Figure 1. DM1 pathogenetic mechanisms and therapeutic strategies. The actions of molecular therapies for DM1 at different pathogenetic levels are illustrated: (1) at DMPK gene, drugs can inhibit CTG-repeat transcription and induce repeat contraction; ZFN, TALEN or CRISPR/Cas9 nucleases can modify gene sequence by inducing CTG-repeat contractions or deletions, or by inserting premature polyadenylation signals; (2) mutated DMPK mRNA can be functionally inactivated by drugs inducing degradation or binding to CUG repeats; (3) MBNL can be released from CUG repeats by disruption of MBNL:CUG interaction through competitive binding, and CELF levels can be regulated by protein kinase C and glycogen synthase kinase 3β; (4) altered signaling pathways downstream of DMPK transcript can be rescued by modulation of splicing and miRNAs; circRNAs and RAN translation could also be targets of future therapies. Other mechanisms involved in DM1 pathogenesis are repeat-associated non-ATG (RAN) translation (reviewed in [24]), which results in the production of toxic protein aggregates containing polyglutamine from antisense CAG-repeated transcripts ^[25][26][25,26], microRNA (miRNA) dysregulation ^{[27][28][29][30]}[27,28,29,30], and upregulation of circular RNA (circRNA) expression ^[31][32][33][31,32,33]. In addition to the ones described above, other signaling cascades are affected by the toxic DMPK RNA and may play important roles in DM1 pathogenesis. For example, MBNL and CELF regulators, besides being key splicing regulators, are likely involved in cytoplasmic pathogenetic processes altering proteostasis and sarcomere structure (reviewed in [34]). Omics studies have added new information to the previous knowledge, revealing several alterations in gene expression, alternative splicing, CpG methylation and proteins levels that potentially contribute to DM1 pathogenesis. In perspective, these new approaches can be crucial to evaluate the degree of therapeutic rescue and the off-target effects of drug candidates (reviewed in [35]).

3. DM1 Cell and Animal Models

In vitro models of DM1 have greatly contributed to clarifying the pathogenetic mechanisms of the disease. Among these, there are engineered cell lines with CTG repeats of different lengths inserted in minigenes ^[36][37][36,37], DM patient-derived primary myoblasts, immortalized myoblasts or MYOD1-converted fibroblasts ^{[38][39][40][41]}[38,39,40,41], and embryonic stem cells [42]. Additionally, induced pluripotent stem cells (iPSCs) and iPSC-derived distinct cell types were used to study tissue-specific DM1 pathological alterations ^{[43][44][45][46]}[43,44,45,46]. Recently, the first 3D in vitro human-muscle model of DM1 was developed by encapsulating patient-derived MYOD1-converted fibroblasts in hydrogels [47]. All of these cell models reproduce molecular alterations typical of DM1 and have been very useful for discovering crucial molecules and cellular pathways involved in the disease and for testing therapeutic strategies. Drosophila models have also been used by several groups to study DM1. Interrupted CTG repeats of various lengths driven by either constitutive or inducible promoters were expressed in flies and DM1-related molecular as well as phenotypic alterations were observed in flies carrying more than 480 repeats ^[48][49][50][48,49,50]. Although DM1 modeling is complicated by the multifaceted impact of the DM1 mutation, many DM1 mouse models have been generated over time through the silencing of the Dmpk gene or the Mbnl family genes; alternatively, mouse models expressing CELF proteins or toxic CTG repeats in various tissues were produced, in order to mimic the different aspects of the disease and to discover therapeutic molecules (Table 1). It is unclear whether the DMPK haploinsufficiency observed in DM1 patients may affect functions of the tissues in which the gene is normally highly expressed, such as muscles and the central nervous system (CNS). To address this question, different Dmpk-knockout (KO) mouse models have been generated and characterized through the years with different results. Initial reports on Dmpk-KO mice described cardiac conduction defects [51] and mild myopathy [52]. Since these mice models were characterized by a mixed genetic background possibly leading to confounding effects, more recently a Dmpk-KO model backcrossed to two different pure genetic backgrounds was generated. This model did not confirm the previous observations, but showed that Dmpk gene deletion does not compromise cardiac or skeletal-muscle function [53]. Additionally, DMPK transcript silencing through antisense oligonucleotides (ASOs) was well tolerated in mice, rats and monkeys [54]. These findings suggest that reduction in DMPK expression should not be a prominent cause of the disease. Given the crucial regulatory role of MBNL proteins in DM1, different Mbnl KO models were generated to elucidate the role of each MBNL protein in the disease. Mbnl1 and Mbnl2 loss of function resulted in muscular and CNS symptom manifestation, respectively ^[55][56][57][55,56,57], while Mbnl3 KO caused a progressive delay in muscle regeneration and embryonic muscle differentiation abnormalities, in agreement with their expression profiles during development [58]. Mice with double Mbnl1/Mbnl2 KO or Mbnl1/Mbnl3 KO exhibited more severe phenotypes compared with the single KO ^[59][60][59,60] and the triple Mbnl1/2/3 KO in muscle tissues recapitulated the severe phenotype observed in congenital DM1, in both newborn and adult mice [61], supporting the idea of a prominent role of MBNL proteins and alternative splicing dysregulation in DM1 pathogenesis. To determine the role of CELF proteins in DM1 pathogenesis, mouse models overexpressing CELF1 and CELF2 in skeletal and/or cardiac muscle were generated ^{[62][63][64][65]}[62,63,64,65]. Overexpression of CELF1 was shown to reproduce DM1-associated histopathological and functional changes [63]. Notably, CELF1/2 overexpressing mouse models have revealed a strong pattern of antagonistic regulation of mRNA levels by CELF and MBNL proteins through competitive binding to 3′ UTR regions [64] (Table 1A).