Abstract
Facioscapulohumeral muscular dystrophy (FSHD) is characterized by misexpression of the double homeobox 4 (DUX4) developmental transcription factor in mature skeletal muscle, where it is responsible for muscle degeneration. Preventing expression of DUX4 mRNA is a disease-modifying therapeutic strategy with the potential to halt or reverse the course of disease. We previously reported that agonists of the β-2 adrenergic receptor suppress DUX4 expression by activating adenylate cyclase to increase cAMP levels. Efforts to further explore this signaling pathway led to the identification of p38 mitogen-activated protein kinase as a major regulator of DUX4 expression. In vitro experiments demonstrate that clinically advanced p38 inhibitors suppress DUX4 expression in FSHD type 1 and 2 myoblasts and differentiating myocytes in vitro with exquisite potency. Individual small interfering RNA–mediated knockdown of either p38α or p38β suppresses DUX4 expression, demonstrating that each kinase isoform plays a distinct requisite role in activating DUX4. Finally, p38 inhibitors effectively suppress DUX4 expression in a mouse xenograft model of human FSHD gene regulation. These data support the repurposing of existing clinical p38 inhibitors as potential therapeutics for FSHD. The surprise finding that p38α and p38β isoforms each independently contribute to DUX4 expression offers a unique opportunity to explore the utility of p38 isoform-selective inhibitors to balance efficacy and safety in skeletal muscle. We propose p38 inhibition as a disease-modifying therapeutic strategy for FSHD.
SIGNIFICANCE STATEMENT Facioscapulohumeral muscular dystrophy (FSHD) currently has no treatment options. This work provides evidence that repurposing a clinically advanced p38 inhibitor may provide the first disease-modifying drug for FSHD by suppressing toxic DUX4 expression, the root cause of muscle degeneration in this disease.
Introduction
Facioscapulohumeral muscular dystrophy (FSHD) is one of the most prevalent muscular dystrophies (Padberg et al., 1995; Deenen et al., 2014), yet there is currently no treatment available and few clinical trials of promising therapies are ongoing. Thus, there is a desperate need to identify drug targets to stop or reverse progression of the disease. As described in its name, FSHD typically presents as facial, shoulder, and upper arm weakness, which eventually progresses to involve nearly all skeletal muscle groups (Tawil et al., 2014). Although most individuals develop symptoms during their second or third decade of life, a small number have a more acute pediatric onset (Goselink et al., 2017). FSHD is caused by misexpression of the double homeobox 4 (DUX4) transcription factor in skeletal muscle. DUX4 is encoded by a retrogene located within each unit of the D4Z4 macrosatellite repeat array on chromosome 4q35 and is normally expressed in preimplantation embryos, where it activates an early developmental program that marks the cleavage stage of embryogenesis (Tawil et al., 2014; De Iaco et al., 2017; Hendrickson et al., 2017; Whiddon et al., 2017).
In normal skeletal muscle, DUX4 is silenced, likely through repeat-mediated epigenetic repression (van Overveld et al., 2003; Zeng et al., 2009; Snider et al., 2010; Daxinger et al., 2015; Das and Chadwick, 2016). In patients with FSHD, deletion of a subset of D4Z4 repeats [FSHD type 1 (FSHD1)] or damaging variants in epigenetic regulators of the D4Z4 array [FSHD type 2 (FSHD2)] (Lemmers et al., 2010, 2012; van den Boogaard et al., 2016) leads to inefficient D4Z4 repression in somatic cells; when combined with a permissive chromosome 4 haplotype that provides a polymorphic polyadenylation site this results in the ectopic expression of DUX4 in muscle cells (Lemmers et al., 2010; Snider et al., 2010; Tawil et al., 2014). DUX4 misexpression in skeletal muscle causes many cellular phenotypes, ultimately leading to cell death (Winokur et al., 2003; Kowaljow et al., 2007; Bosnakovski et al., 2008; Snider et al., 2009; Wallace et al., 2011; Geng et al., 2012; Young et al., 2013; Feng et al., 2015; Rickard et al., 2015; Shadle et al., 2017).
Because of its causative role in FSHD, targeting DUX4 is an obvious therapeutic approach to halt or reverse disease progression (Himeda et al., 2015). Blocking transcription of DUX4 mRNA with small molecule drugs is an attractive option because all downstream pathologic mechanisms are inherently disrupted and small molecules are generally amenable to oral dosing, avoiding delivery complications associated with RNA interference or gene editing technologies. However, subsequent to chromatin decompaction associated with FSHD, the mechanisms responsible for bursts of DUX4 mRNA synthesis are still poorly understood and limited small molecule drug targets have been identified (Block et al., 2013; Campbell et al., 2017; Teveroni et al., 2017; Cruz et al., 2018).
Previously, we reported on the discovery of β-2 adrenergic receptor agonists as modulators of DUX4 expression (Campbell et al., 2017). Since β2 adrenergic signaling has been extensively characterized in skeletal muscle, we began systematically testing small molecule inhibitors that could potentially interfere with the pathway (for a review, see Joassard et al., 2013). One example of kinases that are activated by β2 adrenergic signaling in a protein kinase A–dependent manner is the set of p38 mitogen-activated protein kinases (MAPKs) (Moule and Denton, 1998; Zheng et al., 2000; Aggeli et al., 2002; McAlees and Sanders, 2009). The p38 MAPKs are classically involved in the cellular response to stressful stimuli, including inflammatory cytokines, and have been heavily pursued by pharmaceutical companies for diseases with inflammatory components, such as rheumatoid arthritis, resulting in an abundance of chemical tools from p38α selective to pan-p38 inhibitors (Xing, 2015). In exploring the hypothesis that activation of p38 plays a role in β2-agonist repression of DUX4, we determined that rather than blocking the ability of β2-agonists to suppress DUX4, p38 inhibitors instead potently inhibit the expression of DUX4 in the absence of β2 agonism. Here, we identify p38α and p38β MAPKs as having important roles in the pathogenic expression of DUX4 in FSHD and demonstrate that clinically advanced p38 inhibitors decrease expression of DUX4 and DUX4 target genes in FSHD patient–derived muscle cells at inhibitor levels that do not negatively affect muscle differentiation. In a mouse pharmacology model of FSHD gene regulation, these inhibitors suppress DUX4 at blood levels that are efficacious in published preclinical models of inflammatory diseases and that have been routinely achieved in human clinical trials. These findings suggest that existing clinical p38 inhibitors are repurposing candidate drugs for FSHD therapeutic development.
Materials and Methods
Ethics Statement.
This research used preexisting deidentified human cell lines from approved repositories. These studies were determined to not be human subject research by the Saint Louis University Institutional Review Board. Primary human myoblast cell lines were obtained from the Fields Center at the University of Rochester (http://www.urmc.rochester.edu/fields-center.aspx) and immortalized by retroviral transduction of cyclin-dependent kinase 4 and human telomerase reverse transcriptase (MB200, FSHD2) (Snider et al, 2010; Stadler et al., 2011). An immortalized FSHD1 cell line (54-2) and isogenic control line from a mosaic male patient (54-6, non-FSHD) were also used (Krom et al., 2012).
Cell Culture.
Immortalized myoblasts were grown in Ham’s F-10 Nutrient Mix (Gibco, Waltham, MA) supplemented with 20% Corning USDA–approved source FBS (Corning, Corning, NY), 100 U/100 μg penicillin/streptomycin (Gibco), 10 ng/ml recombinant human fibroblast growth factor (Promega Corporation, Madison, WI), and 1 μM dexamethasone (Sigma-Aldrich, St. Louis, MO). Differentiation of myoblasts into myotubes was achieved by switching confluent myoblast monolayers into Dulbecco’s modified Eagle’s medium/F-12 Nutrient Mix (1:1; Gibco) supplemented with 2% knockout serum replacement (Gibco), 100 U/100 μg penicillin/streptomycin, 10 μg/ml insulin, and 10 μg/ml transferrin (knockout serum replacement media) for 40 hours.
Compounds.
Individual compounds were purchased from Sigma-Aldrich, Tocris Bioscience (A-485 catalog #6387, BIRB 796 catalog #5989, SB203580 catalog #1202, SCIO 469 catalog #3528, PF-3644022 catalog #4279; Bio-Techne Corporation, Minneapolis, MN), MedChemExpress (PH-797804 catalog #HY-10403, losmapimod catalog #10402, acumapimod catalog #HY-16715, MK@-IN-1 catalog #HY-12834; Monmouth Junction, NJ), or Selleck Chemicals (pexmetinib catalog #S7799, BMS-582949 catalog #S8124, neflamapimod catalog #S1458, pamapimod catalog #S8125, ralimetinib catalog #S1494, SB202190 catalog #S1077, TAK-715 catalog #S2928, VX-702 catalog #S6005, eFT-508 catalog #S8275; Houston, TX), dissolved in 100% DMSO as concentrated stocks and stored at −20°C until use. For in vitro experiments, concentrated DMSO stocks were first diluted in 100% DMSO to 2000-fold the final concentration, and then diluted 2000-fold into culture media before addition to cells. Experiments were performed in triplicate with S.D. represented. For determination of the 50% inhibitory concentration (IC50) for each compound, 11-point concentration response curves were generated by first creating 3-fold serial dilutions of compounds from concentrated stocks in 100% DMSO in 96-well plates. Serial dilutions were then further diluted 2000-fold into culture medium before addition to cells. The IC50 values were determined by nonlinear regression using a four-parameter logistic equation (http://www.graphpad.com; GraphPad Prism Software Inc., San Diego, CA). Data are presented as IC50 values with two significant digits.
DUX4 Activity Assay.
DUX4 activity assay was performed as detailed in our previous studies (Campbell et al., 2017). Briefly, control 54-6 (non-FSHD) myoblasts in one well of a six-well plate were cotransfected with 75 ng of the DUX4 expression vector pCS2-DUX4 (Geng et al., 2012) and 2.925 µg of pGL3-promoter vector (Promega) using Lipofectamine 3000 (Thermo Fisher Scientific, Waltham, MA) following the manufacturer’s instructions. Three hours after transfection, cells were trypsinized and distributed to 42 wells of a 96-well plate. Two hours later (a time at which there is a low but detectable level of DUX4 target gene expression), six wells were harvested to represent the baseline gene expression state, while DMSO control or compounds were added to the remaining wells as indicated in the figures. Twenty hours later, the remaining wells were harvested to represent the endpoint gene expression state. DUX4 activity was determined by normalizing DUX4 target gene mRNA levels at the 25-hour endpoint to the levels at the 5-hour baseline and setting that value to 100 in the absence of drug.
mRNA Expression Analyses.
For screening of compounds in 96-well plates for IC50 determinations and DUX4 activity assay, cell lysates were prepared using Cells-to-Ct Bulk Lysis Reagents (Invitrogen, Carlsbad, CA). For other gene expression analyses, total RNA was extracted from whole cells using the E.Z.N.A. Total RNA Kit or xenograft tissues using the E.Z.N.A. Tissue RNA Kit (OMEGA Bio-Tek, Norcrass, GA) and RNA/DNA was isolated from xenograft tissue using the E.Z.NA. DNA/RNA Kit (OMEGA Bio-Tek) according to the manufacturer’s instructions. Quantitative real-time polymerase chain reaction (PCR) was carried out on a QuantStudio 5 (Applied Biosystems, Foster City, CA). For all gene detection except for DUX4, TaqMan Gene Expression Assays (Applied Biosystems) and TaqMan Fast Virus 1-Step Master Mix (Invitrogen) were used. For DUX4 expression, isolated RNA was treated with DNase I (Thermo Fisher Scientific) and reverse transcribed into cDNA using Superscript IV (Thermo Fisher Scientific) and Oligo (dT) Primer (Invitrogen) following the manufacturer’s protocol. Quantitative real-time PCR was performed using a custom Taqman primer/probe set and TaqMan Gene Expression Master Mix (Applied Biosystems). The relative expression levels of target genes were normalized to that of the reference gene ribosomal protein L30 (RPL30), which was included in multiplex (two gene) PCR reactions, using the ΔΔCt method (Livak and Schmittgen, 2001) after confirming equivalent amplification efficiencies of reference and target molecules.
Small Interfering RNA Transfections.
Duplex small interfering RNAs (siRNAs) were obtained from Thermo Fisher Scientific (Silencer Select). Transfections of siRNAs into myoblasts were carried out, in triplicate, using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer’s instructions. Briefly, cells were seeded at 1 × 105 cells/well in 12-well plates and transfected ∼20 hours later with 2 μl Lipofectamine RNAiMAX and 10 pmol of either gene-specific siRNAs or a scrambled nonsilencing control siRNA diluted in 100 μl Opti-MEM Reduced Serum Medium. For myoblast knockdowns, cells were harvested for RNA analysis 72 hours later. For myotube experiments, 24–48 hours following transfection, cells were switched to differentiation medium and harvested for RNA analysis 40 hours later. Where noted, a double transfection protocol was followed to ensure efficient depletion of preexisting proteins. Cells were transfected a second time 24 hours after the first transfection, switched to differentiation medium 24 hours later, and harvested for RNA analysis 40 hours after switching to differentiation medium.
Western Blotting.
Samples from siRNA knockdown experiments were obtained by lysing cells by addition of SDS-PAGE loading buffer. Reduced and boiled samples were run on NuPage 4–12% Bis-Tris precast polyacrylamide gels (Life Technologies, Carlsbad, CA) and transferred to Immobilon-FL polyvinylidene difluoride membrane (Millipore, Burlington, MA). After blocking in 0.2% I-Block (Thermo Fisher Scientific) in Tris-buffered saline/Tween 20 for 1 hour at room temperature, membranes were incubated with appropriate primary antibodies diluted at 1:1000 in 1X Tris-buffered saline/Tween 20 overnight at 4°C. Membranes were then incubated with Li-Cor near-infrared fluorescently labeled secondary antibodies diluted at 1:15,000 in 1X Tris-buffered saline/Tween 20 for 1 hour at room temperature. Blots were scanned and analyzed with the Li-Cor Odyssey CLx Imaging System (Li-Cor Biosciences, Lincoln, NE). Quantitation of bands was normalized to α-tubulin using Image Studio software (Li-Cor Biosciences).
Antibodies.
The following antibodies were used: α-tubulin mouse mAb (926-42213; Li-Cor Biosciences); p38α MAPK polyclonal Rabbit Ab (9218S; Cell Signaling Technology, Danvers, MA); p38β MAPK (C28C2) Rabbit mAb (2339; Cell Signaling Technology); IRDye 680 goat anti-mouse secondary (926-68070; Li-Cor Biosciences); and IRDye 800 goat anti-rabbit secondary (926-32211; Li-Cor Biosciences).
Animals.
Male NOD-Rag immunodeficient mice (strain #007799 NOD.CgRag1tm1Mom Il2rgtm1Wjl/SzJ; Jackson Laboratories) were used for the xenograft model of FSHD because the absence of T, B, and NK cells in this strain makes them suitable for xenograft transplantation (Silva-Barbosa et al., 2005). All protocols were approved by the Institutional Animal Care and Use Committee of Saint Louis University.
Plasma Drug Levels.
Terminal plasma samples were diluted with control naive mouse plasma as appropriate to bring samples into the standard curve range (1–1000 ng/ml) or were run undiluted. Enalapril was used as an internal standard prior to extraction at a 100 ng/ml final concentration. Samples were vortexed for 5 minutes to fully incorporate the internal standard. Afterward, 180 ml of acetonitrile was added, vortexed for 5 minutes, and centrifuged for 5 minutes at 4°C and greater than 2100g. The supernatant was transferred to a 96-well sample plate and heat sealed for liquid chromatography–tandem mass spectrometry analysis using a system consisting of an LC-20AD pump (Shimadzu, Kyoto, Japan), an HTC PAL autosampler (Leap Technologies, Carrboro, NC), and a Sciex API-4000 mass spectrometer in electrospray ionization mode (AB Sciex, Foster City, CA). The multiple reaction monitoring transitions for losmapimod were m/z: 383 > 267. An Amour C18 reverse-phase column (2.1 × 30 mm, 5 μm; Analytical Sales and Services, Pompton Plains, NJ) was used for chromatographic separation. Mobile phases were 0.1% formic acid (aqueous) and 0.1% formic acid in acetonitrile (organic) with a flow rate of 0.35 ml/min. The starting phase was 10% acetonitrile for 0.9 minutes, increased to 90% acetonitrile over 0.4 minutes, maintained for an additional 0.2 minutes, returned to 10% acetonitrile over 0.4 minutes, and then held for 2 minutes. Peak areas were integrated using Analyst 1.5.1 (AB Sciex).
Cell and Barium Chloride Preparations for Injection.
Barium chloride was dissolved in 0.9% saline to a concentration of 2.4% w/v. The barium chloride solution was then sterile filtered (0.2 μm filter; Thermo Scientific). Growing FSHD myoblasts were detached with 0.25% trypsin 2.21 mM EDTA (Corning) and resuspended in growth media for enumeration. Cells were then centrifuged at 900g for 5 minutes and suspended in 30 ml of sterile saline. This saline rinse was repeated one more time before cells were suspended in saline at ∼6.67 × 107 cells/ml. The cells were stored on ice and injected within 1 to 2 hours of preparation. Just prior to injection, cells were combined with an equal volume of barium chloride solution yielding 1.2% w/v barium chloride solution with 3.33 × 107 cells/ml.
Barium Chloride Xenograft Pharmacology Model of FSHD.
NOD-Rag immunodeficient mice were anesthetized with 3%–3.5% isoflurane to effect and the injection site was shaved and cleaned with betadine. Barium chloride cell suspensions were injected using a 26-gauge needle into three sites along the tibialis anterior muscle. Injections were 10 μl/site in volume amounting to 30 μl and 1 × 106 cells injected in total. Mice recovered for 1 to 2 hours prior to administration of test compounds. PH-797804 and losmapimod were dissolved in DMSO prior to addition of the remaining vehicle components. PH-797804 was administered subcutaneously twice a day at 5 ml/kg in 10% DMSO, 40% polyethylene glycol, and 50% saline vehicle. Losmapimod was administered by mouth twice a day at 10 ml/kg in 10% DMSO and 90% 0.5% methylcellulose. At the termination of the study, mice were euthanized via CO2 asphyxiation and blood and tissue samples were collected for bioanalytical or quantitative PCR analysis. For FSHD endpoints, the entire xenograft muscle was harvested, weighed, and homogenized in lysis buffer as described in mRNA Expression Analyses. Lysis buffer was then transferred such that 30 mg of tissue was used in the RNA isolation procedure and 10 mg of tissue was used in the DNA/RNA procedure to prevent column clogging.
Taqman Gene Expression Assay Identification Numbers.
Taqman assays were purchased from Applied Biosystems (Thermo Fisher Scientific): MBD3L2, Hs00544743_m1; LEUTX, Hs01028718_m1; MYH2, Hs00430042_m1; MAPK11, Hs00177101_m1; MAPK14, Hs01051152_m1; MYOG, Hs01072232_m1; RPL30, Hs00265497_m1; ZSCAN4, Hs00537549_m1; and DUX4, primers GCCGGCCCAGGTACCA and CAGCGAGCTCCCTTGCA with probe 6FAM-CAGTGCGCACCCCG-MGBNFQ.
siRNA Assay Identifications and Target Sequences.
All siRNAs were purchased with Silencer Select Chemistry (Ambion; Life Technologies): MAPK14, s3585, CCTAAAACCTAGTAATCT; MAPK11, s11155, GCGACTACATTGACCAGCT; and negative control: Silencer Select Negative Control #1.
Statistical Analysis.
All statistics were performed using GraphPad 4 software. One-way ANOVA was used for multigroup comparisons; Dunnett’s post test was used to compare individual groups to the control. An unpaired two-tailed t test was used for two sample comparison. The P values and statistical comparison used are listed in each figure caption.
Results
p38 Inhibitors Suppress DUX4 in FSHD Myotubes and Myoblasts.
We have previously screened several small molecule compound libraries to identify bromodomain and extra-terminal motif inhibitors and agonists of the β-2 adrenergic receptor as inhibitors of DUX4 expression in FSHD muscle cells (Campbell et al., 2017). Efforts to further explore the signaling pathways regulating DUX4 expression in FSHD led us to screen p38 MAPK inhibitors in a high throughput assay for DUX4. Detection of DUX4 mRNA in FSHD muscle cells for drug screening purposes is challenging for many reasons (Snider et al., 2009, 2010; Geng et al., 2012; Feng et al., 2015). Therefore, DUX4 was measured indirectly by quantitating mRNA levels for the DUX4-regulated gene MBD3L2, which is a sensitive and highly specific marker of DUX4 expression since it is not expressed in normal muscle cultures or tissue (Geng et al., 2012; Yao et al., 2014; Campbell et al., 2017). Differentiating cultures of MB200 (FSHD2) myoblasts that express elevated levels of DUX4 and DUX4-target gene mRNA upon differentiation into multinucleated myotubes (myocytes) were used. Figure 1A shows the concentration-response curve for MBD3L2 RNA levels in MB200 (FSHD2) myotubes differentiated in the presence of the clinical p38α/β inhibitor PH-797804 for 40 hours. The levels of MBD3L2 RNA induced during differentiation are suppressed >95% in a concentration-dependent manner with a potent IC50 value of 0.41 nM, in line with the reported IC50 value of 2.3 nM for PH-797804 on p38α enzymatic activity (Selness et al., 2011). Since the p38 family of kinases is known to play important roles in the myogenic program (Wu et al., 2000), it was important to distinguish effects on DUX4 expression from effects on myocyte differentiation. We noted that only at high concentrations of PH-797804 (>100 nM) was there a noticeable delay in myotube formation. Key markers of myocyte differentiation (myogenin, MYOG, and myosin heavy chain, MYH2) were likewise only inhibited at concentrations greater than 100 nM, and even at 1–10 μM the inhibition was incomplete, reaching a maximum of ∼60% inhibition (Fig. 1A). To further demonstrate that the effects of p38 inhibition on DUX4 target gene expression were independent of myocyte differentiation, cultures of proliferating FSHD myoblasts were treated with PH-797804. Figure 1B shows that multiple DUX4 targets are suppressed with treatment in both 54-2 (FSHD1) and MB200 (FSHD2) proliferating myoblast cultures, consistent with suppression of DUX4 expression being independent of the differentiation state of the cells. To confirm that p38 inhibition by PH-797804 had a direct effect on DUX4, we performed experiments in both FSHD1 and FSHD2 myocytes treated with several drug concentrations for 40 hours during differentiation. Figure 1C (54-2, FSHD1) and Fig. 1D (MB200, FSHD2) demonstrate that p38 inhibitor PH-797804 treatment reduces RNA levels for DUX4 and DUX4 targets (MBD3L2, ZSCAN4, and LEUTX) at drug concentrations that have negligible effects on markers of differentiation MYOG and MYH2. These data confirm that p38 inhibition directly suppresses DUX4 expression, and consequently the expression of DUX4 target genes is similarly reduced.
To determine if PH-797804 was representative of the class of p38 inhibitors, we screened 12 commercially available p38 inhibitors that had reached clinical testing and two commonly cited chemical probes in differentiating FSHD myocytes. All p38 inhibitors were active with concentration-response curves similar to those of PH-797804, exhibiting an approximate 3-fold order of magnitude difference in their ability to suppress DUX4 target MBD3L2 expression versus having an effect on differentiation markers (Supplemental Fig. 1). Inhibitors were similarly active in 54-2 (FSHD1) and MB200 (FSHD2) myocytes with IC50 values reported in Table 1, suggesting that DUX4 expression is exquisitely sensitive to p38 inhibition in both genetic backgrounds. Interestingly, inhibitors of the p38 inflammatory pathway target MAPK-activated protein kinase 2 (MAPKAPK2, MK2) (Mourey et al., 2010) were not active at suppressing DUX4, nor were inhibitors of MAPK-interacting serine/threonine kinase 1/2 (MKNK1/2, MNK1/2) (Reich et al., 2018) (Table 1).
DUX4 Activity.
DUX4 protein is likely a target for post-translational modifications, including acetylation and phosphorylation, which could affect its function. It was possible that p38 inhibition not only suppressed the transcription of DUX4 mRNA as shown previously, but also affected the transcriptional activation function of DUX4 by altering its post-translational modifications. To test this possibility, we used forced expression of DUX4 in normal skeletal muscle myoblasts to determine if p38 inhibition would alter the ability of transiently expressed DUX4 to induce expression of its target genes. The 54-6 (non-FSHD) myoblasts were transfected with the pCS2-DUX4 plasmid (Geng et al., 2012), which drives DUX4 expression from the cytomegalovirus promoter, and then treated with various concentrations of two different p38 inhibitors for 20 hours. Since DUX4 protein interacts with the transcriptional coactivator p300 and utilizes its acetyltransferase activity to induce expression of many DUX4 target genes (Choi et al., 2016), we used the p300 acetyltransferase activity inhibitor A-485 (Lasko et al., 2017) as a control. Figure 2A shows that DUX4-induced increases in targets MBD3L2 and LEUTX are reduced by >95% in a concentration-dependent manner by A-485, demonstrating the utility of this assay in measuring DUX4 activity. Interestingly, DUX4 target ZSCAN4 mRNA levels were not affected by A-485 treatment (Fig. 2A), suggesting that DUX4 activates some target genes independent of p300 acetyltransferase activity. When transfected cells are exposed to PH-797804 during this time frame, there is a slight decrease in DUX4 targets (∼35%–40% maximal inhibition); however, significant transcriptional activation activity remains that is insensitive to p38 inhibition (Fig. 2B). Additionally, the p38α-biased inhibitor pamapimod (Hill et al., 2008) did not block DUX4-induced target gene expression (Fig. 2C). These data indicate that p38 inhibition has a primary activity of suppressing DUX4 expression and potentially minor activity of partially reducing DUX4 transactivation function.
siRNA Targeting p38α and p38β.
The majority of p38 inhibitors to enter clinical testing are p38α/p38β isoform-selective inhibitors with several having bias toward p38α (for reviews, see Goldstein and Gabriel, 2005; Yong et al., 2009; Norman, 2015). An exception is BIRB796, which inhibits the α, β, γ, and δ isoforms (Kuma et al., 2005). The data presented previously suggested that either p38α, p38β, or both isoforms play a role in promoting DUX4 expression in FSHD myoblasts and differentiating myocytes. The consistency of inhibition of DUX4 target genes (Table 1) appeared to track with the potency of the drugs toward the p38α isoform. To determine the relative roles of each isoform, we used siRNAs to knock down individual isoforms in FSHD1 and FSHD2 myoblasts and myotubes. Figure 3A shows that in proliferating 54-2 (FSHD1) myoblast cultures, siRNAs targeting p38α or p38β, individually or in combination, effectively and selectively decrease levels of their respective target mRNAs (left panel). Interestingly, individual knockdown of p38α mRNA resulted in increased levels of p38β mRNA and individual knockdown of p38β mRNA resulted in increased levels of p38α mRNA (Fig. 3A, left panel). Since DUX4 expression in myoblasts is difficult to measure, we looked at the effects of p38 knockdown on DUX4 target gene expression. Individual knockdown of each isoform or combined knockdown of both resulted in decreased expression of multiple DUX4 target genes (Fig. 3A, right panel), suggesting that both isoforms are individually required to support DUX4 mRNA expression. Similar results were obtained using MB200 (FSHD2) myoblasts (Fig. 3B, left and right panels). These results are surprising because instead of having redundant roles, p38α and p38β appear to have independent roles in supporting DUX4 mRNA synthesis. We extended these experiments to differentiating FSHD myotubes where DUX4 expression is increased during the differentiation process (Snider et al., 2010; Jones et al., 2012; Krom et al., 2012; Tassin et al., 2013). For these experiments, confluent monolayers of myoblasts were transfected with siRNAs 1 or 2 days prior to differentiation to allow for efficient depletion of target proteins. Differentiation was then induced for 40 hours until multinucleated myotubes predominated in the cultures. Similar to results in myoblasts, siRNAs targeting p38α or p38β, individually or in combination, effectively decrease levels of their respective target mRNA in 54-2 (FSHD1) and MB200 (FSHD2) myotubes (Fig. 4, A and B, respectively, left panels). Western analysis confirmed that p38α protein levels were decreased only when siRNA targeting p38α was used alone or in combination with siRNA targeting p38β (Supplemental Fig. 2). Similarly, p38β protein levels were decreased only when siRNA targeting p38β was used alone or in combination with siRNA targeting p38α (Supplemental Fig. 2). In myotubes, knockdown of p38β increased the expression of p38α at the mRNA and protein levels, while knockdown of p38α did not significantly affect p38β levels. Inspection of quantitative PCR cycle times indicated that p38β is expressed much less abundantly than p38α in our FSHD myotube cultures at the mRNA level. Nonetheless, individual knockdown of each p38 isoform alone or combined knockdown of both resulted in decreased expression of DUX4 and multiple DUX4 target genes in FSHD1 and FSHD2 myotubes (Fig. 4, A and B, respectively, right panels), although the decrease in DUX4 did not reach significance with individual knockdown of p38α in 54-2 (FSHD1) myotubes. It is worth noting that knockdown of p38β decreases DUX4 expression even as p38α levels are increased and that combined knockdown of both isoforms does not decrease DUX4 expression below the level generated with knockdown of p38β alone. Additionally, knockdown of the p38γ isoform did not affect DUX4 expression (Supplemental Fig. 3). These combined data in FSHD myoblasts and myotubes suggest that p38α and p38β are both necessary and play nonredundant roles in supporting pathogenic DUX4 expression in FSHD muscle cells.
p38 Inhibitors Are Active in a Xenograft Model of FSHD.
To test the ability of clinically advanced p38 inhibitors to suppress DUX4 expression in vivo, we developed a pharmacology model of human FSHD gene regulation. The primate-specific organization of the DUX4 gene imbedded within D4Z4 repeats (Clapp et al., 2007; Leidenroth et al., 2012) and the loss of epigenetic repression of DUX4 within the context of human genetic mutations represent barriers to the development of a transgenic mouse model that recapitulates DUX4 misexpression in FSHD. To most accurately model FSHD epigenetic dysregulation of DUX4, we used a mouse xenograft approach in immunodeficient mice in which tibialis anterior muscles are injured by barium chloride injection to induce a regenerative response (Hardy et al., 2016). The regenerating muscle environment is conducive to engraftment of transplanted human myoblasts that respond to cues from the microenvironment, express appropriate markers of differentiation, and become part of mature myofibers (Silva-Barbosa et al., 2005). Variations of this model have been used to demonstrate engraftment and differentiation of human FSHD patient–derived myoblasts or muscle biopsies (Krom et al., 2012; Chen et al., 2016; Sakellariou et al., 2016). To treat FSHD xenograft mice with p38 inhibitors to suppress DUX4, it was necessary to determine the time frame of DUX4 expression after xenotransplantation. Since it is well established that DUX4 is induced upon differentiation of myoblasts into multinucleated myotubes in vitro in the time frame of 1.5 to 6 days (Snider et al., 2010; Jones et al., 2012; Krom et al., 2012; Tassin et al., 2013; Balog et al., 2015) and that the corresponding elevated levels of DUX4 protein are sufficient to cause myotube death (Block et al., 2013; Rickard et al., 2015), we profiled DUX4 expression during the first week after xenotransplantation of MB200 (FSHD2) myoblasts. Total RNA from xenograft mice tibialis anterior muscles was analyzed using human-specific primers and Taqman probes for DUX4 and DUX4 target gene RNA levels by quantitative real-time PCR. Supplemental Fig. 4 demonstrates that DUX4 mRNA levels increase over the first few days and peak at about the fourth day after xenotransplantation (left panel). RNA levels for representative DUX4 target gene MBD3L2 peak at days 5 to 6 and begin to decline by day 7 (Supplemental Fig. 2, right panel). These data suggest that DUX4 expression is induced as transplanted FSHD myoblasts begin to differentiate in vivo and that this process peaks within the first week.
We reasoned that treating animals for 4 days, starting immediately after the xenograft procedure through the peak of DUX4 expression, was appropriate to evaluate agents intended to suppress DUX4 expression in FSHD. We treated xenograft mice systemically with two different inhibitors: PH-797804 and losmapimod. Figure 5A shows that treatment of xenograft mice with PH-797804 by subcutaneous injection twice daily resulted in dose-dependent decreases in mRNA levels for DUX4 (left panel) and DUX4 targets MBD3L2, ZSCAN4, and LEUTX (right three panels, respectively). The highest dose of PH-797804 (5 mg/kg b.wt.) elicited a decrease of approximately 80% for DUX4 and its targets compared with vehicle. Similarly, treatment of xenograft mice with losmapimod orally twice daily resulted in dose-dependent decreases in DUX4 and DUX4 target mRNA levels (Fig. 5B). In this case, mRNA levels for DUX4 and its targets were decreased by approximately 80% with losmapimod dosed at 6 mg/kg twice daily and by greater than 90% at a dose of 18 mg/kg twice daily. In preclinical models of inflammation, losmapimod has been efficacious at a dose of 12 mg/kg per day in food (Willette et al., 2009; Yeung et al., 2018). To understand how suppression of DUX4 was related to plasma drug levels, we measured trough levels of losmapimod in plasma at sacrifice (∼14 hours after last dose). Supplemental Fig. 6 shows that terminal (trough) plasma levels of losmapimod in the 6 mg/kg dosing group were still above the in vitro IC50 values of losmapimod (average of 16 nM in plasma vs. an IC50 of 2.4 nM in MB200 cells) (see Table 1).
To further assess the potential utility of losmapimod in FSHD and the effects of p38 inhibitor treatment on the differentiation of transplanted FSHD myoblasts, xenograft mice were treated for 14 days with losmapimod dosed at 6 mg/kg twice daily. Muscle RNA was analyzed for the late differentiation marker myosin heavy chain (MYH2) while terminal plasma samples were used to measure trough drug levels that are reported in Supplemental Fig. 6B. Figure 5C shows that compared with the low level present 4 days after xenotransplantation, levels of MYH2 RNA increase dramatically in mice treated with vehicle or losmapimod over the period during which fully formed mouse myofibers develop, between 4 and 14 days after xenotransplantation (left panel) (Hardy et al., 2016). These data indicate treatment of xenograft mice with losmapimod at a level that represses DUX4 by 80% is compatible with robust differentiation of the introduced human FSHD myoblasts. Since DUX4 expression during the differentiation process is sufficient to cause cell death (Block et al., 2013; Rickard et al., 2015), we also looked at the survival of human FSHD cells by measuring the human cell DNA copy number relative to mouse cell DNA after 14 days of losmapimod treatment. Fig. 5C (right panel) shows that xenograft mice treated with losmapimod had an approximately 2-fold increase in human cell copy number relative to mouse. Analysis of the PCR cycle times for the human TERT gene suggests an even greater absolute increase in human cell number per xenograft (5.5-fold) (Supplemental Fig. 5) with a 2.7-fold increase in mouse cell number per xenograft, although this analysis is semiquantitative. These data suggest that p38 treatment suppresses DUX4 expression and promotes the survival and incorporation of differentiating human FSHD cells in regenerating mouse muscle.
Discussion
FSHD is one of the most prevalent muscular dystrophies, yet development of disease-modifying therapies lags far behind that for other dystrophies. The consensus identification of DUX4 protein expression in skeletal muscle as causing the disease has given a clear therapeutic target. Approaches include targeting the DUX4 mRNA with antisense oligonucleotide technology and targeting DUX4 transcriptional activity or the activity of gene products downstream from DUX4 (Himeda et al., 2015; Banerji et al., 2018). These approaches have challenges to overcome, such as muscle delivery for antisense oligonucleotide–based therapeutics. Furthermore, there is incomplete understanding of the pathophysiology of FSHD in relation to the downstream gene targets of DUX4 that mediate its toxic effects. We have focused instead on identifying small molecule drugs capable of suppressing the transcription of DUX4 mRNA such that no DUX4 protein is produced, an approach that bypasses many of these challenges.
In this study, we demonstrated that DUX4 mRNA synthesis in FSHD myoblasts and myotubes is exquisitely sensitive to p38α/β inhibitors. The IC50 vaues shown in Table I are in line with published activities determined for isolated p38α or p38β enzymes (Xing, 2015), strongly suggesting that DUX4 expression is positively regulated by p38α or p38 α/β kinase activity. One desirable characteristic of a drug intended for degenerative muscular dystrophy is that the treatment does not negatively affect the ability of muscle cells to participate in the muscle regenerative process. The suppression of DUX4 by p38 inhibitors is independent of myocyte differentiation, evidenced by the activity of inhibitors in proliferating myoblasts. Furthermore, in differentiating myotubes DUX4 is suppressed at drug concentrations that do not affect markers of differentiation. In fact, decreases in markers of muscle differentiation (MYOG and MHY2) require three orders of magnitude higher drug concentrations than for DUX4 suppression. Even at high drug concentrations, MYOG and MHY2 decreases are not complete and may be an artifact of in vitro differentiation. Systemic administration of losmapimod was able to decrease DUX4 expression and its downstream targets without decreasing maturation markers after chronic dosing over 14 days.
The sensitivity of DUX4 in differentiating muscle cells to p38 inhibition at drug concentrations that do not prevent differentiation is surprising given the literature describing the roles of p38 in skeletal muscle. Four p38 isoforms exist in mammals: p38α, p38β, p38γ, and p38δ (Cuenda and Rousseau, 2007; Cuadrado and Nebreda, 2010). The α, β, and γ isoforms are expressed in skeletal muscle (Wang et al., 2008), and p38α involvement in regulating the skeletal muscle differentiation process is well documented. Activation of p38α/β has been shown to be important in regulating muscle progenitor satellite cell activation from quiescence and reestablishment of the satellite cell pool (Troy et al., 2012). The latter mechanism led to the proposal of using p38 inhibition to help reestablish the muscle stem cell pool in situations where this pool is depleted, such as in aged skeletal muscle (Bernet et al., 2014). Also, p38 has been implicated in key temporally ordered events involving myogenic regulatory transcription factors and the establishment of muscle-specific gene expression (Zetser et al., 1999; Wu et al., 2000; Bergstrom et al., 2002; Penn et al., 2004). Importantly, p38 is involved in several epigenetic mechanisms including recruitment of methyltransferase and nucleosome remodeling complexes to muscle-specific gene promoters (Simone et al., 2004; Rampalli et al., 2007) and inactivation of methyltransferases at specific genes (Chatterjee et al., 2016). In fact, p38α binds to and regulates many promoters during myogenesis (Segalés et al., 2016).
How can inhibition of p38α/β selectively reduce DUX4 without disrupting muscle differentiation? Several observations suggest this is possible. First, DUX4 expression is exquisitely sensitive to p38 inhibition, with much less inhibitor required than in the case of blocking differentiation. For example, SB203580 (Kumar et al., 1997) inhibited DUX4 in differentiating FSHD myocytes with IC50 values of 17 nM (FSHD1) and 9.8 nM (FSHD2), while muscle differentiation gene markers were partially inhibited at IC50 values >1500 nM (Supplemental Fig. 1). Second, mice with muscle-specific deletion of p38α grow and function normally, albeit with smaller muscles and delayed myofiber growth and maturation (Brien et al., 2013). Additionally, muscle-specific p38α knockout mice exhibit an increase in the muscle satellite cell pool, consistent with the previously mentioned role of p38 in regulating satellite cell activation and restricting excess myoblast proliferation (Troy et al., 2012) and the proposal that p38 inhibition may be a therapeutic option in muscle degenerative disorders in which the stem cell niche appears to be diminished (Bernet et al., 2014; Dumont and Rudnicki, 2016). Notably, when we treated xenograft mice for 14 days with the p38 inhibitor losmapimod, the total number of both human and mouse cells increased compared with nontreated animals (Fig. 5; Supplemental Fig. 5), consistent with p38 inhibition promoting myoblast proliferation even as the expression of a late differentiation marker (MYH2) was induced and maintained.
The molecular mechanisms that tie DUX4 expression to p38 activity remain to be elucidated. A surprise finding from these studies is that DUX4 expression is dependent on both α and β p38 isoforms (Figs. 3 and 4). The result that simultaneously knocking down both isoforms does not suppress DUX4 more than solely knocking down p38β implies that each kinase plays a distinct mechanistic role such that when p38β is knocked down, DUX4 is suppressed even when p38α levels are increased. It should be noted that p38β is generally expressed at lower levels than the other isoforms (Cuadrado and Nebreda, 2010) and appears to be dispensable for muscle differentiation (Perdiguero et al., 2007; Ruiz-Bonilla et al., 2008). This offers an opportunity to explore the utility of p38 isoform-selective inhibitors to balance efficacy and safety. Due to its role in inflammation (Allen et al., 2000), previous efforts have focused on developing p38α inhibitors for various inflammatory diseases (Kumar et al., 2003). Most of these inhibitors also inhibit p38β due to the high sequence similarity between the closely related proteins (Kondoh et al., 2016), and no p38β-selective inhibitors have been described. Given the roles of p38α in muscle differentiation as well as its role in inflammation (Page et al., 2010) it is worth pursuing the development of p38β-selective inhibitors. Interestingly, p38β inhibition has independently been suggested as a therapeutic strategy for muscle cachexia (Ding et al., 2017).
Since DUX4 is a transcription factor that is likely regulated by post-translational modifications, it was important to determine if p38 inhibition had any effect on the transcriptional activation function of DUX4. We used the p300 acetyltransferase activity inhibitor A-485 as a control since DUX4 has been reported to interact with p300 to activate target genes (Choi et al., 2016). Indeed, A-485 inhibited induction of DUX4 targets MBD3L2 and LEUTX. Interestingly, ZSCAN4 RNA levels were not decreased by A-485, suggesting that this DUX4 target is induced independent of p300 acetyltransferase activity, even though DUX4 has been shown to induce p300-dependent acetylation of histones in the vicinity of the ZSCAN4 DUX4 binding sites (Choi et al., 2016). The significance of this finding remains to be explored. In the case of p38 inhibitors, there are minor and inconsistent decreases in some DUX4 targets by p38 inhibitors. It is difficult to conclude from these data if p38 plays any role in DUX4 activity or if the small decreases are assay artifacts due to potential interference of cytomegalovirus promoter by p38 inhibitors (Bruening et al., 1998). Nonetheless, high concentrations of p38 inhibitors do not block DUX4 activity and the primary effect of p38 inhibition in FSHD cells is to suppress the synthesis of DUX4 mRNA.
The p38 inhibitors suppress the expression of DUX4 and its targets at clinically relevant doses. It is clear from in vivo testing of two clinically advanced inhibitors that p38 inhibition effectively reduces the peak levels of DUX4 expression in a dose-dependent manner (Fig. 5). The doses of 5 mg/kg twice a day for PH-797804 and 6 mg/kg twice a day for losmapimod reduce DUX4 and DUX4 target RNA levels by approximately 80%. These dosing levels have been used in translational animal models (Willette et al., 2009; Xing et al., 2012). Losmapimod has been extensively characterized in humans. Efficacy and safety studies have used repeated dosing with losmapimod at 7.5 mg, twice daily (O’Donoghue et al., 2016; Fisk et al., 2018); a detailed study of drug exposure demonstrated a stable plasma trough concentration of around 16 ng/ml (39 nM) (Ino et al., 2015). Losmapimod was efficacious in our xenograft mouse model with a terminal drug concentration in the plasma of 16 nM. These data suggest there is sufficient muscle exposure by dosing losmapimod at 7.5 mg twice a day to substantially reduce DUX4 in FSHD patient muscle. A more detailed analysis of pharmacokinetic and pharmacodynamic relationships, specifically in muscle tissue, will enable accurate prediction of human dose requirements.
To conclude, the data presented here support the repurposing of existing p38α or dual p38α/β inhibitors as potential therapeutics to suppress DUX4 expression in FSHD and suggests the development of p38β-selective inhibitors to perhaps more specifically target DUX4. These studies uncover the exciting possibility of a disease-modifying treatment for FSHD by targeting the root cause of disease through p38 MAPK inhibition.
Acknowledgments
We thank Silvère M. van der Maarel for critical reading of the manuscript, Nikolaos Atkinson for technical assistance and screening of compounds, and Mary Campbell and Stacey Arnett for PK analysis and dose selection for PH-797804.
Authorship Contributions
Participated in research design: Oliva, Galasinski, Campbell, Meyers, Tapscott, Sverdrup.
Conducted experiments: Oliva, Richey, Campbell, Modi, Zhong, Sverdrup.
Contributed new reagents or analytic tools: Tawil.
Performed data analysis: Oliva, Richey, Campbell, Sverdrup.
Wrote or contributed to the writing of the manuscript: Oliva, Galasinski, Campbell, Tapscott, Sverdrup.
Footnotes
- Received May 1, 2019.
- Accepted June 10, 2019.
This work was supported by the National Institutes of Health National Institute of Neurologic Disorders and Stroke [Grant NS069539]; the Muscular Dystrophy Association [Grant 576054]; and Ultragenyx Pharmaceutical, Inc.
↵This article has supplemental material available at jpet.aspetjournals.org.
Abbreviations
- DUX4
- double homeobox 4
- FSHD
- facioscapulohumeral muscular dystrophy
- FSHD1
- facioscapulohumeral muscular dystrophy type 1
- FSHD2
- facioscapulohumeral muscular dystrophy type 2
- MAPK
- mitogen-activated protein kinase
- PCR
- polymerase chain reaction
- siRNA
- small interfering RNA
- Copyright © 2019 by The American Society for Pharmacology and Experimental Therapeutics