Synthesis of SMT022357 enantiomers and in vivo evaluation in a Duchenne muscular dystrophy mouse model

Following on from ezutromid, the ﬁ rst-in-class benzoxazole utrophin modulator that progressed to Phase 2 clinical trials for the treatment of Duchenne muscular dystrophy, a new chemotype was designed to optimise its physicochemical and ADME pro ﬁ le. Herein we report the synthesis of SMT022357, a second generation utrophin modulator preclinical candidate, and an asymmetric synthesis of its constituent enantiomers. The pharmacological properties of both enantiomers were evaluated in vitro and in vivo . No signi ﬁ cant difference in the activity or ef ﬁ cacy was observed between the two enantiomers; activity was found to be comparable to the racemic mixture.


Introduction
Duchenne muscular dystrophy (DMD) is an X linked, recessive, fatal, muscle-wasting disease [1]. This disorder affects 1 in 5000 boys and its incidence is continuing to rise in all populations worldwide [2]. Dystrophin is essential for muscle membrane stability as it acts as a linker between the internal cytoskeleton of the muscle cell and the extracellular matrix to form the dystrophinassociated protein complex (DAPC) at the sarcolemma [3]. DMD is caused by mutations in the DMD gene leading to the expression of non-functioning or truncated dystrophin. This in turn gives rise to a less stable DAPC and progressive muscle degeneration. Currently there is no cure for all DMD patients and treatments aim to lessen the symptoms of the disease without addressing the cause. Initial symptoms appear when patients are young children and loss of ambulation occurs in the early teens. Due to the wide-reaching implications of the disease on the human body, the condition usually results in death for patients in their twenties or thirties due to heart or respiratory failure [4,5].
Amongst the strategies being pursued to treat DMD, such as exon-skipping therapy [6] and viral gene therapy [7], the upregulation of utrophin production could provide a long-term treatment for all patients with DMD. In mouse and dog models of the disease, it has been found that the lack of dystrophin could be compensated for by the increased production of the related protein utrophin. Therefore, the administration of small molecules which transcriptionally upregulate the production of utrophin could be an effective treatment for DMD [8].
Utrophin is structurally similar to dystrophin and is able to provide stability to muscle cells in a similar role to dystrophin but at different stages of fibre maturity. This has the advantage over the other treatments in development in that it should be successful in treating all patients, regardless of the genetic mutation responsible for their illness [9], and it does not require immunosuppression [7]. In this context, ezutromid ( Fig. 1), developed by Summit Therapeutics plc, has been the first orally bioavailable small molecule specifically designed to target the utrophin-A promoter. Unfortunately, despite the encouraging interim 24-week data where evidence of target engagement were observed (reduced muscle fibre damage in combination with an increase in utrophin protein), ezutromid was discontinued in June 2018 after primary and secondary endpoints were missed at the conclusion of a Phase 2 trial [10].
In parallel to ezutromid's clinical evaluation, second generation utrophin modulators were also being developed. This new compound series arose from bioisosteric replacement of the sulfone moiety with a phosphinate ester and conferred improved physicochemical properties and a more stable metabolic profile [8]. In particular SMT022357 was found to exhibit in vivo efficacy in the mdx mouse (2-week-old mice, orally dosed daily with 30 mg/kg SMT022357 or vehicle for 5 weeks) leading to increased utrophin expression in skeletal and cardiac muscles [8].
Herein, we report the synthesis of SMT022357, its resolution into, and de novo enantioselective synthesis of, its constituent enantiomers and the preliminary results of their in vivo efficacy.
In an attempt to produce significant quantities and to define the absolute configurations of the two enantiomers, an asymmetric route was used. Although several synthetic strategies have been developed in this field, one of the most common involves the use of chiral auxiliaries. Chiral pool precursors including ephedrine [11], menthol [12,13], and prolinol derivatives [14e16] have been used for inducing asymmetric induction to prochiral organophosphorus compounds. In our case, the absolute configuration of the P-stereocentre was determined by using enantiopure prolinols to afford the corresponding oxazaphospholidine oxide (Scheme 2). This method, first reported by Koizumi and co-workers, has been shown to be a simple but effective way to control selectivity in the P-centre [17]. Moreover, the same group has proved that the resulting cyclic oxazaphospholidine oxide reacts stereoselectively with different Cand O-nucleophiles, providing the corresponding phosphinates in enantioenriched form [18].
Two chiral auxiliaries were used to explore potential improvements in crystallinity, stability, and/or higher levels of diastereoselectivity in the nucleophilic displacement reaction at P. As a result, compound 7 was reacted with either (S)-prolinol 8a or a,a 0dimethyl-(S)-prolinol 8b in the presence of triethylamine, in THF. The resulting cyclic oxazaphospholidine oxides 9a and 9b were obtained with diastereomeric ratios varying from 60:40 to 55:45, depending on the nature of the chiral auxiliary (Scheme 2B). Surprisingly, no improvement in diastereoselectivity was observed by using the more hindered gem-disubstituted prolinol 8b.
It is worth noting that both diastereoisomers ((S P , S)-9a 0 and (S P , S)-9b 0 ) spontaneously decomposed over a few days at ambient temperature leaving the diastereoisomers ((R P , S)-9a and (R P , S)-9b) as the only final products in 13% and 21% yields, respectively (See Scheme 2 and Fig. 2).
The relative configuration within the major diastereoisomer (R P , S)-9b was established unambiguously via single crystal X-ray diffraction analysis of the solvate (R P , S)-9b$CHCl 3 [19]. The absolute (R P , S)-configuration of 9b was assigned by reference to the known (S)-configuration of the prolinol auxiliary, and the determination of a Flack x parameter [20,21] of þ0.013(10) for the structure of (R P , S)-9b$CHCl 3 confirmed this assignment ( Fig. 2A). The relative configuration within oxazaphospholidine (R P , S)-9a (Fig. 2B) was initially assigned by analogy to (R P , S)-9b and later supported by NOESY experiments (Supplementary Data). Given the rigid bicyclic structure of the oxazaphospholidine, it is plausible that protons H 4' and H 6' would display NOE signals with either H 3a or H 6b , depending on the absolute configuration of the phosphorous atom. The NOESY spectrum exhibited NOE signals between H 4' and H 6' and H 3a . No cross-peak with H 6b or any other hydrogens of the Re face of the prolinol ring was observed, consistent with the equatorial orientation of the benzoxazole ring (Fig. 2B). With data to support the absolute and relative configurations of bicyclic oxaphospholidines 9a and 9b, ethylation of the two compounds was attempted (Scheme 3). Unfortunately, no alkylation of diastereoisomer (R P , S)-9b was achieved despite trialing a number of experimental conditions. When (R P , S)-9a was treated with EtMgBr in THF at rt for 16 h, the corresponding amidate was obtained in almost quantitative yield and in diastereomerically pure form.
The resulting phosphinamide of (R P , S)-9a was then subjected to acid-mediated methanolysis affording (þ)-(R P )-4 in 83% ee and 35% overall yield. The absolute configuration of (þ)-(R P )-4 was assigned by analogy with similar examples reported in the literature where a stereospecific overall inversion of configuration at phosphorus was unambiguously demonstrated [14,18]. The comparison of the HPLC traces and the comparable values for specific rotation supports enantiomer 1 as the enantiomer (þ)-(R P )-4 and enantiomer 2 as (À)-(S P )-4, enantiomer 1 and enantiomer 2 refers to the enantiomers' relative retention times, Supplementary material.

In vitro and in vivo evaluation
Examination of the overall profile of SMT022357 and its  Determination of the relative configurations of (R P , S)-9b via single crystal X-ray diffraction analysis (2A) and (R P , S)-9a via NOESY analysis, observed NOE signals are depicted in red (2Bi); predicted NOE signals from diastereomer (S P , S)-9a 0 which were not observed in the NOESY analysis (2Bii).
Scheme 3. Synthesis of phosphinate (þ)-(R P )-4 from (R P , S)-9a by double inversion of configuration [14,18] enantiomers showed there are no significant differences between the biological activity of the two enantiomers and the properties were generally similar to those of the racemic mixture (Table 1). More specifically, the compound's ability to increase utrophin's transcription was assessed in the screening cell line H2K-mdx utrnA-luc, stably transfected with a 8.5 kb fragment of the human utrophin promoter linked to a firefly luciferase reporter gene [8,22]. The two enantiomers showed low micromolar activity in the reporter gene assay, which was comparable to that of the racemic mixture (Table 1). Similarly to the structurally related ezutromid [23], SMT022357 was found to inhibit firefly luciferase, though to a lesser extent. For this reason an orthogonal assay was used to verify the compounds' activities. A homogenous time-resolved fluorescence (HTRF) assay was used to quantify human utrophin protein in cell lysates of iDMD cells, an immortalised muscle cell line derived from a DMD patient. The racemic mixture and the two enantiomers were found to increase the utrophin protein 1.2e1.3 fold (Supplementary data Fig. S1, Table S1). In terms of physicochemical and ADME properties, the three compounds have good aqueous solubility and cell permeability, are not very lipophilic and show similar binding to plasma proteins. In addition, they are metabolically stable in human, but less so in mouse hepatocytes (Table 1).
Previous studies with the racemic mixture SMT022357 showed that oral administration of this compound led to increased expression of utrophin in skeletal and cardiac muscles, improving sarcolemmal stability and preventing the dystrophic pathology in mdx mice [8]. With the two enantiomers of SMT022357 in hand, we wanted to establish if there were any significant differences in their in vivo activity. The effect of the two enantiomers ((þ)-4, (À)-4) was evaluated in 2 week old mdx mice, under the same protocol as before [8]. Briefly, the mice were dosed orally daily with 30 mg/kg ((þ)-4, or (À)-4 or vehicle) for 5 weeks. Immunofluorescence with a utrophin antibody revealed a higher utrophin protein signal after compound treatment and importantly that utrophin is localised at the sarcolemma in the extensor digitorum longus (EDL) muscle (Fig. S2).
Interestingly, the two enantiomers increased utrophin to a similar extent as demonstrated by quantification of the utrophin protein measured by western blot analysis. A 1.6-fold and 1.5-fold increase were observed for (þ)-4 and (À)-4, respectively (Fig. 3A). This is in good agreement with the results obtained with the racemic mixture in our previous study (1.8-fold increase of utrophin in EDL muscle) [8].
Next, we investigated functional improvements in the muscle upon administration of (þ)-4 and (À)-4. Dystrophin-deficient muscles are unable to sustain the mechanical stress induced by forced eccentric contractions and show an excessive force drop which is probably responsible for the activity-dependent damage seen in dystrophic muscles [24,25]. For this reason, the reduction of the force drop caused by eccentric contractions has been used as a quantitative test to measure the efficacy of therapeutic interventions, such as utrophin modulation. Ex vivo analysis of EDL muscles showed a substantial 34.8% (p ¼ 0.00057) and 24.8% (p ¼ 0.004) decrease in force drop compared with vehicle in (þ)-4 and (À)-4 treated mdx mice, respectively (Fig. 3B). Taking into consideration that force is reduced only by 13.6% in wild type mice and up to 64.2% in mdx mice [26] (48.8% in the present experiment, Fig. 3) it is apparent that both enantiomers (þ)-4 and (À)-4 significantly improved EDL muscle pathophysiology and ameliorated the DMD phenotype in mdx mice. These results, are consistent with the in vivo efficacy data previously reported for SMT022357 [8].

Conclusions
A second generation potential preclinical candidate for Duchenne muscular dystrophy with an improved physicochemical and ADME profile was presented. As the new analogue has a P- Table 1 In vitro biological activity and ADME profiles of SMT022357 (±)-4 and enantiomers.  stereocentre, asymmetric synthesis was attempted to establish the absolute configurations of the two enantiomers. Further evaluation of the enantiomers in vitro and in vivo established that there are no significant differences in the biological and ADME profiles of the two. All in all, SMT022357 and enantiomers appear to have an improved ADME profile compared to ezutromid and improve the dystrophic phenotype in mdx mice.

General information
All NMR spectra were recorded on Bruker AV400 and AV500 spectrometers. 1 H and 13 C NMR spectra are reported as chemical shifts (d) in parts per million (ppm) relative to the solvent peak using the Bruker internal referencing procedure (edlock). 19 F NMR spectra are referenced relative to CFCl 3 in CDCl 3 . Coupling constants (J) are reported in units of hertz (Hz). The following abbreviations are used to describe multiplicities e s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet), br. (broad). High resolution mass spectra (HRMS, m/z) were recorded on a Bruker MicroTOF spectrometer using positive electrospray ionization (ESI þ ), on a Micromass GCT spectrometer using filed ionization (FI þ ) or chemical ionization (CI þ ), or on a Waters GCT Classic GCMS using electron impact ionization (EI). Infrared spectra were recorded either as the neat compound or in a solution using a Bruker Tensor 27 FT-IR spectrometer. Absorptions are reported in wavenumbers (cm À1 ). Melting points of solids were measured on an EZ-Melt apparatus and are uncorrected. Solvents were purchased from Sigma-Aldrich, Honeywell or Fisher. When dry solvents were required they were purified by expression through an activated alumina column built according to the procedures described by Pangborn and Grubbs. Chemicals were purchased from Acros, Alfa Aesar, Fisher, Fluorochem, Sigma-Aldrich and used as received. Reactions were monitored by thin-layer chromatography (TLC) carried out on Merck Kiesegel 60 F254 plates, silica gel column chromatography was performed over Merck silica gel C60 (40e60 mm).

HTRF assay: utrophin quantification (Evotec)
The human utrophin HTRF kit was obtained from Cisbio Bioassay (Parc Marcel Boiteux, 30200 Codolet, France). In this, utrophin is detected in a sandwich assay format using two different specific antibodies, one labelled with Eu 3þ -Cryptate (donor) and the second with d2 (acceptor). When the donor/acceptor pair is in close proximity, the excitation of the donor with a light source (laser or flash lamp) triggers a Fluorescence Resonance Energy Transfer (FRET) towards the acceptor, which in turn fluoresces at a specific wavelength (665 nm). The measurement of HTRF emissions at two different wavelengths (620 nm for the donor and 665 nm for the acceptor) allows the ratiometric reduction of data to correct for well-to-well variability and signal quenching from assay components and media.
The specific signal modulates positively in proportion to human utrophin. The assay is run under a two-plate assay protocol as follows: cells are plated, stimulated and lysed in a first culture plate, then lysates are transferred to the assay plate for utrophin detection using HTRF reagents. This protocol gives the ability to monitor the cell viability and confluence. . Add 17 mL of fresh growth medium to homogenize the suspension. Cells are centrifuged (5 min, 1200 rpm) and suspended in 15 mL of fresh growth medium. Add the proper volume of cell suspension (1/3 to 1/10) to a new T225 cm 2 flask. Cells were maintained at 5% CO 2 at 37 C, 90% RH.
The cells were plated in 384-well plates (50 mL/well, 8000 cells/ well, 5 replicates for each condition evaluated). After 24 h at 37 C, 5% CO 2 , 90% RH the cells were serum starved to get synchronised and the next day the cells were dosed with compounds in triplicate from stock solutions (medium, 0.3% DMSO, cpd 10 mM) and incubated for 120 h. The cells were then lysed, a mix of the two antibodies (Eu 3þ /d 2 ) was added and incubated for further 24 h. Timeresolved fluorescence was measured on a EnVision™ 2103 Multilabel plate reader set up for Eu 3þ -cryptate and fluorescence emission was measured at 665 nm and 620 nm.
Results are expressed as HTRF ratio (mean ± standard deviation). Analysis of the raw data was performed using GraphPad Prism 7 software, and data was then compared to the untreated cells. Calculations were performed on Excel software. Statistical analysis was performed with GraphPad Prism 7 software using the unpaired t-test (2 biological replicates, 3 technical replicates each).

Mice and drug treatment
15 day-old male mdx (C57BL/10ScSn-Dmdmdx/J; Charles River Laboratories, MA, USA) littermates were randomly assigned and treated with (±)-4, (þ)-4, (À)-4 (30 mg/kg) or vehicle only [phosphate buffered saline (PBS), 0.1% Tween-20, 5% DMSO] by daily intraperitoneal injection for 1 week and then daily oral gavage for a further 4 weeks. At the end of drug treatment, mice were sacrificed by CO 2 asphyxiation in accordance with Schedule I of the UK Animals (Scientific Procedures) Act 1986 and muscle and blood samples taken.

Histology
Frozen TA muscle sections (10 mm thick) were air dried for 10 min before being placed in hematoxylin solution (Sigma, UK #GHS232) for 8 min, slides were washed in tap water to remove excess stain and placed in 70% ethanol, 1% HCl for 10 s followed by 5 s in 1% Eosin Y solution (Sigma, UK #E4382) in 80% ethanol. Following a series of alcohol washes unbound stain was removed in Histochoice clearing agent (Sigma UK #H2779) and mounted using Histomount (National Diagnostics, UK #HS-103). Slides were examined under an Axioplan 2 Microscope System (Carl Zeiss, Germany) to obtain images.

Isolated muscle function analysis
Peak force, specific force and force drop from the EDL muscle were measured as previously described [8]. In brief, isolated EDL were attached to a lever arm connected to a force transducer and stimulator; the equipment was controlled using the signal interface and the DMC software (Aurora Scientific, Bristol, United Kingdom). After determination of the optimum length (L o ), the optimum fibre length (L f ) was calculated by multiplying L o by the fibre length to muscle length ratio of 0.44. A forceefrequency curve was generated and the maximum isometric force calculated. Absolute force (P o ) are normalized to specific force (sP o ; mN/cm 2 ) using the equation (muscle mass/L f Â1.06). Percentage force drop was calculated by comparing maximum force between the first (EC0) and fifth (EC5), expressed as a percentage of EC0. All data were digitized and analysed using the DMC software.

Protein analyses
Muscles samples were homogenized (Precellys 24 Bertin Technologies) for 2 Â 30 s 5500 rpm in chilled RIPA buffer (R0278-50 mL, Sigma-Aldrich) supplemented with protease inhibitors (1:100; P8340, Sigma-Aldrich). Following BCA quantification (23227; ThermoFisher Scientific), 30 mg of total protein were heatdenatured for 5 min at 100 C before loading onto NuPAGE 3e8% TRIS Acetate Midi Gel (Novex, Life Technologies) and transferred to PVDF membranes (Millipore). Membranes were blocked for 1 h with Odyssey Blocking buffer (926e41090; LI-COR; USA) and then incubated with primary antibodies in Odyssey Blocking buffer PBS þ 0.1% Tween for 2 h at room temperature. Primary antibodies used were: mouse monoclonal anti-utrophin (1:50, MAN-CHO3(84A), gift from G.E. Morris), mouse monoclonal anti-MYH3 (1:100, sc-53091, Santa Cruz Biotechnology, USA), anti-dystrophin  Table 1 physicochemical and ADME characterisation were conducted by Evotec and Xenogesis applying the standard protocols. Solubility refers to kinetic solubility measured by a turbidimetric assay, and metabolic stability was assessed in mouse liver microsomes after 2 h at 37 C. Results described in Table 1 were evaluated by Evotec and Xenogesis (metabolic stability). Solubility refers to kinetic solubility measured by a turbidimetric assay. Metabolic stability was assessed for up to 1 h at 37 C in mouse and human cryopreserved hepatocytes.