show Abstracthide AbstractFSHD is characterized by the misexpression of DUX4 in skeletal muscle. Although DUX4 upregulation is thought to be the pathogenic cause of FSHD, DUX4 is lowly expressed in patient samples, and analysis of the consequences of DUX4 expression has largely relied on artificial overexpression. To better understand the native expression profile of DUX4 and its targets, we first performed pooled RNA-seq on a 6-day differentiation time-course in FSHD2 patient-derived primary myoblasts. We identify a set of 54 FSHD-induced genes upregulated in FSHD2 cells starting at day 2 of differentiation through the end of the time-course. Using single-cell and single-nucleus RNA-seq on FSHD2 myoblasts and day 3 and day 5 differentiated myotubes respectively, we captured, for the first time, DUX4 expressed at the single-nucleus level in a native state. We identified two populations of FSHD myotube nuclei based on low or high enrichment of DUX4 and FSHD-induced genes (FSHD-Lo and “FSHD Hi”, respectively). FSHD-Hi nuclei upregulate many cell cycle related genes with significant enrichment of E2F target genes and p53 signaling activation. In FSHD-Hi myotube nuclei, multiple DUX4 target genes are co-expressed including a set of transcription factors, such as DUXA, ZSCAN4 and LEUTX. DUXA (the DUX4 paralog) is more widely expressed than DUX4, and depletion of DUXA suppressed the expression of LEUTX and ZSCAN4 in late, but not early, differentiation. The results indicate that the DUXA can take over the role of DUX4 and maintain target gene expression. These results may provide explanation as to why it is easier to detect and monitor DUX4 target genes than DUX4 itself in patient cells and suggest a self-sustaining network of gene dysregulation that perpetuates this disease after DUX4 is no longer expressed. Overall design: Full length single-cell and single-nucleus RNA-seq using SmartSeq on the Fluidigm C1 platform for control and FSHD2 myoblast cells and myotube nuclei.