To ascertain separation of the distinct

To ascertain separation of the distinct enzymatic activities found in this protein preparation following separation on the HAP column, a rigorous quantitative analysis was conducted on the nickel–agarose elution and HAP FT pools of protein. For these analyses, each pool of protein was dialyzed in the identical buffer to minimize any difference in reactions conditions that may stimulate or inhibit at eh enzymatic activities of the protein. Comparison of the Ni pool and HAP FT via western blot analysis (Fig. 5A) revealed the presence of a significant level of Artemis in each pool as expected. This result was confirmed in analysis of DNA-PK phosphorylation of each pool (Supplemental Fig. 2). As mentioned previously, the flow-through fractions from the HAP column retained minimal exonuclease activity on single-strand DNA, an activity previously attributed to the Artemis polypeptide. In an effort to directly compare the exonuclease activity collected from each pool of protein, increasing amounts of the nickel pool of protein and the HAP flow-through pool were incubated with a 5′ radiolabeled oligonucleotide and the release of the 5′ deoxynucleoside monophosphate (dNMP) monitored. Increasing concentrations of Ni pool containing Artemis resulted in increased 5′–3′ exonuclease activity. However, the Artemis containing HAP flow-through pool revealed an NMP product barely above the detection limit of our assay (1% of the input DNA) demonstrating that this protein pool does not contain 5′–3′ exonuclease activity (Fig. 5B). Importantly, there are preparation specific variations of exonuclease activity and differences observed in the analysis of column fraction versus pools. In the absence of 5′–3′ exonuclease activity in vitro, the more relevant activity of Artemis, found both in vitro and in vivo, is DNA-PK dependent hairpin-opening activity and the 5′ and 3′ overhang endonuclease activity. In an effort to ensure that the HAP FT pool of [His]6-Artemis devoid of exonuclease activity still retained its endonuclease catalytic activity, we assayed each of the two pools of protein for hairpin-opening activity. In agreement with results demonstrating the importance of Ku in forming the DNA-PK–Artemis complex, we conducted all of our endonuclease assays with purified heterotrimeric DNA-PK [14]. DNA-PK-dependent Artemis-catalyzed endonuclease activity was assessed on a hairpin substrate containing a 6 cl 2 receptor 5′ overhang and yields an endonuclease product of approximately 28 bases in length as previously reported [11]. Both the nickel pool of protein and the HAP FT pool catalyzed hairpin-opening activity in the presence of ATP and DNA-PK on this substrate (Fig. 5C, lanes 6 and 7, and lanes 10 and 11, respectively). Importantly, there is slightly more hairpin-opening activity in the HAP FT pool of protein, as evidenced by the more prominent cleavage products (Fig. 5C, lanes 10 and 11). The intensity of these products could also result from the lack of exonuclease activity in the HAP FT, which would necessarily result in less exonucleolytic removal of the 5′ label on the cleavage product. Also evident, though of reduced intensity compared to the hairpin-opening activity, are products that result from 5′ overhang cleavage activity of Artemis which is anticipated to release a short 5–7 base product. Interestingly, the 5′ dNMP product produced by exonuclease activity cleaving the 5′ label on the hairpin substrate is evident in reactions performed with the nickel pool, but is substantially reduced in reactions performed using the HAP flow-through fraction. This is evident upon reduced exposure of the gel to minimize bleed-over from the Ni-fractions (Fig. 5C bottom panel). Despite rigorous experimental precautions taken, the potential exists that our interpretation of the dNMP observed in Fig. 5B is incorrect. In order to ensure that the radiolabeled nucleotide products observed in Fig. 5B are in fact the result of exonuclease activity and not a dinucleotide product produced by an endonuclease activity, we assessed nuclease activity on a single-strand DNA substrate with a 3′ radiolabeled terminus. Increasing amounts of the nickel pool of protein resulted in a ladder of products consistent with sequential single nucleoside monophosphate removal from the 5′ terminus of the oligonucleotide (Fig. 6A). The HAP flow-through pool of protein again, contained barely detectable 5′–3′ single-strand exonuclease activity. Overexposure of these gels (Fig. 5, Fig. 6) however, reveals products indicative of low level exonuclease activity which comprises less than 1% of those observed in the Ni pool. These experiments confirm that while [His]6-Artemis fractionated over a nickel column retains significant 5′–3′ exonuclease activity, fractionation of this protein over a HAP column results in near quantitative separation of the exonuclease from the Artemis protein.