Frizzled related protein deficiency impairs muscle strength, gait and calpain 3 levels

Background Limb-girdle muscular dystrophy recessive 1 calpain3-related (LGMDR1), previously known as LGMD2A, is a disease caused by mutations in the CAPN3 gene. It is characterized by progressive weakness and muscle degeneration. Frizzled related protein (FRZB), upregulated in LGMDR1, was identified as a key regulator of the crosstalk between Wnt and integrin signalling pathways. FRZB gene silencing showed a recovery in the expression of some of the costamere protein levels in myotubes. Results Here, we performed a comprehensive characterization of Frzb−/− mice muscles to study the absence of Frzb in skeletal muscle and eventual links with the molecular characteristics of LGMDR1 patient muscles. Frzb−/− mice showed reduced muscle size and strength. Gait analysis showed that Frzb−/− mice moved more slowly but no impaired regeneration capacity was observed after muscle injury. Additionally, Frzb−/− mice muscle showed an increased number of mesoangioblasts. Lack of Frzb gene in Frzb−/− mice and its increased expression in LGMDR1 patients, showed contrary regulation of Rora, Slc16a1, Tfrc and Capn3 genes. The reciprocal regulation of Frzb and Capn3 genes further supports this axis as a potential target for LGMDR1 patients. Conclusions Our data confirm a role for Frzb in the regulation of Rora, Slc16a1, Tfrc, and Capn3 genes in muscle cells. In vivo, reduced muscle strength and gait in the Frzb−/− mice are intriguing features. The reciprocal relationship between FRZB and CAPN3 further supports a key role for this axis in patients with LGMDR1.

Muscular dystrophies are a heterogeneous group of genetic disorders characterized by progressive weakness and muscle degeneration. Among these, Limb-girdle muscular dystrophy recessive 1 calpain3-related (LGMDR1), previously known as LGMD2A, is a disease caused by mutations in the CAPN3 gene [1]. LGMDR1 was considered the most frequent type of LGMD worldwide [2][3][4][5] up till now, although new data suggest that this may not apply to some regions of Latin America [6]. In LGMDR1 patients, disease symptoms caused by muscle wasting and gradual degeneration of the proximal muscle groups, usually first present during the second decade of life and progressively worsen with patients becoming wheelchair-dependent after less than 25 years of evolution of the disease [7,8]. The pathophysiological mechanisms underlying the process of muscle degeneration in the absence of functional CAPN3 protein are still largely unknown.
Loss of CAPN3 leads to abnormal sarcomere formation [9] as well as changes in the expression of several genes in the muscles of LGMDR1 patients [10]. Sarcomere assembly and stabilisation are dependent on a protein complex called the costamere [11][12][13]. Its function is to enable the adhesion between the sarcomere in the muscle and the extracellular matrix [14]. This linkage is partially mediated by integrins [15]. In LGMDR1 myotubes, the physiologically required replacement of the integrin β1 isoforms (β1A substitution by β1D) is disturbed and may be the cause of incorrect costamere assembly. Our studies on integrin interacting proteins as well as proteins implicated in costamere regulation, identified frizzled related protein (FRZB) as a key regulator of the crosstalk between integrin and Wnt signalling pathways [16].
FRZB was originally identified as a secreted antagonist of the Wnt signalling pathway, blocking the effects of Wnt-1, Wnt-8, Wnt-5a and Wnt-9a [17][18][19][20]. In muscle tissue of LGMDR1 patients FRZB gene expression is strongly upregulated [10] and FRZB is a negative regulator of myogenesis [21]. Upon silencing of FRZB gene expression in LGMDR1 patients' myotubes, we earlier showed a recovery in the expression of some of the costamere protein levels such as integrin β1D, melusin and anosmin-1 [16].
Frzb-deficient mice were shown to be highly susceptible to the development of osteoarthritis characterized by damage to the articular cartilage, and also have increased cortical bone density [22]. Of note, Frzb −/− mice showed reduced voluntary exercise performance in running wheels [23]. Here, we aimed to further study the effects of absence of Frzb in skeletal muscle of Frzb −/− mice, and eventual links with the molecular characteristics of LGMDR1 patient muscles.
To determine whether the absence of Frzb in mice affects baseline muscle mass and characteristics, we isolated different muscle groups (Gastrocnemius, Tibialis anterior, Quadriceps and Biceps brachii) from male and female wild-type and Frzb −/− mice aged 4 to 5 weeks. Muscle weight for the different muscle groups was significantly lower in Frzb −/− mice compared to wild-type controls. There was a statistically significant interaction between muscle type and genotype on weight [F (1,25) (Fig. 1c). When muscle weight was normalized to body weight, no differences were observed (data not shown) suggesting that the differences may be related to growth retardation.
Earlier analysis in Frzb −/− mice on the Swiss/CD1 background did not reveal differences in type I vs type IIa fibres in the Soleus or in type IIa vs type IIb fibres in the Extensor digitorum longus [23]. Here, we further analysed fibre composition at the gene expression level: different myosin heavy chain isoforms (Myh1, Myh2 and Myh4) were analysed in 10-week-old male and female Frzb −/− and wild-type mice. The distinct genotypes did not show differences in the expression of the respective myosins (Fig. 1d). As expected Myh4 (myosin present in 2B fibre type) followed by Myh1 (myosin present in 2X fibre type) were highly expressed in the Tibialis anterior while Myh2 showed only minimal expression (myosin present in 2A fibre type). In the Soleus expression of Myh1 and Myh2 was balanced while Myh4 was minimally expressed.
To screen for a potential functional impact of these observations, we assessed mouse muscle strength and endurance by the four-limb hanging test. Male Frzb −/− mice were performing inferior to wild-type mice in this set-up, independent of weight [difference between means 895.52 s (95%CI: 10.58-1780.47) in 2-way ANOVA with F(1,12) = 4.861; p = 0.048 for genotype] (Fig. 1e). Taken together, these analyses suggest that Frzb −/− mice have reduced muscle size and strength that is not explained by obvious changes in fibre types.

Gait analysis
We earlier demonstrated that deletion of Frzb reduces voluntary running exercise performance in mice using running wheels [23]. We therefore set-out to analyse the gait of the Frzb −/− mice compared to wild-type controls to evaluate the global mobility of these mice by Catwalk gait analysis [24]. Frzb −/− mice paws spent significantly more time in contact with the glass plate and in the air as shown by stand and swing phase analysis respectively ( Fig. 2a-b). There was a statistically significant interaction between the effects of genotype, sex and front-or hind-paw on stand [F (1,34) (Fig. 2c). Although Frzb −/− mice moved more slowly (increased stand and swing phase), the distance covered by their paws, stride length, did not differ between WT and Frzb −/− mice (Fig. 2d).

Treadmill exercise of mice
To document whether Frzb −/− and wild-type mice react differently to exercise, animals were subjected to treadmill running. All animals in the experiment were able to perform the selected chronic running exercise protocol and no abnormal behaviour or signs of exhaustion were observed. First, mice body weights were measured, differences between WT and Frzb −/− body weight were gradually diminishing over time, and no differences between trained and not-trained groups were seen [2-way ANOVA (F (11,41) = 5.850 (p < 0.001) for interaction age and genotype, F([11,41) = 0.972 (p = 0.97) for interaction age, genotype and running] (Fig. 3a). Fibre crosssectional area of Soleus was measured. Although no overall differences between cross-sectional areas were observed between trained and not-trained groups, Frzb −/− mice fibres were smaller than WT fibres, most notable in the not-trained group [2-way ANOVA (F(1, 11) = 11.13 (p = 0.0066), 0.51 (p = 0.4865), 0.22 (p = 0.64) for genotype, running and interaction respectively, difference between means 2.7*10 − 4 mm 2 (95%CI: 2.05*10 − 5 -5.2*10 − 4 ; p = 0.034) and 1.7*10 − 4 mm 2 (95%CI: − 6.46*10 − 5 -4.1*10 − 4 ; p = 0.17) Bonferroni-corrected for 2 tests in non-trained and trained mice] (Fig. 3c). No exercise-induced damage was observed in any of transverse sections of Soleus or Tibialis anterior (Fig. 3d). Muscle fibre type composition was analysed in both study groups in Tibialis anterior muscles and no differences were observed after exercise or between genotypes (Fig. 3d). Gene expression markers for muscle damage were not changed by exercise or genotype, with Fbx2 as a positive control for the effect of training (difference between the means 44.67, 95%CI: 23.49-65.85) (Fig. 3e).

Muscle regeneration capacity after intramuscular cardiotoxin-induced muscle injury
To test whether muscle regenerative capacities of Frzb −/− mice were different from wild-type mice, acute skeletal muscle injury was triggered by cardiotoxin injection in the Soleus and Tibialis anterior muscles. No differences in pathology characteristics were noted between the two mouse strains. Non-injured skeletal muscle shows polygonal myofibres with peripheral nuclei ( Fig. 4a-b). At day 3 post injection, muscles showed degenerative myofibres and inflammatory cellular infiltration. In the Soleus, most of the fibres were damaged, while in the case of Tibialis anterior, just the cardiotoxin injection area appeared affected ( Fig. 4a-b). One week after injury, small regenerating myotubes with centrally located nuclei were observed with some remnants of inflammation. Regenerating fibers with centrally located nuclei increased their diameter by two weeks and the pattern became homogenous by four weeks (Fig. 4a-b).

In vitro muscle cell differentiation
Taking into account that Frzb affects myogenesis and that myotube differentiation and costamere assembly appear disturbed in LGMDR1 patients and models, we studied muscle cell differentiation in the different genetic backgrounds, using two different cell populations with myogenic potential. First, we isolated satellite cells from four-weeks-old wild-type and Frzb −/− mice using the Biceps brachii, Gastrocnemius, Tibialis anterior and Quadriceps. In proliferation assays, immunofluorescence analysis showed more MyoD positive cells in Frzb −/− mice muscles after enzymatic digestion and increased levels of the proliferation marker protein Ki67, represented as a percentage [P = 0.0046 and P < 0.0001 respectively, t-test] (Fig. 5a). However, after differentiation was induced, at the myotube stage fusion index and numbers of MyoD and myogenin positive nuclei were not different between the wild-type and Frzb −/− mice (Fig. 5b).
In addition, we isolated mesoangioblasts (MABs) from five-week-old wild-type and Frzb −/− mice using the Biceps brachii, Gastrocnemius, Tibialis anterior and Quadriceps as source. Freshly isolated cells were analysed by FACs analysis using alkaline phosphatase (ALP) as a cell surface marker. The obtained ALP+ cell distribution was different between wild-type and Frzb −/− mice. Percentage of ALP+ cells from Frzb −/− mice was on average 39.59 higher than from wild-type animals [95%CI: 26.18-53.00; p < 0.0001), Bonferroni corrected for 2 tests in 2way ANOVA with F(1,22) = 46.73 (p < 0.001) for interaction between genotype and cell type] (Fig. 6a). Further characterization of the ALP+ cells showed that all of them were indeed negative for endothelial cell marker CD31 and hematopoietic cell marker CD45. Nevertheless, while wild-type ALP+ cells were around 81% positive for platelet derived growth factor receptor alpha (CD140A or PDGFRα), Frzb −/− ALP+ cells were only around 43% positive for this marker [p = 0.043 for CD140A, t-test] (Fig. 6b).

Molecular analysis of the muscles
To screen for molecular differences between the muscles of Frzb −/− and wild-type mouse muscles and find eventual links with LGMDR1, we performed gene expression b Immunofluorescence analysis of WT and Frzb −/− myotubes at day 3 of differentiation. Left images stained for nuclear myogenin (red) and cytoplasmatic sarcomeric α-actinin (green). Right images stained for nuclear MyoD (green) and cytoplasmatic MyHC (red). Myotubes fusion index, calculated as the percentage of nuclei inside myotubes on the total amount of nuclei. In all cases nuclei were visualized with Hoechst (blue). Scale bar 250 μm. Error bars indicate mean ± SEM analysis on isolated muscles. First, we focused on muscle-specific genes. On one hand, no differences were found for Pax7. However the expression of Myod was on average 52.73% higher in  (Fig. 7a). Muscle atrophy-related ubiquitin ligases Fbx32 and Murf1 did not show differences between wild-type and Frzb −/− mice (Fig. 7b). The analysis of adipocyte genes Pparg, Adipo and Fasn, showed no differences between wildtype and Frzb −/− mice (Fig. 7c).
Other genes of interest were identified from the analysis of LGMDR1 patient muscles with CAPN3 mutations. Interestingly, Capn3 expression was on average 63 (Fig. 7f). Again, no differences were found in the Tibialis anterior muscles. Of note, Tfrc, Rora and Slc16a1 were not only differentially expressed in muscles from LGMDR1 patients but also in the articular cartilage of Frzb −/− mice [10,26].

FRZB silencing in human muscle cells
The effect of specific genes of interest were translationally validated in human myotubes by silencing the expression of the human FRZB gene. As shown in Fig. 8a (Fig. 8d).

Discussion
The potential role of FRZB in LGMDR1 is intriguing as patients have high levels of FRZB expression in their muscles [10]. In myotubes of LGMDR1 patients, correct costamere assembly and cell fusion appears to be disturbed and this has been linked with the absence of the required integrin isoform replacement from β1A to β1D. Remarkably FRZB silencing in myotubes leads to costamere protein rescue. We therefore suggested that FRZB may be a potential therapeutic target for LGMDR1 patients [16]. To better understand the function of endogenous FRZB in muscles, different aspects of muscle biology in the Frzb −/− mouse model were studied at the functional, cellular and molecular level in an exploratory analysis.
Muscle weakness is the hallmark of muscular dystrophies. Our analysis suggests that Frzb −/− mice have a lower muscle mass that may result in reduced performance in the four limb hanging test. These findings seem to correspond to a stunted growth since no differences in body weight were previously observed in 6-month and 1-year old mice [23]. We have no evidence that weakness would evolve over time as this parameter was only evaluated at one time-point.
We also performed gait analysis of Frzb −/− mice to gain insights into the effect of Frzb deficiency on muscle function. Frzb −/− mice showed a longer step cycle, spending more time with the paw in contact with the walkway (stand time) as well as airborne (swing time). Thus, their limb movement was slower, however with no effect on the covered distance. In a previous report, we showed that Frzb −/− mice ran daily significantly lower distances in a voluntary running wheel setup [23]. This reduced voluntary running exercise performance could be attributed to the lower speed rather than to the fact that they spend less time running. Thus, Frzb −/− mice appear to have detectable issues in speed and mobility compared to wild-type controls. Although interesting from a scientific point of view, these observations cannot be translated into clinically meaningful data that could be applied to dystrophy patients. Anomalies that can be noted in muscular dystrophy patients are linked with weakness of the hip abductor muscles, producing a Trendelenburg gait characterized by "waddling" [27] and differ from what was observed in Frzb −/− mice. In addition, the instability and altered step patterns observed among LGMD patients that have been already characterized in a mouse model [27] are different to what was observed in this study.
To better understand the reasons why gait and muscle strength abnormalities were observed, several other factors were analysed. Jackson and collaborators (2015) showed that depletion of Pax7 expressing satellite cells in muscles resulted in reduced voluntary wheel running performance. Hence, Pax7 expression was analysed. No differences with wildtype animals were shown. Increased adipose tissue infiltration of muscles is a common hallmark in neuromuscular disorders [28][29][30]. Frzb −/− mice have lower body and muscle mass. However, Frzb −/− mice did not show fatty infiltration, centrally located nuclei, fibrosis or other dystrophic features in the analysed muscle sections. Regulators of adipocyte differentiation, peroxisome proliferator activated receptor gamma or adiponectin coded by Pparg and Adipoq respectively were not different between the studied mouse strains. In addition to fat infiltration, atrophy can also cause muscle weakness, by upregulation of atrogenes (MAFbx/Atrogin-1 and MuRF1) that lead to loss of muscle mass [31][32][33]. None of the atrogenes were up-regulated in Frzb −/− muscles. Therefore, we have no evidence that fat or atrophy could be the reason why mice spent more time to complete a step cycle.
Gait and other phenotypes reported here might not only be altered by impairment in muscles; but also by changes in the nervous system or bones. So far, there is no clear evidence of alterations in the nervous system of Frzb −/− mice. Some reports confirm dynamic expression of Frzb in neural cells but we are not aware of specific data on motor neurons, motor cortex or cerebellum that would explain our observations [34][35][36][37][38]. Nevertheless, the transgenic mice are known to have thicker cortical bone, with increased stiffness and higher cortical appositional bone formation after loading of the long bones [22]. These differences with wild-type animals cannot be excluded as factors contributing to the changes observed in Frzb −/− mouse gait and other phenotypes.
Exercise has direct effects on muscles, triggering changes in CSA of the fibres, fibre type distribution, weight change and potentially muscle injury [39]. Although physical training induces beneficial adaptive changes in skeletal muscle of healthy individuals, its effects in patients with muscular dystrophy remain controversial. While some studies attributed a beneficial effect without reporting muscle injury, other studies reported training-induced muscle damage and creatine kinase elevations in high-intensity training programs in patients, or even an earlier onset of symptoms associated with exercise in LGMD2B patients [40][41][42][43][44]. Since distinct muscular dystrophies show different progression of muscle degeneration and strength loss, leading to diverse exercise tolerance, endurance treadmill training tolerance and muscle changes were studied in Frzb −/− model. The lack of structural changes in Frzb −/− mice after exercise suggests the absence of a severe phenotype.
The CSA of muscle fibres in Frzb −/− mice was smaller. Consequently, we studied muscle fibre type composition. Tibialis anterior and Soleus fibre type distribution were within normal range values as has been described for C57Bl/6 mice [45,46]. Previous studies in Soleus and Extensor digitorum longus analysed by immunofluorescence showed, similar fibre composition in WT and Frzb −/− mice [23]. Additionally, myosin gene expression as well as SDH activity were analysed obtaining the same result: fibre composition in Soleus and Tibialis anterior Frzb −/− did not vary from wild-type mice.
Several muscular dystrophy models, such as syntrophin α1 null mice and murine models for LGMDR12 (LGMD2L) and LGMDR1, showed aberrant muscle regeneration with longstanding necrosis and impaired exercise and contractile properties with aberrant neuromuscular junctions [47][48][49]. However, in Frzb −/− mice, after cardiotoxin injection, no aberrant or impaired regeneration capacity or fibrosis was noticed. In summary, Frzb −/− mice muscles showed normal fibre composition and they do not display altered regeneration capacity.
Murine primary cell cultures have been widely studied for myogenesis and muscular dystrophies in vitro analysis such as LGMDR1, LGMD2I or DMD [9,16,[50][51][52][53][54][55]. We isolated satellite cells and we found that Frzb −/− mice cells showed enrichment for MyoD and Ki67 nuclear proteins. FRZB inhibits MyoD expression at RNA and protein level [21,56]. We here show that in the absence of FRZB, MyoD was upregulated in our cells. As myotube formation was not altered, myogenesis may not be strongly impaired, but further studies will be required to identify the consequences of MyoD during myogenesis in the absence of FRZB.
On the other hand, the increased Ki67 expression, which is a proliferation marker, suggested an increased proliferative capacity in Frzb −/− muscles. However, there is controversy about the way in which the presence or absence of FRZB could affect proliferation. Some studies described that FZRB inhibits the growth of mesoangioblasts and suppressed cell proliferation in gastric cancer [57,58], but other authors suggested that Frzb suppression reduced proliferation in alveolar rhabdomyosarcoma [59]. Moreover, tissue dependent differences have been observed in the same model, since Frzb −/− mice chondrocytes proliferated less than those obtained from WT mice, contrary to the observation in satellite cells in the same mice.
Considering that some muscle resident cell population are able to generate muscle, both in vitro and in vivo [60][61][62] pericyte-derived adult MABs were also studied. Pericyte-derived adult MABs are isolated from adult muscles and they retain similar characteristics of embryonic MABs [63][64][65][66]. Detailed analysis of the ALP+ cells showed significantly less PDGFRα expression. So far, two types of pericytes have been described: type-1 and type-2. Both of them differ in their cell surface markers as well as in their differentiation capacity. Type-1 are Nestin−/PDGFRα+ and are characterized by their ability to differentiate into adipocytes while type-2 are Nestin+/ PDGFRα-and do not differentiate into adipocytes but form myotubes in culture [67,68]. The lower PDGFRα expression could indicate that Frzb −/− mice have more type-2 pericytes.
In previous studies, differentially expressed genes have been analysed in the articular cartilage-subchondral bone biomechanical unit of Frzb −/− mice [22,26]. However, gene expression analyses in muscle have not been carried out. Thus, one aim of this study was to establish whether Frzb deficiency impairs muscle gene expression in Frzb −/− mice. The analysis was focused mainly on the Soleus, as Soleus showed the greatest molecular similarities to human skeletal muscles [69] and since together with diaphragm these are the most affected muscles in Capn3 −/− mice [9].
Myogenesis, a process that takes place during growth and regeneration in adult, depends on satellite cell activation of Pax7 cells and it is regulated by muscle-specific transcription factors such as MyoD and Myogenin [70,71]. In the studied samples Pax7 was not upregulated, MyoD was upregulated in Frzb −/− and there was some increase in Myog. The same was observed in the FRZB silenced human samples; MYOD expression was upregulated. In Xenopus, Frzb inhibits axis duplication induced by Xwnt8 and also muscle development by blocking MyoD induction [18,56,72]. In mammals, myogenesis inhibition by Frzb accompanied by reduction in Myf5 and MyoD expression was reported [21]. Although most of the works were carried out during embryonic development, the possibility that Frzb has a role in adult myogenesis or muscle maintenance, modualting MyoD levels, should be considered. Although Myod was upregulated in Frzb −/− mouse muscles, increased myogenesis was not observed (centrally located nuclei were absent and different size fibres were not observed). MyoD increase after siFRZB in LGMDR1 patients could be considered as a beneficial consequence given that muscle degeneration stimulus is occurring and consequently new myofibres formation would be necessary. Nevertheless, further studies will be required to analyse if this increase improves cell physiology.
Among the selected genes deregulated in C3KO mice [25] the Ky gene showed expression changes in Frzb −/− mice. Its protease activity targets different proteins and its absence could disrupt muscle cytoskeleton homeostasis [73]. Natural ky mutant mice has smaller muscles with slower contraction time and are weaker than controls [74,75]. When focusing on deregulated genes in LGMDR1 patients, it is noteworthy to mention that Rora, Slc16a and Tfrc genes showed the same direction of differential expression in cartilage and muscle of Frzb −/− mice. These differences were opposite to what was observed in LGMDR1 patients where FRZB is upregulated.
Tfrc is implicated in muscle biology. On one hand, it participates in iron acquisition in skeletal muscle [76]. On the other hand, Tfrc has been already described as a Wnt target gene [77]. Tfrc is elevated in regenerating fibres in patients with Duchenne muscular dystrophy as well as in facioscapulohumeral muscular dystrophy (FSHD) [78,79]. However, the effects of TFRC upregulation in LGMDR1 patients have so far not been studied.
In skeletal muscle, retinoic acid receptor-related orphan receptor-α (Rorα) has been described as positive regulator of myogenesis by its interaction with MyoD and p300 cofactor which lead to activate muscle-specific genes transcription [80]. Furthermore Rorα is involved in the regulation of glucose and lipid metabolism in skeletal muscle [81,82]. On the other hand, a role in Wnt signalling has been described since Wnt5a/PKCαdependent as well as PGE2/PKCα-dependent Rorα phosphorylation exerts inhibitory function of the expression of Wnt / β-catenin target genes [83,84]. Altogether, rescue of Rora expression by Wnt signalling pathway activation in the absence of FRZB could be beneficial for LGMDR1 patients due to its importance in muscle homeostasis.
Slc16a1, coding for a proton-linked monocarboxylate transporter, is highly expressed in oxidative fibres (type I fibres) consistent with the role of SLC16A1 in mediating lactate uptake for oxidative metabolism [85]. Its deregulation may be responsible for some of the metabolic impairment in LGMDR1 patients. No previous relation between Slc16a1 and the Wnt pathway was reported so far, but its downregulation in mouse muscle and in cartilage as well as its upregulation in LGMDR1 patients, where FRZB is overexpressed, suggests a direct interaction in its regulation.
One of the most striking findings was the upregulation of the Capn3 gene in Frzb −/− mice Soleus and its upregulation after FRZB silencing in human myotubes since no genetic regulatory mechanism of Capn3 expression has been described so far. FRZB upregulation in CAPN3 deficient LGMDR1 patients was already described [10], but the reciprocal regulation has not been reported. However, the increase in Frzb expression has been discarded as a beneficial compensatory mechanism since silencing of the gene increased several proteins that were upregulated in LGMDR1 patients [16]. These findings could be interesting not only for LGMDR1 patients, but for dysferlinopathy and titinopathy patients in whom a secondary reduction of CANP3 has been described [86,87].

Conclusion and limitations
In summary, the result presented here confirm a role for Frzb in the regulation of Rora, Slc16a1, Tfrc, and Capn3 genes, which is of interest in understanding molecular changes observed in LGMDR1 patients. However, the studies and results failed to demonstrate a clinical correlate or clinically meaning effect that can be applied to our understanding of LGMDR1. Our specific in vivo and ex vivo analyses may not have captured all differences between the loss of function and wildtype mice, in particular due to the limitations in experimental design in particular when selecting the age of the mice used in the experiments. Even if not directly clinically revelevant for patients with LGMDR1, the involvement of FRZB in myogenesis was confirmed, since this gene regulates MyoD gene expression and the Frzb −/− mice show reduced muscle strength and gait abnormalities. However, lack of Frzb did not alter skeletal muscle regeneration capacity and neither induced modifications after exercise, with the caveat that our observations can be dependent on the age of the mice studied. Our data thus uncover new roles for FRZB in muscle and support a specific role for FRZB in patients with LGMDR1.

Animals
Frzb knockout (Frzb −/− ) mice were previously generated [22]. C57BL/6 mice were at least in the 19th -20th generation of backcrossing. Wild-type C57Bl/6 mice (WT) were purchased from Janvier (Le Genest St Isle, France). Mice were housed in groups of 4-5 mice in Static micro-insulator cage with Macrolon filter with bedding material, under conventional laboratory conditions (14 h light -10 h dark; 23+/− 2°C), with standard mouse chow food (Sniff, Soest, Germany) and water provided ad libitum. All studies were performed with the approval from the Ethics Committee for Animal Research (P034/ 2016; KU Leuven, Belgium) in accordance with relevant guidelines and regulations. Several groups of mice were used for different experiments, their specific characteristics are available in Table 1.

Human samples
All participants gave informed consent, using forms approved by the Ethics Committee on the Use of Human Subjects in Research at Donostia University Hospital (ASP-FRZ-2017-01) and all the experiments were performed in accordance with relevant guidelines and regulations. Muscle biopsy specimens were obtained from adult patients with LGMDR1 (genetically confirmed) and healthy donors (Supplementary Table S1). Primary human skeletal muscle cell culture and FRZB gene silencing experiments were performed in healthy and LGMDR1 patients' myotubes as previously described [16].

Muscle strength and endurance analysis
The four limb hanging test was used to monitor muscle strength and endurance (Treat-NMD Neuromuscular Network (SOP (ID) Number DMD_M.2.1.005)) [88]. Five to six-week-old mice were placed once on a crosslinked wire grid. The grid was inverted and the 'time to fall' was monitored.

Mouse gait analysis
The CatWalk™ XT system (Noldus, CatWalk 7.1, The Netherlands) was used to assess gait and locomotion [24]. Mice were placed on the runway for three consecutive runs. Runs were analysed separately and an average of these three runs was used as an individual value. The following parameters were evaluated: stand (paw contact time with the glass plate during the step cycle in seconds), swing phase (paw time in the air during the step cycle in seconds), step cycle (the sum of stand and swing time in seconds) and stride length (distance covered by a paw in mm).

Treadmill exercise
Six-week-old mice were subjected to a 5-week chronic exercise protocol on a four-lane modular treadmill (Columbus Instruments). The exercised group ran 30 min on a horizontal treadmill at 12 m/min twice a week for 5 weeks [89,90] after a warm-up exercise consisting in 2 min at 4.2 m/min followed by 8 min at 7.8 m/min. Four WT and 2 Frzb −/− mice were included in a non-training control group. All mice weights were monitored every training day and muscles were dissected.

Cardiotoxin injection
Ten-week-old mice were anesthetized by intraperitoneal injection and 3 μl of 50 μM cardiotoxin (CTX; Latoxan, Portes-lès-Valence, France) was injected into the Tibialis anterior and 3 μl of 16.7 μM of cardiotoxin into the Soleus [91]. Tibialis anterior and Soleus muscles were dissected at three days, one, two and four weeks after cardiotoxin injection.

Histology and immunofluorescence
OCT compound (Tissue-Tek) immersed Tibialis anterior and Soleus were directly frozen into cold 2methylbutane (Thermo Fisher Scientific). Frozen muscles were sectioned and stained with hematoxylin and eosin. For immunohistochemistry muscle cryosections were fixed (4% paraformaldehyde (PFA) (Electron microscopy sciences; Hatfield, PA, USA)) followed by permeation (0. Nuclei were visualized with 10 μg/ml containing Hoechst (Sigma-Aldrich) solution. Fluor Save reagent (Millipore) was used as mounting medium. Muscle structure was analysed using a Nikon 80i microscope and the NIS-Element software. For satellite cell immunofluorescence analysis, the percentage of positive nuclei was counted in randomly selected 6 fields of view. Fusion index was calculated in myotubes as the percentage of nuclei inside myotubes, being MyHC or sarcomeric α-actinin, myotubes markers. Between 5 and 6 field of view were counted for each sample. For fibre type classification SDH enzymatic activity was used [92]. The fibres were assigned to four different groups depending on the intensity (in pixels) value obtained with the ImageJ program (strong< 100, medium-stron 100-150, medium-weak 150-200 and weakly coloured > 200 pixels). All fibres of one section of Tibialis anterior per mouse in to the chronic exercise protocol experiment were measured.

Muscle fibres cross-sectional area
ImageJ software was used to measure fibres crosssectional area (CSA). One hundred fibres from 4 different fields of view were measured.

Mouse primary cells Satellite cells
Murine satellite cells from 4-week-old mice were isolated from Gastrocnemius, Tibialis anterior, Quadriceps and Biceps as previously described [93]. Cells were plated in triplicate and at confluency, the differentiation was induced by switching medium to DMEM high glucose supplemented with 2% horse serum (Gibco-Thermo Fisher Scientific) and 1 mM (100 mg/ml) sodium pyruvate, 100 U/ml penicillin and streptomycin, 2 mM L-glutamine. Cells were incubated at 37°C, 5% CO 2 , 5% O 2 .

Mesoangioblasts
Mesoangioblasts (MABs) were isolated from Gastrocnemius, Tibialis anterior, Quadriceps and Biceps explant culture, by Fluorescence Activated Cell Sorter (FACS; BD FACS ARIA III) for alkaline phosphatase (ALP) + cells (R&D Systems-Biotechne, Minneapolis, MN, USA) from 5 week-old mice as previously described [94]. Flow cytometry analysis of the ALP+ fraction was carried out in 3 WT and 7 Frzb −/− samples. Protein tyrosine phosphatase receptor type C (CD45), platelet and endothelial cell adhesion molecule 1 (CD31) and platelet derived growth factor receptor alpha (PDGFRα or CD140α; Thermo Fisher Scientific) cell surface proteins presence were analysed by flow cytometry (FACs; BD Canto AIG) and analysed by BD FACSDiva software.
The entire list of the custom-designed SYBR green panels can be found as Supplementary Table S2 online. a) muscle specific genes, b) deregulated genes in LGMDR1 patients' muscles [10,95], c) deregulated genes in Frzb −/− mice articular cartilage and LGMDR1 patients' muscles [23,26]

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. All data generated or analysed during this study are included in this published article (and its supplementary information files).
Ethics approval and consent to participate Animal samples: All studies were performed with the approval from the Ethics Committee for Animal Research (P034/2016; KU Leuven, Belgium) in accordance with relevant guidelines and regulations. Human samples: All participants gave informed consent, using forms approved by the Ethics Committee on the Use of Human Subjects in Research at Donostia University Hospital (ASP-FRZ-2017-01) and all the experiments were performed in accordance with relevant guidelines and regulations.
Consent for publication "Not applicable".