Chimeric Nucleases Stimulate Gene Targeting in Human Cells

See allHide authors and affiliations

Science  02 May 2003:
Vol. 300, Issue 5620, pp. 763
DOI: 10.1126/science.1078395

Correction of gene defects in human somatic cells by targeting as has been used in murine embryonic stem cells (1, 2) has been precluded by the low spontaneous rate of gene targeting (3). However, creation of a DNA double-stranded break (DSB) in the genomic target (DSB-GT) can stimulate homologous recombination by over 1000-fold (4).

We can rapidly and quantitatively measure gene targeting by correcting a mutation in a green fluorescent protein (GFP) gene that has been stably integrated into the genome (5) (fig. S1). With an optimized GFP gene targeting system, the introduction of a DSB by I–Sce I (Sce) stimulated GT >40,000-fold and the absolute rate of gene targeting reached 3 to 5% (fig. S2). Such a system, however, depends on the prior introduction of a Sce binding site into the target gene and cannot be used for endogenous genes. Chimeric nucleases (CNs) have the potential to create sequence-specific DSBs (6). CNs—fusions between zinc finger binding DNA binding domains and the endonuclease domain of Fok I—can sitespecifically cleave naked DNA in vitro (6), extrachromosomal DNA in Xenopus oocytes (7), and chromosomal DNA in Drosophila (8). CNs work as dimers, and their efficiency depends on the spacing and orientation of the zinc finger binding sites with respect to the length of the amino acid linker between the DNA binding and endonuclease domains (7, 9).

QQR is an artificial zinc finger DNA binding domain that recognizes the sequence 5′-GGGGAAGAA-3′ with nanomolar affinity (10). We modified QQR chimeric nucleases (QQR-CNs) (7, 9) (Fig. 1A) and tested whether they stimulated gene targeting (Fig. 1B). The background rate of gene targeting was 0.71 events per million transfected cells (fig. S1C). QQRL18-CN stimulated gene targeting 17-fold on target QQR6 and 260-fold on target QQR8 (Fig. 1B). QQRL0-CN did not stimulate gene targeting on target QQR8, but it was as efficient as Sce in stimulating gene targeting by over 2000-fold on target QQR6 (Fig. 1B). QQRL18-CN showed some preference for an 8–base pair (bp) spacing between binding sites whereas QQRL0-CN preferred 6-bp spacing. Thus, removing the linker between the zinc spacfinger and the nuclease domains increased the activity and specificity of the fusion protein in mammalian cells. As controls, we showed that the CNs did not stimulate gene targeting if (i) they lacked a nuclear localization signal, (ii) there was a single binding site rather than an inverted repeat binding site in the target, and (iii) the cognate binding site was changed. Thus, homodimers of CNs are potent stimulators of gene targeting in human somatic cells.

Fig. 1.

Gene targeting induced bychimeric nucleases. In all experiments we co-transfected nuclease with RS2700 [a repair substrate that corrects the mutated GFP gene (fig. S1A)] into 293 cells in which the mutated GFP target had been stablyintegrated. (A) Structures of chimeric nucleases. L0, L3, and L18 represent no, 3, or 18 amino acid linkers, respectively, between the endonuclease domain and the zinc finger DNA binding domain. (B) Gene targeting with homodimers. The 6 and 8 refer to the bps separating the inverted repeat binding sites. Binding sites are adjacent to the Sce recognition site. (C) Gene targeting with heterodimers. The target GFP gene consists of inverted binding sites for the QQR and Zif268 zinc fingers separated by6 bps adjacent to a Sce site. (D) Time course of gene targeting using CNs. CMV, human cytomegalovirus early enhancer/promoter; ATG, start codon; NLS, nuclear localization signal; Fok nuclease, endonuclease domain from Fok I; Stop, in-frame termination codon; Sce Site, recognition site forSce; QQRL0, expression of QQRL0-CN; Zif, expression of Zif-CN.

We next made the chimeric nuclease Zif-CN, which consists of the three-finger, zinc finger DNA binding domain from Zif268 that binds the sequence 5′-GCGTGGGCG-3′ with subnanomolar affinity (11). QQRL0-CN or Zif-CN alone did not stimulate gene targeting on target OQR/Zif6 (Fig. 1C). Cotransfection of both CNs, however, stimulated gene targeting as efficiently as Sce (Fig. 1C). This mimics the situation for an endogenous gene, in which heterodimers of CNs with differing half-site binding specificity would need to cooperate for target gene cleavage.

The number of GFP+ cells remained stable after targeting when Sce was used (fig. S1D), but decreased after CN treatment until day 7 after transfection (Fig. 1D). At that point, the number of GFP+ cells stabilized at approximately 30% of the number at day 3. This decrease probably reflects the toxicity of continuous expression of CNs. Such toxicity might be decreased by improving the specificity of the zinc fingers by in vitro selection (11) or by placing CNs under more regulated control.

Our work establishes a basis for efficient site-specific genomic manipulation in mammalian somatic cells for experimental purposes and raises the possibility of therapeutically correcting mutations by gene targeting.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

References and Notes

Stay Connected to Science

Navigate This Article