Dosage effect of multiple genes accounts for multisystem disorder of myotonic dystrophy type 1

Multisystem manifestations in myotonic dystrophy type 1 (DM1) may be due to dosage reduction in multiple genes induced by aberrant expansion of CTG repeats in DMPK, including DMPK, its neighboring genes (SIX5 or DMWD) and downstream MBNL1. However, direct evidence is lacking. Here, we develop a new strategy to generate mice carrying multigene heterozygous mutations to mimic dosage reduction in one step by injection of haploid embryonic stem cells with mutant Dmpk, Six5 and Mbnl1 into oocytes. The triple heterozygous mutant mice exhibit adult-onset DM1 phenotypes. With the additional mutation in Dmwd, the quadruple heterozygous mutant mice recapitulate many major manifestations in congenital DM1. Moreover, muscle stem cells in both models display reduced stemness, providing a unique model for screening small molecules for treatment of DM1. Our results suggest that the complex symptoms of DM1 result from the reduced dosage of multiple genes.


Introduction
Myotonic Dystrophy type 1 (DM1) is a complex disease with variable pathological phenotypes, disease severity, and onset ages [1][2][3]. The major symptoms include myotonia, muscle wasting, muscle weakness, cardiac conduction defects, cataracts and insulin resistance [3]. DM1 is a genetic disease caused by the expansion of a CTG repeat in the 3'-untranslated region of the Meanwhile, this model cannot mimic CDM symptoms [39]. The potential reason might be that 25 the random insertion of transgenes in chromatin does not affect the expression of Dmwd-Dmpk-Six5 locus, which may be also involved in the complex symptoms of DM1 [39]. Taken together, mouse DM1 models that better recapitulate the varieties of human symptoms are required to fully understand the underlying mechanism of DM1.
The multisystem symptoms of DM1 may be caused by the combination of different 30 mechanisms, leading to down-regulation of multiple genes, including Dmwd-Dmpk-Six5 and Mbnl1 [40,41]. However, the direct evidence to support the notion is missing due to the challenges to generate mice carrying multiple gene mutations simultaneously. It is time and labor consuming to generate mice carrying triple or quadruple mutations using conventional methods.
Recently, mouse androgenetic haploid embryonic stem cells (AG-haESCs) have been successfully developed as sperm replacement to efficiently produce semi-cloned (SC) animals by 5 injection into oocytes (intracytoplasmic AG-haESC injection, ICAHCI) [42][43][44]. AG-haESCs enable one-step efficient and stable generation of mice with multiple heterozygous mutant genes by ICAHCI of haploid cells carrying these mutations [42], allowing production of sufficient numbers of SC offspring for analyses in one generation. Thus, the haploid ESC-mediated SC technology may provide us an ideal tool to generate mouse models with multiple heterozygous 10 mutant genes in one step to mimic the reduced expression of multiple genes in human complex diseases.
In this study, we tested our hypothesis by generating triple and quadruple heterozygous mutant mice in one step through injection of haploid ESCs (O48 cell line used in this study) carrying triple or quadruple mutant genes (Dmpk, Six5 and Mbnl1 or Dmpk, Six5, Mbnl1 and Dmwd) into 15 oocytes. Mice with triple mutations exhibited most of the major pathogenic phenotypes observed in adult-onset DM1 patients. Mice with quadruple mutations could mimic symptoms of patients from the most severe form of DM1, congenital DM1 (CDM). Interestingly, differentiation of muscle stem cells (MuSCs) is defective in both models due to stemness reduction, which recapitulates the defects in human DM1 patients and provides a novel system for screening of 20 drugs to treat muscle problems in DM1 in the future.

Generation of a novel DM1 model carrying mutations in Dmpk, Six5 and Mbnl1
We first examined the feasibility of SC technology to generate mouse models of DM1 carrying single mutant gene by injection of haploid cells carrying mutation in Dmpk, Six5 or Mbnl1, the three well-studied DM1-related genes. A haploid cell line (termed H19 △DMR -IG △DMR -AGH or 5 O48) that has been reported to efficiently support SC mice generation [42] was used in this study.
We then set out to generate SC mice carrying triple mutations of Dmpk, Six5 and Mbnl1 in one 15 step using ICAHCI. We disrupted Six5 and Mbnl in ΔDmpk-O48-1 cells that have been analyzed as shown in Figure S1 and generated stable haploid cell lines carrying triple knockouts (termed ΔDSM-O48) (Figure 1a and 1b). Whole-genome sequencing analysis showed no off-target effect in ΔDSM-O48-1 cells (Table S2). ICAHCI experiments showed that ΔDSM-O48 cells (from two cell lines, i.e., ΔDSM-O48-1 and ΔDSM-O48-2) could reproducibly produce live SC pups 20 (termed DSM-TKO SC mice) after injection into oocytes (Figure 1b and Table S1). Over 90% of DSM-TKO SC pups grew up to adulthood and showed the similar growth profiles as those of WT SC mice (Figure 1c, S4a and Table S1). We then examined muscle phenotypes commonly observed in DM1 patients, including myotonia, muscle weakness and wasting in DSM-TKO SC mice. Electromyography (EMG) tests in skeletal muscle demonstrated myotonia in adult DSM-25 TKO SC mice (Figure 1d). Mouse treadmill assay and grip strength test revealed that DSM-TKO SC mice displayed muscle weakness (Figure 1e and 1f); and rotarod test indicated that DMS-TKO SC mice exhibited severe motor defects (Figure 1g). Histological analysis of tibialis anterior (TA) muscles from DMS-TKO SC mice revealed several main histological hallmarks of muscles from DM1 patients, including increased number of nuclear clump and decreased fiber size (shown as the myofiber cross-sectional area (CSA)), a sign of muscle wasting (Figure 1h and 1i). Meanwhile, consistent with the clinical observations, dystrophin (Dys) expression in DMS-TKO SC mice was normal ( Figure S4b). Fiber type analysis indicated an increased ratio of type I (slow) myofiber and a decreased CSA of type I in DSM-TKO SC mice ( Figure S4c).
Abnormalities of diaphragm muscle and small intestine were also observed ( Figure S4d and S4e), 5 implying the potential breathing and digestive dysfunctions in DMS-TKO SC mice. We next analyzed the cardiac structure and function in adult DMS-TKO SC mice because heart abnormalities are common in DM1 patients [45]. Echocardiography did not show obvious structure and functional abnormalities in 4-month-old DMS-TKO SC mice ( Figure S4f and S4g).
These data demonstrate that Dmpk +/-; Six5 +/-; Mbnl1 +/-SC mice can be generated in one step using haploid cells carrying triple mutations. DSM-TKO SC mice mimic the reduced dosage of three genes and display more severe pathological consequences compared to DM1 models 15 carrying single homozygous mutant (Table S3), providing a new model of DM1. However, DSM-TKO SC mice cannot mimic several phenotypes frequently observed in DM1 patients, such as cataracts and symptoms observed in CDM patients, implying that other genes may be involved.

Dmwd is involved in DM1
It has been shown that the expansion of the CTG repeats in Dmpk produces allele-specific effects on transcription of the two adjacent genes, Six5 (downstream of Dmpk) and Dmwd  (Figure 2b and 2c). By injecting these cells into oocytes, live SC pups could be efficiently produced (Figure 2d and  Table S1). Dmwd +/-SC mice grew up to adulthood normally (Figure 2e and Table S1).
Phenotype analysis showed that the myofiber CSA was dramatically reduced in adult Dmwd +/mice ( Figure 2f-i), suggesting that Dmwd is involved in pathological mechanism of DM1. But other symptoms of DM1, such as cardiomyocytes defects and cataracts, cannot be observed in Dmwd +/mice, suggesting that Dmwd is not sufficient to account for all the complex multisystem 5 symptoms of DM1.

Generation of congenital DM1 (CDM) mice carrying quadruple mutations
Next, we tested whether the compound loss of Dmpk, Six5, Mbnl1 and Dmwd could give rise to a mouse model recapitulating the majority of the symptoms in DM1 patients and the more severe  (Table S2).
ICAHCI analysis of three lines (ΔDSMD-O48-1, -2 and -3) generated Dmpk +/-; Six5 +/-; Mbnl1 +/-; Dmwd +/-SC mice (DSMD-QKO SC mice) efficiently (Table S1, Figure S5a (Table S1). We sacrificed DSMD-QKO SC pups at postnatal day 2 (P2) and observed residual milk in their stomach, excluding the possibility that they died of feeding difficulties. P2 pups exhibited overt abnormal intestines ( Figure S5f implying that other genes, such as MBNL2 or CUGBP1 that might be involved in DM1 patients, should be included in our mouse models in future. Since mis-splicing is a characteristic feature of DM1 and the dosage of Mbln1 is also reduced in our models, we next analyzed splicing profiles of 7 DM1-related genes in P2, P10 and adult mice 10 (4-month old). RT-qPCR analysis demonstrated the misregulation of Ldb3, Serca1, m-TTN, Tmem63b, Sorbs1, and Spag9 mRNA splicing( Figure S8a-b), while the splicing efficiencies of Dysf were unchanged similar as that in the DM1 patients ( Figure S8c). In addition, we also checked the splicing of Clcn1, Ryr1, Ryr2, Tnnt2, and Tnnt3 mRNA, and they were showed unchanged in our mouse models (data not shown), reflecting the heterogeneity of mis-splicing in 15 patients [59, 60] and suggesting that other genes may be involved in our models to mimic these splicing abnormalities. Taken together, DSMD-QKO SC mice can recapitulate most of the symptoms in DM1 and CDM (Table S3), providing another new model for DM1 study. S9c-e). To further confirm the differentiation defect was cell autonomous, we performed MuSC transplantation experiments. We generated QKO SC mice carrying RFP transgenes (termed RFP-QKO SC mice) and control SC mice carrying EGFP transgene (EGFP-WT SC mice) ( Figure   S10a and S10b) and then obtained MuSCs from the two strains, respectively (Figure S10c showed that muscle differentiation-related genes (such as muscle development and contraction) were up-regulated in mutant MuSCs compared to WT cells (Figure 5e and S11c-S11e). In contrast, the expression levels of stem cell markers were similar (Figure 5e and S11d-S11e).

Muscle stem cells are defective in DSM-TKO and DSMD-QKO SC mice
Nevertheless, qRT-PCR analyses indicated that the expression levels of the muscle-30 differentiation-related genes were comparable in differentiating cells from mutant and wild-type MuSCs ( Figure S11f). These results demonstrate that MuSCs of TKO and QKO SC mice are at a more committed state [69]. The premature expression of differentiation genes at stem cell stage suggests the partial loss of stemness and may account for the decreased differentiation potential of mutant MuSCs in vitro and in vivo.

Discussion
Mutations or expression changes occur in multiple genes in most of the diseases, especially in those multisystem syndromes. Generation of mouse models carrying multigene mutations to faithfully recapitulate the complex symptoms is the key to elucidate the mechanism of these syndromes. However, it is time consuming and labor intensive to do so using the conventional 5 methods, leading to the lack of proper animal models for many syndromes. In this proof-ofconcept study, we generated mouse models to mimic the reduced expression of multiple genes in human DM1 in one step by ICAHCI of haploid ESCs carrying three or four mutant genes. Dmwd is also affected (mimicked by DSMD-QKO SC mice), leading to more severe and early onset DM1 (Table S3). The longer the CTG repeat expansion is, the expression of more genes may be affected. Interestingly, MuSC stemness is affected in our DM1 mouse models, consistent with the observations in both DM1 and CDM patients [61, 63, 70] and recent observations in homozygous mice carrying Mbnl3 gene depletion [26]. Taken together, our study provides the 25 direct evidence to support the hypothesis that expansion of the CTG repeats affects the expression level of multiple genes, leading to complex pathogenic phenotypes in DM1 patients.
Meanwhile, TKO and QKO mouse models may provide a novel platform for drug screening to treat DM1.
Our method paves the road towards generating a series of mouse models carrying multiple gene 30 mutations in a short time to mimic different stages and severity of the complex syndromes. A 13 potential application of the approach is to produce mice carrying multiple modifications in candidate loci that have been identified in high-throughput studies or genetic screenings to mimic clinical manifestations of multigenic diseases. In summary, we have established novel DM1 mouse models by one-step injection of haploid cells carrying three or four mutant genes into oocytes, providing suitable models to investigate the molecular mechanisms underlying 5 complex manifestations and perform drug screening. Future analysis of the TKO and OKO mice will reveal more genes involved in the complex manifestations of DM1. Meanwhile, we hope that our haploid ESC-mediated SC technology will promote modeling of other complex diseases in mice.

3.
Udd wrote and revised the manuscript.

Competing Financial Interests:
The authors declare no competing financial interests.

Experimental mice
All animal procedures were carried out in accordance with the guidelines of the Shanghai Institute of Biochemistry and Cell Biology (SIBCB). All mice were housed in specific pathogenfree facilities of SIBCB. Oocytes for micromanipulation were obtained from female mice of 5 B6D2F1 (C57BL/6 × DBA/2) background. HSA LR mice (FVB/n background) and IL2r −/− mice (C57BL/6 background) were purchased from the Jackson Laboratory.

Derivation of gene-modified haploid cell lines
CRISPR-Cas9-mediated Gene manipulation was performed as previously described methods [52,

Grip strength, treadmill, rotarod test and righting assay
The forelimb grip strength of mice (4-month-old) was assessed using a grip strength meter.
Every mouse was tested for five times and the arithmetic average value of the records shall be as the measured value. For treadmill test, following familiarization test (6 m/min for 3 min), the mouse (4-month-old) was placed on a treadmill with a start speed of 10 m/min. Running speed 30 was increased by 2 m/min every two minutes until 30 m/min, and the distance was recorded when the mouse could no longer run. For rotarod performance test, the mouse (4-month-old) was placed on a rotating rod at uniform motion (10 rpm) for 5 min and then an accelerating rotarod, which started with 4 rpm and gradually increased to 40 rpm in 5 minutes. The mice were pretrained for 2 days for adaptation. The duration time on the rotarod before the mice fell off was recorded. As for righting assay, the mice at P5 were placed on its back to determine if the mice 5 could right themselves to normal posture with four paws on the ground.

Statistical analysis
Quantitative values are presented mean ± s.e.m, unless noted otherwise. Statistical differences between groups were determined by using GraphPad Prism 6 using unpaired two-tailed t-test. No statistical method was used to predetermine sample size. Investigators were not blinded to 20 outcome assessment.