Skeletal muscle wasting is a devastating consequence of cancer that contributes to increased complications and poor survival, but is not well understood at the molecular level. Herein, we investigated the role of Myocilin (Myoc), a skeletal muscle hypertrophy-promoting protein that we showed is downregulated in multiple mouse models of cancer cachexia. Loss of Myoc alone was sufficient to induce phenotypes identified in mouse models of cancer cachexia, including muscle fiber atrophy, sarcolemmal fragility, and impaired muscle regeneration. By 18 months of age, mice deficient in Myoc showed significant skeletal muscle remodeling, characterized by increased fat and collagen deposition compared with wild-type mice, thus also supporting Myoc as a regulator of muscle quality. In cancer cachexia models, maintaining skeletal muscle expression of Myoc significantly attenuated muscle loss, while mice lacking Myoc showed enhanced muscle wasting. Furthermore, we identified the myocyte enhancer factor 2 C (MEF2C) transcription factor as a key upstream activator of Myoc whose gain of function significantly deterred cancer-induced muscle wasting and dysfunction in a preclinical model of pancreatic ductal adenocarcinoma (PDAC). Finally, compared with noncancer control patients, MYOC was significantly reduced in skeletal muscle of patients with PDAC defined as cachectic and correlated with MEF2c. These data therefore identify disruptions in MEF2c-dependent transcription of Myoc as a novel mechanism of cancer-associated muscle wasting that is similarly disrupted in muscle of patients with cachectic cancer.
This work identifies a novel transcriptional mechanism that mediates skeletal muscle wasting in murine models of cancer cachexia that is disrupted in skeletal muscle of patients with cancer exhibiting cachexia.
Cancer-associated cachexia is a multifactorial syndrome characterized by the involuntary loss of body and skeletal muscle mass (with or without fat loss), which cannot be completely reversed through nutritional support (1). Skeletal muscle loss reduces tolerance to cancer treatments, increases complications following cancer surgery, and is strongly predictive of reduced survival (2). However, due to the inherent complexity of cancer cachexia, effective therapeutics to preserve muscle mass in patients with cancer are not currently available.
The etiology of cancer-associated muscle wasting involve impaired protein homeostasis and negative energy balance that are driven by a number of factors, including increased metabolism, reduced appetite leading to reduced caloric intake, and chronic systemic inflammation induced by both host- and tumor-derived factors (1, 3, 4). Inflammatory cytokines may play a role in the wasting process through several mechanisms including interference of centrally mediated pathways involving the hypothalamus, which regulate satiety and hunger, and through direct stimulation of pathways, which lead to an imbalance in muscle protein breakdown and synthesis (5). In addition to inflammatory cytokines, other circulating factors dysregulated in tumor-bearing hosts may also contribute to muscle wasting, including TGFβ family members, angiotensin II, parathyroid hormone-related protein, heat shock proteins, and tumor-derived exosomes (1). For a more detailed review of the systemic mechanisms involved in cancer-associated cachexia and muscle loss derived from experimental models and patient samples, readers are referred to the work by Baracos and colleagues (1).
The intracellular mechanisms whereby muscles atrophy in response to cancer have largely been derived from animal models—and, similar to other models of muscle wasting, support increased muscle protein breakdown coupled with reductions in protein synthesis as key mechanisms of wasting (6, 7). Despite this, less convincing data exist to support an imbalance in these pathways in muscle of patients with cancer with cachexia. However, mechanistic studies in mouse models of cancer cachexia indicate that muscle membrane damage (8), coupled with impaired muscle regeneration and repair (9), may also play causative roles in cancer associated muscle loss. In support of these findings, disruptions to membrane architecture (8) and muscle damage (10) have been observed in muscle biopsies from patients with cachectic cancer. Yet, the mechanisms that induce these phenotypes are not known.
Previous studies profiling the transcriptional networks changed in mouse skeletal muscle in response to tumor burden have provided key insight into pathways that may be involved in the muscle wasting phenotype (11), including signaling through the atrophy-associated Forkhead Box O (FoxO) transcription factors (12). Indeed, multiple atrophy–associated genes involved in muscle protein breakdown are upregulated through the FoxO transcription factors (12, 13), which mediate muscle atrophy in mice bearing Lewis lung carcinoma (14) and C26 tumors (12). In addition to upregulating genes involved in muscle proteolysis, we also identified FoxO factors as key upstream factors necessary for the cancer-induced repression of many downstream target genes (12). While the significance of these repressed targets are not yet clear, these findings may be particularly relevant to human cancer cachexia, because muscle biopsies from patients with cachectic cancer display increased levels of FoxO1 (10, 15), and significant gene repression (16).
One FoxO target gene repressed in skeletal muscle in response to C26 tumor burden that could play a role in the atrophy phenotype is Myocilin (Myoc). Although few studies have examined the role of Myoc in skeletal muscle, Myoc has been identified as a prohypertrophic protein that binds and stabilizes the dystrophin glycoprotein complex (DGC; ref. 17), which is well-established to stabilize the muscle fiber membrane. Myoc is most well-known for its role in the pathogenesis of glaucoma, where genetic mutations in MYOC lead to misfolding and aggregation of MYOC protein within the trabecular meshwork, resulting in high intraocular pressure (18). However, Myoc is also expressed in skeletal muscle, where it binds to α1-syntrophin at the sarcolemma (17). In transgenic mice overexpressing Myoc, skeletal muscles show an approximately 40% increase in skeletal muscle mass and suppression of the FoxO signaling axis (17). These mice also exhibit increased protein levels of dystrophin and enhanced binding of α1-syntrophin to dystrophin, implicating Myoc as a key regulator of DGC stability and, by association, sarcolemma stability. Although muscles from mice lacking Myoc have not been fully characterized, these mice show decreased binding of α1-syntrophin to dystrophin (17), suggesting that in conditions of Myoc deficiency, the sarcolemma may be less stable.
In this study, we determined whether Myoc downregulation is a common response in multiple models of cancer cachexia, and further investigated the biological significance of Myoc deficiency on skeletal muscle in isolation and in the context of cancer cachexia. Through these studies, we demonstrate that Myoc is downregulated in multiple models of cancer cachexia and that MYOC gain of function deters muscle wasting in the C26 model of cachexia, and in a murine model of PDAC-associated cachexia (19), involving orthotopic implantation of pancreatic tumor cells isolated from a syngeneic C57BL/6 KRASG12D P53R172H PdxCre+/+ (KPC) mouse. We further identify myocyte enhancer factor 2C (MEF2c) as a key upstream transcription factor necessary for maintaining Myoc expression in skeletal muscle, whose gain of function prevents Myoc downregulation and deters cancer-induced muscle wasting and dysfunction. Finally, using patient biopsies, we demonstrate that MYOC mRNA is reduced in skeletal muscle of patients with cachectic PDAC, and correlates with the levels of MEF2c, thus indicating that these findings may be relevant to muscle wasting in patients with cancer.
Materials and Methods
Mice and mouse models of cancer cachexia
The University of Florida Institutional Animal Care and Use Committee approved all animal procedures described herein. Mice were housed in a temperature-controlled facility with a 12-hour light/dark cycle, with standard diet and water provided ad libitum. Myocilin-null (Myoc−/−) mouse breeding pairs were received from Dr. S Tomarev from the National Eye Institute, and have been described previously (20).
Murine Colon-26 (C26) cells (NCI-DTP catalog no. colon 26, RRID:CVCL_0240) were purchased from the National Cancer Institute Cell Repository (Bethesda, MD). Culturing and subcutaneous inoculation of C26 cells into CD2F1 mice to induce cachexia was performed as described previously (14). Panc02 (RRID:CVCL_D627) murine pancreatic adenocarcinoma cells and Panc02-H7 cells (RRID:CVCL_D628), which are a highly metastatic variant of Panc02 (21), were received from Dr. Min Li at The University of Oklahoma (Norman, OK). For the orthotopic KPC (LSL-KrasG12D/+;LSL-Trp53R172H/+;Pdx-1-Cre) model of cancer cachexia (19), we acquired KPC FC1245 tumor cells derived from a KPC mouse backcrossed to the C57BL/6 genetic background from Dr. David Tuveson (Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, NY). The authenticity of the KPC FC1245 cell line has been described previously (22). Cell lines were verified to be Mycoplasma free prior to receipt or by IDEXX BioAnalytics using PCR evaluation. Tumor cells were maintained in either RPMI (Panc02/Panc02-H7) or DMEM (KPC) supplemented with 10% FBS, 1% penicillin, 1% streptomycin in a humidified chamber at 37°C with 5% CO2. Panc02 and Panc02-H7 cells (1 × 106) or KPC cells (2.5 × 105) were orthotopically injected into the pancreas of 10- to 12-week-old male and female C57BL/6 wild-type (WT) mice (The Jackson Laboratory, Bar Harbor, ME) and Myoc−/− mice as described previously (23). For orthotopic studies, Sham mice underwent a comparable Sham surgery, in which the pancreas was surgically exposed and injected with sterile saline. Mice were euthanized when tumor-bearing groups reached Institutional Animal Care and Use Committee (IACUC)-mandated experimental endpoint criteria based on deterioration of body condition score, which is reflective of estimated tumor-free body weight loss of ≥15% and/or signs of pain or distress, including hunched posture and failure to groom. Cell lines were verified to be Mycoplasma free prior to receipt or by IDEXX BioAnalytics using PCR evaluation within the last 6 months. All cell lines were utilized prior to passage 13.
Additional preclinical models of cancer cachexia, including those that utilize human L3.6 pancreas-liver (L3.6pl) cells (RRID:CVCL_0384) and the PDAC patient-derived xenograft (PDX) model, in which tibialis anterior (TA) muscles were utilized for qRT-PCR analyses of Myoc and Mef2c, were described previously (23–25). Briefly, human L3.6pl pancreatic cancer cells were injected orthotopically into the pancreas or subcutaneously into the flank of immunocompromised NOD.Cg-Prkdcscid IL2rgtm1Wjl/SzJ (NSG) mice. PDAC tumor fragments were either sutured to the pancreas or subcutaneously implanted into the flank of NSG mice. Sham mice for the orthotopic L3.6pl and PDX studies underwent a comparable surgery, in which the pancreas was surgically exposed and either injected with sterile saline (L3.6pl studies) or a suture placed loosely around the pancreas and then removed (PDX studies). The L3.6pl cell line was authenticated within 6 months of receipt by short tandem repeat analysis.
Attainment of biospecimens from noncancer control subjects undergoing benign abdominal surgery and from patients with PDAC undergoing tumor resection surgery was compliant with an approved Institutional Review Board protocol, with written informed consent received from all patients. Detailed demographic information on the patients included in this study were previously published (10), and can be found in Supplementary Tables S1 and S2.
Body composition analysis
Preoperative CT scans were identified. An axial image at the level of the third lumbar (L3) vertebra was accessed for each patient; the lower aspect of the L3 transverse processes was used to guide image acquisition. Cross-sectional area at the L3 vertebra correlates to total body composition (26, 27). Manual image segmentation was performed using sliceOmatic version 5.0 (Tomovision). Specific tissue thresholds were used that correspond to the radiodensities of skeletal muscle and adipose. Cross-sectional skeletal muscle area (cm2) was calculated by the summative area of −30 and 0 Hounsfield units (HU), 1 and 25 HU, and 26 and 150 HU to further characterize myosteatosis (28). Intermuscular adipose cross-sectional area was measured between -150 and -50 HU. Skeletal muscle area was normalized to height (m2) yielding a lumbar skeletal muscle index (SMI, cm2/m2). Mean muscle attenuation (MA) is reported in HU for the entire muscle area at the L3 vertebrae.
Downhill running and assessment of muscle damage
Mice were subjected to downhill treadmill running at a decline of 140 at a speed of 8 to 10 m/minute, for 45 minutes as described previously (29). To assess muscle damage, MRI was performed on hindlimb muscles in a 4.7T, horizontal bore magnet (Agilent; VMJ version 3.1) immediately following completion of the running protocol. This protocol is detailed in Supplementary Methods, and has been described previously (30, 31). Six hours following treadmill running, mice received an intraperitoneal injection of 2.5 mg Evan's blue dye (EBD) per 25 g body weight dissolved in 100 μL sterile saline, and the next day muscles were harvested for cryosectioning and histologic analyses of EBD uptake and morphology.
To study muscle regenerative capacity and subsequent regrowth, TA muscles of 10- to 12-week-old male and female WT and Myoc−/− mice were injected with 75 μL of 10 μmol/L cardiotoxin (CTX; Calbiochem) and harvested 2, 4, or 6 weeks postinjury.
AAV vectors and delivery and plasmid injections
AAV9-tMCK-MYOC-polyA-CMV-GFP-polyA and AAV9-tMCK-MEF2c-polyA-CMV-GFP-polyA vectors containing codon-optimized mouse transgenes were custom created by SignaGen Laboratories. A truncated muscle creatine kinase (tMCK) promoter was selected to drive expression of MYOC and MEF2c in adult muscle fibers only. AAV9-CMV-GFP was used as control. Two to three weeks prior to beginning tumor studies, AAV9 vectors (1 × 10E11 vg) dissolved in 20 μL lactated Ringer solution were injected through the skin and into the mid-belly of TA muscles, through both heads of the gastrocnemius muscles to optimally target the soleus, and/or into the intrapleural cavity to target the diaphragm, as described previously (12).
The constitutively active FoxO1 and FoxO3a plasmids, also known as triple phosphorylation mutants (TM), due to mutations within the three Akt phosphorylation sites, which renders them unable to be inactivated by Akt, have been described previously (12). The plasmid encoding for dominant negative (d.n.) MEF2c (amino acids 1 to 105) contains the DNA-binding domain, but lacks the transcriptional activation domain, and was kindly provided by Eric Olson (University of Texas Southwestern Medical Center, Dallas, TX; ref. 32). The 3XMEF2-luc plasmid was obtained from Addgene (plasmid #32967), and was deposited by Ron Prywes (Columbia University, New York, NY). Expression plasmids or empty vector (EV) plus or minus the MEF2 luciferase reporter dissolved in 50 μL 1× PBS were injected and electroporated into rat soleus muscles, harvested 4 days later, and processed for either qRT-PCR or measurement of luciferase activity (12). The Myoc promoter reporter was custom-created by Switch Gear Genomics (an Active Motif Company) and comprises of a 521 base pair (bp) fragment of the mouse Myocilin promoter region, including 432 bp upstream, and 89 bp downstream, of the transcription start site (TSS; C57BL/6J chromosome 1, GRCm38.p4 C57BL/6J, 162638718 to 162639238). The Myoc promoter fragment was cloned into the pLightSwitch reporter vector immediately upstream of the coding sequence for an optimized Renilla luminescent reporter gene (WT Myoc promoter reporter). To create the mutated (m)MEF2 Myoc promoter reporter, the conserved MEF2-binding motif located 169 bp upstream of the TSS (TTCAAAATAGC), was mutated to TTCACGCTAGC, to prevent MEF2 binding. The WT or mMEF2 Myoc promoter reporter plasmids were injected and electroporated into TA muscles of mice (10 μg/25 μL 1× PBS) and muscles harvested 4 days later. Muscles were homogenized in 1:10 weight/volume passive lysis buffer (Promega) and luciferase activity measured using a LightSwitch Luciferase Assay Kit according to manufacturer's guidelines (SwitchGear Genomics) and a GloMax Single Tube Luminometer (Promega).
Histochemistry and cross-sectional area analyses
Processing of skeletal muscle tissue for histologic analysis, including hematoxylin and eosin (H&E) staining, was performed as described previously (33). Oil Red O and Trichrome staining was performed by the Molecular Pathology Core at the College of Medicine, University of Florida (Gainesville, FL), on fresh frozen sections (10 μm) provided by our lab. Determination of transfection efficiency (based on GFP fluorescence) and EBD uptake (using Texas-Red filter) were performed on muscle cross-sections immediately following cryosectioning using direct fluorescent microscopy. Sections were visualized and imaged using a Leica DM5000B microscope attached to a digital camera (Leica Microsystems), with representative images spanning each region of the muscle captured for analyses. A Leica Application suite, version 3.5.0 software or ImageJ were used to trace and measure cross-sectional area (CSA) of muscle fibers, while ImageJ was used to quantify extracellular tissue surrounding myofibers in H&E-stained images, both of which have been described previously (33).
Homogenization of mouse skeletal muscle tissue, isolation of myofibrillar fractions, Western blotting, and quantification using the Odyssey Infrared System (LI-COR) were performed as described previously (34, 35). Primary antibodies were used according to manufacturer's directions: [anti-Myocilin (MABN866, Sigma Aldrich); anti-α-tubulin (T6199, Sigma-Aldrich); anti-Spectrin-β1 (#MA3-062, Thermo Scientific)].
RNA isolation, qRT-PCR, and human microarray data
RNA isolation, cDNA synthesis, and qRT-PCR were performed as described previously (34) using a 7300 Real-time PCR system and TaqMan gene expression primer sets from Applied Biosystems, listed in Supplementary Methods. Microarray data from rectus abdominis muscle biopsies from noncancer control patients and patients with PDAC (series GSE130563) was described previously (10). Additional details can be found in Supplementary Methods. Correlative relationships between specific transcripts of interest and clinical variables in these patients are presented in Supplementary Tables S3 and S4.
The methods and solutions used for studies of mouse diaphragm muscle isometric function have been described by us previously (35). Mice were anesthetized using 5% isoflurane. Upon reaching a surgical plane of anesthesia, the diaphragm was excised and immediately placed in a dissecting chamber containing a Krebs-Hensleit solution equilibrated with 95% O2–5% CO2 gas. A muscle strip, including the tendinous attachment at the central tendon and rib cage was dissected from the mid-costal region. The strip was suspended vertically between two lightweight plexiglass clamps with one end connected to an isometric force transducer (dual-mode lever system, Aurora Scientific) within a jacketed tissue bath. The force output was recorded via a computerized data acquisition system (LabView 8.6, National Instruments). After a 15-minute equilibration period, in vitro diaphragm contractile measurements were made. The muscle strip was stimulated along its entire length with platinum wire electrodes (S48 stimulator, Grass Instruments) to determine the optimum contractile length (Lo). Lo was determined by systematically adjusting the length of the muscle and evoking single twitches. To measure the force frequency response, each strip was stimulated supramaximally with 120-V pulses at 15, 30, 60, 100, and 160 Hz while at Lo. The duration of each train was 500 ms to achieve a force plateau. Contractions were separated by a 2-minute recovery period. Diaphragm force production was normalized to specific optimal force. The total muscle cross-sectional area, measured at a right angle to the long axis, was calculated by the following algorithm: total muscle cross-sectional area (mm2) = [muscle mass/(fiber length × 1.056)], where 1.056 is the density of muscle (in g/cm2). Fiber length was expressed in centimeters measure at Lo.
GraphPad Prism statistical software, version 8.0 were used for data analysis. All variables are presented as mean ± SE, unless otherwise stated. Mann–Whitney test was used to compare two groups and Kruskal–Wallis to compare groups of three or more, with Dunn multiple comparisons post hoc test, when appropriate. Two-way ANOVA was used to determine main effects and interactions of two factors, followed by Tukey multiple comparison test to determine differences between individual groups when necessary. In the absence of significant interaction (36), Sidak multiple comparisons post hoc test was used to determine differences within groups for factors showing significant main effects. Univariate and multivariate correlations were performed using Spearman rank-order correlation coefficient. A P value of <0.05 was considered statistically significant.
Myocilin is decreased in multiple models of cancer cachexia
To determine the time course in which tumor burden decreases skeletal muscle expression of Myoc, we harvested TA muscles from sham and C26 tumor-bearing mice at various time points post tumor cell inoculation that correspond to different stages of the muscle atrophy process, which was assessed on days 18 and 23 via MRI imaging (Supplementary Fig. S1A–S1I). TA muscles were subsequently processed for qRT-PCR or Western blot analyses. The magnitude of Myoc downregulation was greatest in muscle of tumor-bearing mice at time points prior to the development of significant muscle atrophy (≤ day 18), yet remained decreased at experimental endpoint (day 26) when mice display significant muscle loss (Fig. 1A–C; refs. 12, 35). We next determined whether the transcriptional downregulation of Myoc in skeletal muscle is a consistent response across other mouse models of cancer cachexia. We thus measured Myoc mRNA at IACUC-mandated experimental endpoints in skeletal muscle of three different cohorts of PDAC-PDX mice (and their respective sham groups), in which PDAC tumor fragments from three different patients were either implanted subcutaneously into the flank (flank model) or directly sutured to the mouse pancreas (orthotopic model) and shown to induce cachexia (24, 25). We also measured Myoc mRNA at experimental endpoint in skeletal muscle of sham mice and mice inoculated with human L3.6pl pancreatic cancer cells subcutaneously into the flank or orthotopically into the pancreas (23). Compared with their respective sham groups, each of the PDAC-PDX and L3.6pl tumor-bearing cohorts showed significantly decreased skeletal muscle expression of Myoc (Fig. 1D and E).
Loss of Myocilin causes muscle fiber atrophy, sarcolemmal fragility, and impaired muscle regeneration following injury
Because the muscle phenotype of mice lacking Myoc (Myoc−/− mice) has not been fully characterized, we first compared the gross morphology of skeletal muscles from adult Myoc−/− mice (Fig. 2A) to WT mice. Compared with WT mice, muscle mass of the TA was marginally reduced in in Myoc−/− mice, while gastrocnemius muscle complex (also known as the triceps surae) mass was not significantly different (Fig. 2B). However, because muscle mass may be a reflection of not only muscle fiber size, but the relative levels of extracellular connective tissue and water, we further stained sections from TA muscles with H&E (Fig. 2C) and quantified muscle fiber CSA. Compared with WT, muscles from Myoc−/− mice displayed visible increases in extracellular space, and significantly decreased (∼18%) muscle fiber CSA (Fig. 2D and E).
Because Myoc has previously been established to stabilize the DGC, which plays a critical role in stabilizing the muscle fiber membrane, we determined whether mice lacking Myoc show sarcolemmal fragility and increased susceptibility to damage induced by downhill running. To do this we injected EBD, which is a membrane impermeable dye, into cage-restricted mice, or into mice subjected to downhill running. While only faint staining of EBD was observed in myofibers of cage-restricted Myoc−/− mice, downhill running induced significant uptake of EBD into myofibers within the diaphragm and limb muscles of Myoc−/− mice, but not WT mice (Fig. 2F and G). Some limb muscles of Myoc−/− mice were so damaged following downhill running that approximately 50% of the total muscle area was positive for EBD, with the majority of myofibers in these areas already degraded and replaced by mononuclear cells. Thus, these data support the notion that Myoc deficiency induces significant sarcolemmal fragility. Importantly, because tumor-bearing mice have significantly reduced levels of Myoc, we further determined whether cachectic C26 tumor-bearing mice also display sarcolemmal fragility. We found that diaphragm muscles of cage-restricted C26 mice showed clear evidence of myofiber uptake of EBD in the absence of downhill running, which was exacerbated in response downhill running (Fig. 2H), indicating significant sarcolemmal fragility and damage of myofibers within the diaphragm. In contrast, hindlimb muscles of C26 mice did not show similar evidence of sarcolemmal fragility and damage, which was assessed via both MRI imaging (Supplementary Fig. S2A and S2B) and EBD uptake assays (Supplementary Fig. S2C and S2D).
We also assessed the ability of TA muscles from Myoc−/− mice to regenerate following muscle injury, because muscle regeneration has been linked with muscle wasting in the LLC and C26 models of cancer cachexia (8, 9). To do this, we injected TA muscles of WT and Myoc−/− mice with CTX and harvested muscles for histologic analyses at multiple time points postinjury (Fig. 3A). Although the morphology of injured muscles from WT and Myoc−/− mice were largely comparable 2 weeks following CTX injury, at later time points Myoc−/− mice showed reduced CSA of regenerating muscle fibers (Fig. 3B and C), coupled with increased extracellular connective tissue (Fig. 3D), indicating an important role for Myoc in the later stages of muscle regeneration.
Because our data indicate that Myoc−/− mice have both increased susceptibility to contraction-induced damage and impaired muscle regeneration, we hypothesized that Myoc deficiency throughout the lifespan could lead to pathologic muscle remodeling. Indeed, pathologic muscle remodeling is a known consequences of repeated cycles of muscle damage, coupled with impaired regeneration (37). To test this hypothesis, we harvested and histologically analyzed TA and diaphragm muscles from 18-month-old WT and Myoc−/− mice using H&E. While reduced muscle quality is expected with the normal aging process, we observed a notable increase in connective tissue deposition within diaphragm muscles of Myoc−/− mice compared with WT mice (Fig. 3E; Supplementary Fig. S3A). Through additional staining of muscle cross-sections with Masson' Trichrome and Oil Red O, we further confirmed the increased connective tissue to be composed of both fibrotic scar tissue (Supplementary Fig. S3B) and fatty tissue deposition (Supplementary Fig. S4). TA muscles of Myoc−/− mice also showed evidence of enhanced fibrotic scarring (Supplementary Fig. S5A and S5B), but were less affected than the diaphragm, which is consistent with findings in models of muscular dystrophy in which the diaphragm is more severely affected than limb muscles (38).
Loss of Myocilin mediates tumor-associated muscle wasting
To determine whether the downregulation of Myoc in skeletal muscle in response to tumor burden plays a causative role in tumor-induced muscle wasting, we injected TA muscles with rAAV9-tMCK-Myoc-GFP (or rAAV9-GFP as control), in which Myoc expression is driven by a muscle-specific (muscle creatine kinase) promoter. Two weeks later, mice were subcutaneously inoculated with C26 tumor cells (or PBS in sham mice), and tissues harvested at experimental endpoint, when mice show significant body wasting (Supplementary Fig. S6A) and muscle wasting (Fig. 4A). Upregulation of MYOC significantly increased muscle mass (normalized to total body mass) in both sham and C26 tumor-bearing mice (Fig. 4A), despite the same tumor burden (Supplementary Fig. S6B). Successful transduction of muscle with AAV vectors was confirmed via measurement of Myoc via qRT-PCR (Fig. 4B), and through visualization of GFP fluorescence in muscle cross-sections (Fig. 4C). Measurement of muscle fiber CSA showed a significant 38% decrease in muscle fiber CSA in TA muscles of tumor-bearing mice transduced with AAV9-GFP (Fig. 4D). In contrast, C26 mice transduced with AAV9-tMCK-MYOC-GFP showed only 21% fiber atrophy (vs. Sham-AAV9-GFP), an approximately 45% sparing of muscle fiber CSA. Despite these findings, no effects of MYOC rescue were observed on the activation of muscle atrophy biomarkers, Fbxo32/atrogin-1/MAFbx, Trim63/MuRF1, or Fbxo30/MUSA1 (Supplementary Fig. S6C). Thus, the partial protection against muscle wasting by MYOC gain-of-function likely occurs independently of pathways that induce protein degradation.
We next determined whether skeletal muscles from mice lacking Myoc exhibit exacerbated muscle wasting in response to tumor burden. To test this, we orthotopically injected the pancreas of WT and Myoc−/− null mice with either Panc02 or Panc02-H7 murine pancreatic cancer cells, which are syngeneic with the C57BL/6 background of Myoc−/− mice. As shown in Kaplan–Meier survival curves in Supplementary Fig. S6D, there was a trend towards decreased survival in Myoc−/− mice versus WT mice bearing orthotopic Panc02/Panc02-H7 tumors (P = 0.0877). Indeed, 7 of 16 tumor-bearing Myoc−/− mice (44%), died, or reached IACUC-mandated endpoint (related to deterioration of body condition score) prior to the expected endpoint of 25 days postsurgery, compared with only 2 of 13 tumor-bearing WT mice (15%). No significant differences were observed in primary tumor mass or in primary tumor-free body mass between WT and Myoc−/− mice (Supplementary Fig. S6E and S6F), which could be related to marked metastasis observed in this model. However, Panc02-H7 tumor burden induced significant loss of gonadal fat mass in both WT and Myoc−/− mice, consistent with cachexia (Supplementary Fig. S6G). Moreover, despite similar tumor-burden, Myoc−/− mice showed significantly greater loss of muscle mass than WT mice (Fig. 4E). Compared with Sham mice, gastrocnemius muscle mass of Myoc−/− mice decreased by 22.9% (P < 0.001) in response to Panc02-H7 tumor burden, while WT mice showed a 10.8% decrease (P < 0.01). A similar trend was evident in the TA muscle (Fig. 4F), but was not significant between genotypes. However, further measurement of muscle fiber CSA in TA muscle cross-sections stained with H&E (Fig. 4G) revealed significant main effects of both Panc02-H7 tumor-burden and Myoc deficiency (Fig. 4H). Similar to our findings with MYOC overexpression, the absence of Myoc did not significantly alter the activation of muscle atrophy biomarkers, despite a trend towards increased levels of Musa1 (Supplementary Fig. S6H–S6J).
Myocilin is transcriptionally regulated by MEF2c
To gain further insight into the mechanisms whereby Myoc is transcriptionally downregulated in response to tumor burden, we analyzed the human and mouse Myoc gene promoters. Despite our previous identification of Myoc as a downstream target of FoxO, we were unable to identify a consensus FoxO binding motif, suggesting FoxO factors likely downregulate Myoc through an indirect mechanism. However, within a conserved regulatory module approximately 200-bp upstream of the TSS, we identified a conserved binding motif for MEF2. Of the Mef2 isoforms expressed in skeletal muscle, Mef2c is significantly decreased at the mRNA (Supplementary Fig. S7A and S7B; refs. 11, 12) and protein level (39) in skeletal muscle of tumor-bearing mice along with several downstream targets of MEF2c (39), thus also supporting reduced MEF2c-dependent transcription. We therefore reasoned that Myoc downregulation in response to tumor burden could be mediated, at least in part, through FoxO-dependent inhibition of MEF2. In support of this, we demonstrate here that transfection of muscle with plasmids encoding constitutively active mutants of FoxO1 or FoxO3a, is sufficient to decrease global MEF2 transcriptional activity (Fig. 5A), and decrease the gene expression of both Myoc and Mef2c (Fig. 5B), the latter of which is also known to be regulated by MEF2 activity (40).
To determine whether a reduction in MEF2c activity alone is sufficient to decrease Myoc transcription, we transfected rodent muscles with an empty vector or d.n. MEF2c expression plasmid and, 4 days later, measured Myoc mRNA. Compared with empty vector, d.n.MEF2c repressed Myoc gene expression by approximately 50%, thus highlighting MEF2c as a key activator of Myoc gene transcription (Fig. 5C). In support of this, mutation of the conserved MEF2-binding site within the Myoc proximal promoter (Fig. 5D) significantly decreased its promoter activity in skeletal muscle, in vivo, as measured via transfection of muscle with luciferase reporter plasmids driven by the Myoc proximal promoter containing either the WT or mutated MEF2 (mMEF2) motif (Fig. 5E).
MEF2c gain of function blocks the cancer-associated downregulation of Myoc and muscle wasting
We next determined whether reduced activity of MEF2c specifically within adult muscle fibers plays a causative role in the C26-induced downregulation of Myoc and mediates muscle loss. To do this, we injected mouse TA muscles with AAV9-GFP or AAV9-tMCK-MEF2c-GFP, and 2 weeks later inoculated mice with C26 cells. Tissues were subsequently harvested at experimental endpoint. Successful targeting of the Mef2c-expressing vector to the TA was confirmed via qRT-PCR (Fig. 5F). We found that MEF2c gain of function in skeletal muscle significantly increased muscle mass across both sham and tumor-bearing groups, which maintained muscle mass in C26 mice to levels comparable with Sham (Fig. 5G). Aligned with this finding, TA muscles transduced with AAV9-tMCK-MEF2c-GFP were significantly protected against C26-induced muscle fiber atrophy (Fig. 5H and I). In support of this, qRT-PCR analyses further demonstrated that maintaining MEF2c activity was sufficient to deter not only the C26-induced downregulation of Myoc (Fig. 5J and K), but significantly blocked the activation of muscle atrophy biomarkers involved in muscle protein breakdown, atrogin-1/MAFbx, and MuRF1/Trim63 (Fig. 5L).
MEF2c gain of function prevents muscle wasting and dysfunction in a preclinical model of PDAC-associated cachexia
While the data generated thus far using the C26 model offer novel insight into mechanisms of cancer-induced muscle wasting, we aimed to determine whether our findings carry over to muscle wasting associated with PDAC, a cancer type that shows high prevalence of cachexia (1). We therefore conducted additional studies using a murine model of PDAC-associated cachexia, involving orthotopic implantation of pancreatic tumor cells isolated from a syngeneic C57BL/6 KRASG12D P53R172H PdxCre+/+ (KPC) mouse, that has been established to recapitulate key features of PDAC-associated cachexia (19). We found that skeletal muscle expression of Myoc was significantly reduced in KPC mice as early as 8 days postsurgery (Fig. 6A), and remained repressed until IACUC-mandated experimental endpoint, which was reached approximately 15–17 days postsurgery. The mRNA for Mef2c was significantly reduced by day 12 postsurgery, and remained decreased until study endpoint. To determine whether MYOC or MEF2c gain of function would provide similar protection against PDAC-associated muscle wasting, as we found in the C26 model, we transduced TA, soleus, and diaphragm muscles of mice with AAV9-GFP, AAV9-tMCK-MYOC, or AAV9-tMCK-MEF2c. Three weeks later, mice were inoculated with KPC cells (or saline for Sham), and tissues harvested on days 13–14 postsurgery, when KPC mice display significant body and muscle wasting (Fig. 6B–D). We found that both MYOC and MEF2c gain of function were protective against KPC-induced loss of muscle mass in the TA (Fig. 6C) and solei (Fig. 6D). Because MEF2c acts upstream of MYOC, and is well-established to regulate additional genes critical to muscle structure and function (39, 41), we further determined whether maintaining MEF2c activity would also provide protection against KPC-induced dysfunction of the diaphragm, which is implicated in cachexia-associated mortality. We found that the decrease in muscle-specific force observed in the diaphragm of KPC mice transduced with GFP was blunted in muscles transduced with MEF2c across multiple force frequencies (Fig. 6E), with maximum specific force significantly higher in diaphragm muscles of KPC mice transduced with MEF2c (Fig. 6F).
MYOC is reduced in patients with PDAC exhibiting cachexia
We next determined whether our findings are relevant to patients with PDAC exhibiting cachexia. To do this, we extracted the normalized levels of MYOC mRNA from a recently published microarray analyses that we performed on rectus abdominis muscle biopsies from n = 16 noncancer control patients and n = 20 patients with PDAC (Supplementary Table S1; ref. 10). Using CT images taken at the level of the third lumbar vertebra from these same patients, we subsequently performed CT-based measurements of skeletal muscle index (SMI) and muscle attenuation (MA) using SliceOmatic software, which are quantitative measures of skeletal muscularity and myosteatosis, respectively (28). From this cohort of patients, CT scans were available from n = 15 control patients and n = 19 patients with PDAC. Using previously established sex- and BMI-specific thresholds (42), patients were further classified as having normal muscularity or muscle depletion (low SMI), and having normal or low MA. We subsequently compared the mRNA levels of MYOC and its upstream regulators FOXO1 and MEF2c, in noncancer control patients with normal muscularity (n = 9) to PDAC patients defined as cachectic based on body weight loss of >8%, in combination with muscle depletion and low MA, which are cachexia thresholds previously established to associate with short survival in patients with cancer (42). Using these cachexia criteria, n = 9 patients with PDAC (47%) were defined as cachectic (Supplementary Table S2). Compared with noncancer controls, patients with cachectic PDAC showed significantly reduced levels of MYOC mRNA (P = 0.0207, Fig. 7A), as well as increased levels of FOXO1 (P = 0.0315, Supplementary Fig. S8A). MEF2c mRNA was not statistically different between groups (P = 0.1903, Supplementary Fig. S8B). We also measured MYOC protein levels, but found high variability within the control group, and no statistical difference between groups (Supplementary Fig. S9A–S9C). We also further examined in patients with PDAC alone, whether the mRNA levels of MYOC are associated with the mRNA levels of FOXO1 and MEF2c, and with clinical parameters including sex, age, cancer-stage, BMI, % BW loss, SMI, MA, and survival time postsurgery using a multivariable correlation matrix (Supplementary Table S3). A significant positive correlation was identified between skeletal muscle MYOC mRNA and survival time postsurgery (P = 0.0008, r = 0.700, Fig. 7B). In support of this, stratification of patients with PDAC based on survival time of >1 year versus <1 year postsurgery revealed significantly lower levels of MYOC in patients surviving <1 year postsurgery (P = 0.005, Fig. 7C). We also identified a positive correlation between the mRNA levels of MYOC and MEF2c (r = 0.4754, P = 0.0397), and a negative correlation between the mRNA levels of FOXO1 and MEF2c (r = −0.4947, P = 0.0266). These data therefore highlight the clinical relevance of our findings in mice, and collectively implicate the dysregulation of a FoxO1–MEF2c–Myoc axis in skeletal muscle of patients with cancer that is associated with cachexia.
In this study, we identify Myoc as a downstream target gene repressed by the FoxO transcription factors whose downregulation precedes and contributes to muscle loss in two different models of cancer cachexia, and that is mediated through reduced activity of MEF2c. We further establish that MEF2c gain of function not only prevents Myoc downregulation in response to tumor burden, but protects against muscle wasting and weakness in a preclinical model of PDAC-associated cachexia. We also demonstrate significant clinical relevance to these findings, showing that MYOC is significantly downregulated in skeletal muscle biopsies from patients with cachectic PDAC who also show increased levels of FoxO1, and correlates with the levels of MEF2c.
Through additional characterization of mice lacking Myoc we demonstrate the biological significance of Myoc in skeletal muscle. In this regard, we show that Myoc deficiency induces muscle fiber atrophy, increased susceptibility of the sarcolemma to damage and impaired muscle regeneration, which are phenotypes implicated in muscle loss associated with cancer, at least in the C26 and LLC models of cachexia (8, 9). Moreover, in aged mice lacking Myoc, we also observed a significant decline in skeletal muscle quality, characterized by increased fatty deposition and fibrotic scar tissue, which are known consequences of chronic damage and impaired regeneration (37). Moreover, although abundant deposition of fat and fibrotic tissue are not phenotypes typically associated with skeletal muscle wasting in rodent models of cancer cachexia, recent work from our lab identified fibrotic remodeling as a feature of cachexia in the orthotopic PDAC-PDX model, which was particularly evident in the diaphragm (43). Moreover, low muscle attenuation (myosteatosis, i.e., fatty muscle) identified via CT imaging is a well-established feature of cachexia in patients with cancer (28), which also indicates reduced muscle quality as a feature of human cancer cachexia. In support of this, increased fat and fibrotic tissue have also been observed histologically in skeletal muscle biopsies from patients with cachectic PDAC (10). However, it is important to note that the reduced muscle quality observed in muscles of aged mice lacking Myoc could be related to the absence of Myoc in not only muscle, but other cell types, because Myoc deletion was not restricted to skeletal muscle. In contrast, in our cancer cachexia studies, MYOC gain-of-function experiments were accomplished using AAV9 and a muscle-specific promoter to selectively upregulate MYOC within muscle fibers only.
Through the studies herein, we further identify MEF2c as a key upstream transcription factor that directly activates Myoc transcription under baseline condition, whose reduced function mediates Myoc downregulation in response to tumor burden. In support of this, previous work has established that the protein expression of MEF2c, and several bonafide downstream target genes of MEF2 are also reduced in skeletal muscle of C26 tumor-bearing hosts (39), which together support reduced MEF2c transcriptional activity. We extend the significance of these findings by demonstrating that MEF2c gain-of-function not only prevents the cancer-associated downregulation of Myoc, but also significantly blocks cancer-induced muscle wasting and dysfunction. Moreover, the protection against muscle loss conferred by MEF2c was also associated with a significant inhibition in the cancer-induced activation of atrogin-1 and MuRF1, thus further linking MEF2c dysfunction with the activation of transcriptional pathways, which promote muscle proteolysis.
MEF2c is a well-characterized transcription factor involved in multiple steps of skeletal muscle development, including myoblast differentiation and fusion, and various processes, which regulate postnatal skeletal muscle homeostasis and plasticity (44–46). In this regard, skeletal muscle expression of MEF2c is necessary for the maintenance of postnatal muscle fiber integrity (41), normal muscle fiber-type specification (47), as well as normal growth and glucose metabolism (48). MEF2c also promotes muscle regeneration and regrowth following injury (49), and when constitutively active, can promote muscle hypertrophy (50). Thus, MEF2c gain of function in skeletal muscle may protect against cancer-induced muscle wasting through its regulation of a wide range of target genes involved in these processes, which warrants further investigation.
The mechanisms whereby MEF2c-dependent transcription is reduced in response to tumor burden are not clear. However, our data indicate that FoxO activation may be involved, as both FoxO1 and FoxO3a are sufficient to decrease Mef2c mRNA and global MEF2 transcriptional activity, and blocking FoxO activation prevents the cancer-induced downregulation of several MEF2c target genes, including Myoc and Mef2c (12). However, MEF2 activity is well-established to be regulated through neural activity (51) and calcium-dependent signaling pathways such as calcium/calmodulin-dependent kinase CaMK) signaling (52), which activate MEF2 through preventing its association with histone deacetylase (HDAC) proteins, which are potent repressors of MEF2-dependent transcription. Because HDAC inhibitors have shown significant efficacy in blocking cancer-induced muscle wasting in the C26 and LLC models (53), exploring the link between HDACs and MEF2 in the context of cancer cachexia warrants further study.
Limitations of this study include the assumptions that mechanisms previously identified using the well-characterized C26 and LLC models of cancer cachexia carry over to more clinically relevant models of cancer cachexia, such as the orthotopic KPC model used herein. In this regard, additional studies are needed to determine whether FoxO activation, DGC disruptions, and impaired regeneration similarly contribute to muscle wasting in the KPC model and other models of PDAC-associated cachexia. The small sample size of patients with PDAC and noncancer controls used for analyses herein is also a significant limitation of the study, as is the possibility of sampling bias, thus highlighting the need to validate our findings using a larger cohort of patients with cancer.
In summary, our data identify MEF2c-dependent transcription of Myoc as a novel transcriptional pathway repressed by FoxO that mediates muscle loss in two different models of cancer cachexia and that is similarly dysregulated in patients with cachectic cancer. These findings thus provide novel insight into mechanisms of cancer-associated muscle wasting and dysfunction that could be further explored as therapeutic targets.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: S.M. Judge, A.R. Judge
Development of methodology: S.M. Chrzanowski, G.A. Walter
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.M. Judge, M.R. Deyhle, D. Neyroud, R.L. Nosacka, A.C. D'Lugos, M.E. Cameron, R.S. Vohra, A.J. Smuder, B.M. Roberts, C.S. Callaway, P.W. Underwood, S.M. Chrzanowski, A. Batra, J.D. Heaven, J.G. Trevino, A.R. Judge
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.M. Judge, M.R. Deyhle, R.L. Nosacka, M.E. Cameron, R.S. Vohra, A.J. Smuder, B.M. Roberts, S.M. Chrzanowski, A. Batra, M.E. Murphy, G.A. Walter, A.R. Judge
Writing, review, and/or revision of the manuscript: S.M. Judge, M.R. Deyhle, D. Neyroud, R.L. Nosacka, A.C. D'Lugos, M.E. Cameron, R.S. Vohra, A.J. Smuder, B.M. Roberts, C.S. Callaway, P.W. Underwood, S.M. Chrzanowski, A. Batra, M.E. Murphy, G.A. Walter, J.G. Trevino, A.R. Judge
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.R. Deyhle, S.M. Chrzanowski, M.E. Murphy
Study supervision: A.R. Judge
This work was supported by the National Institute of Arthritis, Musculoskeletal and Skin Diseases (R01AR060209 to A.R. Judge); the NCI (R21CA194118 to A.R. Judge); the UF Health Cancer Center (bridge funding to A.R. Judge); the UF Clinical and Translational Science Institute (CTSI; pilot award to A.R. Judge and S.M. Judge) [the UF CTSI is supported by the National Center For Advancing Translational Sciences of the NIH (UL1TR001427)]; and the V Foundation for Cancer Research (V2015-021 to J.G. Trevino). D. Neyroud is supported by a Swiss National Science Foundation grant (P4000PM_180814/1). R.L. Nosacka is supported by a National Institute of Child Health and Human Development Grant (T32-HD-043730), and A.C. D'Lugos is supported by a National Heart, Lung and Blood Institute Grant (T32HL134621).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.