Wadia et al. (7) recently claimed that apoptin, a virally encoded protein with tumor-selective apoptosis activity, contains a concentration-dependent nuclear localization signal (NLS) that is not tumor selective as previously reported (1,2,4). Their Fig. 1B shows, however, that under all the conditions tested, nuclear import of green fluorescent protein (GFP)-apoptin 70-121 remains 200 to 300% higher in tumor cells than in primary cells. The apparent concentration dependence of apoptin's NLS is intriguing. We agree with their interpretation that presentation of multiple apoptin NLS domains by a molecular aggregate could generate more efficient nuclear trafficking than NLS exposure at a single site. This result is, however, somewhat academic, as full-length apoptin can only be harvested as a multimer from live cells, with a dissociation rate constant (k off ) that is so slow that it can hardly be measured, whereas the C-terminal fragment forms only monomers (5, 9). This could explain why Wadia et al. could measure a cooperative, concentration-dependent effect of apoptin's NLS, as they used a GFP fusion of a C-terminal fragment of apoptin lacking the multimerization domain (5). Had they used full-length apoptin, all the NLS sequences would likely have been clustered, resulting in effective nuclear import over all concentration ranges. Nevertheless, we suggest caution in using GFPapoptin fusions in functional studies. For example, fusion of GFP to full-length apoptin results in increased levels of nuclear GFP-apoptin versus wild-type apoptin in primary cells (Fig. 1A).The current hypothesis is that nuclear trafficking and tumorspecific phosphorylation of apoptin at Thr108 are essential for induction of apoptosis (3,6,8). Wadia et al. reported a failure to detect phosphorylation with radioactive labeling of GFPapoptin 70-121 ; we do not know why this is so, as phosphorylation of full-length apoptin has been thoroughly documented by the use of mass spectrometry and a phospho-specific antibody (6). Our preliminary experiments suggest that abolishing the phosphorylation site of apoptin does not significantly disrupt its nuclear import in tumor cells. In attempting to address this observation with a GFP-apoptin 70-121 T108A mutant, Wadia et al. used a construct that is not phospho-null in vivo; in our hands, as Thr108 is the last in a run of three threonine residues, the adjacent Thr107 becomes opportunistically phosphorylated instead, which yields the same phenotype. Mutation at both positions 107 and 108 is required to eliminate phosphorylation and function. As Fig. 1B indicates, phosphorylation of apoptin is required for apoptosis induced by its Cterminal death domain (3); a control experiment confirms that the apoptosis-competent fragment is phosphorylated on Thr108 in vivo (Fig. 1C).Apoptin's tumor-specific activity does not result from a single characteristic. Its multimerization behavior, its potential to be phosphorylated, its cooperativity in nuclear trafficking, and many of its other physical, chemical, and f...