Disease-Proportional Proteasomal Degradation of Missense Dystrophins

Disease-proportional proteasomal degradation of missense dystrophins

Dana M. Talsness, Joseph J. Belanto, and James M. Ervasti1

Department of Biochemistry, Molecular Biology, and Biophysics, University of Minnesota–Twin Cities, Minneapolis, MN 55455

Edited by Louis M. Kunkel, Children’s Hospital Boston, Harvard Medical School, Boston, MA, and approved September 1, 2015 (received for review May 5, 2015) The 427-kDa protein dystrophin is expressed in striated muscle insertions or deletions (indels) represent ∼7% of the total DMD/ where it physically links the interior of muscle fibers to the BMD population (13). When indel mutations cause a frameshift, they extracellular matrix. A range of mutations in the DMD gene encod- can specifically be targeted by current exon-skipping strategies (15). ing dystrophin lead to a severe muscular dystrophy known as Du- Patients with missense mutations account for only a small percentage chenne (DMD) or a typically milder form known as Becker (BMD). of dystrophinopathies (<1%) (13), yet they represent an orphaned Patients with nonsense mutations in dystrophin are specifically tar- subpopulation with an undetermined pathomechanism and no cur- geted by stop codon read-through drugs, whereas out-of-frame de- rent personalized therapies. letions and insertions are targeted by exon-skipping therapies. Both The first missense mutation reported to cause DMD was L54R treatment strategies are currently in clinical trials. Dystrophin mis- in ABD1 of an 8-y-old patient (16). Another group later reported sense mutations, however, cause a wide range of phenotypic se- L172H, a missense mutation in a structurally analogous location of verity in patients. The molecular and cellular consequences of such ABD1 (17), yet this patient presented with mild symptoms at 42 mutations are not well understood, and there are no therapies spe- years of age. We first hypothesized that differential effects on cifically targeting this genotype. Here, we have modeled two rep- actin-binding affinity could account for differences in disease se- resentative missense mutations, L54R and L172H, causing DMD and verity between L54R and L172H. However, biochemical assays BMD, respectively, in full-length dystrophin. In vitro, the mutation reported only nominal effects of six different missense mutations, associated with the mild phenotype (L172H) caused a minor de- including L54R and L172H, on actin-binding affinity (18). In crease in tertiary stability, whereas the L54R mutation associated vitro biophysical analyses of the full-length proteins instead with a severe phenotype had a more dramatic effect. When stably revealed that each of the mutations rendered dystrophin more expressed in mammalian muscle cells, the mutations caused steady- insoluble and less stable during thermal denaturation. Missense state decreases in dystrophin protein levels inversely proportional mutations analyzed in purified ABD1 fragments also exhibited to the tertiary stability and directly caused by proteasomal degra- significant aggregation as assessed by thioflavin T fluorescence, dation. Both proteasome inhibitors and heat shock activators were Congo red staining, and electron microscopy (19). able to increase mutant dystrophin to WT levels, establishing the Although previous research indicates that missense mutations new cell lines as a platform to screen for potential therapeutics personalized to patients with destabilized dystrophin. in dystrophin ABD1 cause protein instability and aggregation, no in vitro measurements were able to distinguish between missense Duchenne muscular dystrophy | missense mutation | protein folding | mutations that cause severe DMD versus those that cause mild proteasome inhibitor | heat shock activator BMD symptoms. There is also currently no understanding of the in vivo cellular consequences of dystrophin instability. If such protein instability does indeed cause protein aggregation in the n striated muscle cells, the contraction of sarcomeres is coupled cell, therapeutic methods may aim at removal or resolubilization Ito the sarcolemma through the costamere protein network (1). One substructure of the costamere, known as the dystrophin– glycoprotein complex (DGC), physically links the cytoskeleton to Significance the extracellular matrix, contributing to myofiber integrity during contraction and relaxation (2). One of the key components of the Duchenne and Becker muscular dystrophies occur at a frequency DGC, dystrophin, is a 427-kDa structural protein composed of an of 1:4,000 live male births. Patients experience progressive muscle N-terminal actin binding domain (ABD1), a central rod region weakness and often succumb to cardiac or respiratory failure. made of 24 spectrin-like repeats that contain binding domains for Many laboratories are investigating potential treatments, in- several proteins (3–6), a cysteine-rich (CR) globular domain, and cluding several therapies that target subsets of patients with an intrinsically disordered C-terminal tail (CT) (7). Dystrophin is certain types of mutations. Yet there are currently no treatments encoded by the DMD gene, which is located on the X chromo- specific to patients with missense mutations in the affected pro- some and spans over 2.4 Mb of DNA, making it the largest in the tein, dystrophin. Here, we have shown missense mutant dystro- human genome (8). phin is degraded by the cell, which can be prevented by several Males who harbor a disease-causing mutation in their single copy small molecules targeting the cellular degradation machinery. In of the DMD gene develop a muscular dystrophy either with severe addition to determining disease mechanism, this work has generated new cell-based models that may aid in the discovery of symptoms known as Duchenne (DMD) or typically milder symp- personalized medicines for dystrophinopathy patients afflicted toms referred to as Becker (BMD) (9), occurring with an incidence with missense mutations. of ∼1 in 4,000 live male births (10). Muscle degeneration is pro-

gressive and boys lose ambulation in their teenage years. Death Author contributions: J.M.E. designed research; D.M.T. performed research; D.M.T. and occurs in the second or third decade of life, commonly due to J.J.B. contributed new reagents/analytic tools; D.M.T. analyzed data; and D.M.T. and J.M.E. cardiac or respiratory failure (11). The types of mutations causing wrote the paper. DMD and BMD include insertions, deletions, point mutations, and The authors declare no conflict of interest. splice site mutations (12). A recent collaborative data bank has This article is a PNAS Direct Submission. calculated that patients with nonsense mutations account for ∼10% 1To whom correspondence should be addressed. Email: [email protected]. of dystrophinopathies (13), all of whom are the target population This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. for stop codon read-through therapies (14). Patients with small 1073/pnas.1508755112/-/DCSupplemental.

12414–12419 | PNAS | October 6, 2015 | vol. 112 | no. 40 www.pnas.org/cgi/doi/10.1073/pnas.1508755112 Downloaded by guest on September 25, 2021 of aggregates akin to treatments for polyglutamine diseases or protein submissions to the SNP database (rs145668843). To determine the aggregate myopathies (20, 21). If, however, the unstable dystrophin stability of the tertiary protein structures, each purified protein was protein is degraded, then more effective disease treatments may in- measured by DSF over a temperature gradient from 20 to 90 °C (Fig. stead focus on stabilizing the mutant protein. Here, we use differ- 1B). WT dystrophin showed a cooperative unfolding transition (Tm = ential scanning fluorimetry (DSF) to assess the in vitro tertiary 43.36 °C) and the SNP I232M gave a similar melt curve. Mutant structure of L54R and L172H mutant dystrophins, and found that the dystrophin L172H was slightly left shifted (Tm = 42.08 °C), indicating more severe mutation (L54R) caused greater tertiary instability. asmalldecreaseinproteinstability. L54R was more left-shifted Additionally, we generated stable myoblast lines expressing L54R and (Tm = 40.03 °C), suggesting either greater thermal instability or the L172H dystrophin, which revealed that the mutant proteins are de- presence of a partially unfolded state of the protein at initiation of the graded by the proteasome to a degree that is inversely proportional to experiment. These DSF results indicate that L54R and L172H cause their in vitro stabilities. Flow cytometry used to detect dystrophin improper folding of dystrophin, with the mutation leading to severe showed several small molecules capable of increasing the mutant disease symptoms (L54R) having a greater deleterious effect. protein levels, supporting the cell lines’ potential for future high- As a sensitive measure of protein aggregation, the hydrodynamic throughput screens for dystrophin stabilizers. radius of full-length protein was measured by dynamic light scattering (DLS) (Fig. 1 C and D). DLS analysis of dystrophin isoforms Results Dp71 and Dp260, and full-length dystrophin yielded z-average sizes L54R and L172H Cause Different Degrees of Protein Instability in that correlated positively with increasing molecular weight as Vitro. We expressed and purified full-length WT, L54R, and expected (Fig. S1). Boiling full-length WT dystrophin protein in- L172H dystrophins (Fig. 1A) using the previously established bacu- creased its z-average size by approximately fourfold (Fig. 1C), lovirus system and FLAG-affinity chromatography (22). As an demonstrating the large effect of thermal-induced aggregation on additional control, we also expressed and purified full-length z-average size. However, the two missense mutations, L54R and dystrophin with the non–disease-causing amino acid substitution L172H, and the SNP I232M each showed z-average values similar I232M (Fig. 1A), which has been reported by several independent to WT dystrophin, and the polydispersity index was also equivalent across all four proteins (Fig. 1D). Together, the DSF and DLS data indicate that the DMD and BMD missense mutations cause measurable thermal instability, but do not cause the large-order aggregation more typical of protein aggregation diseases.

L54R and L172H Mutations Lead to Decreased Dystrophin Levels in Myoblasts. To elucidate the cellular fate of protein instability caused by the L54R and L172H missense mutations, C2C12 myoblast cell lines were generated to stably express full-length WT, L54R, L172H, and I232M dystrophins with amino-terminal GFP tags. The amount of transgene insertion was measured by quantitative PCR (qPCR) of genomic DNA (Fig. 2A).TheL172Hlinewasfoundtohaveap- proximately twofold higher incorporation of transgene compared with the WT, L54R, and I232M lines. Transcript levels measured by RT- qPCR (Fig. 2B) were elevated in the L172H and I232M lines compared with WT and L54R lines. Despite elevated or similar transcript levels for L172H and L54R, respectively, quantitative Western blot analysis (Fig. 2C) revealed graded decreases in the GFP-dystrophin protein levels of L172H and L54R lines compared with WT. Quan- tification of three separate radioimmunoprecipitation assay (RIPA) lysates revealed that both missense mutations significantly decreased the steady-state level of dystrophin, whereas the SNP I232M increased it significantly (Fig. 2D). Due to the differences in mRNA expression inherent to stable cell lines (Fig. 2 A and B), we normalized the protein levels to their respective transcript level (Fig. 2E). The steady-state ratio of protein/mRNA revealed that L54R expressed only 13% of WT protein and L172H expressed 46% of WT protein, whereas I232M was comparable to WT. The steady-state levels of WT, severe mutant L54R, mild mutant L172H, and SNP I232M dystrophins in live muscle cells exhibited an inverse relationship with the degree of protein instability measured in vitro (Fig. 1). The C2C12 myoblast lines were then switched to a low serum media and analyzed after 0, 3, and 7 d to assess whether their Fig. 1. In vitro stability and aggregation state of full-length dystrophin. differentiation into myotubes had any effect on the steady-state (A) Coomassie-stained gel of purified full-length dystrophin protein. (B) Differ- levels of GFP-dystrophin. Western blot analysis of the protein ential scanning fluorimetry (DSF) monitored at 610 nm over a temperature desmin (Fig. 3A), a marker of myotube differentiation (23), range of 20–90 °C. Fluorescent values are normalized from 0 to 1 fraction showed an increase from day 0 to day 7 for each of the cell lines unfolded. WT protein is shown in black, missense mutations are in red, analyzed. For both WT and L54R, there were statistically signif- and SNP is in green. (C) Calculated z average for full-length dystrophin protein icant increases in desmin protein from day 0 to day 7 (Fig. 3B). measured by dynamic light scattering (DLS) of a technical triplicate; units are in nanometers of hydrodynamic radius. “BOIL” samples were measured as a These results indicate that expression of neither WT dystrophin positive control of aggregation. (D, Top) Calculated z average of each dys- nor a missense dystrophin had any effect on the ability of the trophin protein measured in triplicate. (Bottom) Polydispersity Index mea- C2C12 cell line to differentiate. GFP-dystrophin protein levels sured on a scale from 0 to 1, where 0 indicates complete monodispersity and increased in the WT dystrophin line, but the increase was not CELL BIOLOGY 1 indicates a completely heterogeneous mixture. significant (Fig. 3 A and C). Most importantly, there was no

Talsness et al. PNAS | October 6, 2015 | vol. 112 | no. 40 | 12415 Downloaded by guest on September 25, 2021 MG132 was taken as the maximum protein abundance for each cell line. The ratio of GFP-dystrophin protein at 10 μMMG132to initial protein at 0 μM MG132 was calculated for each line. L54R dystrophin increased significantly upon proteasomal inhibition compared with WT (Fig. 4D). These results together indicate that the observed differences in dystrophin protein levels are due to protein degradation. Previous studies have reported an inverse correlation between dystrophin abundance and disease severity in several mouse models (24–26). Here, we have shown a mutant associated with mild disease (L172H) that decreases dystrophin levels approximately by one-half and a mutant associated with severe disease (L54R) that decreases dystrophin to 10% of WT levels in our myoblast model. We therefore hypothesized that abundance of dystrophin protein might correlate with disease severity and that increasing mutant dystrophin levels could ameliorate disease symptoms. Our novel cell lines were administered several compounds to identify small molecules that had a dose-dependent effect on missense dystrophin steady-state levels. First, we screened several osmolytes (Fig. 5A), which are thought to populate the interior of the cell at high concentrations and therefore enhance thermodynamic stability of proteins (27, 28). The osmolyte trehalose is commonly synthesized by bacteria and yeast under thermodynamic stress (29, 30), whereas N Fig. 2. Expression levels of transgenic dystrophin in myoblast cell lines. trimethylamine -oxide (TMAO) is thought to stabilize proteins (A) qPCR analysis of genomic DNA. Sequence amplified within CMV promoter against the osmotic pressure in salt water animals (31). However, to measure degree of transgene insertion into genome. n = 3. Measurements neither TMAO nor trehalose induced measurable increases in relative to reference gene ADH and normalized to a WT sample. ANOVA L54R-dystrophin when incubated with concentrations up to 200 mM, F < 0.0001, *P < 0.05 compared with WT. (B) RT-qPCR of total RNA. Sequence and cells did not survive higher concentrations of 400 mM amplified within GFP sequence to measure levels of transgenic transcript. TMAO or 500 mM trehalose. The kidney osmolyte betaine (32) = n 3. Measurements relative to reference transcript Hprt and normalized to a caused an increase in GFP-dystrophin in the mutant L54R line, WT sample. ANOVA F < 0.0001, *P < 0.05 compared with WT. (C) Represen- tative Western blot of cell lysates. Blot was probed with GFP primary anti- but only at a concentration (400 mM) that argues against its bodies to detect the transgenic GFP-Dystrophin fusion protein (455 kDa). potential as a therapeutic. (D) Quantification of n = 3 separate cell lyses and Western blot analyses. Measurements normalized to WT sample on each blot. ANOVA F < 0.0001, *P < 0.05 compared with WT. (E) The ratio of the normalized protein levels to the normalized transcript levels. All error bars represent SEM.

measureable increase in GFP-dystrophin protein in the L54R line (Fig. 3 A and C), indicating that the L54R dystrophin was not stabilized upon differentiation of C2C12 myoblasts into myotubes. Transcript analysis by RT-qPCR confirmed that there was no significant change in GFP-dystrophin transcript levels upon differentiation as expected (Fig. S2).

L54R and L172H Mutant Dystrophins Are Degraded by the Proteasome. The decreased mutant dystrophin levels visible by Western blot analysis (Figs. 2 and 3) may have been due to dystrophin aggregates partitioning into RIPA-insoluble compartments of the C2C12 cells. Therefore, the insoluble pellets from the RIPA cell lysates were resolubilized by sonication in 1% SDS and analyzed by Western blot. However, there was no mutant dystrophin de- tectable in the resolubilized pellets (Fig. 4A). To determine whether mutant dystrophin was the target of a degradative pathway, the autophagy inhibitor bafilomycin and the proteasome inhibitor MG132 were incubated at increasing concentrations with the L54R cell line and then analyzed for GFP-dystrophin expression by Western blot analysis (Fig. 4B). In contrast to the positive control p62, no increase in mutant dystrophin protein was observed in cells incubated with increasing concentrations of the autophagy inhibitor bafilomycin (Fig. 4B). However, there was a clear, dose-dependent increase in L54R dystrophin levels in cells treated with MG132 (Fig. 4B). These data suggest that Fig. 3. Differentiation of stable transgenic cell lines. (A) Representative the ubiquitin–proteasome system is the major pathway of mis- Western blots of desmin and transgenic dystrophin expression in cell lines sense dystrophin degradation. Each of the stable C2C12 cell differentiated for 0, 3, and 7 d. (B and C) Quantification of Western blots represented in A. n = 3. (B) Desmin protein was normalized to nontransgenic lines was then treated with increasing concentrations of MG132 line day 0 on each blot. ANOVA F < 0.001, *P < 0.05. (C) GFP-Dystrophin to determine the relative degree of proteasomal degradation for protein was normalized to WT line day 0 on each blot. ANOVA F < 0.001; WT each genotype (Fig. 4C). Because higher doses of MG132 were and L54R statistically different from all nontransgenic samples; n.s., not lethal to the cells, the GFP-dystrophin immunoreactivity at 10 μM significant within genotype. All error bars represent SEM.

12416 | www.pnas.org/cgi/doi/10.1073/pnas.1508755112 Talsness et al. Downloaded by guest on September 25, 2021 ubiquitination in the cell, emphasizing the close connection between the heat shock and the ubiquitin–proteasome pathways. These data also corroborate previous studies that have found Hsp90 inhibition increases ubiquitination (42). The heat shock system was confirmed to be up-regulated upon treatment with each of the small-molecule compounds (Fig. S4). Finally, treatment of the L54R and L172H GFP-dystrophin cell lines with submaximal doses of gedunin, bortezomib, or epoxomicin each caused marked increases in the cellular GFP fluorescence measured by flow cytometry (Fig. 5 C and D). These results both confirm the Western blot results and validate flow cytometry as a high-throughput method to screen the cell lines for stabilizers of misfolded dystrophin. Discussion Previous studies have demonstrated that several missense mutations in ABD1 of dystrophin cause protein instability in vitro (18, Fig. 4. Degradation pathway of mutant dystrophin. (A) Western blot 19). Here, we were able to distinguish between a missense muta- analysis of soluble (S) versus insoluble (I) fractions of transgenic mammalian tion associated with DMD (L54R) from one associated with BMD cell lines. (B) Western blot analyses of WT cell line compared with L54R line (L172H) by monitoring protein tertiary structure in vitro (Fig. 1B). treated with bafilomycin or MG132 to inhibit autophagy or proteasomal Our myoblast cell lines expressing full-length WT and missense degradation, respectively. (C) Representative Western blot analyses of each of the cell lines treated with varying concentrations of MG132 blotted with dystrophins have revealed that tertiary structure instability leads GFP antibody to detect GFP-Dystrophin fusion protein. (D) Ratio of protein at to a corresponding decrease in the steady-state cellular level of 10 μM MG132 to protein at 0 μM MG132, effectively showing the total increase in protein with maximal proteasome inhibition. Calculated for n = 4 experiments represented in C. ANOVA F < 0.01. *P < 0.05 compared with WT. All error bars represent SEM.

Although the heat shock system is directly linked to the proteasome degradation pathway (33), we found that there was no global up-regulation in heat shock proteins associated with expression of mutant GFP-dystrophins alone (Fig. S3). Therefore, exogenous up-regulation of the heat shock system has the potential to affect missense dystrophin abundance. The small- molecule celastrol activates the conserved heat shock factor 1 (HSF1) (34, 35), but caused no increase in mutant L54R GFP- dystrophin levels at concentrations up to 1 μM. The compound BGP-15 specifically induces one of the heat shock proteins, HSP72, and has been shown to alleviate symptoms in models of diabetes and muscular dystrophy (36, 37). However, we saw no direct molecular effect on L54R GFP-dystrophin levels at tolerated concentrations up to 50 μM. Finally, we tested gedunin, which binds to and inhibits Hsp90, thereby releasing bound HSF1 for transcriptional activity (38, 39). Increasing L54R GFP-dystrophin was observed in cells treated with 10–80 μM gedunin, suggesting that pharmacologic activation of the heat shock system can facilitate stabilization of mutant dystrophin. Given that the general proteasome inhibitor MG132 effectively increased mutant GFP-dystrophin levels (Fig. 4), we evaluated several additional proteasome inhibitors (Fig. 5A). The MG132 derivative, MG262, gave similar results at even lower concentrations than MG132. Bortezomib is a member of the boronic acid class of proteasome inhibitors (40) and was recently demonstrated to increase levels of missense mutated dysferlin in vivo (41). Bortezomib had a strong concentration-dependent effect on increasing L54R GFP-dystrophin protein levels. Epoxomycin is in the epoxyketone Fig. 5. Small-molecule therapeutics screen to increase dystrophin protein class of proteasome inhibitors (40) and also had a robust con- levels. (A) Western blot analyses of WT cell line compared with L54R line centration-dependent effect on mutant GFP-dystrophin levels. We treated with a panel of small molecules using GFP antibody to detect GFP- further investigated the effects of three of the positive hits: gedunin, dystrophin fusion protein in each top panel and C4 actin as a loading control in bortezomib, and epoxomicin. Both L54R and L172H myoblast lines each bottom panel. *These cells did not survive the given dosage. Bz, borte- were treated in parallel with submaximal doses of each small mol- zomib; epoxo, epoxomicin. (B) Cells were treated with positive hits of screen: ecule (Fig. 5B). Western blot analysis of the GFP-dystrophin pro- 40 μM gedunin, 0.5 μM bortezomib, or 0.5 μM epoxomicin. Western blot tein demonstrated marked increases in the L54R line for all three analyses after drug treatment with GFP antibody to detect the fusion protein, small molecules. The L172H mutant dystrophin was also increased C4 to detect actin as a control, and ubiquitin antibody. (C and D) Live cells were analyzed for GFP fluorescence by flow cytometry. (C)Percentageofmaximum when treated with either bortezomib or epoxomicin. Western blot μ μ B Bottom histogram for L54R line untreated and treated with 40 M gedunin, 0.5 M analysis of ubiquitin (Fig. 5 , ) revealed that both protea- bortezomib, or 0.5 μMepoxomicin.(D) Percentage of maximum histogram for some inhibitors caused a large increase in ubiquitination across the L172H line untreated and treated with 40 μMgedunin,0.5μM bortezomib, or CELL BIOLOGY proteome. The heat shock activator gedunin had a similar effect on 0.5 μMepoxomicin.

Talsness et al. PNAS | October 6, 2015 | vol. 112 | no. 40 | 12417 Downloaded by guest on September 25, 2021 dystrophin (Fig. 2E), which appears to be directly controlled by actin-binding affinity (18). Indeed, proteasome inhibition in a proteasomal degradation (Fig. 4B). The C2C12 myoblast model missense dysferlin model has been shown to restore functional thereby recapitulates reported symptoms of the modeled patients. protein (49). If restored missense dystrophin proves to be at least Dystrophin Western blot analysis of muscle from the L54R DMD partially functional, then proteasome inhibition may be an ef- patient revealed a large decrease in full-length dystrophin protein, fective, personalized treatment for the missense subpopulation whereas immunofluorescence analysis showed dystrophin properly of dystrophinopathy patients. localized to the sarcolemma, but at reduced intensity (16). Similarly, Proteasome inhibitors such as bortezomib and epoxomicin muscle from the L172H BMD patient presented with decreased derivative carfilzomib are currently being used in the clinic for levels of full-length dystrophin assessed by Western blot and correct treatment of multiple myeloma and other cancers (40). In con- localization by immunofluorescence microscopy (17). Although our trast to the acute treatment regimen necessary to treat cancer data suggest that a reduction in total dystrophin levels caused by patients, further work is needed to determine whether chronic proteasomal degradation may contribute to the dystrophic pheno- proteasome inhibition could be tolerated by missense mutant type in patients afflicted with missense mutations, the severe phe- dystrophinopathy patients requiring treatment over their life- notype of L54R in a human patient with residual sarcolemmal times. For example, up to 30% of even WT proteins are esti- localization (16) leaves open the possibility that the mutation also mated to be degraded by the proteasome due to misfolding (50) perturbs dystrophin function. In support of this possibility, we pre- so global proteasome inhibition may have adverse consequences. viously demonstrated a significant reduction in the actin-binding Alternatively, pharmacologic inhibition of the ubiquitination li- affinity of mouse dystrophin bearing the L54R mutation (18). gases specific to skeletal muscle or even dystrophin could be more The absence of large-order structures seen by DLS (Fig. 1D), effective therapeutics with fewer side effects. Indeed, the cancer the high solubility in cell culture (Fig. 4A), and the lack of effect of biology field, which has been using proteasome inhibitors for an autophagy inhibitor (Fig. 4B) all indicate that the L54R and treatment of multiple myeloma (51), has recently begun to target L172H missense mutations do not cause dystrophin to form the cancer-specific E3 ligases responsible for ubiquitination of p53 aggregates in the cell. These data differentiate missense mutant and c-Myc (52). Although identifying the specific E1–E2–E3 cas- dystrophinopathy from the protein aggregate myopathies (PAMs) cade for a given substrate is a difficult task (53, 54), our stable (21, 43). In PAM diseases, the causative mutant protein evades missense dystrophin cell lines provide a high-throughput screening degradation, forming fibrils or aggregation conglomerates along platform (Fig. 5 C and D) that could allow for the efficient iden- with other muscle proteins. Often a gain-of-function, toxic effect is tification of the enzymes responsible for ubiquitination and sub- caused by the aggregates, and therefore treatment for these dis- sequent degradation of misfolded dystrophin. eases is twofold: removing the aggregate and replacing it with a Although missense mutations represent a very small percent- functional protein. The L54R and L172H mutant dystrophins age of dystrophinopathy patients, therapies directed at such expressed in myoblasts are degraded by the proteasome, and patients may also be more broadly applicable to any patient with therefore treatment is potentially onefold: inhibiting degradation compromised dystrophin stability. BMD is often the result of in- of the mutant protein. Our small-molecule screen found several frame deletions or insertions, resulting in internally truncated compounds that were able to increase mutant dystrophin levels dystrophin protein. In addition to potentially lacking functional (Fig. 5A). The osmolyte betaine stabilized mutant dystrophin at domains, internally truncated dystrophin has been shown to be very high concentrations, and the heat shock activator gedunin thermally unstable in vitro (55), and BMD patients often present increased dystrophin protein levels at moderate concentrations. with a decreased abundance of dystrophin (56). In addition, The efficacy of gedunin supports heat shock activators as a class of small-molecule stabilizers of misfolded dystrophin or targeted E3 compounds with potential to treat DMD patients with missense ligase inhibition may also be useful adjuvants to other person- mutations, and a broader screen may identify heat shock activators alized therapies dependent on the functionality of sequence- effective at lower doses. altered dystrophins such as exon-skipping (15), virus-based gene The most successful class of small molecules for increasing therapy (57, 58) and stop codon read-through drugs (14). mutant dystrophin levels was proteasome inhibitors. Each of the Therefore, development of any therapy that stabilizes and/or tested proteasome inhibitors restored mutant dystrophin to WT prevents degradation of misfolded dystrophin could potentially levels at concentrations less than 10 μM. Several previous studies benefit the DMD patient community as a whole. have tested proteasome inhibition to treat the nonsense mutation mdx mouse model of dystrophinopathy (44–47) but have Materials and Methods reported contradictory results. Although proteasome inhibition Detailed descriptions of cloning, protein expression and purification, DSF, by MG132 in the mdx mouse restored the amino-terminal frag- DLS, generation of mammalian cell lines, cell culture and differentiation, ment of dystrophin translated up to the premature stop codon qPCR, RT-qPCR, Western blot analysis, solubility assay, small-molecule treat- (44), prolonged treatment with MG132 resulted in no improve- ments, flow cytometry, and statistical analysis are provided in SI Materials and Methods. ment of muscle function (46). It has therefore been hypothesized that proteasome inhibition of nonsense dystrophin may treat ACKNOWLEDGMENTS. We thank Dr. Davin Henderson for his intellectual some of the symptoms of dystrophinopathy such as decreased input and guidance at the initiation of this study, Dr. Frank Niessen for his help DGC expression but does not treat the primary cause of disease with the DSF experiments, Christopher Scherber for his technical assistance, (48). Our data demonstrate that proteasome inhibition restores and the University of Minnesota Flow Cytometry Core for use of their facilities. This work was supported by National Institutes of Health (NIH) Grant R01 full-length, missense dystrophin to WT levels. It is possible that AR042423 (to J.M.E.), American Heart Association Fellowship 12PRE12040402 the missense dystrophin could be largely functional given that (to D.M.T.), the NIH Training Program in Muscle Research (AR007612), and a missense mutations in the ABD1 did not dramatically affect University of Minnesota Doctoral Dissertation Fellowship (to J.J.B.).

1. Ervasti JM (2003) Costameres: The Achilles’ heel of Herculean muscle. J Biol Chem 5. Bhosle RC, Michele DE, Campbell KP, Li Z, Robson RM (2006) Interactions of in- 278(16):13591–13594. termediate filament protein synemin with dystrophin and utrophin. Biochem Biophys 2. Ervasti JM, Campbell KP (1991) Membrane organization of the dystrophin-glycopro- Res Commun 346(3):768–777. tein complex. Cell 66(6):1121–1131. 6. Prins KW, et al. (2009) Dystrophin is a microtubule-associated protein. JCellBiol186(3):363–369. 3. Brenman JE, Chao DS, Xia H, Aldape K, Bredt DS (1995) Nitric oxide synthase com- 7. Koenig M, Monaco AP, Kunkel LM (1988) The complete sequence of dystrophin plexed with dystrophin and absent from skeletal muscle sarcolemma in Duchenne predicts a rod-shaped cytoskeletal protein. Cell 53(2):219–228. muscular dystrophy. Cell 82(5):743–752. 8. Koenig M, et al. (1987) Complete cloning of the Duchenne muscular dystrophy (DMD) 4. DeWolf C, et al. (1997) Interaction of dystrophin fragments with model membranes. cDNA and preliminary genomic organization of the DMD gene in normal and af- Biophys J 72(6):2599–2604. fected individuals. Cell 50(3):509–517.

12418 | www.pnas.org/cgi/doi/10.1073/pnas.1508755112 Talsness et al. Downloaded by guest on September 25, 2021 9. Hoffman EP, Brown RH, Jr, Kunkel LM (1987) Dystrophin: The protein product of the 36. Chung J, et al. (2008) HSP72 protects against obesity-induced insulin resistance. Proc Duchenne muscular dystrophy locus. Cell 51(6):919–928. Natl Acad Sci USA 105(5):1739–1744. 10. Mendell JR, et al. (2012) Evidence-based path to newborn screening for Duchenne 37. Gehrig SM, et al. (2012) Hsp72 preserves muscle function and slows progression of muscular dystrophy. Ann Neurol 71(3):304–313. severe muscular dystrophy. Nature 484(7394):394–398. 11. Rall S, Grimm T (2012) Survival in Duchenne muscular dystrophy. Acta Myol 31(2): 38. Brandt GEL, Schmidt MD, Prisinzano TE, Blagg BSJ (2008) Gedunin, a novel hsp90 117–120. inhibitor: Semisynthesis of derivatives and preliminary structure-activity relationships. 12. Aartsma-Rus A, Van Deutekom JCT, Fokkema IF, Van Ommen G-JB, Den Dunnen JT J Med Chem 51(20):6495–6502. (2006) Entries in the Leiden Duchenne muscular dystrophy mutation database: An 39. Neef DW, Jaeger AM, Thiele DJ (2011) Heat shock transcription factor 1 as a thera- overview of mutation types and paradoxical cases that confirm the reading-frame peutic target in neurodegenerative diseases. Nat Rev Drug Discov 10(12):930–944. rule. Muscle Nerve 34(2):135–144. 40. Kisselev AF, van der Linden WA, Overkleeft HS (2012) Proteasome inhibitors: An ex- 13. Bladen CL, et al. (2015) The TREAT-NMD DMD Global Database: Analysis of more than panding army attacking a unique target. Chem Biol 19(1):99–115. 7,000 Duchenne muscular dystrophy mutations. Hum Mutat 36(4):395–402. 41. Azakir BA, Erne B, Di Fulvio S, Stirnimann G, Sinnreich M (2014) Proteasome inhibitors 14. Peltz SW, Morsy M, Welch EM, Jacobson A (2013) Ataluren as an agent for thera- increase missense mutated dysferlin in patients with muscular dystrophy. Sci Transl peutic nonsense suppression. Annu Rev Med 64(2):407–425. Med 6(250):250ra112. 15. Goemans NM, et al. (2011) Systemic administration of PRO051 in Duchenne’s mus- 42. Basak S, Pookot D, Noonan EJ, Dahiya R (2008) Genistein down-regulates androgen cular dystrophy. N Engl J Med 364(16):1513–1522. receptor by modulating HDAC6-Hsp90 chaperone function. Mol Cancer Ther 7(10): – 16. Prior TW, et al. (1993) A missense mutation in the dystrophin gene in a Duchenne 3195 3202. muscular dystrophy patient. Nat Genet 4(4):357–360. 43. Vrabie A, Goebel H (2007) Protein aggregation in muscle fibers and respective neu- 17. Hamed S, Sutherland-Smith A, Gorospe J, Kendrick-Jones J, Hoffman E (2005) DNA romuscular disorders. Protein Misfolding, Aggregation, and Conformational Diseases. sequence analysis for structure/function and mutation studies in Becker muscular Part B: Molecular Mechanisms of Conformational Diseases. Protein Reviews, eds – dystrophy. Clin Genet 68(1):69–79. Uversky VN, Fink AL (Springer, New York), Vol 6, Chap 18, pp 365 389. 18. Henderson DM, Lee A, Ervasti JM (2010) Disease-causing missense mutations in actin 44. Bonuccelli G, et al. (2003) Proteasome inhibitor (MG-132) treatment of mdx mice binding domain 1 of dystrophin induce thermodynamic instability and protein ag- rescues the expression and membrane localization of dystrophin and dystrophin- associated proteins. Am J Pathol 163(4):1663–1675. gregation. Proc Natl Acad Sci USA 107(21):9632–9637. 45. Bonuccelli G, et al. (2007) Localized treatment with a novel FDA-approved protea- 19. Singh SM, Kongari N, Cabello-Villegas J, Mallela KMG (2010) Missense mutations in some inhibitor blocks the degradation of dystrophin and dystrophin-associated pro- dystrophin that trigger muscular dystrophy decrease protein stability and lead to teins in mdx mice. Cell Cycle 6(10):1242–1248. cross-β aggregates. Proc Natl Acad Sci USA 107(34):15069–15074. 46. Selsby J, Morris C, Morris L, Sweeney L (2012) A proteasome inhibitor fails to atten- 20. Jimenez-Sanchez M, Thomson F, Zavodszky E, Rubinsztein DC (2012) Autophagy and uate dystrophic pathology in mdx mice. PLoS Curr 4:e4f84a944d8930. polyglutamine diseases. Prog Neurobiol 97(2):67–82. 47. Rougier JS, Gavillet B, Abrie H (2013) Proteasome inhibitor (MG132) rescues Nav1.5 21. Schröder R (2013) Protein aggregate myopathies: The many faces of an expanding protein content and the cardiac sodium current in dystrophin-deficient mdx5cv mice. disease group. Acta Neuropathol 125(1):1–2. Front Physiol 4:51. 22. Rybakova IN, Patel JR, Davies KE, Yurchenco PD, Ervasti JM (2002) Utrophin binds 48. Hollinger K, Selsby JT (2013) The physiological response of protease inhibition in laterally along actin filaments and can couple costameric actin with sarcolemma dystrophic muscle. Acta Physiol (Oxf) 208(3):234–244. when overexpressed in dystrophin-deficient muscle. Mol Biol Cell 13(5):1512–1521. 49. Azakir BA, Di Fulvio S, Kinter J, Sinnreich M (2012) Proteasomal inhibition restores 23. Kislinger T, et al. (2005) Proteome dynamics during C2C12 myoblast differentiation. biological function of mis-sense mutated dysferlin in patient-derived muscle cells. Mol Cell Proteomics 4(7):887–901. J Biol Chem 287(13):10344–10354. 24. Phelps SF, et al. (1995) Expression of full-length and truncated dystrophin mini-genes 50. Wang F, Durfee LA, Huibregtse JM (2013) A cotranslational ubiquitination pathway in transgenic mdx mice. Hum Mol Genet 4(8):1251–1258. for quality control of misfolded proteins. Mol Cell 50(3):368–378. 25. Li D, Yue Y, Duan D (2008) Preservation of muscle force in Mdx3cv mice correlates 51. Kumar SK, et al. (2008) Improved survival in multiple myeloma and the impact of with low-level expression of a near full-length dystrophin protein. Am J Pathol 172(5): novel therapies. Blood 111(5):2516–2520. – 1332 1341. 52. Weathington NM, Mallampalli RK (2014) Emerging therapies targeting the ubiquitin 26. van Putten M, et al. (2012) The effects of low levels of dystrophin on mouse muscle proteasome system in cancer. J Clin Invest 124(1):6–12. function and pathology. PLoS One 7(2):e31937. 53. Deshaies RJ, Joazeiro CA (2009) RING domain E3 ubiquitin ligases. Annu Rev Biochem 27. Kumar R (2009) Role of naturally occurring osmolytes in protein folding and stability. 78:399–434. – Arch Biochem Biophys 491(1-2):1 6. 54. Li W, et al. (2008) Genome-wide and functional annotation of human E3 ubiquitin 28. Arakawa T, Ejima D, Kita Y, Tsumoto K (2006) Small molecule pharmacological ligases identifies MULAN, a mitochondrial E3 that regulates the organelle’s dynamics chaperones: From thermodynamic stabilization to pharmaceutical drugs. Biochim and signaling. PLoS One 3(1):e1487. – Biophys Acta 1764(11):1677 1687. 55. Henderson DM, Belanto JJ, Li B, Heun-Johnson H, Ervasti JM (2011) Internal deletion 29. Castillo K, et al. (2013) Trehalose delays the progression of amyotrophic lateral scle- compromises the stability of dystrophin. Hum Mol Genet 20(15):2955–2963. – rosis by enhancing autophagy in motoneurons. Autophagy 9(9):1308 1320. 56. Anthony K, et al. (2011) Dystrophin quantification and clinical correlations in Becker 30. Tanaka M, et al. (2004) Trehalose alleviates polyglutamine-mediated pathology in a muscular dystrophy: Implications for clinical trials. Brain 134(Pt 12):3547–3559. mouse model of Huntington disease. Nat Med 10(2):148–154. 57. Wang B, Li J, Xiao X (2000) Adeno-associated virus vector carrying human mini- 31. Yancey PH, Clark ME, Hand SC, Bowlus RD, Somero GN (1982) Living with water stress: dystrophin genes effectively ameliorates muscular dystrophy in mdx mouse model. Evolution of osmolyte systems. Science 217(4566):1214–1222. Proc Natl Acad Sci USA 97(25):13714–13719. 32. Burg MB (1995) Molecular basis of osmotic regulation. Am J Physiol 268(6 Pt 2): 58. Harper SQ, et al. (2002) Modular flexibility of dystrophin: Implications for gene F983–F996. therapy of Duchenne muscular dystrophy. Nat Med 8(3):253–261. 33. Shiber A, Ravid T (2014) Chaperoning proteins for destruction: Diverse roles of Hsp70 59. Hartley JL, Temple GF, Brasch MA (2000) DNA cloning using in vitro site-specific re- chaperones and their co-chaperones in targeting misfolded proteins to the protea- combination. Genome Res 10(11):1788–1795. some. Biomolecules 4(3):704–724. 60. Niesen FH, Berglund H, Vedadi M (2007) The use of differential scanning fluorimetry 34. Trott A, et al. (2008) Activation of heat shock and antioxidant responses by the to detect ligand interactions that promote protein stability. Nat Protoc 2(9): natural product celastrol: Transcriptional signatures of a thiol-targeted molecule. Mol 2212–2221. Biol Cell 19(3):1104–1112. 61. Legardinier S, et al. (2009) A two-amino acid mutation encountered in Duchenne 35. Westerheide SD, et al. (2004) Celastrols as inducers of the heat shock response and muscular dystrophy decreases stability of the rod domain 23 (R23) spectrin-like repeat cytoprotection. J Biol Chem 279(53):56053–56060. of dystrophin. J Biol Chem 284(13):8822–8832. CELL BIOLOGY

Talsness et al. PNAS | October 6, 2015 | vol. 112 | no. 40 | 12419 Downloaded by guest on September 25, 2021