FAP adipogenesis drives early spinal muscle loss in dysferlinopathy and reveals a therapeutic target

Fatty replacement of paraspinal muscles is a feature of dysferlinopathy patients

To identify changes in axial muscles in patients with dysferlinopathy, we retrospectively identified a cohort of dysferlinopathic patients who underwent muscle MRI at two different locations (South Korea and France). This resulted in data from 20 dysferlinopathic patients with biallelic mutations in the dysferlin gene confirmed by next-generation sequencing (NGS) (Table 1). All patients suffered from progressive limb-girdle muscle weakness and had significantly elevated serum creatine kinase (CK) levels (Table 1). Based on the clinical features, these patients were diagnosed as either (A) distal myopathy (Miyoshi type), (B) Limb girdle myopathy (LGMD type), or (C) CK elevation with/without exercise intolerance (Table 1). The T1-weighted lumbar spine MRI scans from patients were analyzed for individual spinal muscles (Fig. 1A). This showed that, unlike the images of unaffected controls, dysferlinopathic patients exhibit high T1 signal intensity of erector spinae muscles, indicating fatty replacement of these muscles in patients (Fig. 1B–E). While these images showed a marked degeneration of several spinal muscles in patients, MRI images of the adjacent pelvic girdle muscles that support hip and back flexion (psoas and quadratus lumborum) as well as limb muscles remained unaffected (Figs. 1F and EV1). Detailed analysis of the T1 images showed that the lower back muscles were more affected than the lower dorsal muscles, while the upper dorsal and cranial muscles were the least affected. Degeneration of the spinal muscle manifested as iliocostalis > longissimus > spinalis (Fig. 1F). Use of serum CK as an indicator of disease severity identified that a higher serum CK level was associated with an earlier age of disease onset and age at which spinal MRI needed to be performed (Fig. 1G,H). The extent and severity of spinal muscle degeneration vary among patients and among different spinal muscles. For example, in patient #4, high signal intensity was observed only in the iliocostalis muscle, not in multifidus, longissimus, and iliocostalis, which make up the erector spinae muscles (Fig. 1E). However, irrespective of how widespread fatty degeneration was, the extent of signal in the spinal muscles was greater than in the girdle muscles. Together, these results demonstrate degeneration of the erector spinae muscles, with the iliocostalis muscles being the first or the most severely affected spinal muscle.

Fatty degeneration of spinal muscles precedes degeneration of quadriceps in BLA/J mice

The observations above suggest dysferlinopathy causes early-onset spinal muscle adipogenic degeneration. However, an alternate possibility is that spinal muscle degeneration in dysferlinopathic patients is a secondary manifestation of limb weakness in these patients, leading to greater axial load on the spine muscles to stabilize the vertical biped posture and walking gait of the patient. If so, we reasoned, this feature will be limited to the patients (biped) and not to the mouse (quadruped) model. To examine this, we performed T1-weighted MR imaging of the spinal muscles using WT and BLA/J mouse model of dysferlinopathy (Fig. 2A). At their symptomatic age (10 months old), BLA/J mice showed a significant increase in the fatty degeneration of their limb and the spinal muscles such that the spinal muscles showing 25-fold greater fatty replacement as compared to fivefold greater degeneration of the limb (quadriceps) muscle of BLA/J mice when compared to the WT mice (Fig. 2B,F,G). To monitor the age of onset of this fatty degeneration, we performed MRI analysis of BLA/J mice at 6, 8, 10, and 22 months of age and compared the results with the WT mice (Fig. 2B–E). This showed fatty degeneration first manifests at the spinal muscles by 6 months of age, when there is little to no fatty MRI-detectable degeneration of the quadriceps muscle (Fig. 2E,H). The adipogenic loss of spinal muscle progressed further by 8, 10, and 22 months, being first detectable in the iliocostalis muscles and progressing to the longissimus and multifidus and the entire spinal muscle cluster (Fig. 2C,F,H). The quadriceps muscle lagged behind the spinal muscle in the onset of fatty degeneration, but followed a similar trajectory, showing increased fatty degeneration detectable in BLA/J by 10 months of age, which progresses throughout the muscle by 22 months of age (Fig. 2E,G,H). The WT mice lacked fatty degeneration of spinal or quadriceps muscles with minimal signs of degeneration even at 22 months of age (Fig. 2B,D). These findings suggest that fatty degeneration of spinal muscles is a common and primary deficit associated with dysferlinopathy, not a consequence of the degeneration of the limb/girdle muscles. Further, it showed that the loss of spinal muscle is not a feature of the biped gait/posture of the human, but a feature of dysferlinopathy.

Spinal muscle degeneration is linked to spine rigidity and abnormal gait

In the early stages of the disease, the gait pattern of patients with dysferlinopathy is characterized by a normal step length but a wider stride width (Mahjneh et al, 2001). After 10 years of disease, patients manifest a rapid decrease in the swing phase, an increase in hip tilt, and hyperlordosis (Chawla et al, 2025; Mahjneh et al, 2001). To investigate whether adipogenic degeneration of the spinal muscle of BLA/J mice affects their locomotion on a treadmill, we conducted gait analysis. Healthy wild-type (WT) mice typically have a natural spine curvature that creates a slightly hunched posture. In contrast, symptomatic BLA/J mice lost this curvature, resulting in a noticeably flatter posture (Fig. 3A). To quantify these posture changes, we generated posture plots of WT and BLA/J mice walking on inclined (Fig. 3B–H), declined (Fig. 3I–O), and flat (Fig. EV2) treadmills. We quantified the axial and horizontal body lengths marked by vertical gold line for body’s midline length and horizontal gold lines marking the position of the forepaws and hindpaws (Figs. 3C,J and EV2B). In all locomotory conditions, BLA/J mice had a longer body length than WT mice reflecting their flattened posture (Figs. 3D,K and EV2D). Limb positioning relative to the midline (magenta line) remained unchanged, indicating that spinal muscle loss did not affect limb placement in the BLA/J mice (Fig. EV2J,L).

We next assessed gait parameters and found that on both inclined (Fig. 3B–H) and declined treadmills (Fig. 3I–O), BLA/J mice showed increased stride length, stride duration, and stance duration, with mice on the decline treadmill also showing increased swing duration (Figs. 3F–H,M–O and EV2K,M). As the treadmill was maintained at a constant speed (15 cm/s), these alterations in locomotory parameters in BLA/J mice were matched by a concomitant reduction in their forelimb and hindlimb stride frequency. We then measured gait and found that BLA/J mice had longer strides and spent more time in each step phase (stance and stride duration) on both uphill and downhill treadmills. On the downhill treadmill, they also had longer swing times (Figs. 3F–H,M–O and EV2K,M). Since the treadmill speed was fixed at 15 cm/s, these changes meant the mice stepped less often with both their front and back legs. (Fig. 3E,L). The increase in BLA/J body length as compared to the WT mice caused by a flattened spine manifested even during horizontal (0° incline) locomotion (Fig. EV2B,D). However, the locomotory deficit due to limb placement (relative to the midline) was no longer evident during horizontal locomotion (Fig. EV2B,C). Their stride frequency, length, duration, stance, and swing durations were also similar to WT mice during horizontal locomotion (Fig. EV2E–I). The findings show that rigidity and spinal muscle loss in BLA/J mice lead them to adapt during incline and decline locomotion by walking with fewer but longer and slower steps, affecting both forelimb and hindlimb stride frequency. This adjustment helps them maintain mobility and locomotory speed despite structural changes to the spinal muscle. Recognizing such compensatory strategies in humans could help identify early signs of patient adaptations to the disease and inform rehabilitation approaches aimed at delaying functional decline in mobility.

BLA/J mice show progressive myofiber degeneration and fatty replacement of the spinal muscle

While gait modification is an avenue for rehabilitative care of DYSF-mutated patients, to gain therapeutic insights into the underlying cellular pathology responsible for spine degeneration, we performed histopathological assessment. We first examined the spinal muscles from BLA/J mice from early to late symptomatic stages (6 months to 10 months of age) and matched the WT control. Hemotoxylin and eosin staining revealed centrally nucleated fibers, immune cell infiltration, and fibrofatty muscle degeneration in BLA/J spinal muscles, as early as 6 months of age (Fig. 4A). This worsens progressively by 10 months of age, with widespread replacement of myofibers with fibrofatty deposits. In comparison to the spine, and similar to our previous studies (Bittel et al, 2022; Hogarth et al, 2019), we found quadriceps were less severe at 6 and 8 months of age, with predominantly centrally nucleated fibers and immune infiltration and minimal fibrofatty replacement of muscle but increased infiltrating immune cells and necrotic fibers by 10 months of age (Fig. 4B). These phenotypes are in contrast to the WT mice, where the spinal and quadriceps muscles show no such extensive histopathological signs of muscle degeneration even at 22 months of age (Fig. 4C,D).

Immune infiltration is a physiological response of muscle damage, which guides the process of muscle repair by supporting myogenesis and extracellular matrix (ECM) remodeling by fibroadipogenic progenitors (FAPs) (Hogarth et al, 2025; Wosczyna and Rando, 2018). This sequence of events often goes awry in dystrophies due to chronic inflammation and leads to fibrofatty muscle degeneration (Hogarth et al, 2025; Hogarth et al, 2022). To assess these pathogenic features, we immunostained early symptomatic spinal muscles of BLA/J mice for signs of myofiber damage (IgM) and myofiber fatty degeneration (perilipin) (Fig. 5). IgM staining showed that the most prominent myofiber damage occurred in the iliocostalis muscles even at 6 months of age (Fig. 5A,D,E). The number of these IgM⁺ fibers in the iliocostalis increased significantly at 8-10 months of age (Fig. 5B–E). Using perilipin to mark areas with degenerated myofibers in the iliocostalis showed a significant increase starting at 8 months of age, which progressively worsened by 10 months of age (Fig. 5F). Compared to the iliocostalis, the longissimus and multifidus muscles showed relatively limited myofiber damage and adipogenic degeneration by 10 months of age (Fig. 5D,F). Taken together, these histological analyses of BLA/J mice reveal that muscle damage, followed by immune cell infiltration to clear the debris, which becomes chronic and results in the fibrofatty muscle degeneration, is the mechanistic feature of spinal muscles, and addressing these sequelae of dysferlin deficiency on spinal muscle is a potential therapeutic target.

Spinal FAPs show increased adipogenic potential amenable to therapeutic intervention

As FAPs are central players in fibrofatty muscle degeneration, they are attractive and viable cellular targets to inhibit muscle loss in muscular dystrophy (Hogarth et al, 2025). As PDGFRα expression marks the FAPs, we immunostained the spinal muscles of 6-month-old BLA/J mice for PDGFRα to assess FAP abundance in the spinal muscle. We observed that the areas enriched in IgM-labeled damaged myofibers are also highly enriched in PDGFRα-labeled FAPs (Fig. 6A). Upon differentiation, FAPs and pre-adipocytes lose PDGFRα expression and instead adopt the differentiated cell marker, perilipin, which labels adipocytes (Contreras et al, 2019; Uezumi et al, 2010; Vaziri and Faller, 1996). Consistent with this feature of FAPs, we observed reduced staining for PDGFRα (FAPs) in areas enriched in perilipin labeling (adipocytes) (Fig. 6A). With the difference in the extent of fatty degeneration of the spinal muscles in WT versus BLA/J mice, we next examined the adipogenic potential of FAPs isolated from the iliocostalis muscle of 8-month-old WT and BLA/J mice. Adipogenic differentiation of FAPs was monitored by staining with the neutral lipid probe BODIPY to visualize lipid accumulation. This showed significant adipogenesis of FAPs from the iliocostalis muscles of BLA/J mice as compared to the WT mice (Fig. 6B,C). Next, we examined spontaneous adipogenesis of FAPs isolated from the iliocostalis versus quadriceps muscles of BLA/J mice and observed the iliocostalis-derived FAPs to have significantly greater adipogenic potential (Fig. 6D,E). This identified greater commitment indicated by increased propensity of FAPs from iliocostalis muscles of BLA/J to undergo adipogenesis as compared to the FAPs from WT iliocostalis or from BLA/J quadriceps muscles, indicating targeting FAP adipogenesis as a potential target for therapy.

Potential therapeutic strategies to inhibit spinal FAP adipogenesis

To identify therapeutic targets to inhibit the adipogenic potential of FAPs from dysferlinopathic muscles, we performed RNAseq analysis of freshly isolated FAPs from muscles of 10-month-old WT and BLA/J mice (n = 4 each). Principal component analysis (PCA) of this transcriptomic dataset identified distinct clustering of FAPs from the WT and BLA/J mice (Fig. 7A). We found 1035 differentially expressed genes (DEGs - Log₂ fold change >1; P value < 0.05) with 762 genes downregulated and 273 genes upregulated in the BLA/J FAPs (Fig. 7B). Gene ontology (GO) analysis identified BLA/J FAPs exhibit upregulation of genes involved in cell cycle, chromatin organization, and inflammatory responses, while genes that regulate muscle function are downregulated (Fig. 7C,D).

These broad-ranging cellular functions are known to be regulated by Histone deacetylases (HDACs), which function to deacetylate proteins involved in regulating cell cycle and chromatin organization, affecting cell proliferation (Lagger et al, 2002; Telles and Seto, 2012). HDAC activity also alters expression of adipogenic transcription factors (TFs), including PPARγ, C/EBPβ, and SREBP1α (Lagace and Nachtigal, 2004; Lv et al, 2021). It also acetylates TFs like C/EBPα among others, altering their chromatin accessibility and adipogenic metabolism through proteins including Otub1, Pcna, Eno1, and Actg1, among others (Haberland et al, 2010; Xu et al, 2013). We found that several of these genes were also dysregulated in BLA/J FAPs (Fig. 7E). Due to these features, the ability of HDAC inhibitors to regulate expression of these genes, and recent approval of the HDAC inhibitor Givinostat for Duchenne Muscular Dystrophy treatment (Consalvi et al, 2013), we tested if this could prevent adipogenesis of spinal FAPs from dysferlinopathic mice. Our previous work has identified that MMP inhibitor Batimastat can inhibit dysferlinopathic FAP adipogenesis (Hogarth et al, 2019), therefore we used this drug as a positive control. Givinostat treatment of spinal FAPs from 8-month-old BLA/J mice inhibited their spontaneous adipogenesis to the same extent as Batimastat (Fig. 7G,H).

A larger group of differentially expressed genes between the WT and BLA/J FAPs included those linked to the Wnt/β-catenin pathway. Notably, these genes include the classic transcriptional targets of β-catenin - Cyclin D1 (Ccnd1), p21(CIP1/WAF1- Cdkn1a), and Axin (Axin2) (Jho et al, 2002; Shtutman et al, 1999; van de Wetering et al, 2002) (Fig. 7F). They also included adipogenic co-regulators of Wnt signaling - nuclear co-receptor (Ncoa-1), fatty acid synthase (Fasn), fatty acid binding protein 4 (Fabp4), stearoyl CoA desaturase 1 (Scd1) (Bagchi et al, 2020a; Bagchi et al, 2020b; de Winter and Nusse, 2021; Ross et al, 2000; Triki et al, 2017; Yang Loureiro et al, 2023) (Fig. 7F). As GSK3 is the central regulator that controls the Wnt/β-catenin pathway to modulate adipogenesis (Reggio et al, 2020; Zhang et al, 2024), we tested the effect of a potent GSK3 inhibitor LY2090314. Unlike the partial inhibition of spontaneous BLA/J FAP adipogenesis by HDAC and MMP inhibition, GSK3 inhibition fully abolished spontaneous adipogenesis of dysferlinopathic FAPs (Fig. 7G,H). These results identify GSK3 inhibition as a particularly strong druggable target and a promising therapeutic approach. Together, the molecular analyses and targeted in vitro drug testing open avenues to counter the enhanced adipogenic commitment of BLA/J spinal FAPs, offering potential strategies to nearly abolish the adipogenic degeneration of dysferlinopathic spinal muscle described in this study.

Canonical Wnt signaling enables BLA/J FAP adipogenesis, and GSK3 inhibition reverses it

Given the strong inhibitory effect of GSK3 inhibition on the adipogenic fate of dysferlinopathic FAPs, we investigated the mechanisms underlying this response. Analysis of Wnt ligands identified by RNA-seq comparing WT and BLA/J FAPs revealed increased expression of several canonical and non-canonical Wnt ligands in BLA/J FAPs (Fig. 8A). To validate these findings and determine the effect of GSK3 inhibition on Wnt ligand mRNA expression, we isolated FAPs from WT and BLA/J mice and treated them with LY2090314 or DMSO (control). qPCR analysis confirmed that the top three Wnt ligands Wnt1, Wnt5b, and Wnt10a identified by RNAseq showed a trend or a significant overexpression in BLA/J FAPs and noncanonical Wnt5a, which showed a lower increase by RNAseq, was unchanged in BLA/J FAPs by qPCR (Fig. 8A,B). Treatment of BLA/J FAPs with LY2090314 suppressed the expression of all of the above Wnt ligands (Fig. 8B). To determine whether altered canonical Wnt signaling contributes to the adipogenic phenotype, we assessed TCF/LEF activity by measuring Axin2 and Ccnd2. Both transcripts were reduced in BLA/J compared to WT FAPs, and LY2090314 treatment trended to reverse this inhibition, indicating restoration of canonical Wnt signaling (Fig. 8B).

At the protein level, dysferlinopathic FAPs showed that the high adipogenicity of BLA/J FAPs aligned with an elevated level of PPARγ in BLA/J FAPs that was decreased by treatment with LY2090314 (Fig. 8C,D). Concomitant with this, we observed reduced levels of active β-catenin and a decrease in the ratio of active to total β-catenin with no change in the level of total β-catenin (Fig. 8C,D). This increase in β-catenin activity in BLA/J FAPs was associated with reduced inhibitory Ser9 phosphorylation of GSK3β with no change in the level of total or Tyr216/279 activated GSK3β (Fig. 8C,D). While LY209314 treatment did not affect total or Ser9-inhibited GSK3β, it significantly reduced the level of Tyr216/279-activated GSK3β (Fig. 8C,D). Together, these findings demonstrate that the canonical Wnt/β catenin signaling is impaired in BLA/J FAPs and is restored by GSK3 inhibition. In contrast, analysis of non-canonical signaling, showed highly variable c-Jun phosphorylation in WT FAPs, with no change in the ratio of phosphorylated to total c-Jun between WT and BLA/J FAPs with or without treatment with LY2090314 (Fig. 8C,D).

Together, these findings support a model in which increased adipogenesis of dysferlinopathic BLA/J FAPs is due to elevated Wnt ligand expression with impairment of canonical Wnt/β-catenin signaling due to increased GSK3β activity. This leads to reduced active β-catenin, increased PPARγ expression, and enhanced adipogenesis. Treatment of BLA/J FAPs with LY2090314 reverses the impaired canonical Wnt/β-catenin signaling by reducing the activated pTyr216/279-GSK3β, which restores β-catenin signaling and inhibits PPARγ expression, inhibiting their adipogenic differentiation.

GSK3 inhibition prevents fatty degeneration of dysferlinopathic muscle

The above results support the potential therapeutic value of GSK3 inhibition to limit adipogenic loss of dysferlinopathic muscle. To test this potential, we carried out in vivo GSK3 inhibition by LY2090314. We treated 7-months old BLA/J mice with LY2090314 on alternate days for 6 weeks to assess if this extended period of treatment will impact the spontaneous onset of adipogenic degeneration of spinal muscles in these mice (Fig. 9A). Examination of the gross pathology of the spinal muscle from untreated and drug-treated BLA/J muscle showed a notable preservation of the spinal muscle in drug treatment mice (Figs. 9B and EV3A). To examine this further, we used perilipin staining to mark the muscle areas replaced by intramuscular adipose tissue, indicative of fatty degeneration of the myofibers. In the placebo-treated control BLA/J mice, we observed abundant perilipin staining in of the spinal muscle, which was lacking in the LY2090314-treated mice (Figs. 9C,D and EV3B).

To determine whether the effect of LY2090314 on adipogenic muscle loss is mediated through inhibition of injury-triggered adipogenic fate specification in BLA/J FAPs, we performed focal injury of the asymptomatic TA muscles in the BLA/J mice using the approach we previously described (Hogarth et al, 2019). This approach involved three consecutive rounds of injury with 2 weeks in between each injury to allow muscles to repair (Fig. 9E). We previously identified that this can be prevented by in vivo treatment with the hedgehog signaling inhibitor Batimastat (Hogarth et al, 2019). In our current study, we treated mice with alternate-day dosing with LY2090314 starting from the day before injury and continuing for the next 6 weeks. Assessment of perilipin immunostaining of the injured TA muscles from control-treated mice confirmed that 3 bouts of successive injury triggered robust intramuscular fat after 3-bouts of injury (Figs. 9F,G and EV3C). In contrast, TA muscles from LY2090314-treated mice were completely protected from injury-induced adipogenesis (Figs. 9F,G and EV3C). Despite the noted benefits of the LYS2090314 treatment in preventing adipogenic muscle loss, the treated mice showed multiple adverse effects, which ranged from scarring of the skin and the diaphragm (site of intraperitoneal injection) and enlargement of the spleen (Fig. EV3D,E). While these effects should be considered in future efforts to develop long-term therapeutic use of this GSK3 inhibitor, our in vivo analysis validates the in vitro studies, identifying the suitability of GSK3 as a target to inhibit adipogenesis of dysferlinopathic muscle resident FAPs and establish its therapeutic potential to prevent adipogenic muscle loss in dysferlinopathy.

Human subjects

A retrospective analysis was conducted on twenty patients diagnosed with dysferlinopathy at Jeonbuk National University Hospital, Pusan National University Yangsan Hospital, and Hôpital Pitié-Salpêtrière. Patients were included based on the following criteria: progressive proximal muscle weakness, elevated serum creatine kinase (CK) levels, degenerative muscle changes observed on CT or MRI, and genetic confirmation of either two pathogenic heterozygous mutations or one or two homozygous mutations in the DYSF gene and absence of dysfelin on muscle biopsy or monocytes. All patients were diagnosed between 2009 and 2022 at Pusan, Jeonbuk National University Hospital and between 1997 and 2023 at Hôpital Pitié-Salpêtrière. The patient cohort was ethnically diverse, comprising 6 Asian patients, 9 White/Caucasian patients, and 5 Arabic patients. All enrolled patients underwent muscle MRI, including imaging of spinal musculature, and serum CK levels were assessed at disease onset. Specifically, five patients underwent lumbar spine MRI, and 15 underwent whole-body MRI [all 14 French patients underwent whole-body MRI]. MRI scans were performed using 3-T clinical MRI systems (PrismaFit, or Magnetom Verio, Siemens Healthcare, Erlangen, Germany), and individuals were imaged by T1-weighted imaging. The Mercuri score was used to grade fatty involvement of muscles (Mercuri et al, 2007). The control group consisted of age- and gender-matched individuals who underwent spine MRI for minor trauma or back pain and had no history of neuromuscular disease. The study protocol was approved by the Institutional Review Boards of Jeonbuk National University Hospital (IRB No. 2022-11-051) and Pusan National University Yangsan Hospital (IRB No. 05-2023-043). The French patients were enrolled in the Clinical Outcome Study for Dysferlinopathy (Jain COS) - ClinicalTrials.gov # NCT01676077. The study was approved by the local ethics committee (Comité de Protection des Personnes [CPP] Île de France VI, ID RCB 2012-A01689-34), and it was registered in accordance with French data protection regulations (CNIL, Commission nationale de l’informatique et des libertés). Informed consent was obtained from all human subjects, and the MRI protocols conformed to the principles set out in the WMA Declaration of Helsinki.

Mouse model

All animal experiments used in this study were approved by the Children’s National Medical Center Institutional Animal Care and Use Committee (Protocol #30709) and were conducted in accordance with guidelines for the care and use of laboratory animals. WT (C57BL/6 J) and BLA/J (B6.A-Dysf^prmd/GeneJ) mice were purchased from Jackson Laboratory (Bar Harbor, ME) and maintained in-house. All animals were maintained in a germ-free facility and individually vented cage system under a controlled 12-h light/ dark cycle with free access to food and water. Anatomical T1 MRI images for mice were acquired on a Bruker Avance400 (Bruker Biospin, Billerica, MA, USA) 9.4 T 89 mm vertical bore MRI with a 26 mm ID volume coil. Two sets of axial T1-weighted fast spin echo scans were acquired, one aligned with the spine and the other aligned with the femur, each scanned twice, with and without fat saturation. Scans were acquired using the RARE (Rapid Acquisition with Refocusing Echoes) sequence: TE = 8.5 ms, TR = 4000 ms, RARE Factor = 4, NA = 12, FOV = 25.6 × 25.6 mm, MTX = 256 × 256, 85 × 0.3 mm slices. To quantify the fat fractions, fat-only images were created by subtracting the T1-weighted with fat-saturation image from the T1-weighted without fat-saturation image using ImageJ image analysis software (version 1.53t, National Institutes of Health, Bethesda, Maryland). We applied a region-of-interest (ROI) approach to test specific structures throughout the lumbar spine and quadriceps in axial view and selected. The ROIs of the spinal muscle and quadriceps muscle were established based on the deep fascia surrounding the muscle observed in T1-weighted MR. The fat fraction was measured within the ROI defined as muscle, and the fat percentage within the spinal muscle or quadriceps was expressed as a percentage by measuring the fat fraction area, respectively, when each muscle ROI was set to 100%. All ROIs were drawn and quantified manually by a blinded experimenter.

Histology and immunofluorescence

Lumbar spine and quadriceps muscles of mice were collected. The resected lumbar spine with spinal muscles was fixed in 4% paraformaldehyde and decalcified in Calci-Clear Rapid (National Diagnosis, GA, USA) for three days. The spine tissue was cut into 10-µm-thick sections based on the 4th lumbar vertebra. For quadriceps, frozen sections 8-µm thick were cut based on the mid area. To evaluate the histological findings, thawed tissue sections were processed for H&E and immunofluorescence. Immunofluorescence was performed by blocking with 5% BSA and incubation with primary antibodies against PDGFRα (1:250, Cell Signaling, #3174S), perilipin (1:250, Sigma, #P1873), laminin-2α-chain (1:100, Sigma, #clone 4H8-2), and anti-IgM (1:100, Invitrogen, #A-21043). Alexa Fluor 488 and/or 568 conjugated (1:500, ThermoFisher) secondary antibodies were used for visualization and co-stained with wheat germ agglutinin (1:500, ThermoFisher, #W11261) and DAPI (ThermoFisher, #P36941). Images were captured using a VS120 slide-scanning microscope (Olympus America) at 200x magnification and quantified using CellSens software (Olympus America) and ImageJ software (National Institutes of Health).

Microscopy and image analysis

Frozen sections, 8-µm thick, were cut from the collected mouse muscles. Immunofluorescence was performed by fixing sections in chilled 4% Paraformaldehyde, blocking with 5% BSA in PBST, and incubation with a primary antibody against perilipin-1 (1:250, Sigma, P1873). Myofiber membranes were marked using AlexaFluor 488-conjugated wheat germ agglutinin (1:500, ThermoFisher, #W11261), which was added during incubation with secondary antibody. Coverslips were mounted using ProLong Gold with DAPI (ThermoFisher, #P36941). Microscopy was performed using an Olympus VS200 Virtual Slide Scanning System. Quantification was performed using QuPath 0.6.0 (Bankhead et al, 2017). The quantification was done using the pixel classifier function to get the total area of the muscle section and the area of the perilipin-1 staining, which was used to calculate the total positive area relative to the muscle section area. Analysis of the spinal muscles was done by averaging values across three randomly picked ROIs (800 µm × 800 µm) in the spinal muscle of each mouse and measuring the perilipin-1-stained area over the total area.

Motor function analysis

Gait dynamic and locomotion analysis were performed using the DigiGait imaging system (Mouse Specifics, Quincy, MA). WT and BLA/J mice were placed on a motorized transparent treadmill within a plexiglass compartment, moving at 18 cm/s (horizontal degree) or 15 cm/s (15° incline and 15° decline). Digital videos of the ventral surfaces of the fore and hind paw placement were acquired using a camera mounted beneath the treadmill. The videos were analyzed using DigiGait software, which automatically processes from nine to twelve strides to calculate gait parameters. Approximately 4 s videos were collected from each mouse.

Acute muscle injury

Notexin injuries were performed by applying Nair to the anterior hind limbs before intramuscular injection of 50 µL notexin (5 μg/ml, Latoxan, #L8104) into the tibialis anterior using a 0.3-ml ultrafine insulin syringe (BD Biosciences, #324906). Mice were allowed to recover for 14 days before the next injury was performed. Injury was repeated a total of three times, and was collected two weeks after the final injury. For drug treatment, mice at 30 weeks of age were given an intraperitoneal dose of LY2090314 (25 mg/kg; MedChemExpress-HY-16294). The drug was dissolved in DMSO and then diluted in corn oil, 5% and 95%, respectively, for IP injection. Mice were dosed 3 days a week for 6 weeks, starting the day before the first Notexin injury.

Cell culture studies

FAPs were isolated from the iliocostalis or quadriceps of 8-month-old WT and BLA/J mice. Muscle digestive procedure was previously described (Hogarth et al, 2019). Primary cells isolated from muscles were stained with anti-CD140a-APC (Biolegend, #135908) and isotype control (Biolegend, #400512) for FACS. Stained cells with anti-CD140a-APC were sorted on an Influx cell sorter (Becton Dickinson, #646500). The CD140a⁺ cells were plated in 96-well plates at 15,000 cells/well density. Cells were cultured in FAPs growth media (DMEM with 20% FBS, 1% penicillin, and bFGF (2.5 ng/ml, Biolegend, #579604)) for 3 days to reach 80% confluence and then were switched to adipogenic maintenance media (DMEM with 10% FBS and insulin (10 µg/ml)) for 3 or 5 days. For drug treatment, the CD140a⁺ cells were plated in 96-well plates at 15,000 cells/well density. After reaching 80% confluency, they were switched into the adipogenic maintenance media and then were treated with either DMSO (Sigma-Aldrich), Batimastat (10 µM, Sigma-Aldrich, #SML0041) (Hogarth et al, 2019), Givinostat (50 nM, Selleckchem, #ITF2357) (Milan et al, 2018), or LY2090314 (10 nM, Selleckchem, #S7063) (Reggio et al, 2020) for 3 days. Cells were then fixed in 4% paraformaldehyde for 30 min and stained with boron difluoride dipyrromethene (BODIPY, ThermoFisher, #D3922) and Hoechst (ThermoFisher, #33342) for 30 min. Images were captured with a Leica DMi8 microscope (Leica, Germany) and quantified using ImageJ software (Version 1.54p).

RNA isolation, qPCR, and western blotting

To evaluate the changes in Wnt signaling and the efficacy of LY2090314 during FAP adipogenesis from WT and BLA/J iliocostalis, we isolated CD140a⁺ from 8-month-old WT or BLA/J mice’s iliocostalis using a cell sorter. The CD140a⁺ cells were plated in 12-well plates at 2 × 10⁵ cells/well density. Cells were cultured in FAPs growth media for 3 days to reach 80% confluence and then were switched to adipogenic maintenance for 3 days with DMSO or LY2090314 (20 nM).

To obtain total RNA, cells were subjected to TRIzol reagent (Invitrogen) extraction. Purified RNA (1 μg) was reverse-transcribed using a first-strand cDNA synthesis kit (Applied Biosystems, Foster City, CA, USA). Quantitative PCR was performed in 384-well plates using the ABI Prism 7900HT Sequence Detection System (Applied Biosystems). Gene expression for all mRNA targets was normalized to Actb mRNA transcript levels. RNA from each sample was analyzed individually. The primer sequences used for PCR are listed in Table S1.

We extracted protein from cells using a protein extraction reagent (Thermo Fisher Scientific, 78505). Homogenates (20 μg of total protein) were then separated by 10–15% SDS-PAGE and transferred to nitrocellulose membranes. After blocking the samples with 5% skim milk, the blots were probed with primary antibodies for overnight at 4 °C. They were then incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Immunoreactive bands were detected using a Las-4000 imager (GE Healthcare Life Science, Pittsburgh, PA, USA). The primary antibodies used for Western blotting were PPARγ (1:2500, Santa Cruz Biotechnology, #sc-7273), β-catenin (1:2500, Santa Cruz Biotechnology, #sc-7963), activie β-catenin (1:2500, Cell Signaling, #8814), GSK3β (1:2500, Cell Signaling, #9315), pGSK3β (Ser9) (1:2500, Cell Signaling, #9336), pGSK3β/3α (Tyr216/Tyr279) (1:2500, Abcam, #ab68476), p-c-Jun (1:1000, Cell Signaling, #2361), c-Jun (1:2500, Cell Signaling, #9165), and HSP90 (1:2500, Enzo Life Sciences, #ADI-SPA-836-F).

Statistical analysis

GraphPad Prism 9.0.0 was used for statistical analyses of data. Sample sizes were based on prior studies. As detailed in Appendix Table S1, we used simple linear regression to assess relationships between continuous variables, unpaired t-tests and Mann–Whitney tests to compare two independent groups, and two-way ANOVA to evaluate the effects of two factors and their interaction. Holm–Šídák multiple-comparisons tests were used following ANOVA to identify specific group differences while controlling for multiple testing. A Wald test with Benjamini–Hochberg correction was also used to assess statistical significance while controlling the false discovery rate. A P value < 0.05 was considered significant. Data plots were reported as scatter plots with mean ± SD. Animals used in the experiments described in the manuscript were chosen for their age and sex as noted, but were randomized for litter and for weight and other physical parameters. The experimenters were not blinded to the genotype, but during data analysis, efforts were made to minimize unblinding of the experimenters until after the analysis.

RNA-seq analyses

FAPs were isolated from muscle as described above and previously (Hogarth et al, 2019). Cells were counted using a hemocytometer and lysed in TriZol (ThermoFisher Scientific), and total RNA was isolated using a Direct-zol RNA Miniprep (Xymo Research #R2051). For RNA sequencing, libraries were prepared using poly(A) selection and sequenced on an Illumina NovaSeq X platform with 150-bp paired-end reads, generating approximately 30 million reads per sample. Quality control was performed on FASTQ files generated from sequencing using FastQC (version 0.11.9) for individual samples and multiqc (version 1.29) (Ewels et al, 2016) for all samples. After quality trimming with Trim Galore (version 0.6.10), the reads were aligned to the human reference (hg38) using STAR (version 2.7.11b) (Dobin et al, 2013), and the counts were estimated with RSEM (version 1.3.1) (Li and Dewey, 2011). For differential expression, we first filtered out low read counts (<10 reads) and then estimated sample quality using principal component analyses (PCA). Normalization and differential expression were performed with DESeq2 (version 1.46.0) (Love et al, 2014) using an adjusted p-value threshold of <0.05 and an average log2 fold change of >1. Gene set enrichment analysis was done with clusterProfiler (version 4.14.6) (Yu et al, 2012) package, utilizing gene sets from the MSigDB (version 2024.1) (Liberzon et al, 2015) database. Visualization of differentially expressed genes using volcano plots was performed using EnhancedVolcano (version 1.24.0) (https://github.com/kevinblighe/EnhancedVolcano), whereas genes associated with GSK3, HDAC, and chromatid-associated pathways were visualized using pheatmap (version 1.24.0).

The authors JKJ and MWH hold US Patent #12,594,278, which is related to the work reported here. The remaining authors declare no competing interests.