The evolution of DUX4 gene regulation and its implication for facioscapulohumeral muscular dystrophy

Abstract

Double homeobox 4 (DUX4) is an early embryonic transcription factor whose expression in the skeletal muscle causes facioscapulohumeral muscular dystrophy (FSHD). Despite decades of research, our knowledge of FSHD and DUX4 biology is incomplete, and the disease has currently no cures or targeted therapies. The unusual evolutionary origin of DUX4, its extensive epigenetic and post-transcriptional gene regulation, and various feedback regulatory loops that control its expression and function all contribute to the highly complex nature of FSHD pathogenesis. In this Minireview, I synthesize the current state of knowledge in DUX4 and FSHD biology to highlight key areas where further research is needed to better understand DUX4 regulation. I also emphasize post-transcriptional regulation of and by DUX4 via changes in RNA and protein stability that might underlie key features of FSHD pathophysiology. Finally, I discuss the various feedback loops involved in DUX4 regulation and the context-specific consequences of its expression, which could be key to developing novel therapeutic approaches to combat FSHD.

1. Introduction

Double homeobox 4 (DUX4) is best known for its role in causing facioscapulohumeral muscular dystrophy (FSHD) [1–3]. FSHD is an autosomal dominant muscle disease that often manifests in teens and young adults, and, in 1 in 5 cases, results in wheelchair dependence [3]. As the third most common muscular dystrophy, FSHD impairs the quality of life of a significant fraction of the worldwide population with an estimated 1:8,000 individuals affected [4]. The genetic defect that causes FSHD was first mapped to the DUX4 gene locus in chromosome 4q in 1990 [5]. It was another 20 years before a unified disease model was put forth confirming DUX4 as the disease-causing gene [6]. Why such a delay in identifying the molecular cause of FSHD? Simply put, DUX4 as the molecular cause of FSHD was an extraordinary claim, and it took extraordinary evidence for the FSHD community to build consensus around this model. Below, I will review the unusual biology of DUX4 and its regulation and highlight open questions for further exploration.

2. Evolutionary origins of Double homeobox 4

The homeobox gene family represents a class of ancient transcription factors that can be found in almost all eukaryotes, and play essential roles in development and body patterning [7]. Early in the evolution of metazoans, homeobox genes underwent significant expansion, likely via cis duplication and functional divergence, based on their presence in the genome as large clusters [7]. The double homeobox (DUX) proteins arose from a unique duplication event that created a second homeobox domain in an ancestral single homeobox DUX gene (sDUX) in the eutherian mammal lineage [8]. The subsequent evolutionary history of the DUX family of genes – painstakingly put together by careful phylogenetic and syntenic analyses [8–12] – is nothing short of fascinating (Figure 1).

Figure 1. Evolutionary origins of Double homeobox 4.

In the eutherian lineage, a homeodomain duplication in the ancestral sDUX gene gave rise to the double homeobox (DUX) family of genes. Among these genes, DUXC underwent tandem duplication and acquired a tandem repeat array organization. In the afrotherian, primate, and gilres lineages (the clade consisting of rodents), a retrotransposition event replaced every copy of DUXC with an intronless DUX4 or Dux gene. Perhaps uniquely in primates, the intronless DUX4 gene gained introns in its 3’ untranslated region (UTR), rendering the resultant transcript a target for the RNA quality control mechanism, nonsense-mediated RNA decay.

Many homologs of DUX genes exist in placental mammals including DUXA, DUXB, Duxbl, and DUXC, all of which contain introns within the DUX open reading frame [9,10]. Of these genes, DUXC has a unique genomic organization and is found in highly repetitive tandem arrays in the subtelomeric or pericentromeric chromosomal regions [10]. The lack of synteny among the DUXC arrays across various placental mammal suggests that their localization to such chromosomal regions is likely due to convergent evolution, though the underlying selective pressures are poorly understood [8]. Phylogenetic analyses show that the intronless DUX4 and Dux genes were derived from DUXC through independent retrotransposition events that occurred in Afrotheria, Primates, and rodents [8–10]. The retrotransposon-derived DUX4 and Dux subsequently swept through the whole DUXC tandem repeat array and replaced every copy of DUXC [10]. In fact, every species that contains the retrotransposed DUX gene has lost the ancestral DUXC gene [10]. We know now that the tandem repeat array organization of DUX4 is intricately linked to how and why their expression is (mis)regulated during FSHD [2,13].

A frequent fate of retrotransposon-derived genes is to accumulate mutations and become pseudogenized – but not in the case of DUX4 and Dux, which show a high degree of functional conservation. In 2017, three independent studies discovered that DUX4 and Dux are expressed in the cleavage-stage embryos of human and mouse, respectively, and are involved in zygotic genome activation [14–16]. Despite carrying out orthologous functions, DUX4 and Dux have diverged in the DNA sequence motif recognized by their homeodomains [14,17–19]. While the second homeodomain facilitates binding to a highly conserved set of cleavage-stage genes in both mice and humans, the first homeodomain appears to regulate a rapidly evolving set of retroviral elements that drive repeat expression, showing signs of coevolution [14,19]. Interestingly, DUX4, but not Dux, subsequently gained introns in its 3’ untranslated region (UTR). This gain of introns in the 3’ UTR confers a unique layer of regulation to the DUX4 transcripts, namely destabilization via the highly conserved RNA quality control pathway, nonsense-mediated RNA decay (NMD) [20].

Several unanswered questions remain regarding DUX4 evolution that could impact our understanding of its role in FSHD. Does the intron containing DUXC play a role in early embryonic development like its retrotransposed counterparts DUX4 and Dux? A recent analysis of canine DUXC shows that this ancestral gene indeed produces an intact double homeodomain protein that induces a pluripotent program similar to human DUX4 [21]. Is the tandem repeat organization of DUXC/DUX4/Dux genes important for their biological function? Given the critical role of epigenetic regulation in DUX4 expression, the repeat organization is likely key, although this remains to be experimentally confirmed. Finally, is the gain of 3’ UTR introns and the resultant susceptibility to NMD unique to human (or primate) DUX4? A recent study found that mouse Dux is indeed a target of NMD although it lacks 3’ UTR introns, by virtue of having a long 3’ UTR [22] – a feature known to trigger NMD, albeit weakly [23]. Thus, the targeting of DUX4/Dux transcripts by NMD likely evolved in both lineages through convergent evolution. Better understanding of the various evolutionary pressures that may have shaped DUX4 evolution could hold important clues to modulating DUX4 activity.

3. Epigenetic regulation of DUX4 gene locus

Early studies identified the FSHD-causing genetic defect as the contraction of a 3.3 kb macrosatellite repeat array called D4Z4 on chromosome 4q35 (Figure 2A), from 11-100 copies in healthy individuals to 1-10 copies in FSHD-affected individuals [11,24,25]. Two hypotheses were put forth to explain how D4Z4 contraction could cause disease – either the deletion caused the loss of a gene important for muscle function, or the contraction affected the chromatin structure and had a resultant position effect on an FSHD gene within or near the D4Z4 repeats [26,27]. The autosomal dominant pattern of FSHD inheritance argued against simple deletion of a gene. Furthermore, a complete loss of the D4Z4 region did not cause FSHD [28], proving that the disease is not the result of haploinsufficiency of a gene product. The presence of the DUX4 open reading frame with coding potential within each D4Z4 repeat [29] and its evolutionary conservation [27] strongly suggested a function for this gene. Yet, the difficulty in detecting DUX4 expression in FSHD muscle caused much skepticism in the gain-of-function model of DUX4 causing FSHD. The presence of a nearly identical D4Z4 repeat array containing DUX4 on chromosome 10q [12], contractions in which do not cause FSHD, also contributed to this skepticism. Several genes in the vicinity of D4Z4 were considered as potential candidates, but none could be definitively linked to FSHD through biochemical or genetic studies [30]. one such candidate was a close homolog of DUX4 called DUX4c, produced from a single inverted D4Z4 unit distal to the repeat array and lacking a portion of the C-terminal domain. However, DUX4c was later shown to not be necessary to develop FSHD [31].

Figure 2. Epigenetic regulation of DUX4-containing D4Z4 macrosatellite repeats.

A) The subtelomeric region of chromosome 4 contains the D4Z4 repeat array. Each unit of the D4Z4 repeat encodes the DUX4 open reading frame. Distal to the last repeat, the “pLAM” sequence provides the signal for polyadenylation (PAS; Polyadenylation site) and is followed by beta satellite repeats. Upstream of the D4Z4 array are two DUX4 myogenic enhancer elements, DME1 and DME2. B) Epigenetic derepression of the D4Z4 repeats in the context of the 4qA haplotype causes FSHD.

It took a confluence of irrefutable genetic evidence from the study of several FSHD families to establish that DUX4 is indeed the FSHD-causing gene [6]. For example, a study found that contraction of the D4Z4 repeats only caused FSHD if it occurred in the context of a “permissive” 4qA variant, but not the 4qB variant (Figure 2B) [32]. The 4qA variant differs from 4qB in containing a 260 bp sequence (called pLAM) as well as a beta-satellite repeat distal to the last D4Z4 unit (Figure 2A) [32,33], which was shown to be essential for the production of stable, polyadenylated DUX4 transcripts [34]. Yet, this finding did not explain why D4Z4 contractions in the 10qA chromosomal context are nonpathogenic.

Another large genetic study of FSHD-affected individuals revealed that further polymorphisms in 4qA that are not found in 10qA are essential for FSHD pathogenesis, though their functional contribution to FSHD was unclear [35]. Finally, a landmark study in 2010 characterized the DUX4 transcripts produced from several permissive and non-permissive D4Z4 repeat arrays and found that a polymorphism in the 4qA haplotype of FSHD-affected individuals conferred efficient polyadenylation, while non-permissive haplotypes did not [6], ending the debate on whether DUX4 was really the causative gene in FSHD. Evidence of DNA hypomethylation [36] and the loss of repressive histone marks and HP1γ/Cohesin binding [37] in the contracted D4Z4 array solidified the idea that the gain of DUX4 expression due to derepression of the D4Z4 repeat region causes FSHD [38].

While most individuals affected by FSHD have a contraction of the D4Z4 repeat array, a subset of affected individuals (<5%) do not display this contraction. Yet, they display marked hypomethylation of the normally heterochromatic D4Z4 region [39]. Further analysis of such individuals led to the discovery of epigenetic modifiers including SMCHD1 [39], DNMT3B [40], and LRIF1 [41], mutations in which cause FSHD in the context of a permissive D4Z4 haplotype. While such contraction-independent FSHD cases were initially called FSHD2, we now understand FSHD as a spectrum of disease conditions that are modified by several variables including the D4Z4 repeat number, mutations in various epigenetic factors, and other polymorphisms [2,42,43].

4. Post-transcriptional regulation of DUX4 transcript isoforms

Extensive characterization of D4Z4-derived transcripts has led to the identification of many DUX4 transcript isoforms (Figure 3). All functional DUX4 transcripts are thought to be derived from the last D4Z4 unit through splicing to exons distal to the repeat that provide polyadenylation signals [34,44,45]. In FSHD cells, the capped and polyadenylated transcripts contain the full DUX4 open reading frame (termed DUX4-fl) with two introns in the 3’ UTR [34,46]. The first intron is alternatively spliced, while the second intron is constitutively spliced [34]. Healthy myoblasts, on the other hand, utilize a cryptic splice site within the first exon to produce an N-terminal fragment of DUX4 (termed DUX4 short, or DUX4-s) as well as two introns in the 3’ UTR [47]. While DUX4-s is found at very low levels in a variety of healthy somatic tissues including skeletal muscle, ovary, heart, and liver [47], DUX4-fl is found in the testis of healthy individuals albeit with a different set of exons in the 3’ UTR [34,47], as well as in the thymus [48]. It is currently unknown whether there is aberrant expression of DUX4 in tissues other than the skeletal muscle in FSHD-affected individuals. Interestingly, a transgenic mouse model of FSHD that mimics the epigenetic regulation of contracted D4Z4 arrays (D4Z4-2.5; [49]) shows expression of DUX4 in many non-muscle somatic tissues, suggesting that DUX4 expression may not be limited to muscle contexts when the D4Z4 locus is derepressed. Finally, it is unclear how loss of epigenetic silencing of the DUX4 locus causes a switch from the short to long isoform of DUX4, and how different 3’ UTR exons influence stability or other aspects of post-transcriptional regulation.

Figure 3 |. Post-transcriptional regulation of DUX4 transcript isoforms.

The last D4Z4 repeat generates DUX4 transcripts that utilize distal polyadenylation sequences (PAS) to generate mature transcripts that are spliced in various configurations in different expression contexts. (*) indicates the location of translation stop codon. Two protein products result from these transcripts – a short DUX4-s protein that lacks the C-terminal domain and the full-length DUX4-fl containing the entire coding sequence. All DUX4 transcript isoforms are natural targets of the RNA quality control pathway, nonsense-mediated RNA decay (NMD).

The presence of introns in the 3’ UTR of essentially every DUX4 isoform raises another conundrum. These introns, upon splicing, result in the deposition of exon-junction complexes in the 3’ UTR and induce rapid degradation of transcripts via the RNA quality control pathway, nonsense-mediated RNA decay (NMD) [23]. How can an unstable transcript produce sufficient DUX4 protein to carry out its various functions? In FSHD cells, DUX4 can inhibit NMD in order to amplify its own transcript level [20]. Whether a similar mechanism exists in the germline or the early embryo is unknown. In fact, the study that reported mouse Dux to be an NMD target invoked NMD-mediated degradation as the mechanism by which Dux mRNA is cleared after the 2-cell stage of embryogenesis [22], indirectly suggesting that either Dux or DUX4 in the context of the early embryo may not inhibit NMD. In that case, what is unique about the muscle context of DUX4 expression that allows NMD inhibition? Further characterization of NMD inhibition by DUX4/Dux in the context of muscle versus the early embryo is needed to shed light on this important aspect of DUX4 regulation.

5. Post-transcriptional gene regulation induced by DUX4 expression

In addition to being subjected to post-transcriptional regulation itself, DUX4 also indirectly orchestrates a multitude of post-transcriptional changes to other genes and pathways. This aspect of DUX4 biology has long been underappreciated since DUX4 is a canonical transcription factor and most studies on DUX4 have rightfully focused on transcriptional changes mediated by its expression [18,46,50].

A key example of post-transcriptional regulation by DUX4 is the stabilization of hundreds of aberrant transcripts normally degraded by the NMD pathway due to DUX4-mediated NMD inhibition [20]. In healthy cells, NMD eliminates transcripts containing premature translation termination codons (PTCs) to prevent the formation of C-terminally truncated proteins [23]. Upon DUX4 expression, hundreds of truncated proteins are translated that could theoretically account for many of the hallmarks of FSHD pathology including cell stress, protein aggregation, proteasome inhibition, and the eventual cell death [51,52]. In addition to degrading aberrant, PTC-containing transcripts, NMD also regulates the expression of many normal endogenous mRNAs [23]. For example, endogenous mRNAs that contain upstream open reading frames, unusually long 3’ UTRs, and introns downstream of the normal termination codon are also recognized and degraded by NMD [23]. Many such “regulated” NMD substrates are stress response factors that regulate homeostatic pathways including oxidative stress response [53] and UPR [54], which are known to become dysregulated in DUX4-expressing cells [55–57]. Thus, studying NMD and its precise contribution to DUX4-induced cytotoxicity is of potentially high significance to our understanding of FSHD.

Interestingly, the mechanism by which DUX4 inhibits NMD involves another post-transcriptional gene regulatory mechanism, namely protein stability. Core components of the NMD pathway, including UPF1, XRN1, and possibly SMG6 and UPF3B, show reduced protein abundance with no observable change in transcript abundance upon DUX4 expression [20,55]. At least in the case of UPF1, addition of a small molecule inhibitor of the ubiquitin-proteasome system, MG132, restores its protein level in the presence of DUX4 [20]. The ubiquitin-proteasome system carries out the important function of protein degradation in all cells [58]. Selectivity of the proteasome to specific substrates is conferred by E3 ubiquitin ligases, which recognize specific target proteins and catalyze ubiquitination through a series of enzymatic reactions mediated by E1 activating enzymes, E2 conjugating enzymes and other cofactors [58]. It is key to note that DUX4 upregulates the expression of about ~20 E3 ligases in addition to a small number of E2 conjugating enzymes, deubiquitinases, and other cofactors [46,50,55]. DUX4 expression also results in the downregulation of a similar number of E1, E2, E3 enzymes, and cofactors [46,50,55]. In short, DUX4 appears to change the regulatory landscape of protein degradation in cells, resulting in discordant RNA and protein levels for a large number of genes [55]. Thus, proteasomal rewiring and the resultant change to the proteome is another significant and understudied aspect of DUX4 biology.

Finally, multiple studies have observed aggregation and/or mislocalization of proteins upon DUX4 expression, which could have wide ranging consequences to cellular homeostasis. Examples include TDP-43 [52,59], FUS [59], and EIF4A3 [60] – key RNA-binding proteins – found to aggregate in DUX4-expressing muscle cells. Multiple mechanisms potentially underlie this phenomenon, including altered protein ubiquitination [52], production of double stranded RNA that sequester RNA-binding proteins [60], and the lack of turnover of hundreds of NMD targets that exist in complex with essential RNA binding proteins [20,51].

Taken together, the many post-transcriptional changes induced by DUX4 to gene expression is likely to have a cascading effect on cellular homeostasis, and ultimately contribute to FSHD pathology. Given that transcription factors are some of the hardest to develop specific inhibitors for, dissecting the precise contribution of each of these post-transcriptional pathways to DUX4 toxicity may yield other druggable targets for FSHD therapeutic development.

6. Feedback regulation of DUX4 expression

In the early embryo, the expression of DUX4 is precisely timed: DUX4 spikes during the transition from the 2-cell to 4-cell embryo and then disappears, suggesting mechanisms that tightly regulate its expression [15,16]. In FSHD myoblasts, DUX4 expression is sporadic, with only 1 in 1000 cells showing DUX4 [47,61]. This expression pattern is consistent with strong epigenetic repression of the D4Z4 locus with stochastic activation in a subset of nuclei. In fact, a locus specific proteomics study identified the association of epigenetic repressors NuRD and CAF-1 to endogenous D4Z4 repeats [62], adding to the long list of ways in which DUX4 is silenced [2].

Given the strong epigenetic repression of D4Z4, how is DUX4 expression activated in the cleavage-stage embryo? A recent study found that DNA damage response in the early embryo activates p53, which subsequently binds to the D4Z4 locus and induces DUX4 transcription [63]. Given the long-standing controversy around the involvement of p53 in DUX4-mediated muscle cell death [64,65], it is possible that p53 signaling potentiates endogenous DUX4 expression in models of inducible DUX4 expression, thereby contributing to DUX4-induced pathology, while not inherently necessary for DUX4-induced cell death.

Once activated, DUX4 has many mechanisms in place to sustain its expression. For example, DUX4 induces the expression of MBD3L family proteins, which can antagonize NuRD, thus constituting a positive feedback loop that further activates DUX4 expression [62]. The inhibition of NMD by DUX4 also generates a feedback loop [20]. NMD normally serves to limit DUX4 transcripts by targeting them for degradation. By inducing proteasomal degradation of the NMD factor, UPF1, DUX4 inhibits NMD [20,55]. This mutual inhibition of DUX4 and the NMD pathway constitutes a double negative feedback loop that also contributes to the amplification of DUX4. Next, many genes induced by DUX4 are themselves transcription factors (e.g., DUXA, LEUTX) and might further drive the observed DUX4 gene expression program [66]. Similarly, the DUX4-induced histone genes, H3.X and H3.Y, incorporate within the nucleosomes of DUX4 target genes, making them more sensitive to subsequent activation [67]. Such positive feedforward loops likely play a key role in the perdurance of the DUX4-activated gene expression program and perhaps in sustaining toxicity even after DUX4 expression ceases [68].

Consistent with the above ideas, when FSHD myoblasts are differentiated to form myotubes, stochastic activation of DUX4 in one nucleus appears to spread to adjacent nuclei in the shared syncytium [68]. This mode of spreading and amplification through various feedback and feedforward loops could lead to high DUX4 levels in a subset of myotubes and underlie toxicity [61,68]. Interfering with DUX4 amplification could be a potential therapeutic avenue if we could better understand and target the underlying mechanisms. An important consideration in this goal is to determine how DUX4 is silenced after it plays its role in the cleavage-stage embryo. DUX4 gene regulation resembles a bistable system with both positive and negative feedback loops in place to switch between the “ON” and “OFF” states [69]. FSHD reveals the inadvertent consequence of aberrantly switching “ON” DUX4 expression in the wrong context. The solution to this problem is likely to come from better understanding of how DUX4 is kept “OFF” in all the right contexts in healthy systems.

7. Context-specific consequences of DUX4 expression

Full-length DUX4 is expressed in both the cleavage-stage embryo and in FSHD muscle, triggering a highly similar transcriptional program but vastly different downstream consequences. In the embryo, DUX4 serves to activate the zygotic genome and contributes to the start of life [15,16], while in FSHD muscle, it causes toxicity and cell death [50,64,64,65,68]. The mechanism by which DUX4-expressing muscle cells die has been studied extensively (reviewed in [1,2,70]), and every homeostatic pathway examined to date has been shown to be dysregulated in cells succumbing to DUX4 toxicity. How does the embryo avoid this fate?

One possibility is that the transient nature of DUX4 expression in the early embryo is compatible with life while sustained expression in the FSHD muscle is not. Supporting this idea, a short pulse of DUX4 expression was indeed shown to be not toxic in muscle cells [67]. One caveat to this observation is that it is unknown whether the strength and duration of the DUX4 pulse in muscle cell culture is equivalent to what naturally occurs in the embryo – though this is arguably a difficult point to address. Another possibility is that DUX4 expression in a proliferative cell environment (i.e., the early embryo) is innocuous, but its expression in the post-mitotic muscle cell is incompatible with cell survival. The observation that lymphoblastoid cells (a highly proliferative cell type) isolated from FSHD-affected individuals express robust levels of DUX4 constitutively and can be stably maintained in culture supports this hypothesis [71]. Or there could be minor differences in the gene expression activated by DUX4 in the embryo versus muscle due to distinct epigenetic landscapes and such differences might underlie the distinct fates of these cells. Finally, although DUX4 expression in most cell types shows cell-autonomous toxicity, involvement of the immune system is also a possibility in causing muscle cell death. Whatever the reason, the natural contrast provided by DUX4 expression in the embryo versus muscle is likely to be a fruitful area of research that could hold important clues to better understand FSHD pathogenesis.

An open question that remains is why FSHD is a muscle disease while the mutations that cause FSHD are present in every cell of the body. It is possible that D4Z4 derepression only causes DUX4 expression in the muscle. In fact, the expression of SMCHD1, an epigenetic repressor of the D4Z4 region, was found to go down during muscle differentiation, accompanied by an increase in DUX4 level [72]. The presence of two DUX4 myogenic enhancer sequences (DME1 and DME2; Figure 2A) proximal to the D4Z4 array with binding sites for various myogenic transcription factors supports this hypothesis [63,73]. However, the robust DUX4-fl expression observed in lymphoblastoid cells isolated from FSHD-affected individuals [71] suggests that DUX4 expression may not be exclusive to the muscle. An alternative explanation is that DUX4 expression in the muscle is particularly toxic compared to other tissue contexts. There are many precedents for a germline mutation causing tissue-specific disease. For example, many mutations that induce misregulation of mRNA metabolism underly neuromuscular diseases including myotonic dystrophy and oculopharyngeal muscular dystrophy [74]. These observations suggest that muscle, or perhaps even certain subgroups of muscle, could be uniquely susceptible to different toxic RNAs or RNA-binding proteins and the resultant defects in RNA metabolism. Given that the NMD inhibition caused by DUX4 generates many truncated RNA binding proteins [51], it is tempting to speculate that a common thread of mRNA misprocessing and disrupted mRNA metabolism connects FSHD with other muscle diseases. A better understanding of RNA toxicity in FSHD could aid therapeutic discovery and provide an opportunity for cross-application of treatment solutions to other muscle diseases.

8. Conclusions and Perspectives.

Many therapeutic approaches targeting DUX4 expression and function, such as DUX4 gene silencing and small molecule mediated DUX4 inhibition, are currently being explored [2,3]. Yet, both significant barriers and unexplored opportunities are revealed by the complicated biology of DUX4 gene regulation. For example, DUX4 protein expression is highly sporadic and transient, with only a subset of nuclei containing DUX4 protein at any given time in FSHD myotube cultures [61], and cells exposed to DUX4 may succumb to toxicity even after DUX4 expression ceases [66,68]. If we can better understand the many layers of regulation that ensure that DUX4 is silenced in the somatic tissues of healthy individuals, we may uncover novel therapeutic avenues to silence DUX4 in the FSHD muscle. Similarly, the mechanism by which DUX4 is turned off in the early embryo could also offer clues for DUX4 silencing in the muscle. Once DUX4 is activated, many feedback and feedforward loops contribute to its amplification and perdurance within the muscle. Perturbing such gene regulatory circuits may offer a window of opportunity to limit DUX4 expression and interfere with FSHD progression. The post-transcriptional gene regulatory mechanisms set in motion by DUX4 including misregulation of RNA stability and protein turnover also offer alternative druggable targets that can be modulated to circumvent various pathological consequences of DUX4 expression. In summary, continued investigation into the basic biology of DUX4 evolution, regulation at the epigenetic, transcriptional, and post-transcriptional levels, and the mechanism by which it orchestrates context-specific consequences will greatly benefit ongoing efforts to develop effective therapies for FSHD.

Highlights.

• FSHD is caused by the inappropriate expression of DUX4 in skeletal muscle

• Poor understanding of FSHD pathogenesis hampers therapeutic development

• Critical consideration of all facets of gene regulation of and by DUX4 is essential for progress

ABBREVIATIONS

DUX4

Double homeobox 4

FSHD

Facioscapulohumeral muscular dystrophy

UTR

Untranslated region

ORF

Open Reading Frame

NMD

Nonsense-mediated RNA decay

DME

DUX4 myogenic enhancer