The p53 tumor suppressor network

Mutations in the gene p53 occur in a wide range of human cancers and are often associated with aggressive tumor behavior and poor patient prognosis. Wild-type p53 is activated by DNA damage and various forms of oncogenic stress, inducing genes that promote cell-cycle blockade, apoptosis, senescence, differentiation and/or autophagy. In fact, activated p53 can suppress epigenetic reprogramming of differentiated cells into induced pluripotent stem cells. In addition to its cell-autonomous activities, p53 can promote the secretion of a variety of factors that influence the tissue microenvironment in a non–cell autonomous manner.

Which of these p53 activities is most relevant for the protein’s tumor suppressor role has been widely debated and is likely context-dependent.

Our lab has a long-standing interest in p53 stemming from our early observations that p53 could promote apoptosis in response to oncogenes and DNA-damaging cytotoxic drugs (8),(9). Indeed, we showed that p53 loss compromised the ability of certain cells to undergo therapy-induced apoptosis, thereby providing some of the first evidence that tumor-cell response to therapy could be dictated by cancer genotype and that p53 mutations could promote drug resistance. We also showed that p53 can be activated by oncogenes and have explored the molecular basis for this tumor-suppressive effect.   

Over the last decade, we have focused on the roles and regulation of p53 in vivo. For example, we have shown that p53 loss is required for cancer maintenance (5), (18). We are currently exploring the role of specific p53 mutants on cancer metastasis and the processes by which p53 acts to limit cellular plasticity and liver carcinogenesis. 

Figure 1. Co-suppression of RB and E2F7 promotes transformation. Figure 1. Co-suppression of RB and E2F7 promotes transformation. A) Knockdown of Rb and E2F7 alone and in combination using a tandem shRNA vector (shTan). B) Active H-ras induces senescence in MEFs, characterized by SA-bgal staining and impaired colony formation. Simultaneous silencing of Rb and E2F7 results in senescence bypass. C) Representative images of mice 20 days after injection of 10^6 MEFs. D) Quantification of tumor volumes after MEF injection. (Aksoy et al., Genes Dev, 2012)

Cellular senescence in tumor suppression and tissue pathology

The cellular senescence program is a tumor suppressive process controlled by p53 of particular interest to our group. Senescence is a stable form of proliferative arrest that acts as a potent barrier to cancer development and may contribute to various age-related diseases.

Initially defined by the phenotype of cultured fibroblasts undergoing replicative exhaustion, senescence can be triggered in many cell types by a range of cellular stresses. Our interest in the process originated from the observation that oncogenic Ras proteins could trigger senescence in primary cells through Rb and p53 dependent mechanisms, thereby preventing oncogenic transformation (16). Accordingly, we proposed that “oncogene-induced senescence” acts as a cellular failsafe mechanism to suppress tumorigenesis, a hypothesis that is now supported by numerous animal and human studies.   

Our recent efforts in studying senescence have explored mechanisms whereby oncogenic signaling can trigger a stable cell cycle arrest using a combination of genetics, cell biology, and biochemical approaches. We identified Jarid1a/b-mediated H3K4 demethylation as a contributor to silencing of retinoblastoma targeting genes in senescent cells (3). We also identified the transcription factor E2F7 as an important senescence regulator that provides a novel link between the p53 and Rb pathways (Figure 1) (1). Concomitant suppression of E2F7 and Rb by RNAi cooperate to bypass oncogenic Ras induced senescence and transformation.

In addition to cell autonomous aspects of the program that produce a stable cell cycle arrest, senescent cells secrete a set of inflammatory cytokines and other factors as part of a process termed the “senescence-associated secretory phenotype,” or SASP. Using in vivo models of hepatocellular carcinoma, we found that induction of senescence in liver tumor cells via restoration of p53 can trigger the clearance of cancer cells by cells of the innate immune system (18). In addition, we showed (Figure 2) that ablation of p53 and its associated senescence program in hepatic stellate cells in the context of chronic liver damage promotes fibrosis and cirrhosis as well as transformation of hepatocytes into hepatocellular carcinoma (7),(10). p53-mediated senescence in these stromal cells protects against malignancy by secreting factors that polarize macrophages toward an M1 state, which renders them capable of attacking senescent tumor cells, while p53-deficiency in stellate cells results in a pro-inflammatory M2 macrophage polarization that further enhances tumorigenesis. These studies underscore non-cell autonomous roles for p53 that function to protect against malignancy. Efforts are ongoing to mechanistically dissect these and other tumor-host interactions further.

Figure 2. p53 signaling through SASP modulates macrophage function. Figure 2. p53 signaling through SASP modulates macrophage function. A) and B) Cytokines and chemokines regulated by p53 (in blue) in senescent (S) hepatic stellate cells, compared with proliferating (P) cells. C) Macrophages (red) specifically target senescent hepatic stellate cells (green) (Lujambio et al., Cell, 2013).

Cancer gene discovery using mouse models and RNAi

Cancer genomes are complex and can harbor cancer-promoting “driver” mutations together with “passenger” mutations that have no biological effect. To identify genes that actively contribute to tumorigenesis, we use mosaic mouse models to filter through candidates obtained through cancer genomics research. Our method is based on the following assumptions: 1) Recurrent amplifications and deletions in human tumors are enriched for oncogenes and tumor suppressors, respectively; and 2) lesions that give rise to cancer in humans often do so in mice. This approach has led to the identification and validation of over 50 oncogenes and tumor suppressor genes over the last several years.

Figure 3. Loss of eIF5A and AMD1 cooperate in lymphoma progression. Enlarge Image Figure 3. Loss of eIF5A and AMD1 cooperate in lymphoma progression. A) Deletions of eIF5A (17p) and AMD1 (6q21) are significantly co-occur in human lymphoma. B) Survival curves for single and tandem in vivo knockdown We developed strategies to multiplex this approach by transducing tissue stem and progenitor cells with pools of complementary DNA (cDNAs) or short hairpin RNAs (shRNAs) corresponding to genes that are amplified or deleted in human tumors, and select for those constructs that promote tumorigenesis following transplantation into recipient mice. Initial efforts surveyed nearly 400 recurrently deleted genes in human hepatocellular carcinoma (HCC) and identified 12 new tumor suppressor genes (19). In a conceptually similar screen performed by Scott Powers using full-length cDNAs, we identified ten new oncogenes in human HCC, including FGF19 as a therapeutic target for existing neutralizing antibodies(14).   

Most recently, we completed a screen for novel tumor suppressor genes in lymphoma. We screened a shRNA library targeting genes deleted in human lymphomas to identify genes that, when suppressed, promote tumorigenesis in a mouse lymphoma model (15). Among the new tumor suppressors we identified were adenosylmethionine decarboxylase 1 (AMD1) and eukaryotic translation initiation factor 5A (eIF5A), two genes associated with hypusine, a unique amino acid produced as a product of polyamine metabolism through a highly conserved pathway. Through a secondary screen surveying the impact of all polyamine enzymes on tumorigenesis, we established the polyamine-hypusine axis as a new tumor suppressor network regulating apoptosis. Unexpectedly, heterozygous deletions encompassing AMD1 and eIF5A often occur together in human lymphomas and co-suppression of both genes cooperate to promote lymphomagenesis in mice. Thus, some tumor suppressor functions can be disabled through a two-step process targeting different genes acting in the same pathway (Figure 3).

Beyond identifying new activities of cancer-relevance, our approach is revealing unexpected principles about the nature and organization of cancer genes. For example, we were surprised to find that so many tumor suppressors are haploinsufficient, encode secreted proteins, or have pro-oncogenic activities in other contexts. We also did not expect recurrent amplifications and deletions to contain more than one relevant activity, yet our results imply this is the rule rather than the exception.

Indeed, in a systematic study of genes recurrently located on 8p, we showed that shRNAs targeting multiple genes in this region could be tumor promoting, and that suppression of gene combinations produced cooperative effects (17). We believe that cancer associated deletions contribute to cancer phenotypes in a manner that is distinct from single gene mutations and should be considered and studied as distinct mutational events. We hope that further efforts will facilitate the functional annotation of the genomic alterations occurring in human cancers and identify vulnerabilities these lesions create.

Identifying tumor maintenance genes

We are also using RNAi to systematically identify genotype-specific drug targets. While many investigators focus on human-derived cancer cell lines for these studies, our efforts heavily use murine model systems, owing to the defined nature of the driving genetic events and our ability to rapidly extend results to in vivo systems. By combining inducible RNAi systems and linked fluorescent reporters, many candidate tumor maintenance genes can be identified and rapidly triaged to focus on those that show the most promise. Although our efforts to combine technologies for target discovery focus on a variety of cancers, our initial successes have come largely from the study of hematopoietic malignancies.   

One area of investigation focuses on the identification of tumor maintenance genes in mouse models of refractory acute myeloid leukemia (AML). For example, by characterizing how MLL fusion proteins promote self-renewal, we identified the Myb transcription factor as specifically required for the maintenance of AMLs that are refractory to conventional chemotherapy, such that transient suppression of Myb eradicates these leukemias (20). With Christopher Vakoc (CSHL), we screened an shRNA library targeting “epigenetic” regulators and have identified several histone-modifying activities whose suppression causes the selective arrest or death of leukemic cells. These approaches identified Brd4 as a therapeutic target for acute myeloid leukemia (21). More recently, we showed that IDH2 mutant genes found in human AML cooperate with oncogenic ras or altered Flt3 to drive aggressive leukemia in mice, and that IDH2 could have anti-leukemic effects(2). Interestingly, these IDH2 mutant leukemias were also sensitive to Brd4 inhibition, which worked rapidly to eliminate the leukemic cells.

Recent efforts have expanded our target discovery efforts to solid tumors. As one example that illustrates our overall approach, we developed and used a mouse model that recapitulates the genetics and pathology of human cholangiocarcinoma (bile duct cancer) to validate the FIG-ROS fusion protein as a therapeutic target in this disease (Figure 4). Specifically, our studies produced a rapid “mosaic” model of cholangiocarcinoma and tested the impact of constitutive and conditional FIG-ROS expression in this system (12). We showed that FIG-ROS could dramatically accelerate disease onset and that subsequent inhibition of the fusion protein in vivo could have anticancer effects. With Brian Druker (OHSU), we then used these models to demonstrate efficacy of a new ROS inhibitor that is now in clinical trails against FIG-ROS mutant cancers (4). We are continuing to perform similar screening and modeling approaches to discover and validate new therapeutic targets in several tumor types. 

Figure 4.  A mouse model of intrahepatic cholangiocarcinoma used to validate FIG-ROS as a therapeutic target. Figure 4. A mouse model of intrahepatic cholangiocarcinoma used to validate FIG-ROS as a therapeutic target. Hepatoblasts are isolated from Kras mutant/p53 mutant mice and transduced with vectors expressing a gene or shRNA of interest. Cells are transplanted orthotopically into the liver where they produce a cancer mirroring the human disease (Saborowski et al., Proc Natl Acad Sci U S A, 2013).

Accelerating cancer discovery using mouse models

Genetically engineered mouse models (GEMMs) have greatly expanded our knowledge of human cancer and serve as a critical tool to identify and evaluate new treatment strategies. However, the cost and time required to generate conventional cancer GEMMs limits their use for investigating novel genetic interactions in tumor development and maintenance. To expand and expedite gene function studies in mice, we have developed a conceptually new approach, referred to as “ESC-GEMM” models (11). This methodology is based on the derivation of mouse embryonic stem cells (ESCs) carrying conditional disease-associated alleles that, in combination, generate a functional model of disease. Additionally, we have layered into these models inducible shRNA technology optimized in our laboratory over the last several years as well as powerful new genome-editing techniques to facilitate gene disruption at different stages of disease. By integrating these technologies (see Figure 5), ESC-GEMMs provide a platform to evaluate the contribution of a broad range of candidate “disease genes” in parallel or to develop large cohorts of tailored experimental animals for pre-clinical treatment studies without further intercrossing of engineered mouse strains.

Figure 5. ESC-GEMM method speeds production of complex mouse models Enlarge Image Figure 5. ESC-GEMM method speeds production of complex mouse models A) ESC-GEMM approach combines technology from different areas to produce fast and scalable mouse models of disease. B) Schematic representation of approach to generate multi-allelic ES cells carrying defined disease-sensitizing alleles. Additional genetic complexity (via shRNAs / genome editing) can be introduced during ESC culture. C) Comparison of the time for generation of complex mouse models using conventional and ESC-GEMM approach. The prototypical model consists of ES cells carrying at least four alleles: 1) one or more disease predisposing conditional alleles that can be activated by Cre recombinase (e.g. floxed Apc gene, or a “lox-stop-lox” KrasG12D); 2) a tissue-specific Cre recombinase; 3) a Cre-activatable lox-stop-lox rtTA3 allele to enable tissue specific induction of a tet-responsive transgene; and 4) a “homing cassette” that enables site-specific integration of the tet-responsive transgene (cDNA, shRNA, miRNA) downstream of the col1a1 locus using recombination mediated cassette exchange (RMCE) (6). Other iterations omit the tissue-specific Cre (enabling activation by adenovirus expressing Cre), or link the rtTA allele to a fluorescent reporter (providing lineage tracing of Cre-recombined cells). These ESC- GEMMs can be manipulated in vitro (to introduce additional genetic complexity), for instance, by recombining a genetic element into the homing cassette, and used to produce cohorts of experimental mice by blastocyst injection or tetraploid complementation (Figure 6). Thus, the ESC-GEMM approach enables the production of experimental animals in less than two months. Moreover, as additional genetic alterations are introduced in cell culture, many different genes can be assessed in parallel, without the need for large, expensive breeding colonies.

Using the above principles, we developed a series of ESC-GEMMs to explore the role of tumor suppressor genes in tumor initiation and maintenance, to validate new therapeutic targets, and to explore the on-target toxicities of target inhibition in normal tissues. In one example, we used an ESC-GEMM model of pancreatic cancer to demonstrate that suppression of the PTEN tumor suppressor can accelerate carcinoma formation, and that its re-establishment can lead to tumor regression (13). We are currently producing a range of new models integrating new alleles and genome editing tools into the process. More broadly, we hope through these efforts we can accelerate the study of gene function in vivo for cancer and other diseases. 

Figure 6. Figure 6. ESCs harboring a lox-stop-lox KrasG12D mutant allele, the collagen homing cassette, a Pdx1-cre allele, and a CAGs-lsl-rtTA-IRES Katuska allele were targeted with a GFP-linked shRNA against PTEN. Chimeric mice rapidly develop tumors on dox (to induce the Pten shRNA), and the resulting tumors suppressed PTEN in the epithelial compartment, and expressed the rtTA and shRNA Kate and GFP reporters. Dox withdrawal led to Pten restoration and GFP silencing as expected. This led to dramatic and sustained tumor regression (Saborowski et al., Genes Dev, 2014).

  1. Aksoy, O., Chicas, A., Zeng, T., Zhao, Z., McCurrach, M., Wang, X., and Lowe, S.W. (2012). The atypical E2F family member E2F7 couples the p53 and RB pathways during cellular senescence. Genes Dev 26, 1546-1557.
  2. Chen, C., Liu, Y., Lu, C., Cross, J.R., Morris, J.P.T., Shroff, A.S., Ward, P.S., Bradner, J.E., Thompson, C., and Lowe, S.W. (2013). Cancer-associated IDH2 mutants drive an acute myeloid leukemia that is susceptible to Brd4 inhibition. Genes Dev 27, 1974-1985.
  3. Chicas, A., Kapoor, A., Wang, X., Aksoy, O., Evertts, A.G., Zhang, M.Q., Garcia, B.A., Bernstein, E., and Lowe, S.W. (2012). H3K4 demethylation by Jarid1a and Jarid1b contributes to retinoblastoma-mediated gene silencing during cellular senescence. Proc Natl Acad Sci U S A.
  4. Davare, M.A., Saborowski, A., Eide, C.A., Tognon, C., Smith, R.L., Elferich, J., Agarwal, A., Tyner, J.W., Shinde, U.P., Lowe, S.W., et al. (2013). Foretinib is a potent inhibitor of oncogenic ROS1 fusion proteins. Proc Natl Acad Sci U S A 110, 19519-19524.
  5. Dickins, R.A., McJunkin, K., Hernando, E., Premsrirut, P.K., Krizhanovsky, V., Burgess, D.J., Kim, S.Y., Cordon-Cardo, C., Zender, L., Hannon, G.J., et al. (2007). Tissue-specific and reversible RNA interference in transgenic mice. Nat Genet 39, 914-921.
  6. Dow, L.E., Premsrirut, P.K., Zuber, J., Fellmann, C., McJunkin, K., Miething, C., Park, Y., Dickins, R.A., Hannon, G.J., and Lowe, S.W. (2012). A pipeline for the generation of shRNA transgenic mice. Nat Protoc 7, 374-393.
  7. Krizhanovsky, V., Yon, M., Dickins, R.A., Hearn, S., Simon, J., Miething, C., Yee, H., Zender, L., and Lowe, S.W. (2008). Senescence of activated stellate cells limits liver fibrosis. Cell 134, 657-667.
  8. Lowe, S.W., Ruley, H.E., Jacks, T., and Housman, D.E. (1993). p53-dependent apoptosis modulates the cytotoxicity of anticancer agents. Cell 74, 957-967.
  9. Lowe, S.W., Schmitt, E.M., Smith, S.W., Osborne, B.A., and Jacks, T. (1993). p53 is required for radiation-induced apoptosis in mouse thymocytes. Nature 362, 847-849.
  10. Lujambio, A., Akkari, L., Simon, J., Grace, D., Tschaharganeh, D.F., Bolden, J.E., Zhao, Z., Thapar, V., Joyce, J.A., Krizhanovsky, V., et al. (2013). Non-Cell-Autonomous Tumor Suppression by p53. Cell 153, 449-460.
  11. Premsrirut, P.K., Dow, L.E., Kim, S.Y., Camiolo, M., Malone, C.D., Miething, C., Scuoppo, C., Zuber, J., Dickins, R.A., Kogan, S.C., et al. (2011). A rapid and scalable system for studying gene function in mice using conditional RNA interference. Cell 145, 145-158.
  12. Saborowski, A., Saborowski, M., Davare, M.A., Druker, B.J., Klimstra, D.S., and Lowe, S.W. (2013). Mouse model of intrahepatic cholangiocarcinoma validates FIG-ROS as a potent fusion oncogene and therapeutic target. Proc Natl Acad Sci U S A 110, 19513-19518.
  13. Saborowski, M., Saborowski, A., Morris, J.P.t., Bosbach, B., Dow, L.E., Pelletier, J., Klimstra, D.S., and Lowe, S.W. (2014). A modular and flexible ESC-based mouse model of pancreatic cancer. Genes Dev 28, 85-97.
  14. Sawey, E.T., Chanrion, M., Cai, C., Wu, G., Zhang, J., Zender, L., Zhao, A., Busuttil, R.W., Yee, H., Stein, L., et al. (2011). Identification of a therapeutic strategy targeting amplified FGF19 in liver cancer by Oncogenomic screening. Cancer Cell 19, 347-358.
  15. Scuoppo, C., Miething, C., Lindqvist, L., Reyes, J., Ruse, C., Appelmann, I., Yoon, S., Krasnitz, A., Teruya-Feldstein, J., Pappin, D., et al. (2012). A tumour suppressor network relying on the polyamine-hypusine axis. Nature 487, 244-248.
  16. Serrano, M., Lin, A.W., McCurrach, M.E., Beach, D., and Lowe, S.W. (1997). Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell 88, 593-602.
  17. Xue, W., Kitzing, T., Roessler, S., Zuber, J., Krasnitz, A., Schultz, N., Revill, K., Weissmueller, S., Rappaport, A.R., Simon, J., et al. (2012). A cluster of cooperating tumor-suppressor gene candidates in chromosomal deletions. Proc Natl Acad Sci U S A 109, 8212-8217.
  18. Xue, W., Zender, L., Miething, C., Dickins, R.A., Hernando, E., Krizhanovsky, V., Cordon-Cardo, C., and Lowe, S.W. (2007). Senescence and tumour clearance is triggered by p53 restoration in murine liver carcinomas. Nature 445, 656-660.
  19. Zender, L., Xue, W., Zuber, J., Semighini, C.P., Krasnitz, A., Ma, B., Zender, P., Kubicka, S., Luk, J.M., Schirmacher, P., et al. (2008). An oncogenomics-based in vivo RNAi screen identifies tumor suppressors in liver cancer. Cell 135, 852-864.
  20. Zuber, J., Rappaport, A.R., Luo, W., Wang, E., Chen, C., Vaseva, A.V., Shi, J., Weissmueller, S., Fellmann, C., Taylor, M.J., et al. (2011). An integrated approach to dissecting oncogene addiction implicates a Myb-coordinated self-renewal program as essential for leukemia maintenance. Genes Dev 25, 1628-1640.
  21. Zuber, J., Shi, J., Wang, E., Rappaport, A.R., Herrmann, H., Sison, E.A., Magoon, D., Qi, J., Blatt, K., Wunderlich, M., et al. (2011). RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia. Nature 478, 524-528.