A disease‐linked DNA repeat array rich in potential guanine quadruplexes

Abstract

Potential G‐quadruplex sequences (PQS) can form non‐B structures containing four equal‐length runs of Hoogsteen‐paired G residues. They have extremely high thermodynamic stability and form hairpin structures from four clustered G runs, bimolecular‐type structures from two clustered G runs, or tetramolecular‐type structures (e.g., at telomeres) from four G runs on different strands. D4Z4 arrays at 4q35.2, which are linked to facioscapulohumeral muscular dystrophy (FSHD), have a high density of PQS. In D4Z4's 3.3‐kb repeat unit we found seven hairpin (G3‐5N1‐7G3‐5N1‐7G3‐5 N1‐7G3‐5) PQS. Our evidence for G‐quadruplex formation from these consensus sequences and other D4Z4 oligonucleotide sequences that contain runs of only 2 G's or only two runs of 4 or 5 G's came from characteristic circular dichroism (CD), gel electrophoresis, partial resistance to PCR, and highly targeted deletion and mutation products arising spontaneously upon DNA replication in vitro. D4Z4 arrays of 3.3 ‐ 33 kb at 4q are usually pathogenic, and those of >36 kb are almost always phenotypically neutral. Given the overrepresentation of PQS in promoters of myogenic genes, it is attractive to propose that G‐quadruplexes help organize D4Z4 chromatin structure as part of special changes in D4Z4's conformation accounting for the near‐threshold effect of D4Z4 size on FSHD status. (Supported in part by NIH Grant NS048859)