1068 NATURE MEDICINE • VOLUM E 4 • N UM BER 9 • SEPTEM BER 1998 ARTICLES The p53 tumor suppressor protein binds to both cellular and viral proteins, which influence its biological activity. One such protein is the large E1b tumor antigen 1 (E1b58kDa) from aden- oviruses (Ads), which abrogates the ability of p53 to transacti- vate various promoters 2 . This inactivation of p53 function is believed to be the mechanism by which E1b58kDa contributes to the cell transformation process 2 . Although the p53–E1b58kDa complex occurs during infection 3 and is con- served among different serotypes 4 , there are limited data demonstrating that it has a role in virus replication. However, loss of p53 expression occurs after adenovirus infection of human cells 4,5 and an E1b58kDa deletion mutant (Onyx-015, also called dl 1520) selectively replicates in p53-defective cells 6,7 . These (and other) data indicate a plausible hypothesis is that loss of p53 function may be conducive to efficient aden- ovirus replication. However, wild-type (wt) Ad5 grows more ef- ficiently in cells expressing a wt p53 protein 5 . These studies indicate that the hypothesis may be an oversimplification. Here, we show that cells expressing wt p53, as well as p53-de- fective cells, allow adenovirus replication, but only cells ex- pressing wt p53 show evidence of virus-induced cytopathic effect. This correlates with the ability of adenovirus to induce cell death. Our data indicate that p53 plays a necessary part in mediating cellular destruction to allow a productive adenovirus infection. In contrast, p53-deficient cells are less sensitive to the cytolytic effects of adenovirus and as such raise questions about the use of E1b58kDa-deficient adenoviruses in tumor therapy 6,7 . To determine whether p53 regulates adenovirus growth, we screened a panel of human cell lines that either express a wt p53, a mutant p53 or no p53 for their ability to support the growth of wt Ad5 and the E1b58kDa-defective mutants dl 338 and dl 1520 (Onyx-015). To do this, a cytopathic effect (CPE) assay was used 5–7 . A549 (wt p53), HepG2 (wt p53), III CF/c (p53 null) and 293 control cells (expressing the E1a and E1b genes 8 ), as well as the isogenic cell lines HT1080 (wt p53) and HT1080 6TGc5 (mutant p53), were infected with wt Ad 5 (Fig. 1). We have qualitatively summarized these results with the results of another eleven cell lines (Table). These data show that wt Ad5 caused extensive CPE in all seven cell lines expressing a wt p53 protein but there was little CPE in eight of nine cell lines that have a mutant p53. When the E1b58kDa mutants were tested in the same assay, they (like wt Ad 5) also produced more CPE in cells expressing wt p53 than in cells defective in p53 (Table). These data do not sup- port the hypothesis of selective replication of E1b58kDa-defec- tive mutants in p53-defective cells 6,7 . The only obvious exception to this apparent requirement for wt p53 for efficient adenovirus replication is the C33A cell line, which contains an R273C mutation 9 ; however, this p53 (mutant) protein is conformationally normal and may retain some func- tional activity 10 . This is the cell line in which selective replication by Onyx-015 was first demonstrated 6 . RKO p53.13 cells had a low level of CPE with the E1b58kDa deficient viruses (scored as +/- in Table), possibly because of incomplete inactivation of wt p53 by the transfected dominant negative mutant. These data indicate that p53 has a determining role in allow- ing adenovirus to induce CPE and probably, therefore, in virus replication. Moreover, although the E1b58kDa protein is not re- quired for the induction of CPE, it may enhance the efficiency with which CPE is generated. The absence of CPE in mutant p53 cells may be due to an in- ability of adenovirus to replicate its DNA (ref. 5). However, some mutant p53 cell lines allow considerable virus DNA repli- cation, as well as expression of late (structural) proteins (data not shown). Such data indicate that infectious adenovirus might be produced in mutant p53 cells, despite the absence of an obvious CPE. We therefore infected several p53 deficient cell lines with wt Ad5, and 3 days later lysed the cells and collected the supernatant, which should contain any infectious aden- ovirus. We then reinfected 293 cells with aliquots of super- natant and scored wells for CPE. Results of this experiment using A549 cells and the mutant p53-expressing cells 143BTK – , III CF/c, J82 and T98G show that although CPE was not evident in any mutant p53 cell at the time of collection, all four mutant p53 cell lines produced sufficient virus to generate CPE on 293 cells (Fig. 2). Indeed, at least as much virus was collected from some of these cells as could be collected from the A549 control cells (for example, III CF/c and 143BTK – ). Thus, the defect in adenovirus-induced CPE in mutant p53 cells is not because of an absence of virus infection or growth, but instead results from a defect in the ability of these cells to die or detach from the wells (that is, in the process of CPE generation). We have shown low levels of virus DNA replication in three cell lines (143BTK – , J82 and T98G) containing p53 mutations 5 . Therefore, p53 may be required for efficient viral DNA replication. The data here, however, indicate that the production of infectious virus in these same cell lines is similar to that in cell lines containing wt p53 (Fig. 2). It seems that not all viral DNA is encapsulated into infectious viral particles 11 . To explain the p53-dependence of adenovirus-induced CPE, we next considered the possibility that CPE results from cell death. Given that wt p53 induces apoptosis in response to aden- ovirus E1a proteins 12 , the absence of virus induced CPE in mu- p53-dependent cell death/apoptosis is required for a productive adenovirus infection ANTHONY R. H ALL, BRETT R. DIX, SIMON J. O’C ARRO LL & ANTONY W. BRAITHWAITE Cell Transformation Group, Pathology Department, Dunedin School of Medicine, University of Otago, PO Box 913, Dunedin, New Zealand A.H. and B.D. contributed equally to this work. Correspondence should be addressed to A.B.; email: antony.braithwaite@stonebow.otago.ac.nz