ARTICLE https://doi.org/10.1038/s41467-019-10438-z OPEN Fibroadipogenic progenitors are responsible for muscle loss in limb girdle muscular dystrophy 2B Marshall W. Hogarth1, Aurelia Defour1, Christopher Lazarski2, Eduard Gallardo3,4, Jordi Diaz Manera3,4, Terence A. Partridge1,5, Kanneboyina Nagaraju1,5,6 & Jyoti K. Jaiswal 1,5 Muscle loss due to fibrotic or adipogenic replacement of myofibers is common in muscle 1234567890():,; diseases and muscle-resident fibro/adipogenic precursors (FAPs) are implicated in this process. While FAP-mediated muscle fibrosis is widely studied in muscle diseases, the role of FAPs in adipogenic muscle loss is not well understood. Adipogenic muscle loss is a feature of limb girdle muscular dystrophy 2B (LGMD2B) – a disease caused by mutations in dysferlin. Here we show that FAPs cause the adipogenic loss of dysferlin deficient muscle. Progressive accumulation of Annexin A2 (AnxA2) in the myofiber matrix causes FAP differentiation into adipocytes. Lack of AnxA2 prevents FAP adipogenesis, protecting against adipogenic loss of dysferlinopathic muscle while exogenous AnxA2 enhances muscle loss. Pharmacological inhibition of FAP adipogenesis arrests adipogenic replacement and degeneration of dysferlin- deficient muscle. These results demonstrate the pathogenic role of FAPs in LGMD2B and establish these cells as therapeutic targets to ameliorate muscle loss in patients. 1 Research Center for Genetic Medicine, Children’s Research Institute, Children’s National Health System, Washington, DC 20010, USA. 2 Children’s Research Institute, Children’s National Health System, Washington, DC 20010, USA. 3 Unitat and Laboratori de Malaltdies Neuromusculars, Institut de Recerca, Hospital de la Santa Creu i Sant Pau, Universitat Autònoma de Barcelona, Barcelona 08041, Spain. 4 Centro de Investigación Biomédica en Red sobre Enfermedades Raras (CIBERER), Unitat de Malalties Neuromusculars, Servei de Neurologia, Hospital de la Santa Creu i Sant Pau de Barcelona, Barcelona 08041, Spain. 5 Department of Genomics and Precision Medicine, George Washington University School of Medicine, Washington, DC 20052, USA. 6 Department of Pharmaceutical Sciences, School of Pharmacy and Pharmaceutical Sciences, Binghamton University, Binghamton, NY 13902, USA. Correspondence and requests for materials should be addressed to J.K.J. (email: jkjaiswal@cnmc.org) NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z A lthough composed of terminally differentiated multi- and if AnxA2 accumulation is a mediator of this process. By using nucleated myofibers, adult skeletal muscle maintains a dysferlinopathic patient and mouse models, we show that FAP remarkable ability to regenerate from injury. This ability accumulation and their adipogenic differentiation are key con- depends on mono-nucleated cells that reside amongst the skeletal tributors to this muscular dystrophy. Importantly, the presence myofibers and those that enter the muscle following injury. With of extracellular AnxA2 promotes FAP proliferation and adipo- the ability of the satellite cells to proliferate and fuse to regenerate genic differentiation, and the loss of AnxA2 or pharmacologically damaged myofibers, they have long been identified as the primary inhibiting FAP adipogenesis significantly ameliorates the driver of regeneration. Accordingly, ablating the Pax7+ satellite dysferlin-deficient muscle pathology. This work identifies FAPs cells in adult mice blocks myofiber regeneration1–3. However, and their adipogenic differentiation as a major contributor to there is growing evidence that myofiber regeneration involves dysferlin-deficient muscle loss. By identifying approaches to tar- complex multicellular and extracellular matrix interactions get FAP proliferation and adipogenic differentiation, we provide creating a regenerative niche that consists of secreted factors, novel therapeutic targets for treating LGMD2B. immune cells, myogenic and non-myogenic progenitors4–7. Fibro/adipogenic precursors (FAPs) are muscle-resident non- Results myogenic progenitors of mesenchymal origin that are marked by Muscle adipogenesis determines LGMD2B onset and severity. cell surface expression of platelet-derived growth factor receptor MRI and histological analyses have identified fatty replacement of alpha (PDGFRα) and stem cell antigen-1 (Sca1) that proliferate muscle in symptomatic dysferlinopathic patients34 and mouse in response to injury and can undergo fibrogenic or adipogenic models29,35. By direct histological analysis of muscle sections differentiation8–10. Muscle injury triggers an acute myofiber from LGMD2B patients and mouse model, we examined how this repair response, failure of which causes myofiber death and association relates to disease severity. We obtained muscle resulting tissue infiltration by inflammatory cells11–13. These cells biopsies from LGMD2B patients who exhibited mild to severe clear the debris from the injury site and activate both satellite clinical symptoms described in Supplementary Table 1. As a first cell and FAP proliferation14,15. A critical element in the regen- step, we used the neutral lipid stain Oil Red O to score the adi- erative process is transition of the pro-inflammatory cells to pogenic status of muscle sections from these patients. While become pro-regenerative within 2–3 days after injury16,17. This sections of healthy muscle showed little to no oil red staining, coincides with the apoptotic clearance of FAPs and with satellite extensive staining was noted between the myofibers in sympto- cell fusion leading to myogenesis. Timely occurrence of the above matic patient muscle sections, which increased with the severity cellular choreography between inflammatory, fibro/adipogenic, of the patient’s clinical symptoms and the extent of muscle loss and satellite cells has been implicated in successful muscle (Fig. 1a and Supplementary Fig. 1). To further examine if the regeneration. Consequently, disrupting inflammatory infiltration adipogenic deposits were originating from within the myofibers and FAP homeostasis impairs regeneration, resulting in fibrotic we examined the localization of Perilipin-1, a protein that coats and adipogenic degeneration of injured muscle8,18. This deficit the adipocyte lipid droplets36. No perilipin-1 staining was has been demonstrated in the mouse model of Duchenne mus- detectable in healthy muscle, but patient muscles showed exten- cular dystrophy (mdx mice), where impaired FAP clearance sive perilipin-1 labeling, which increased with disease severity and results in muscle loss and fibrosis19. Facilitating apoptotic clear- localized exclusively in the extracellular matrix space between the ance of FAPs reduces muscle loss and improves mdx muscle myofibers (Fig. 1b, c). Patient myofibers did not show internal function in vivo19. perilipin-1 labeling even when they are adjacent to lipid deposits Adipogenic differentiation of FAPs has been implicated in (Fig. 1c and Supplementary Fig. 1), suggesting extra-myofiber muscle loss following rotator cuff injury in mice20. While adi- origin of these lipids. pogenic muscle replacement is prevalent in muscular dystrophies, To independently assess the link between disease severity and it remains to be determined if FAPs are responsible for this. the extent of adipogenic replacement of muscle, we examined The dysferlinopathies represent a heterogeneous group of late- muscles from 12-month old dysferlin-deficient (B6A/J) mice onset muscle disease, including limb girdle muscular dystrophy (Fig. 1d). Disease severity, as indicated by the extent of damage type 2B (LGMD2B), which are caused by mutations in the and regeneration (myofiber central nucleation) showed a dysferlin gene21–25. Lack of dysferlin compromises myofiber progressive increase between muscles in the following order: repair, alters calcium homeostasis, and causes chronic muscle TA, gastrocnemius, quadriceps, psoas (Fig. 1e). Labeling with inflammation21,26,27. However, these deficits do not explain perilipin-1 showed a parallel increase in lipid accumulation the late and abrupt disease onset, progressive nature, or specific between the myofibers across these same muscles (Fig. 1d, f). muscle involvement seen in patients or mouse models. Recently it Similarly, muscle (gastrocnemius) collected from mice with has been demonstrated that affected muscles of dysferlinopathic increasing age (6–18 months old) showed progressively increas- patients and mouse model show adipogenic replacement28. ing adipogenic replacement marked by increased labeling with Unlike the myofiber repair deficit and inflammation, adipogenic either Oil Red O (Fig. 1i) or perilipin-1 (Fig. 1j). Interestingly, replacement is observed only in symptomatic patient and mouse while increased perilipin-1 staining is detected starting 12 months muscle28,29. Further, eccentric exercise exacerbates this pheno- (Fig. 1h), these muscles showed an increased central nucleation type in patients30, suggesting a link between myofiber injury and starting from 6 months (Fig. 1g), suggesting that myofiber adipogenic replacement of LGMD2B muscle. Muscle damage damage and regeneration precedes their later adipogenic and disease severity in LGMD2B patients correlate with increased replacement. Again, in the mouse muscle we observed perilipin- expression of another membrane repair protein Annexin A2 1 labeling only in the extracellular matrix and not in adjacent (AnxA2)31,32. Recently we described that dysferlin-deficient mice myofibers (Fig. 1k), suggesting that the lipid does not originate in lacking AnxA2 have reduced myofiber repair ability, but are muscle fibers, and instead is produced by muscle interstitial cells. surprisingly protected from adipogenic myofiber loss33. This suggested that loss of AnxA2 disrupts the link between injury and adipogenic replacement of dysferlin-deficient myofibers. FAPs cause the adipogenic loss of dysferlinopathic muscle. The Here, we study the effect of dysferlin loss on the homeostasis above results suggest the extent of muscle damage and regen- of muscle-resident FAPs. We examine if altered FAP biology can eration preempts the degree of adipogenic replacement of dys- explain the late onset muscle-specific symptoms in LGMD2B ferlinopathic muscle. With the known proliferation of FAPs 2 NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z ARTICLE a Healthy Mild Moderate Severe c LGMD2B biopsy Laminin-DAPI Oil Red O b Perilipin-1-DAPI Perilipin-1-DAPI d 12Mo B6A/J - TA 12Mo B6A/J - GST 12Mo B6A/J - QUAD 12Mo B6A/J - PSOAS Perilipin-1-DAPI Merge e f g h k 12Mo B6A/J GST Perilipin-1 area (μm2/100 μm2) 40 15 30 Perilipin-1 area (μm2/100 μm2) 3 ** * * ** *** *** *** *** Central nuclei (%) Central nuclei (%) 30 10 20 2 Laminin-DAPI 20 ** ** 5 10 1 10 0 0 0 0 TA GST QUAD PSOAS TA GST QUAD PSOAS WT B6A/J B6A/J B6A/J B6A/J WT B6A/J B6A/J B6A/J B6A/J 12Mo 2Mo 6Mo 12Mo 18Mo 12Mo 2Mo 6Mo 12Mo 18Mo i 12Mo WT 6Mo B6A/J 12Mo B6A/J 18Mo B6A/J Perilipin-1-DAPI Oil Red O j Perilipin-1-DAPI Merge Fig. 1 Adipogenic replacement of muscle correlates with disease severity in dysferlinopathic patients and mice. Mild, moderate, and severely dystrophic LGMD2B patient and non-dystrophic control muscle cross-section stained with a Oil Red O or for b Perilipin-1 protein. c Confocal images of LGMD2B patient muscle sections showing that perilipin-1 marked lipid deposits (red) accumulate outside the boundaries of laminin-marked myofiber borders (green). Scale bar = 20 µm. d 12Mo B6A/J TA, gastrocnemius, quadriceps, and psoas muscles stained for perilipin-1. Scale Bar = 100 µm. Quantification (mean ± SD) of e myofiber central nucleation and f perilipin-1 area across 12Mo B6A/J muscles. Statistical comparisons performed via t-test between adjacent groups, n = 4 muscles/group. Quantification (mean ± SD) of g myofiber central nucleation and h perilipin-1 area from B6A/J gastrocnemius with advancing age/pathology, n = 4 mice/group. Statistical comparisons performed via ANOVA with Holm–Sidak multiple comparisons test for all means with that of 12Mo WT, *p < 0.05 **p < 0.01, ***p < 0.001. i Oil Red O and j Perilipin-1 labeling of gastrocnemius from 6, 12, and 18Mo B6A/J and 12Mo WT. Scale bar = 100 µm. k Confocal image of gastrocnemius muscle sections showing that perilipin-1 marked lipid deposits (red) localize outside the boundaries of laminin-marked myofiber borders (green) in 12Mo B6A/J. Scale bar = 20 µm in the muscle interstitium following injury, and their adipogenic severity (Fig. 2a, d). Analysis of mouse muscles showed a similar potential, we examined if FAPs lead to adipogenic conversion of interstitial accumulation of PDGFRα in B6A/J, which correlates LGMD2B muscle. Using PDGFRα to label FAPs, control biopsies both to extent of disease severity across muscles at the same age showed minimal interstitial PDGFRα staining, which was evident (Fig. 2b, e), and to increasing disease severity due to age of the in LGMD2B patients and increased with worsening clinical muscle (gastrocnemius) from 6 months onwards (Fig. 2c, f). This NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z a Healthy Mild Moderate Severe g PDGFRα-DAPI Perilipin-1-DAPI Merge PDGFRα-WGA-DAPI b 12Mo B6A/J-TA 12Mo B6A/J-GST 12Mo B6A/J-QUAD 12Mo B6A/J-PSOAS h Uninduced Induced PDGFRα-WGA-DAPI Whole Suspension c 12Mo WT 6Mo B6A/J 12Mo B6A/J 18Mo B6A/J i k 4 10 PDGFRα-WGA-DAPI 3 10 PDGFRα – – PDGFRα 2 FITC 10 + 1 PDGFRα 10 0 10 0 1 2 3 4 10 10 10 10 10 Isotype-APC d e f j 10 4 PDGFRα+ area (μm2/100 μm2) 5 PDGFRα+ area (μm2/100 μm2) 15 PDGFRα+ area (μm2/100 μm2) 15 3 4 *** 10 PDGFRα – Severe *** 10 + 10 FITC PDGFRα 3 2 10 2 ** 1 PDGFRα + 5 Moderate 5 ** 10 * 1 Mild * 0 10 0 0 0 0 1 2 3 4 12Mo 2Mo 6Mo 12Mo 18Mo 10 10 10 10 10 Healthy LGMD TA GST QUAD PSOAS WT B6A/J B6A/J B6A/J B6A/J PDGFRα-APC Fig. 2 FAP accumulation and differentiation dictates the extent of adipogenic replacement of dysferlin-deficient muscle. PDGFRα staining of FAPs in a mild, moderate, and severe LGMD2B patient and non-dystrophic control muscle cross-sections; b muscles with increasing pathology from 12Mo B6A/J and c gastrocnemius muscle cross-sections from 6, 12, and 18Mo B6A/J and 12Mo WT control. Scale bar = 20 µm. Quantification (mean ± SD) of muscle area labeled with PDGFRα in d human and mouse muscle with e increasing pathology and f increasing age. n = 4 mice/group. Statistical comparisons performed via ANOVA with Holm–Sidak multiple comparisons test for all means with 12Mo TA or 12Mo WT respectively, *p < 0.05, **p < 0.01, ***p < 0.001. g Gastrocnemius muscle cross-section of 12Mo B6A/J co-labeled for PDGFRα and Perilipin-1. Scale bar = 50 µm. h Oil Red O staining of primary cell suspension isolated from hindlimb muscle of 6Mo B6A/J prior to (uninduced) or following (induced) induction for adipogenic differentiation. Scale bar = 200 µm. Inset shows a zoomed image of the box drawn in h. Oil Red O staining of sorted cells i lacking or j expressing PDGFRα at their cell surface. Scale bar = 200 µm. k FACS scatter plots showing cell surface labeling of PDGFRα in primary cell suspension prepared from mouse hindlimb muscle. The regions drawn mark the cells that express or lack cell surface PDGFRα expression is a feature of the dysferlinopathic mice, and is not observed in However, PDGFRα+ cells formed adipocytes spontaneously, the healthy, wild type (WT) mice (Supplementary Fig. 2). which increased further upon adipogenic induction (Fig. 2j). PDGFRα-labeled FAP increase occurred concomitantly with the These results identify FAPs as the muscle interstitial cells that increase in central nucleation, but prior to adipogenic replace- contribute to the adipogenic loss of dysferlin-deficient muscle. ment of the myofibers (Fig. 2f compared to Fig. 1f). This obser- vation suggests that persistent myofiber injury and regeneration causes FAP accumulation. Subsequent adipogenic differentiation Adipogenic muscle loss depends on age and myofiber injury. of these FAPs likely causes adipogenic replacement of the myo- Based on above observations we hypothesized that age, chronic fibers, as indicated by the formation of lipid deposits in extra- injury, and poor repair of dysferlinopathic myofibers create a cellular matrix regions being enriched for PDGFRα-marked niche that promotes progressive interstitial FAP accumulation FAPs (Fig. 2g). and subsequent differentiation into adipocytes. To test this, To directly assess the role of FAPs in adipogenic replacement we examined if injury and age of the TA (a muscle that is of dysferlinopathic myofibers, we obtained a mixed primary cell minimally affected in B6A/J) can unmask the adipogenic poten- suspension from hindlimb muscles of 6Mo B6A/J mice. Inducing tial of the FAPs and effect of tissue injury and age on this process. cells with adipogenic media caused a proportion of cells to TA muscles from 3Mo and 12Mo B6A/J mice were injured by differentiate into oil red-labeled adipocytes, such that even when notexin and after allowing 4 weeks for myofibers to fully regen- these adipocytes were found adjacent to myotubes, the myotubes erate, the extent of adipogenic muscle replacement was scored themselves were not producing the lipids (Fig. 2h, and inset). This and compared to the uninjured 12mo B6A/J TA (Supplementary further supports that non-myogenic cells produce the adipogenic Fig. 3). The uninjured 12Mo, and the injured 3Mo B6A/J TAs material that eventually replace myofibers. To establish the both lack significant adipogenic replacement (assessed by identity of these adipogenic cells we used fluorescence-activated perilipin-1 staining) (Fig. 3a–c). In contrast, the injured 12Mo cell sorting to isolate PDGFRα-labeled (PDGFRα+) FAPs and B6A/J TA showed substantially increased adipogenic foci and PDGFRα-unlabeled (PDGFRα−) cells from a primary muscle myofiber areas replaced by adipogenesis (Fig. 3a–c). This showed cell suspension (Fig. 2k). Even upon adipogenic induction, no that increasing age increases injury-triggered adipogenic repla- adipocyte formation was observed in PDGFRα− cells (Fig. 2i). cement of areas otherwise occupied by myofibers. This link is 4 NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z ARTICLE a 12Mo B6A/J-Uninjured 3Mo B6A/J-Single injury 12Mo B6A/J-Single injury b c Perilipin-1 area (μm /100μm ) 2.5 200 2 **** **** **** ** No. of perilipin-1 foci Perilipin-1-WGA-DAPI 2.0 150 2 1.5 100 1.0 50 0.5 0.0 0 12Mo B6A/J 3Mo B6A/J 12Mo B6A/J 12Mo B6A/J 3Mo B6A/J 12Mo B6A/J Uninjured Single injury Single injury Uninjured Single injury Single injury d 12Mo WT-Single injury 12Mo B6A/J-Single injury 12Mo B6A/J-Repeat injury e f *** **** Perilipin-1 area (μm /100μm ) 5 300 * ** 2 * *** No. of perilipin-1 foci Perilipin-1-WGA-DAPI 4 2 200 3 2 100 1 0 0 12Mo WT 12Mo B6A/J 12Mo B6A/J 12Mo WT 12Mo B6A/J 12Mo B6A/J Single injury Single injury Repeat injury Single injury Single injury Repeat injury g 12Mo B6A/J-Uninjured 12Mo WT-Single injury 3Mo B6A/J-Single injury 12Mo B6A/J-Single injury h **** 2.5 *** PDGFRα area (μm /100μm ) 2 * Perilipin-1-WGA-DAPI 2.0 Control Notexin 2 1.5 1.0 * + 0.5 0.0 12Mo WT 3Mo B6A/J 12Mo B6A/J Fig. 3 Injury induces FAP proliferation in older but not younger dysferlinopathic muscle. Images showing Perilipin-1 staining in cross-section of TA muscles from 2 and 12Mo in B6A/J or WT Bl6 mice following 4 weeks after a a single or d three rounds of notexin injury. Scale bar = 50 µm. Quantification (mean ± SD) in TA muscle cross-section of b, e total area and c, f number of foci labeled with perilipin-1 after 4 weeks of b, c single or e, f 3-rounds of notexin injury. n = 6 per group. Statistical comparisons performed via ANOVA with Holm–Sidak multiple comparisons test to compare all means with Uninjured 12Mo B6A/J, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. g Images showing PDGFRα staining after single notexin injury of TA muscle from 12Mo WT, 2Mo and 12Mo B6A/J. Scale bar = 20 µm. h Quantification (mean ± SD) of PDGFRα-labeled area of the muscle after notexin injury, n = 6 mice/group. Statistical comparisons between control and notexin injured performed via two-tailed t-test for each group. Comparisons between notexin injured muscles is by ANOVA with Holm–Sidak multiple comparisons test to compare all means with 12Mo WT, *p < 0.05, ***p < 0.001, ****p < 0.0001 further strengthened by the observation that adipogenic repla- AnxA2 links muscle injury to FAP activation and adipogenesis. cement occurs at the site of notexin injection (marked by tattoo The above link between FAP accumulation and adipogenic dye), while adjacent uninjured regions remained non-adipogenic replacement of dysferlinopathic muscle suggests that accumu- (Supplementary Fig. 3). These results establish muscle injury and lation and adipogenic differentiation of FAPs is responsible for age enhance adipogenic replacement of dysferlinopathic muscle. the decline in dysferlinopathic muscle function, and reversing To confirm that this post-injury lipid formation was due to the this could provide a therapy for LGMD2B. AnxA2 is a mem- absence of dysferlin, we also injured TAs from 12Mo WT, which brane repair protein that is elevated upon muscle injury and showed no significant lipid formation after injury (Fig. 3d–f). increases in LGMD2B patient muscle in a manner that corre- Dysferlinopathic muscle undergoes repeat bouts of injury over the lates with disease severity31,32. Previously, we showed that disease course, which we hypothesize leads to adipogenic deletion of AnxA2 in dysferlinopathic muscle reduces adipo- replacement of the muscle. To test this, we performed three genic replacement and improves muscle function despite no consecutive injuries to the TA muscle in dysferlinopathic mice, improvement in myofiber repair33. Thus, we examined if allowing 2-week regeneration periods between the injuries AnxA2 contributes to FAP proliferation/adipogenic differ- (Supplementary Fig. 3). This resulted in more adipogenic entiation in dysferlinopathic muscle. Immunostaining for replacement of the TA muscle than in the single injured TA AnxA2 showed that in dysferlinopathic mice AnxA2 level muscle (Fig. 3d–f), suggesting that repeat rounds of myofiber increases with disease severity and that this increase is in the injury and regeneration drive the adipogenic conversion of level of AnxA2 in the interstitium (Fig. 4a, b, e). Co-labeling dysferlinopathic muscle. To examine if this injury-dependent with PDGFRα showed that the FAPs are enriched in the regions adipogenic replacement was a consequence of increased FAP with accumulation of AnxA2 (Fig. 4c, d). AnxA2 expression accumulation caused by the injuries, we stained these muscles increases after muscle injury, and extracellular AnxA2 can for PDGFRα (Fig. 3g). Compared to uninjured B6A/J muscles activate inflammation via toll-like receptor 4 (TLR4)37,38. We and WT muscles, there was a significant increase in PDGFRα- thus investigated if the inflammatory cells are enriched at the labeled FAPs in injured B6A/J muscles (Fig. 3h). Injured 12Mo sites of AnxA2 accumulation, and found this to be the case; B6A/J muscle accumulates significantly more FAPs than both F4/80 marked macrophages accumulated in interstitial regions the 3Mo B6A/J and the 12Mo WT injured muscles, indicating enriched for AnxA2 (Fig. 4f), and are located adjacent to the that excessive FAP accumulation after injury is a feature of PDGFRα-labeled FAPs (Fig. 4g). To analyze if AnxA2 accu- dysferlin-deficient muscle with advancing pathology, and likely mulation is causally linked to adipogenic replacement via FAPs underpins the adipogenic loss of these muscles. we examined adipogenic replacement and FAP accumulation NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z 12Mo WT 6Mo B6A/J 12Mo B6A/J 18Mo B6A/J a AnxA2-WGA-DAPI b 15 c 80 d *** % AnxA2 overlapping PDGFR *** μm2) PDGFRα-AnxA2-DAPI 60 (μm2/100 10 *** *** *** 40 * area 5 AnxA2+ 20 * 0 0 12Mo 2Mo 6Mo 12Mo 18Mo 12Mo 2Mo 6Mo 12Mo 18Mo WT B6A/J B6A/J B6A/J B6A/J WT B6A/J B6A/J B6A/J B6A/J AnxA2-laminin-DAPI AnxA2-F4/80-DAPI PDGFRα-F4/80-DAPI e f g 12Mo B6A/J 12Mo A2-B6A/J 12Mo B6A/J 12Mo A2-B6A/J h I i Perilipin-1-DAPI Oil Red O Fig. 4 Extracellular accumulation of AnxA2 drives FAP accumulation in dysferlinopathic muscle. a AnxA2 staining in cross-sections from 6, 12, and 18Mo B6A/J and 12Mo WT mouse gastrocnemius muscle. Scale bar = 20 µm. b Quantification (mean ± SD) of total AnxA2-labeled area across the entire muscle cross-section, n = 4 mice/group. Statistical comparisons performed via ANOVA with Holm–Sidak multiple comparisons test to compare all means with 12Mo WT, *p < 0.05, **p < 0.01, ***p < 0.001. c Quantification (mean ± SD) of d co-localization between AnxA2 and PDGFRα in 12Mo B6A/J gastrocnemius, n = 4 mice/group. Statistical comparisons performed via ANOVA with Holm–Sidak multiple comparisons test to compare all means with Uninjured 12Mo WT, *p < 0.05, ***p < 0.001. Confocal images from 12Mo B6A/J gastrocnemius co-labeled for e AnxA2 and laminin, f AnxA2 and F4/80, and g PDGFRα and F4/80. Scale bar = 20 µm. h Oil Red O and i Perilipin-1 labeling of gastrocnemius cross-sections from 12Mo B6A/J and A2-B6A/J. Scale bar = 50 µm in mice lacking dysferlin and AnxA2 (A2-B6A/J). Oil Red O As we find macrophages are enriched at sites of PDGFRα and perilipin-1 labeling both reveal significantly less adipogenic and AnxA2 accumulation, we next examined if the lack of replacement of muscle in A2-B6A/J than B6A/J mice (Fig. 4h, i, extracellular AnxA2 works by inhibiting muscle inflammatory Supplementary Fig. 4 and quantified in Fig. 5d). This indicates response. F4/80 staining of 12Mo gastrocnemius muscles showed that extracellular AnxA2 accumulation contributes to the pro- that compared to the WT mice, both B6A/J and A2-B6A/J adipogenic niche as AnxA2 deletion arrests the adipogenic mice show a robust increase in macrophage infiltration (Fig. 5a, b). conversion of dysferlinopathic muscle. The extent of macrophage infiltration in these mouse muscles 6 NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z ARTICLE WT B6A/J A2-B6A/J a F4/80-WGA-DAPI b * c d 5 *** * 30 *** 1.5 Perilipin-1 area (μm2/100 μm2) *** *** ** ** F4/80+ area (μm2/100 μm2) 4 Central nuclei (%) 20 1.0 3 2 10 0.5 1 0 0 0.0 WT B6A/J A2-B6A/J WT B6A/J A2-B6A/J WT B6A/J A2-B6A/J e 2.5 f 5 *** ** B6A/J ** 2.0 4 A2-B6A/J PDGFRα+ area PDGFRα+ area (μm2/100 μm2) (μm2/100 μm2) 1.5 3 ** 1.0 2 * 0.5 1 0.0 0 WT B6A/J A2-B6A/J 6Mo 12Mo >18Mo 6Mo A2-B6A/J 12Mo A2-B6A/J 24Mo A2-B6A/J g PDGFRα-WGA-DAPI Fig. 5 Presence of AnxA2 is required for adipogenic conversion of dysferlinopathic muscle. a F4/80 staining of gastrocnemius of 12Mo WT, B6A/J and A2- B6A/J. Scale bar = 20 µm. Quantification (mean ± SD) of b F4/80 labeled area, c myofiber central nucleation, d Perilipin-1 labeled area, and e PDGFRα labeled area from 12Mo WT, B6A/J and A2-B6A/J, n = 4/genotype. Statistical comparisons performed via ANOVA with Holm–Sidak multiple comparisons test to compare all means, *p < 0.05, **p < 0.01, ***p < 0.001. f Quantification (mean ± SD) of g PDGFRα-labeled area in 6, 12, and 24Mo A2-B6A/J gastrocnemius (scale bar = 20 µm) shown in comparison to B6A/J (as presented in Fig. 2f), n = 3 mice/group. Statistical comparisons performed by two- tailed t-test between B6A/J and A2-B6A/J at each timepoint, *p < 0.05, **p < 0.01 is in line with the extent of myofiber injury, indicated by the did not increase with age (Fig. 5f, g). This indicates that lack number of centrally nucleated myofibers in B6A/J and A2-B6A/J of AnxA2 prevents FAP accumulation, contributing to the mice as compared to WT mice (Fig. 5c). In contrast to the suppression of adipogenic replacement of dysferlin-deficient higher injury and inflammation of A2-B6A/J muscle (as muscle. compared to the WT), FAP (PDGFRα) accumulation and lipid We next examined if AnxA2 can also contribute to the (perilipin-1) formation were comparable between the A2-B6A/J adipogenic fate of the FAPs in dysferlin-deficient muscle. For this and WT muscle (Fig. 5d, e). This raises the possibility that we isolated FAPs from 12Mo WT, B6A/J and A2-B6A/J muscle the beneficial effect of the lack of AnxA2 may be by way of its and quantified their spontaneous adipogenesis in vitro. A small effect on FAP proliferation and adipogenic differentiation. To proportion of WT FAPs undergo spontaneous adipogenesis examine the effect of AnxA2 on FAP accumulation and on after 14 days in culture, but B6A/J FAPs show significantly higher preventing their adipogenic differentiation, we first quantified rates of adipogenesis (Fig. 6a, b). The spontaneous adipogenesis FAP accumulation using PDGFRα labeling. While B6A/J mice of B6A/J FAPs suggests these cells are committed to adipogenesis showed increasing accumulation of FAPs with age, muscles in prior to extraction, which may be caused by the pro-adipogenic the A2-B6A/J mice showed fewer FAPs, and their numbers niche in dysferlinopathic muscle. The lack of spontaneous NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z WT B6A/J A2-B6A/J a b 50 *** *** *** Oil Red O area (%) 40 Oil Red O 30 20 10 0 WT B6A/J A2-B6A/J B6A/J FAPs-B6A/J Niche B6A/J FAPs-A2-B6A/J Niche d 40 A2-B6A/J-NTX c Oil Red O area (%) 30 I PDGFRα-WGA-DAPI * Oil Red O 20 10 0 FAPs B6A/J B6A/J Niche B6A/J A2-B6A/J B6A/J FAPs-untreated B6A/J FAPs-AnxA2 f A2-B6A/J-NTX + AnxA2 e 30 *** PDGFRα-WGA-DAPI Oil Red O area (%) 20 Oil Red O 10 0 Untreated AnxA2 A2-B6A/J-NTX A2-B6A/J-NTX + AnxA2 h j 8 * 4 g * Perilipin-1-WGA-DAPI PDGFRα + area 6 3 Perilipin-1 area (μm2/100 μm2) (μm2/100 μm2) 4 2 2 1 0 0 NTX NTX + AnxA2 NTX NTX + AnxA2 Fig. 6 AnxA2 drives FAP adipogenesis and injury-triggered lipid formation in dysferlinopathic muscle. a Oil Red O staining and b quantification of spontaneous adipogenesis of 20,000 FAPs each, isolated from 12Mo WT, B6A/J and A2-B6A/J after 10 days in culture, n = 3 replicates/genotype. Scale bar = 50 µm. c Oil Red O staining and d quantification (normalized to 20,000 FAPs) of spontaneous adipogenesis from a 1:1 mixed culture of 24Mo B6A/J FAPs (10,000 cells) with niche cells (10,000 cells) from either 24Mo B6A/J or A2-B6A/J. Scale bar = 50 µm. e Oil Red O staining and f quantification of spontaneous adipogenesis of 20,000 24Mo B6A/J FAPs left untreated or treated with 100 nM recombinant AnxA2. Scale bar = 50 µm. g Perilipin-1 staining from 10Mo A2-B6A/J TA muscles 28d after single notexin injury with (NTX + AnxA2) or without (NTX) co-administration of 10 µg recombinant AnxA2. Scale bar = 100 µm. h Quantification of total perilipin-1 area after injury. i PDGFRα staining from 10Mo A2-B6A/J TA muscles 28d after single notexin injury and administration of 10 µg recombinant AnxA2 compared to notexin injury alone. Scale bar = 20 µm. j Quantification of total PDGFRα area after injury. All data presented as mean ± SD, average Oil Red area compared across genotypes by ANOVA with Holm–Sidak multiple comparisons test, all other data compared via two-tailed t-test, *p ≤ 0.05 ***p < 0.001 adipogenesis in A2-B6A/J FAPs could be caused by the lack of a staining/20,000 FAPs. Compared to the B6A/J niche, the A2- pro-adipogenic niche or restricted adipogenic potential of the B6A/J niche restricted the spontaneous adipogenesis of B6A/J FAPs in the absence of AnxA2 (Supplementary Fig. 4). As FAPs (Fig. 6c, d). Thus, the absence of AnxA2 in the niche cells adipogenic replacement of muscle is enhanced with age, to reduces the adipogenicity of the dysferlinopathic FAPs. To allow for the niche cells to reach their maximal potential we directly test if it is the AnxA2 or another factor secreted by the isolated the niche (PDGFRα−) and FAP (PDGFRα+) cells from niche cells that influences FAP adipogenesis we treated a purified 24Mo B6A/J and A2-B6A/J muscles. These FAP and niche cells population of B6A/J FAPs (20,0000 FAPs) with 100 nM AnxA2 were then mixed in a 1:1 ratio (20,000 cells total; 10,000 FAPs) and their spontaneous adipogenesis was quantified as the extent from the same or different genetic background and their of oil red staining/20,000 FAPs. Compared to the AnxA2 spontaneous adipogenesis was quantified as the extent of oil red untreated FAPs, treatment with purified AnxA2 alone caused a 8 NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z ARTICLE >2-fold increase in the spontaneous adipogenic differentiation of adipogenic differentiation of FAPs. Batimastat is a small molecule these dysferlinopathic FAPs (Fig. 6e, f). drug that has previously been shown to restrict adipogenesis of Taken together, the above results show that AnxA2 produced both cultured adipogenic precursors39 and WT mouse FAPs40. by the B6A/J muscle niche cells can potentiate the adipogenic To test the ability of batimastat to prevent the spontaneous adi- differentiation of B6A/J FAPs in vitro. We next examined if this pogenesis of B6A/J FAPs we treated 40,000 FAPs isolated from was also true in vivo. For this we used the AnxA2-naive A2-B6A/J 12Mo B6A/J mice with 10 µM batimastat starting from 3 days in mice and notexin injured their muscle with or without the culture. Using Oil Red O, we quantified the extent of adipogenic addition of 10 µg purified AnxA2 at the site of injury. The differentiation of these FAPs after 14 days in culture in vitro presence of exogenous AnxA2 at the site of muscle regeneration (Fig. 7a). We find that treatment with batimastat resulted in in these otherwise AnxA2-deficient muscles resulted in increased reduced spontaneous adipogenesis of the B6A/J FAPs (Fig. 7b). accumulation of the PDGFR+ FAPs as well as perilipin-1 labeled We next tested whether this effect of batimastat to repress FAP adipogenic deposits (Fig. 6g–j). This provides direct evidence adipogenesis can improve muscle histopathology in vivo. 12Mo supporting that extracellular AnxA2 is not only necessary, but B6A/J mice were treated for 10 weeks with batimastat (2 mg/kg i. also sufficient for driving FAP-mediated in vivo adipogenic p. 3× weekly). As B6A/J gastrocnemius muscle showed significant conversion of the regenerating dysferlinopathic muscles. adipogenic replacement starting from the age of 12Mo (Fig. 1i), we assessed for the extent of adipogenic loss in this muscle (Fig. 7c). Compared to the untreated mice, muscles from Blocking FAP differentiation arrests adipogenic muscle loss. batimastat treated mice showed significantly reduced perilipin-1 Our data from the A2-B6A/J model suggests that restricting the labeled area (Fig. 7d). To determine if this reduction in adipogenic conversion would preserve the dysferlinopathic adipogenic replacement was due to an effect on just FAP muscle and is thus a potential therapeutic target. To test this, and differentiation or also on FAP proliferation, we quantified FAP to independently confirm the benefit of inhibiting FAP adipo- accumulation by PDGFRα labeling and saw no effect of genesis for dysferlinopathy, we used a drug to inhibit the batimastat treatment on FAP number (Supplementary Fig. 5). Untreated Batimastat a b 40 Oil Red O area (%) 30 Oil Red O 20 *** 10 0 Untreated Batimastat d 1.5 Perilipin-1 area (μm2/100 μm2) c Perilipin-1-WGA-DAPI * 1.0 0.5 0.0 Untreated Batimastat f e 4 Perilipin-1 area (μm2/100 μm2) Perilipin-1-WGA-DAPI 3 * 2 1 0 Untreated Batimastat Fig. 7 Batimastat blocks FAP adipogenesis and reduces adipogenic loss of dysferlinopathic muscle. a Oil Red O staining and b quantification (normalized to 20,000 FAPs) of spontaneous adipogenesis from 40,000 FAPs each, isolated from 12Mo B6A/J, after 10 days in culture (n = 3 replicates/group). Cells were treated with 10 µM batimastat starting day 3 in culture. Scale bar = 50 µm. c Perilipin-1 staining of 14Mo gastrocnemius muscle sections after 10 weeks of batimastat treatment. Scale bar = 100 µm. d Quantification of perilipin-1 stained area in gastrocnemius muscle, n = 3 muscles/group. e Perilipin-1 staining of TA muscles from 14Mo B6A/J mice after 3 successive notexin injuries while being treated (or left untreated) with batimastat. Scale bar = 100 µm. f Quantification of injury-triggered lipid formation in notexin injured TA muscles corresponding to panel (e), n = 6 muscles/group. All quantifications shown are mean ± SD, and statistical comparisons were performed via two-tailed t-test, *p < 0.05, ***p < 0.001 NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z Similarly, batimastat treatment of 12Mo B6A/J mice did not a b decrease the extent of myofiber central nucleation (Supplemen- Injury tary Fig. 5). Thus, inhibition of adipogenesis by batimastat is not due to reduced FAP numbers or improved myofiber repair. To further confirm the ability of batimastat to prevent dysferlino- pathic muscle from injury-triggered FAP adipogenesis, we employed the repeat notexin-induced TA injury approach which induces significant adipogenic replacement of the muscle (Fig. 3d, e). Compared to the untreated mice, TA muscles from batimastat Regeneration treated mice showed a nearly 40% reduction in perilipin-1 positive area following repeat notexin injury (Fig. 7e, f). Again, d Therapeutic avenues c Degeneration Block FAP Arrest FAP this reduction was due to restriction of adipogenic differentiation accumulation adipogenesis of the FAPs and not due to an effect on their proliferation, as batimastat treatment did not reduce the FAP accumulation caused by repeat notexin injuries (Supplementary Fig. 5). Given that FAP numbers are unaffected by batimastat treatment, we also examined whether batimastat treatment causes cells to adopt a fibrogenic fate by blocking their adipogenic differentiation. However, we did not observe a change in intra-muscular fibrosis as a result of batimastat treatment in either the gastrocnemius or injured TA (Supplementary Fig. 5). Taken together, the above results both confirm the essential nature of FAP adipogenesis for Undifferentiated FAPs Inflammatory cells Injury site the adipogenic conversion of dysferlinopathic muscle, and Lipid deposits Adipogenic FAPs AnxA2 highlight the potential of batimastat to arrest the progressive and even late stage adipogenic replacement of dysferlinopathic Fig. 8 FAPs control the onset and severity of disease in LGMD2B. a Healthy muscle. and/or pre-symptomatic LGMD2B muscle contains resident FAPs. b After myofiber injury, inflammatory cells invade and trigger FAP proliferation. Successful regeneration involves a switch between pro-inflammatory and Discussion pro-regenerative signaling, causing the removal of inflammatory cells and Myofiber loss and the associated muscle weakness is a feature of FAPs. c In symptomatic LGMD2B muscle, there is a gradual accumulation many muscular dystrophies. Thus, in the search for innovative of extracellular AnxA2, which prolongs the pro-inflammatory environment, therapies, it is important to elucidate the cellular mechanisms causing excessive FAP proliferation. This cellular niche becomes pro- that link the initial genetic defect to disease onset and progres- adipogenic over time, allowing for differentiation of FAPs and the sion. In LGMD2B, the absence of dysferlin in myofibers inhibits adipogenic conversion of muscle. d Blocking aberrant signaling due to sarcolemmal repair21, disrupts proper calcium homeostasis at AnxA2 buildup blocks FAP accumulation and thus preventing adipogenic the t-tubules27,41,42, and alters the response of innate immune loss of dysferlinopathic muscle. Similarly, use of a MMP-14 inhibitor cells43,44. Restoration of dysferlin expression in myogenic cells45 (Batimastat) inhibits FAP adipogenesis offering a potential drug-based and blockade of innate immune activation46 reduce pathology in therapy to prevent adipogenic loss of dysferlinopathic muscle dysferlin-deficient mice, indicating that both the myofiber and inflammatory cell-specific deficits contribute to disease symptoms in dysferlinopathy. However, these deficits fail to explain the commitment to adipogenesis (Figs. 4 and 5). The presence of abrupt and late onset of disease in patients and the observation AnxA2 expressing niche cells in the muscle increases FAP adi- that adipogenic replacement is a feature of symptomatic dysfer- pogenesis and purified AnxA2 in regenerating muscle is capable linopathic muscle28,29,35. Here, we provide evidence that disease of increasing FAP accumulation and adipogenesis in regenerating onset and progression in dysferlinopathy is not driven solely by dysferlinopathic muscle (Figs. 6 and 7). These in vitro and in vivo the myofiber and inflammatory cell-specific defects, but by analyses independently confirm an active role of AnxA2 in the creation of an extracellular niche resulting in proliferation and adipogenic conversion of dysferlinopathic muscle. adipogenic differentiation of muscle-resident FAPs. In addition to the direct action of AnxA2 on FAPs, it may also From early in the disease, the primary myofiber defects lead act indirectly via inflammatory or other cell types. In support to persistent myofiber damage. AnxA2, a dysferlin interacting of such a role of AnxA2 we find that dysferlinopathic muscle protein that accumulates at the injured membrane and aids in lacking AnxA2 shows reduction in the extent of macrophage its repair is released at the site of plasma membrane injury47,48. infiltration (Fig. 5). This could be due to the ability of AnxA2 to This AnxA2 released at the site of injured myofibers can trigger function as an agonist for innate immune response38. Previously, muscle inflammation, which in acute injury facilitates myofiber we showed that AnxA2 knockout in B6A/J mice down-regulates regeneration49,50. However, in LGMD2B patients AnxA2 levels TLR4 signaling33. Inhibiting TLR signaling by deletion of the are chronically increased in a manner that correlates with disease central TLR adapter protein Myd88 reduces pathology in dys- severity31,32. Given that AnxA2 expression is ubiquitous, AnxA2 ferlinopathic (A/J) mice46. Innate immune signaling has been can also be produced by myofibers or other cells present in the implicated in the dysferlinopathic symptoms44, and dysferlin- injured muscle, including the FAPs, endothelial and inflamma- deficient muscles show a greater abundance of pro-inflammatory tory cells. AnxA2 released chronically in the extracellular matrix as compared to pro-regenerative macrophages51. Together, these of dysferlinopathic muscle creates a niche which favors increased may contribute to delayed/impaired myogenesis52, further proliferation and subsequently, adipogenic differentiation of enabling FAP accumulation14. Repeat rounds of myofiber injury, FAPs (Fig. 8). Consistent with this hypothesis, we show that chronic inflammation, and FAP accumulation as the muscle ages AnxA2 is a critical component of the pro-adipogenic FAP niche may set up a feed-forward loop linking myofiber damage to the in dysferlinopathic muscle, as deletion of AnxA2 both decreases formation of a pro-adipogenic niche over time, which in turn extracellular matrix FAP accumulation and prevents their contributes to myofiber damage (Fig. 8). In such a system, 10 NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z ARTICLE adipogenic accumulation becomes the nucleating event that significant advance towards the development of the therapy for results in clinical onset, and an abrupt decline in muscle function this disease, which currently lacks any effective or approved in patients. Thus, while dysferlinopathy is driven by a myofiber- therapy. Unlike other inflammatory muscle diseases, where sup- specific deficit, it is the impaired cellular interactions between pression of inflammation with corticosteroids is effective, treat- myofibers, inflammatory cells, and FAPs that is causative for ment of dysferlinopathic patients with glucocorticoids has been disease initiation and severity. This view of the disease opens up without success60, which may be due to the role of conventional previously unrecognized avenues to intervene, as has been rea- corticosteroids in inducing myofiber damage and activating FAP lized in DMD where inhibition of TGFβ-induced FAP accumu- adipogenesis61,62. Further, the use of tyrosine kinase inhibitors lation reduces muscle fibrosis and leads to therapeutic benefits19. prevents excessive FAP proliferation in DMD mouse model19,20, Use of mdx mice shows that aberrant FAP accumulation and has toxicity associated with their long-term use required for differentiation drives disease pathogenesis in DMD19. While these LGMD2B. Our identification of inhibiting adipogenesis in dys- FAPs primarily exhibit fibrogenic fate via TGFβ signaling, in vitro ferinopathic muscle by targeting FAPs by MMP-14 inhibitors they also exhibit adipogenic potential19,53. Their adipogenic (batimastat) also opens avenues for the use of other candidate potential diminishes with age, advancing pathology, and by the drugs like promethazine, which also inhibits FAP adipogenesis63. use of HDAC inhibitors53,54. This contrasts with the in vivo In principle, the direct manipulation of PDGFRα splicing by increase in adipogenic fate of FAPs we observe in dysferlinopathic morpholinos may also be beneficial by preventing FAP pro- mice. This difference between the dystrophin (mdx) and dysferlin liferation64. Irrespective of the precise therapeutic approach that (B6A/J) mouse models suggests spontaneous fibrogenic com- would be efficacious, our study identifies the accumulation mitment of the mdx and adipogenic commitment of B6A/J FAPs. and adipogenic differentiation of FAPs as a central target to However, the mechanism by which FAPs can choose between prevent the precipitation of cellular deficits into the abrupt these fates remains elusive. Inhibition of MMPs is known to onset of disease in dysferlinopathies. Moreover, such therapies decrease the conversion of 3T3-L1 and primary rat preadipocytes would be complementary to the ongoing efforts to restore dys- into adipocytes39. Analyses of the FAPs show that pharmacolo- ferlin expression in terminally differentiated myofibers. gical inhibition of matrix metalloprotease (MMP)-14 represses C/EBPδ and PPARγ in FAPs by way of cilial hedgehog signaling Methods and this reduces the adipogenic fate of FAPs40. Adipogenesis Patient biopsies. Patient biopsies were obtained under informed consent and was of dysferlinopathic muscle has been linked with an increase in approved by the Ethics Committee of Hospital de la Santa Creu i Sant Pau de Barcelona. Frozen muscle biopsies from LGMD2B patients with 2 confirmed C/EBPδ and PPARγ mRNA in the muscle28. Both of these are mutations in dysferlin were used. These were classified mild, moderate, and severe essential transcription factors in adipocyte differentiation, and based on their clinical phenotype (Supplementary Table 1) for analysis. As a MMP-14 is suggested to be one of the extracellular signals that control, frozen muscle biopsies were obtained from young adults with no known triggers adipogenic differentiation of FAPs. MMP-14 is released neuromuscular conditions and without any histopathological features to serve as a by myofibers and is critical to successful myogenesis during comparison. muscle regeneration55. Interestingly, MMP-14 expression in the dysferlin-null SJL mouse quadriceps increases by 3-fold between Animals. All animal procedures were conducted in accordance with guidelines for the care and use of laboratory animals, and were approved by the Children’s 2-month old (presymptomatic) and 9-month old (symptomatic) National Medical Center Institutional Animal Care and Use Committee. C57BL/6J muscles56. Similar analysis of SJL mice has identified that the (WT) and B6.A-Dysfprmd/GeneJ (B6A/J) mice were obtained from the Jackson levels of AnxA1 and AnxA2 increase as these mice transition Laboratory (Bar Harbor, ME) and maintained as homozygous colonies for the from 2 to 8 months of age57. Consistent with the putative role for purpose of this study. A2-B6A/J mice were generated as part of our previous study33, and are maintained in-house. All animals were maintained in an indivi- MMP-14 in adipogenic conversion of dysferlinopathic muscle, we dually vented cage system under a controlled 12 h light/dark cycle with free access show that pharmacological inhibition of MMP-14 (by batimastat) to food and water. Mice were used at the timepoints indicated in the study ± reduces FAP adipogenesis in vitro and ameliorates injury- 2 weeks. triggered lipid formation in vivo (Fig. 7). This suggests that gradual loss of FAP ciliation and/or repression of the Hh pathway Immunohistochemical analysis of muscle sections. Frozen sections 8 µm thick may contribute to pro-adipogenic niche formation in dysferli- were cut from human biopsies and the midbelly of mouse muscles. Lipid was visualized using an Oil Red O staining kit (American MasterTech, #KTORO) nopathic muscle. Batimastat treatment has also been shown to according to manufacturer’s instructions. Immunofluorescence was performed by reduce fibrosis and increase muscle function in mdx mice58, fixing sections in chilled 10% neutral buffered formalin, blocking with 5% BSA which is suggestive of an anti-fibrotic effect on FAPs in and incubation with primary antibodies against perilipin-1 (1:250, Sigma, #P1873), dystrophin-deficient muscle. Because batimastat broadly inhibits PDGFRα (1:250, Cell Signaling, #3174S), Annexin A2 (1:250, Santa Cruz, #SC-9061) and F4/80 (1:500, Serotec, #MCA497). For co-labeling anti-PDGFRα MMPs, it is not clear whether this is due to specific inhibition of (1:100, R&D Systems, #AF1062) was used. Staining was visualized using relevant MMP-14 or other MMPs upregulated in mdx muscle58. But, it is secondary antibodies conjugated to AlexaFluor 488 and/or 568 (1:500, Thermo- clear that more insight into the mechanisms by which FAPs Fisher). Myofiber membranes were marked using AlexaFluor 488-conjugated choose between fibrosis and adipogenesis and therapies targeting wheat germ agglutinin (1:500, ThermoFisher, #W11261) and coverslips were these pathways are of great interest in treating muscular dystro- mounted using ProLong Gold with DAPI (ThermoFisher, #P36941). phies and other degenerative muscle diseases. AnxA2-mediated MMP secretion has been shown to cause Microscopy and image analysis. Microscopy was performed using an Olympus BX61 VS120-S5 Virtual Slide Scanning System with UPlanSApo 40×/0.95 objective, joint destruction in rheumatoid arthritis59, suggesting AnxA2/ Olympus XM10 monochrome camera or Allied vision Pike F-505C color camera, MMP interactions may play a role in FAP-dependent adipogen- and Olympus VS-ASW FL 2.7 imaging software. Confocal images were acquired esis in dysferlinopathic muscle. AnxA2 has also been shown to using an Olympus FV1000 Confocal Microscope with UPlanFLN 40×/1.30 oil influence Hh signaling via the AnxA2 receptor in endothelial objective and Olympus FV-ASW version 4.2 imaging software. Perilipin-1 quantification was performed by thresholding the image to exclude non-specific cells58, providing additional mechanisms by which AnxA2 and staining and then calculating the area of each lipid deposit encircled by perilipin-1 MMP-14 may be linked during adipogenic niche formation in (excluding those in the epi- or perimysium) using MetaMorph software (Molecular dysferlinopathic muscle. Such a role of AnxA2 is consistent with Devices). The total area encircled by perilipin-1 was calculated for all lipid deposits our previous study showing that loss of AnxA2 uncouples the across the muscle and expressed relative to the total cross-sectional area. PDGFRα and AnxA2 positive area was calculated using CellSens software (Olympus) by repair defect of dysferlinopathic myofibers from the eventual thresholding to remove non-specific staining and calculating the total positive adipogenic replacement of the muscle33, identifying AnxA2 and area (again, excluding any epi- or perimysial staining) relative to the entire MMP-14 as therapeutic targets for LGMD2B. This represents a muscle cross-section. NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z Isolation and in vitro adipogenesis of FAPs. FAPs were isolated from the hin- is listed in the legend for each figure. To visualize the distribution of data, we dlimb muscles of 6Mo B6A/J mice in a modified protocol from previously pub- plotted the individual data points for each plot in Supplementary Figs. 7 and 8. lished studies8,10. Mice were euthanized and the tibialis anterior, extensor digitorum longus, gastrocnemius, soleus, quadriceps, and psoas were immediately dissected. Non-muscle tissue including tendon, nerve, and overlying fascia were Reporting summary. Further information on research design is available in carefully removed, and muscles were minced finely in a sterile dish and incubated the Nature Research Reporting Summary linked to this article. in Collagenase II (2.5 U/ml, ThermoFisher, #17101015) in PBS for 30 min at 37 °C. The resulting slurry was washed with sterile PBS before further digest in Col- Data availability lagenase D (1.5 U/ml, Sigma Aldridge, #COLLD-RO) and Dispase II (2.4 U/ml, The authors declare that all data supporting the findings of this study are available within Sigma Aldridge, #D4693) in PBS for 60 min at 37 °C. Resulting slurries were passed the paper and its Supplementary Information. through 100 and 40 µm strainers and primary cells were resuspended in 1 ml PBS with 2% FBS and 2 nM EDTA. 300 µl of the primary cell suspension was reserved for plating and the remaining was stained with anti-PDGFRα-APC (1.0 µg per 106 Received: 19 December 2017 Accepted: 10 May 2019 cells in 100 µl, Biolegend, #135908) and isotype control (1.0 µg per 106 cells in 100 Published online: 03 June 2019 µl, Biolegend, #400512) for FACS. Cells stained with isotype control were used as a control for gating positive and negative events. Cells stained with anti-PDGFRα- APC were then sorted on an Influx cell sorter (Becton Dickinson, #646500) using two-way sorting to separate positive and negative PDGFRα-expressing cells for further analysis (Supplementary Fig. 6). The freshly isolated primary cell, PDGFRα− and PDGFRα+ populations were References 1. Lepper, C., Partridge, T. A. & Fan, C.-M. An absolute requirement for plated in Matrigel-coated Nunc Lab-Tek chamber slides (ThermoFisher, #154534) at a density of 20,000 cells/well. Cells were cultured in DMEM (Lonza, #12-604F) Pax7-positive satellite cells in acute injury-induced skeletal muscle supplemented with 20% fetal bovine serum, 1% penicillin, and 2.5 ng/ml bFGF regeneration. Development 138, 3639–3646 (2011). (BioLegend, #579604) for 3 days. Adipogenic differentiation was induced by 2. Murphy, M. M., Lawson, J. A., Mathew, S. J., Hutcheson, D. A. & Kardon, G. exposure to DMEM with 10% FBS, 0.5 mM IBMX (Sigma Aldridge, #I5879), 0.25 Satellite cells, connective tissue fibroblasts and their interactions are crucial µM dexamethasone (Sigma Aldridge, #D2915), and 10 µg/ml insulin (Sigma for muscle regeneration. Development 138, 3625–3637 (2011). Aldridge, #I0516) for 3 days. Following this, cells were cultured in adipogenic 3. Sambasivan, R. et al. Pax7-expressing satellite cells are indispensable for adult maintenance media (DMEM with 10% FBS and 10 µg/ml insulin) for 3 days. skeletal muscle regeneration. Development 138, 3647–3656 (2011). Uninduced cells were not exposed to the adipogenic differentiation media, but 4. Farup, J., Madaro, L., Puri, P. & Mikkelsen, U. R. Interactions between muscle instead cultured for 6 days in the adipogenic maintenance media, with the media stem cells, mesenchymal-derived cells and immune cells in muscle changed after 3 days. At the beginning of day 10, cells were fixed for 30 min in homeostasis, regeneration and disease. Cell Death Dis. 6, e1830 (2015). chilled, neutral buffered formalin before Oil Red O staining to visualize lipid. For 5. Yin, H., Price, F. & Rudnicki, M. A. Satellite cells and the muscle stem cell studies involving AnxA2 treatment, FAPs were isolated and plated as described niche. Physiol. Rev. 93, 23–67 (2013). above. From day 3 onwards, cells were continuously treated with 100 nM 6. Otto, A. et al. Canonical Wnt signalling induces satellite-cell proliferation recombinant AnxA2 (RayBiotech, #230-30023) until being fixed for oil red staining during adult skeletal muscle regeneration. J. Cell Sci. 121, 2939–2950 (2008). on day 10. 7. Serrano, A. L., Baeza-Raja, B., Perdiguero, E., Jardí, M. & Muñoz-Cánoves, P. Interleukin-6 is an essential regulator of satellite cell-mediated skeletal muscle hypertrophy. Cell Metab. 7, 33–44 (2008). Notexin injury. Single injury was performed by carefully shaving the anterior 8. Joe, A. W. et al. Muscle injury activates resident fibro/adipogenic progenitors hindlimb before intramuscular injection of 40 µl notexin (5 μg/ml, Latoxan, that facilitate myogenesis. Nat. Cell Biol. 12, 153–163 (2010). #L8104) into the tibialis anterior using a 0.3 ml ultrafine insulin syringe (BD 9. Uezumi, A. et al. Identification and characterization of PDGFRα+ Biosciences, #324906). Immediately prior to injection, the needle was dipped in mesenchymal progenitors in human skeletal muscle. Cell Death Dis. 5, e1186 green tattoo dye (Harvard Apparatus, #72-9384) to mark the needle track. The (2014). contralateral leg was left uninjured as a control. Mice were allowed to recover for 10. Uezumi, A., Fukada, S., Yamamoto, N., Takeda, S. & Tsuchida, K. 28 days before the animal was euthanized and muscles collected for analysis. For Mesenchymal progenitors distinct from satellite cells contribute to ectopic repeat injuries we performed 3 separate intramuscular notexin injections, each fat cell formation in skeletal muscle. Nat. Cell Biol. 12, 143–152 (2010). 14 days apart, and allowed 28 days of recovery following the final injury before 11. Orimo, S., Hiyamuta, E., Arahata, K. & Sugita, H. Analysis of inflammatory tissue collection. Using the superficial mark on the skin from the tattoo dye, we attempted to perform each injury as close to the site of the previous injury as cells and complement C3 in bupivacaine‐induced myonecrosis. Muscle Nerve possible so as to repeatedly injure the same myofibers each time. For AnxA2- 14, 515–520 (1991). notexin studies, 10 µg recombinant AnxA2 (RayBiotech, #230-30023) was added to 12. McLennan, I. S. Resident macrophages (ED2- and ED3-positive) do not 40 µl notexin (5 μg/ml, Latoxan, #L8104) and injected into the mid-belly of the phagocytose degenerating rat skeletal muscle fibres. Cell Tissue Res. 272, right tibialis anterior. The contralateral (left) tibialis anterior was injected with 40 193–196 (1993). µl notexin only for comparison. Again, these muscles were harvested for analysis 13. St Pierre, B. & Tidball, J. G. Differential response of macrophage 28 days after injury. subpopulations to soleus muscle reloading after rat hindlimb suspension. J. Appl. Physiol. 77, 290–297 (1994). 14. Heredia, J. E. et al. Type 2 innate signals stimulate fibro/adipogenic Batimastat treatment. For in vitro studies, PDGFRα+ FAPs were isolated from progenitors to facilitate muscle regeneration. Cell 153, 376–388 (2013). 12Mo B6A/J muscle as described above, and plated in Matrigel-coated Nunc Lab- 15. Mounier, R. et al. AMPKα1 regulates macrophage skewing at the time of Tek chamber slides at 40,000 cells/well. Cells were cultured in DMEM (Lonza, #12- resolution of inflammation during skeletal muscle regeneration. Cell Metab. 604F) supplemented with 20% fetal bovine serum, 1% penicillin, and 2.5 ng/ml 18, 251–264 (2013). bFGF (BioLegend, #579604) for 3 days. After which, cells were treated with 10 µM 16. Varga, T. et al. Highly dynamic transcriptional signature of distinct batimastat (Sigma-Aldridge, #SML0041) added to the adipogenic maintenance macrophage subsets during sterile inflammation, resolution, and tissue repair. media for 6 days. At the beginning of day 10, cells were fixed for 30 min in chilled, J. Immunol. 196, 4771–4782 (2016). neutral buffered formalin before Oil Red O staining to visualize lipid. 17. Arnold, L. et al. Inflammatory monocytes recruited after skeletal muscle For in vivo studies, 12Mo B6A/J mice were treated thrice weekly with injury switch into antiinflammatory macrophages to support myogenesis. batimastat (Sigma-Aldridge, #SML0041), 2 mg/kg i.p. for 10 weeks. At the end of J. Exp. Med. 204, 1057–1069 (2007). the treatment period, the effect of batimastat on disease pathology was evaluated by 18. Warren, G. L. et al. Chemokine receptor CCR2 involvement in skeletal muscle quantification of perilipin-1 marked lipid area in the gastrocnemius and compared regeneration. FASEB J. 19, 413–415 (2005). to untreated controls. In addition, we tested injury-triggered adipogenesis using the 19. Lemos, D. R. et al. Nilotinib reduces muscle fibrosis in chronic muscle injury repeat notexin injury protocol described above and in Supplementary Fig. 5. by promoting TNF-mediated apoptosis of fibro/adipogenic progenitors. During the 10-week treatment period, 12Mo B6A/J mice were subjected to 3 Nat. Med. 21, 786–794 (2015). intramuscular notexin injections into the TA, each 2 weeks apart, beginning on day 20. Shirasawa, H. et al. Inhibition of PDGFR signaling prevents muscular fatty 2 and ending on day 29. After 10 weeks of treatment, injury-induced lipid infiltration after rotator cuff tear in mice. Sci. Rep. 7, 41552 (2017). formation was quantified by perilipin-1 area in the injured TA and compared between batimastat treated and untreated controls. For both experiments, 21. Bansal, D. et al. Defective membrane repair in dysferlin-deficient muscular batimastat was first dissolved in DMSO, before being reconstituted in sterile 5% dystrophy. Nature 423, 168–172 (2003). saline for treatment. 22. Laval, S. & Bushby, K. Limb‐girdle muscular dystrophies—from genetics to molecular pathology. Neuropathol. Appl. Neurobiol. 30, 91–105 (2004). 23. Liu, J. et al. Dysferlin, a novel skeletal muscle gene, is mutated in Statistical analysis. Data were analyzed using Prism GraphPad software. The Miyoshi myopathy and limb girdle muscular dystrophy. Nat. Genet. 20, 31–36 precise statistical test employed varied depending on the nature of the analysis, and (1998). 12 NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10438-z ARTICLE 24. Lo, H. P. et al. Limb-girdle muscular dystrophy: diagnostic evaluation, 54. Bettica, P. et al. Histological effects of givinostat in boys with Duchenne frequency and clues to pathogenesis. Neuromuscul. Disord. 18, 34–44 (2008). muscular dystrophy. Neuromuscul. Disord. 26, 643–649 (2016). 25. Nguyen, K. et al. Phenotypic study in 40 patients with dysferlin gene mutations: 55. Snyman, C. & Niesler, C. MMP-14 in skeletal muscle repair. J. Muscle Res. Cell high frequency of atypical phenotypes. Arch. Neurol. 64, 1176–1182 (2007). Motil. 36, 215–225 (2015). 26. Nagaraju, K. et al. Dysferlin deficiency enhances monocyte phagocytosis: a 56. Suzuki, N. et al. Expression profiling with progression of dystrophic change in model for the inflammatory onset of limb-girdle muscular dystrophy 2B. Am. dysferlin-deficient mice (SJL). Neurosci. Res. 52, 47–60 (2005). J. Pathol. 172, 774–785 (2008). 57. Lennon, N. J. et al. Dysferlin interacts with annexins A1 and A2 and mediates 27. Kerr, J. P. et al. Dysferlin stabilizes stress-induced Ca2+ signaling in the transverse sarcolemmal wound-healing. J. Biol. Chem. 278, 50466–50473 (2003). tubule membrane. Proc. Natl. Acad. Sci. USA 110, 20831–20836 (2013). 58. Kumar, A., Bhatnagar, S. & Kumar, A. Matrix metalloproteinase inhibitor 28. Grounds, M. D. et al. Lipid accumulation in dysferlin-deficient muscles. batimastat alleviates pathology and improves skeletal muscle function in Am. J. Pathol. 184, 1668–1676 (2014). dystrophin-deficient mdx mice. Am. J. Pathol. 177, 248–260 (2010). 29. Terrill, J. R. et al. Oxidative stress and pathology in muscular dystrophies: 59. Zhao, W. et al. The discoidin domain receptor 2/annexin A2/matrix focus on protein thiol oxidation and dysferlinopathies. FEBS J. 280, metalloproteinase 13 loop promotes joint destruction in arthritis through 4149–4164 (2013). promoting migration and invasion of fibroblast‐like synoviocytes. Arthritis 30. Angelini, C., Peterle, E., Gaiani, A., Bortolussi, L. & Borsato, C. Rheumatol. 66, 2355–2367 (2014). Dysferlinopathy course and sportive activity: clues for possible treatment. Acta 60. Walter, M. C. et al. Treatment of dysferlinopathy with deflazacort: a double- Myol. 30, 127 (2011). blind, placebo-controlled clinical trial. Orphanet J. Rare Dis. 8, 26 (2013). 31. Cagliani, R. et al. Mutation finding in patients with dysferlin deficiency and 61. Dong, Y., Silva, K. A. S., Dong, Y. & Zhang, L. Glucocorticoids increase role of the dysferlin interacting proteins annexin A1 and A2 in muscular adipocytes in muscle by affecting IL-4 regulated FAP activity. FASEB J. 28, dystrophies. Hum. Mutat. 26, 283 (2005). 4123–4132 (2014). 32. Kesari, A. et al. Dysferlin deficiency shows compensatory induction of 62. Sreetama, S. C. et al. Membrane stabilization by modified steroid offers a Rab27A/Slp2a that may contribute to inflammatory onset. Am. J. Pathol. 173, potential therapy for muscular dystrophy due to dysferlin deficit. Mol. Ther. 1476–1487 (2008). https://doi.org/10.1016/j.ymthe.2018.07.021 (2018). 33. Defour, A. et al. Annexin A2 links poor myofiber repair with inflammation 63. Kasai, T. et al. Promethazine hydrochloride inhibits ectopic fat cell formation and adipogenic replacement of the injured muscle. Hum. Mol. Genet. 26, in skeletal muscle. Am. J. Pathol. 187, 2627–2634 (2017). 1979–1991 (2017). 64. Mueller, A. A., Van Velthoven, C. T., Fukumoto, K. D., Cheung, T. H. & 34. Jin, S. et al. Heterogeneous characteristics of MRI changes of thigh muscles in Rando, T. A. Intronic polyadenylation of PDGFRα in resident stem cells patients with dysferlinopathy. Muscle Nerve 54, 1072–1079 (2016). attenuates muscle fibrosis. Nature 540, 276–279 (2016). 35. Nagy, N. et al. Hip region muscular dystrophy and emergence of motor deficits in dysferlin-deficient Bla/J mice. Physiol. Rep. 5, e13173 (2017). 36. Greenberg, A. S. et al. Perilipin, a major hormonally regulated adipocyte- Acknowledgements specific phosphoprotein associated with the periphery of lipid storage droplets. This work is supported by a MDA Career Development Award (MDA477331) to J. Biol. Chem. 266, 11341–11346 (1991). M.W.H. Additional financial support was provided by NIAMS (R01AR055686) and 37. Zhao, P. et al. In vivo filtering of in vitro expression data reveals MyoD targets. MDA (MDA277389) to J.K.J. and NIH (K26OD011171; R24HD050846, P50AR060836) C. R. Biol. 326, 1049–1065 (2003). grants to K.N. The authors acknowledge Carsten Bönnemann’s contribution to this 38. Swisher, J. F., Burton, N., Bacot, S. M., Vogel, S. N. & Feldman, G. M. Annexin study. Microscopy imaging was performed at the CRI Cellular Imaging Core, which is A2 tetramer activates human and murine macrophages through TLR4. Blood supported by funds from CRI and NICHD (U54HD090257). 115, 549–558 (2010). 39. Chavey, C. et al. Matrix metalloproteinases are differentially expressed in Author contributions adipose tissue during obesity and modulate adipocyte differentiation. J. Biol. This study was conceptualized by M.W.H. and J.K.J., with contributions from K.N. and Chem. 278, 11888–11896 (2003). T.A.P. Experiments were designed by M.W.H. and J.K.J. and performed by M.W.H., with 40. Kopinke, D., Roberson, E. C. & Reiter, J. F. Ciliary Hedgehog signaling contributions from A.D. and C.L. Patient biopsies and clinical histories were collected restricts injury-induced adipogenesis. Cell 170, 340–351 (2017). and provided by E.G. and J.D.M. Data analysis was performed by M.W.H., with assis- 41. Ampong, B. N., Imamura, M., Matsumiya, T., Yoshida, M. & Takeda, S. tance from J.K.J. Funding for this project was obtained by M.W.H., J.K.J., and K.N. The Intracellular localization of dysferlin and its association with the initial manuscript was written by M.W.H. and J.K.J., and reviewed by T.A.P., K.N., C.L., dihydropyridine receptor. Acta Myol. 24, 134–144 (2005). and E.G. who all made contributions to the final manuscript text. 42. Klinge, L. et al. Dysferlin associates with the developing T-tubule system in rodent and human skeletal muscle. Muscle Nerve 41, 166–173 (2010). 43. Gallardo, E. et al. Inflammation in dysferlin myopathy: immunohistochemical Additional information characterization of 13 patients. Neurology 57, 2136–2138 (2001). Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- 44. Rawat, R. et al. Inflammasome up-regulation and activation in dysferlin- 019-10438-z. deficient skeletal muscle. Am. J. Pathol. 176, 2891–2900 (2010). 45. Millay, D. P. et al. Genetic manipulation of dysferlin expression in skeletal Competing interests: The authors declare no competing interests. muscle: novel insights into muscular dystrophy. Am. J. Pathol. 175, 1817–1823 (2009). Reprints and permission information is available online at http://npg.nature.com/ 46. Uaesoontrachoon, K. et al. The effects of MyD88 deficiency on disease reprintsandpermissions/ phenotype in dysferlin‐deficient A/J mice: role of endogenous TLR ligands. J. Pathol. 231, 199–209 (2013). Peer review information: Nature Communications thanks the anonymous reviewers for 47. Chen, J. et al. The PEPR GeneChip data warehouse, and implementation their contribution to the peer review of this work. Peer reviewer reports are available. of a dynamic time series query tool (SGQT) with graphical interface. Nucleic Acids Res. 32, D578–D581 (2004). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in 48. Jaiswal, J. K. et al. S100A11 is required for efficient plasma membrane repair published maps and institutional affiliations. and survival of invasive cancer cells. Nat. Commun. 5, 3795 (2014). 49. Chazaud, B. Inflammation during skeletal muscle regeneration and tissue Open Access This article is licensed under a Creative Commons remodeling: application to exercise-induced muscle damage management. Attribution 4.0 International License, which permits use, sharing, Immunol. Cell Biol. 94, 140–145 (2016). adaptation, distribution and reproduction in any medium or format, as long as you give 50. Tidball, J. G. Regulation of muscle growth and regeneration by the immune appropriate credit to the original author(s) and the source, provide a link to the Creative system. Nat. Rev. Immunol. 17, 165–178 (2017). Commons license, and indicate if changes were made. The images or other third party 51. Cohen, T. V. et al. Upregulated IL-1β in dysferlin-deficient muscle attenuates material in this article are included in the article’s Creative Commons license, unless regeneration by blunting the response to pro-inflammatory macrophages. indicated otherwise in a credit line to the material. If material is not included in the Skelet. Muscle 5, 24 (2015). article’s Creative Commons license and your intended use is not permitted by statutory 52. Cohen, T. V., Cohen, J. E. & Partridge, T. A. Myogenesis in dysferlin-deficient regulation or exceeds the permitted use, you will need to obtain permission directly from myoblasts is inhibited by an intrinsic inflammatory response. Neuromuscul. the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Disord. 22, 648–658 (2012). licenses/by/4.0/. 53. Mozzetta, C. et al. Fibroadipogenic progenitors mediate the ability of HDAC inhibitors to promote regeneration in dystrophic muscles of young, but not old Mdx mice. EMBO Mol. Med. 5, 626–639 (2013). © The Author(s) 2019 NATURE COMMUNICATIONS | _(2019) 10:2430_ | https://doi.org/10.1038/s41467-019-10438-z | www.nature.com/naturecommunications 13