A Novel Minor Groove Binder as a Potential Therapeutic Agent for Myotonic Dystrophy Type 1

Abstract

Myotonic dystrophy type 1 (DM1) is a multisystemic neuromuscular disorder that is inherited in an autosomal dominant manner. DM1 originates in a (CTG·CAG) repeat expansion in the 3’-UTR of the dystrophia myotonic protein kinase (DMPK) gene on chromosome 19. One of the transcripts, r(CUG)^exp is toxic in various ways. Herein, we report a rationally designed small molecule with a thiazole peptidomimetic unit that can serve as a minor groove binder for the nucleic acid targets. This peptide unit linked to two triaminotriazine recognition units selectively binds to d(CTG)^exp to inhibit the transcription process, and also targets r(CUG)^exp selectively to improve representative DM1 pathological molecular features, including foci formation and pre-mRNA splicing defects in DM1 model cells. As such, it represents a new structure type that might serve as a lead compound for future structure-activity optimization.

Introduction

Myotonic dystrophy type 1 (DM1) is one of the incurable neuromuscular disorders. It originates from a dominantly inherited (CTG·CAG) repeat expansion in the 3’-UTR of dystrophia myotonic protein kinase (DMPK) gene on chromosome 19.^[1] The molecular basis of DM1 is that DMPK is transcribed into an mRNA transcript with CUG repeat expansions that form hairpin secondary structures. These hairpin structures sequester proteins such as the muscleblind-like (MBNL) family of proteins. MBNL proteins are responsible for post-transcriptional splicing and polyadenylation site regulation.^[2,3] The sequestration of MBNL proteins prevents them from performing their proper functions, leading to hundreds of pre-mRNA splicing defects (Figure 1).^[4] In addition, the CUG RNA-binding protein Elav-Like family member 1 (CELF1) is elevated in DM1 patients and this protein also impacts alternative splicing events. Thus, mis-splicing patterns are observed in multiple transcripts, including muscle chloride channel (CLCN1), insulin receptor (IR), cardiac troponin T (cTNT), cardiac sodium channel (SCN5A), and dystrophin (DMD), that each lead to specific clinical symptoms including myotonia, muscle weakness, cardiac arrhythmias, and cognitive dysfunction.^[5–9] The clinical findings have shown that the severity of the symptoms generally correlates with CTG repeat size.^[10–12] Although the pathogenesis of DM1 has been studied thoroughly and is quite well understood, there is currently no cure or effective therapeutic intervention for DM1.

Figure 1.

Schematic illustration of DM1 pathogenesis. The (CTG·CAG) repeat expansion undergoes transcription process to form r(CUG)^exp transcript that can form hairpin structures and sequester MBNL proteins. The depletion of MBNL proteins inside cells leads to hundreds of pre-mRNA splicing defects and various disease phenotypes. Ligand 5 serves as a minor groove binder for CTG repeat and CUG repeat, inhibits transcription process, and releases sequestered MBNL1 proteins.

A majority of therapeutic efforts have been focused on development of small molecules to target r(CUG)^exp and inhibit the sequestration of MBNL proteins. In 2009, our group reported the first structure-based design of a small molecule (1) that selectively complexes CUG repeats in RNA (K_d = 390 ± 80 nM) and CTG repeats in DNA (K_d = 430 ± 110 nM).^[13] This ligand contains a triaminotriazine unit linked to an acridine intercalator through a methylene linker chain that together act as a “stacked intercalator” (Figure 2a). The triaminotriazine unit was designed to recognize the U-U or T-T mismatches through Janus-wedge type hydrogen bonding or base flipping and Watson-Crick-Franklin base-pair.^[14] We recently reported a structure of this ligand in a complex with DNA containing three consecutive CTG repeats with three T-T mismatches, revealing that ligand 1, as designed, intercalated the CTG repeats with three hydrogen bonds formed with one thymine base, an additional hydrogen bond to the phosphate backbone. The structure shows the opposite thymine base to be flipped out of the helix.^[15] Unfortunately, ligand 1 is not sufficiently cell-permeable and has high cytotoxicity, although analogous cell-permeable ligands were developed and shown to revert some DM1 pathologies in cell culture.^[16,17] Berglund and co-workers reported pentamidine 2 and its analogues to function as transcription inhibitors for d(CTG)^exp and to rescue some DM1 related defects in cell and mouse model studies of DM1.^[18] In 2004, our group reported a new RNA-groove binding inhibitor 3. This ligand contains two triaminotriazine units connected by a bisamidinium linker, and binds in the major groove of r(CUG)^exp with low micromolecular affinity (K_d = 8 ± 2 μM) but not seem to bind d(CTG)^exp. It has been shown to reduce MBNL1-r(CUG)^exp ribonuclear foci and restore mis-regulated splicing of IR and cTNT in a DM1 cell culture.^[19] This ligand 3 has been further developed into a multivalent ligand 4 as a mixture of oligomers ranging from 4mer to 8mer to target d(CTG)^exp and r(CUG)^exp in DM1. Ligand 4 showed improved activities in DM1 cells and DM1 liver mouse model to reduce prominent DM1 features.^[20] Herein, we report a novel minor groove binder 5 as a charge neutral analogue of 3. We show that ligand 5 can selectively target pathogenic nucleic acids. In particular, it selectively binds to d(CTG)^exp to inhibit transcription process, binds to r(CUG)^exp to release sequestered MBNL1 proteins, and rescues downstream pre-mRNA splicing defects. As such, it represents a new structure type that might serve as a lead for future structure-activity optimization.

Figure 2.

(a) Structure of ligand 1 and the interaction between triaminotriazine ring with one thymine base and another thymine base is flipped out of the helix. (b) Structure of pentamidine. (c) Structure of ligand 3. (d) Structure of ligand 4, mixture of oligomers, ranging from 4mer to 8mer.

Results and Discussion

Design of Minor Groove Binding Ligand 5.

Ligand 5 was designed with two triaminotriazine units as recognition moieties for T-T and U-U mismatches, and a thiazole-based peptidomimetic linker to serve as a generic minor groove binder (Figure 1). Thiazole-containing compounds have been widely utilized in medicinal chemistry, such as antibiotics, bacteriostatics, CNS regulants, diuretics, fungicides, and herbicides.^[21] For example, Austin and co-workers reported a series of thiazole moiety-containing compounds as splicing regulators for survival motor neuron gene 2 (SMN2), a potential therapeutic strategy for spinal muscular atrophy (SMA).^[22] Dash and co-workers identified several thiazole moiety-containing ligands as binders for HIV-1 transactivation response element (TAR) through an in situ cycloaddition screening process.^[23]

Computational Docking of Ligand 5 to r[(CUG)₆]₂.

To validate our molecular design, we performed a molecular modeling study using Molecular Dynamics (MD) simulations. The crystal structure of r[(CUG)₆]₂ was obtained from the Protein Data Bank (code 3GM7).^[24] Ligand 5 was docked with triaminotriazine units located at alternating U-U mismatches based on the approach and results of modeling on our previously reported ligand 3.^[19] Ligand 5 was docked in the minor groove of the RNA duplex. Each triaminotriazine ring formed three hydrogen bonds with one U base, with the opposite U base flipped out of the double helix. A single hydrogen bond between the triaminotriazine ring and the phosphate backbone of the flipped U base was observed, consistent with the triaminotriazine interactions observed in the crystal structure described above. The thiazole peptidomimetic unit sits in the minor groove of RNA duplex with hydrogen bonds formed between amide groups and thiazole rings with the RNA backbone (Figure 3). The length of methylene linker was also optimized through this modeling study.

Figure 3.

Molecular modeling of ligand 5 with r[(CUG)₆]₂. Energy minimized structures show ligand 5 docked in the minor groove of duplex with recognition of U-U mismatches (in green) of triaminotriazine rings through hydrogen bonding (in red). N atoms are in blue. C atoms are in black. S atoms are in yellow. C bases are in light grey. G bases are in dark grey.

Synthesis of Ligand 5.

The synthesis of ligand 5 was achieved through stepwise amide coupling reactions (Scheme 1). Thus, thiazole peptidomimetic unit 6 was coupled to compound 7 using PyBOP/DIPEA to give the corresponding amide whose ethyl ester was hydrolyzed with LiOH-H₂O to give compound 8. A second peptide coupling with triaminotriazine unit 9 afforded the final ligand 5 (See Supporting Information for details).

Scheme 1.

a) PyBOP, DIPEA, DCM, r.t., 24 h. b) LiOH-H₂O, THF-MeOH-H₂O (32% for two steps). c) PyBOP, DIPEA, DMF/DMSO, 50 °C; 7%.

Binding Affinity to CTG and CUG Repeats.

To assess the binding affinity and selectivity of ligand 5 toward the nucleic acid targets, isothermal titration calorimetry (ITC) experiments were performed (Figure S1). It was found that ligand 5 bound to d(CTG)₁₆ with a micromolar affinity (K_d = 141 ± 25 μM), and to r(CUG)₁₆ with a slightly higher binding affinity (K_d = 55 ± 1 μM). On the contrary, the binding of ligand 5 toward random duplex was very weak, with minimal heat generated and no measurable K_d value. These data suggest that ligand 5 can selectively target d(CTG)^exp and r(CUG)^exp.

In Vitro Transcription Inhibition Assay.

We further studied the ability and selectivity of ligand 5 to serve as a transcription inhibitor. An in vitro transcription inhibition assay was conducted with a (CTG·CAG)₉₀-containing plasmid and a non-repeat-containing control plasmid.^[25] Ligand 5 strongly inhibited the production of (CUG)₉₀ transcript in a dose-dependent manner with an IC₅₀ value of approximately 300 μM. Nearly 100% inhibition was achieved with 800 μM of ligand 5. The inhibition of the control template was much less in comparison (Figure 4a,b).

Figure 4.

(a) Representative gel of the in vitro transcription inhibition assay. (b) Quantification of r(CUG)₉₀ transcript. A dose-dependent inhibition of the transcription process was observed with treatment of ligand 5. At least three independent experiments were conducted and error bars represent standard error of the mean. (c) Reduction of cellular r(CUG)^exp levels in the DM1 model cells with treatment of ligand 5. At least three independent experiments were performed and error bars indicate standard error of the mean. * P < 0.05, ** P < 0.01, *** P < 0.001.

Bioactivity of ligand 5 in DM1 model cells.

Encouraged by the promising in vitro results, the bioactivity of ligand 5 in DM1 model cells was further evaluated. The cytotoxicity and cell permeability of ligand 5 were tested, as these criteria are important to indicate the concentration range and possibility for future development. The cytotoxicity of ligand 5 was determined using a sulforhodamine B toxicity assay with HeLa cells. No obvious cytotoxicity was observed up to 1 mM treatment of ligand 5 for 3 d (Figure S2a). Ligand 3 was included for a head-to-head comparison. More than 20% cell death was observed with 800 μM of ligand 3, indicating that ligand 5 possesses a better cytotoxicity profile (Figure S2b). The cell permeability of ligand 5 was analyzed using HPLC. The cells were treated with ligand 5 at 10 μM for 1 d, harvested, and lysed using NP-40 lysis buffer solution. The absolute concentration of ligand 5 in cell lysate was determined to be 1.4 ± 0.5 μM through HPLC analysis using a standard curve of ligand 5 (Figure S3). The results indicated that ligand 5 is cell-permeable and has low cytotoxicity.

In addition to sequestration of MBNL proteins, r(CUG)^exp can cause toxicity through other mechanisms. For example, it interferes with transcriptional network of MEF2 which causes microRNA dysregulation in DM1 heart tissues.^[26] The r(CUG)^exp can also undergo repeat associated non-ATG (RAN) translation to produce toxic homopeptides.^[27] For these reasons, we next studied the ability of ligand 5 to reduce cellular toxic r(CUG)^exp production in DM1 model cells. The DM1 model cells were constructed by transfecting HeLa cells with GFP-DT960 or GFP-DT0 plasmids containing 960 or 0 CTG repeats, respectively, in the exon 15 of a truncated DMPK gene.^[28] The DM1 model cells were treated with ligand 5 for 3 d and harvested. The cellular r(CUG)^exp mRNA level was determined by measuring the level the of exon 15 upstream of r(CUG)^exp, relative to PABP mRNA, which was used as an internal control. The data were normalized to the level measured in untreated DM1 model cells. The results indicated that ligand 5 reduced the cellular r(CUG)^exp level in a dose-dependent manner, and the IC₅₀ was determined to be around 400 μM (Figure 4c).

As one of the hallmarks in DM1, ribonuclear foci are formed by r(CUG)^exp sequestration of MBNL1 proteins inside the nucleus. The ability of ligand 5 to reduce the foci formation was evaluated in DM1 model cells. The ribonuclear foci can be readily visualized with confocal microscopy, using immunofluorescent staining for MBNL1 proteins and a fluorescent in situ hybridization (FISH) probe for r(CUG)^exp. To quantify the foci reduction in DM1 model cells, the foci area was measured in treated cells and compared to the untreated model cells. Upon treatment of ligand 5 for 2 d at various concentrations, the foci area decreased significantly in the DM1 model cells. At a concentration of 800 μM, the ribonuclear foci area was reduced to less than 20% of that in untreated cells, which indicated that ligand 5 can successfully disturb the foci formation in the DM1 model cells (Figure 5).

Figure 5.

(a) Representative confocal images of ligand 5 treated DM1 model cell culture. MBNL1 proteins were visualized with mouse anti-MBNL1 antibody and goat anti-mouse secondary antibody Alexa 647. r(CUG)^exp was probed using Cy3-(CAG)₁₀ FISH probe. Nuclei were stained with 10 μg/mL Hoechst 33342. Scale bar = 10 μm. (b) Quantification of foci area per cell in treated DM1 model cell culture, relative to untreated cells. At least three independent experiments were conducted. Error bars indicate standard error of the mean. * P < 0.05, ** P < 0.01.

To further examine the therapeutic efficacy of ligand 5 in the important downstream alternative splicing regulation, we tested the ability of ligand 5 to rescue insulin receptor (IR) pre-mRNA splicing defect in DM1 model cells. Alternative splicing of the IR pre-mRNA is aberrantly regulated in DM1, resulting in predominant expression of the low-signaling isoform A, lacking exon 11, which leads to decreased insulin sensitivity.^[29] The IR splicing defect was of particular interest because it has been reported to be more difficult to rescue compared to other DM1 splicing defects.^[30] Following the method described above, HeLa cells was transfected with DT960 or DT0 plasmids containing 960 or 0 interrupted CTG repeats, respectively, in the exon 15 of a truncated DMPK gene. In DM1 model cells, the relative amount of isoform B (including exon 11) was around 30%, compared around 53% in DT0 normal cells. Treatment of ligand 5 in DM1 model cells successfully rescued the IR pre-mRNA splicing defect in a dose-dependent manner (Figure 6). Full rescue was achieved with around 400 μM treatment of ligand 5. Ligand 3 at 100 μM was introduced as a comparison, and around 40% of isoform B was achieved, slightly better than then efficacy of ligand 5 at 100 μM.

Figure 6.

(a) Schematic description of IR pre-mRNA splicing pattern. There are two patterns depending on exclusion or inclusion of exon 11, generating isoform A and isoform B, respectively. In DM1 model cells, isoform A is predominant. (b) Treatment of ligand 5 rescued the mis-splicing defect and full rescue was achieved at around 400 μM. Ligand 3 was included at 100 μM for comparison. At least three independent experiments were conducted. Error bars indicate the standard error of the mean. * P < 0.05, ** P < 0.01, *** P < 0.001.

Conclusion

In this study, we reported a novel small molecule 5 that can selectively recognize the unique secondary structure formed by T-T and U-U mismatches in DM1. Ligand 5 contains triaminotriazine units as recognition moieties and a thiazole peptidomimetic unit as a minor groove binding linker. Ligand 5 exhibited bioactivity both in vitro and in DM1 model cells. At the transcriptional level, 5 functioned as a transcriptional inhibitor and reduced cellular toxic r(CUG)^exp significantly in DM1 model cells. In addition, 5 effectively reduced the ribonuclear foci formation and rescued the downstream IR pre-mRNA splicing defect in a dose-dependent manner. Though ligand 3 showed slightly better efficacy in rescuing the IR splicing defect, ligand 5 exhibits a better cytotoxicity profile and thus if of interest as a scaffold for further optimization.

Herein, we have demonstrated a new structure type of ligand with a triazole peptidomimetic unit as a minor groove binder to selectively target both d(CTG)^exp and r(CUG)^exp. The new thiazole-based scaffold is different from previously reported acridine- and bisamidinium-based compounds. Thus, the thiazole-based ligand is neutral and highly cell permeable with a promising cytotoxicity profile. It can serve as a lead for future structure-activity optimization and a potential therapeutic agent.

Experimental Section

Materials and Methods.

Detailed information of the chemicals and instrumentation methods used in this work can be found in the Supporting Information.

Synthetic Procedures.

The detailed procedures and characterizations are included in the Supporting Information.

In Vitro Transcription Inhibition Assay.

The DNA template for transcription was prepared by linearization of (CTG)₉₀-containing pSP72 plasmid by cleavage at BamH1 site.^[25] The control DNA template was prepared by PCR of a cloned plasmid, which has 38 bp of random sequence (5’-GCG TAG CTT ACA TAG TTC ACA TAG TCT GAT ACT TCC GA-3’) inserted into the pGEM-Teasy vector (Promega). The ligand was pre-incubated with transcription reaction solution including 15 ng of DNA template, 0.5 mM each rATP, rUTP, rCTP, rGTP, 1X RNAPol reaction buffer (NEB, 1 mM DTT, 40 mM Tris-HCl, 6 mM MgCl₂, 2 mM spermidine, pH 7.9) at room temperature for 3 h. After pre-incubation, 50 U of T7 RNA polymerase (NEB, M0251S) was added to each reaction tube and the solution was incubated at 37 °C for 1.5 h. Reactions were quenched by adding 8 μL of quenching solution (7 μL of 8 M urea (aq), 1 μL of denaturing dye (95% formamide, 5 mM EDTA, 0.025% each xylene cyanol and bromophenol blue)). The solutions were heated at 95 °C for 5 min before gel loading. An aliquot of 15 μL was run on 8% denaturing PAGE in 1X TBE. The gel was post-stained with ethidium bromide and imaged with Gel Doc XR+ system (Bio-Rad). Band intensity was quantified with ImageJ software. The band intensity of (CUG)₉₀ was normalized to that from the untreated transcription reaction.

Cellular r(CUG)^exp Level Measurement.

Approximately 60,000 HeLa cells were plated in each well of a 12-well plate in DMEM medium supplemented with 10% FBS 1 d before transfection. HeLa cells were transfected with 1 μg of GFP-DT0 or GFP-DT960 plasmids using Lipofectamine 2000 (Life Technologies) following the recommended protocol. After 4 h, the transfection cocktail was replaced with the DMEM medium supplemented with 10% FBS, and the cells were treated with 1 μg of Dox. Cells were treated with ligand at desired concentrations for 3 d. Fluorescence microscopy was utilized to check cells for a GFP signal as a marker for successful transfection. Cells were harvested and total mRNA was isolated using E.Z.N.A. total RNA kit I (Omega) according to the manufacturer protocol. From the RNA isolated, 1.5 μg of total mRNA was subjected to DNase treatment to remove all DNA contaminants. cDNA was synthesized using iScript cDNA synthesis kit (Bio-Rad) and used as template for real-time PCR using SYBR master mix (Applied Biosystem). The results from real-time PCR experiments were analyzed using the ΔΔC_t method.^[31] The mRNA levels of exon 15 upstream of (CUG)^exp were measured relative to PABP mRNA levels. The difference in the expression level of exon 15 RNA transcript between treated GFP-DT0 and GFP-DT960 samples was compared with the one of untreated sample that were normalized to 100%. The primers used in the experiments were: E15upF: 5’-TCG GAG CGG TTG TGA ACT-3’; E15upR: 5’-GTT CGC CGT TGT TCT GTC-3’; PabpF: 5’-CTG CTG TTC ATG TGC AAG GT-3’; PabpR: 5’-CAA CAG CAT GCC AGT GAT T-3’.

Foci Dispersion Study with Confocal Microscopy.

Approximately 120,000 HeLa cells were plated onto coverslips in each well of a 6-well plate in DMEM medium supplemented with 10% FBS. Cells were transfected at 70-80% confluence with 1 μg of DT960 plasmid with Lipofectamine 2000 (Life Technologies) according to the recommended protocol. After 4 h, the transfection cocktail was changed to DMEM medium supplemented with 10% FBS and the compound was added to reach the final concentration. Cells were incubated at 37 °C for 2 d, and fixed for 15 min at room temperature with 4% PFA with subsequent washing with 1X PBS 5 times at RT. For the fluorescence in situ hybridization (FISH) procedure, cells were permeabilized with 0.5% triton X-100 in 1X PBS at room temperature for 5 min and washed twice with 1X PBS. As a pre-equilibration, the cells were incubated with 50% formamide in 2X SSC for 10 min at RT. The cells were then probed for 2 h at 37 °C in dark with 1 ng/μL Cy3-(CAG)₁₀ probe (Integrated DNA Technologies), 2 μg/mL BSA, 2 mM Vanadyl complex, and 66 μg/mL yeast tRNA in 50% formamide in 2X SSC. The cells were washed with 50% formamide in 2X SSC for 30 min at 37 °C, with 1X SSC for 30 min at 37 °C, and then with 1X PBS twice at room temperature.

For immunofluorescent (IF) staining, the cells were probed with anti-MBNL antibody (mouse Ab, 1:1000 dilution in 1X PBS) overnight at 4 °C. The cells were washed twice with 1X PBS at RT. The cells were incubated with rabbit anti-mouse lg (DyLight647 secondary Ab, Pierce PROD# SA5-1015, 1:1000 dilution in 1X PBS) for 2 h at RT. The cells were washed twice with 1X PBS at RT. To the cells was added 10 μg/mL Hoescht33342 (1:1000 dilution in 1X PBS) and cells were incubated for 5 min at RT. The cells were washed 4 times with 1X PBS at RT. ProLong anti-fading oil was dropped onto the glass slides, and cover slips mounted onto glass slides. The boundary of cover slip on glass slide was sealed with nail polish. The slides were kept in dark before being imaged on Zeiss confocal laser scanning microscope (LSM) 700. Foci area were defined by the overlapping signal between Cy3-(CAG)₁₀ probe and DyLight647 labelled secondary antibody. The foci area was quantified using ImageJ software.

IR Pre-mRNA Mis-splicing Assay.

Approximately 120,000 HeLa cells were plated in each well of a 6-well plate in DMEM medium supplemented with 10% FBS the day before transfection. Cells were then transfected with 500 ng IR and 750 ng DT960 or DT0 plasmid with Lipofectamine 2000 (Life Technologies) according to the manufacturer’s recommended procedures. After 4 h, the transfection cocktail was changed to DMEM medium supplemented with 10% FBS and the ligand was added to the medium to reach the final testing concentration. The cells were harvested after incubation at 37 °C for 3 d. The cells were washed once with 1X PBS and detached using trypsin with 0.05% EDTA (Fisher Mediatech). RNA was immediately isolated using Total RNA kit I (Omega Bio-Tek) following manufacturer procedures. An aliquot of 1 μg of isolated RNA was reverse transcribed to cDNA with iScript cDNA synthesis kit (Bio-Rad). The reverse transcription reaction was purified using QIAquick PCR purification kit (Qiagen). From the resulting cDNA, 70 ng of cDNA was subjected to 35 cycles of PCR amplification using gene specific primers. The PCR products were run on an 8% polyacrylamide gel with 1X TBE buffer at 250 V for 30 min. The gel was post stained with ethidium bromide and subsequently imaged with Gel Doc XR+ system (Bio-Rad). The band intensity was quantified using ImageJ (NIH). The forward primer was 5’-GTA CCA GCT TGA ATG CTG CTC CT-3’, and the reverse primer was 5’-CTC GAG CGT GGG CAC GCT-4’.