We have identified a human nuclease that specifically cleaves four-stranded DNA stabilized by G quartets (G4 DNA). This nuclease, GQN1 (G quartet nuclease 1), cuts within the single-stranded region 5 of the barrel formed by stacked G quartets. GQN1 does not cleave duplex or single-stranded DNA, Holliday junctions, or G4 RNA. Cleavage depends on DNA structure and not on flanking sequence. Activity is elevated in but not restricted to B cells, making GQN1 a strong candidate for function in immunoglobulin heavy chain class switch recombination. Identification of a mammalian nuclease that specifically cleaves G4 DNA provides further support for the notion that DNA structures stabilized by G quartets form in vivo and function in regulated recombination and genomic evolution.T he mammalian genome contains domains that are characteristically G-rich, including the immunoglobulin heavy chain (IgH) class switch regions, the rDNA, and the telomeres. These domains all have specialized properties in recombination. The IgH switch (S) regions participate in a process of regulated DNA deletion, during which one or more constant regions are excised to join the expressed variable to a new constant region (1-3). The rDNA exists as several hundred repeats, which must undergo recombination to erase mutations which would otherwise accumulate with each cell division (4). The telomeric repeats [(TTAGGG) n ] are normally established and maintained by telomerase, but there also exist alternative pathways for telomere maintenance that depend on recombination (5).DNA oligonucleotides that are G-rich can interact in vitro to form four-stranded structures, called G4 DNA (6, 7). The repeating unit of G4 DNA is a G quartet (Fig. 1A), a planar array of four guanines joined by hydrogen bonds, with a monovalent cation at the center (8). Hydrogen bonds between guanines in each quartet and stacking of hydrophobic G quartets on one another contribute to the stability of G4 DNA. Runs of three or more guanines are sufficient to allow G4 DNA formation. G quartets form readily and spontaneously in vitro, and G-rich synthetic oligonucleotides typically exist in solution as equilibrium mixtures of single strands and G4 DNA. Although G4 DNA has not been directly observed in vivo, eukaryotic cells contain enzymes that specifically recognize and attack G4 DNA, further consistent with the hypothesis that this structure occurs during normal replication, transcription, or recombination. Mammalian proteins that bind with high affinity and specificity to G4 DNA and appear to function in G-rich chromosomal domains include the major nucleolar protein, nucleolin (9, 10), and the candidate telomere binding protein, hnRNP D (9, 11). Strikingly, the eukaryotic RecQ family helicases BLM and Sgs1p preferentially unwind G4 DNA (12, 13). Both BLM (14) and Sgs1p (15) localize to the nucleolus, where the G-rich rDNA is transcribed into rRNA. Deficiency in Sgs1 causes nucleolar fragmentation and deletion of rDNA circles (15, 16), and Sgs1 participates in telomere maintenance ...