Deregulations of miR‐1 and its target Multiplexin promote dilated cardiomyopathy associated with myotonic dystrophy type 1

Abstract Myotonic dystrophy type 1 (DM1) is the most common muscular dystrophy in adults. It is caused by the excessive expansion of noncoding CTG repeats, which when transcribed affects the functions of RNA‐binding factors with adverse effects on alternative splicing, processing, and stability of a large set of muscular and cardiac transcripts. Among these effects, inefficient processing and down‐regulation of muscle‐ and heart‐specific miRNA, miR‐1, have been reported in DM1 patients, but the impact of reduced miR‐1 on DM1 pathogenesis has been unknown. Here, we use Drosophila DM1 models to explore the role of miR‐1 in cardiac dysfunction in DM1. We show that miR‐1 down‐regulation in the heart leads to dilated cardiomyopathy (DCM), a DM1‐associated phenotype. We combined in silico screening for miR‐1 targets with transcriptional profiling of DM1 cardiac cells to identify miR‐1 target genes with potential roles in DCM. We identify Multiplexin (Mp) as a new cardiac miR‐1 target involved in DM1. Mp encodes a collagen protein involved in cardiac tube formation in Drosophila. Mp and its human ortholog Col15A1 are both highly enriched in cardiac cells of DCM‐developing DM1 flies and in heart samples from DM1 patients with DCM, respectively. When overexpressed in the heart, Mp induces DCM, whereas its attenuation rescues the DCM phenotype of aged DM1 flies. Reduced levels of miR‐1 and consecutive up‐regulation of its target Mp/Col15A1 might be critical in DM1‐associated DCM.

The primary cause of DM1 is the gain-of-function of toxic transcripts carrying expanded noncoding CUG repeats that aggregate into foci in nuclei, sequestering RNA-binding protein Muscleblindlike 1 (MBNL1) (Fardaei et al, 2001(Fardaei et al, , 2002. Reduced MBNL1 levels and concomitant stabilization of another RNA-binding protein, Elavlike family member 1 (CELF1) (Kuyumcu-Martinez et al, 2007), lead to the deregulation of alternative splicing with the abnormal expression of embryonic splice variants in adult tissues (Lee & Cooper, 2009). For example, the CELF1-dependent inclusion of fetal exon 5 in the adult isoform of cardiac troponin T (cTnT) has been associated with cardiac conduction defects in DM1 (Philips et al, 1998). Besides their roles as alternative splicing regulators, both MBNL1 and CELF1 are involved in mRNA translation (Timchenko et al, 2005(Timchenko et al, , 2006Dasgupta & Ladd, 2012;de Haro et al, 2013), deadenylation and decay (Vlasova et al, 2008;Dasgupta & Ladd, 2012;Masuda et al, 2012), and in RNA editing (Dasgupta & Ladd, 2012). MBNL1 is specifically involved in miRNA processing (Rau et al, 2011). These diverse functions of MBNL1 and CELF1 mean that DM1 may involve the deregulation of multiple pathways.
The reduction in MBNL1 and stabilization of CELF1 are thought to be involved in most DM1 phenotypes. Indeed, Mbnl1 knockout mice develop muscle myotonia, weakness/wasting, and cardiac defects including dilated cardiomyopathy and heart conduction block (Lee et al, 2013). Mice overexpressing CELF1 in the heart show conduction abnormalities and dilated cardiomyopathy (Koshelev et al, 2010) thus confirming the contribution of MBNL1 sequestration and CELF1 upregulation to DM1 pathogenesis. Overall, the mouse models reproduced multiple DM1 features including RNA foci formation and various alternative splice defects.
We generated a series of inducible Drosophila DM1 lines bearing UAS-iCTG constructs with 240, 480, 600, and 960 CTGs (Picchio et al, 2013). These lines were used to model DM1 in larval somatic muscles showing not only nuclear foci formation and Mbl sequestration but also muscle hypercontraction, splitting of muscle fibers, reduced fiber size, and myoblast fusion defects leading to impaired larva mobility (Picchio et al, 2013). The severity of phenotypes in these Drosophila models could be correlated with repeat size (Picchio et al, 2013), as also observed in DM1 patients. Finally, the overexpression of Drosophila CELF1 ortholog Bru3 and attenuation of MBNL1 counterpart mbl (Picchio et al, 2018; Auxerre-Planti e et al, 2019) offer further valuable models for identifying gene deregulations underlying DM1.
Among molecular mechanisms associated with DM1, the deregulation of miRNAs and in particular reduced levels of evolutionarily conserved muscle-and heart-specific miRNA, miR-1, has been reported in DM1 patients (Rau et al, 2011) and in DM1 models including mouse (Kalsotra et al, 2014) and Drosophila (Fernandez-Costa et al, 2013). However, the impact of miR-1 down-regulation on DM1-associated phenotypes has not yet been analyzed.
Here, we made use of Drosophila DM1 models to explore miR-1 involvement in cardiac dysfunction in DM1. We observed that dmiR-1 level was reduced in the cardiac cells of DM1 flies and that its down-regulation in the heart led to DCM, thus suggesting that reduced dmiR-1 levels contribute to DM1-associated DCM. Among potential dmiR-1 regulated genes from in silico screening, we identified Multiplexin (Mp)/Collagen15A1 (Col15A1) as a new cardiac dmiR-1 target involved in DM1. Both Mp and Col15A1 proteins were highly enriched in cardiac cells of DCM-developing DM1 flies and in heart samples from DM1 patients with DCM, respectively. Moreover, the heart-targeted overexpression of Mp was sufficient to induce DCM, whereas its attenuation rescues the DCM phenotype in DM1 flies. miR-1 and its target Mp/Col15A1 thus emerge as molecular determinants of DM1-associated DCM.

Heart-targeted dmiR-1 attenuation causes DCM in Drosophila
Reduced miR-1 levels had previously been observed in mice developing DCM (Isserlin et al, 2014) and in cardiac samples from patients with end-stage DCM (Ikeda et al, 2007). It had also been shown that miR-1 knockout mice display the DCM phenotype (Wei et al, 2014). However, whether miR-1 attenuation specifically within the heart leads to DCM has not been assessed. Here we tested the heart-specific knockdown (KD) of dmiR-1 in Drosophila using a sponge line that contains multiple complementary binding sites for dmiR-1. When expressed at high levels under the control of a cardiac-specific Hand-Gal4 driver, sponge sequences inhibit the activity of dmiR-1 in fly heart, thus preventing its function (Fulga et al, 2015). In the Hand > dmiR-1 sponge context, the adult fly hearts showed a larger diameter with an enlarged cardiac lumen (Fig 1B and B 0 ) in comparison to the control (Fig 1A and A 0 ). Consistent with this observation, analyses of semi-intact Hand > dmiR-1 sponge heart preparations and generated M-modes (Fig 1C and C 0 ) confirmed enlargement of the cardiac tube and showed significantly increased diastolic and systolic heart diameters in young and aged flies (Fig 1D and E) concomitant with myofibrillar disarray in the heart tube ( Fig EV1C and C 0 ). We then characterized the effects of dmiR-1 down-regulation on heart contractility by assessing percent fractional shortening (%FS), which refers to the size of the cardiac tube at the end of systole and the end of diastole. The cardiac dilation in Hand > dmiR-1 sponge flies was associated with a significant reduction in heart contractility at both 1 and 5 weeks of age ( Fig 1F). Sponge-driven dmiR-1 attenuation in the Drosophila heart thus leads to DCM. This finding is supported by significantly increased diastolic and systolic heart diameters observed in heterozygous dmiR-1 KO À/+ flies at 5 weeks of age (Fig EV2A-C).
DM1 flies develop a DCM phenotype and show a reduced dmiR-1 level in cardiac cells DCM accounts for fatal cardiac involvements in DM1 patients, but the gene deregulations underlying DM1-associated DCM have not been identified. To address this issue, we first tested whether Drosophila DM1 models developed DCM. We tested three heart-specific DM1 contexts, namely (i) overexpression of 960CTG repeats, (ii) overexpression of CELF1 ortholog Bru3, and (iii) attenuation of MBNL1 ortholog mbl. Since the severity of cardiac phenotypes in DM1 increases with age, we performed all analyses at 1 and 5 weeks of age. From the three heart-specific DM1 contexts the DCM phenotype was present in Hand > Bru3 and Hand > mblRNAi models at both 1 and 5 weeks of age (Fig 2A-F) but not in Hand > 960CTG (Fig EV1E-G). The DCM phenotype was also associated with myofibrillar disarray in the Hand > mblRNAi and Hand > Bru3 heart tubes ( Fig EV1A 0 and B 0 ) in comparison to the controls (Fig EV1A  and B). The Hand > 960CTG line presents cardiac dilation at 5 weeks of age characterized by a significant increase in diastolic and systolic diameters (Fig EV1E and F). Nonaffected contractility in these DM1 models ( Fig EV1G) could be due to a milder effect of 960CTG repeats on Bru3 and mbl levels compared with GAL4-driven overexpression of Bru3 and RNAi-knockdown of mbl.
Given that dmiR-1 attenuation leads to DCM and that Hand > Bru3 and Hand > mblRNAi DM1 models develop a DCM phenotype, we tested whether cardiac cells of Hand > Bru3 and Hand > mblRNAi flies showed reduced dmiR-1 levels. Using in situ hybridization (ISH) with dmiR-1-specific miRCURY LNA probe we found a significantly reduced dmiR-1 level in cardiac cells of both DCM-developing DM1 contexts (Figs 2G, G 0 , H, and I, and EV3A-K). Pre-dmiR-1 expression was also lower in young DM1 flies (Fig EV2D), whereas hearts from aged Hand > mblRNAi flies showed an increased pre-dmiR-1 level, most probably owing to its impaired processing (Fig EV2E). DCM-developing Drosophila DM1 models could thus serve to test and identify genes deregulated in the DM1-associated DCM context in response to a reduced dmiR-1 level.

dmiR-1 target Multiplexin is up-regulated in DCM-developing DM1 flies
To identify dmiR-1 target genes involved in DM1-associated DCM, we first performed in silico screening by mapping Drosophila-specific dmiR-1 seed sites annotated in miRBase (Griffiths-Jones et al, 2006; http:// www.mirbase.org/) on 3 0 UTR regions of Drosophila genes up-regulated in the hearts of two DCM-developing DM1 contexts at 5 weeks of age (Table EV1; Fig 3A) (Auxerre-Planti e et al, 2019). We identified a set of 124 candidate genes that contain a potential dmiR-1 seed site and among them Multiplexin (Mp) ( Table EV1, scheme in Fig 3A).
Mp, a Drosophila ortholog of mammalian Collagen XVIII and Collagen XV, belongs to the family of multi-domain collagens. It is composed of an N-terminal thrombospondin-related domain, followed by multiple Collagen repeats, a Collagen trimerization domain, and a Cterminal endostatin domain (Meyer & Moussian, 2009). Mp also contains consensus glycosaminoglycan (GAG) attachment sites, and biochemical analysis of the protein extracted from embryonic tissues revealed the presence of chondroitin sulfate (CS) chains, making it more like human collagen XV than collagen XVIII in this respect. In the embryonic Drosophila heart, Mp is expressed in cardioblasts of the heart properly but not in aorta (  A-C 0 (A) Control (UAS-dmiR-1 sponge) and (B) mutant (Hand > dmiR-1 sponge) adult hearts from 1-week-old flies labeled for F-actin (red). (A 0 , B 0 ) Cross-sections of cardiac tubes 3D-reconstructed using Imaris software. White line shows the enlargement of lumen cardiac tube in Hand > dmiR-1 sponge (B 0 ) in comparison with control (UAS-dmiR-1 sponge) (A 0 ). M-mode records from 1-week-old control (UAS-dmiR-1 sponge) (C) and Hand > dmiR-1 sponge (C 0 ) flies showing increased diastolic (green) and systolic diameters (red) in the dmiR-1 sponge context (C 0 ). D-F Box plots showing diastolic (D) and systolic (E) diameters and cardiac contractility analyses (percent fractional shortening) (F) performed using SOHA approach for controls (Hand-Gal4 and UAS-dmiR-1 sponge) and Hand > sponge dmiR-1 contexts at ages 1 and 5 weeks. n = 20 hearts. Central band corresponds to median. Whiskers correspond to Min to Max. Box corresponds to interquartile range from 25 th to 75 th percentile.  . The Mp was shown to be deposited in a polarized way along the heart lumen during cardiac tube formation (Meyer & Moussian, 2009) and involved in lumen shaping by enhancing Slit/Robo activity (Harpaz et al, 2013). Mp overexpression in the developing embryonic heart leads to an enlargement of the heart lumen and is sufficient to promote an increase of the embryonic aorta diameter to that of the heart proper (Harpaz et al, 2013). We thus reasoned that Mp could be involved in DM1-associated DCM.
Here, we show that Mp is also expressed in the adult fly heart (Fig EV4A-C″). Consistent with the predicted location of an extracellular matrix protein, Mp was detected on the luminal and external surfaces of the cardiomyocytes and was also present on the underlying adult heart ventral longitudinal muscles (VLM) (Fig EV4A-C″).
The predicted dmiR-1-binding site within the Mp3 0 UTR region ( Fig 3B) is expected to negatively regulate Mp transcript level in the presence of dmiR-1. To assess the biological relevance of this binding site in vivo, we cloned the Mp3 0 UTR fragment carrying the dmiR-1 seed site downstream of the GFP coding sequence to generate the UAS-GFP-Mp3 0 UTR transgenic line. In parallel, the UAS-GFP-ΔMp3 0 UTR line was created in which the dmiR-1 binding site was deleted from the 3 0 UTRMp sequence. Both GFP sensor lines were then combined with the UAS-dmiR-1 line to generate double transgenic lines UAS-GFP-Mp3 0 UTR; UAS-dmiR-1 ( Fig 3C, C 0 , D, and D 0 ) and UAS-GFP-ΔMp3 0 UTR; UAS-dmiR-1 ( Fig 3E, E 0 , F, and F 0 ), respectively. We found that the expression of dmiR-1 in Hand > GFP-3 0 UTRMp hearts significantly reduced the GFP signal in cardiac and pericardial cells (Fig 3D 0 ), suggesting that dmiR-1 binds to the predicted seed site and negatively regulates Mp mRNA expression in the adult fly heart. The deletion of the dmiR-1 binding site in Hand > GFP-Δ3 0 UTRMp; dmiR-1 heart prevented the repressive effects of dmiR-1 (Fig 3F 0 ), demonstrating that the GFP silencing observed in Hand > GFP-Mp3 0 UTR; dmiR-1 hearts (Fig 3D 0 ) is dmiR-1 dependent. This finding is also consistent with increased Mp protein levels detected in hearts of the Hand > dmiR-1-sponge context ( Fig EV4J) mimicking miR-1 attenuation and leading to DCM.
We then tested whether Mp protein level increased in DM1 contexts with DCM characterized by a reduced dmiR-1. We detected a significant increase in Mp protein level on the surface of cardiomyocytes in DCM-developing lines (Hand > mblRNAi and Hand > Bru3) in both young and aged flies (Fig 3G-L). A similar increase in Mp expression was found in DM1 pericardial cells (Fig EV4D-I).

Heart-targeted Mp overexpression leads to DCM
Mp protein level increases in DCM-developing DM1 hearts and in the heart-specific dmiR-1 attenuation context causing DCM. . The diastolic and systolic diameters of hearts overexpressing Mp were also significantly increased at both 1-and 5week-old Hand > Mp flies relative to controls (Fig 4E, F, H, and H 0 ) with myofibrillar disarray associated with the enlarged heart tubes ( Fig EV1D and D 0 ). The cardiac dilation was concomitant with significantly reduced fractional shortening and thus adversely affected cardiac contractility ( Fig 4G). These findings demonstrate that a heart-specific increase in Mp protein level is detrimental to cardiac function, leading to DCM in flies.
To further analyze the effects of Mp expression level on heart morphology and function, we tested the impact of cardiac-specific Mp knockdown at 1 and 5 weeks of age. The Hand > Mp RNAi adult flies expressed a lower Mp protein level in the heart ( Fig 5B 0 and C) compared with control (UAS-Mp RNAi) (Fig 5A 0 ). We observed a reduced heart lumen in Hand > Mp RNAi flies (Fig 5B) compared with controls ( Fig 5A) and a significant decrease in diastolic and systolic diameters at ages 1 and 5 weeks (Fig 5D and E) without effect on cardiac contractility ( Fig 5F).
Taken together, our findings demonstrate that Mp expression level plays a critical role in setting the size of the cardiac lumen and the systolic and diastolic fly heart variables, and its overexpression in the heart leads to DCM. We thus infer that the reduced dmiR-1 leading to the up-regulation of its direct target Mp promotes the development of DCM in the DM1 context (see scheme in Fig 4I).

Mp counterpart Col15A1 is up-regulated and miR-1 is downregulated in cardiac samples of DM1 patients with DCM
Given the increased Mp level in DM1 fly models developing DCM, we examined whether the expression of its human ortholog Col15A1 was also up-regulated in cardiac cells of DM1 patients. Of three DM1 cardiac tissue samples tested, two were from DM1 patients with DCM (Fig 6A and B). We observed that DM1 cardiac cells showed an increase in Col15A1 expression compared with cardiac cells from healthy donors that are nonsignificant at RNA levels ◀ Figure 2. DM1 flies develop DCM phenotype with a reduced dmiR-1 level in cardiac cells.
A-F Box plots showing cardiac tube size (diastolic (A, D) and systolic (B, E) diameters) and contractility (percent fractional shortening (C, F)) analyses performed using SOHA approach for controls (Hand-Gal4, UAS-Bru3, and UAS-mblRNAi) and DM1 contexts (Hand > Bru3 and Hand > mblRNAi), at ages 1 and 5 weeks. n = 20 hearts. Central band corresponds to median. Whiskers correspond to Min to Max. Box corresponds to interquartile range from 25 th to 75 th percentile. G, G 0 Representative spot views generated using Imaris from in situ hybridization with miRCURY LNA probe for dmiR-1 and used for quantification of dmiR-1 levels. Spot views of dmiR-1 in hearts of 1-week-old control (UAS-mblRNAi) (G) and DM1 flies (Hand > mblRNAi) (G 0 ) are shown. Each spot represents a pool of dmiR-1 transcripts detected in the same area. The zoom region in G corresponds to area in which FISH signal was quantified. H, I Scatter plot graph showing the signal intensity quantified in cardioblasts of 1-and 5-week-old flies for controls (UAS-Bru3, UAS-mblRNAi) and DM1 contexts (Hand > Bru3, Hand > mblRNAi). n = 9 hearts. Bars correspond to mean AE SD.
Ó 2023 The Authors EMBO reports 24: e56616 | 2023 ( Fig 6A) but significant at protein levels ( Fig 6B). We also noticed differentially higher Col15A1 levels in DM1 patients with DCM compared with cardiac cells from DM1 patients without DCM (Fig 6A  and B).
Since the up-regulation of Mp was associated with dmiR-1 lossof-function in DM1 flies, we tested the expression of miR-1 in heart samples from DM1 patients. We observed that DM1 patients showed significant down-regulation of miR-1 in cardiac tissue in comparison with controls ( Fig 6C). Thus, like in Drosophila DM1 models, the down-regulation of miR-1 and the concomitant upregulation of Col15A1 in the heart correlate with DCM in DM1 patients. Heart-specific attenuation of Mp rescues DM1-associated DCM phenotype To determine whether the increased Mp/Col15A1 levels could contribute to the DCM phenotype in DM1, we applied our DM1 fly models and performed genetic rescue experiments by attenuating Mp expression in the DCM-developing Hand > Bru3 context and we used UAS-UPRT; Bru3 as a negative control (UPRT encodes uracil phospho-ribosyltransferase enzyme). Young 1-week-old Hand > UPRT; Bru3 flies do not present DCM (Fig EV5A-C). However, 5week-old flies from this line show a significant increase in diastolic and systolic diameter in comparison to the control (Fig 6D and E) with significant decrease in heart contractility measured by fractional shortening (Fig 6F). When Mp expression is attenuated via RNAi in Hand > Bru3 DM1 context (Hand > Bru3;Mp RNAi), diastolic and systolic diameters are reduced in comparison to DM1 flies (Hand > UPRT; Bru3) (Fig 6D, E, G, H, and I), and the cardiac contractility defect is ameliorated and becomes similar to that of aged control flies (UAS-UPRT; Bru3) ( Fig 6F). Thus, heart-specific attenuation of Mp rescues the DCM phenotype in aged DM1 flies.

Discussion
Myotonic dystrophy type 1 is the most common muscular dystrophy in adults. Cardiac repercussions including DCM are among the main causes of death in DM1 (Groh et al, 2008

DM1 fly models show reduced dmiR-1 in cardiac cells and develop DCM
In humans, DCM is characterized by left ventricular dilation and systolic dysfunction defined by a depressed ejection fraction. Similarly, in DCM-developing flies, the cardiac tube is enlarged and shows an increased diastolic and systolic diameter with reduced contractility. The loss of cardiac miRNAs and in particular miR-1 has already been correlated to DCM and heart failure in mice (Rao et al, 2009;Wei et al, 2014) and observed in patients with end-stage DCM (Ikeda et al, 2007). miR-1 sequence is highly conserved between Drosophila and Human (Fig EV2F), and it is well known that it regulates genes involved in cardiac development and function including Nkx2.5, SRF, and components of WNT and FGF signaling pathways (Kura et al, 2020) and that its level is reduced in the pathological context of DM1 (Rau et al, 2011). However, it was not known whether the low miR-1 level caused DM1-associated DCM, nor what were the downstream miR-1 targets. Here, we show that two heart-targeting Drosophila DM1 models, Hand > mblRNAi and Hand > Bru3 mimicking sequestration of MBNL1 and stabilization of CELF1, respectively, developed DCM and showed a reduced expression of dmiR-1 in cardiac cells including cardiomyocytes and pericardial cells. Regarding the influence of Hand-Gal4 driven expression in pericardial cells on the DM1 heart phenotypes, we previously tested all our DM1 models using cardioblast-specific Tin-GAL4 driver. DM1 cardiac phenotypes such as conduction defects observed in the Hand > Bru3 model (Auxerre-Planti e et al, 2019) and DCM observed in Hand > mblRNAi and Hand > Bru3 models (Auxerre-Planti e et al, 2019 and this work) are observed when using Tin-Gal4 driver. These results suggest that the cardiac phenotypes observed in the DM1 Drosophila heart, including DCM, are mainly due to gene deregulations within the cardiomyocytes. Because the overexpression of CELF1 (Koshelev et al, 2010) and the loss of MBNL1 (Lee et al, 2013) also result in DCM in mice, Drosophila appears well-suited to assessing the impact of reduced miR-1 in DM1-associated DCM. One mechanism explaining why miR-1 levels fall in the DM1 context is the sequestration of MBNL1, which can no longer play its physiological role in pre-miR-1 processing into mature miR-1 (Rau et al, 2011). Here, we observed reduced dmiR-1 also upon the cardiac overexpression of CELF1 ortholog Bru3. How CELF1/Bru3 impinges on miR-1 levels is not fully understood, but it was demonstrated that CELF1 could bind UG-rich miRNAs (such as miR-1) and mediate their de-adenylation and degradation by recruiting poly(A)-specific ribonuclease (PARN) (Katoh et al, 2015). Given that Drosophila DM1 models developing DCM showed markedly reduced dmiR-1 in cardiac cells, we sought to determine Ó 2023 The Authors EMBO reports 24: e56616 | 2023 whether heart-targeted attenuation of dmiR-1 was sufficient to induce DCM: dmiR-1 knockdown in the heart mimics DM1associated DCM.
Col 15A1 ortholog Mp is a new miR-1 target involved in DM1associated DCM To identify candidate dmiR-1 target genes involved in DCM we performed in silico screening for dmiR-1 seed sites in the 3 0 UTR regions of genes up-regulated in cardiac cells at 5 weeks of age (Auxerre-Planti e et al, 2019) in DM1 models developing DCM (see Table EV1). Among 1,189 3 0 UTR sequences tested, we found that 162 bore potential dmiR-1 seed sites, including the 3 0 UTR of Multiplexin (Mp). Mp codes for extracellular matrix protein belonging to a conserved collagen XV/XVIII family. We top-ranked Mp because of its known role in setting the size of the cardiac lumen (Harpaz et al, 2013). The embryos overexpressing Mp display an enlarged cardiac tube and conversely, Mp À/À embryos were found to present a narrower lumen with reduced contractility of the heart tube (Harpaz et al, 2013). In parallel, the mouse mutants of Mp ortholog, Col15A1, showed age-related muscular and cardiac deterioration linked to a degraded organization of the collagen matrix (Eklund et al, 2001;Rasi et al, 2010). This prompted us to examine how Mp was expressed in the adult fly heart and what the effect of its overexpression was. Using Mp specific antibody we detected Mp on the surface of the cardiac cells and found that it accumulated to a high level in both Hand > mblRNAi and Hand > Bru3 DM1 lines. We also tested whether the in silico identified dmiR-1 seed site was required for the regulation of Mp expression and confirmed that Mp is a direct in vivo target of dmiR-1 in cardiac cells. As the potential binding site for human miR-1 is present also in 3 0 UTR of Col15A1 transcript (Fig EV2G) we hypothesize that Mp/Col15A1 are evolutionarily conserved miR-1 targets. Consistent with its role downstream of dmiR-1, Mp overexpression in the heart mimicked the dmiR-1 knockdown phenotype, leading to a significantly enlarged heart with reduced contractility. Moreover, heart-specific attenuation of Mp expression in the Hand > Bru3 DM1 context reduced heart dilation and rescued DCM phenotype in aged flies, thus demonstrating that increased Mp levels contribute to DCM observed in DM1 flies. Previous reports (Gil-Cayuela et al, 2016; Louzao-Martinez et al, 2016) revealed increased expression levels of different collagens associated with DCM in both animal models and patients. Here, we report evidence that Col15A1 is specifically upregulated at both transcript and protein levels in cardiac samples from DM1 patients and in particular in those with DCM, with downregulation of miR-1. Altogether, the observations that Col15A1 expression level is abnormally elevated in DCM-developing DM1 patients and that attenuation of its Drosophila ortholog Mp could ameliorate the DCM phenotype suggest that Col15A1 could be a novel therapeutic target in DM1.

DCM, a complex cardiac condition with a poor prognosis in DM1
A large number of genes have so far been implicated in DCM, attesting to the complex molecular origin of this cardiac condition. For example, in Drosophila, DCM was observed in mutants of genes encoding contractile and structural muscle proteins such as Troponin I (TpnI), Tropomyosin 2 (Tm2), d-sarcoglycan and Dystrophin but also associated with deregulations of EGF, Notch, Cdc42 and CCR4-Not signaling pathway components (reviewed in Wolf, 2012). In humans, DCM-causing mutations were also identified in a large number of genes including those encoding cytoskeletal proteins such as FLNC, nuclear membrane protein LMNA or involved in sarcomere stability (Titin, TNNT2, MYH7, TPM1) but also RNAbinding protein RBM20 (McNally & Mestroni, 2017).
Here, we focused on DCM associated with DM1. A previous study on a mouse model overexpressing CELF1 and developing DCM (Wang et al, 2015), identified down-regulation of Transcription factor A mitochondrial (Tfam), Apelin (Apln), and Long-chain fatty acid-CoA ligase 1 (Acsl1) as potentially associated with DCM. It was suggested that CELF1 might regulate their mRNA stability by binding to their 3 0 UTR regions and causing destabilization and degradation of their transcripts (Chang Kuei-Ting et al, 2017). In this DCM-developing mouse DM1 model, Col15a transcripts were elevated (Wang et al, 2015), but the role of Col15a in DCM was not analyzed. Here, using Drosophila DM1 models with a DCM phenotype, we identified up-regulation of Col15A1 ortholog Mp as a molecular determinant of DM1-associated DCM. We also link reduced miR-1 levels in DCM-developing DM1 cardiac cells to the up-regulation of Mp, establishing that Mp is an in vivo target of dmiR-1.
Importantly, our findings show that in DM1 patients, Collagen 15A1 is up-regulated in the hearts of patients with DCM. In DM1 patients, the DCM phenotype appears several years after onset and Scheme presenting cardiac role of dmiR-1 and its target Mp in DCM-developing Drosophila DM1 models. In wild-type, Drosophila heart mbl promotes pre-dmiR-1 processing. Bru3 has a potential antagonistic role in the destabilization of dmiR-1. As a result, in this context dmiR-1 and its target Mp levels are moderate. In DM1, mbl is sequestered and Bru3 is stabilized, causing inefficient processing of pre-dmiR-1 and destabilization of mature dmiR-1. As a result, in the DM1 context, the dmiR-1 level is reduced while its target Mp level is high. This leads to an enlarged heart with adversely affected contractility and myofibrillar disarray in the heart tubes. Ó 2023 The Authors EMBO reports 24: e56616 | 2023 is less common than the conduction system defects and arrhythmias (Nguyen et al, 1988;Lin et al, 1989). However, it is frequently associated with poor prognosis and indication for heart transplant (Papa et al, 2018). In summary, we report evidence for the importance of miR-1dependent gene deregulations in DM1 and identify Mp as a new miR-1 target involved specifically in DM1-associated DCM. We also demonstrate that Mbl depletion and Bru3 up-regulation in the heart have overlapping impacts on DM1 pathogenesis, both leading to reduced miR-1, up-regulation of Mp, and so to DCM (see scheme in Fig 4I).
Our conclusion is that in a physiological context, Mp level is moderately triggered by Mbl-dependent regulation of dmiR-1 processing and Bru3-dependent regulation of dmiR-1 stability. However, in the DM1 context, Mbl is sequestered in nuclear foci while Bru3 levels increase, leading to a reduced dmiR-1 and the upregulation of its target gene Mp. Considering the known role of Mp as a positive regulator of cardiac lumen size (Harpaz et al, 2013) we would expect Mp accumulation in the adult heart also to promote heart tube enlargement, leading to the DCM phenotype. Whether like in embryos this Mp function involves the Slit/Robo signaling pathway remains to be investigated, but the finding that Robo2 is among identified miR-1 targets up-regulated in DCM-developing DM1 flies (Table EV1) supports this possibility. Finally, the fact that Mp ortholog Col15A1 is highly elevated in cardiac samples from DM1 patients with DCM and that reducing Mp rescues the DCM  Ó 2023 The Authors EMBO reports 24: e56616 | 2023 phenotype in DM1 fly model suggests that Mp/Col15A1 could be an attractive diagnostic and/or therapeutic target for DM1-associated DCM.

Materials and Methods
Drosophila stocks All D. melanogaster stocks were grown and crossed at 25°C on standard fly food. In this study, we used three Drosophila DM1 models: UAS-960CTG (Picchio et al, 2013), UAS-mblRNAi (28732, VDRC Vienna, Austria), and UAS-Bru3 (Picchio et al, 2013). For dmiR-1 loss-and gain-of-function, we used UAS-dmiR-1 sponge (Fulga et al, 2015), All inducible lines were crossed with the driver line Hand-Gal4 (provided by Laurent Perrin, TAGC, Marseille, France) to induce the expression of transgenes specifically in the fly heart (cardioblasts and pericardial cells, with a low expression in VLM). Control lines were generated by crossing the above-cited UAS lines with w 1118 line. Hand-Gal4/+ control was generated also by crossing Hand-Gal4 driver line with w 1118 . w 1118 is a mutant strain with a recessive white-eye phenotype.

Heartbeat analyses of adult Drosophila hearts
Physiological analyses of adult Drosophila hearts were performed on 1-and 5-week-old female flies using the Semi-automatic Optical Heartbeat Analysis (SOHA) approach protocol of Ocorr et al (2009). For each genotype, about 20 flies were analyzed. First, we proceeded to dissection and preparation of semi-dissected hearts: adult Drosophila flies were anesthetized with Fly Nap for 5 min, then placed, dorsal side down, into a petri dish coated with a thin layer of petroleum jelly. The head, the ventral thorax, and the ventral abdominal cuticle were removed in order to expose the abdominal contents. The ovaries and other organs as well as fats were then removed in order to expose heart tube attached to the dorsal cuticle. Dissection was performed in an oxygenated, artificial hemolymph (AH) solution composed of 108 mM Na + , 5 mM KCl, 2 mM CaCl 2 , 8 mM MgCl 2 , 1 mM NaH 2 PO 4 , 4 mM NaHCO 3 , 10 mM sucrose, 5 mM trehalose, 5 mM HEPES (pH 7.1, all reagents from Sigma Chemicals). The solution was oxygenated by air-bubbling for 15 to 20 min. The beating hearts were filmed by a digital camera on 30 s movie with the speed of 150 frames/s (Digital camera C9300, Hamamatsu, McBain Instruments, Chatsworth, CA) using microscope Zeiss (Axiophot, Zeiss) using immersion objective 10×. The heartbeats were recorded in A3 and A4 segments. The SOHA program based on Matlab R2009b software has been used for film analysis: The cardiac tube membrane during maximum diastole (relaxation) and maximum systole (contraction) were defined manually. One pair of marks identified the diastolic diameter, and one pair identified the systolic diameter. From this vertical row of pixels, an Mmode was generated to analyze the contraction and relaxation intervals: diastolic (DD), and systolic (SD) diameters, heart period (HP), and systolic (SI) and diastolic (DI) intervals. The diastolic and systolic diameters were used to calculate the fractional shortening (FS) using the formula: (((Diastolic diameter À Systolic diameter)/Diastolic diameter) × 100) (Ocorr et al, 2009).

Quantification of Multiplexin immunolabeling
Fluorescent-labeled heart tissues were all processed and stained under the same conditions. The level of fluorescent signal was quantified using ImageJ software by CTCF (Corrected Total Cell Fluorescence) approach. CTCF is calculated according to the formula: Measured by the software Integrated Optical Density À (Area of select × mean fluorescence of background readings). For each heart, CTCF was measured in two regions from segment A3 and two regions from segment A4. For each genotype, 9 hearts were analyzed for cardioblasts with 4 measurements in regions for each, and 3 pericardial cells were analyzed from each heart.

Fluorescence in situ hybridization (FISH) using LNA-enhanced probes
For the detection of dmiR-1, we used fluorescence in situ hybridization (FISH) using double-labeled (enhanced) miRCURY LNA probes. First, flies 1-and 5-week-old were dissected and fixed for 30 min with 4% paraformaldehyde; rinsed at PBS1X-Tween 0.1% threetime, 5 min each. Samples were dehydrated through a series of increasing ethanol concentrations, transferred sequentially to 25, 50, 75, and 100% ethanol baths for 10 min each. After removing ethanol tissues were rehydrated through a series of decreasing ethanol concentrations, transferred sequentially to 50%, 25% ethanol baths for 10 min each, and then postfixed for 30 min with 4% paraformaldehyde. Then incubated with a solution composed of 50% PBT and 50% hybridization buffer (5 ml Formamide, 0.5 ml SSC 20X, 100 ll ssDNA, 20 ll tRNA 50 ng/ll, 5 ll Heparin 100 ng/ll, 10 ll Tween) for 5 min. The samples were finally incubated with hybridization solution for 1 h 30 min at 50°C. The hybridization mix including specific DIG-labeled probes (1 nM for U6 snRNApositive control ref 339459, 20 nM for negative control scramble ref 339459 and 40 nM for DME-miR-1-3p ID: 339111) was added in each tube and samples were incubated at 50°C overnight. To remove nonspecific RNA hybridization, samples were washed with ISH (5 ml Formamide, 0.5 ml SSC 20X, 100 ll ssDNA, 5 ll Heparin 100 ng/ll, 10 ll Tween) for 30 min each at 50°C then with PBT. Slides were blocked in western blockage reagent (3 ml blockage reagent + 12 ml PBT) for 30 min, incubated with sheep anti-DIG POD antibodies (11633716001, Roche) for 2 h, treated with the TSA Plus Cyanine 3 System 1% (PerkinElmer, USA) for 5 min, blocked in 20% NGS diluted in PBT for 30 min. Finally, samples were incubated with primary antibodies (anti-Actin 1/250) overnight at 4°C and then with secondary antibodies coupled with Alexa448 (1/150). The preparations were mounted in a Vectashield with a DAPI medium. Immunofluorescence labeling preparations have been then analyzed using a confocal microscope Leica SP8.

FISH quantification
Stained Drosophila hearts were imaged in 3D to allow the quantification of RNA transcripts in the total volume of cardiac and pericardial cells. The transcripts hybridized with miRCURY LNA-enhanced probes appear on the images as luminous fluorescent spots. Each spot represents several miR-1 transcripts. The 3D images were analyzed using Imaris (version 9.3.1) that allows the detection, the visualization of spatial distribution, and the quantification of intensity for each spot detected. To detect the real RNA spots of interest we have adjusted the threshold of intensity detection according to scramble. To determine the lower threshold in pericardial cells, we have analyzed the spots detected in 27 pericardial cells in 9 heart preparations labeled by scramble. We adjusted the lower threshold above the mean intensity of scramble spots (30) and the upper threshold (255) by default. With this background subtracting parameter, Imaris software generated dmiR-1 spots and calculated the mean intensity for each spot detected and the average of the mean intensities for all dmiR-1 spots. Similar approach has been applied for dmiR-1 FISH quantification in cardioblasts.

RNA extraction and RT-qPCR on adult fly heart samples
Total RNA was isolated from 20 adult hearts from 1-and 5-week-old female flies, using Direct-zol TM RNA Microprep (ref: R2060) from Zymo Research following the manufacturer's instructions. RNA quality and quantity were, respectively, assessed using Agilent RNA Screen Tape Assay on 4200 TapeStation System (Agilent Technologies) and Qubit RNA HS assay kit on a Qubit 3.0 Fluorometer (Thermo Fischer Scientific). Then, 150 ng total RNA was reverse transcribed using SuperScript IV Reverse Transcriptase kit (Invitrogen) with random primer mix, in a 20 ll reaction. Quantitative PCR was performed in 4 replicates in the final volume of 10 ll using Light SYBR Green PCR Master Mix (Roche, Applied Science) on a LightCycler 480 Real-Time PCR System (Roche, Applied Science). A 2 ll (3 ng) of cDNA were added to an SYBR Green Master Mix. We used Rp49 as a reference gene. The following pairs of primers were used: Rp49: forward GCTTCAAGGGACAGTATCTG and reverse AAACGCGGTTCTGCATGAG; pre-dmiR-1: forward TTCAGCCTTTG AGAGTTCCATG and reverse CGCCAGATTTCGCTCCATAC. The relative quantifications of transcripts were obtained with the DDCt method. Finally, nonparametric Mann-Whitney tests were performed to compare control samples and samples of interest.

Generating transgenic fly lines
To validate Mp as a direct target of dmiR-1 in vivo, we have generated double transgenic fly lines. For the generation of UAS-GFP-3 0 UTRmp line, approximately 470 base pairs surrounding the predicted dmiR-1-target site in 3 0 UTR of Mp were amplified directly from the genomic DNA from the w 1118 flies using the primers Mp-F1: ATAACTAGTTGAGCGGAAACGGAAGGAAGAAGAGGAG and Mp-R1: ATATCTAGATGTTGTGAATGATGACGTTAGG and a highfidelity DNA Polymerase enzyme (Thermo Scientific Phusion High-Fidelity DNA Polymerase) kit. For the UAS-GFP-Δ3 0 UTRmp line, the same steps were proceeded with primers Mp-F2: ATAACTAGTTGA TAAAACAAAACAAATCACAGCAC and Mp-R1: ATATCTAGA TGTTGTGAATGATGACGTTAGG to amplify about 350 bp without predicted dmiR-1-target site. The SpeI and XbaI restriction sites were incorporated into primers and introduced by PCR. PCR products were purified using NucleoSpin Plasmid clean-up kit after validation of inserts by electrophoresis on 1% agarose gel. After digestion of purified 3 0 UTRMp fragments and pUASP-PL-Venus vector by SpeI and XbaI enzymes, we performed ligation between 3 0 UTRMp fragments and the vector using the T4 DNA ligase kit (Invitrogen) according to the manufacturer 0 s instructions. Purified vectors were then microinjected by the Fly Facility platform to generated transgenic lines. Finally, UAS-GFP-3 0 UTRMp and UAS-GFP-Δ 3 0 UTRMp transgenic lines were combined with UAS-miR-1 (41125, Bloomington, USA) to generate UAS-GFP-3 0 UTRMp; UAS-miR-1 and UAS-GFP-Δ3 0 UTRMp; UAS-miR-1, respectively, and then crossed with the driver line Hand-GAL4.

RNA extraction, RT-qPCR, and immunoblot on DM1 human hearts
Human ventricular cardiac muscle tissues were obtained at autopsy from 3 DM1 patients and 3 normal controls following informed consent. All experimental protocols were approved by the Institutional Review Board at Osaka University and carried out in accordance with the approved guidelines. Total mRNA was extracted and first-strand complementary DNA synthesized as described previously (Nakamori et al, 2008). RT-qPCR was performed using TaqMan Gene Expression assays (Hs00266332_m1 and 4333760F, Applied Biosystems) on an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems), as described previously (Nakamori et al, 2011). The level of COL15A1 mRNA was normalized to 18 S rRNA. For miR-1 expression analysis, micro-RNA was extracted from human cardiac muscle tissues by using miRNeasy Micro Kit (Qiagen). Follow-up qPCRs were performed using TaqMan Advanced miRNA cDNA Synthesis Kit (Applied Biosystems) and TaqMan Advanced miRNA Assays (477820_mir) normalized by miR-16-5p (477860_mir). For protein analysis, cardiac muscle tissues were homogenized in a 10× volume of radioimmunoprecipitation assay buffer (25 mM Tris-HCl; pH 7.5; 150 mM NaCl; 1% NP-40; 1% sodium deoxycholate; and 0.1% sodium dodecyl sulfate) containing a protein inhibitor cocktail (Sigma-Aldrich). The homogenate was centrifuged for 10 min at 10,000 g and the supernatant was collected. Equal amounts of protein (40 lg) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto Immobilon-P membranes (Millipore), as previously described (Nakamori et al, 2008). Blots were blocked for nonspecific protein binding with 5% (w/v) nonfat milk and then incubated with a 1:500-diluted antibody against COL15A1 (Thermo Fisher Scientific) or 1:3,000-diluted antibody against GAPDH (glyceraldehyde 3-phosphate dehydrogenase) (Sigma-Aldrich). After repeated washings, the membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG (Thermo Fisher Scientific). The ECL Prime Western Blotting Detection Reagent (Cytiva) and ChemiDoc Touch Imaging System (Bio-Rad) were used to detect the proteins.

Statistics
Nonparametric Mann-Whitney tests were performed to compare control samples and samples of interest in Drosophila model, and the t-test was performed to compare controls to DM1 context from heart samples of DM1 patients. All statistical analyses were performed using GraphPad Prism (version 8.0.1) software. Results are reported with P-value < 0.05 considered statistically significant.

Data availability
No primary datasets have been generated and deposited. Previously generated sequencing data have been deposited with the GEO-NCBI tracking system under accession number GSE109370 (http://www. ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE109370) and used here to identify potential dmiR-1 targets listed in Table EV1.
Expanded View for this article is available online. Figure EV1. Flies showing DCM present myofibrillar disarray in the heart tubes.

▸
A-D 0 Z stack projection from hearts of controls (UAS-mblRNAi, UAS-Bru3, UAS-dmiR-1 sponge and UAS-Mp) (A-D) and mutants (Hand > mblRNAi, Hand > Bru3, Hand > dmiR-1 sponge and Hand > Mp) (A 0 -D 0 ) aged of 5 weeks. The arrows point to irregularity in myofibrillar density and arrangement in the heart tubes. Scale bar = 20 lm. E-G Box plots showing cardiac size analyses (diastolic (E) and systolic (F) diameters) and percent fractional shortening (G) performed by SOHA approach for controls (Hand-Gal4 and UAS-960CTG) and mutant (Hand > 960CTG) at 1 and 5 weeks of age. Aged Hand > 960CTG flies present cardiac dilation characterized by a significant increase in diastolic and systolic diameters in comparison with control Hand > Gal4 (E, F). n = 20 hearts. Central band corresponds to median. Whiskers correspond to Min to Max. Box corresponds to interquartile range from 25 th to 75 th percentile.
A-C Scatter plot showing cardiac size analyses (diastolic (A) and systolic (B) diameters) and percent fractional shortening (C) performed by SOHA approach for control wt 1118 and heterozygous mutant context (dmiR-1 KO À/+ ) at 1 and 5 weeks of age. Aged Hand > dmiR-1 KO À/+ flies present cardiac dilation characterized by a significant increase in diastolic and systolic diameters in comparison with control. n = 20 hearts. Bars correspond to mean AE SD. D, E RTq-PCR analysis for pre-dmiR-1 transcript in adult heart of 1 and 5 weeks of age for control (Hand > LacZ) and DM1 contexts (Hand > mblRNAi, Hand > Bru3). n = 4 biological replicates. F Alignment of Drosophila and human miR-1 sequences. G Potential binding site of human miR-1 in 3 0 UTR of COL15A1 (COL15A1-201 transcript position 4428 to 4447).