Science Spotlight

When proteins diverge to stay the same

From the Tapscott Lab, Human Biology Division

Researchers in the laboratory of Dr. Stephen Tapscott study the human transcription factor DUX4 that causes facioscapulohumeral muscular dystrophy (FSHD) when mutations lead to its inappropriate expression in skeletal muscle. Two recent publications from the Tapscott lab identify DUX4 and mouse  Dux as the factors that activate the first wave of gene expression in the early "cleavage-stage" embryo– where fertilized embryos have undergone the first round of mitotic division to become two cells. Generally, genes that serve such a fundamental developmental role would be highly conserved, but the sequence of mouse Dux and human DUX4 have significantly diverged. This led Dr. Jennifer Whiddon and colleagues to investigate which functions of this divergent gene family were conserved or unique in mice and humans, and the implications for mouse models of FSHD muscular dystrophy.

First, Dr. Whiddon showed that mouse Dux expressed in mouse muscles cells caused a dramatic change to the RNA profile that was significantly enriched in genes normally expressed in cleavage-stage embryos. When this profile of up-regulated genes was compared to the RNA profile of human muscle cells expressing human DUX4 there was a high degree of overlap. Basically finding that mouse DUX in mouse cells and human DUX4 in human cells turn on similar genes, as might be expected for orthologs that regulate an early developmental program. While this makes sense, the mechanism through which these divergent proteins regulate the same genes was unclear. Dux/DUX4 must bind to DNA through their tandem homeodomains in order to activate gene expression; however at the amino acid level the first homeodomain is only 35% identical between the mouse and human proteins and homeodomain two is 58% identical. These differences do in fact alter the DNA binding behavior of Dux and DUX4. Dr. Whiddon used chromatin immunoprecipitation (ChIP) to isolate each protein from cells and sequenced the DNA they had bound. Overall both DUX and DUX4 bound ~10 nucleotides each; the first five nucleotides differed between DUX and DUX4, while the last five nucleotides were the same between both proteins.

Mouse and human DUX4-family members use partially conserved, partially diverged consensus binding motifs to drive expression of a conserved genetic transcriptome and an entirely diverged retroposon transcriptome. Image provided by Dr. Jennifer Whiddon

The unique binding sequences are specific to the protein – not the environment. When the human protein DUX4 was expressed in mouse muscle cells it bound the same recognition sequence as it did in human cells. Next, Dr. Whiddon compared the genes mouse DUX and human DUX4 activated in mouse muscle cells. This showed that mouse Dux stimulated gene expression from both standard promoters and from repetitive gene elements (retroposons), but human DUX4 did not. Such genetic features are the remnants of retrovirus genomes that in some cases have been co-opted to drive expression of host genes. In the case of mouse cells, Dux was particularly potent at driving expression from the MERV-L class of retroposons, which was further validated using a luciferase reporter expressed by a MERV-L element. Deeper analysis of the ChIP data from DUX4 expressed in human muscle cells revealed that DUX4 also activates genes driven by retroposons; however, it binds a related, but unique class called ERVL-MaLR elements.

A simple explanation for these findings is that the 58% similar homeodomain two makes contact with the DNA that is recognized by both Dux and DUX4, while the less conserved homeodomain one allows for the unique behavior of Dux and DUX4. Thus researchers made chimeric versions of Dux and DUX4 where they replaced the mouse homeodomain one or two with the human sequence and measured gene expression of known Dux/DUX4 targets. Surprisingly, the homeodomains did not appear modular for the genes tested, “We are intrigued by the idea of homeodomain modularity. It's hard for me to imagine a scenario where only half the binding motif would be conserved between species, and each homeodomain didn't contribute a unique function. If the two homeodomains function as a pair, we would imagine selection to behave similarly across the two homeodomains and the conservation level of the binding motif to be even from 5' to 3'. The data we have so far is mixed; we are clearly still missing a piece of the puzzle” explained Dr. Whiddon.

This research was simultaneously published with an article arising from a collaboration between the Tapscott Lab and the Cairns Lab in Utah. These publications along with previous work suggest that the DUX family of genes diverged in response to the activation of different retroposon elements. “Although human DUX4 and mouse Dux both regulate the conserved early gene activation in the cleavage stage embryo, each species also uses a DUX4-family gene to drive expression of retroposons at the same developmental time, yet the retroposons are species-specific and the genes that mouse Dux activates via retroposons are mouse-specific. These findings beg the question as to the function of retroposon transcription in early development and, given the divergence in the individual retroposons involved, would we expect this function to be conserved or not? In the future, we are interested to learn whether retroposon transcription might contribute to the development of the trophoblast and placenta, the extra-embryonic tissues that arose at the same time as the double homeobox family of genes” said Dr. Whiddon.

National Institutes of Health, National Science Foundation, and Friends of FSH Research funded this work.


Whiddon JL, Langford AT, Wong CJ, Zhong JW, Tapscott SJ. 2017 Conservation and innovation in the DUX4-family gene network. Nature Genetics. Jun;49(6):935-940. Epub 2017 May 1.

Hendrickson PG, Doráis JA, Grow EJ, Whiddon JL, Lim JW, Wike CL, Weaver BD, Pflueger C, Emery BR, Wilcox AL, Nix DA, Peterson CM, Tapscott SJ, Carrell DT, Cairns BR. 2017. Conserved roles of mouse DUX and human DUX4 in activating cleavage-stage genes and MERVL/HERVL retrotransposons. Nat Genet. Jun;49(6):925-934. Epub 2017 May 1.