenerating area, the number of cells positive for CD56, myogenin, ki67, CD68/iNOS, CD68/Cox2 (COX2), CD68/Arg1, CD206, and CD163 was evaluated.
All in vitro experiments were performed at least three times using independent MP and MPC primary cultures. In vivo analyses have been performed on three different subjects. Results are expressed as mean ± SEM. Means were compared using t test or ANOVA for statistical analyses. p < .05 was considered significant.
Human MPC cultures were used in the in vitro study. More than 97% of these cells expressed CD56, a canonical marker for myogenic cells in human (supporting information Fig. S1). After an expansion phase during which they proliferate, the majority of human MPCs enter into terminal differentiation and express myogenin . A subset of 15%–20% of MPCs returns to quiescence to constitute a reservoir [29, 30]. Differentiation is associated with migration of the cells into “areas” where fusion takes place . As previously shown, human MP activation triggers specific inflammatory profiles: M1 MPs secrete high levels of TNFα and IL-1β . Flow cytometry analysis revealed that, as expected, M1 MPs expressed high levels of the costimulatory antigen CD80, M2a MPs expressed CD206, and M2c MPs expressed CD163. However, M2c MPs also expressed high levels of CD206 (supporting information Fig. S1).
Polarized MPs Exert Their Effect on MPCs Through Soluble Factors
We checked whether the effects of activated/polarized MPs were triggered by soluble factors or whether contact between the two cell types was essential for the regulation of myogenesis. MPCs were either cocultured with MPs or treated with MP-conditioned medium for 3 days, and expression of myogenin and fusion of the cells were measured as final outcomes of in vitro myogenesis. Figure 1 shows that M1 MPs slightly decreased cell fusion (p < .05) while M2a MPs and M2c MPs stimulated both the expression of myogenin (Fig. 1A, 1C) and myotube formation (both p < .05) (Fig. 1B, 1C). Myogenin expression (Fig. 1A, 1C) and myotube formation (Fig. 1B, 1C) by MPCs were similar in coculture and with MP-conditioned medium, indicating that MPs mainly exerted their effects on myogenesis through soluble factors. The following experiments were then performed with MP-conditioned media.
M2a MPs and M2c MPs Stimulate MPC Commitment into Myocytes
Commitment into myocytes, which is the entry into terminal differentiation, was analyzed through the expression of myogenin in low-density cell cultures, to prevent migration/fusion of the cells. M1 MP-conditioned medium did not alter the expression of myogenin by MPCs while M2a MP- and M2c MP-conditioned media strongly increased myogenin expression, by 2.7- and 4.2-fold, respectively (p < .05) (Fig. 2). These data show that M2a MPs and M2c MPs stimulate the differentiation of MPCs into myocytes.
M1 MPs Stimulate MPC Motility
Figure 3A shows that a M1 MP-conditioned medium did not alter MPC motility compared to untreated cells. However, the speed of the cells was lower under M2a MP- and M2c MP-conditioned media (22%–24% less, p < .05). The median velocity was lower in cells treated with M2a MP- and M2c MP-conditioned medium (Fig. 3B). Mean velocities of the cell populations above and below the median were also decreased (data not shown), indicating a decreased motility of the entire MPC population and not of a subset of MPCs.
M1 MPs Do Not Attract MPCs While They Migrate Toward MPCs
Cell migration toward a gradient/other cell type was analyzed in a double chamber silicone device (Fig. 3C). Under M2a MP- and M2c MP-conditioned media, MPCs displayed the same degree of migration as toward other MPCs, while their migration toward M1 MPs was much lower (−45%, p < .05) (Fig. 3D). Conversely, when examining MP migration, M1 MPs were strongly attracted by MPCs, compared to M2a MPs and M2c MPs (Fig. 3E) (2.5 and twofold, respectively, p < .05).
M1 MPs Inhibit MPC Fusion
Already differentiated myocytes (preventing the differentiation parameter) were plated at high density (preventing the migration parameter) in MP-conditioned media to evaluate fusion after desmin immunolabeling (Fig. 4A). Fusion index was strongly decreased in MPCs cultured with M1 MP-conditioned medium (63%, p < .05) (Fig. 4B). The myotube distribution was significantly different under all MP conditions tested compared to MPC alone (p < .05). In particular, M1 MP-conditioned medium inhibited the formation of large myotubes (75% inhibition for myotubes containing more than 10 nuclei, p < .05) while the formation of small myotubes (three to five nuclei) was unaffected (Fig. 4C). As a whole, the fusion index of MPCs was not affected by M2a MP- and M2c MP-conditioned media (Fig. 4B). However, the distribution of myotubes was significantly different from that of MPCs cultured alone (both p < .05). M2c MP-conditioned medium stimulated the formation of large myotubes (68% of increase for myotubes more than 10 nuclei, p < .05) (Fig. 4C). Similarly, M2a MP-conditioned medium stimulated the formation of large myotubes (52%, p < .05) (Fig. 4C). In both cases, the formation of small myotubes was not altered. These results suggest that activated MPs differentially control the formation of mature myotubes.
Effectors Secreted by Activated MPs Regulate In Vitro Myogenesis
Some MP-released effectors are known or believed to have an impact on MPCs. However, the role of these effectors in the regulation of myogenesis through secretion by MPs has not been established. We tested the involvement of a series of seven molecules in the regulation of myogenesis by MPs through the use of blocking antibodies. Blocking IL-6 (Fig. 5A) increased myotube formation (by 38%, p < .05) only in MPCs cultured with M1 MP-conditioned medium. Blocking IL-1β similarly increased myotube formation only in MPCs cultured with M1 MP-conditioned medium (+49%, p < .05) (Fig. 5B). The blockade of VEGF (Fig. 5C) increased myotube formation in MPCs treated with all MP-conditioned media, although the effect was more pronounced for M1 MPs (+58%, +20%, and +26% for M1 MPs, M2a MPs, and M2c MPs, respectively, all p < .05). As TNFα is known to have differential effects on myogenesis depending on its concentration , we used two dilutions of blocking antibodies (Fig. 5D). At a low dilution, anti-TNFα decreased myotube formation in M2a MP- and M2c MP-conditioned medium-treated MPCs (30 and 31%, p < .05) while it had no effect on M1 MP-conditioned medium-treated cells (Fig. 5D, white bars). At a higher concentration, anti-TNFα still decreased myotube formation in M2a MP- and M2c MP-conditioned medium-treated cells—to the same extent (p < .05)—and increased myotube formation of MPCs incubated with M1 MP-conditioned medium (+27% of fusion, p < .05) (Fig. 5D, gray bars). Blockade of the TNFα-family member TWEAK did not affect the formation of myotubes by MPCs in all conditions (Fig. 5E). Anti-IL-13 antibodies induced an increase of myotube formation in MPCs cultured only in the presence of M1 MP-conditioned medium (+50%, p < .05) (Fig. 5F). Adding blocking anti-TGFβ antibodies significantly reduced the formation of myotubes in MPC cultures treated with M2a MP-and M2c MP-conditioned media (−26% and −27%, respectively, p < .05) (Fig. 5G). Except for the TWEAK antibody, none of these blocking antibodies affected the formation of myotubes by MPCs cultured alone (Fig. 5H). Although it is known that MPCs synthesize some of these cytokines and growth factors [33, 34], MPs secrete much higher levels of these compounds [11, 26]. These results show that effectors differentially secreted by activated MPs regulate myotube formation according to the state of activation of MPs.
During In Vivo Human Muscle Regeneration, MPs Sequentially Express Proinflammatory and Anti-Inflammatory Markers and Are Associated with Proliferating and Differentiating MPCs, Respectively
To explore inflammation during in vivo human muscle regeneration, we used a protocol combining lengthening contractions and electrostimulation . This resulted in myofiber damage, and the kinetics of tissue repair can be followed during the ensuing muscle regeneration. Supporting information Figure S2 shows that in normal untreated human muscle (day 0), CD56-positive cells (CD56pos) were observed throughout the muscle, at the normal place of satellite cells, while no staining was observed for the embryonic form of myosin heavy chain, which is a characteristic of regenerated/regenerating myofibers . On day 7 after the damage induction, the myogenic marker CD56 stained numerous areas while embryonic myosin heavy chain stained several regenerating myofibers (supporting information Fig. S2, day 7). These areas associated with CD56pos cells and young new myofibers were observed in regions of damaged myofibers and where regeneration took place. These regions, which we called regenerating areas, were distributed throughout the muscle biopsy cross-section (supporting information Fig. S2, day 7). On day 30 after damage, the muscle had recovered its normal histology, except for very rare myofibers with a centrally located myonucleus (supporting information Fig. S2, day 30). That the dynamic processes of damage and regeneration can be followed so clearly in this model makes it suitable for studying the relationship between MP activation and myogenesis in vivo.
To assess MP activation in vivo, we analyzed on serial sections the expression of various proinflammatory and anti-inflammatory markers, together with myogenic markers, during normal human skeletal muscle regeneration. Pax7 is a transcription factor expressed by quiescent and cycling, but not differentiating, MPCs. CD56 is expressed by quiescent, cycling, differentiating MPCs, and young regenerated myofibers. Myogenin is a transcription factor only expressed by differentiating myogenic cells. In human, CD68 is believed to widely label MPs while expression of CD206 and CD163 has been associated with anti-inflammatory phenotypes. When expressed by MPs, iNOS and COX2 are proinflammatory markers while Arg1 is expressed by M2a MPs [14, 15].
In normal muscle at steady state, almost no CD56pos cells were ki67pos indicating that they were quiescent (Fig. 6, day 0), as previously reported . As expected, no labeling for myogenin was observed (data not shown). Few cells were observed in the interstitial space that were positive for CD68, CD206, or CD163; no positivity was observed for iNOS, COX2, or Arg1 (Fig. 6, day 0). At day 7 after damage, numerous regenerating areas were present, containing proliferating cells (ki67posCD56pos) and/or differentiating cells (myogeninpos) (Fig. 6, day 7, Fig. 7). The vast majority (more than 80%) of the regenerating areas was associated with MPs that were positive for proinflammatory and for anti-inflammatory markers. Less than 20% of the regenerating areas were associated with only one type of MPs: 17% contained MPs expressing only anti-inflammatory markers while 0.4%–2% was associated with only proinflammatory MPs (when considering three and two markers, respectively). This suggests that MPs expressing proinflammatory and anti-inflammatory markers coexisted in the same regenerating areas. To analyze whether early and later regenerating areas were preferentially associated with different types of MPs, they were divided according to the absence (approximately 28% of the total regenerating areas) or the presence (approximately 72%) of myogeninpos cells, independently of the number of myogeninpos cells. The number of MPs per myofiber did not differ between myogeninpos- and myogeninneg-containing regenerating areas. An MP/MPC ratio of 2.5 ± 0.7 was found in regenerating areas, the median being 1.8 (the absolute range was from 0 to 12 MPs/area). When the regenerating areas were divided according to the presence or absence of myogeninpos cells, MP inflammatory markers were segregated (Fig. 6, day 7, Fig. 7). Proinflammatory markers such as iNOS and COX2 were expressed by more MPs in regenerating areas containing only myogeninneg cells compared to those containing at least one myogeninpos MPC: 51.3% versus 22.2% and 44.6% versus 18.6%, respectively (p < .05) (Fig. 6, day 7, Fig. 7). Conversely, Arg1-expressing MPs were more numerous in regenerating areas containing myogeninpos MPCs, compared to those without differentiating MPCs: 52.8% versus 30.9% (p < .05) (Fig. 6, day 7, Fig. 7). The actual expression of iNOS, COX2, and Arg1 by CD68 MPs was assessed by confocal microscopy (Fig. 7). Immunostainings showed that most of the regenerating areas contained MPs expressing CD206 and CD163. However, regenerating areas containing myogeninpos MPCs were associated with 1.8-fold more CD206pos MPs and CD163pos MPs (p < .05) (Fig. 6, day 7, Fig. 7). It has been shown that murine myogenic cells express CD206 . Colabeling of regenerating human muscle with CD68 and CD206 showed that almost all CD206pos cells observed in the regenerating areas were costained with CD68 indicating their macrophagic origin (supporting information Fig. S3). Thirty days after damage, no cycling or differentiating cells were observed (Fig. 6, day 30). With regard to the untreated muscle at day 0, few MPs expressing CD68, CD206, or CD163 were observed between the myofibers, and they did not express iNOS or COX2. Some MPs expressed Arg1 (Fig. 6, day 30). Taken together, these data reveal that different MP subsets are preferentially associated with areas containing proliferating or differentiating MPCs.
This study shows for the first time in human that MPs differentially act on MPCs depending on their activation/polarization state. They exert these trophic roles through the secretion of a series of cytokines and growth factors, thereby controlling the fate of the muscle progenitors at each step of in vitro myogenesis. We also show for the first time in human that these in vitro-activated MPs correspond to MP subsets in vivo. Indeed, during human muscle regeneration, regenerating areas are first preferentially associated with MPs bearing proinflammatory markers. Then, at the time of myogenic differentiation, MPs expressing anti-inflammatory markers are preferentially found in the regenerating areas.
Trophic activities for MPs have recently emerged in various contexts of tissue development, morphogenesis, and angiogenesis [37, 38]. In normal adult tissues, recent advances have been made in identifying supportive roles for MPs in stem cell homeostasis. MPs promote erythroblast survival and differentiation through very late activation antigen 4 and CD163 [39–41]. MPs support stem cell homeostasis in the bone marrow, and in intestinal and mammary epithelia [42–45]. In the nervous system, microglia rescues neurogenesis in neural stem cells by the secretion of neurogenic factors . MPs deliver anti-inflammatory cues beneficial to spinal cord recovery . They secrete TGFβ which promotes oligodendrocyte differentiation and myelinogenesis . Microglia activated by proinflammatory cues block neural progenitor differentiation while IL-4-activated microglia stimulates the formation of oligodendrogenesis, possibly through the production of IGF-1 [16, 49]. In skeletal muscle, we have shown in vitro that MPs rescue MPCs from apoptosis and stimulate their growth and proliferation through a panel of secreted and membrane-bound effectors [25, 33].
In our in vitro coculture system, based on a 1:3 MP/MPC ratio, we here showed that MPs mainly exerted their effects on myogenesis through the secretion of soluble factors. We have previously shown in vitro that M1 MPs stimulate MPC proliferation while M2a MPs and M2c MPs do not . Here, we explored each step of in vitro myogenesis using an experimental setup that allows for analysis of each step independently of the others: commitment into myocyte, migration and fusion . During in vitro myogenesis, MPCs migrate to the injured area and then toward each other in order to form myotubes. Differentiating and differentiated MPCs express a wide range of chemokine receptors that bind to chemokines in a redundant way . Activated MPs have been shown to secrete specific panels of chemokines [14, 15] and likely other attracting molecules. Chemotaxis is a global integrated cell response to signaling by several individual receptors. Therefore, it is difficult to decipher which signals secreted by MPs govern MPC motility and chemotaxis. However, our results present a new property of human MPs in the regulation of progenitor motility. M2c MPs and M2a MPs were observed to decrease the motile properties of MPCs and were better attractants for MPC directional migration. We have previously shown in vitro that MPCs attract circulating monocytes . It is therefore tempting to propose, knowing that M1 MPs are the first to enter the damaged muscle in mouse , that monocyte-derived M1 MPs have a great migratory capacity toward MPCs at the beginning of muscle regeneration. It can be hypothesized that these M1 MPs would maintain MPC migratory capacities thus preventing fusion from occurring too early, a high migratory speed being detrimental for good fusion . M2a MPs and M2c MPs would further attract MPCs, decrease their motility, thus favoring contact between cells, and deliver prodifferentiating cues. Further analyses are required to confirm these hypotheses.
Myogenic commitment into myocytes is the expression of the myogenic program after cell-cycle withdrawal and precedes fusion, which is the last step of in vitro myogenesis. In all the functional assays we tested in vitro, M2a MPs and M2c MPs displayed opposing effects to those of M1 MPs. M1 MPs inhibited MPC fusion, particularly the formation of large myotubes while M2a MPs and M2c MPs strongly stimulated the commitment of MPCs into myocytes and promoted the formation of large myotubes. Our results suggest that the formation of nascent myotubes is not controlled by MP-conditioned media. Most of the molecular systems involved in prefusion migration and myocyte fusion are transmembrane proteins and a few soluble effectors have been identified . Among them IL-4 is involved in the formation of mature myotubes in mouse . IL-4 production by lung MPs has been associated with the presence of M2a MPs [54, 55]. IL-4 may, therefore, be one of the effectors of the stimulation of fusion by M2a MPs and M2c MPs.
Few molecules that can account for the effects of MPs on muscle precursors are known. Recent work in mouse showed that Granulocyte-Colony Stimulating Factor (G-CSF) is beneficial to skeletal muscle regeneration by stimulating MPC proliferation . This study did not explore which cells were the source of G-CSF. However, MPs secrete G-CSF, particularly M1 MPs, and G-CSF has been shown to be associated with inflammation and activation by Th-1 cytokines [57, 58]. Exploring the role of CCR2 expressed by MPs, Lu et al.  have shown that MPs are likely to be the main source of IGF-1 in regenerating muscle; they further showed that the intramuscular Ly6Cneg MP subset, which presents an anti-inflammatory profile, expresses high levels of IGF-1, which promotes muscle repair. Here, we show the involvement of several cytokines and growth factors in the regulation of myogenesis by activated MPs. The proinflammatory cytokines IL-6 and IL-1β are mainly secreted by M1 MPs compared to M2a MPs and M2c MPs [11, 26], and blocking IL-6 or IL-1β affected only this MP subtype by increasing the formation of myotubes. Accordingly, IL-6 has a proliferative effect on MPCs [60, 61], while IL-1β has been shown to induce IL-6 production  and to reduce myogenic differentiation [63, 64]. TNFα is also produced in greater amounts by M1 MPs [11, 26]. The effects of TNFα inhibition depended on the concentration of antibodies used, in accordance with the dose-dependent effect of TNFα on muscle cell proliferation . At low concentrations, TNFα stimulates myogenic differentiation . Accordingly, decreased myotube formation was observed upon TNFα inhibition of M2a MPs and M2c MP-conditioned media, which contain low levels of TNFα. Conversely, high levels of TNFα stimulate myogenic cell proliferation  and M1 MPs increased myotube formation when the cytokine was completely inhibited by blocking antibodies. VEGF stimulates MPC growth  and differentiation [66, 67]. M1 MPs produce more VEGF than M2a MPs and M2c MPs . However, blocking VEGF induced an increase in myotube formation in all conditions, suggesting that the effect of VEGF on MPCs is mainly on cell growth. IL-13 regulates the secondary recruitment of reserve cells into myotubes, particularly on IGF-1-stimulated human cells  and increases fusion into myotubes . Unexpectedly, blocking IL-13 induced increased myotube formation in MPCs cultured with M1 MP-conditioned medium, suggesting that IL-13 participates in the nondifferentiating property of M1 MPs. TGFβ is expressed to a greater extent by M2a MPs and M2c MPs . TGFβ inhibits cell myogenesis at the time of differentiation [70–72] while it decreases cell growth and activates myogenesis when added to proliferating myoblasts . Here, TGFβ participated in the prodifferentiating cues delivered by M2a MPs and M2c MPs to MPCs since its blockade reduced myotube formation in these two conditions. These results identify an initial series of effectors differentially secreted by activated MPs which are involved in the regulation of in vitro myogenesis through their effects on MPCs.
It is likely that the MP activation states used for in vitro studies are not fully representative of MP subsets found in vivo [14, 74]. Therefore, we followed, to our knowledge for the first time in human, the regeneration of skeletal muscle in muscle biopsies obtained from healthy young men who had performed maximal lengthening contractions of the quadriceps muscles—a model that has previously been used in many human studies of muscle damage [21, 75-77]. However, not all studies using this model have observed damage in muscle biopsies collected in the days after exercise [20, 78]. Furthermore, in a previous study , electrical stimulation was reported to result in greater muscle damage than voluntary contractions. In light of this, and in order to ensure initiation of damage and regeneration processes, we increased the load delivered to the muscle by electrically stimulating the vastus lateralis muscle while the subject was contracting the muscle voluntarily. Analysis of muscle biopsies collected before and after this stimulus to the muscle enables distinct muscle regenerative events to be followed clearly in humans in vivo.
We investigated the expression of a series of MP markers, together with myogenic markers, during the time course of regeneration in human muscle. The first lesson from this study is that MP subsets cannot definitively be identified by a sole marker, and that multiple labelings are required to attempt to define MP subsets in vivo. We analyzed the expression of iNOS, COX2, and Arg1 by CD68pos cells. Colabeling and analysis at the single-cell level are necessary since many other cells, including endothelial cells, interstitial cells, and myogenic cells, also express these molecules [79–81]. CD206 and CD163 are considered as markers for anti-inflammatory subsets (supporting information Fig. S1). Careful examination of the immunostainings showed that several levels of fluorescence brightness were observed for these markers (data not shown), in accordance with their increased expression by anti-inflammatory MPs [82, 83]. The second lesson from this series of experiments is conceptual. Our results clearly show that, according to the six markers used here to identify MPs in vivo, several subsets of MPs are present in the same areas at the same time. These results argue against the overly simplistic view that a biological event is associated with one type of MPs. Also found in the same regenerating areas were cycling myogenic cells (CD56pos or Pax7pos ki67pos) together with differentiating cells (myogeninpos). Therefore, although general conclusions regarding the time course of some major biological events can be drawn, one must keep in mind that these events spatially and temporally overlap.
We present here the first time course of analysis of inflammation during human skeletal muscle regeneration. Our results confirm that both in normal muscle and once regeneration is completed, almost no MP is present in the muscle parenchyma. Regenerating areas are associated with MPs, the vast majority (more than 80%) being associated with MPs expressing proinflammatory and anti-inflammatory markers. However, despite this complex situation, our analysis depicts a time course of inflammation during human skeletal muscle repair that resembles what has been observed in mouse [11, 84]. When analyzing regenerating areas on the basis of whether they did or did not contain myogeninpos cells, a segregation in MP distribution was observed. Regenerating areas that contained only myogeninneg cells were preferentially associated with MPs expressing proinflammatory markers. Conversely, regenerating areas containing at least one myogeninpos cell were preferentially associated with MPs expressing anti-inflammatory markers. The evidence of such localization indicates privileged interactions between the two cell types and argues for a synchronous regulation of adult myogenesis and inflammation during muscle repair. The distinction between M2a MPs and M2c MPs during muscle regeneration is not yet clear. It is likely that the so-called anti-inflammatory subset(s) include(s) some MPs that stimulate matrix remodeling and deposition (thus resembling M2a MPs). Perdiguero et al.  showed that three MP subsets are successively present during muscle repair: inflammatory MPs, anti-inflammatory MPs, and a deactivated MP subset associated with the last steps of muscle repair. “Canonical” M2a MPs have been identified under very specific conditions, that is, parasitic challenge; it is therefore not known whether this type of MPs may be also—even transiently—present during normal tissue repair.
Combining our in vivo and in vitro data indicates that human skeletal muscle regeneration is sequentially associated with several MP subsets. Furthermore, we provide evidence of new functions of MPs in the coordinated regulation of MPC fate during muscle repair. In vivo analyses in a model of skeletal muscle regeneration in the healthy adult revealed that regenerating areas at the time of MPC proliferation are associated with M1 MPs, while in vitro experiments accordingly showed that M1 MPs migrate toward MPCs, stimulate their proliferation, and prevent premature differentiation through the secretion of IL-6, IL1-β, high TNFα, VEGF, and IL-13. In vivo immunostainings showed that later, at the time of terminal differentiation, regenerating areas are associated with M2c MPs and/or M2a MPs. In line with this, in vitro experiments showed that M2a MPs and M2c MPs attract MPCs and promote in vitro myogenesis by stimulating their commitment into myocytes and the formation of mature myotubes, through the secretion of TGFβ and low TNFα. Thus, MPs orchestrate an effective skeletal muscle repair process by finely tuning the sequential steps of adult myogenesis. These results strengthen the emerging concept that inflammation regulates stem cell homeostasis.
This work was supported by European Community's Seventh Framework Programme in the project FP7-Health—2009 ENDOSTEM 241440, the Agence Nationale pour la Recherche (ANR), the Association Française contre les Myopathies (AFM), the Nordea Foundation (Healthy Aging grant), the Danish Ministry of Health, and MYOAGE (nr. 223576) funded by the European Commission under FP7. The F5D monoclonal antibody developed by Woodring E. Wright was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. We thank the “Banque de Cellules APHP—Hôpital Cochin,” the Plate-forme d'Imagerie Cellulaire Pitié Salpêtrière for the migration experiments, and the Cochin Imaging Facility.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
None of the authors have a conflict of interest to report.