Development of an orally available inhibitor of CLK1 for skipping a mutated dystrophin exon in Duchenne muscular dystrophy

Duchenne muscular dystrophy (DMD) is a fatal progressive muscle-wasting disease. Various attempts are underway to convert severe DMD to a milder phenotype by modulating the splicing of the dystrophin gene and restoring its expression. In our previous study, we reported TG003, an inhibitor of CDC2-like kinase 1 (CLK1), as a splice-modifying compound for exon-skipping therapy; however, its metabolically unstable feature hinders clinical application. Here, we show an orally available inhibitor of CLK1, named TG693, which promoted the skipping of the endogenous mutated exon 31 in DMD patient-derived cells and increased the production of the functional exon 31-skipped dystrophin protein. Oral administration of TG693 to mice inhibited the phosphorylation of serine/arginine-rich proteins, which are the substrates of CLK1, and modulated pre-mRNA splicing in the skeletal muscle. Thus, TG693 is a splicing modulator for the mutated exon 31 of the dystrophin gene in vivo, possibly possessing therapeutic potential for DMD patients.

and the other lacking only exon 31. The dystrophin mRNA lacking the 111-bp exon 31 is in-frame to produce an internally deleted, but functional dystrophin protein. The expression of the functional truncated protein is enhanced by TG003 treatment.
Pharmacological efficacy in target tissues is an unavoidable issue for developing therapeutic modalities. While TG003 was effective in modulating CLK activity and mRNA splicing in cellular models of DMD 8 , its metabolic instability hinders its clinical application. In this study, we identified an orally available selective CLK1 inhibitor, TG693, which promoted exon skipping and production of a truncated functional dystrophin protein in immortalized cells derived from a patient with DMD (c.4303G > T) to the same extent as TG003. Additionally, TG693 was metabolically stable and transiently detected in the skeletal muscle at a bioactive level, supporting its therapeutic potential as an orally available drug candidate.

Results
Development of the metabolically stable CLK1 inhibitor TG693. We screened for stable exon-skipping inducers of the mutant DMD exon 31 (c.4303G > T) in our original chemical library and identified the small molecule 5-(4-pyridinyl)-1H-indazole (henceforth referred to as TG693) as an exon-skipping inducer (Fig. 1a). To test the in vivo stability of TG003 and TG693, mice were administered a single subcutaneous injection of either compound and then serum concentrations were monitored by liquid chromatography-mass spectrometry (LC/ MS) at serial time points. While only trace amounts of TG003 were detected, the average TG693 serum concentration was 13 μ M at 6 h post-injection, indicating that TG693 was more stable than TG003 in the blood plasma (Fig. 1b). Subsequent analyses on CLK1 inhibition with in vitro kinase assays revealed that TG693 inhibited the phosphorylation of a synthetic SRSF1 RS domain peptide with a half-maximal inhibitory concentration (IC 50 ) of 112.6 nM, approximately 10-fold weaker than that of TG003 (IC 50 = 13.1 nM) (Fig. 1c). We also examined whether TG693 served as an ATP-competitive inhibitor. Hanes-Woolf plots showed parallel lines with y-intercepts that correspond to TG693 concentration, demonstrating that TG693 competes with ATP for a single site in CLK1 ( Fig. 1d; K m = 14.88 μ M, K I = 105.3 nM). Moreover, kinetic parameters (K m apparent , V max apparent , α , α ') were determined by Michaelis-Menten model fitting (Supplementary Table S1), which indicated that TG693 was an ATP-competitive type I inhibitor that antagonized CLK1. Further, in order to evaluate the specificity of TG693mediated kinase, we investigated the in vitro inhibitory activity of 1 μ M TG693 against a panel of 313 recombinant kinases. TG693 inhibited CLK1 and Haspin activity by over 90% and kinases over 50% inhibition were shown in a kinase dendrogram (Fig. 1e). These suggested that TG693 was a potent and selective inhibitor of CLK1.
TG693 inhibits CLK1 kinase activity in cells. TG003 administration to cultured cells inhibits the CLK1-mediated phosphorylation of SR proteins, particularly that of SRSF4 17 . To examine the cellular activity of TG693, HeLa cells were treated with increasing concentrations of compounds, and then SR protein phosphorylation was assessed by western blotting with an anti-pan-phospho-SR antibody that recognizes canonical SR protein family members 25 . Notably, TG693 and TG003 inhibited the phosphorylation of SRSF4 at 5 μ M and SRSF6 at 20 μ M and 10 μ M, respectively (Fig. 2a), confirming that these compounds act via the same biochemical mechanism.
TG693 induces exon-skipping of a DMD splicing reporter. Since TG693 was shown to be a metabolically stable CLK1 inhibitor, we evaluated the dose-dependent effects of TG693 on the skipping of mutant DMD exon 31 (c.4303G > T) with HeLa cells expressing a H492-Dys Ex31m splicing reporter 8 . Interestingly, the effects of TG693 and TG003 on the exon-skipping were comparable, despite their differences in suppressing CLK1 activity in in vitro kinase assays (Fig. 2b, upper panel); and TG693 had no effect on wild type transcript splicing even at 30 μ M (Fig. 2c). To determine if this activity was specific to exon 31 mutation, we then evaluated the function of TG693 on exon-skipping with the H492-dysEx27m splicing reporter containing a point deletion in exon 27 (c.3613delG (p.Glu1205LysfsX9)) that causes a frameshift-induced premature termination codon. As shown in Supplementary Fig. S1, TG693 induced exon 27-skipping from H492-dysEx27m reporter, similar to that previously observed with TG003 8 .
CLK1 pre-mRNA splicing is regulated by CLK1 kinase activity 14 . In this mechanism, abundant CLK1 levels induce SR phosphorylation, resulting in the removal of exon 4 from CLK1 pre-mRNA and nonsense-mediated mRNA decay 26,27 . In previous reports, administration of TG003 to cultured cells promoted full-length CLK1 expression 14,17 . Consistently, CLK1 exon 4-skipped transcript expression was attenuated in the presence of 5 μ M TG693 (Fig. 2b, middle panel), suggesting that TG693 exhibited a considerable inhibitory effect on CLK1 and concomitant exon-skipping, similar to that of TG003 8 .

TG693 restored dystrophin expression in patient-derived cells.
We previously reported that TG003 restored dystrophin production in patient-derived myotubes 8 . Therefore, we investigated the effect of TG693 on dystrophin expression in an immortalized cell line derived from a DMD patient with the c.4303G > T mutation, which was by established by the forced expression of constitutively active cyclin-dependent kinase 4, cyclin D1, and telomerase 28,29 . Notably, TG693 promoted the skipping of mutant exon 31 (Fig. 3a), but did not affect any other exons (Supplementary Fig. S2 and Table S4). Moreover, TG693 was sufficient to increase dystrophin protein expression in a dose-dependent manner with a 130% and 200% increase at 10 and 20 μ M, respectively    Oral administration of TG693 modulated splicing in mouse skeletal muscle. To analyze the in vivo effects of TG693, mice were orally administered TG693 (30 mg kg −1 ). Notably, a significant amount of TG693 (> 4 μ M) was detected in the tibialis anterior (TA) muscle of mice (Fig. 4a), as well as a marked concomitant reduction of SR protein phosphorylation, particularly that of SRSF4 (Fig. 4b). Further analysis of splicing modulation showed that TG693 promoted full-length Clk1 transcript expression in TA, suggesting that treatment suppressed the excision of exon 4 induced by CLK1 activity (Fig. 4c). TG693 was active in other muscle tissues, such as the heart and diaphragm, as well as skeletal muscle (Fig. 4d). We also confirmed that TG693 had no apparent toxicity in rats at up to 100 mg kg −1 per os ( Supplementary Fig. S3). No mortalities or gross abnormalities were observed in TG693-treated animals during the experimental period, nor did it affect body weight in mice administrated an overdose of the compound. Thus, these results suggest that TG693 is a useful tool for inhibition of CLK1 activity in vivo per os.

Discussion
We succeeded in identifying a selective ATP-competitive inhibitor of CLK1, named TG693. TG693 is structurally distinct from TG003 (Fig. 1a), but it similarly shows potent CLK1 inhibition as well as dystrophin exon 31 skipping enhancement in cells ( Fig. 2a and b), strengthening our original hypothesis that CLK1 can be the molecular target for splicing manipulation 8,14 . TG693 restored the dystrophin protein expression in cells harboring a point mutation (c.4303G > T) in exon 31 of the dystrophin gene by inducing exon-skipping (Fig. 3). The c.4303G > T point mutation in exon 31 attenuates exon recognition by disruption and creation of splicing elements such as exonic splicing enhancer and silencer 8,30,31 . Thus, mutated exon 31 is targeted by TG693 to suppress exon recognition without altering wild type transcript splicing (Fig. 2c). We also showed TG693 promotes skipping of a mutated exon 27 in another patient ( Supplementary Fig. S1). In vivo mouse experiments revealed that oral TG693 administration decreased SR protein phosphorylation in muscle tissue and modulated the splicing pattern of endogenous Clk1 (Fig. 4). Thus, these results indicated that TG693 could modulate the splicing pattern of the mutated dystrophin gene in the patient by oral administration. This represents a significant advance in drug-induced exon skipping therapy for DMD.
Splicing modulation has drawn increasing attention as a therapeutic strategy for DMD. In 2016, two antisense drugs that modulate splicing were approved by the U.S. Food and Drug Administration (FDA) (e.g. Eteplirsen and Nusinersen), and exon-skipping therapy has become a reality. While AONs offer a specific way to modulate splicing in a therapeutic manner, some hurdles must be overcome for the orally available antisense drugs 32,33 . Compared to AONs, small chemical compounds may not be as specific, but more advantageous for the synthesis cost and the oral availability. Thus, splicing modulation with small molecules can become an alternative therapeutic approach in addition to AONs, and can expand applications of splicing modulation therapy. CLK1 inhibitors have attracted extensive attention in recent years. Notably, TG693 is the first and only CLK1 inhibitor which is orally available and metabolically stable in blood 34,35 . As such, we believe that the present study will provide a foundation for the continued development of therapies other diseases associated with aberrant CLK1 activity-such as Alzheimer's and Influenza 36,37 .
Our results show that TG693 mediated CLK1 inhibition elicits changes in SR protein phosphorylation, particularly that of SRSF4 and SRSF6 (Figs 2a and 4b). SR proteins regulate a multitude of splicing events in the phosphorylation dependent manner 21 , raising the possibility that reduction of SR proteins phosphorylation may cause off-target splicing alteration. Therefore, we previously checked transcriptional wide effect of TG003, which similarly inhibits phosphorylation of SR proteins as TG693 in vitro (Fig. 2a), and we found that only 0.3% (110/37487) 24 of registered alternative splicing events were skipped enhanced by TG003. TG693 is fairly specific in its inhibition profile, showing significant off target effects (> 70%) to only 4 of 313 kinases (Fig. 1e). This solidifies the mode-of-action evidence for TG693. Although we confirmed TG693 had no apparent acute toxicity in rats at up to 100 mg kg −1 per os ( Supplementary Fig. S3), more preclinical studies with other animal models are necessary to consider the clinical application of TG693. We have also started a project to seek next generation CLK1 inhibitors with a pharmaceutical company to improve the efficacy and selectivity of TG693 based on the information written here.
TG693 is a simple compound with indazole and pyridine groups (Fig. 1a) far more metabolically stable than TG003 which is a benzothiazole compound. We believe that TG003 is easily metabolized via O-demethylation and subsequent sulfate conjugation of the hydroxyl group 38 and likely interacts with CLK1 via hydrogen bonds in the ATP-binding pocket 39 . In the case of TG693, metabolically stable indazole moiety is suspected to form a hydrogen bond with an amino acid residues of CLK1 40 , but structural studies will be necessary to confirm this hypothesis.
Our results suggest that exon-skipping therapy with orally available small molecule has therapeutic potential to genetic diseases caused by disruption/creation of splicing elements, including Duchenne muscular dystrophy. We believe that our approach is potentially applicable to other genetic diseases caused by nonsense or frameshift mutations in the internal coding exons whose length is a multiple of three. On this criteria, we found 6,244 pathogenic mutations in the NIH ClinVar database that may be target candidates for exon-skipping therapy with TG693. Our efforts have focused on expanding application of TG693 to other diseases, addition to the Duchenne muscular dystrophy.

Methods
Materials. TG693 was synthesized as detailed in the Supporting Information. TG003 was prepared as described previously 14 . Both compounds were dissolved in dimethylsulfoxide (DMSO) (Hybri-MAX TM , Sigma-Aldrich, St. Louis, MO, USA). Carboxymethyl cellulose sodium salt was obtained from Nacalai Tesque (Kyoto, Japan).
Antibodies. The source and catalog numbers of commercially available primary antibodies have been provided in Supplemental Table 2. Mouse monoclonal antibodies against phospho-SR proteins (1H4) and α -tubulin (DM1A) were purchased from Invitrogen (Carlsbad, CA, USA) and Thermo Fisher Scientific (Yokohama, Japan), respectively. Rabbit polyclonal antibody against dystrophin was obtained from Abcam (Cambridge, UK). Goat polyclonal antibody against Lamin B (M-20) was purchased from Santa Cruz Biotechnology (Santa Cruz, Dallas, TX, USA). HRP-conjugated anti-rabbit, anti-goat IgG, and anti-mouse IgG secondary antibodies were purchased from GE Healthcare Life Sciences (Pittsburgh, PA, USA), Jackson ImmunoResearch (West Grove, PA, USA), and Abcam, respectively.
In vitro kinase assay. CLK1 kinase activity was assessed as previously reported 14 with minor modifications.
The reaction mixture containing serially diluted inhibitors, 10 mM MOPS-KOH (pH 6.5), 10 mM magnesium chloride, 200 μ M EDTA, 1 μ M ATP, 0.167 μ Ci of [γ -32 P] ATP, 0.417 μ g of synthetic RS peptide, and recombinant GST-tagged human CLK1 (Cat #04-126, Carna Biosciences, Kobe, Japan) was prepared in a final volume of 25 μ L. The reaction mixture was incubated at 30 °C for 30 min, and phosphoric acid was added at a final concentration of 5% to stop the reaction. Then, 25 μ L of the reaction mixture was dispensed on a P81 phosphocellulose membrane (Cat #20-134, Merck Millipore, Darmstadt, Germany) and washed four times in 5% phosphoric acid. Cherenkov light from the incorporated 32 P was measured using a liquid scintillation counter. The kinase assay conditions, including the incubation period and concentrations of kinases and substrates, were optimized to maintain linearity during incubation. The net radioactivity was determined by subtracting the background count from the reaction mixture without kinases. The IC 50 of each compound was calculated by interpolation on a log-concentration-response curve fitted with a four-parameter logistic equation.
Kinase profiling. The effect of 1 μ M TG693 on 313 kinases (listed in Supplementary Table S2) was examined with immobilized metal ion affinity-based fluorescence polarization (IMAP) assay or off-chip mobility shift assay (MSA) depending on the kinase being examined with an ATP concentration at the K m or 1 mM. These assays rely on the phosphorylation of small synthetic peptides developed for each kinase, and detect activity by changes in electrophoretic mobility (MSA assay) or by binding of a fluorescently labelled peptide to a large micro-particle and concomitant polarization increases (IMAP assay). Reaction conditions for each kinase are described in Supplementary Table S5. All kinase assays were carried out at Carna Biosciences (Kobe, Japan). The inhibitory map was constructed using Kinome Render 41 .
Immobilized metal ion affinity-based fluorescence polarization (IMAP) Assay. Substrate/ATP/ metal (4× ) and kinase (2× ) solutions were prepared using with assay buffer (20 mM HEPES, 0.01% Tween-20, 2 mM DTT, pH 7.4), and then mixed in 384-well black plates for 1 hour at room temperature. IMAP binding solution (IMAP Screening Express kit; Molecular Devices, Sunnyvale, CA, USA) was added to each well and incubated for 30 min. The level of kinase activity was then evaluated by fluorescence polarization at 485/530 nm (Ex/Em).

Off-Chip Mobility Shift Assay (MSA).
Inhibitor (4× ), substrate/ATP-Metal (4× ), and kinase (2× ) solutions were prepared with assay buffer (20 mM HEPES, 0.01% Triton X-100, 2 mM DTT, pH 7.5) and mixed in 384-well plates at room temperature for 1 or 5 h depending on the kinase. Reactions were stopped by the addition of termination buffer (QuickScout Screening Assist MSA; Carna Biosciences). The entire reaction mixture was then applied to a LabChip system (Perkin-Elmer, Waltham, MA, USA) and the product and substrate peptide peaks were separated and quantified, and then heights of product (P) and substrate (S) peptide peaks were calculated. The level of kinase activity was evaluated with the following equation: activity = P/(P + S).
ATP kinetics assay. Kinetic reactions were performed with the same method used for in vitro kinase assays, except that 0.05-0.4 μ Ci of [γ -32 P] ATP was used. K m and K i values were calculated with a competitive inhibition model in Prism 6 software (GraphPad Software, San Diego, CA, USA).
Western blot analysis. Proteins were extracted from HeLa cells, immortalized DMD cells, and mouse muscle tissue using CelLytic M (Sigma-Aldrich, St. Louis, MO, USA), 1× Cell Lysis Buffer (Cell Signaling Technology, Danvers, MA, USA), and CelLytic MT (Sigma-Aldrich), respectively. All buffers were supplemented with protease and phosphatase inhibitor cocktail (#25955 and #0757461, Nacalai). Protein concentrations were quantified with Protein Assay Reagent (Pierce by Thermo Fisher Scientific). The HeLa cell and mouse tissue lysates were mixed with Sample buffer (#0949914, Nacalai), denatured at 95 °C for 5 min, and then electrophoresed in a 10% SuperSep TM Ace gel (Wako). The lysates of immortalized cells were mixed with NuPAGE LDS Sample buffer (Thermo Scientific Fisher), denatured at 70 °C for 10 min, and electrophoresed in a 3-8% NuPAGE Novex Tris-Acetate gels (Thermo Scientific Fisher). The samples were then transferred to PVDF membranes (Pall Corporation, Port Washington, NY, USA). Antibody reactions were performed with Can Get Signal ® Immunoreaction Enhancer Solution (Toyobo, Osaka, Japan). Peroxidase activities were visualized with ImmunoStar ® LD (Wako) and a ChemiDoc TM MP Imaging System (Bio-Rad).
Mice and TG693/TG003 administration. Seven-week-old, male Jcl:TCR mice were obtained from Charles River Laboratories (Shiga, Japan). All procedures were performed in accordance with the US National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee of Kyoto University Graduate School of Medicine. Mice were subcutaneously injected with 30 mg kg −1 TG693 or TG003 suspended in 5% DMSO, 5% Solutol, 9% Tween-80, and 81% saline. TG693 was also administered orally at a dose of 30 mg kg −1 suspended in 0.5% carboxymethylcellulose. In vivo pharmacokinetic analysis. TG693 metabolism and tissue bioavailability were assessed in mouse plasma and muscle tissue. To collect plasma, mice were anesthetized with isoflurane (Mylan, Osaka, Japan) at the indicated time point. Whole blood samples were collected, allowed to clot, and then centrifuged to isolate serum. TG693 or TG003 serum levels were analyzed by LC/MS with an Agilent 6420 Q-TOF mass spectrometer (Agilent Technologies) and ZORBAX HILIC Plus (TG693) or ZORBAX Eclipse Plus C18 (TG003) columns (Agilent Technologies). For muscle tissue bioavailability, mice were sacrificed by cervical dislocation and the tibialis anterior (TA) muscles, heart, and diaphragm dissected. TA homogenates were prepared using a Bead Crusher μ T-12 shaking machine (Taitec, Kyoto, Japan) in saline and used to assess TG693 levels by LC/MS with a Poroshell 120 PFP (Agilent Technologies).
Statistical analysis. Statistical analysis was performed using Student's t-tests, except for the comparison of exon-skipping ratio in patient-derived cells analyzed by one-way ANOVA with independent post hoc Tukey's multiple comparison testing. Results are reported as the means ± SD. Statistical significance was defined as p < 0.05. IC 50 values were calculated from data fitted to a four-parameter logistic curve (variable slope) in Prism 6.0 software.