- Email: [email protected]
Tumor senescence as a determinant of drug response in vivo
Drug Resistance Updates 5 (2002) 204–208
Tumor senescence as a determinant of drug response in vivo Igor B. Roninson∗ Department of Molecular Genetics, University of Illinois at Chicago, Chicago, IL 60607-7170, USA Received 16 July 2002; received in revised form 3 August 2002; accepted 5 August 2002
Abstract It has become apparent in the last few years that induction of apoptosis is insufficient to account for the therapeutic effect of anticancer agents. Chemotherapy and radiation induce two other antiproliferative responses in tumor cells, cell death through mitotic catastrophe and terminal growth arrest through the program of senescence. Different types of tumor cells were found to develop the senescent phenotype upon drug treatment in vitro and in vivo. Cell culture studies demonstrated that this phenotype marks tumor cells that survive drug exposure but lose the ability to proliferate, and that such cells activate multiple growth-inhibitory genes. A recent study demonstrated that senescence, along with apoptosis, is a key determinant of in vivo response to chemotherapy in a transgenic mouse model of B-cell lymphoma. This review discusses the results of the latter study, as well as the differences between the genetic determinants of treatment-induced senescence in murine lymphoma and in human solid tumor cells. © 2002 Elsevier Science Ltd. All rights reserved. Keywords: Senescence; Apoptosis; p53; p16/Ink4A; Lymphoma; Murine tumor models
1. Background Just a few years ago, most investigators in the field of experimental therapeutics regarded apoptosis (programmed cell death) as the principal, if not the only antiproliferative response of tumor cells to chemotherapeutic drugs. More recently, this concept became a subject of debate, fueled by frequent failures to find a correlation between the apoptotic response and the overall treatment outcome in human tumors and tumor cell lines (Borst et al., 2001; Schmitt and Lowe, 2001; Brown and Wouters, 2001) The gap between the induction of apoptosis and the overall growth inhibition can be filled, however, by two other antiproliferative responses. These responses, the subject of our recent review in Drug Resistance Updates (Roninson et al., 2001) are tumor cell death by mitotic catastrophe and terminal growth arrest resulting from cell senescence. A comparison of the principal characteristics of apoptosis, mitotic catastrophe and senescence is shown in Table 1. Senescence of drug-treated or irradiated cells, in contrast to the “classical” replicative senescence, is a rapid response that is not mediated by telomere shortening. On the other hand, replicative senescence of normal cells and damage-induced senescence of tumor cells involve the same phenotypic changes and at least two common genes, p53 ∗
Tel.: +1-312-996-3486; fax: +1-312-413-8358. E-mail address: [email protected] (I.B. Roninson).
and p53-inducible cyclin-dependent kinase (CDK) inhibitor, p21Waf1/Cip1/Sdi1 . The role of senescence as the determinant of drug-induced permanent growth arrest has been conclusively demonstrated in vitro, by the analysis of separated tumor cell populations that did or did not proliferate after drug treatment (Chang et al., 1999a; Chang et al., 2002). The cells that did not die after treatment but failed to divide and form colonies displayed the senescent phenotype, i.e. enlarged and flattened shape, increased granularity and the induction of senescence-associated -galactosidase (SA–-gal), a commonly used surrogate marker of senescent cells (Dimri et al., 1995). In contrast, the non-senescent fraction looked and proliferated just like the untreated cells. The widespread occurrence of drug-induced senescence in different types of human tumor cell lines has been demonstrated by a survey of SA–-gal induction upon drug treatment in cell culture (Chang et al., 1999a). SA–-gal staining also showed that the senescent phenotype develops in human tumor xenografts grown in nude mice and treated in vivo with a retinoid (Chang et al., 1999a) or with doxorubicin (Roninson et al., 2001). Most recently, the senescent phenotype was found to be induced by chemotherapy (the CAF regimen: cyclophosphamide, doxorubicin (Adriamycin) and 5-fluorouracil) in clinical samples of breast cancer (te Poele et al., 2002). In contrast to the mechanistic studies in vitro, these staining-based in vivo surveys could not prove that drug-induced senescence is in fact, a determinant of long-term response in the treated tumor.
1368-7646/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved. PII: S 1 3 6 8 - 7 6 4 6 ( 0 2 ) 0 0 1 1 0 - 3
I.B. Roninson / Drug Resistance Updates 5 (2002) 204–208
205
206
I.B. Roninson / Drug Resistance Updates 5 (2002) 204–208
This gap has now been filled by the new study of Schmitt et al. (2002), which took advantage of the powerful tools of mouse genetics to demonstrate the role of treatment-induced senescence and to elucidate some of its determinants in a transgenic murine tumor model.
2. Senescence and apoptosis in E-myc lymphoma model This study has been carried out by the group of Scott Lowe, who had originally demonstrated the importance of the pro-apoptotic role of p53 in drug response of transformed mouse cells (Lowe et al., 1993, 1994). Schmitt et al. (2002) used E-myc transgenic mice, which overexpress the c-myc oncogene in the B-cell lineage and therefore inevitably develop B-cell lymphomas. The genotype of E-myc mice can be modified by breeding with mice that carry germ line knockout or other mutations in specific genes. E-myc lymphomas can also be transduced with retroviral vectors that overexpress any gene of interest. This model allows one, therefore, to investigate the roles of multiple genes in the development or treatment of B-lymphoma. Lowe’s group has previously demonstrated that treatment of E-myc lymphomas with various chemotherapeutic agents involves a strong and rapid apoptotic response (Schmitt et al., 1999, 2000). Both the overall treatment response and the induction of apoptosis were inhibited in this model by the knockout of p53 or by the overexpression of the anti-apoptotic gene bcl2. These effects of p53 knockout were originally interpreted as indicating that the pro-apoptotic function of p53 is responsible for its overall effect on the treatment response. In the new study, Schmitt et al. (2002) compared side-byside the effects of p53 knockout and bcl2-overexpression on the response of E-myc lymphoma, treated in vivo with a single dose of cyclophosphamide. In the short-term, bcl2-overexpression was much more potent than the loss of p53 in preventing tumor shrinkage, due to the inhibition of apoptosis. The long-term survival assays showed, however, that p53-deficient tumors were more resistant to treatment than bcl2-overexpressing tumors with wild-type p53. These findings indicated that a block of apoptosis was not enough to provide maximal protection from cyclophosphamide, and that the pro-apoptotic function of p53 could not fully account for its chemosensitizing effect. Schmitt et al. then investigated the nature of treatment response in bcl2-overexpressing tumors. Seven days after treatment, such tumors showed apparently complete cessation of DNA replication and mitosis and drastic induction of the senescence marker SA–-gal. The loss of p53, however, prevented both cell cycle arrest and SA–-gal induction. These results showed that apoptosis-resistant tumors respond to treatment through the induction of senescence, and that both apoptosis and senescence are mediated through p53. In another series of experiments, Lowe’s group analyzed the effects of the CDK inhibitor p16, one of the principal
mediators of senescence in normal cells. p16 is encoded by the Ink4a locus, where p16 overlaps with the ARF gene. The best-known function of ARF is positive regulation of p53. In an earlier study, a deletion in Ink4a that inactivates both ARF and p16 was shown to make E-myc lymphomas more resistant to treatment, which was attributed to the effect of ARF on p53 (Schmitt et al., 1999). In the new study, Schmitt et al. (2002) analyzed the responses of E-myc lymphomas deficient in either ARF alone or in both ARF and p16. Surprisingly, the knockout of ARF alone had no effect on the treatment outcome, but inactivation of both ARF and p16 made the lymphomas more resistant to treatment and prevented the induction of SA–-gal. The p16 protein was upregulated in the tumors by cyclophosphamide, and this treatment also selected for the loss of p16. The authors concluded that p16 contributes to treatment outcome in vivo through its pro-senescence activity. This effect of p16 parallels its known role in the maintenance of permanent growth arrest in normal senescent cells, but another p16-related finding was quite unexpected. p53-null E-myc lymphomas expressed very high levels of p16 relative to p53 wild-type cells, even without treatment. Furthermore, p53-null lymphomas easily tolerated infection with a p16-carrying retrovirus, but p16 could not be expressed in cells with wild-type p53. Based on these results, Schmitt et al. (2002) suggested that the antiproliferative effect of p16 is somehow dependent on p53. The results of Schmitt et al. (2002) allow one to estimate the relative contribution of apoptosis and senescence to cyclophosphamide response in E-myc lymphoma. p53-null tumors, which are resistant to both apoptosis and senescence, are also the most resistant to cyclophosphamide, followed by bcl2-overexpressing tumors (resistant to apoptosis but not to senescence), and then by p16-deficient tumors (resistant to senescence but not to apoptosis), with the wild-type tumors being the most sensitive. Both apoptosis and senescence, therefore, contribute to treatment response, but apoptosis appears to make a somewhat bigger contribution in this tumor model.
3. Genetic determinants of tumor senescence in mice and men The article of Schmitt et al. (2002) provides convincing evidence that the induction of senescence is an important determinant of tumor response to chemotherapy in vivo. The finding that treatment resistance of p16-deficient E-myc lymphomas is due to the inhibition of senescence makes it possible to compare the contribution of senescence to the response of this tumor to different classes of anticancer agents. It is important, however, to point out the essential differences between the genetic regulation of drug-induced senescence in E-myc mouse lymphoma, which comes out of the study by Lowe’s group, and the regulation of this response in human tumor cells.
I.B. Roninson / Drug Resistance Updates 5 (2002) 204–208
In the study of Schmitt et al. (2002), drug-induced senescence was utterly undetectable in lymphomas that were deficient in p53 or p16, according to the images of SA–-gal staining presented in the paper. A large majority of human tumors are deficient for either p53 or p16, and a literal translation of the conclusions from E-myc lymphoma to human cancer would suggest that treatment-induced senescence must be uncommon in human tumors. This, however, is not the case. In particular, the original survey of the effects of doxorubicin in solid-tumor derived human cell lines (Chang et al., 1999a) showed SA–-gal induction in 11 of 14 lines, including 3 of 6 lines with p53 mutations and both of the tested cell lines where p53 was inhibited by the E6 oncogene. The inhibition or knockout of p53 in HCT116 colon carcinoma and in HT1080 fibrosarcoma cells decreased drug-induced senescence several-fold, but it did not prevent the emergence of a readily detectable population of senescent cells, identified by both SA–-gal expression and growth arrest (Chang et al., 1999b). The requirement for p53 in the senescence of human tumor cells therefore is not as strict as it appears to be in E-myc lymphoma. Furthermore, HCT116 and HT1080 showed the strongest senescence response among the cell lines surveyed by Chang et al. (1999a). Both of these lines are p16-deficient, indicating that p16 is not required for senescence in human tumor cells, in contrast to E-myc lymphoma. On the other hand, te Poele et al. (2002) found a significant correlation between p16 expression and SA–-gal staining in clinical samples of treated breast cancers, suggesting that p16 may contribute to senescence in human tumors that have not inactivated this gene. In another apparent contradiction with E-myc lymphoma, the growth-inhibitory effect of p16 in human tumor cells shows no correlation with the p53 status (Craig et al., 1998). Which genes other than p53 or p16 mediate senescence in human tumor cells? Chang et al. (2002) showed that doxorubicin-induced senescence of p16-deficient HCT116 cells is associated with sustained induction of multiple growth-inhibitory genes, including several tumor suppressors. These include intracellular growth inhibitors, such as p21, BTG1, BTG2 and EPLIN, as well as secreted proteins with growth-suppressing activity, such as Maspin, MIC-1 (pTGF), and IGFBP-6. With the exception of p21, induction of these growth inhibitors in drug-treated cells was either unaffected or only mildly diminished by the knockout of p53 (Chang et al., 2002). Co-induction of these growth-inhibitory genes (and probably some others) should account for drug-induced senescence even in the absence of p16 and p53. Importantly, not all of the growth-controlling genes that are induced in senescent cells act to inhibit cell growth. A subset of senescence-associated genes encode secreted factors that promote the growth of neighboring non-senescent cells, as demonstrated by in vitro and in vivo assays (Chang et al., 2002; Chang et al., 2000; Krtolica et al., 2001). The two opposite effects of senescence need to be taken into account when considering the role of the senescence response in cancer treatment.
207
Finally, the role of p53 in the treatment response is not always as clear-cut as it appears to be in E-myc lymphoma, where the loss of p53 greatly increases tumor resistance. In human tumor cells, however, there are well-documented examples where p53 deficiency makes cells more sensitive rather than more resistant to treatment. For example, p53 knockout in HCT116 carcinoma makes these cells more resistant to 5-fluorouracyl but at the same time sensitizes them to doxorubicin and ionizing radiation (Bunz et al., 1999). Inhibition of p53 in HT1080 fibrosarcoma sensitizes these cells to doxorubicin, ionizing radiation, aphidicolin, cytarabine, cisplatin and taxol (Chang et al., 1999b). The chemosensitizing effect of p53 inhibition is primarily due to the role of p53 in G1 and G2 cell cycle checkpoints. When these checkpoints are suppressed, the treated cells enter mitosis before repairing the damage and die through the process of mitotic catastrophe (Roninson et al., 2001). Mitotic catastrophe (Table 1) is the most common cytotoxic response to all the tested chemotherapeutic drugs and radiation in human tumor cell lines (Chang et al., 1999a), but this process has not been examined in E-myc lymphoma. Thus, p53 promotes two antiproliferative responses, apoptosis and senescence, but inhibits the third response, mitotic catastrophe. Analyzing and understanding all three responses is essential for predicting treatment response in clinical cancer and for developing more efficient anticancer drugs.
Acknowledgements I thank Milos Dokmanovic for helpful discussions. The work in the author’s laboratory is supported by NIH grants R01 CA62099, R01 CA95727 and RO1 AG17921 and grant 025-01 from Mary Kay Ash Charitable Foundation. References Borst, P., Borst, J., Smets, L.A., 2001. Does resistance to apoptosis affect clinical response to antitumor drugs? Drug Resistance Updates 4, 129– 131. Brown, J.M., Wouters, B.G., 2001. Apoptosis: mediator or mode of cell killing by anticancer agents? Drug Resistance Updates 4, 135–136. Bunz, F., Hwang, P.M., Torrance, C., Waldman, T., Zhang, Y., Dillehay, L., Williams, J., Lengauer, C., Kinzler, K.W., Vogelstein, B., 1999. Disruption of p53 in human cancer cells alters the responses to therapeutic agents. J. Clin. Invest. 104, 263–269. Chang, B.D., Broude, E.V., Dokmanovic, M., Zhu, H., Ruth, A., Xuan, Y., Kandel, E.S., Lausch, E., Christov, K., Roninson, I.B., 1999a. A senescence-like phenotype distinguishes tumor cells that undergo terminal proliferation arrest after exposure to anticancer agents. Cancer Res. 59, 3761–3767. Chang, B.D., Xuan, Y., Broude, E.V., Zhu, H., Schott, B., Fang, J., Roninson, I.B., 1999b. Role of p53 and p21waf1/cip1 in senescence-like terminal proliferation arrest induced in human tumor cells by chemotherapeutic drugs. Oncogene 18, 4808–4818. Chang, B.D., Watanabe, K., Broude, E.V., Fang, J., Poole, J.C., Kalinichenko, T.V., Roninson, I.B., 2000. Effects of p21Waf1/Cip1/Sdi1 on cellular gene expression: implications for carcinogenesis, senescence,
208
I.B. Roninson / Drug Resistance Updates 5 (2002) 204–208
and age-related diseases. Proc. Natl. Acad. Sci. USA 97, 4291– 4296. Chang, B.D., Swift, M.E., Shen, M., Fang, J., Broude, E.V., Roninson, I.B., 2002. Molecular determinants of terminal growth arrest induced in tumor cells by a chemotherapeutic drug. Proc. Natl. Acad. Sci. USA 99, 389–394. Craig, C., Kim, M., Ohri, E., Wersto, R., Katayose, D., Li, Z., Choi, Y.H., Mudahar, B., Srivastava, S., Seth, P., Cowan, K., 1998. Effects of adenovirus-mediated p16INK4A expression on cell cycle arrest are determined by endogenous p16 and Rb status in human cancer cells. Oncogene 16, 265–272. Dimri, G.P., Lee, X., Basile, G., Acosta, M., Scott, G., Roskelley, C., Medrano, E.E., Linskens, M., Rubelj, I., Pereira-Smith, O., 1995. A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proc. Natl. Acad. Sci. USA 92, 9363–9367. Krtolica, A., Parrinello, S., Lockett, S., Desprez, P.Y., Campisi, J., 2001. Senescent fibroblasts promote epithelial cell growth and tumorigenesis: a link between cancer and aging. Proc. Natl. Acad. Sci. USA 98, 12072–12077. Lowe, S.W., Ruley, H.E., Jacks, T., Housman, D.E., 1993. p53-dependent apoptosis modulates the cytotoxicity of anticancer agents. Cell 74, 957–967.
Lowe, S.W., Bodis, S., McClatchey, A., Remington, L., Ruley, H.E., Fisher, D.E., Housman, D.E., Jacks, T., 1994. p53 status and the efficacy of cancer therapy in vivo. Science 266, 807–810. Roninson, I.B., Broude, E.V., Chang, B.D., 2001. If not apoptosis, then what? Treatment-induced senescence and mitotic catastrophe in tumor cells. Drug Resistance Updates 4, 303–313. Schmitt, C.A., McCurrach, M.E., de Stanchina, E., Wallace-Brodeur, R.R., Lowe, S.W., 1999. INK4a/ARF mutations accelerate lymphoma genesis and promote chemoresistance by disabling p53. Genes Dev. 13, 2670– 2677. Schmitt, C.A., Rosenthal, C.T., Lowe, S.W., 2000. Genetic analysis of chemoresistance in primary murine lymphomas. Nat. Med. 6, 1029– 1035. Schmitt, C.A., Lowe, S.W., 2001. Apoptosis is critical for drug response in vivo. Drug Resistance Updates 4, 132–134. Schmitt, C.A., Fridman, J.S., Yang, M., Lee, S., Baranov, E., Hoffman, R.M., Lowe, S.W., 2002. A senescence program controlled by p53 and p16INK4a contributes to the outcome of cancer therapy. Cell 109, 335–346. te Poele, R.H., Okorokov, A.L., Jardine, L., Cummings, J., Joel, S.P., 2002. DNA damage is able to induce senescence in tumor cells in vitro and in vivo. Cancer Res. 62, 1876–1883.
Tumor senescence as a determinant of drug response in vivo Igor B. Roninson∗ Department of Molecular Genetics, University of Illinois at Chicago, Chicago, IL 60607-7170, USA Received 16 July 2002; received in revised form 3 August 2002; accepted 5 August 2002
Abstract It has become apparent in the last few years that induction of apoptosis is insufficient to account for the therapeutic effect of anticancer agents. Chemotherapy and radiation induce two other antiproliferative responses in tumor cells, cell death through mitotic catastrophe and terminal growth arrest through the program of senescence. Different types of tumor cells were found to develop the senescent phenotype upon drug treatment in vitro and in vivo. Cell culture studies demonstrated that this phenotype marks tumor cells that survive drug exposure but lose the ability to proliferate, and that such cells activate multiple growth-inhibitory genes. A recent study demonstrated that senescence, along with apoptosis, is a key determinant of in vivo response to chemotherapy in a transgenic mouse model of B-cell lymphoma. This review discusses the results of the latter study, as well as the differences between the genetic determinants of treatment-induced senescence in murine lymphoma and in human solid tumor cells. © 2002 Elsevier Science Ltd. All rights reserved. Keywords: Senescence; Apoptosis; p53; p16/Ink4A; Lymphoma; Murine tumor models
1. Background Just a few years ago, most investigators in the field of experimental therapeutics regarded apoptosis (programmed cell death) as the principal, if not the only antiproliferative response of tumor cells to chemotherapeutic drugs. More recently, this concept became a subject of debate, fueled by frequent failures to find a correlation between the apoptotic response and the overall treatment outcome in human tumors and tumor cell lines (Borst et al., 2001; Schmitt and Lowe, 2001; Brown and Wouters, 2001) The gap between the induction of apoptosis and the overall growth inhibition can be filled, however, by two other antiproliferative responses. These responses, the subject of our recent review in Drug Resistance Updates (Roninson et al., 2001) are tumor cell death by mitotic catastrophe and terminal growth arrest resulting from cell senescence. A comparison of the principal characteristics of apoptosis, mitotic catastrophe and senescence is shown in Table 1. Senescence of drug-treated or irradiated cells, in contrast to the “classical” replicative senescence, is a rapid response that is not mediated by telomere shortening. On the other hand, replicative senescence of normal cells and damage-induced senescence of tumor cells involve the same phenotypic changes and at least two common genes, p53 ∗
Tel.: +1-312-996-3486; fax: +1-312-413-8358. E-mail address: [email protected] (I.B. Roninson).
and p53-inducible cyclin-dependent kinase (CDK) inhibitor, p21Waf1/Cip1/Sdi1 . The role of senescence as the determinant of drug-induced permanent growth arrest has been conclusively demonstrated in vitro, by the analysis of separated tumor cell populations that did or did not proliferate after drug treatment (Chang et al., 1999a; Chang et al., 2002). The cells that did not die after treatment but failed to divide and form colonies displayed the senescent phenotype, i.e. enlarged and flattened shape, increased granularity and the induction of senescence-associated -galactosidase (SA–-gal), a commonly used surrogate marker of senescent cells (Dimri et al., 1995). In contrast, the non-senescent fraction looked and proliferated just like the untreated cells. The widespread occurrence of drug-induced senescence in different types of human tumor cell lines has been demonstrated by a survey of SA–-gal induction upon drug treatment in cell culture (Chang et al., 1999a). SA–-gal staining also showed that the senescent phenotype develops in human tumor xenografts grown in nude mice and treated in vivo with a retinoid (Chang et al., 1999a) or with doxorubicin (Roninson et al., 2001). Most recently, the senescent phenotype was found to be induced by chemotherapy (the CAF regimen: cyclophosphamide, doxorubicin (Adriamycin) and 5-fluorouracil) in clinical samples of breast cancer (te Poele et al., 2002). In contrast to the mechanistic studies in vitro, these staining-based in vivo surveys could not prove that drug-induced senescence is in fact, a determinant of long-term response in the treated tumor.
1368-7646/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved. PII: S 1 3 6 8 - 7 6 4 6 ( 0 2 ) 0 0 1 1 0 - 3
I.B. Roninson / Drug Resistance Updates 5 (2002) 204–208
205
206
I.B. Roninson / Drug Resistance Updates 5 (2002) 204–208
This gap has now been filled by the new study of Schmitt et al. (2002), which took advantage of the powerful tools of mouse genetics to demonstrate the role of treatment-induced senescence and to elucidate some of its determinants in a transgenic murine tumor model.
2. Senescence and apoptosis in E-myc lymphoma model This study has been carried out by the group of Scott Lowe, who had originally demonstrated the importance of the pro-apoptotic role of p53 in drug response of transformed mouse cells (Lowe et al., 1993, 1994). Schmitt et al. (2002) used E-myc transgenic mice, which overexpress the c-myc oncogene in the B-cell lineage and therefore inevitably develop B-cell lymphomas. The genotype of E-myc mice can be modified by breeding with mice that carry germ line knockout or other mutations in specific genes. E-myc lymphomas can also be transduced with retroviral vectors that overexpress any gene of interest. This model allows one, therefore, to investigate the roles of multiple genes in the development or treatment of B-lymphoma. Lowe’s group has previously demonstrated that treatment of E-myc lymphomas with various chemotherapeutic agents involves a strong and rapid apoptotic response (Schmitt et al., 1999, 2000). Both the overall treatment response and the induction of apoptosis were inhibited in this model by the knockout of p53 or by the overexpression of the anti-apoptotic gene bcl2. These effects of p53 knockout were originally interpreted as indicating that the pro-apoptotic function of p53 is responsible for its overall effect on the treatment response. In the new study, Schmitt et al. (2002) compared side-byside the effects of p53 knockout and bcl2-overexpression on the response of E-myc lymphoma, treated in vivo with a single dose of cyclophosphamide. In the short-term, bcl2-overexpression was much more potent than the loss of p53 in preventing tumor shrinkage, due to the inhibition of apoptosis. The long-term survival assays showed, however, that p53-deficient tumors were more resistant to treatment than bcl2-overexpressing tumors with wild-type p53. These findings indicated that a block of apoptosis was not enough to provide maximal protection from cyclophosphamide, and that the pro-apoptotic function of p53 could not fully account for its chemosensitizing effect. Schmitt et al. then investigated the nature of treatment response in bcl2-overexpressing tumors. Seven days after treatment, such tumors showed apparently complete cessation of DNA replication and mitosis and drastic induction of the senescence marker SA–-gal. The loss of p53, however, prevented both cell cycle arrest and SA–-gal induction. These results showed that apoptosis-resistant tumors respond to treatment through the induction of senescence, and that both apoptosis and senescence are mediated through p53. In another series of experiments, Lowe’s group analyzed the effects of the CDK inhibitor p16, one of the principal
mediators of senescence in normal cells. p16 is encoded by the Ink4a locus, where p16 overlaps with the ARF gene. The best-known function of ARF is positive regulation of p53. In an earlier study, a deletion in Ink4a that inactivates both ARF and p16 was shown to make E-myc lymphomas more resistant to treatment, which was attributed to the effect of ARF on p53 (Schmitt et al., 1999). In the new study, Schmitt et al. (2002) analyzed the responses of E-myc lymphomas deficient in either ARF alone or in both ARF and p16. Surprisingly, the knockout of ARF alone had no effect on the treatment outcome, but inactivation of both ARF and p16 made the lymphomas more resistant to treatment and prevented the induction of SA–-gal. The p16 protein was upregulated in the tumors by cyclophosphamide, and this treatment also selected for the loss of p16. The authors concluded that p16 contributes to treatment outcome in vivo through its pro-senescence activity. This effect of p16 parallels its known role in the maintenance of permanent growth arrest in normal senescent cells, but another p16-related finding was quite unexpected. p53-null E-myc lymphomas expressed very high levels of p16 relative to p53 wild-type cells, even without treatment. Furthermore, p53-null lymphomas easily tolerated infection with a p16-carrying retrovirus, but p16 could not be expressed in cells with wild-type p53. Based on these results, Schmitt et al. (2002) suggested that the antiproliferative effect of p16 is somehow dependent on p53. The results of Schmitt et al. (2002) allow one to estimate the relative contribution of apoptosis and senescence to cyclophosphamide response in E-myc lymphoma. p53-null tumors, which are resistant to both apoptosis and senescence, are also the most resistant to cyclophosphamide, followed by bcl2-overexpressing tumors (resistant to apoptosis but not to senescence), and then by p16-deficient tumors (resistant to senescence but not to apoptosis), with the wild-type tumors being the most sensitive. Both apoptosis and senescence, therefore, contribute to treatment response, but apoptosis appears to make a somewhat bigger contribution in this tumor model.
3. Genetic determinants of tumor senescence in mice and men The article of Schmitt et al. (2002) provides convincing evidence that the induction of senescence is an important determinant of tumor response to chemotherapy in vivo. The finding that treatment resistance of p16-deficient E-myc lymphomas is due to the inhibition of senescence makes it possible to compare the contribution of senescence to the response of this tumor to different classes of anticancer agents. It is important, however, to point out the essential differences between the genetic regulation of drug-induced senescence in E-myc mouse lymphoma, which comes out of the study by Lowe’s group, and the regulation of this response in human tumor cells.
I.B. Roninson / Drug Resistance Updates 5 (2002) 204–208
In the study of Schmitt et al. (2002), drug-induced senescence was utterly undetectable in lymphomas that were deficient in p53 or p16, according to the images of SA–-gal staining presented in the paper. A large majority of human tumors are deficient for either p53 or p16, and a literal translation of the conclusions from E-myc lymphoma to human cancer would suggest that treatment-induced senescence must be uncommon in human tumors. This, however, is not the case. In particular, the original survey of the effects of doxorubicin in solid-tumor derived human cell lines (Chang et al., 1999a) showed SA–-gal induction in 11 of 14 lines, including 3 of 6 lines with p53 mutations and both of the tested cell lines where p53 was inhibited by the E6 oncogene. The inhibition or knockout of p53 in HCT116 colon carcinoma and in HT1080 fibrosarcoma cells decreased drug-induced senescence several-fold, but it did not prevent the emergence of a readily detectable population of senescent cells, identified by both SA–-gal expression and growth arrest (Chang et al., 1999b). The requirement for p53 in the senescence of human tumor cells therefore is not as strict as it appears to be in E-myc lymphoma. Furthermore, HCT116 and HT1080 showed the strongest senescence response among the cell lines surveyed by Chang et al. (1999a). Both of these lines are p16-deficient, indicating that p16 is not required for senescence in human tumor cells, in contrast to E-myc lymphoma. On the other hand, te Poele et al. (2002) found a significant correlation between p16 expression and SA–-gal staining in clinical samples of treated breast cancers, suggesting that p16 may contribute to senescence in human tumors that have not inactivated this gene. In another apparent contradiction with E-myc lymphoma, the growth-inhibitory effect of p16 in human tumor cells shows no correlation with the p53 status (Craig et al., 1998). Which genes other than p53 or p16 mediate senescence in human tumor cells? Chang et al. (2002) showed that doxorubicin-induced senescence of p16-deficient HCT116 cells is associated with sustained induction of multiple growth-inhibitory genes, including several tumor suppressors. These include intracellular growth inhibitors, such as p21, BTG1, BTG2 and EPLIN, as well as secreted proteins with growth-suppressing activity, such as Maspin, MIC-1 (pTGF), and IGFBP-6. With the exception of p21, induction of these growth inhibitors in drug-treated cells was either unaffected or only mildly diminished by the knockout of p53 (Chang et al., 2002). Co-induction of these growth-inhibitory genes (and probably some others) should account for drug-induced senescence even in the absence of p16 and p53. Importantly, not all of the growth-controlling genes that are induced in senescent cells act to inhibit cell growth. A subset of senescence-associated genes encode secreted factors that promote the growth of neighboring non-senescent cells, as demonstrated by in vitro and in vivo assays (Chang et al., 2002; Chang et al., 2000; Krtolica et al., 2001). The two opposite effects of senescence need to be taken into account when considering the role of the senescence response in cancer treatment.
207
Finally, the role of p53 in the treatment response is not always as clear-cut as it appears to be in E-myc lymphoma, where the loss of p53 greatly increases tumor resistance. In human tumor cells, however, there are well-documented examples where p53 deficiency makes cells more sensitive rather than more resistant to treatment. For example, p53 knockout in HCT116 carcinoma makes these cells more resistant to 5-fluorouracyl but at the same time sensitizes them to doxorubicin and ionizing radiation (Bunz et al., 1999). Inhibition of p53 in HT1080 fibrosarcoma sensitizes these cells to doxorubicin, ionizing radiation, aphidicolin, cytarabine, cisplatin and taxol (Chang et al., 1999b). The chemosensitizing effect of p53 inhibition is primarily due to the role of p53 in G1 and G2 cell cycle checkpoints. When these checkpoints are suppressed, the treated cells enter mitosis before repairing the damage and die through the process of mitotic catastrophe (Roninson et al., 2001). Mitotic catastrophe (Table 1) is the most common cytotoxic response to all the tested chemotherapeutic drugs and radiation in human tumor cell lines (Chang et al., 1999a), but this process has not been examined in E-myc lymphoma. Thus, p53 promotes two antiproliferative responses, apoptosis and senescence, but inhibits the third response, mitotic catastrophe. Analyzing and understanding all three responses is essential for predicting treatment response in clinical cancer and for developing more efficient anticancer drugs.
Acknowledgements I thank Milos Dokmanovic for helpful discussions. The work in the author’s laboratory is supported by NIH grants R01 CA62099, R01 CA95727 and RO1 AG17921 and grant 025-01 from Mary Kay Ash Charitable Foundation. References Borst, P., Borst, J., Smets, L.A., 2001. Does resistance to apoptosis affect clinical response to antitumor drugs? Drug Resistance Updates 4, 129– 131. Brown, J.M., Wouters, B.G., 2001. Apoptosis: mediator or mode of cell killing by anticancer agents? Drug Resistance Updates 4, 135–136. Bunz, F., Hwang, P.M., Torrance, C., Waldman, T., Zhang, Y., Dillehay, L., Williams, J., Lengauer, C., Kinzler, K.W., Vogelstein, B., 1999. Disruption of p53 in human cancer cells alters the responses to therapeutic agents. J. Clin. Invest. 104, 263–269. Chang, B.D., Broude, E.V., Dokmanovic, M., Zhu, H., Ruth, A., Xuan, Y., Kandel, E.S., Lausch, E., Christov, K., Roninson, I.B., 1999a. A senescence-like phenotype distinguishes tumor cells that undergo terminal proliferation arrest after exposure to anticancer agents. Cancer Res. 59, 3761–3767. Chang, B.D., Xuan, Y., Broude, E.V., Zhu, H., Schott, B., Fang, J., Roninson, I.B., 1999b. Role of p53 and p21waf1/cip1 in senescence-like terminal proliferation arrest induced in human tumor cells by chemotherapeutic drugs. Oncogene 18, 4808–4818. Chang, B.D., Watanabe, K., Broude, E.V., Fang, J., Poole, J.C., Kalinichenko, T.V., Roninson, I.B., 2000. Effects of p21Waf1/Cip1/Sdi1 on cellular gene expression: implications for carcinogenesis, senescence,
208
I.B. Roninson / Drug Resistance Updates 5 (2002) 204–208
and age-related diseases. Proc. Natl. Acad. Sci. USA 97, 4291– 4296. Chang, B.D., Swift, M.E., Shen, M., Fang, J., Broude, E.V., Roninson, I.B., 2002. Molecular determinants of terminal growth arrest induced in tumor cells by a chemotherapeutic drug. Proc. Natl. Acad. Sci. USA 99, 389–394. Craig, C., Kim, M., Ohri, E., Wersto, R., Katayose, D., Li, Z., Choi, Y.H., Mudahar, B., Srivastava, S., Seth, P., Cowan, K., 1998. Effects of adenovirus-mediated p16INK4A expression on cell cycle arrest are determined by endogenous p16 and Rb status in human cancer cells. Oncogene 16, 265–272. Dimri, G.P., Lee, X., Basile, G., Acosta, M., Scott, G., Roskelley, C., Medrano, E.E., Linskens, M., Rubelj, I., Pereira-Smith, O., 1995. A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proc. Natl. Acad. Sci. USA 92, 9363–9367. Krtolica, A., Parrinello, S., Lockett, S., Desprez, P.Y., Campisi, J., 2001. Senescent fibroblasts promote epithelial cell growth and tumorigenesis: a link between cancer and aging. Proc. Natl. Acad. Sci. USA 98, 12072–12077. Lowe, S.W., Ruley, H.E., Jacks, T., Housman, D.E., 1993. p53-dependent apoptosis modulates the cytotoxicity of anticancer agents. Cell 74, 957–967.
Lowe, S.W., Bodis, S., McClatchey, A., Remington, L., Ruley, H.E., Fisher, D.E., Housman, D.E., Jacks, T., 1994. p53 status and the efficacy of cancer therapy in vivo. Science 266, 807–810. Roninson, I.B., Broude, E.V., Chang, B.D., 2001. If not apoptosis, then what? Treatment-induced senescence and mitotic catastrophe in tumor cells. Drug Resistance Updates 4, 303–313. Schmitt, C.A., McCurrach, M.E., de Stanchina, E., Wallace-Brodeur, R.R., Lowe, S.W., 1999. INK4a/ARF mutations accelerate lymphoma genesis and promote chemoresistance by disabling p53. Genes Dev. 13, 2670– 2677. Schmitt, C.A., Rosenthal, C.T., Lowe, S.W., 2000. Genetic analysis of chemoresistance in primary murine lymphomas. Nat. Med. 6, 1029– 1035. Schmitt, C.A., Lowe, S.W., 2001. Apoptosis is critical for drug response in vivo. Drug Resistance Updates 4, 132–134. Schmitt, C.A., Fridman, J.S., Yang, M., Lee, S., Baranov, E., Hoffman, R.M., Lowe, S.W., 2002. A senescence program controlled by p53 and p16INK4a contributes to the outcome of cancer therapy. Cell 109, 335–346. te Poele, R.H., Okorokov, A.L., Jardine, L., Cummings, J., Joel, S.P., 2002. DNA damage is able to induce senescence in tumor cells in vitro and in vivo. Cancer Res. 62, 1876–1883.