The Dimeric Form of 1,3‐Diaminoisoquinoline Derivative Rescued the Mis‐splicing of Atp2a1 and Clcn1 Genes in Myotonic Dystrophy Type 1 Mouse Model

Abstract Expanded CUG repeat RNA in the dystrophia myotonia protein kinase (DMPK) gene causes myotonic dystrophy type 1 (DM1) and sequesters RNA processing proteins, such as the splicing factor muscleblind‐like 1 protein (MBNL1). Sequestration of splicing factors results in the mis‐splicing of some pre‐mRNAs. Small molecules that rescue the mis‐splicing in the DM1 cells have drawn attention as potential drugs to treat DM1. Herein we report a new molecule JM642 consisted of two 1,3‐diaminoisoquinoline chromophores having an auxiliary aromatic unit at the C5 position. JM642 alternates the splicing pattern of the pre‐mRNA of the Ldb3 gene in the DM1 cell model and Clcn1 and Atp2a1 genes in the DM1 mouse model. In vitro binding analysis by surface plasmon resonance (SPR) assay to the r(CUG) repeat and disruption of ribonuclear foci in the DM1 cell model suggested the binding of JM642 to the expanded r(CUG) repeat in vivo, eventually rescue the mis‐splicing.

Abstract: Expanded CUG repeatR NA in the dystrophia myotonia protein kinase( DMPK)g ene causes myotonic dystrophy type 1( DM1) and sequesters RNA processing proteins, such as the splicingf actor muscleblind-like 1 protein (MBNL1). Sequestration of splicing factors results in the mis-splicing of some pre-mRNAs. Small molecules that rescue the mis-splicing in the DM1 cells have drawn attention as potentiald rugs to treat DM1.H erein we report an ew molecule JM642 consisted of two 1,3-diaminoisoquinoline chromophores having an auxiliary aromatic unit at the C5 position. JM642 alternates the splicing pattern of the pre-mRNA of the Ldb3 gene in the DM1 cell model and Clcn1 and Atp2a1 genes in the DM1 mouse model. In vitro bindinga nalysis by surface plasmon resonance (SPR) assay to the r(CUG) repeat and disruptiono f ribonuclear foci in the DM1 cell model suggested the bindingo fJM642 to the expanded r(CUG) repeat in vivo, eventually rescuethe mis-splicing.
Molecules modulating the splicing pattern of genes in DM1 cells have drawn attention as potentiald rugs treating this devastatingn eurologicald isorder. [1][2][3][4] DM1 is an autosomald ominant neuromuscular disorder,c haracterizedb ym yotonia (delayed relaxation of muscles after contraction), progressive weakness, cardiacc onduction defects, and cognitive impairments.T he aberrant expansion of the CTG repeat in the 3' un-translated regions of the DMPK gene is the cause of the disease. [5][6][7] The transcript of the DMPK gene with the long CUG repeats equesters the RNA-binding proteins,s uch as the splicing factor MBNL1 in the nucleus. [8,9] As ac onsequence, several genesi nD M1 cells showed different splicing patterns from those observed in the wild type cells. [10,11] In the splicing of pre-mRNA of the LIM domain binding3( Ldb3)gene, exon 11 is excluded from mRNA by about 80 %i nt he wild type, but exon 11-included mRNA is produced about5 0% in DM1 cells. [3,12] In the splicingo fp re-mRNAs encoding muscle-specific chloride channel (Clcn1), the mRNA withoute xon 7a is dominant in wild type, whereas exon 7a-included mRNAi sa bundant in DM1 cells. [13] Similarly,s plicing of pre-mRNAo ft he Atp2a1 gene codings arcoplasmicr eticulumc alcium-ATPase 1 (SERCA1) produces mRNA containing exon 22 in the wild type cells, whereas mRNA withoute xon 22 predominates in DM1 cells. [14] Misregulated alternative splicing is af undamentalm olecularf eature of DM1, having good potential to function as biomarkers of severitya nd therapeutic response. [11] These differences in the splicing patterns betweent he wild type and DM1 encouraged studies focused on the modulation of the splicingp atterns. Besides prominent approaches using oligonucleotides, [15] severalg roups have reported small molecules binding to the CUG repeatsa nd modulating the splicing pattern in DM1c ells. [16][17][18][19][20][21][22][23][24][25][26][27][28][29] We herer eport that the dimeric form of 1,3-diaminoisoquinoline derivative JM642 (Figure 1) rescued the mis-splicing in Ldb3 pre-mRNA in the DM1 cell model and Clcn1 and Atp2a1 pre-mRNAsi nD M1 mousem odel in ad osedependentm anner.S PR assay showed the binding of JM642 to the r(CUG) 9 -immobilized sensor surface, and JM642 led to the disruptiono fr ibonuclear foci in DM1 cell model expressing r(CUG) 800 repeat,d emonstrating that JM642 would be au seful molecular tool for the deeperu nderstanding of the pathogenesis of DM1 and studies on the therapeutic potential of small molecules targeting DM1.
We have reported different types of molecules that bind to r(CUG) repeat and modulate the alternative splicing in DM1 cells. [30][31][32][33] After structure-activity studies on small molecules targetingt he r(CUG) repeat, we revisited 1,3-diaminoisoquinoline derivativesw ith an additional aromatic unit at the C5 position and found am onomeric 1,3-diaminoisoquinoline ligand JM608 and its dimericf orm JM642. ( Figure 1) While the detail of structure-activity studies will be reported elsewhere, in brief, the substituent at the C5 position of the 1,3-diaminoisoquinoline chromophore showed as ignificant effect on the binding to the CUG repeatR NA. JM608 was synthesized by Suzuki-Miyaura cross-coupling [34] of the 1-amino-5-bromo-3chloroisoquinoline derivative 3 with ap iperazine-substituted pyridinyl pinacol boronic ester 8 followedb yB uchwald-Hartwig cross-coupling [35] of the resulting 4 at the C3 positionw ith the Boc-protected carbamate 9.( Scheme 1) Deprotection of all Boc groups in 5 furnished the synthesis of JM608. JM642 was obtainedb yc oupling of 4 with aC bz-protected carbamate 10, deprotection of the Cbz group in 6,d imerizationo f7 with a pentafluorophenyl-activated biscarboxylic acid 11, [36] and deprotection of the Boc groups.
The effect of JM608 and JM642 on alternative splicing was investigated on pre-mRNA of the Ldb3 gene in the C2C12 DM1 cell model conditionallye xpressing r(CUG) 800 repeatR NA. [37] ( Figure 2a)I nt he controlc ells without expression of r(CUG) 800 , the percentage of exon 11 exclusion in the Ldb3 gene was about 81 AE1.7 %, while the fraction in the DM1 cell model expressing r(CUG) 800 was 53 AE 1.9 %( Figure2b). After the treatment of the DM1 cell model with JM642 for two days, the mis-splicing of Ldb3 pre-mRNA wass ignificantly rescuedi na dose-dependentm anner,i ncreasing exon 11 exclusion up to 77 AE 2.5 %w ith 80 mM. The observed rescue effect of JM642 on the mis-splicing is statistically significant (**P < 0.01) at the concentrations higher than 30 mM. The effect of am onomer JM608 on the recovery in mis-splicing was 58 AE 1.9 %a t8 0mM. For the reference, cytotoxicity of JM608 and JM642 to the C2C12 DM1 cell model was not apparent over the treatment range (data not shown).
We then investigated the effect of JM642 on the mis-splicing of pre-mRNAs in the DM1 mouse model( HSA LR ), which expresses r(CUG) 220 ande xhibits the mis-splicing of Clcn1 and Atp2a1 pre-mRNAs. [38] JM642 (10 mg kg À1 or 20 mg kg À1 per day) was administrated to the HSA LR mice (n = 3i ne ach group) by daily intraperitoneal (i.p.) injection for five days. The fraction of exon 7a exclusion for the Clcn1 gene was 85 AE 0.53 %f or the wild type mice and 44 AE 2.4 %f or the HSA LR mice. (Figure 3a). Mis-splicing of Clcn1 has been suggested to causem yotonia. [13] Treatmento ft he HSA LR mice with i.p. JM642 (10 and 20 mg kg À1 )r escued the mis-splicing in the Clcn1 gene, leading   to an exclusion rate of 61 AE 2.3 %( P = 0.03) and7 0AE 2.3 %( P = 0.01), respectively,a lthough an improvement of phenotypic myotonia was not apparent due to the partial rescue of splicing. The rescue effect of JM642 was also observedi nAtp2a1 mis-splicing. In thew ild-type mice, the inclusion rate for exon 22 is 100 AE 0%,w hereast he inclusion fraction of exon 22 in the HSA LR mice was 16 AE 2.4 % ( Figure 3b). After administration of 10 and 20 mg kg À1 of JM642,t he inclusion rate improved to 32 AE 2.5 %( P < 0.05) and 74 AE 6.0% (P < 0.01), respectively.T oxicity was not observed within the mouse modelo ver this treatment range. These results demonstrated the rescue effect of JM642 on mis-splicing of Clcn1 and Atp2a1 pre-mRNAsi nD M1 in vivo.
Having observed the significant effect of JM642 and somewhat moderate effect of JM608 on the rescue in mis-splicing of genes in DM1 cell and mouse models, we have investigated the origin of these biological effects. The currenth ypothesis on the expected therapeutic effects of small molecules in the treatment of DM1 stems from the competitive bindingo fs mall molecules with RNA-binding proteins to the aberrantly expanded CUG repeat RNA in the nucleus. [1][2][3] To know if JM608 and JM642 could fit this hypothesis, we lookeda tt he binding of these molecules to the CUG repeat RNA with the SPR assay. The biotin-labeledr (CUG) 9 repeat RNA and r(CCG) 9 repeat RNA as control were immobilizedt hrough the tri-ethylene glycol linker to the avidin-coated sensors urface, and the analyte molecule was sequentially added with the increasedc oncentration to the surface (single cycle kinetic analysis).
The SPR response curveso btained for JM608 and JM642 from the same sensors urface of r(CUG) repeat RNA were quite differenti nt erms of the shape of the curve, which characterizes the association and dissociation kinetics as well as the affinity.T he SPR profiles obtained for JM608 showed the rectangular shape indicating ar apid associationa nd dissociation kinetics. (Figure 4a)T he lowest concentration necessary for the significant SPR response under the conditions was 63 nm.T he apparentd issociation constant (K d(app) )o fJM608 to the r(CUG) 9 repeat RNA was determined1 .2 mm based on the assumed 1:1 binding isotherm.
In contrast,S PR responsec urveso btained for JM642 were quite characteristic, showingabroad parabola shape without any plateau region. (Figure 4b)T he lowest concentration of JM642 for producing as ignificant SPR response was 6.3 nm, which is one order of magnitude smaller than that of JM608, suggesting ap ositive effect of dimerization of JM608 on the binding to the CUG repeat. The parabola shape observed for the response curvesi su nique and is likely due to the dimeric form. The SPR response increased as the duration in applying JM642 prolonged. However,t he SPR signal startedt od ecrease while JM642 was kept applying to the surface. In general,S PR responses reach the plateauo rs teadily increase due to the equilibrium shift toward the ligand-bound state from the free unbound state in the bulk solution. The characteristic phenomena in SPR analysiso fJM642 are likely due to conformational changes on the JM642-CUG RNA complexo nt he surface after initial complex formation. The significant effects of the linker length and structure connecting two isoquinoline chromophores on the binding to r(CUG) 9 observed in the SPR analysis may support the above speculation. (FigureS1i nt he Supporting Information) SPR responses on the r(CCG) repeat RNA surface were weak for both JM608 and JM642,e ven at 1.0 and 0.1 mm,respectively.
To gain insight into the possibility of competitive binding of JM642 with RNA-binding proteins on CUG repeatR NA,w e have investigated the disruption of the ribonuclear foci in DM1 patient-derived myoblast cells by JM642 treatment. Untreated DM1 myoblasts showed the formation of ribonuclear foci (Figure 5a). The percentage of cells showing foci positive nucleus was 41 AE 7.0 %a mong 255 cells examined. Upon treatment with 30 mm JM642,t he number of cells showing the foci positive nucleus dropped to 6.7 AE 1.3 %a mong2 86 cells counted. Since the FISH probes capture the CUG repeatR NA, the CUG repeat RNA was suggested to dissociate from the aggregates forming foci in the cell nucleusupon JM642 treatment.
In summary,anewly developed small molecule JM642,adimeric form of 5-substituted-1,3-diaminoisoquinoline derivative  JM608,r escued the mis-splicing in both DM1 cell and mouse models. The increaseda ffinity and the different modes of the binding of JM642,a sc ompared to JM608 in SPR assayi nv itro likely attributed to the differenceo fb iological activity in the DM1 cell model. Disruption of the ribonuclear foci in the DM1 cell model also supported the possibility of JM642-binding competitively with RNA-bindingp roteins.O verall, JM642 could be au seful molecular tool for the studies on the biological responses induced by expanded CUG repeat.

Experimental Section
Studies on rescueeffect of small molecules on the DM1 cell model Ac onditional cell model for the analysis of MBNL1 splicing regulatory activity has been established, as reported previously. [37] Briefly, C2C12 mouse muscle cells were co-transfected with pLC16 containing 800 CTG repeats and plasmid PhiC318 encoding PhiC31 integrase (Addgene, Cambridge, MA). Transfection was performed using Nucleofector technology (Lonza, Basel, Switzerland) according to the manufacturer's program B-32. Stably transfected clones were selected with puromycin (1.25 mgmL À1 ). Transcription across the expanded repeat was activated by Cre recombinase-mediated excision of at ranscription terminator cassette. C2C12 cells with recombination were selected using hygromycin B( 300 mLmL À1 ). RNA was harvested after 2days of incubation with JM608 and JM642. RNA extraction and analysis of the splicing pattern were carried out as described below.W ST-1 assay was performed according to the manufacturer's instructions (Roche, Basel, Switzerland).

Studies on rescueeffect of small molecules on the DM1 mouse model
Mouse handling and experimental procedures were performed following the Osaka University guidelines for the welfare of animals and were approved by the institutional review board. Homozygous HSA LR transgenic mice of line 20b (FVB inbred background) were described previously. [38] Gender-and age-matched (< 3months old) mice were treated with JM642 at indicated dose and period by daily i.p. injection. After treatments, mice were sacrificed, and the rectus femoris (quadriceps) muscle was obtained for splicing analysis. RNA extraction and analysis of the splicing pattern were carried out as described below.

RNA extraction and splicing analysis
To tal RNA extraction from model cells, cDNA synthesis, and polymerase chain reaction (PCR) amplification were performed as described previously. [39] The PCR products were separated by agarose gel electrophoresis, and the gel was stained with GelRed (Biotium, Hayward, CA). The gel was imaged using aT yphoon laser fluorimager (GE Healthcare, Pittsburgh, PA)a nd the products quantified using ImageQuant (GE Healthcare).