Alternative utrophin mRNAs contribute to phenotypic differences between dystrophin‐deficient mice and Duchenne muscular dystrophy

Duchenne muscular dystrophy (DMD) is a fatal disorder caused by absence of functional dystrophin protein. Compensation in dystrophin‐deficient (mdx) mice may be achieved by overexpression of its fetal paralogue, utrophin. Strategies to increase utrophin levels by stimulating promoter activity using small compounds are therefore a promising pharmacological approach. Here, we characterise similarities and differences existing within the mouse and human utrophin locus to assist in high‐throughput screening for potential utrophin modulator drugs. We identified five novel 5′‐utrophin isoforms (A′,B′,C,D and F) in adult and embryonic tissue. As the more efficient utrophin‐based response in mdx skeletal muscle appears to involve independent transcriptional activation of conserved, myogenic isoforms (A′ and F), elevating their paralogues in DMD patients is an encouraging therapeutic strategy.

Duchenne muscular dystrophy (DMD) is a fatal disorder caused by absence of functional dystrophin protein. Compensation in dystrophin-deficient (mdx) mice may be achieved by overexpression of its fetal paralogue, utrophin. Strategies to increase utrophin levels by stimulating promoter activity using small compounds are therefore a promising pharmacological approach. Here, we characterise similarities and differences existing within the mouse and human utrophin locus to assist in high-throughput screening for potential utrophin modulator drugs. We identified five novel 5 0 -utrophin isoforms (A 0 , B 0 ,C,D and F) in adult and embryonic tissue. As the more efficient utrophinbased response in mdx skeletal muscle appears to involve independent transcriptional activation of conserved, myogenic isoforms (A 0 and F), elevating their paralogues in DMD patients is an encouraging therapeutic strategy.
Keywords: Duchenne muscular dystrophy; therapy; utrophin Counteracting the absence of functional dystrophin in Duchenne muscular dystrophy (DMD) individuals via upregulation of its autosomal paralogue, utrophin [1], has been an appealing option for over two decades. The discovery and preliminary characterization of utrophin presented an ideal candidate to devise a replacement therapeutic strategy given its extensive protein sequence homology, size and functional properties with dystrophin, including association with the dystroglycan complex and F-actin [2][3][4][5]. Furthermore, developmental mouse (Utrn) and human (UTRN) transcription precedes that of dystrophin, and as such is considered a fetal precursor [6]. The latter is particularly evident during skeletal muscle development in mice and humans, where sarcolemmal Utrn/UTRN mRNAs rapidly decline upon dystrophin expression [7] resulting in postnatal confinement primarily to neuromuscular and myotendinous junctions [6,8,9]. Indeed, utrophin has been shown to physiologically rescue dystrophin-deficient mdx [10,11] mouse skeletal muscle by varied means, including transgenic upregulation, viral delivery and oral compound administration (reviewed in [12]). Consequently, discovery and exploitation of regulatory mechanisms that raise endogenous human utrophin mRNA levels are of great interest.
Interestingly, the relatively mild mdx mouse phenotype appears conferred via postnatal mechanisms apparently absent in DMD patients, where limb muscle degeneration is effectively countered by reversion to fetal myogenic programming, including Utrn re-expression [19,20]. Furthermore, mice deficient in both full-length dystrophin and utrophin (dko) are a closer DMD phenocopy [21,22], implicating the regulatory control of utrophin in contributing to this mdx/DMD phenotypic disparity.
Given that the protein localization of A-and Butrophin appears conserved within skeletal muscle [15][16][17], we postulated that mouse and human loci respond differently to counteract dystrophin deficiency by transcriptional regulation of alternative mRNA isoforms. We support this hypothesis by illustrating that novel, independently regulated transcripts are differentially regulated in the mdx mouse, and that conserved myogenic isoforms, in particular utrophin A 0 and F, are valid screening targets for our small compoundmediated utrophin upregulation strategy. Importantly, this approach can be employed in methodologies

Immunoblotting
Specificity of utroF was assessed by dot-blots of (a) serially diluted immunising peptide (

Immunofluorescence
Tissue sectioning and immunofluorescence (IF) was performed as described [31].

Sample numbers and statistical analysis
sqRT-PCR (human tissue panel) was performed n = 3 per isoform using individually synthesised cDNA template mixes, quantified on separate agarose gels. All qRT-PCR reactions were conducted in triplicate, numbers as follows. C2C12 myogenesis and C3H studies were performed using cDNA synthesized from up to 4 or 6 wells, respectively (nonpooled), with standard deviation calculated if n ≥ 3 separate cDNA preparations used. For prepooled samples (aorta-/skeletal muscle-derived mdx mesoangioblasts, D16 and D351 clonal cell lines, mouse total embryo, human embryonic tissue), three separate cDNA preparations were performed in triplicate. Data presented in colour scale (Figs 1B-C, 2A-B) are presented in numerical form in Table S1-S2. Postnatal mouse tissue samples were extracted from n ≥ 3 subjects; two tailed t-tests and P-values on data in Fig. 2A were conducted using Prism7 (GraphPad, La Jolla, USA) and listed in Table S2.

Animal use and ethics statement
The source and use of animal tissue was performed under strict compliance with required standards outlined in the Animals (Scientific Procedures) Act 1986, revised 2012 (ASPA) and European Directive 63/2010 relating to the protection of animals used for scientific purposes.

Results
Transcriptional complexity of mouse and human utrophin loci revealed by 5 0 RACE and genome analysis As the genomic location and splicing configuration of utrophin A and B are conserved between mouse (Utrn) and human (UTRN) gene loci [13,14], we hypothesised unidentified novel regulatory transcripts/elements within the~100 kb 5 0 region spanning exons 1A-3 may contribute towards the mdx/DMD genotypephenotype disparity. Novel exons were identified by 5 0 RACE using embryonic tissue, when utrophin expression peaks [8]. PCR products were cloned and sequenced on both strands to determine transcription start sites. The presence of novel exons were then authenticated using isoform-specific RT-PCR, sequenced (Fig. S1-S4) and prefixed according to established nomenclature. Prime symbols ( 0 ) were allocated to exons that share sequences previously thought unique to previously identified full-length utrophin A-and B-mRNAs (A 0 and B 0 respectively, Fig. 1A).
Utrophin exon 1A 0 lies in a conserved regioñ 550 bp 3 0 to 1A and gives rise to UtrnA 0 mRNA (Fig. S1a) via cosplicing to translated exon 2A. Thus, any protein product arising from utrophin 1A 0 would be indistinguishable from utrophin 1A using mouse [17] and human [15] A-utrophin specific N-terminal antibodies. Conversely, utrophin B 0 isoforms share a common first exon (1B), but divergent splicing results   (Table S2). P values represent differences between wild-type and mdx tissues at the same time point (P < 0.0001••• P < 0.001••, P < 0.05•, 99% cl). qRT-PCR analysis of Utrn-A,-A 0 and -F levels in specific hindlimb muscles (tibilialis anterior, quadricep and soleus) compared to diaphragm ( Fig. 2A) Table S2. in distinct open reading frames that retain actin binding potential (Fig. S2). Three novel first exons splicing to exon 3 (1C,D and F) were also identified ( Fig. 1A; Fig. S3-S4). As mouse exon 1C is untranslated, potential use of an in-frame methionine in exon 4 would give rise to a protein indistinguishable using antibodybased approaches, whereas murine exon 1D contains two potential initiating methionines (Fig. S3). Human Utrn-C and Utrn-D paralogues were not detected, indicating these mRNAs are either (i) mouse-specific or (ii) absent in either the source tissue selected/within the conserved region selected for UTRN-specific 5 0 RACE. In contrast, exon 1F was of particular interest; similar to exons 2A and 1B [17], Utrn-/UTRN-F mRNAs contain open-reading frames and have the capacity to encode unique N-terminal protein isoforms (Fig. S4). Identification of these novel mRNAs highlights the complexity of Utrn/UTRN loci, which differ in their splicing and transcriptional attributes.

Utrophin isoforms are independently regulated in embryonic and adult tissues
With the exception of Utrn-A 0 , developmental profiling indicates mouse Utrn transcription peaks at embryonic day 11 (Fig. 1B, Table S1) with subsequent decline concomitant with the onset of dystrophin transcription in smooth muscle (e11) and organs such as lung (e12) and brain (e13) [36,37]. During early embryogenesis (e7-11), utrophin 2A-and 1B-containing transcripts initially represent major isoforms, but become progressively less apparent when Utrn levels are lowest and the overall contribution from alternative mRNAs becomes more pronounced (e15-17; Fig. 1B). As total UTRN tissue patterning is similar to that of Utrn, we analysed the transcriptional attributes of human paralogues in embryonic and human tissue panels using qRT-PCR and sq-RT-PCR respectively (Fig. 1C, Fig. S5a, Table S1). Comparison of mRNAs that comprise total utrophin A (2A-:UTRN-A/A 0 ) or B (UTRN-B/B 0 ) indicate that their individual mRNA transcripts have similar isoform-specific, but overall distinctive, expression profiles, a phenomenon previously indistinguishable by prior studies [13,14,17,38]. Although human utrophin isoforms appear independently regulated, common regulatory attributes may also be shared, particularly with UTRN-B/B 0 where the main utrophin B 0 mRNA start site lies within exon 1B (Fig. S2). Conversely, UTRN-F is particularly distinctive in its tissue specificity, with abundant distribution in mesenchymal lymphatic tissue and highly vascularised organs (Fig. 1C). The latter observation is solidified using UCSC ENCODE DNaseI-seq data [34], where human exon 1F resides within an open chromatin region in not only embryonic stem cells, but in adult tissues; notably epithelial-derived cell lines (Fig. S6). Extending analysis to corresponding murine A and F utrophin genomic regions, DNAseI-seq data (Fig. S6-S7) indicates utrophin exons 1A,A 0 and F all reside in open chromatin regions across a broad variety of embryonic and adult tissues.
Postnatal Utrn mRNA variability in wild-type and mdx tissue We also wished to establish whether mouse utrophin transcript levels differ between age-matched wild-type and mdx tissue and thus performed isoform-specific qRT-PCR using tissue sourced at the preweaning stage (2 weeks) to determine transcript levels prior to increased activity and onset of visible muscle pathology. These data were directly compared to identical tissues isolated from 6-week old mice; after the primary 'wave' of muscle necrosis (3-4 weeks), when active regeneration, functional restoration and Utrn reexpression occurs [39,40] (Fig. 2A). Postnatal Utrn is most abundant in lung [38,41,42] total Utrn-A and Utrn-B mRNA levels agree with prior studies [38]; and low levels of Utrn-C, D and F ( Fig. 2A, Table S2) are comparable with wild-type mRNA profiles. We also assessed heart, given that cardiomyopathy represents a leading cause of DMD patient death [43]. Utrn-A and -B heart mRNAs are upregulated in mdx (1.7-fold), with total 1B-/2A-containing mRNA profiles similar to prior studies [14,17,38]. Other isoforms show comparable wild-type/mdx profiles that remain stable or decrease with age ( Fig. 2A, Table S2). However, as Utrn overexpression in diaphragm ameliorates cardiac symptoms [44], the role of individual isoforms in perpetuating milder mdx cardiomyopathy was not pursued.
In diaphragm, the sole mdx tissue with severe DMD pathophysiology ( [45]), utrophin A transcripts are markedly elevated in mdx at 6-weeks (Utrn-A; 2.9-fold and Utrn-A 0 ; 3.2-fold, Fig. 2A, Table S2), mirrored to a lesser extent by Utrn-B (1.8-fold). Conversely, Utrn-B 0 and -F upregulation in mdx at the 2 week stage is not retained. Surprisingly, elevated Utrn-F hindlimb levels in mdx at the preweaning stage (2.3-fold) almost doubles at 6-weeks (4.2-fold) representing a direct contrast to the concurrent, lower and nonsustained profile in diaphragm ( Fig. 2A). As Utrn-A 0 /F upregulation may be linked to postnatal hindlimb regeneration in mdx [19,20], we wished to determine the myogenic capacity of Utrn isoforms for comparison with their human paralogues.

Myogenic upregulation of endogenous mouse utrophin mRNA in vitro
Links between Utrn and efficient postnatal mdx regeneration are established during reversion to fetal myogenic programming and muscle precursor activation [19,20,39,46]. We therefore initially documented levels of endogenous Utrn mRNAs during myogenic differentiation in C2C12 cells [23,24] and show Utrn-A (2.5-fold) and -A 0 (2.2-fold; Fig. 2B) contribute to the established 2-fold myogenic upregulation of 2A-containing mRNAs [30,38,47]. Furthermore, Utrn-C (1.4-fold) and Utrn-F (1.7-fold) mRNAs increase during myogenesis, whereas other isoform mRNAs were barely detectable. After consideration of the tissue and myogenic properties of novel transcripts, further characterization of murine utrophin isoforms B 0 /C/D were not pursued due to the absence of (a) a human paralogue (b) an efficient myogenic response and (c) overall contribution to utrophin transcript levels in multiple tissues.

Utrophin isoforms exhibit species variability during myogenic differentiation
Although C2C12 cells represent an established myogenic model [23,24], they do not reflect the in vivo transcriptional decline of utrophin during skeletal muscle development [39,48] nor represent the absence of dystrophin. We therefore compared the transcriptional response of utrophin under wild-type and dystrophindeficient conditions using immortalised cell lines from H-2K b -tsA58 mice (H2K [26] Fig. 2C) and human skeletal muscle ( [27] Fig. 2D). Although both systems recapitulate in vivo Utrn/UTRN myogenic profiles, two marked differences between species exist in conferring this response. The first is the ratio of 2A-containing mRNAs, where Utrn-A 0 mRNAs are more abundant in H2K-mdx than UTRN-A 0 in dystrophin-deficient hDMD cells. Secondly, a reciprocal relationship exists where elevated Utrn-F and low Utrn-B mRNA levels are reversed in their human paralogues. The impeded myogenic/dystrophic response of UTRN-A 0 /-F and expression of UTRN-B over human UTRN-F may be linked to differential myogenic signalling involving, in part, E-Box-mediated mechanisms.

The molecular basis of Utrn-A 0 and Utrn-F myogenic induction
As the mouse and human utrophin A myogenic response relies on a functional CANNTG E-Box motif (CAGGTG: [30]), we sought to establish if Utrn-A 0 transcription is dependent on the Utrn-A promoter, or is regulated by the uncharacterised 1A/1A 0 interexonic genomic region (A ie ) lying between exons 1A/1A 0 ( Fig. 3 and Fig. S1). Analysis of the A ie region using UCSC ENCODE-Caltech C2C12 ChIP-seq data [35] indicates extensive methylation (H3K4me2/3) and acetylation (H3K27ac) histone H3 marks that hallmark active transcription and positively acting regulatory regions (Fig. 3C), particularly in the absence of repressive H3K27me3 histone modifications [49,50]. Combined with additional ChIP-and RNA-seq data, this observation suggests the murine A ie region is (a) recognised by myogenic factors (with particular affinity for myogenin containing complexes) specifically the E-box motif located proximal to the transcription start site of Utrn-A 0 and (b) produces mRNA levels commensurate with Utrn-1A (Fig. 3C). The corresponding human region was also assessed using UCSC ENCODE RNA-ChIP-and DNAseI-seq data [34,35] from wild-type LHCN-M2 [27] and HSMM immortalised skeletal muscle [51] cell lines (Fig. S6), indicating UTRN-1A 0 also resides in a region of open, transcriptionally active chromatin by DNAseI-and RNA-seq, the latter albeit at lower levels than UTRN-1A. Furthermore, as focal myogenic regulatory factor (myoD/myogenin) enrichment over Utrn-A/A 0 transcription start sites/E-Box motifs can be distinguished by intensity and differentiation-coupled increase, we postulated that the CAGCTG motif (a preferential hexamer for myogenin (myoG) and myoD [52]) within the mouse A ie plays an important role in Utrn-A 0 -independent myogenic responsivity (Fig. 3B).
Given the location and preliminary characterization of the A ie , numerous possibilities therefore exist in terms of its regulatory capability, including acting as an enhancer element, or driving transcription of an antisense transcript to exon 1A. Subsequent assessment of mouse and human utrophin A ie function using luciferase assays (Fig. 3A) indicates that both can function directionally to yield 1A 0 mRNAs (Fig. 3B), however, the myogenic response observed for mouse A ie (1.8fold increase; compared to 2.1-fold for Utrn-A 0 mRNA; Fig. 3A) is not recapitulated in its human, nor E-Box mutant counterpart. This observation supports the hypothesis that the murine A ie E-Box can act as a myogenic element. Furthermore, as human utrophin A 0 exhibits endogenous myogenic activity (Fig. 2D), additional genomic elements residing outside the A ie are clearly implicated in regulating this response.
Conversely, Utrn-F transcript levels in C2C12 myoblasts are lower compared to Utrn-A/-A 0 (3.2/2.5-fold higher respectively) and increase at a later myogenic stage (d9; 1.7-fold, Fig. 2B). We thus employed a more direct, electrophoretic mobility shift assay (EMSA) using C2C12 nuclear extracts to establish the functional capability of the conserved Utrn-/UTRN-F E-Box motif (Fig. 4B). Interestingly, mouse and human 1F promoter regions have different affinities for myogenic factor-containing complexes to UTRN-A [30], including myoD and myoG. It is tempting to speculate the unique UTRN-A: protein complex represents myoD homodimers, as the Utrn-/UTRN-F CAGTTG motif illustrates preference for heteromeric myoG-E12 and MyoD-E2A complexes [53]. This would account for weakened human UTRN-F probe binding in comparison to mouse (despite identical core hexamers; Fig. 4B) as heterodimerised complexes are particularly reliant on flanking nucleotides for affinity [53].

F-utrophin distribution overlaps A-and B-utrophin in mdx skeletal muscle
Similar to 2A-and 1B-containing mRNAs, exon 1F potentially encodes a unique N-terminal full-length utrophin protein (Fig. 4, Fig. S4). Given the potential of F-utrophin to contribute to regeneration/sarcolemmal stabilization of mdx skeletal muscle, we sought to   identify where resultant protein is localised, particularly as prior studies indicate that total utrophin can be accounted for by A-utrophin and -B only [17].

Utrophin mRNAs in perivascular mesenchymal stem cells parallel species-specific myogenic profiling
Given the localization of F-utrophin and its speciesspecific dystrophic response at the mRNA level appear myogenic, we also wished to determine if this phenomena could be linked to vessel-derived muscle regeneration. Vascular-derived mesenchymal precursors potentiate myogenic regeneration in diseased muscle [54,55]. In particular, the mesoangioblast (MAB) pericytic subset can restore dystrophin-deficient muscle and thus represent an attractive DMD stem cell-therapy approach ( [61], for review [62]).
Links between lineage preference and dystrophic Utrn upregulation ability were then assessed using wild-type/mdx aortic MABs. Direct comparison using cells representing different stages of pluripotency proved complex, as Utrn-A-/F transcripts were elevated; however, unlike Utrn-A, high D351 Utrn-A 0 levels did not further increase in mdx MABs isolated directly from aorta (Fig. S5c). We thus specifically compared mdx and DMD MABs sourced from skeletal muscle, which not only recapitulated differences between UTRN/Utrn transcript levels and ratios observed between dystrophin-deficient myoblasts, but were more pronounced, particularly in DMD MABs (Fig. 2C,D). This furthers our hypothesis that mouse and human loci have evolved separate, possibly myogenic, strategies in response to dystrophin deficiency.

Discussion
Utrophin is regarded as a regeneration-associated protein, as its expression is linked to effective myogenic re-programming that enables functional restoration of mdx skeletal muscle. In the mdx mouse model, postnatal Utrn re-expression is hallmarked by upregulation within the sub-sarcolemma and vascular perimysium, a compensatory mechanism that correlates with DMD disease severity but without the protective effects observed in mdx mice (reviewed in [66,67]). As dystrophin-utrophin (dko) knockout mice are a more accurate reflection of the DMD phenotype [21,22], our hypothesis was that the mouse utrophin locus possesses a transcriptional response to dystrophic conditions that differs from its human equivalent. We therefore focused on the isolation of mouse (Utrn) and human (UTRN) utrophin transcripts with novel 5 0 sequences and characterising their transcriptional and species-specific attributes. Encouragingly, the observed enhanced capability in mdx appears to be conferred by differential regulation of Utrn isoforms with human paralogues, rather than solely murine-specific mRNAs.
First, this phenomenon is manifest for utrophin A [13] and B [14] which this study identifies both as two independently regulated mRNAs (A/A 0 and B/B 0 ). The 2A-containing transcripts share myogenic attributes; however, their independency is marked by preferences for Utrn-A 0 /UTRN-A upregulation during dystrophindeficient myogenesis, and within myogenic stem cells such as MABs. Myogenic E-Box mediated mechanisms within the mouse A ie may elevate levels of extrajunctional Utrn-A 0 mRNA that can contribute to sarcolemmal A-utrophin protein in mdx. This phenomena is possibly compromised in DMD patients given the absence of an E-Box in the human A ie region. Importantly, however, both DMD myoblasts and MABs exhibit the presence of myogenically responsive UTRN-A 0 (albeit at lower levels), suggesting elements outside the A ie are likely to confer positive mediation of myogenic signalling or increase post-transcriptional stability. The role of 1B exon derived isoforms appears more complex. Although UTRN-B 0 illustrates similarities in its transcriptional profile with its shared exon 1B counterpart (UTRN-B), the lack of myogenic capability, coupled with nonconserved splicing/putative translation attributes, indicates that an UTRN-B 0 -derived strategy is unlikely to be translated to human benefit.
Secondly, we identify a novel transcript, utrophin F, which demonstrates such marked regulatory differences between mouse and humans that we immediately considered its transcriptional control as a primary utrophin candidate in modulating the mdx phenotype. Indeed, Utrn-F is the sole non-2A-containing transcript upregulated in mdx skeletal muscle, during in vitro myogenesis and within MABs to levels unparalleled by its DMD paralogue. Links between Utrn-F and enhanced mdx regeneration were provided by robust transcriptional upregulation in regenerationcompetent hindlimb (particularly soleus), that is absent in regeneration-inhibited diaphragm. Using confocal microscopy, we localised F-utrophin protein with regeneration-rich regions in mdx skeletal muscle, including immature myofibres and the interstitial endomysium, where mesenchymal-lymphocyte mediated regeneration occurs [68]. This region contains multipotent vascular-derived/mesenchymal precursors with myogenic potential such as pericytes and MABs with increased Utrn levels [65,68] and assist, incorporate or even form, myofibres (reviewed in [55]). Our hypothesis that Utrn-F is linked to these events is further bolstered by preferential transcription in mesenchymal cells that retain pluripotency, and upregulation in mdx MABs. Critically, as Utrn-F does not appear to be directly regulated via TGF-b1-inducible pathways, upregulation within the myofibre niche is unlikely to induce fibrosis or inhibit vital regenerative processes such as satellite cell proliferation, myofibre fusion or muscle-specific gene transcription, a hurdle that currently prevents direct TGF-b1 targeting strategies from entering the clinic. Isolation of Utrn-/ UTRN-F thus allows development of a novel upregulation strategy linked to the vascular-derived regeneration process to combat nonsarcolemmal DMD pathology.
Dissecting transcriptional and genomic properties of established and novel isoforms alike, we illustrate the utrophin locus displays a considerably more complex pattern of transcriptional regulation and interspecies variation than previously expected. Encouragingly, utrophin A 0 and F isoforms exhibiting desirable upregulation traits in mouse (such as the ability to respond to myogenic and regenerative stimuli) have independently regulated human paralogues, providing novel pharmacological targets to supplement promoter-based small compound approaches focussed on UTRN-A. We thus envisage that a multitargeting utrophin upregulation approach has therapeutic promise in countering sarcolemmal and nonsarcolemmal pathology within DMD skeletal muscle.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article: Fig. S1. Mouse and human sequence alignment of utrophin exon 1A and 1A 0 genomic region.      Fig. S7. UCSC DNAseI-seq information for mouse and human utrophin 1F. Fig. S8. F-utrophin N-terminal sequence and specificity of UtroF antibody. Fig. S9. Utrophin F transcript levels and protein distribution in skeletal muscle. Table S1. Utrn/UTRN qRT-PCR and sqRT-PCR values to accompany Fig. 1. Table S2. Utrn qRT-PCR values and statistics to accompany Fig. 2.