Che-ating death: CHE1/AATF protects from p53-mediated apoptosis

Abstract

EMBO J (2012) 31 20, 3961–3975 doi:; DOI: 10.1038/emboj.2012.236; published online August 21 2012

The tumour suppressor p53 directs cells towards different fates depending on the cell type and the stimulus. The decision to direct a cell towards apoptosis rather than cell-cycle arrest or senescence has important implications for tumour suppression in normal cells and drug response in tumour cells. Cells that undergo senescence and growth arrest can persist and contribute to organismal ageing (Campisi, 2005), or they can contribute to tumour relapse (Jackson et al, 2012). In this issue of The EMBO Journal, Höpker et al (2012) show in a comprehensive study that the RNA PolII binding protein CHE1/AATF is a factor that determines the fate of cells that have activated the p53 pathway.

Using a phospho-proteomic screen for novel proteins that regulate the DNA damage response (DDR) pathway, the authors discovered MRLC3, a cytoskeletal protein that bound strongly to peptide sequences resembling the phosphorylation motifs of DDR targets, but only when the peptide was in the unphosphorylated form. A yeast two hybrid screen validated by co-immunoprecipitation experiments identified phosphorylated CHE1/AATF as an MRLC3 interacting protein. The phosphorylation of CHE1/AATF was induced by UV or the DNA damaging chemotherapeutic agent Doxorubicin. Indeed, CHE1/AATF shuttled from the cytoplasm to the nucleus in cells exposed to UV or hyperosmotic stress in a phosphorylation-dependent manner.

The finding that the MRLC3–CHE1/AATF complex is cytoplasmic before DNA damage and CHE1/AATF translocates to the nucleus after DNA damage, suggests that the DDR kinase responsible for the phosphorylation is cytoplasmic. Indeed, the DNA damage activated kinase MK2 phosphorylated CHE1/AATF on Threonine 366 in vitro and CHE1/AATF did not translocate to the nucleus after DNA damage in MK2-null MEFs. 3D modelling suggested that the MRLC3–CHE1/AATF interaction would be disrupted by a phosphorylation on T366, consistent with the biochemical experiments.

To determine how CHE1/AATF levels affect patient survival, the authors examined 164 human neuroblastoma samples and corresponding patient data. Interestingly, when patients were stratified according to CHE1/AATF expression, those with low levels had a greater probability of survival than those expressing higher levels. These data suggest that, unlike other targets of the DDR, CHE1/AATF might not actually be activating a cell death pathway, but may play a role in survival of tumour cells.

Of note, neuroblastomas are typically TP53 wild type. p53 is a major DDR protein that activates transcription of numerous targets involved in apoptosis, cell-cycle arrest, senescence, metabolism, and paradoxically, survival. In p53 wild-type cells, repression of CHE1/AATF by shRNA knockdown did not affect the number of cells that escaped cell-cycle arrest following DNA damage, and overexpression of CHE1/AATF only slightly improved the arrest. However, knockdown of CHE1/AATF enhanced apoptosis following UV or DNA damaging chemotherapy treatment in several p53 wild-type cell lines, but not an isogenic variant of one of these cell lines that lacks p53, or two other p53 mutant cell lines. Importantly, following UV, p53 targets that induce apoptosis such as PUMA were not induced unless CHE1/AATF was knocked down, while induction of growth arrest targets of p53, such as p21, was unaffected. The generality of this response to other stimuli is unknown. The inhibition of p53-induced apoptotic target genes was direct: CHE1/AATF associated with chromatin in the promoters of PUMA, BAX and BAK following UV treatment of MEFs, but not p21 or cell-cycle arrest targets. The induced association with chromatin was p53 independent, and p53 association with its target gene promoters was independent of CHE1/AATF nuclear localization. In MEFs that lack MK2 and hence sequester CHE1/AATF in the cytoplasm, CHE1/AATF no longer associated with p53 apoptotic promoters following UV stress. Whether this shifts the response in the MEFs from arrest to apoptosis, similar to CHE1/AATF knockdown in human tumour cell lines, is currently unknown.

Another important question addressed by the authors was the clinical benefit of shifting the p53-mediated gene expression of treated cells to favour apoptotic targets. Established xenografted tumours (HCT116) with CHE1/AATF knockdown had more dramatic tumour regression following genotoxic chemotherapy treatment than the parental cell line, and concomitant greater staining for PUMA in treated tumour sections. Further, HCT116 xenografts expressing a mutant CHE1/AATF that mimics the phosphorylated state of the protein (which is constitutively nuclear and active) showed a dampened response to chemotherapy treatment.

It was also found that in endometrial cancer specimens, tumours with low level p53 IHC staining (used as a surrogate for detection of wild-type TP53) had largely nuclear CHE1/AATF, suggesting selective pressure for tumours with wild-type p53 to have phosphorylated CHE1/AATF in the nucleus, repressing transactivation of apoptotic targets. Tumours with high level of p53 IHC staining (used as a surrogate for detecting TP53 mutation) did not have the selective pressure to translocate high levels of CHE1/AATF to the nucleus.

Why p53 directs cells towards apoptosis in some cell types and growth arrest in others has been an area of intense research (Vousden and Lu, 2002). This study provides evidence that following DNA damage, the nuclear localization of CHE1/AATF protects cells from apoptosis. How this regulator of p53-mediated apoptosis versus arrest acts in concert with or in opposition to other mediators of this decision such as p53 post-translational modification (Tang et al, 2006; Jansson et al, 2008) and interaction with other cofactors is unknown (Samuels-Lev et al, 2001; Moumen et al, 2005; Das et al, 2007).

It is interesting to note in this study that following UV treatment in the HCT116 model system, pro-apoptotic genes were not induced unless CHE1/AATF was knocked down. In contrast, when other cell types including fibroblasts and breast cancer cells are exposed to DNA damaging stimuli, robust induction of apoptotic genes is observed, but the cell fate remains growth arrest, not apoptosis (Oda et al, 2000; Nakano and Vousden, 2001; Jackson et al, 2012). If the level of apoptotic gene induction observed in these studies is insufficient to induce apoptosis, would knockdown of CHE1/AATF mean even greater induction and apoptosis? If so, then could inhibition of CHE1/AATF activity, achieved pharmacologically most likely through MK2 kinase inhibition, improve the efficacy of drugs in these cell types, shifting response to apoptosis (Figure 1)? Another potential area of clinical utility for CHE1/AATF inhibition would be to augment treatment with the p53 activating drug Nutlin. This drug, which disrupts the binding of p53 to its inhibitor Mdm2, induces cell-cycle arrest, not apoptosis, in many cell types (Xia et al, 2011).

Figure 1.

Model of pharmacological inhibition of CHE1/AATF phosphorylation. In some cell types, following DNA damaging chemotherapy, the DDR signals to the kinase MK2 to phosphorylate CHE1/AATF, resulting in translocation to the nucleus and inhibition of p53-mediated transcription of its apoptotic target genes (A). Pharmacological inhibition of CHE1/AATF nuclear localization and binding to the p53 apoptotic target gene promoters, as depicted here with a kinase inhibitor of MK2 (KI), could increase expression of BAX, PUMA and BAK, resulting in a shift in response from arrest to cell death (B).

The role of CHE1/AATF in other cancer cell types is also not well understood, notably those of lymphoid origin, which undergo p53-mediated apoptosis following treatment with DNA damaging drugs (Fan et al, 1994; Weller, 1998). Perhaps, CHE1/AATF fails to translocate to the nucleus and to promoters of apoptotic genes in these cell types, but this awaits testing by immunohistochemical analysis of CHE1/AATF localization in these tumour types before and following treatment. It will be fascinating to see whether CHE1/AATF localization and activity explain the differences in response to irradiation in organs such as the thymus, haematopoietic system and embryonic brain that undergo p53-dependent apoptosis and other organs that do not such as adult brain, lung and kidney (Gudkov and Komarova, 2003).

The complexity of the p53 pathway in cell fate decisions continues to grow with new discoveries. How these factors can be used to benefit treatment response in cancer patients is an area of immense interest. The identification of CHE1/AATF as an important factor in this decision is likely to kindle many future studies.