Background: All types of facioscapulohumeral muscular dystrophy (FSHD) are caused by the aberrant activation of the somatically silent DUX4 gene, the expression of which initiates a cascade of cellular events ultimately leading to FSHD pathophysiology. Typically, progressive skeletal muscle weakness becomes noticeable in the second or third decade of life, yet there are many individuals who are genetically FSHD but develop symptoms much later in life or remain relatively asymptomatic throughout their lives. Conversely, FSHD may clinically present prior to 5–10 years of age, ultimately manifesting as a severe early-onset form of the disease. These phenotypic differences are thought to be due to the timing and levels of DUX4 misexpression. Methods: FSHD is a dominant gain-of-function disease that is amenable to modeling by DUX4 overexpression. We have recently created a line of conditional DUX4 transgenic mice, FLExDUX4, that develop a myopathy upon induction of human DUX4-fl expression in skeletal muscle. Here, we use the FLExDUX4 mouse crossed with the skeletal muscle-specific and tamoxifen-inducible line ACTA1-MerCreMer to generate a highly versatile bi-transgenic mouse model with chronic, low-level DUX4-fl expression and cumulative mild FSHD-like pathology that can be reproducibly induced to develop more severe pathology via tamoxifen induction of DUX4-fl in skeletal muscles. Results: We identified conditions to generate FSHD-like models exhibiting reproducibly mild, moderate, or severe DUX4-dependent pathophysiology and characterized progression of pathology. We assayed DUX4-fl mRNA and protein levels, fitness, strength, global gene expression, and histopathology, all of which are consistent with an FSHD-like myopathic phenotype. Importantly, we identified sex-specific and muscle-specific differences that should be considered when using these models for preclinical studies. (Continued on next page) * Correspondence: email@example.com Department of Pharmacology, School of Medicine, University of Nevada, Reno, Reno, NV 89557, USA Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Jones et al. Skeletal Muscle (2020) 10:8 Page 2 of 28 (Continued from previous page) Conclusions: The ACTA1-MCM;FLExDUX4 bi-transgenic mouse model has mild FSHD-like pathology and detectable muscle weakness. The onset and progression of more severe DUX4-dependent pathologies can be controlled via tamoxifen injection to increase the levels of mosaic DUX4-fl expression, providing consistent and readily screenable phenotypes for assessing therapies targeting DUX4-fl mRNA and/or protein and are useful to investigate certain conserved downstream FSHD-like pathophysiology. Overall, this model supports that DUX4 expression levels in skeletal muscle directly correlate with FSHD-like pathology by numerous metrics. Introduction disease presentation among healthy, FSHD1-affected, Facioscapulohumeral muscular dystrophy (FSHD) afflicts and FSHD1-asymptomatic subjects . Together, this females and males of all ages and an estimated 1:8,300– data supports that the level of somatic DUX4-fl expres- 22,000 people worldwide [1–4]. All forms of FSHD share sion, which is inducible and affected by the epigenetic a common pathogenic mechanism, increased somatic stability in the region, is the key determinant of disease expression of the DUX4 (Double homeobox 4) retrogene onset and severity. caused by the loss of stable epigenetic repression of the In addition to the complex genetic and epigenetic con- chromosome 4q35.2 D4Z4 macrosatellite array [5–12]. ditions that are required to develop clinical FSHD , Epigenetic dysregulation of the locus is caused by large the pathogenic mechanism is also unusual among neuro- deletions of the D4Z4 array on a single 4q allele, redu- muscular diseases. While FSHD is an autosomal domin- cing it to 1–10 repeat units (RU) (classified as FSHD1) ant gain-of-function disease, the pathogenic DUX4 gene [6, 13, 14], or by mutations in genes encoding repressive is typically expressed in only a small fraction (< 1%) of epigenetic regulators of the locus (classified as FSHD2) cells, ultimately leading to debilitating muscle pathology [15, 16]. In addition, all forms of FSHD require a per- over time [9, 18, 28]. This may account in part for the missive DUX4 polyadenylation signal (PAS) in cis distal more common adult onset of clinical symptoms in to a dysregulated chromosome 4q D4Z4 array . FSHD patients [29, 30]. In addition, it appears that Pathology ultimately results from the aberrant increase FSHD pathology is caused by sporadic bursts of in- in stable somatic expression of the dysregulated DUX4 creased DUX4-fl expression in differentiated myocytes, mRNA from the distal-most RU [5, 7, 9–12, 17–20]. which are epigenetically suppressed in healthy and The DUX4 gene encodes several mRNA isoforms gen- asymptomatic subjects [10, 21, 31, 32]. Since these cells erated by alternate 5′splice site usage in the first exon are syncytial, the detrimental effects of aberrant DUX4- ; however, only the DUX4-full length (DUX4-fl) FL expression may be found throughout an FSHD myo- mRNA is pathogenic when expressed in muscle [9, 17, fiber despite expression initially being restricted to a 21]. DUX4-fl encodes a paired homeobox domain tran- small percentage of myonuclei at any one time . Re- scription factor (DUX4-FL) normally expressed in gardless, DUX4-fl expression in FSHD myocytes and healthy human testis, pluripotent cells, and cleavage- skeletal muscle, even when bursting, is still extremely stage embryos, and the DUX4-mediated transcriptional rare, highly variable, and difficult to detect [9, 18, 21]. program is key for zygotic genome activation, all of Since DUX4-FL is a transcription factor not typically which supports an important role for DUX4-FL during expressed in healthy muscle, its aberrant expression al- early embryonic development [5, 9, 22–25]. Adult som- ters the mRNA profiles of numerous genes in myocytes atic cells from healthy individuals are typically devoid of [26, 33–35]. Many of these DUX4 target genes, including detectable DUX4-fl expression [9, 17, 18]. However, indi- germline-specific genes, cleavage-stage genes, and im- viduals meeting the genetic criteria for FSHD express mune system regulators, are not normally expressed in stable DUX4-fl mRNA and protein in their skeletal mus- healthy myogenic cells [24, 26, 33, 34]. In addition, cles, which aberrantly activate an embryonic gene ex- DUX4-FL expression ultimately initiates a cascade of nu- pression profile [9, 17, 18, 26, 27], ultimately leading to merous potentially detrimental events including the dis- FSHD pathophysiology. Interestingly, low somatic ex- ruption of proteostasis, nonsense-mediated decay, and pression of DUX4-fl mRNA per se is not necessarily mRNA splicing mechanisms [21, 36, 37], altered myo- pathogenic since expression can be detected in some genesis [38–41], and induction of apoptosis [7, 21, 42– rare cultures of myogenic cells and muscle biopsies from 44]. These DUX4-mediated changes, either alone or in healthy and asymptomatic FSHD subjects, albeit at levels combination, lead to progressive muscle cell death and significantly lower than in equivalent cells and tissues ultimately pathology [20, 45]. Thus, the DUX4-fl mRNA from FSHD-affected subjects [10, 18, 19]. Epigenetic and protein are prime targets for therapeutic interven- analysis of the pathogenic 4q D4Z4 RU shows that the tion, and animal models for FSHD should be based on stability of DUX4 epigenetic repression correlates with DUX4-fl expression in adult skeletal muscle. Jones et al. Skeletal Muscle (2020) 10:8 Page 3 of 28 We have previously reported the generation of the 5′ untranslated region (UTR), the endogenous PAS, and FLExDUX4 (FLExD) conditional DUX4-fl transgenic line the distal auxiliary elements (DAEs) that enhance DUX4 of mice, which contains a DUX4 transgene engineered mRNA cleavage and polyadenylation events [5, 9, 52– into the Rosa26 locus using the FLEx directional switch 54]. Here, we show that the FLExD mouse model, when system [46, 47] to bypass the embryonic lethality from crossed with the skeletal muscle-specific and tamoxifen leaky embryonic expression of this transgene . Upon (TMX)-inducible ACTA1-Mer-cre-Mer mouse , pro- Cre-mediated induction, the transgene recombines to duces a bi-transgenic model with chronic low levels of express DUX4-fl under control of the Rosa26 gene pro- mosaic DUX4-fl expression in skeletal muscle and repro- moter (Fig. 1a). Importantly, since many elements within ducibly recapitulating many aspects of FSHD patho- DUX4 are prime targets for sequence-based therapies physiology. Further, when treated with TMX, this model [49–52], the DUX4 transgene maintains the exon/intron can be induced to develop more severe pathology structure of the endogenous human gene, including the dependent upon the amount of TMX used, thus Fig. 1 Transgene map and FSHD-like model generation. a The synthesized FLExDUX4 (FLExD) transgene, flanked by heterologous lox sites (loxP and lox511), was inserted in the antisense orientation to the Rosa26 promoter and maintains the intron/exon structure and cis mRNA regulatory features, including the PAS and DAE, of the human chromosome 4q35 DUX4 gene. When crossed with ACTA1-MCM mice, the bi-transgenic offspring have the capacity for dosage-dependent, TMX-inducible unidirectional recombination of the transgene, resulting in DUX4 expression exclusively in skeletal muscle and transcribed from the Rosa26 promoter, processed, and terminated in exon 3 using the DUX4 PAS. b Genomic PCR indicating percent transgene recombination in different muscles from FLExD/+ mice and bi-transgenic ACTA1-MCM;FLExD mice with no TMX or 3 days after a single IP injection of 5 mg/kg TMX (low), or two IP injections of 10 mg/kg TMX (high). TA, tibialis anterior; Sol, soleus; GA, gastrocnemius; Qua, quadriceps; Diaph, diaphragm Jones et al. Skeletal Muscle (2020) 10:8 Page 4 of 28 providing several suitable models of FSHD severity for Tamoxifen (TMX) injections therapeutic interventions targeting DUX4-fl mRNA, pro- Tamoxifen free base (Sigma T5648) was dissolved in tein, and certain downstream pathways and pathology. 100% ethanol (200 mg/ml) at 55 °C and added to pre- warmed clean corn oil (ThermoFisher S25271) to make Methods 20 mg/ml frozen stocks. Stock TMX aliquots were Mouse husbandry warmed to 37 °C, diluted with pre-warmed sterile corn All mice for this study were bred and maintained in a oil either 10- or 20-fold just prior to use, and adminis- specific-pathogen-free barrier animal facility at the Uni- tered to mice by intraperitoneal (IP) injection at a final versity of Nevada, Reno (UNR), School of Medicine. Staff concentration of 5 or 10 mg/kg. For the moderate were required to wear disposable personal protective model, ACTA1-MCM; FLExD bi-transgenic animals clothing including hair bonnets, coveralls, gloves, and shoe were injected once with 5 mg/kg TMX. For the severe covers and use the high-level surface disinfectant Clidox-S model, ACTA1-MCM; FLExD bi-transgenic mice were (Pharmacal Research Laboratories, Inc., Naugatuck, CT). injected on two consecutive days with 10 mg/kg TMX. Mice were housed on autoclaved corncob bedding (Teklad 7092, Envigo RMS, Indianapolis, IN) in solid-bottom and X-Gal staining for evaluating TMX dosing individually ventilated polysulfone cages (Tecniplast USA, Male ACTA1-MCM mice were crossed with female NZG West Chester, PA) with the air handler set to 70 air Rosa26 mice to assess induction of Cre-mediated re- changes per hour. Reverse-osmosis purified water and ir- combination using our TMX dosage regimens. Control or radiated maintenance diet (Teklad 2919) was provided ad bi-transgenic mice (n = 3) were treated with TMX, as libitum. Structural enrichment of rodent cages included above. Whole gastrocnemius muscles were fixed in LacZ Alpha-twist paper nesting material (Shepherd Specialty fixative (0.2% glutaraldehyde, 5 mM EGTA, 100 mM MgCl Papers, Watertown, TN) and high-temperature polycar- in PBS pH 7.4) at 4 °C for 4 h, cryopreserved in cold 30% bonate shelters (Bio-Serv, Flemington, NJ). The animal sucrose/PBS at 4 °C overnight, then frozen in O.C.T. for room was maintained at 70 °F in a 12:12-h light:dark cycle sectioning. Cryosections (8 μm) were fixed with 4% parafor- and 10–12 air changes per hour with 40% relative humid- maldehyde (PFA)/PBS and then stained with X-Gal staining ity. Fresh cages were provided every 14 days using a solution (1× PBS with 2 mM MgCl , 5 mM potassium ferri- HEPA-filtered downdraft laminar airflow cage changing cyanide, 5 mM potassium ferrocyanide, 0.01% Nonidet P- station. The soiled-bedding sentinel mouse health surveil- 40, 0.1% sodium deoxycholate, 1 mg/mL X-Gal) for 50 min. lance system verified absence of all murine pathogens for Stained sections (n = 10/muscle analyzed) were post-fixed all periods during these studies, including Clostridium in buffered formalin at 4 °C for 5 min, rinsed three times piliforme, Mycoplasma pulmonis, CAR bacillus, Ectrome- with PBS, and then deionized water before mounting. Im- lia virus, EDIM virus, lymphocytic choriomeningitis virus, ages were captured using an Olympus CX41 microscope mouse adenoviruses 1 and 2, mouse cytomegalovirus, with PixeLINK camera under bright field. mouse hepatitis virus, murine norovirus, mouse parvovi- ruses, polyomavirus, reovirus type 3, Sendai virus, Thei- Transgene recombination assay ler’s murine encephalomyelitis virus, Encephalitozoon The direction of floxed DUX4 transgene in FLExD mouse cuniculi, Helicobacter spp., Pasteurella pneumotropica,the will be recombined to sense direction by Cre recombinase. pinworms Aspiculuris tetraptera and Syphacia spp., and The non-recombined transgene (NR) was detected by the fur mites Myobia spp., Myocoptes spp., and Radfordia PCR of 100-ng genomic DNA using primers; Forward: 5′- spp. All procedures were approved by the UNR Institu- CAATACCTTTCTGGGAGTTCTCTGCTGC-3′ and Re- tional Animal Care and Use Committee. The UNR animal verse: 5′- CTCGTGTAGACAGAGCCTAGACAATT care and use programs are fully accredited by AAALAC, TGTTG-3′, and recombined transgene (R) was detected International. using different reverse primer, 5′- AGGCTCGCAGGGCC TCGCTT-3′. PCR products were visualized by ethidium Transgenic mice bromide staining after agarose gel electrophoresis, and the The ACTA1-MCM Cre driver line refers to B6.Cg- intensities of PCR products were quantified using volume Tg(ACTA1-cre/Esr1*)2Kesr/J (JAX 025750)  and tool of BioRad Image Lab software ver.4, global subtrac- NZG Rosa26 refers to FVB.Cg-Gt(ROSA)26Sor< tion method. Recombination rate was calculated as follows tm1(CAG-lacZ,-EGFP)Glh>/J (JAX 012429) , both %=[R/(NR + R)] × 100. purchased from Jackson Labs (Bangor, Maine). FLEx- DUX4 (FLExD) mice were generated and characterized Gene expression analysis by RT-PCR by the PL Jones lab  and are available from Jackson Specific mRNA expression was determined for each se- Labs (JAX 028710). Mice were genotyped as described verity model using 3–4 mice per model. Total RNA was . extracted from dissected mouse muscles homogenized Jones et al. Skeletal Muscle (2020) 10:8 Page 5 of 28 in 10 volumes of TRIzol (ThermoFisher) using the Tis- determine the number of pixels stained red and the total sueLyser LT (Qiagen), as per manufacturer’s instructions number of pixels stained. Muscles from 3–4 mice for followed by on-column DNaseI treatment and clean-up each treatment were analyzed. Significance was calcu- using the RNeasy mini kit (Qiagen). Quantitative DUX4- lated by one-way ANOVA using Prism 7 (Graphpad). fl mRNA expression was analyzed using nested qRT- PCR, as described [48, 57]. Expression of DUX4 target Immunofluorescence (IF) genes was analyzed by qPCR using 5 to 10 ng cDNA, as The 10-μm cryosections of muscle tissues were fixed described [33, 48]. All oligonucleotide primer sequences with 4% PFA/PBS on ice for 10 min, permeabilized with for DUX4-fl, downstream targets, and 18S rRNA are pre- 0.25% TritonX-100/PBS for 10 min, incubated with viously reported . Expression of Myostatin mRNA blocking solution (5% normal goat serum, 2% BSA, was analyzed by qRT-PCR using primers 5′-AGTGGA 0.01% TritonX-100/PBS) for 30 min, and then incubated TCTAAATGAGGGCAGT-3′ (forward) and 5′- with primary antibody at 4 °C overnight. The following GTTTCCAGGCGCAGCTTAC-3′ (reverse). Significance day, sections were incubated with secondary antibody at was calculated using the two-tailed t test. room temperature for 40 min, then mounted in ProLong Gold with DAPI for nuclear staining. DUX4 was de- Histology tected using E5-5 antibody (Abcam ab124699) diluted 1: Freshly dissected muscles were kept moist, coated with 200. Embryonic fast myosin heavy chain was detected O.C.T. compound, snap-frozen in liquid nitrogen-cooled using the F1.652 monoclonal antibody, developed by the isopentane, and stored at − 80 °C until sectioning. Cryo- Baxter Lab for Stem Cell Biology, Stanford University, sections (12 μm) were mounted on slides and air-dried and obtained from the Developmental Studies Hybrid- before staining or storage. Sections were used for oma Bank, created by the NICHD of the NIH and main- hematoxylin and eosin (H&E) staining  or picrosirius tained at The University of Iowa, Department of Biology, red staining. All histological analyses were done with the Iowa City, IA 52242. Dystrophin was detected using an identity of the sample blinded to the person performing anti-dystrophin rabbit polyclonal antibody (Abcam, the assessment. ab15277). Immunofluorescent images were captured using the Leica DMi8, DFC365 FX camera, and LAS X Centralized nuclei software (Leica Microsystems Inc.). Three non- H&E staining was performed as described . A series consecutive sections from each muscle of 3–4 mice per of micrographs from each muscle section were captured group were analyzed. using a × 20 objective on a Leica DM2000 and reconsti- tuted to form a whole muscle cross section using LAS Apoptosis assay 4.12 software (Leica Microsystems, Inc.). Muscle sections Apoptotic events were analyzed by TUNEL (terminal taken from similar points in the soleus or tibialis anter- deoxynucleotide transferase dUTP nick end labeling) ior muscles were used across the models. The number staining using the In Situ Cell Death Fluorescein Kit of muscle fibers with or without internally located nuclei (Roche/SIGMA 11684795910). The 10-μm cryosections were counted using Image J software. of TA muscles were mounted on slides and fixed with 4% paraformaldehyde for 20 min. Staining was per- Picrosirius red (SR) staining formed as per manufacturer’s instructions. Three non- SR staining was performed as described . Cryosec- consecutive sections from each muscle of 3–4 mice per tions (12 μm) cut mid-belly of the tibialis anterior group were analyzed. muscle were fixed with 4% PFA/PBS, pH 7.4 for 10 min, rinsed with dH O, and dehydrated with a series of 1 min Treadmill exhaustion test ethanol washes (70%, 95%, 100%), and air-dried. Sections All treadmill tests were performed with an Exer3/6 were then stained for 1 h in SR solution (0.1% direct red treadmill and shock detection system (Columbus Instru- 80, 1.3% saturated picric acid), and washed three times ments) in the mouse mode with electric shocking grid. with dH O. Stained sections were dehydrated with a All mice were acclimated to the apparatus before run- series of 1 min ethanol washes (70%, 95%, and 100%), ning by placing them on an unmoving treadmill for 5 cleared with xylene for 5 min, and mounted. A series of min, then at a speed of 5 m/min for 5 min at 0° incline micrographs from each muscle section were captured with electric shock grid on. Mice were rested for 2 days using a × 10 objective on a Leica DM2000 and reconsti- before the first test. The exhaustion test was established tuted to form a whole muscle cross section using LAS for these FSHD-like model mice after several modifica- 4.12 software (Leica Microsystems, Inc.). Whole cross tions of the Treat-NMD protocol DMD_M.2.1.003. The section images (n = 2 per muscle) were divided into 3–5 test was performed using a 7° incline and an initial speed sections and processed with MATLAB (Mathworks) to of 5 m/min with speed increasing by 0.5 m every minute. Jones et al. Skeletal Muscle (2020) 10:8 Page 6 of 28 Mice were run until they were unable to maintain a Genome annotation, read mapping, and gene expression speed to remain off the shock grid for more than 5 sec estimation (time of fatigue) or a maximum of 20 min (approximate Human (hg19) and mouse (mm10) genome annotations maximum speed is 15 m/min and maximum distance is were created by merging the UCSC knownGene , 200 m). This testing was performed at least three times Ensembl 71 , and MISO v2.0  annotations. Se- per mouse with at least 2 days of rest in between tests. quence reads were mapped to these annotations as pre- There were 3–4 mice analyzed per group. Significance viously described . In brief, RSEM v1.2.4  was was calculated by two-way ANOVA using Prism 7 modified to call Bowtie  with the option “-v 2” and (Graphpad). then used to map all reads to the merged genome anno- tation. Remaining unaligned reads were then mapped to the genome and a database of potential splice junctions Ex vivo muscle contractile properties with TopHat v2.0.8b . All gene expression estimates EDL muscles (n indicated in figure legends) were excised were normalized using the trimmed mean of M values from deeply anesthetized mice with 3% isoflurane, hung (TMM) method . from a computer-controlled servomotor (300B, Aurora Differentially expressed genes (Table S1 and GEO acces- Scientific, Inc), and mounted in a heated (30 °C) oxygen- sion number GSE122562) in DUX4-induced versus con- ated tissue bath containing a physiologic salt solution as trol experiments were defined at a threshold of 1.4-fold described [59, 60]. Experiments were performed, and and 2-fold change and Bayes factor ≥ 10 (computed with data were analyzed using DMA software (Aurora Scien- Wagenmakers’s framework ), as observed in all pos- tific, Inc) and Prism 7 (Graphpad), as previously de- sible pair-wise comparisons of experimental replicates. A scribed (Treat-NMD SOP, DMD_M.1.2.002, and ). gene that was differentially expressed was a gene that was Significance for twitch and tetanus was calculated by significantly upregulated or downregulated in all 3 repli- one-way ANOVA, and for the force frequency analysis, cate comparisons. Genes listed in Table S1 were at 1.4- it was calculated by two-way ANOVA using Prism 7 fold. Gene ontology (GO) terms that were enriched (GraphPad). among genes that exhibited increased or decreased ex- pression in DUX4-induced versus control samples were identified with the GOseq method , with a false- RNA-seq discovery rate threshold of 0.01. GO superterms (Table RNA-seq was performed by Genewiz LLC (South Plain- S4) are all terms that are descendent of GO terms con- field, NJ). Total RNA (5 μg) was isolated from gastrocne- taining the following key terms, respectively: mius muscles of 13-week-old mice, 3 mice per group Cell cycle: cell cycle, mitosis, mitotic, chromosome seg- (ACTA1-MCM, FLExD, ACTA1-MCM;FLExD mild/no regation, cell division, nuclear division, proliferation, TMX, ACTA1-MCM;FLExD moderate TMX, and chromosome condensation, kinetochore, spindle, cyclin ACTA1-MCM;FLExD severe TMX), as described above. Apoptosis: apoptosis, cell death, apoptotic, cell killing The mRNAs were purified using poly(A) selection and Muscle: muscle, sarco, myofibril then fragmented. First-strand cDNA synthesis used ran- Immune: immune, chemokine, interferon, wound, dom priming followed by second-strand synthesis. The virus, cytokine, cytokine-, leukocyte, interleukin, Toll- resulting double-strand cDNA was end repaired, phos- like, toll-like, bacteri, inflamma, defense, immunological, phorylated, and A-tailed. Adapters were ligated and PCR immunoglobulin receptor, MHC, immunoglobulin, viral amplification was performed. The library was sequenced Differentially expressed genes from the comparison of using the Illumina HiSeq2500platform in a 1 × 50 base 9 control and 9 FSHD1 muscle biopsies  were de- pair single-read configuration in Rapid Run mode, with a fined as genes that were consistently differentially total of at least 120 million reads per lane. Sequence reads expressed at a threshold of Bayes factor ≥ 10 (computed were trimmed to remove adapter sequences and poor with Wagenmakers’s framework ) in more than 50 quality nucleotides (error rate < 0.05) at the ends. Se- out of 81 possible pair-wise comparisons between con- quence reads shorter than 50 nucleotides were discarded. trol and FSHD1 muscles. The intersection of these genes with orthologous differentially expressed mouse genes from DUX4-induced vs control experiments are summa- Differential RNA-seq expression and gene ontology (GO) rized in Table S3. analysis Data sources Data analysis and visualization RNA-seq reads for C2C12 expressing DUX4-fl or con- All data analysis was performed in the R programming trol were downloaded from the Gene Expression Omni- environment and relied on Bioconductor , dplyr bus (accession number GSE87282) . , and ggplot2 . Venn diagrams were plotted using Jones et al. Skeletal Muscle (2020) 10:8 Page 7 of 28 the venneuler package. For significance, fold change prolapse, and soft stools as commonplace in the hetero- thresholds of 1.4-fold (i.e., log2 fold change of 0.5) or 2- zygous FLExD/+ transgenic mouse and more even more fold (i.e., log2 fold-change of 1), depending on the Venn pronounced in the homozygous FLExD/FLExD line (Fig- diagram. ure S1 and Figs. 2 and S4 in ref ), indicating they were linked to the low leaky levels of DUX4-fl expres- Alternative splicing analysis sion that were higher in the transgene homozygote. In The MISO computational method  was used to addition, these gastrointestinal (GI) phenotypes sug- characterize skipped exon (SE) and retained intron (RI) gested there may be some gut inflammation exacerbated alternative splicing events for 12 RNA-seq samples from by low levels of leaky DUX4 expression. This GI pheno- the control, FLExD, moderate, and severe models (Table type may not be acceptable to veterinary staffs at some 3 and Table S6). MISO (version 0.5.3) was executed with institutions and prevent import and use. However, in mapping results produced with TopHat, using the fol- 2017, the PL Jones lab moved to the University of Nev- lowing configuration: --read-len:51, min_event_reads = ada, Reno School of Medicine (UNR Med), and relocated 20, burn_in = 500, lag = 10, num_iters = 5000, and the FLExD mouse colony used in this current study. In num_chains = 6. order to enter the UNR Med barrier facility, mice from UMMS were analyzed for undesirable bacteria and vi- Results ruses. In addition to being found positive for murine The FLExDUX4 (FLExD) line of transgenic mice was de- norovirus (MNV), analysis of the gut flora from the veloped and first characterized by the PL Jones lab while UMMS housed mice revealed the presence of Pasteur- at the University of Massachusetts Medical School ella pneumotropica and Helicobacter mastomyrinus, (UMMS), was deposited into Jackson Labs for distribu- which necessitated re-derivation of the line to allow im- tion (#028710), and is described elsewhere . The ori- port. Helicobacter species (ssp) have been known to ginal description of the model reported alopecia, rectal cause rectal prolapse and diarrhea in some immune Fig. 2 All bi-transgenic ACTA1-MCM;FLExD mouse models show mosaic patterns of nuclear DUX4-FL expression. Tibialis anterior muscles were analyzed by IF for DUX4-FL protein expression over 28 days (moderate model, panels c–j) or 9 days (severe model, panels o–t). Muscles from the mild model, bi-transgenic ACTA1-MCM;FLExD without TMX (panels m and n), showed low steady levels of DUX4-FL immunostaining. Muscles from ACTA1-MCM mice treated with TMX at MD14 (panels a and b) or SD9 (k and l) showed no DUX4-FL signal and served as negative controls. Red, DUX4 IF; Blue, DAPI staining; Green, WGA staining. Scale bar = 50 μm Jones et al. Skeletal Muscle (2020) 10:8 Page 8 of 28 deficient or otherwise susceptible animals . After re- Expression of the nLacZ reporter gene was visualized by derivation, the UNR Med housed mice tested negative X-Gal staining  in the gastrocnemius muscles iso- for MNV, Helicobacter ssp, and Pasteurella pneumotro- lated 2 weeks after TMX injection. The results indicated pica biotypes Jawetz and Heyl. Interestingly, the prior that both dosing regimens produced mosaic patterns of alopecia phenotype was much less pronounced in the Cre-mediated recombination in skeletal muscle, with the UNR Med housed mice and the soft stools and rectal high TMX dose producing ~ 1.5× more X-Gal stained prolapse have not been seen in the UNR Med FLExD nuclei than the low-dose (Figures S2 and S3). There was mouse colony over 3 years (Figure S1), supporting these no significant difference in transgene recombination be- were a result of different gut microbiomes. This is a tween male and female mice (Figure S4). Surprisingly, in NZG good illustration of potential overlooked sources of vari- the absence of TMX the ACTA1-MCM;R26 mice ability between labs when using the same transgenic also showed transgene recombination, although only in mouse models of human disease [75, 76]. Investigators a very small fraction of skeletal muscle nuclei (Figure should be aware of these adverse effects, which may also S2C). Interestingly, this low-level recombination and impact experimental outcomes, when using the FLExD nLacZ expression was not uniform across all nuclei; ra- mouse model and similarly that they can potentially be ther, it also showed a mosaic pattern of recombination overcome by mouse facility conditions. (~ one third the number of recombined nuclei induced by the low TMX dose, Figure S3C compared with S3D). Generation of mice with mosaic DUX4-fl expression in Since there was no indication of recombination in the NZG skeletal muscles R26 single transgenic animals treated with TMX FSHD is caused by mosaic expression of DUX4-fl mRNA (Figure S2B), we interpret this result as indicating a low and its encoded protein from the normally silent DUX4 level of MerCreMer protein leaking into the nucleus in gene in a small fraction of differentiated adult skeletal the absence of TMX, either at a level that only occasion- myocytes [9, 18, 21]. Previously we generated a condi- ally leads to recombination or the protein is only sporad- tional DUX4-fl expressing transgenic mouse, FLExD, and ically leaky into a few nuclei. Regardless, this suggested showed that high expression of DUX4-fl in skeletal that bi-transgenic animals generated with the ACTA1- muscle can produce a very severe myopathy with some MCM line may be useful for generating very low mosaic FSHD-like pathology . However, high levels of expression of a transgene in the absence of TMX induc- DUX4-fl expression in many skeletal myonuclei does not tion. Overall, we concluded that the ACTA1-MCM line recapitulate the mosaic rare myonuclear expression of of mice was suitable for generating a range of mosaic DUX4-fl found in FSHD subjects and preclinical testing transgene expression models. for different candidate FSHD therapeutics targeting Generating FSHD-like model mice with differing levels DUX4-fl mRNA, and protein expression will likely re- of pathophysiology required adjusting the level of mosai- quire different criteria, such as lesser degrees and slower cism with respect to DUX4-fl expression. Therefore, progression of pathophysiology, dependent upon the ap- ACTA1-MCM mice were crossed with FLExD mice  proach. To address this issue, we generated and charac- (Fig. 1a), and 13–14-week-old ACTA1-MCM;FLExD bi- terized a highly reproducible series of phenotypic FSHD- transgenic mice were treated with the above TMX dos- like transgenic mouse models varying in severity and ing regimens to induce differing levels of mosaic trans- pathogenic progression based on differing levels of mo- gene recombination. Genomic DNA was isolated 3 days saic expression of the pathogenic DUX4-fl mRNA iso- post-injection (DPI) of TMX and assayed for transgene form of human DUX4 in adult murine skeletal muscle. recombination by genomic PCR (Fig. 1b). The FLExD We identified the tamoxifen (TMX) inducible and skel- hemizygous mice showed no transgene recombination in etal muscle-specific Cre expressing transgenic mice, the absence of Cre, and the bi-transgenic animals ACTA1-MerCreMer (or ACTA1-MCM) as a strong showed variable low levels (2–10%) of transgene recom- candidate for the generation of the desired phenotypes. bination in skeletal muscles, but not in the heart or liver, To test if these could be used to generate mosaic expres- in the absence of TMX, due to the abovementioned sion in skeletal muscles and to optimize TMX dosing, sporadic leaky nuclear Cre activity in ACTA1-MCM NZG the ACTA1-MCM mice were crossed with R26 Cre mice. When injected with TMX to induce Cre nuclear reporter mice  that produce readily detectable nu- activity and stimulate transgene recombination in skel- clear ß-galactosidase (nLacZ) expression in all cells etal muscles, the bi-transgenic mice showed increased where Cre is functional in the nucleus (Figure S2). Two recombination (8–30%) in response to the low TMX dosing regimens of TMX were tested in the ACTA1- dose and an even higher level of recombination (38– NZG MCM; R26 bi-transgenic offspring, a single low-dose 55%) in response to the high TMX dose. Surprisingly, in (5 mg/kg) intraperitoneal (IP) injection and two IP injec- all three bi-transgenic models (no, low, and high TMX), tions on consecutive days of a higher dose (10 mg/kg). there were muscle-specific differences in the transgene Jones et al. Skeletal Muscle (2020) 10:8 Page 9 of 28 recombination rate; however, these differences were con- referred to as the moderate FSHD-like model (Table 1, sistent between the TMX treated lines (Fig. 1b). For ex- discussed in detail below), were assayed over 28 DPI. ample, quadriceps muscles showed the lowest DUX4-FL expression appeared by 3 DPI (moderate day 3, recombination in both TMX-induced models, followed by or MD3), peaking at MD14, and then was greatly reduced the tibialis anterior (TA) and gastrocnemius muscles with by MD28, likely due to the death of DUX4-positive cells intermediate levels, while the diaphragm and soleus showed (Fig. 2a–j). Mice injected with the high-dose TMX regi- the highest recombination levels. As expected when using men were so severely affected by 9 DPI that they had to the ACTA1-MCM driver line, the heart and liver showed be sacrificed and could not be assessed further. These no transgene recombination in any of the bi-transgenic ani- mice, which will be referred to as the severe FSHD-like mals. Since this assay measures transgene recombination of model, similarly showed DUX4-FL expression by 3 days a single copy transgene per nucleus, the results showing less after the first injection (severe day 3, or SD3) and peak than 100% recombination represent mosaic recombination DUX4-FL expression at SD6. DUX4-FL expressing myo- which should translate into the desired mosaic DUX4-fl nuclei had decreased in number by SD9, likely due to transgene expression. Importantly, variable muscle-specific massive cell death. Overall, the mild FSHD-like model ex- recombination efficiencies were not significantly different hibited very low numbers of DUX4-FL-positive myonu- between male and female mice (Fig. S4). We conclude that clei, with the moderate model exhibiting a significant we have identified three conditions that reproducibly pro- increase (> 5-fold) in DUX4-FL-positive myonuclei, and duce differing levels of mosaic transgene recombination, the severe model had the highest numbers of DUX4-FL- which we will refer to as mild, moderate, and severe (Table positive myonuclei, with a > 10-fold increase compared 1), based on the subsequent characterizations described with the mild model (Figure S5). DUX4-FL protein was below. Importantly for future studies, we show that differ- not detectable in the heart or skeletal muscles from the ent skeletal muscles show different levels of recombination non-recombined FLExD mice or ACTA1-MCM controls. in response to these TMX treatments. Thus, each model consistently displays mosaic nuclear To assess if the variable rates of transgene recombin- DUX4-FL protein expression patterns (Fig. 2) and pro- ation in each model translated similarly to variable levels vides an indication of the relative abundance of DUX4- of protein expression at the single nucleus level, muscle FL-expressing nuclei in bi-transgenic animals and in re- tissues were analyzed by immunofluorescence (IF) using sponse to two different TMX treatments. an anti-DUX4-FL antibody. Mosaic patterns of nuclear In order to quantitate the changes in gene expres- DUX4-FL expression were readily detected in cross sec- sion for each severity model, qRT-PCR was used to tions of TA muscle from all three models (Fig. 2). The bi- measure overall DUX4-fl mRNA levels (Fig. 3a). How- transgenic ACTA1-MCM;FLExD mild model showed few ever, we have previously shown that this assay is a DUX4-FL-positive nuclei in the absence of TMX (Fig. 2m poor measure of DUX4-fl transgene expression using and n), consistent with a very low level of transgene re- FLExD mouse models , likely due to DUX4- combination (Fig. 1b). However, since DUX4-FL is cyto- induced cell death, and DUX4-fl mRNA is even diffi- toxic and it is not known how long DUX4-FL expressing cult to detect in muscle biopsies from FSHD-affected cells may remain in the muscles, an IF time course study subjects . Since DUX4-FL functions as a tran- was performed for the TMX-injected mice and the scriptional activator in both human and mouse cells DUX4-FL protein-positive myonuclei were quantified [33, 78], expression of DUX4-FL direct target genes (Figs. 2 and S5). The low-dose mice, which appeared to be has proven to be a more accurate indicator of DUX4- moderately affected phenotypically over time and will be FL expression levels in both species [26, 34, 35, 48]. Therefore, in addition to DUX4-fl mRNA, the mRNA levels of two mouse homologs of DUX4-FL direct tar- Table 1 FSHD-like mouse severity models get genes, Wfdc3 and Trim36 [48, 78], were also Genotype TMX Transgene DUX4-fl Phenotype assayed (Fig. 3b and c). Detectable DUX4-fl mRNA recombination expression levels were extremely low in gastrocnemius muscles ACTA1-MCM/ 5 mg/kg × NA No Control + 1IP moderate from all models (Fig. 3a), consistent with previous studies . Interestingly, there were no significant ACTA1-MCM/ 10 mg/kg NA No Control + ×2IP Severe changes detected in DUX4-fl mRNA levels between ACTA1-MCM; None 2-10% Yes Mild the mild, moderate, and severe models 9 days after FLExD/+ TMX treatments, a time point with prominent differ- ACTA1-MCM; 5 mg/kg × 7–28% Yes Moderate ences in DUX4-FL protein expression (Figs. 2 and FLExD/+ 1IP S5). In contrast, both DUX4-FL target genes assayed ACTA1-MCM; 10 mg/kg 38–55% Yes Severe showed significant induction in all models compared FLExD/+ ×2IP with the FLExD/+ mice, indicating the presence of Jones et al. Skeletal Muscle (2020) 10:8 Page 10 of 28 Fig. 3 DUX4-FL target genes, but not DUX4-fl mRNA, show TMX dose-dependent increased expression. Gastrocnemius muscles from 13–14- week-old female mice isolated 9 days after TMX administration were assayed for gene expression levels by qRT-PCR. a Levels of DUX4-fl mRNA are not significantly different between the FLExD mice and the three bi-transgenic models. However, mRNA expression of b Wdfc3 and c Trim36 are increased significantly in all bi-transgenic models compared with FLExD. Both Wdfc3 and Trim36 are significantly increased by moderate and severe TMX induction compared with no TMX. Trim36 mRNA levels are significantly increased in the severe compared to moderate models. All expression is normalized to RpL37 expression. MT, equivalent level of cDNA from 17Abic FSHD1 myotubes . Data are mean ± S.D. with significance calculated using Welch’s t test, *p < 0.05, **p < 0.01 DUX4-FL protein. Wdfc3 and Trim36 mRNA levels treatment, resemble FSHD patient muscle biopsies, are each significantly increased in muscles from the which show increased expression of known DUX4-FL moderate and severe models compared with the mild target genes compared with control biopsies . model, and Trim36 mRNA levels are significantly in- creased in the severe model compared to the moder- Characterization of muscle function for three levels of ate model (Fig. 3b and c). We conclude that the bi- phenotypic severity transgenic mice, which show increased DUX4-FL ex- To determine if increases in DUX4-FL protein and tar- pression correlating with the degree of TMX get gene expression levels in skeletal muscles correlate Jones et al. Skeletal Muscle (2020) 10:8 Page 11 of 28 with altered muscle function and strength, we performed to fatigue, with a maximum assay time of 20 min. We treadmill exhaustion assays and ex vivo muscle physi- found these particular conditions (see the “Methods” ology studies. Adult ACTA1-MCM control and bi- section) provided highly consistent results and a readily transgenic mice (mild model) showed size differences assayable window for these three mouse models. Both between males, with bi-transgenic males being signifi- male and female ACTA1-MCM control mice, injected cantly smaller than controls (~ 26 g and ~ 23 g, respect- with the appropriate TMX dose for the group being ively; Figure S6B and D), while females for both assayed, showed steady levels of near-maximum tread- genotypes showed no significant size differences prior to mill running fitness over the course of the assays (Fig. 4, TMX injection (~ 22 g and ~ 21 g, respectively; Figure blue lines). Similarly, treadmill fitness for male and fe- S6A and C). Therefore, male and female mice were ana- male mild model mice was not significantly different lyzed separately to assess potential sex differences in from the ACTA1-MCM controls (Fig. 4, green lines). presentation of the phenotypes. The treadmill analysis However, moderate and severe model mice were signifi- protocol consisted of running the mice on a slightly in- cantly less fit compared with controls (Fig. 4, red lines). clined treadmill, slowly increasing the speed until the Interestingly, both the moderate and severe models mice could not run any longer, and measuring the time showed sex-specific treadmill fitness profiles in which Fig. 4 The moderate and severe FSHD-like mouse models show significant decline in treadmill performance. Mild, moderate, and severe FSHD- like mice were assessed for treadmill performance, as described in the methods. a The moderately affected female mice (red line) were assayed prior to TMX injection (D0) and 3, 6, 10, 16, 23, and 29 days post-injection (DPI) and compared with age-matched female bi-transgenic mice (green line) and female ACTA1-MCM mice similarly injected with TMX (blue line). b The moderately affected male mice (red line) were assayed prior to TMX injection (D0) and at 2, 6, 10, 14, 17, 21, 24, and 28 DPI and compared with aged-matched male bi-transgenic mice (green line) and male ACTA1-MCM mice similarly injected with TMX (blue line). c Severely affected female mice (red line) were assayed prior to TMX injection (D0) and 2, 5, and 7 DPI (# D8, mice were too affected to be assayed) and compared with female bi-transgenic mice (green line) and ACTA1-MCM mice similarly injected with TMX (blue line). d Severely affected male mice (red line) were assayed prior to TMX injection (D0) and 2, 5, 7, and 8 DPI and compared with age-matched male bi-transgenic mice (green line) and ACTA1-MCM mice similarly injected with TMX (blue line). Data are mean ± S.D. with significance, *p < 0.05, **p < 0.01, ***p < .001, calculated between bi-transgenic +TMX and bi-transgenic no TMX (green) or ACTA1-MCM +TMX (blue) Jones et al. Skeletal Muscle (2020) 10:8 Page 12 of 28 the female mice (Fig. 4a, c) were more severely affected both male and female mild models were less responsive to than the male mice (Fig. 4b, d). Moderate female mice mid to high stimulation frequencies (65–180 Hz) com- showed significant declines in fitness 10 days after TMX pared with controls (Figs. 5cand S8F). We conclude that induction (MD10), dropping from near-maximum run- chronic, low, mosaic DUX4-FL expression in even a few ning times to less than 3 min. This decline remained at myofibers reproducibly produced a measurable decrease MD16 before beginning to recover to ~ 4 min running at in isometric contractile strength of muscle that does not MD23 and ~ 10 min by MD29. In comparison, all male appear to affect treadmill fitness. Thus, this data suggests moderate model mice showed a significant decline start- that these mild FSHD-like model mice may present a ing at MD6 (running ~ 15 min) which progressed more model of pre-symptomatic FSHD or an early symptomatic slowly than in females, running for ~ 10 min at MD10 mild FSHD. and ~ 5 min at MD14 and MD17 before recovering to ~ We similarly assayed the moderate and severe FSHD- 15 min by MD28. Overall, although all moderately af- like model mice for ex vivo muscle function, both of fected mice showed significant declines in treadmill fit- which showed significant differences in treadmill fitness ness over the course of 4 weeks, female mice were more compared with the ACTA1-MCM controls and the mild severely affected and recovered more slowly than male model mice upon TMX injection (Fig. 4). For the moder- mice. For the severe model mice, female mice declined ate model analysis, female mice were run on the treadmill more rapidly than males, although both showed a signifi- until exhaustion, as above, prior to TMX injection, and cant decline in running time before fatigue, decreasing then run twice per week until sacrificed and assayed at by over 10 min compared with controls by SD5, a point MD14, the peak of DUX4-FL protein expression (Fig. 2g at which the moderate model mice were still unaffected. and h) and impaired treadmill running (Fig. 4a). The iso- Female mice could not even begin the assay at SD7, lated EDL muscles from the moderate model mice while their male counterparts were slightly less fatigued were significantly weaker and stiffer compared with at SD5 and were able to run a few seconds at SD7. the ACTA1-MCM controls (Figs. 5 and S6); however, These male mice reached the point of being unable to these muscles only showed small, but significant, de- safely start the assay at SD8. All of these assay points in- creases in specific force measurements and stiffness dicated a much more severe phenotype than at any point when compared with the mild model. The DUX4- during the assessment of the moderate model, and these dependent impairment of muscle function was much severely affected mice had to be humanely sacrificed by more striking in the severe model mice. Male and fe- SD10, with general movement in the cage greatly im- male mice were run on the treadmill until exhaustion, paired and no signs of recovery. These sex-specific dif- as before, prior to TMX injection and again at SD2, ferences make it vitally important to analyze and SD5, and SD7 (female) or SD8 (male), then sacrificed compare mice of the same sex when performing fitness at SD10, the point of maximal fitness decline (Fig. 4c assays using both of these FSHD-like models. and d). The EDL muscles from both female (Fig. 5) Treadmill assays indicated that muscle use and/or func- and male (Figure S8) severe model mice showed ~ 3- tion was impaired by induction of DUX4-fl expression in fold decreases in all specific force measurements, and the moderate and severe mouse models. Ex vivo muscle muscles were significantly stiffer after eccentric contractile studies using the isolated extensor digitorum stretches when compared with ACTA1-MCM controls longus (EDL) were then performed on these mice to assess and mild model mice. In addition, the severe model changes in muscle strength [59, 60, 79] (Figs. 5, S7,and mice were virtually non-responsive to low-frequency S8). First, EDL muscles isolated from the mild model were stimuli (< 30 Hz) and were severely impaired across compared with age-matched ACTA1-MCM mice (control) the force frequency assay range (Figs. 5cand S8F). treated with the appropriate dose of TMX for the respect- When compared with the moderate model mice, mus- ive group isolated at TMX D14 or TMX D10 from the cles from the severe model were again significantly first injection. Interestingly, despite having similar tread- weaker. Overall, we conclude that murine skeletal mill endurance profiles as controls (Fig. 4), the EDL mus- musclescan tolerateaverylow levelofDUX4-FL ex- cles from female mild model mice consistently showed pression without significant impairment of function; significantly lower maximum absolute force (Figure S7) however, increases in DUX4-FL expression eventually and specific force (maximum force normalized by cross lead to muscle weakness and loss of function in a sectional area, CSA) compared with ACTA1-MCM con- dose-dependent manner. trols (Fig. 5). Male mild model mice were less affected and only showed a significant strength difference from con- Global mRNA expression analysis for moderate and trols with respect to maximum twitch force and max- severe FSHD-like mice imum force frequency, and no significant change in To begin to identify the mechanisms and pathways dis- specific force measurements (Figure S8A-F). However, rupted by DUX4-fl expression in murine skeletal muscle, Jones et al. Skeletal Muscle (2020) 10:8 Page 13 of 28 Fig. 5 Ex vivo muscle function analysis shows DUX4-induced muscle weakness. EDL muscles from female ACTA1-MCM/+ with TMX (MCM, blue, n = 9), mild model (MCM;FLExD Mild, green, n = 9), moderate model (MCM;FLExD Mod, yellow, n = 6), and severe model (MCM;FLExD Severe, red, n = 3) were assayed for maximum twitch, maximum tetanus, and force frequency (Figure S7, not normalize) and then normalized, here, to muscle cross sectional area to provide a specific twitch, b specific tetanus, and c specific force frequency at MD14 and SD10. d Muscle stiffness calculated from eccentric contractions analysis showed all three model mice have stiffer muscle than control. Data are mean ± S.D. with significance *p < 0.05, **p < 0.01, ***p < .001. For panel c, blue asterisks indicate significance compared with MCM +TMX control mice, green asterisks indicate significance compared with mild bi-transgenic mice, and yellow asterisks indicate significance compared with moderate model mice RNA-seq analysis was performed on gastrocnemius mus- identified (Table S2). The ACTA1-MCM/+ and FLExD/ cles isolated from control (ACTA1-MCM/+), FLExD/+ + single transgenic mice only showed 3 genes differen- hemizygous, mild, moderate (MD9), and severe (SD9) tially expressed (2 upregulated and 1 downregulated) be- bi-transgenic FSHD-like model mice (13 weeks old, n = tween the mice (Fig. 6a and Table S2), which is 3 each), and analyzed for global mRNA expression levels consistent with our previous qRT-PCR studies showing (Table S1). Genes with significant differential expression no significant differences in expression of DUX4-fl or (> 1.4-fold) from the ACTA1-MCM controls were several prominent DUX4-FL targets between these mice Jones et al. Skeletal Muscle (2020) 10:8 Page 14 of 28 . In contrast, the transcriptomes for the mild, mod- retroelements that are lacking in or poorly conserved erate, and severe models were significantly different with mice . from the ACTA1-MCM controls when analyzing genes Recently, an analysis of human FSHD transcriptome significantly misregulated > 1.4-fold (Fig. 6b–d and Table data has suggested that PAX7 regulated genes are aber- S2). The mild model showed 663 genes differentially up- rantly repressed in FSHD [83, 84]. PAX7 expression can regulated and 192 genes differentially downregulated, suppress DUX4-FL cytotoxicity  in vitro and the the moderate model showed 1295 genes differentially PAX7 and DUX4-FL homeodomains are functionally upregulated and 852 genes differentially downregulated, interchangeable . However, DUX4 and PAX7 have while the severe model showed 2577 genes differentially non-overlapping expression profiles during in vitro myo- upregulated and 1962 genes differentially downregulated genesis of FSHD-derived stem cells , single-cell compared with ACTA1-MCM controls. This included RNA-seq show DUX4-negative cells have a PAX7 re- known murine DUX4 target genes previously identified pression signature , and there are no published re- from C2C12 cells such as Wfdc3 . However, when ports of PAX7/DUX4 double-positive cells being found DUX4-induced genes (> 2-fold) from murine skeletal in FSHD biopsies or cell culture, all suggesting the muscle were compared with differential gene expression PAX7-related mis-regulation is likely downstream of profiles from human FSHD patient muscle biopsies  DUX4 expression in FSHD. Since PAX7/Pax7 are essen- there was very little overlap, with only 127 upregulated tially functionally conserved between human and mouse genes and 10 downregulated genes being the same , we analyzed our RNA-seq data similarly  for a (Table S3). This is actually not surprising since human Pax7 repression RNA signature (Figure S9). However, DUX4 expressed in cultured mouse cells induces a gene our analysis revealed that levels of both Pax7 and Pax7- set that only partially overlaps with DUX4 expression in induced genes are in fact upregulated in response to cultured human cells [80, 81], and many functional DUX4 levels in the mouse models, which may be indica- DUX4 binding sites in the human genome are within tive of stimulated muscle repair and regeneration. Fig. 6 Comparison of gene expression in muscle from FSHD-like models compared with controls. a–d Scatter plots showing differentially expressed genes in RNA-seq data derived from gastrocnemius muscles isolated from a FLExD/+ mice, b bi-transgenic mild model mice, c bi- transgenic moderate model mice, and d bi-transgenic severe model mice (y-axis), each compared with ACTA1-MCM control mice (x-axis). Significantly upregulated genes are indicated in red, and downregulated genes are indicated in blue. e–h Above scatter plots highlighting the DUX4-induced genes similarly induced by DUX4 in C2C12 myoblasts  Jones et al. Skeletal Muscle (2020) 10:8 Page 15 of 28 Despite the low overlap in specific misregulated genes reveals differences between mouse muscle expressing between muscles from FSHD patients and FSHD-like human DUX4-fl and cultured murine cells expressing mice, there is some conservation in DUX4-activated human DUX4-fl, the most prominent being the inver- pathways in human and mouse , and the general sion of the immune genes group. However, since the DUX4-induced myopathic phenotype appears to be con- DUX4-expressing C2C12 cells express higher levels of served across several species [42, 89–92]. DUX4 is pro- DUX4 than any of the mouse models, we also performed apoptotic and its expression is detrimental to muscle de- the GO analysis at a higher threshold (> 2-fold) for ex- velopment and differentiation [7, 21, 38, 40, 42–44, 85, pression changes (Figure S10) for an additional 89, 91, 93]. Expression of DUX4 stimulates genes that comparison. modulate the immune response , and FSHD patient DUX4-FL expression has been shown to inhibit biopsies show expression of immune genes, features of nonsense-mediated decay, resulting in accumulation of inflammation visualized by MRI, and immune cell infil- mRNAs that would be typical substrates for NMD . tration [26, 35, 94–96]. Therefore, we performed gene Therefore, we analyzed the RNA-seq data for aberrant ontology (GO) enrichment analysis on the differentially alternative splicing events resulting in skipped exons expressed genes (> 1.4-fold change, Fig. 7) from the (SE) or retained introns (RI) using MISO . MISO FSHD-like mild, moderate, and severe mouse models fo- identifies alternative splicing events based on prede- cusing on pathways consistent with FSHD (Superterms: fined splicing events, taking advantage of existing Apoptosis, Cell cycle, Immune, and Muscle; Figs. 7 and knowledge and therefore is more accurate in character- S10 and Tables 2, S4, and S5). A similar GO analysis izing alternative splicing events in RNA-seq data. The using RNA-seq performed on C2C12 cells with inducible MISO analysis of the RNA-seq data revealed that the DUX4-fl expression  is shown for comparison and moderate and severe disease models had significant Fig. 7 GO enrichment analysis of differentially expressed genes in FSHD-like severity model mice. GO enrichment analysis for DUX4-induced genes identified in (left to right) C2C12 cells , mild FSHD-like mice, moderate FSHD-like model mice, and severe FSHD-like model mice. The size of the beige-colored circles indicates the total number of differentially expressed genes (> 1.4-fold), colored circles indicate the proportion of genes from significantly enriched GO terms, that are in turn offspring of GO superterms (see the “Methods” section and Table S4) relating to mus. Upregulated genes are represented in the top half and downregulated genes are represented in the lower half. The number of genes in each grouping is indicated (n) Jones et al. Skeletal Muscle (2020) 10:8 Page 16 of 28 Table 2 GO superterms enriched in FSHD-like models Table 3 Alternative splicing analysis Comparison Status GO % of genes Phenotype Mouse ID* #SE #RI superterm associated ACTA1-MCM/+ MCM-1127 9619 1905 iDUX4 Dox vs vehicle Up Apoptosis 16.42 MCM-1246 9775 1934 iDUX4 Dox vs vehicle Up Cell cycle 12.55 MCM-1349 9890 1940 iDUX4 Dox vs vehicle Up Immune 0.53 9761 ± 95 1926 ± 14 iDUX4 Dox vs vehicle Down Cell cycle 39.25 FLExD/+ FLExD-1358 9704 1910 iDUX4 Dox vs vehicle Down Immune 7.07 FLExD-1367 9745 1913 iDUX4 Dox vs vehicle Down Muscle 4.99 FLExD-1378 9828 1934 Moderate model vs Up Immune 30.06 9759 ± 46 1919 ± 10 MCM p = 0.489896 p = 0.30383 Moderate model vs Up Cell cycle 24.91 MCM Moderate model dTGM-1236 10,987 2278 Moderate model vs Up Apoptosis 10.29 dTGM-1308 10,917 2261 MCM dTGM-1331 10,940 2255 Moderate model vs Down Muscle 8.23 10,948 ± 26 2265 ± 9 MCM p = 0.000064 p < 0.00001 Severe model vs MCM Up Immune 31.05 Severe model dTGS-1047 11,691 2505 Severe model vs MCM Up Cell cycle 26.43 dTGS-1066 11,658 2456 Severe model vs MCM Up Apoptosis 16.95 dTGS-1079 11,560 2466 Severe model vs MCM Up Muscle 0.70 11,636 ± 51 2476 ± 20 Severe model vs MCM Down Muscle 14.84 p = 0.000014 p < 0.00001 Superterms listed in Table S4 Genes enriched for each comparison are listed in Table S5 p values calculated compared with ACTA1-MCM/+ *Mouse ID correlates with RNA-seq data in Table S1 increases in SE and RI events and that the severe model had significantly more SE and RI events than the mod- MD3, MD6, MD14, and MD28 (Fig. 8), and using male erate model (Tables 3 and S6). This data supports that mice for the severe model, with mice sacrificed at SD3, increased DUX4-fl expression and increased disease se- SD6, and SD9 (Fig. 9). Sex and age-matched ACTA1- verity is correlated with increased SE and RI, which MCM mice injected with appropriate levels of TMX for suggests a decrease in RNA quality control. the model were used as controls, and the mild bi- transgenic (No TMX) model mice were assayed for com- Histological analysis of FSHD-like model mice parison (D0). The three severity levels of myopathic mouse pheno- Analysis of the H&E histology indicated that for the types (Table 1) were analyzed for FSHD-like DUX4- mild model muscles (Fig. 8 e–hand Fig. 9 e–h), very low dependent histopathology [30, 35, 97]. The initial ana- mosaic DUX4-fl expression leads to minor changes in lysis of the mouse models showed different, muscle- histology, the most notable being increased percentage of specific levels of transgene recombination in all three bi- myofibers containing centralized nuclei compared with transgenic severity models (Fig. 1). Therefore, cryosec- ACTA1-MCM control muscles (3–10% vs < 1%, respect- tions for histological analysis were generated from mul- ively; Fig. 10). Interestingly, there are anatomical muscle- tiple muscles, including TA, gastrocnemius, soleus, specific effects that correlate with the level of transgene quadriceps, and heart, for all three models. These muscle recombination. The soleus muscles, which have the lowest sections were then analyzed by hematoxylin and eosin levels of leaky transgene recombination in muscle assayed (H&E) staining to assess fiber morphology, number, and for the mild model, have ~ 3% myofibers with central nu- centralized nuclei (Figs, 8, 9, and 10 and S11 and S12), clei, while the TA muscles, which have a higher recombin- embryonic myosin heavy chain (eMyHC/Myh3) IF and ation rate and thus more DUX4-FL-expressing nuclei (Fig. myostatin expression to assess muscle fiber regeneration 1b), have ~ 10% myofibers with central nuclei (Fig. 10), (Figs. 11, S13, and S14), TUNEL assay to assess apop- suggesting that DUX4-FL expression is driving the forma- tosis (Figs. 12, 13, and S15), and picrosirius red (SR) tion of fibers with central nuclei. staining to assess fibrosis (Figs. 14, S16,and S17). The In contrast to the mild model, the histology from the previously described time-courses of treadmill exhaus- moderate (Fig. 8i–x) and severe (Fig. 9 i– Fig. 10 t, and tion running were performed using female mice for the S11) model mice showed greater effects of DUX4-fl ex- moderate model, with mice sacrificed for histology at pression, including variability in skeletal muscle fiber Jones et al. Skeletal Muscle (2020) 10:8 Page 17 of 28 Fig. 8 Skeletal muscles from the mild and moderately affected FSHD-like mouse models show signs of dystrophic histopathology. Representative cryosections of tibialis anterior, gastrocnemius, soleus, and quadriceps of indicated transgenic animals were analyzed with H&E staining. Female (a–d) ACTA1-MCM mice treated with 1× 5 mg/kg TMX, (e–h) mild model bi-transgenic mice without TMX, and (i–x) moderate model bi- transgenic mice treated 1× with 5 mg/kg TMX. The TMX-injected ACTA1-MCM control mice were assayed at MD28. The TMX-injected bi- transgenic mice were assayed at (i–l) MD3, (m–p) MD6, (q–t) MD14, and (u–x) MD 28. All mice had performed the treadmill exhaustion analysis, as described, 2× per week starting the week prior to TMX injection. Scale bar = 100 μm size, round and triangular fiber shapes, few fiber num- TA. Importantly, as seen in the mild model, these bers, significant increases in the percentage of fibers with models showed anatomical muscle-specific differences in centralized nuclei, an influx of mononuclear cells, and histopathogenic features, correlating with muscle- ultimately an apparent decrease in structural integrity of specific levels of transgene recombination (Fig. 1b), con- the muscle. This histopathology accumulated over time sistent with DUX4-FL expression leading to an atrophic and correlated with loss of muscle function. For ex- phenotype . For example, the soleus muscle, which ample, in the moderate model histology, the few mono- shows the highest level of TMX-responsive transgene re- nuclear cells that have infiltrated by MD6 dramatically combination, also appears to have the greatest degree of increase by MD14. Similarly, the percentage of fibers histopathology for each model, including mononuclear with centralized nuclei jumps significantly between MD6 cell infiltration, disrupted muscle integrity (Fig. 8 k, o, s (4%, 10%) and MD14 (20%, 35%) and MD28 (38%, 52%) and Fig. 9 O, S), and fewer fibers per cross section albeit in both the soleus and TA, respectively (Fig. 10). The se- with a higher percentage of centralized nuclei (Fig. 10). vere model has a similar percentage of myofibers with The quadriceps showed the lowest level of transgene re- centralized nuclei at SD6; however, there is an increase combination and similarly showed the least amount of by SD9 showing ~ 10% in the soleus and ~ 20% in the histopathology for the analyzed skeletal muscles. The Jones et al. Skeletal Muscle (2020) 10:8 Page 18 of 28 Fig. 9 Skeletal muscle from the severely affected FSHD-like mice shows signs of a severe myopathy. Representative cryosections of tibialis anterior, gastrocnemius, soleus, and quadriceps from indicated transgenic animals were analyzed with H&E staining. a–d ACTA1-MCM mice treated with 2× 10 mg/kg TMX, e–h mild model bi-transgenic mice without TMX, and (i–t) severe model bi-transgenic mice treated 2× with 10 mg/kg TMX. The TMX-injected controls were assayed at SD9. The TMX-injected bi-transgenic mice were assayed at (i–l) SD3, (m–p) SD6, and (q– t) SD9. All mice had performed the treadmill exhaustion analysis, as described, 1 week prior to TMX injection, then 2, 5, and 7 DPI. Scale bar = 100 μm heart, which showed no transgene recombination or ex- The centrally positioned nuclei in myofibers are a pression, did not show any pathology even in the severe common feature of many myopathies, including FSHD model (Figure S12). Overall, the skeletal muscle histo- , and are considered a sign of repair and regener- pathology in these models was progressive and the in- ation of the myofiber . Interestingly, all three severity creased severity reflected increased DUX4-FL expression models of FSHD-like mice show increased centralized levels. This further supports that the extent of patho- nuclei, especially following DUX4-FL induction (Fig. 10). physiology in these mouse models correlates with the Therefore, we analyzed muscles from these models for level of transgene recombination (Fig. 1b) and DUX4-fl expression of eMyHC protein (Figs. 11 and S12), a expression (Figs. 2 and 3). marker for newly regenerating myofibers and dystrophic Jones et al. Skeletal Muscle (2020) 10:8 Page 19 of 28 Fig. 10 Skeletal muscles from FSHD-like mouse models develop centralized nuclei. Soleus and TA muscle histological sections from control ACTA1-MCM (blue) and mild (green), moderate (yellow), and severe (red) FSHD-like model mice that had undergone treadmill exhaustion analysis 2× per week (see Figs. 7 and 8), were stained with H&E and counted for myofibers with centralized nuclear. Muscles were assayed at the indicated time points for each model. The ACTA1-MCM +TMX controls were assayed at MD28 and SD9, as appropriate. Data is plotted as percent of myofibers with centralized nuclei. Data are mean ± S.D. with significance calculated by one-way ANOVA using Prism 7. *p < .05, **p < .01, ***p < .001, p < .001 for ACTA1-MCM compared with all other samples in the group. muscle , and expression of Myostatin (Mstn)/Gdf-8 and severe model (Figs. 11M–R and S12) showed high (Figure S13), a negative regulator of muscle growth and levels of eMyHC, peaking at MD14 and SD9, respect- remodeling in adult muscle . Adult mouse skeletal ively, thus confirming that the spike of DUX4-FL expres- muscles have very few myofibers that express the sion at these moderate and severe levels activates the eMyHC isoform under normal healthy conditions . skeletal muscle regeneration program. In contrast to the However, regenerating myofibers re-express many devel- increased eMyHC levels, expression of Mstn mRNA de- opmental isoforms of muscle proteins, including creased in TA muscles as DUX4 levels and severity of eMyHC, which can be detected within 3 days of injury pathology increased (Figure S13), which similarly would and whose expression can persist for up to 3 weeks correlate with the induction of muscle regeneration. To- [102–104]. Therefore, a time course study of eMyHC ex- gether, these results show increased DUX4 expression pression after DUX4-fl induction was performed on TA leads to the muscle remodeling and regeneration. muscles (Fig. 11). As expected, the ACTA1-MCM con- The activation of muscle fiber regeneration after in- trols showed no detectable expression of eMyHC in re- duction of DUX4-FL expression suggests that DUX4-FL sponse to TMX (Figs. 11a–c and S12). Interestingly, the is causing muscle damage. The GO analysis of differen- mild model mice, which have chronic low mosaic ex- tially induced genes showed that muscles from the mod- pression of DUX4-FL, similarly showed no detectable erate and severe models are enriched for genes in expression of eMyHC (Figs. 11K, L, S, T and S12), indi- apoptotic pathways (Fig. 7, Table 2). As mentioned, cating that myofibers with centralized nuclei in the mild DUX4-FL is a pro-apoptotic protein, its expression is model are remnants of an old regeneration event. In highly toxic to muscle cells in culture [7, 21, 44, 105], contrast, both the moderate model (Figs. 11E–J and S12) and an increased apoptosis rate compared is a feature of Jones et al. Skeletal Muscle (2020) 10:8 Page 20 of 28 Fig. 11 Induced DUX4-fl expression causes regeneration of muscle fibers marked with eMyHC expression. Histological cross sections from TA muscles were dissected from mild, moderate, and severe FSHD-like model mice that had undergone treadmill exhaustion analysis 2× per week (see Figs. 7 and 8). Age-matched female control mice were used for the moderate model analysis and male control mice for the severe model analysis. Muscles were assayed from ACTA1-MCM TMX controls at a MD14, b SD3, and c SD9. Representative images of mild model (k, l, s, t), female moderate model (e–i and magnified images f–j), and male severe model (m–q and magnified images n–r) at indicated time point. Quantitative analysis of data is shown in Figure S12. Scale bars = 100 μm FSHD muscle . To assess apoptosis in the three (Figs. 12 and 13, respectively and Figure S14). In the FSHD-like severity models, TUNEL assays were per- moderate model, TUNEL-positive nuclei appeared by formed on TA muscles across the prior DUX4-FL induc- MD6, were prevalent at MD14, and were nearly absent tion time-courses in both the moderate and severe by MD28 (Fig. 12a–c). Similarly, the severe model models as well as age-matched mild (No TMX) mice showed TUNEL-positive nuclei by SD6, with a 2-fold Jones et al. Skeletal Muscle (2020) 10:8 Page 21 of 28 Fig. 12 Muscles from the moderate FHSD-like model mice undergo apoptosis in response to induced DUX4-fl expression. TA muscles from female moderate model mice that had undergone treadmill exhaustion analysis 2× per week (see Fig. 8) were assayed at MD6 (a, e), MD14 (b, f), and MD28 (c, g) or at MD28 for matched mild model mice (d, h). Green signal in the TUNEL assay (a–d) indicates nuclei undergoing apoptosis, compared with DAPI staining (e–h) showing all nuclei in the same histological sections. Scale bar = 200 μm. Quantitative analysis of data is shown in Figure S14 increase by SD9 (Fig. 13a–c). Interestingly, muscles from and Figure S15). The severe model showed increased fi- the mild model showed no indication of apoptosis and brosis by SD3 and a maximal 2.5-fold increase in fibrosis were similar to muscles from ACTA1-MCM controls (5.5% fibrotic area) at SD9 (Fig. 14 h–jand Figure S14). (Fig. 13, compare panels D and L). Thus, the moderate Overall, the level of accumulated fibrosis is small in the and severe models show dose-dependent DUX4-FL- moderate and severe models; however, it should be noted induced apoptosis following a similar time course as that these models only provide a short time frame for fi- muscle weakness and decreased function and correlating brosis to form. The heart, which does not express detect- with the activation of muscle regeneration. able DUX4-fl in any of the models, showed no signs of Muscle biopsy data indicate fibro-fatty replacement in fibrosis in any of the models (Figure S16). muscles of FSHD patients increasing with severity [30, 35, 97]. Muscle fibrosis, caused by stimulated fibroblast Discussion growth resulting in deposits of extracellular matrix be- Modeling FSHD in transgenic mice has historically been tween myofibers, leads to the loss of muscle architecture very difficult despite the fact that FSHD is a gain-of- and decreased muscle function [30, 107–110]. The extent function disease seemingly amenable to transgenic DUX4 of fibrosis is typically quantified using histological overexpression [78, 90, 92]. While the human DUX4 gene methods and staining for collagen [58, 111]. Therefore, has a conserved developmental role with Dux family cross sections of the TA muscles isolated from the above members found in other mammals, including the mouse, series of control, mild, moderate, and severe FSHD-like there is significant divergence at the DNA and protein se- mice were assayed for the extent of fibrosis developing quence level as well as in the spectrum of species-specific over time using SR staining (Figs. 14, S15,and S16). The target genes [25, 80, 112]. In addition, because DUX4-FL muscles from control ACTA1-MCM/+ and mild FSHD- is highly cytotoxic for many somatic cells, leaky expression like model mice had similar low levels (~ 2% fibrotic area) during development has been problematic, and surviving of fibrosis (Fig. 14a, f, g and Figure S15). The moderate mice can be severely phenotypic and difficult to breed model had similar control levels of fibrosis at MD3 and . However, the FLExD mice we recently developed MD6; however, this model showed a small but significant overcame many of these previous limitations . Both 50% increase in fibrosis by MD14 and MD28 (Fig. 14B–E male and female mice are fertile and easy to breed, male Jones et al. Skeletal Muscle (2020) 10:8 Page 22 of 28 Fig. 13 Muscles from the severe FSHD-like model mice start to undergo apoptosis within 6 days of induced DUX4-fl expression. TA muscles from male severe model mice that had undergone treadmill exhaustion analysis (see Fig. 9) were assayed at SD3 (a, e), SD6 (b, f), and SD9 (c, g). ACTA1-MCM (k, l, o, p) and mild FSHD-like model mice (D, H) were analyzed at indicated time point. Green signal in the TUNEL assay (a–d, i–l) indicates nuclei undergoing apoptosis, compared with DAPI staining (e–h, m–p) showing all nuclei in the same histological sections. DNase-I treated histological section of ACTA1-MCM (i, m) and TUNEL staining of the severe model SD9 sample without the TdT enzyme (j, n) are used as positive and negative controls for TUNEL staining, respectively. Scale bar = 200 μm. Quantitative analysis of data is shown in Figure S14 or female transgenic mice can be produced either as trans- controlled increased expression of DUX4-fl mRNA and gene heterozygotes or homozygotes, which can live more protein resulting in FSHD-like muscle pathology . than 1.5 years, and, when mated with an inducible Cre line Here, we report the characterization of a series of pheno- of mice, bi-transgenic mice allow for investigator- typic FSHD-like mouse models varying in severity from Jones et al. Skeletal Muscle (2020) 10:8 Page 23 of 28 Fig. 14 Increased DUX4-fl expression in the moderate and severe FSHD-like model mice leads to accumulation of fibrotic tissue in skeletal muscles. Histological cross sections of TA muscles dissected from mild, moderate, and severe FSHD-like model mice that had undergone treadmill exhaustion analysis 2× per week (see Figs. 7 and 8). Age-matched female control mice were used for the moderate model analysis and male control mice for the severe model analysis. Muscles were assayed from ACTA1-MCM TMX controls at MD28 (a) and SD9 (b), from mild model at MD28 (g), from female moderate model at MD3 (b), MD6 (c), MD14 (d), and MD28 (e), and from male severe model at SD3 (h), SD6 (i), and SD9 (j). Representative histology images are shown. Quantitative analysis of data is shown in Figure S15. Scale bars = 500 μm mild to severe, generated using the FLExD conditional (Fig. 1b, e.g., soleus is low and quadriceps is high). Thus, DUX4-fl transgenic mouse line crossed with ACTA1- the situation in this mild model is similar to that in pre- MCM TMX-inducible mice and identify important con- symptomatic or asymptomatic FSHD patients. These siderations for using these models. We demonstrate that mice live with a chronic, low-level, mosaic expression of these bi-transgenic mouse severity models recapitulate DUX4-FL in a fraction (< 10%) of muscle fibers, yet many aspects of FSHD pathophysiology, thus providing phenotypically the mice appear healthy and behave nor- suitable models for therapeutic interventions targeting mally, with no changes in overall fitness or lifespan. DUX4-fl mRNA, protein, and potentially certain down- Although these mild mice appear outwardly healthy, stream pathways, with several key caveats. In particular, it the low-level recombination produces several assayable is important for those working on these models to keep in phenotypes. The mild phenotype manifests as increased mind the described anatomical muscle-specific and sex- expression of DUX4-FL target genes (Fig. 3), 5–10% of specific differences in pathology and disease progression myofibers with centralized nuclei at 10–12 weeks of age as well as the potential impact the microbiome and differ- (Figs. 8 and 10), and 40% decreased capacity for muscle ent mouse facilities may have on phenotypes [75, 76]. force generation in female (but not male) mice, assayed ex vivo (Figs. 5, S7, and S8). There are also small but sig- Mild FSHD-like mouse model nificant differences between males and females with re- The ACTA1-MCM; FLExD bi-transgenic mouse, in the spect to histology and centralized nuclei that need to be absence of any TMX induction, has mosaic expression taken in to account. Overall, there is some low-level of DUX4-fl mRNA and protein and provides an excel- muscle pathology and new fiber regeneration, however, lent model of mild, pre-symptomatic FSHD. This is be- the mice do not exhibit increased apoptosis, immune cell cause the TMX-inducible Cre fusion protein produced infiltration, or increased fibrosis. The model is readily in skeletal muscles by the ACTA1-MCM transgenic line scalable, highly reproducible, and, considering that these exhibits leakiness into nuclei, resulting in Cre activity in mice live > 1.5 years (female n = 43, male n = 20, to the absence of TMX in a fraction of nuclei. Thus, in the date), provides a DUX4-fl expression model that is ACTA1-MCM;FLExD bi-transgenic mice, this results in amenable to longevity studies for efficacy of putative recombination of the DUX4 transgene and low mosaic DUX4-targeted therapeutics. Thus, while the mild model expression of DUX4-FL protein in skeletal muscles is likely the easiest to work with and imposes no time throughout their lifetime. Importantly, although this limits on treatments, investigators must be careful in re- leaky recombination is specific to skeletal muscle, it is spect to sex, using a significant number of both sexes not uniform among skeletal muscles, with different ana- and analyzing them separately, which ages are used since tomical muscles exhibiting different, but consistent, this is a chronic accumulation pathology model, and levels of recombination and thus DUX4-fl expression which specific muscles to assay or treat. Jones et al. Skeletal Muscle (2020) 10:8 Page 24 of 28 Moderate FSHD-like mouse model analyses, as the DUX4-induced pathology is so severe Similar model-specific effects need to be taken into ac- that the mice require sacrifice no later than SD10. These count when injecting TMX to generate the moderate and mice do not show any signs of recovery. As in the mod- severe models, but these are even more pronounced. Im- erate model, there are consistently different levels of portantly, while the moderate model showed consistently transgene recombination and DUX4-fl expression different levels of TMX-induced recombination between among different anatomical muscles (Fig. 1b). However, anatomical muscles (Fig. 1b), these were different from different muscles show the same patterns as in the mod- the leaky recombination in the mild model and more erate model (e.g., soleus is highest and quadriceps is likely a reflection of the difference in TMX accessibility to lowest for both models), supporting the idea that acces- various muscles. The moderate model mice also showed sibility to the TMX that is responsible for the variability. significant differences between males and females with re- In addition, as seen for both the mild and moderate spect to weight (Figure S6), treadmill profiles (Fig. 4), and models, females and males of the severe model showed muscle physiology (Figs. 5 and S7), with females being sex-specific differences, with females again being more more severely affected by all metrics. With respect to ap- severely affected. Upon induction of DUX4-fl expression, pearance and progression of pathology, and, as opposed to eMyHC expression (Figs. 11 and S13), apoptosis (Figs. the mild model, the moderate model mice show DUX4- 13 and S15), and fibrosis (Figs. 14 and S16) appear by induced apoptosis (Fig. 12). DUX4-FL protein expression SD6 and are all at the highest levels for any model at and apoptosis peak at MD14, at which point there is also SD9. Treadmill stamina is significantly affected by SD6, an increase in eMyHC positive, newly regenerated fibers and mice are immobile by SD9, with muscles producing as wells as fibrosis (Figs. 11, 14,and S16). Muscles at this ~ 25% of the force generated by ACTA1-MCM controls stage produce ~ 60% of the force produced by ACTA1- and ~ 50% of the force produced in the moderate model MCM controls (Fig. 5). Analysis of global differential gene (Figs. 4, 5, and S7). These markers for pathology are sup- expression supports the activation of apoptosis, the im- ported by global differential gene expression profiles that mune response, and the cell cycle, all three of which are show greater enrichment for induced genes relating to much larger groupings than enriched in C2C12 cells ex- apoptosis, immune response, and cell cycle, compared to pressing DUX4. This illustrates a key difference between the moderate model, while many muscle biology genes performing studies in vitro using single-cell types overex- are significantly decreased (Fig. 7). pressing DUX4 compared with studies of intact muscle expressing mosaic levels of DUX4 and containing all asso- Conclusions ciated cell types (Fig. 7). The goal in this study was to generate differing levels of A key component of the moderate model is that after FSHD-like severity using our FLExD mouse model and MD14 both male and female mice recover on their own, characterize the progression of pathology in ways useful regaining ~ 50% treadmill running by MD28, which coin- to those performing preclinical testing of candidate cides with decreases in apoptosis, and eMyHC staining, al- DUX4-targeted therapeutics. We have provided an initial though fibrosis remains. This is likely because DUX4-FL- molecular, phenotypic, physiological, histological, and positive myofibers die and are being repaired and replaced transcriptome characterization of three severity levels of using satellite cells that did not undergo transgene recom- FSHD-like model mice based on three levels of mosaic bination and thus have not activated DUX4 expression. DUX4-fl expression in skeletal muscles. A key feature of Therefore, it is imperative to have the proper controls these models is that the ACTA1-MCM;FLExD bi- when using these models for preclinical testing of thera- transgenic mice express chronic low-level mosaic and peutics. However, this also presents an opportunity skeletal muscle-specific DUX4-fl mRNA and protein whereby the mice can be re-injected with TMX at some without any TMX induction throughout their lifetime. point during treatment and progressive decline of the Thus, the moderate and severe models are not introdu- model can be assessed over several months. Preliminary cing DUX4 expression to a naive system, instead provid- experiments in our lab suggest that this is a viable possi- ing a situation similar to the bursts of DUX4 expression bility that needs further investigation and characterization. seen in FSHD myocytes . It is well documented that In addition, TMX can be adjusted to single higher-dose cells from asymptomatic FSHD subjects express DUX4 injections or multiple lower-dose injections to refine the and cells from relatively healthy muscle biopsies from model to meet investigational needs. The flexibility and clinically affected FSHD patients express significant tunability of this model are almost endless. levels of DUX4 [10, 18, 19]. Thus, these models are re- capitulating how we envision the DUX4 expression situ- Severe FSHD-like mouse model ation in FSHD whereby DUX4 levels in skeletal muscles While the moderate model decline can be assayed over correlate with pathology . Thus, these bi-transgenic 2 weeks, the severe model is only useful for short-term FSHD-like models allow investigators to recapitulate the Jones et al. Skeletal Muscle (2020) 10:8 Page 25 of 28 chronic, low-levels of DUX4 using the mild model, or Additional file 2: Table S1. Significantly altered gene expression in investigate more severe DUX4-mediated pathology by an FSHD-like models investigator-controlled increase in DUX4 expression Additional file 3: Table S2. Differentially expressed genes using TMX. In addition, the bi-transgenic model is the Additional file 4: Table S3. Intersection of misregulated genes in models with FSHD patients only available DUX4 model mouse that lives a normal Additional file 5: Table S4. GO term superterms lifespan (up to 2 years) while continually expressing de- Additional file 6: Table S5. GO superterm genes tectable levels of mosaic DUX4-FL protein throughout Additional file 7: Table S6. MISO alternative splicing analysis for SE its skeletal musculature [78, 91, 113], making it the only and RI choice for long-term therapeutic knockdown studies of DUX4-fl. Abbreviations Overall, these dose-dependent DUX4-fl FSHD-like DPI: Days post-injection; EDL: Extensor digitorum longus; FLExD: FLExDUX4; phenotypic mouse models strongly support the DUX4 FSHD: Facioscapulohumeral muscular dystrophy; GA: Gastrocnemius; H&E: Hematoxylin & eosin; IF: Immunofluorescence; IP: Intraperitoneal; misexpression model for levels of DUX4 expression me- MCM: MerCreMer; MD: Moderate model day; Mstn: Myostatin; nLacZ: Nuclear diating levels of FSHD pathology [17, 20] and provide a ß-galactosidase; PCR: Polymerase chain reaction; qRT-PCR: Quantitative useful and highly flexible tool for performing FSHD pre- reverse transcriptase PCR; QUA: Quadriceps; RNA-seq: RNA sequencing; S.D.: Standard deviation; SD: Severe model day; SOL: Soleus; TA: Tibialis clinical testing of therapeutic approaches targeting anterior; TMX: Tamoxifen; TUNEL: Terminal deoxynucleotide transferase dUTP DUX4-fl mRNA and protein. Importantly for future ana- nick end labeling lyses, we have shown sex-specific differences, anatomical Acknowledgements muscle-specific differences, and model-specific differ- We thank Jennifer Burgess; Chris Carrino; Mick Hitchcock, Ph.D.; Dr. Bill R. ences that must be taken into account when using these Lewis Sr. and Duncan Lewis; and Daniel P. Perez for supporting our FSHD FSHD-like mice. Within a single mouse, one can assess mouse work. We thank Dr. Charis Himeda for helpful discussions and editing the manuscript. This work was funded by grants from the Chris Carrino differentially affected muscles. Studying both sexes from Foundation for FSHD, the FSH Society (FSHS-22012-01), the Muscular a cross provides more fine-tuning of effects as well, with Dystrophy Association (MDA383364), and the National Institute of Arthritis females being slightly but significantly more affected and Musculoskeletal and Skin Diseases (R01 AR070432 and R21 AR070438) to PLJ and from the National Institute of Neurological Disorders and Stroke P01 than the males. This provides even greater flexibility and NS069539 to RKB. PLJ is supported by the Mick Hitchcock, Ph.D. Endowed utility for the model as a tool for studying FSHD and Chair of Medical Biochemistry at UNR Med. RKB is a Scholar of The Leukemia testing potential therapeutic approaches targeting the & Lymphoma Society. DUX4-fl mRNA and/or protein. Authors’ contributions TJ conceived of the study, performed experiments, analyzed data, and wrote the manuscript. GW analyzed data and wrote the manuscript. PB performed Supplementary information experiments and analyzed data. SS performed experiments and analyzed Supplementary information accompanies this paper at https://doi.org/10. data. MR performed experiments and analyzed data. RW performed 1186/s13395-020-00227-4. experiments, analyzed data, and wrote the manuscript. DB conceived of the study and wrote the manuscript. RB analyzed data and wrote the manuscript. PJ conceived of the study, analyzed data, and wrote the Additional file 1: Figure S1. FLExDUX4 transgenic mice kept in manuscript. All authors read and approved the final manuscript. different housing facilities acquire different mouth and gut microbiomes and have differing alopecia and GI health. Figure S2. Mosaic tamoxifen Availability of data and materials dose-dependent recombination in gastrocnemius muscle of ACTA1- NZG Most data generated and analyzed during this study are included in the MCM;R26 bi-transgenic mice. Figure S3. Increased TMX dosage leads manuscript and supplemental data. Any data not included is available from to increased mosaic recombination in skeletal muscle of ACTA1- NZG the corresponding authors upon request. Raw RNA-seq data generated in MCM;R26 bi-transgenic mice. Figure S4. There is no significant differ- this study has been deposited in the NCBI GEO database under accession ence in the transgene recombination rate between male and female number GSE122562. ACTA1-MCM/FLExD bi-transgenic FSHD-like mice. Figure S5. Quantifica- tion of DUX4-FL protein positive myonuclei. Figure S6. The moderate Ethics approval and consent to participate and severe FSHD-like mouse models show significant weight loss. Figure All animal procedures were approved by the local IACUC committee at the S7. Maximum isometric forces of the female FSHD-like mouse models. University of Nevada, Reno (#0701). Figure S8. Maximum and specific isometric forces of the male mild and severe FSHD-like mouse models. Figure S9. Pax7 target genes are not Consent for publication significantly misexpressed in the FSHD-like mouse models. Figure S10. Not applicable GO enrichment analysis of differentially expressed genes (>2-fold) in the different FSHD-like severity model mice. Figure S11. Fiber number per Competing interests cross-section does not significantly change with severity. Figure S12. The authors declare that they have no competing interests. The heart is not affected by TMX treatment in control or bi-transgenic an- imals. Figure S13. Quantification of eMyHC positive muscle cells in dif- Author details ferent FSHD-like mouse models. Figure S14. Mstn gene mRNA Department of Pharmacology, School of Medicine, University of Nevada, expression decreases with increased DUX4-fl expression in the FSHD-like Reno, Reno, NV 89557, USA. Computational Biology Program, Public Health mouse models. Figure S15. Quantification of TUNEL positive nuclei in Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA FSHD-like mouse models. Figure S16. Quantification of SR staining 98109, USA. Basic Sciences Division, Fred Hutchinson Cancer Research shows significant fibrosis in late stages of the moderate and severe Center, Seattle, WA 98109, USA. Current Address: The Cancer Science FSHD-like mouse models. Figure S17. Cardiac muscle from FSHD-like Institute of Singapore, National University of Singapore, Singapore, mouse models shows no signs of increased fibrosis. Singapore. Jones et al. Skeletal Muscle (2020) 10:8 Page 26 of 28 Received: 21 February 2020 Accepted: 5 March 2020 differences in CpG methylation at D4Z4 correlate with clinical variability in FSHD1 and FSHD2. Hum Mol Genet. 2015;24:659–69. 20. Tawil R, van der Maarel SM, Tapscott SJ. Facioscapulohumeral dystrophy: the path to consensus on pathophysiology. Skelet Muscle. 2014;4:12. 21. Rickard AM, Petek LM, Miller DG. Endogenous DUX4 expression in FSHD References myotubes is sufficient to cause cell death and disrupts RNA splicing and cell 1. Padberg GW, Frants RR, Brouwer OF, Wijmenga C, Bakker E, Sandkuijl LA. migration pathways. Hum Mol Genet. 2015;24:5901–14. Facioscapulohumeral muscular dystrophy in the Dutch population. Muscle 22. Dixit M, Ansseau E, Tassin A, Winokur S, Shi R, Qian H, Sauvage S, Matteotti Nerve. 1995;2:S81-S84. C, van Acker AM, et al. DUX4, a candidate gene of facioscapulohumeral 2. van der Maarel SM, Miller DG, Tawil R, Filippova GN, Tapscott SJ. muscular dystrophy, encodes a transcriptional activator of PITX1. Proc Natl Facioscapulohumeral muscular dystrophy: consequences of chromatin Acad Sci U S A. 2007;104:18157–62. relaxation. Curr Opin Neurol. 2012;25:614–20. 23. De Iaco A, Planet E, Coluccio A, Verp S, Duc J, Trono D. DUX-family 3. Deenen JC, Arnts H, van der Maarel SM, Padberg GW, Verschuuren JJ, transcription factors regulate zygotic genome activation in placental Bakker E, Weinreich SS, Verbeek AL, van Engelen BG. Population-based mammals. Nat Genet. 2017;49:941–5. incidence and prevalence of facioscapulohumeral dystrophy. Neurology. 24. Hendrickson PG, Dorais JA, Grow EJ, Whiddon JL, Lim JW, Wike CL, Weaver 2014;83:1056–9. BD, Pflueger C, Emery BR, et al. Conserved roles of mouse DUX and human 4. Prevalence and incidence of rare diseases: Bibliographic data [http://www. DUX4 in activating cleavage-stage genes and MERVL/HERVL orpha.net/orphacom/cahiers/docs/GB/Prevalence_of_rare_diseases_by_ retrotransposons. Nat Genet. 2017;49:925–34. alphabetical_list.pdf]. 25. Whiddon JL, Langford AT, Wong CJ, Zhong JW, Tapscott SJ. Conservation 5. Gabriels J, Beckers MC, Ding H, De Vriese A, Plaisance S, van der Maarel SM, and innovation in the DUX4-family gene network. Nat Genet. 2017;49:935– Padberg GW, Frants RR, Hewitt JE, et al. Nucleotide sequence of the partially deleted D4Z4 locus in a patient with FSHD identifies a putative gene within 26. Yao Z, Snider L, Balog J, Lemmers RJ, Van Der Maarel SM, Tawil R, Tapscott each 3.3 kb element. Gene. 1999;236:25–32. SJ. DUX4-induced gene expression is the major molecular signature in 6. van Overveld PG, Lemmers RJ, Sandkuijl LA, Enthoven L, Winokur ST, Bakels FSHD skeletal muscle. Hum Mol Genet. 2014;23:5342–52. F, Padberg GW, van Ommen GJ, Frants RR, et al. Hypomethylation of D4Z4 27. Campbell AE, Belleville A, Resnick R, Shadle SC, Tapscott SJ. in 4q-linked and non-4q-linked facioscapulohumeral muscular dystrophy. Facioscapulohumeral dystrophy: activating an early embryonic Nat Genet. 2003;35:315–7. transcriptional program in human skeletal muscle. Hum Mol Genet. 2018. 7. Kowaljow V, Marcowycz A, Ansseau E, Conde CB, Sauvage S, Matteotti C, Arias C, Corona ED, Nunez NG, et al. The DUX4 gene at the FSHD1A locus 28. Tassin A, Laoudj-Chenivesse D, Vanderplanck C, Barro M, Charron S, Ansseau encodes a pro-apoptotic protein. Neuromuscul Disord. 2007;17:611–23. E, Chen YW, Mercier J, Coppee F, et al. DUX4 expression in FSHD muscle cells: how could such a rare protein cause a myopathy? J Cell Mol Med. 8. de Greef JC, Lemmers RJ, van Engelen BG, Sacconi S, Venance SL, Frants RR, 2013;17:76–89. Tawil R, van der Maarel SM. Common epigenetic changes of D4Z4 in 29. Padberg GW: Facioscapulohumeral Disease [thesis]. Leiden, the Netherlands: contraction-dependent and contraction-independent FSHD. Hum Mutat. Leiden University; 1982. 2009;30:1449–59. 30. Wang LH, Tawil R. Facioscapulohumeral dystrophy. Curr Neurol Neurosci 9. Snider L, Geng LN, Lemmers RJ, Kyba M, Ware CB, Nelson AM, Tawil R, Rep. 2016;16:66. Filippova GN, van der Maarel SM, et al. Facioscapulohumeral dystrophy: 31. Himeda CL, Debarnot C, Homma S, Beermann ML, Miller JB, Jones PL, Jones incomplete suppression of a retrotransposed gene. PLoS Genet. 2010;6: TI. Myogenic enhancers regulate expression of the facioscapulohumeral e1001181. muscular dystrophy associated DUX4 gene. Mol Cell Biol. 2014;34:1942–55. 10. Jones TI, King OD, Himeda CL, Homma S, Chen JC, Beermann ML, Yan C, Emerson CP Jr, Miller JB, et al. Individual epigenetic status of the 32. Haynes P, Bomsztyk K, Miller DG. Sporadic DUX4 expression in FSHD pathogenic D4Z4 macrosatellite correlates with disease in myocytes is associated with incomplete repression by the PRC2 complex facioscapulohumeral muscular dystrophy. Clin Epigen. 2015;7:37. and gain of H3K9 acetylation on the contracted D4Z4 allele. Epigenetics 11. Himeda CL, Jones PL. The genetics and epigenetics of facioscapulohumeral Chromatin. 2018;11:47. muscular dystrophy. Annu Rev Genomics Hum Genet. 2019;20:265–91. 33. Geng LN, Yao Z, Snider L, Fong AP, Cech JN, Young JM, van der Maarel SM, Ruzzo WL, Gentleman RC, et al. DUX4 activates germline genes, 12. Johnson NE, Statland JM. FSHD1 or FSHD2: That is the question: the answer: it's all just FSHD. Neurology. 2019;92:881–2. retroelements, and immune mediators: Implications for facioscapulohumeral dystrophy. Dev Cell. 2012;22:38–51. 13. Wijmenga C, Hewitt JE, Sandkuijl LA, Clark LN, Wright TJ, Dauwerse HG, 34. Jagannathan S, Shadle SC, Resnick R, Snider L, Tawil RN, van der Maarel SM, Gruter AM, Hofker MH, Moerer P, et al. Chromosome 4q DNA Bradley RK, Tapscott SJ. Model systems of DUX4 expression recapitulate the rearrangements associated with facioscapulohumeral muscular dystrophy. transcriptional profile of FSHD cells. Hum Mol Genet. 2016;25:4419–31. Nat Genet. 1992;2:26–30. 14. van Deutekom JC, Wijmenga C, van Tienhoven EA, Gruter AM, Hewitt JE, 35. Wang LH, Friedman SD, Shaw D, Snider L, Wong CJ, Budech CB, Poliachik Padberg GW, van Ommen GJ, Hofker MH, Frants RR. FSHD associated DNA SL, Gove NE, Lewis LM, et al. MRI-informed muscle biopsies correlate MRI rearrangements are due to deletions of integral copies of a 3.2 kb tandemly with pathology and DUX4 target gene expression in FSHD. Hum Mol Genet. repeated unit. Hum Mol Genet. 1993;2:2037–42. 2019;28:476–86. 15. Lemmers RJ, Tawil R, Petek LM, Balog J, Block GJ, Santen GW, Amell AM, van 36. Homma S, Beermann ML, Boyce FM, Miller JB. Expression of FSHD-related der Vliet PJ, Almomani R, et al. Digenic inheritance of an SMCHD1 mutation DUX4-FL alters proteostasis and induces TDP-43 aggregation. Annals of and an FSHD-permissive D4Z4 allele causes facioscapulohumeral muscular clinical and translational neurology. 2015;2:151–66. dystrophy type 2. Nat Genet. 2012;44:1370–4. 37. Feng Q, Snider L, Jagannathan S, Tawil R, van der Maarel SM, Tapscott 16. van den Boogaard ML, Lemmers RJ, Balog J, Wohlgemuth M, Auranen M, SJ, Bradley RK. A feedback loop between nonsense-mediated decay and Mitsuhashi S, van der Vliet PJ, Straasheijm KR, van den Akker RF, et al. the retrogene DUX4 in facioscapulohumeral muscular dystrophy. eLife. Mutations in DNMT3B modify epigenetic repression of the D4Z4 repeat and 2015;4. the penetrance of facioscapulohumeral dystrophy. Am J Hum Genet. 2016; 38. Vanderplanck C, Ansseau E, Charron S, Stricwant N, Tassin A, Laoudj- 98:1020–9. Chenivesse D, Wilton SD, Coppee F, Belayew A. The FSHD strophic myotube phenotype is caused by DUX4 expression. PLoS One. 2011;6:e26820. 17. Lemmers RJ, van der Vliet PJ, Klooster R, Sacconi S, Camano P, Dauwerse JG, Snider L, Straasheijm KR, van Ommen GJ, et al. A unifying genetic model for 39. Dmitriev P, Bou Saada Y, Dib C, Ansseau E, Barat A, Hamade A, Dessen P, facioscapulohumeral muscular dystrophy. Science. 2010;329:1650–3. Robert T, Lazar V, et al. DUX4-induced constitutive DNA damage and 18. Jones TI, Chen JC, Rahimov F, Homma S, Arashiro P, Beermann ML, King oxidative stress contribute to aberrant differentiation of myoblasts from OD, Miller JB, Kunkel LM, et al. Facioscapulohumeral muscular dystrophy FSHD patients. Free Radic Biol Med. 2016;99:244–58. family studies of DUX4 expression: evidence for disease modifiers and a 40. Bosnakovski D, Toso EA, Hartweck LM, Magli A, Lee HA, Thompson ER, quantitative model of pathogenesis. Hum Mol Genet. 2012;21:4419–30. Dandapat A, Perlingeiro RCR, Kyba M. The DUX4 homeodomains mediate 19. Lemmers RJ, Goeman JJ, Van Der Vliet PJ, Van Nieuwenhuizen MP, Balog J, inhibition of myogenesis and are functionally exchangeable with the Pax7 Vos-Versteeg M, Camano P, Ramos Arroyo MA, Jerico I, et al. Inter-individual homeodomain. J Cell Sci. 2017. Jones et al. Skeletal Muscle (2020) 10:8 Page 27 of 28 41. Bosnakovski D, Gearhart MD, Toso EA, Ener ET, Choi SH, Kyba M. Low level 62. Flicek P, Ahmed I, Amode MR, Barrell D, Beal K, Brent S, Carvalho-Silva D, DUX4 expression disrupts myogenesis through deregulation of myogenic Clapham P, Coates G, et al. Ensembl 2013. Nucleic Acids Res. 2013;41:D48– gene expression. Sci Rep. 2018;8:16957. 55. 42. Wuebbles RD, Long SW, Hanel ML, Jones PL. Testing the effects of FSHD 63. Katz Y, Wang ET, Airoldi EM, Burge CB. Analysis and design of RNA candidate gene expression in vertebrate muscle development. Int J Clin Exp sequencing experiments for identifying isoform regulation. Nat Methods. Pathol. 2010;3:386–400. 2010;7:1009–15. 64. Dvinge H, Ries RE, Ilagan JO, Stirewalt DL, Meshinchi S, Bradley RK. Sample 43. Wallace LM, Garwick SE, Mei W, Belayew A, Coppee F, Ladner KJ, Guttridge processing obscures cancer-specific alterations in leukemic transcriptomes. D, Yang J, Harper SQ. DUX4, a candidate gene for facioscapulohumeral Proc Natl Acad Sci U S A. 2014;111:16802–7. muscular dystrophy, causes p53-dependent myopathy in vivo. Ann Neurol. 2011;69:540–52. 65. Li B, Dewey CN. RSEM: accurate transcript quantification from RNA-Seq data 44. Shadle SC, Zhong JW, Campbell AE, Conerly ML, Jagannathan S, Wong CJ, with or without a reference genome. BMC Bioinformatics. 2011;12:323. Morello TD, van der Maarel SM, Tapscott SJ. DUX4-induced dsRNA and MYC 66. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient mRNA stabilization activate apoptotic pathways in human cell models of alignment of short DNA sequences to the human genome. Genome Biol. facioscapulohumeral dystrophy. PLoS Genet. 2017;13:e1006658. 2009;10:R25. 45. Himeda CL, Jones TI, Jones PL. Facioscapulohumeral muscular dystrophy as 67. Trapnell C, Pachter L, Salzberg SL. TopHat: discovering splice junctions with a model for epigenetic regulation and disease. Antioxidants & redox RNA-Seq. Bioinformatics. 2009;25:1105–11. signaling. 2015;22:1463–82. 68. Robinson MD, Oshlack A. A scaling normalization method for differential 46. Schnutgen F, Doerflinger N, Calleja C, Wendling O, Chambon P, Ghyselinck expression analysis of RNA-seq data. Genome Biol. 2010;11:R25. NB. A directional strategy for monitoring Cre-mediated recombination at 69. Wagenmakers EJ, Lodewyckx T, Kuriyal H, Grasman R. Bayesian hypothesis the cellular level in the mouse. Nat Biotechnol. 2003;21:562–5. testing for psychologists: a tutorial on the Savage-Dickey method. Cogn 47. Schnutgen F, De-Zolt S, Van Sloun P, Hollatz M, Floss T, Hansen J, Psychol. 2010;60:158–89. Altschmied J, Seisenberger C, Ghyselinck NB, et al. Genomewide production 70. Young MD, Wakefield MJ, Smyth GK, Oshlack A. Gene ontology analysis for of multipurpose alleles for the functional analysis of the mouse genome. RNA-seq: accounting for selection bias. Genome Biol. 2010;11:R14. Proc Natl Acad Sci U S A. 2005;102:7221–6. 71. Huber W, Carey VJ, Gentleman R, Anders S, Carlson M, Carvalho BS, Bravo 48. Jones T, Jones PL. A cre-inducible DUX4 transgenic mouse model for HC, Davis S, Gatto L, et al. Orchestrating high-throughput genomic analysis investigating facioscapulohumeral muscular dystrophy. PLoS One. 2018;13: with Bioconductor. Nat Methods. 2015;12:115–21. e0192657. 72. dplyr: A Grammar of Data Manipulation. R Package Version 0.4.3. [http:// 49. Wallace LM, Garwick-Coppens SE, Tupler R, Harper SQ. RNA interference cran.r-project.org/package=dplyr ]. improves myopathic phenotypes in mice over-expressing FSHD region 73. Wickham H: ggplot2: Elegant graphics for data analysis: Springer-Verlag gene 1 (FRG1). Mol Ther. 2011;19:2048–54. New York; 2009. 50. Marsollier AC, Ciszewski L, Mariot V, Popplewell L, Voit T, Dickson G, 74. Helicobacter species [(https://www.criver.com/sites/default/files/resources/ Dumonceaux J. Antisense targeting of 3' end elements involved in DUX4 HelicobacterspeciesTechnicalSheet.pdf)]. mRNA processing is an efficient therapeutic strategy for 75. Sundberg JP, Schofield PN. Living inside the box: environmental effects on facioscapulohumeral dystrophy: a new gene-silencing approach. Hum Mol mouse models of human disease. Dis Model Mech. 2018;11. Genet. 2016;25:1468–78. 76. Treuting PM, Clifford CB, Sellers RS, Brayton CF. Of mice and microflora: 51. Ansseau E, Vanderplanck C, Wauters A, Harper SQ, Coppee F, Belayew A. considerations for genetically engineered mice. Vet Pathol. 2012;49:44–63. Antisense oligonucleotides used to target the DUX4 mRNA as therapeutic 77. Gary RK, Kindell SM. Quantitative assay of senescence-associated beta- approaches in FaciosScapuloHumeral muscular Dystrophy (FSHD). Genes galactosidase activity in mammalian cell extracts. Anal Biochem. 2005;343: (Basel). 2017:8. 329–34. 52. Peart N, Wagner EJ. A distal auxiliary element facilitates cleavage and 78. Krom YD, Thijssen PE, Young JM, den Hamer B, Balog J, Yao Z, Maves L, polyadenylation of Dux4 mRNA in the pathogenic haplotype of FSHD. Hum Snider L, Knopp P, et al. Intrinsic epigenetic regulation of the D4Z4 Genet. 2017;136:1291–301. macrosatellite repeat in a transgenic mouse model for FSHD. PLoS Genet. 53. Snider L, Asawachaicharn A, Tyler AE, Geng LN, Petek LM, Maves L, Miller 2013;9:e1003415. DG, Lemmers RJ, Winokur ST, et al. RNA transcripts, miRNA-sized fragments 79. Moorwood C, Liu M, Tian Z, Barton ER. Isometric and eccentric force and proteins produced from D4Z4 units: new candidates for the generation assessment of skeletal muscles isolated from murine models of pathophysiology of facioscapulohumeral dystrophy. Hum Mol Genet. 2009; muscular dystrophies. Journal of visualized experiments. JoVE. 2013:e50036. 18:2414–30. 80. Sharma V, Harafuji N, Belayew A, Chen YW. DUX4 differentially regulates 54. Lemmers RJ, van der Vliet PJ, van der Gaag KJ, Zuniga S, Frants RR, de Knijff transcriptomes of human rhabdomyosarcoma and mouse C2C12 Cells. PLoS P, van der Maarel SM. Worldwide population analysis of the 4q and 10q One. 2013;8:e64691. subtelomeres identifies only four discrete interchromosomal sequence 81. Knopp P, Krom YD, Banerji CR, Panamarova M, Moyle LA, den Hamer B, van transfers in human evolution. Am J Hum Genet. 2010;86:364–77. der Maarel SM, Zammit PS. DUX4 induces a transcriptome more 55. McCarthy JJ, Srikuea R, Kirby TJ, Peterson CA, Esser KA. Inducible Cre characteristic of a less-differentiated cell state and inhibits myogenesis. J transgenic mouse strain for skeletal muscle-specific gene targeting. Skelet Cell Sci. 2016;129:3816–31. Muscle. 2012;2:8. 82. Young JM, Whiddon JL, Yao Z, Kasinathan B, Snider L, Geng LN, Balog J, 56. Yamamoto M, Shook NA, Kanisicak O, Yamamoto S, Wosczyna MN, Camp Tawil R, van der Maarel SM, et al. DUX4 binding to retroelements creates JR, Goldhamer DJ. A multifunctional reporter mouse line for Cre- and FLP- promoters that are active in FSHD muscle and testis. PLoS Genet. 2013;9: dependent lineage analysis. Genesis. 2009;47:107–14. e1003947. 57. Jones TI, Himeda CL, Perez DP, Jones PL. Large family cohorts of lymphoblastoid 83. Banerji CRS, Panamarova M, Hebaishi H, White RB, Relaix F, Severini S, cells provide a new cellular model for investigating facioscapulohumeral Zammit PS. PAX7 target genes are globally repressed in muscular dystrophy. Neuromuscul Disord. 2017;27:221–38. facioscapulohumeral muscular dystrophy skeletal muscle. Nat Commun. 58. Smith LR, Barton ER. Collagen content does not alter the passive 2017;8:2152. mechanical properties of fibrotic skeletal muscle in mdx mice. Am J Physiol 84. Banerji CRS, Panamarova M, Pruller J, Figeac N, Hebaishi H, Fidanis E, Saxena Cell Physiol. 2014;306:C889–98. A, Contet J, Sacconi S, et al. Dynamic transcriptomic analysis reveals 59. Sperringer JE, Grange RW. In vitro assays to determine skeletal muscle suppression of PGC1alpha/ERRalpha drives perturbed myogenesis in physiologic function. Methods Mol Biol. 2016;1460:271–91. facioscapulohumeral muscular dystrophy. Hum Mol Genet. 2019;28:1244–59. 60. Van Ry PM, Wuebbles RD, Key M, Burkin DJ. Galectin-1 protein therapy 85. Bosnakovski D, Xu Z, Gang EJ, Galindo CL, Liu M, Simsek T, Garner HR, Agha- prevents pathology and improves muscle function in the mdx mouse Mohammadi S, Tassin A, et al. An isogenetic myoblast expression screen model of Duchenne muscular dystrophy. Mol Ther. 2015;23:1285–97. identifies DUX4-mediated FSHD-associated molecular pathologies. EMBO J. 61. Meyer LR, Zweig AS, Hinrichs AS, Karolchik D, Kuhn RM, Wong M, Sloan CA, 2008;27:2766–79. Rosenbloom KR, Roe G, et al. The UCSC Genome Browser database: 86. Haynes P, Kernan K, Zhou SL, Miller DG. Expression patterns of FSHD- extensions and updates 2013. Nucleic Acids Res. 2013;41:D64–9. causing DUX4 and myogenic transcription factors PAX3 and PAX7 are Jones et al. Skeletal Muscle (2020) 10:8 Page 28 of 28 spatially distinct in differentiating human stem cell cultures. Skelet Muscle. 110. MacDonald EM, Cohn RD. TGFbeta signaling: its role in fibrosis formation 2017;7:13. and myopathies. Curr Opin Rheumatol. 2012;24:628–34. 87. Banerji CRS, Zammit PS. PAX7 target gene repression is a superior FSHD 111. Whittaker P, Kloner RA, Boughner DR, Pickering JG. Quantitative assessment biomarker than DUX4 target gene activation, associating with pathological of myocardial collagen with picrosirius red staining and circularly polarized severity and identifying FSHD at the single-cell level. Hum Mol Genet. 2019; light. Basic Res Cardiol. 1994;89:397–410. 28:2224–36. 112. Clapp J, Mitchell LM, Bolland DJ, Fantes J, Corcoran AE, Scotting PJ, Armour 88. Blake JA, Ziman MR. Pax genes: regulators of lineage specification and JA, Hewitt JE. Evolutionary conservation of a coding function for D4Z4, the progenitor cell maintenance. Development. 2014;141:737–51. tandem DNA repeat mutated in facioscapulohumeral muscular dystrophy. Am J Hum Genet. 2007;81:264–79. 89. Mitsuhashi H, Mitsuhashi S, Lynn-Jones T, Kawahara G, Kunkel LM. 113. Giesige CR, Wallace LM, Heller KN, Eidahl JO, Saad NY, Fowler AM, Pyne NK, Expression of DUX4 in zebrafish development recapitulates Al-Kharsan M, Rashnonejad A, et al. AAV-mediated follistatin gene therapy facioscapulohumeral muscular dystrophy. Hum Mol Genet. 2013;22:568–77. improves functional outcomes in the TIC-DUX4 mouse model of FSHD. JCI 90. Lek A, Rahimov F, Jones PL, Kunkel LM. Emerging preclinical animal models Insight. 2018;3. for FSHD. Trends Mol Med. 2015;21:295–306. 91. Bosnakovski D, Chan SSK, Recht OO, Hartweck LM, Gustafson CJ, Athman LL, Lowe DA, Kyba M. Muscle pathology from stochastic low level DUX4 Publisher’sNote expression in an FSHD mouse model. Nat Comm. 2017;8:550. Springer Nature remains neutral with regard to jurisdictional claims in 92. Dandapat A, Bosnakovski D, Hartweck LM, Arpke RW, Baltgalvis KA, Vang D, published maps and institutional affiliations. Baik J, Darabi R, Perlingeiro RC, et al. Dominant lethal pathologies in male mice engineered to contain an X-linked DUX4 transgene. Cell reports. 2014; 8:1484–96. 93. Corona ED, Jacquelin D, Gatica L, Rosa AL. Multiple protein domains contribute to nuclear import and cell toxicity of DUX4, a candidate pathogenic protein for facioscapulohumeral muscular dystrophy. PLoS One. 2013;8:e75614. 94. Arahata K, Ishihara T, Fukunaga H, Orimo S, Lee JH, Goto K, Nonaka I. Inflammatory response in facioscapulohumeral muscular dystrophy (FSHD): immunocytochemical and genetic analyses. Muscle Nerve Suppl. 1995:S56– 95. Tasca G, Pescatori M, Monforte M, Mirabella M, Iannaccone E, Frusciante R, Cubeddu T, Laschena F, Ottaviani P, et al. Different molecular signatures in magnetic resonance imaging-staged facioscapulohumeral muscular dystrophy muscles. PLoS One. 2012;7:e38779. 96. Frisullo G, Frusciante R, Nociti V, Tasca G, Renna R, Iorio R, Patanella AK, Iannaccone E, Marti A, et al. CD8(+) T cells in facioscapulohumeral muscular dystrophy patients with inflammatory features at muscle MRI. J Clin Immunol. 2010. 97. Statland JM, Shah B, Henderson D, van der Maarel S, Tapscott SJ, Tawil R. Muscle pathology grade for facioscapulohumeral muscular dystrophy biopsies. Muscle Nerve. 2015. 98. Folker ES, Baylies MK. Nuclear positioning in muscle development and disease. Front Physiol. 2013;4:363. 99. Schiaffino S, Rossi AC, Smerdu V, Leinwand LA, Reggiani C. Developmental myosins: expression patterns and functional significance. Skelet Muscle. 2015;5:22. 100. McPherron AC, Lawler AM, Lee SJ. Regulation of skeletal muscle mass in mice by a new TGF-beta superfamily member. Nature. 1997;387:83–90. 101. Murgia M, Nagaraj N, Deshmukh AS, Zeiler M, Cancellara P, Moretti I, Reggiani C, Schiaffino S, Mann M. Single muscle fiber proteomics reveals unexpected mitochondrial specialization. EMBO Rep. 2015;16:387–95. 102. Esser K, Gunning P, Hardeman E. Nerve-dependent and -independent patterns of mRNA expression in regenerating skeletal muscle. Dev Biol. 1993;159:173–83. 103. Sartore S, Gorza L, Schiaffino S. Fetal myosin heavy chains in regenerating muscle. Nature. 1982;298:294–6. 104. Carraro U, Catani C, Saggin L, Zrunek M, Szabolcs M, Gruber H, Streinzer W, Mayr W, Thoma H. Isomyosin changes after functional electrostimulation of denervated sheep muscle. Muscle Nerve. 1988;11:1016–28. 105. Bosnakovski D, Choi SH, Strasser JM, Toso EA, Walters MA, Kyba M. High- throughput screening identifies inhibitors of DUX4-induced myoblast toxicity. Skelet Muscle. 2014;4:4. 106. Statland JM, Odrzywolski KJ, Shah B, Henderson D, Fricke AF, van der Maarel SM, Tapscott SJ, Tawil R. Immunohistochemical characterization of facioscapulohumeral muscular dystrophy muscle biopsies. Journal of neuromuscular diseases. 2015;2:291–9. 107. Schessl J, Zou Y, Bonnemann CG. Congenital muscular dystrophies and the extracellular matrix. Semin Pediatr Neurol. 2006;13:80–9. 108. Klingler W, Jurkat-Rott K, Lehmann-Horn F, Schleip R. The role of fibrosis in Duchenne muscular dystrophy. Acta Myol. 2012;31:184–95. 109. Lieber RL, Ward SR. Cellular mechanisms of tissue fibrosis. 4. Structural and functional consequences of skeletal muscle fibrosis. Am J Physiol Cell Physiol. 2013;305:C241–52.
Skeletal Muscle – Springer Journals
Published: Apr 11, 2020