Expression of wild-type and mutant ARTEMIS proteins in human cells. (A) HEK293T cells were transfected with pcDNA6 expression plasmids coding for myc-His fusion proteins of either wild-type (ART-WT) or mutant (ARM11–19) ARTEMIS. As controls, HEK293 and HEK293T cells were transfected with the empty vector. The expression of the ARTEMIS-myc-His fusion proteins, as detected by immunoblot analysis, is shown in the upper panel; SV40 large T protein expression is shown below. (B) HEK293T cells were transfected with pcDNA6 expression plasmids coding for myc-His fusion proteins of wild-type (ART-WT) and mutant (ARM11–19) ARTEMIS proteins. The subcellular localization of the myc-His fusion proteins was detected by immunostaining using a mouse anti-human myc antibody. An empty vector transfection served as a control.

In vitro nuclease assay of ARTEMIS and ARTEMIS mutants. (A) The 5′ overhang processing and hairpin-opening activities of ARM11–ARM19. A 20-bp hairpin with 6-nt 5′ overhang was used as the substrate. The asterisks indicate the positions of the radioactive label. Wild-type ARTEMIS, histidine to alanine mutants ARM11–ARM15, and aspartic acid to asparagine mutants ARM16–ARM19 were incubated with the substrate indicated in the absence and presence of DNA-PKcs. Autoradiographs of sequencing gels are shown. M1 marks the position of the hairpin-opening product if the hairpin was opened at the tip. Positions and sizes of the major hairpin-opening and overhang processing products are indicated by arrows and numbers (in nt). The exonucleolytic product by ARTEMIS (1 nt) is indicated by ‘Exo' adjacent to the size of the product. Diagrams adjacent to the arrows are depicted to show the cleavage positions in the substrate that result in the corresponding products. The dotted boxes mark the parts of the gels overexposed and shown beneath. The arrows mark the hairpin products generated by the ARM13:DNA-PKcs complex. (B) A 21-bp DNA with 15-nt 3′ overhang was used as the substrate. The asterisks indicate the positions of the radioactive label. Diagrams adjacent to the arrows are depicted to show the cleavage positions in the substrate that result in the corresponding products. M2 and M3 were generated by incubating the labelled strand of the 3′ overhang substrate with Klenow enzyme at room temperature for 30 and 60 min, respectively.

Mutagenesis of single conserved residues of the human ARTEMIS protein. The positions of the amino-acid changes of the mutant human ARTEMIS proteins (ARM11–19) are indicated; in addition, the exons of the Artemis gene as well as the amino-acid positions of known ARTEMIS domains are shown. In our analyses, the ARTEMIS protein consists of 692 amino acids with the first codon starting at nucleotide 41 of the GenBank sequence AW954867. This is in contrast to the translational start position nucleotide 62, which was assumed by Moshous et al and which leads to a protein of only 685 amino acids (Moshous et al, 2001; Ma et al, 2002). Furthermore, compared to the GenBank sequence AW954867, our cDNA exhibits a G1718T exchange leading to a leucine instead of a valine at codon 560 and a silent C to T exchange at nucleotide 1849. ARTEMIS cDNAs derived from human cell lines always contained a T at nucleotide 1718 as well as at nucleotide 1849.

DNA-PKcs assay of ARTEMIS and ARTEMIS mutants. The ARTEMIS mutants ARM11–ARM19 are indistinguishable from the wild-type ARTEMIS in the ability of being phosphorylated by DNA-PKcs. Wild-type ARTEMIS and mutant ARTEMIS ARM11–ARM19 were subjected to a DNA-PKcs phosphorylation assay in the absence and presence of exogenous 35 bp DNA. The positions of phosphorylated DNA-PKcs, phosphorylated ARTEMIS (p-ARTEMIS), and free ATP are indicated.

Model of the active center of the metallo-β-lactamase/β-CASP domain of the human ARTEMIS protein. Active site residues between ARTEMIS and metallo-β-lactamases, which have been mutated in this analysis, are modeled on the crystal structures of metallo-β-lactamases from S. maltophilia and B. cereus (Protein Data Bank entries 1SML and 1BVT, respectively). The mutated residues are conserved between ARTEMIS and metallo-β-lactamases and are shown in a ball-and-stick representation (with yellow carbons, blue nitrogen, and red oxygen atoms). The conserved clustering of histidine and aspartate residues in ARTEMIS suggests that two metal ions (magenta spheres) are positioned in the active site as has been observed in metallo-β-lactamase crystal structures. Asp37 and histidines 33, 35, 38, 115, and 319 directly coordinate the two active site metals in 1BVT1 (solid lines). Aspartic acids 17, 136, and/or 165 do not directly bind to the metals but likely orient histidines 38 and 33 for efficient metal coordination (dashed lines). A putative water molecule might complete the coordination sphere of the active site metal ions and was modeled (red sphere). Two additional coordination sites on the active site metals (green spheres) might bind the DNA phosphate backbone at the scissile bond and the attacking hydroxyl ion, as observed in the two-metal active site of the nuclease Mre11. The mutations in this analysis therefore likely perturb metal binding to ARTEMIS and correct formation of the active site, suggesting that the endonuclease activity and hairpin-opening activity but not the exonuclease activity are strongly dependent on correct formation of the active site environment.