A novel drug-combination screen in zebrafish identifies epigenetic small molecule candidates for Duchenne muscular dystrophy

Duchenne muscular dystrophy (DMD) is a severe neuromuscular disorder and is one of the most common muscular dystrophies. There are currently few effective therapies to treat the disease, although many small-molecule approaches are being pursued. Specific histone deacetylase inhibitors (HDACi) can ameliorate DMD phenotypes in mouse and zebrafish animal models and have also shown promise for DMD in clinical trials. However, beyond these HDACi, other classes of epigenetic small molecules have not been broadly and systematically studied for their benefits for DMD. Here, we performed a novel chemical screen of a library of epigenetic compounds using the zebrafish dmd model. We identified candidate pools of epigenetic compounds that improve skeletal muscle structure in dmd zebrafish. We then identified a specific combination of two drugs, oxamflatin and salermide, that significantly rescued dmd zebrafish skeletal muscle degeneration. Furthermore, we validated the effects of oxamflatin and salermide in an independent laboratory. Our results provide novel, effective methods for performing a combination small-molecule screen in zebrafish. Our results also add to the growing evidence that epigenetic small molecules may be promising candidates for treating DMD.


Introduction
Duchenne muscular dystrophy (DMD) is a severe neuromuscular disorder caused by an X-linked mutation in the DMD gene, which encodes the protein dystrophin (Hoffman et al., 1987;Monaco et al., 1986). Dystrophin is a key component of the dystrophin-associated protein complex (DAPC), which serves as a stabilizing link between the cytoskeleton and extracellular matrix during muscle fiber contraction. Loss of dystrophin and DAPC function makes muscle cell membranes susceptible to contraction-induced damage and leads to progressive calcium dysregulation, satellite cell dysfunction, inflammation, fibrosis, and necrosis (reviewed in Allen et al., 2016 and in Spinazzola and Kunkel, 2016).
DMD is the most common type of muscular dystrophy, affecting approximately 1 in 3,500-5,000 male births (Emery, 1991;Mah et al., 2014). It first presents as motor difficulties in early childhood and progresses rapidly, leaving most affected boys in need of a wheelchair in their early teens and in need of respiratory aid in their 20s (reviewed in Crone and Mah, 2018).
The disease is usually fatal in the third or fourth decade due to respiratory or cardiovascular failure.
Treatment options for DMD are still quite limited. The current standard of care is corticosteroid treatment, which delays the progression of muscle dysfunction but has serious side effects (Bushby et al., 2010;Crone and Mah, 2018;Kinnett et al., 2015). DMD gene therapy and gene editing approaches are very promising but face many challenges (Chamberlain and Chamberlain, 2017;Conboy et al., 2018;Duan, 2018;Min et al., 2019). Many small molecule approaches are being identified that could benefit DMD by modulating different pathological mechanisms downstream of the dystrophin mutation (Crone and Mah, 2018;Guiraud and Davies, 2017;Hoffman, 2019;Spinazzola and Kunkel, 2016). A current view in the DMD field is that a combination of therapies targeting different pathological mechanisms may ultimately be most beneficial for patients (Guiraud and Davies, 2017;Hoffman, 2019).
Histone deacetylase inhibitors (HDACi) are one class of small molecules that has shown promise for DMD in the mdx mouse, in dmd zebrafish, and in DMD clinical trials (Bettica et al., 2016;Johnson et al., 2013;Minetti et al., 2006;ClinicalTrials.gov NCT02851797, NCT03373968;reviewed in Consalvi et al., 2014). HDACi are examples of "epigenetic drugs", small molecules that target chromatin modifications and transcriptional regulators. Histone acetylation, which is often linked with open chromatin and active transcription, is one example of an epigenetic modification and is the target of many HDACi (Kouzarides, 2007). There are four different classes of histone deacetylase (HDAC) proteins, with different functions and tissue expression patterns, and different HDACi often selectively inhibit specific HDACs (Xu et al., 2007). Studies in the mdx mouse have identified epigenetic components to the pathogenesis of DMD, including constitutive activation of HDAC2 and abnormal levels of certain histone modifications (Chang et al., 2018;Colussi et al., 2008;Colussi et al., 2009;Saccone et al., 2014).
Zebrafish are an outstanding model for DMD, particularly for screening and evaluating novel drug therapies (Bassett et al., 2003;Bassett and Currie, 2004;Berger et al., 2010;Maves, 2014;Widrick et al., 2019). A zebrafish dmd mutant strain, also known as sapje, has a nonsense mutation in exon 4 and is a dystrophin-deficient model of DMD (Bassett et al., 2003;Berger et al., 2010;Granato et al., 1996). The zebrafish dmd mutation is autosomal recessive, and approximately 25% of the offspring from a heterozygous dmd+/-cross exhibit degenerative muscle lesions by about 3 days post-fertilization. dmd zebrafish exhibit many aspects of human DMD pathology; in particular, skeletal muscle fibrosis and inflammation, including infiltration of mononuclear cells (Bassett and Currie, 2004;Berger et al., 2010). dmd zebrafish offer several advantages for screening and evaluating drugs, including being amenable to rapid and highthroughput screening and affording an exceptional range of approaches for assessing treatment outcomes (Kawahara et al., 2011;reviewed in Maves, 2014 andin Widrick et al., 2019).
Zebrafish eggs can be rapidly produced in large numbers, the resulting embryos readily absorb drug compounds, and muscle development and structure can easily be observed in vivo through birefringence techniques (Berger et al., 2012;Granato et al., 1996;Kawahara et al., 2011).
Because the zebrafish dmd mutation, in contrast to the mdx mouse and DMD mdx rat, is lethal during juvenile stages, survival can be used as an outcome measure for drug treatments (Hightower et al., 2020;Kawahara et al., 2011;Kawahara et al., 2014). Large-scale drug screens have highlighted the potential of dmd zebrafish for identifying new therapeutic compounds and targets as well as for understanding the molecular mechanisms behind DMD (Kawahara et al., 2011;Kawahara et al., 2014;Waugh et al., 2014). In addition, dmd zebrafish can replicate drug effects seen in mdx mice (Hightower et al., 2020;Johnson et al., 2013;Kawahara et al., 2011;Li et al., 2014;Winder et al., 2011). Because of this strong conservation, insight from zebrafish can inform mammalian DMD studies, while also taking advantage of the utility of high-throughput analysis.
Here we performed a pilot screen of the commercially-available Cayman Chemical Epigenetics Screening Library to identify additional small molecule epigenetic regulators that could improve the dmd zebrafish muscle phenotype. We identified candidate pools of epigenetic compounds that improve skeletal muscle structure in dmd zebrafish. We then identified a specific combination of two drugs, oxamflatin and salermide, that rescued dmd zebrafish skeletal muscle degeneration. Overall, our study provides novel and effective methods for performing a pooled drug-combination chemical screen and gives further evidence that epigenetic small molecules may be good candidates for successfully treating DMD.

Zebrafish husbandry
All experiments involving live zebrafish (Danio rerio) were carried out in compliance with the Institutional Animal Care and Use Committee guidelines of Seattle Children's Research Institute and the University of Maine. Zebrafish were raised and staged as previously described (Westerfield, 2007). Time (hpf or dpf) refers to hours or days postfertilization at 28.5°C. The wild-type stock and genetic background used was AB. The zebrafish dmd ta222a mutant line (also known as sapje) has been described and is a null allele (Bassett et al., 2003;Granato et al., 1996). dmd ta222a genotyping was performed as previously described (Berger et al., 2011). Eggs were collected from about 20-30-minute spawning periods and raised in petri dishes in ICS water (300 mg Instant Ocean/L, 0.56 mM CaCl2, 1.2 mM NaHCO3) at 28.5⁰C.

Small molecules
Epigenetic small molecule library screening was performed using the Cayman Chemical Epigenetics Screening Library (Item No. 11076, Batch No. 0455098). The Library was received as 10 mM stocks of each chemical in Dimethyl Sulfoxide (DMSO, Sigma). The composition of the Cayman Chemical Epigenetics Screening Library can vary. The version we obtained contained 94 chemicals distributed over two 96-well plates. The identities of the chemicals and plate lay-out are shown in Table 1. Additional chemicals were purchased individually from Cayman Chemical.

Drug treatments
To test toxicity and assess working doses of the Epigenetics Screening Library chemicals, wild-type AB embryos were treated with individual drugs beginning at 4 hpf. Embryos were not dechorionated prior to these dose tests. Three concentrations of each drug were tested: 10 µM, 1 µM, and 100 nM. Embryos were treated with drugs with 1% DMSO in embryo medium (EM), or with 1% DMSO in EM as a vehicle control. 12-well plates were used; each well contained 25 embryos and 3 mL of drug treatment. The drug treatment was changed every 24 hours until 4 dpf. Over the 4-day treatments, developmental abnormalities and survival were noted. Most drugs showed little or no effect on embryo health and survival at 1 µM doses, so we determined 1 µM to be the working dose for drug pool screening for most of the Library drugs. C646 and Apicidin were toxic at 1 µM but not at 100 nM. CAY10669, GSK-J1, GSK-J4, and Tenovin-1 exhibited strong toxicity and thus were removed from drug pool screening.
For drug pool screening and dmd rescue tests, embryos from dmd+/-crosses were used.
At about 24 hpf, the chorions were manually removed from the embryos, and embryos were sorted into fresh petri dishes or wells for drug treatments. 1000X stocks of drugs were made in DMSO just prior to each drug treatment experiment. Embryos were treated with drugs with 1% DMSO in embryo medium (EM), or with 1% DMSO in EM as a vehicle control. For the drug pool screen, 25 embryos each were sorted into two wells of a 12-well plate and treated with 3 mL of EM containing either the drug treatments or vehicle control. Treatments were started at about 24 hpf and continued for 3 days, changing the drug treatment or vehicle control EM solution every 24 hours. Larvae were fixed in 4% paraformaldehyde (PFA) at 4 dpf and stored at 4°C. Larval heads were removed for genotyping, and tails were maintained in 4% PFA.

Imaging and scoring muscle lesions
Larvae fixed in 4% PFA were rinsed in PBS with 0.025% Tween (PBSTw) prior to imaging. Animals were placed in PBSTw in a 60 mm glass Petri dish. An Olympus SZX16 stereomicroscope with attached Olympus DP72 camera was set up with one sheet of polarizing film over the trans-illumination base and another sheet of the film placed over the objective lens such that the two films were crossed. For drug pool screening, larvae were sorted under birefringence into muscle lesion positive or negative groups. The number of larvae in each group was recorded.
For quantitative birefringence measurements, we developed an approach inspired by a previously described method (Berger et al., 2012). In our approach, larvae were placed in a glassbottom petri dish in 2.5% methyl cellulose and oriented to maximize the brightness of the muscle tissue through the crossed polarizers. Birefringence was imaged in individual animals as above.
Images were acquired using Olympus Cellsens Dimensions software. Exposure time was adjusted so that only a few saturated pixels were present. ImageJ was used to outline the trunk musculature, using the wand tool, and to calculate the average pixel intensity within the resulting selection.

Statistical analyses
Statistical tests used are provided in the Figure Legends. Statistical analyses were performed and graphs were constructed using GraphPad Prism 8.

Epigenetic small-molecule library pooling approach
In a previous study, we showed that the pan-HDACi TSA could improve the zebrafish dmd muscle lesion phenotype (Johnson et al, 2013). We wanted to test whether we could identify new epigenetic drugs, or epigenetic drug combinations, that improve the zebrafish dmd muscle lesion phenotype. We turned to the Cayman Chemical Epigenetics Screening Library. The version of the Library that we obtained contained 94 chemicals distributed over two 96-well plates (Table 1). In order to efficiently test combinations of drugs in the Library, we designed a grid system consisting of a set of 393 different drug pools (Supplemental Table 1). In this system, drug A2 of Plate 1 is pooled with the remaining drugs of Row A of Plate 1 for one combination pool. Then drug A2 is pooled with Row B drugs of Plate 1 for a second combination pool, and then drug A2 is pooled with the remaining rows of drugs on both Plate 1 and Plate 2. Drug A3 of Plate 1 is then combined with drugs of Row B of Plate 1, and so on (Supplemental Table 1). In this grid system, each drug would be tested in combination with each of the other Library drugs in at least 1 pool (Supplemental Table 1).
Prior to screening these chemicals on dmd embryos, we performed dose and toxicity testing of each individual Library drug on wild-type embryos (see Materials and Methods). Most drugs showed little or no detrimental effect on embryo health and survival at 1 µM dose, so we selected 1 µM to be the working dose for drug screening for most of the Library drugs. Four drugs exhibited strong toxicity, so we removed these from the drug pool grid (CAY10669, GSK-J1, GSK-J4, and Tenovin-1; Table 1; Supplemental Table 1). We also removed Library drugs that are described as negative control drugs from the drug pool grid ((-)-JQ1, MI-nc, GSK-J2, and GSK-J5; Table 1; Supplemental Table 1).
To determine if screening pools of epigenetic small molecules was sensitive enough to identify drugs that could improve the zebrafish dmd muscle phenotype, we tested the pool of 10 drugs from Plate 2 Row D, which contains TSA (Table 1) These results indicate that a pool of drugs containing a known beneficial epigenetic drug, TSA, can improve the zebrafish dmd muscle phenotype. These results suggest that the epigenetic drugpooling design of our screen is sensitive enough to pick up positive hits of small molecules that improve the zebrafish dmd muscle phenotype.

Pilot screen of epigenetic small-molecule combination pools
We then proceeded to test epigenetic drug pools from our drug-combination grid system (Supplemental Table 1). We tested 93 of the 393 possible drug pool combinations ( Figure 3A; Supplemental Table 1). As above, animals from dmd+/-crosses were treated from 1-4 days and then scored for muscle birefringence. Each pool was tested in duplicate. We determined the average percentage of affected animals from each drug pool, and from each DMSO control tested, and we plotted the results as a heat map on our drug pool grid ( Figure 3A). This heat map grid reveals that, while no specific individual drug well appears to have a strong effect on reducing the percentage of animals affected (horizontal rows in Figure 3A), the majority of pools tested that included Plate 2 Row A reduce the percentage of animals affected (vertical column 5 in Figure 3A). To test whether the Plate 2 Row A pools are significantly improving the percentage of animals affected, we plotted the combined results from all pools containing each plate row compared with the combined results from DMSO control treatments ( Figure 3B). This analysis shows that treatment pools using Plate 2 Row A led to a significant reduction in the percentage of affected animals relative to the DMSO controls ( Figure 3B). On average, treatment with drug pools that included Plate 2 Row A resulted in only about 10% affected animals, compared to the 25% affected average observed in the DMSO controls ( Figure 3B). Figure 3C-D shows an example of the improved birefringence images from animals treated with a representative drug pool from the Plate 2 Row A pools (Pool 59), compared to sibling DMSO controls. While these results do not rule out beneficial effects of compounds or treatment pools from our screen that did not include Plate 2 Row A, these results do highlight that compounds within Plate 2 Row A are having significant beneficial effects on dmd zebrafish.

Analysis of Plate 2 Row A drugs identifies the beneficial combination of oxamflatin and salermide
In order to determine which drug or drugs are mediating the effect of Plate 2 Row A, we wanted an approach that would allow us to more quantitatively assess the effects of drugs on dmd birefringence. We developed a method that uses gray value measurements to quantitate the brightness level of the muscle birefringence of each animal ( Figure 4A). For each animal, we outlined a unilateral area of trunk muscle birefringence and measured the average pixel brightness within that area ( Figure 4A). dmd animals show lower pixel intensity than their control siblings ( Figure 4B,E). As a test of this approach, we confirmed that TSA treatment significantly improved the average pixel brightness of dmd birefringence ( Figure 4D-E).
We next tested each drug from Plate 2 Row A to see if any of the individual drugs could improve the dmd birefringence phenotype. Animals from dmd+/-crosses were treated from 1-4 days and then scored for muscle birefringence. When tested individually, none of the drugs showed a significant effect on the DMSO-treated dmd muscle birefringence brightness ( Figure   4F). These results indicate that a combination of drugs from Plate 2 Row A is likely needed to improve the dmd birefringence phenotype.
We then tested for drugs required for the effects of Plate 2 Row A by removing individual drugs. We decided to test drug screen Pool 72 (Plate 2 Row A + Plate 1 Well C3; 11 total drugs), because this pool represented an average effect from the Plate 2 Row A pools ( Figure 3A-3B). As above, animals from dmd+/-crosses were treated from 1-4 days and then scored for muscle birefringence. Treatment with the full 11 drugs from Pool 72 showed significant improvement of the dmd birefringence using our quantitative analysis ( Figure 5A).
However, the removal of four different drugs (chaetocin, oxamflatin, salermide, and delphinidin chloride) each inhibited the ability of Pool 72 to improve dmd birefringence ( Figure 5A), indicating that these four drugs may be involved in producing the dmd rescue effect of Pool 72. We next focused on the roles of chaetocin, oxamflatin, and salermide. We decided to not further test the role of delphinidin chloride because it was not a component of the Plate 2 Row A drugs and because it did not exhibit beneficial activity on its own (data not shown). To further test the roles of chaetocin, oxamflatin, and salermide in improving dmd muscle birefringence, we tested these three drugs together and each pair-wise combination ( Figure 5B). The combination of oxamflatin and salermide significantly improved dmd muscle birefringence, similar to Pool 72, whereas the other pair-wise combinations and the three-drug combination did not ( Figure 5B-F). These results show that the combination of oxamflatin and salermide mediate the beneficial effects of Pool 72 on improving dmd muscle birefringence.

An independent laboratory test validates the beneficial effects of oxamflatin+salermide
To validate these effects of oxamflatin and salermide, we performed treatments from 1 dpf-4 dpf at an independent site, in the laboratory of Dr. Henry. We see similar improvement of dmd muscle birefringence with oxamflatin+salermide treatments performed in the Henry Lab ( Figure 5G) as we do with treatments performed in the Maves Lab ( Figure 5B). This independent test is a strong validation of the benefits of this drug combination for dmd zebrafish.

Discussion
In this study, we performed a screen of epigenetic-modifying small molecules to identify compounds that prevented muscle degeneration in the zebrafish model of DMD. We developed a novel drug-pooling approach to screen combinations of epigenetic drugs, and we performed a pilot screen of 93 novel epigenetic drug pools. While our pilot screen identified several new candidate drug pools that have beneficial effects on dmd zebrafish ( Figure 3A), our drug-pooling strategy and subsequent analysis highlighted the activity of drugs from Plate 2 Row A (Figure 3).
We were able to resolve a novel drug combination, oxamflatin and salermide, that significantly reduced the number of dmd animals with muscle lesions when compared to DMSO-treated controls. Our results provide support for continued efforts in screening epigenetic small molecules for their beneficial effects in dmd zebrafish, and they support further investigations of oxamflatin and salermide as potential therapeutic compounds for DMD.
The development of significantly beneficial therapies for DMD will likely involve the simultaneous use of a combination of drugs that target dystrophin and/or downstream pathological mechanisms (Guiraud and Davies, 2017;Hoffman, 2019). Epigenetic drugs may represent a promising component of a DMD combination therapy. Epigenetic drugs are outstanding candidates for a DMD pharmacological therapy for many reasons. Certain epigenetic drugs have been shown to benefit DMD and other neuromuscular disorders (Campbell et al., 2017;Consalvi et al., 2014;Gordon et al., 2013;Riessland et al., 2010). Several HDACi, including SAHA, and other epigenetic drugs are already FDA-approved for use in cancers (Abdelfatah et al., 2016;Ahuja et al., 2016;Bai et al., 2019). While some epigenetic drugs have off-target or solubility issues, there is potential for optimizing dosing or using these drugs as lead compounds for further analyses (Abdelfatah et al., 2016;Bai et al., 2019). In addition, there are growing numbers of examples showing that epigenetic small molecules synergize with each other or with other drugs to target different diseases, including cancer and heart disease (Abdelfatah et al., 2016;Ahuja et al., 2016;Alexanian et al., 2019;Bai et al., 2019;Huber et al., 2011;Mazur et al., 2015). In our work here, we show an example of a combination of two epigenetic drugs that are beneficial for dmd zebrafish. Because of the advantages of the zebrafish animal model, we expect that dmd zebrafish will be an increasingly important model for investigating combination therapies for DMD.
Oxamflatin is an HDACi that inhibits Class I and II HDACs and is chemically similar to TSA (Huber et al., 2011;Kim et al., 1999). Salermide is a Class III HDACi that inhibits the NAD+-dependent deacetylases SIRT1 and SIRT2 (Lara et al., 2009), and so represents a new class of HDACi for DMD. In previous studies, overexpression of SIRT1 has been shown to ameliorate DMD pathophysiology in mdx mice (Chalkiadaki et al., 2014), while salermide inhibition of SIRT1 has been shown to protect muscle cells against oculopharyngeal muscular dystrophy in Caenorhabditis elegans (Pasco et al., 2010). SIRT1 and SIRT2 can have opposing effects in promoting angiogenesis and in providing neuroprotective effects in neurodegenerative disease models (Keskin-Aktan et al., 2018;Shi et al., 2014). It will be important to determine the mechanisms by which salermide and oxamflatin are together providing skeletal muscle benefits in dmd zebrafish.
In conclusion, we developed a novel approach for screening combinations of small molecules in dmd zebrafish, and we identified a promising new combination of epigenetic compounds, oxamflatin and salermide. Future studies will include screening larger libraries of epigenetic small molecules in dmd zebrafish and pursuing validations of additional candidate drug pools that were identified in our pilot screen. Also, a critical next step will be to validate the effects of oxamflatin and salermide in mammalian DMD models, such as mdx mice, the Dmd rat, or human induced pluripotent stem cell models of DMD (Choi et al., 2016;reviewed in McGreevy et al., 2015;and Wells, 2018). In this work, we have already taken an important step in small molecule validation for DMD by demonstrating that oxamflatin and salermide show beneficial effects on dmd zebrafish in two independent laboratories. We predict that small molecule validation in independent laboratories and in multiple model systems will increase the potential for future translation of epigenetic drug therapies and other pharmacological approaches for DMD.