p53 and PUMA: A Deadly Duo

See allHide authors and affiliations

Science  09 Sep 2005:
Vol. 309, Issue 5741, pp. 1685-1686
DOI: 10.1126/science.1118232

Despite its importance in protecting us from malignancies, the specific activities of the p53 protein that allow it to function as a tumor suppressor have been hard to reconcile. A paper by Chipuk et al. on page 1732 of this issue (1) now promises to reduce the complexity of the p53 response to cellular stresses by bringing together two seemingly separate activities of p53—one nuclear and one cytoplasmic—into a single, unified model.

The ability of p53 to function as a transcription factor is generally considered to be its main physiological property (2). Expression array technology has revealed long lists of potential gene targets of p53, with the accompanying problem of sorting the wheat from the chaff. Analyses of cells defective in some of these p53-target genes have established several of them as important mediators of the p53 response. For example, p21Waf1/Cip1, an inhibitor of an enzyme (a cyclin-dependent kinase) that controls the cell division cycle, is key to the proliferative arrest induced by p53. More recent studies have indicated a surprisingly strong dependence of p53-mediated apoptosis on the presence of PUMA, a protein with clear apoptotic potential (3). Combined with compelling evidence that the ability to function as a transcription factor is essential for the antineoplastic activity of p53, it is tempting to conclude that regulation of gene expression is the key to its suppressive effects on tumorigenesis.

However, complicating this simple model have been persistent indications that p53 also functions through nonnuclear, transcriptionally independent mechanisms (4). Early studies showing that p53 fragments lacking part of its DNA binding domain could still induce cell death, and that p53-mediated death was not necessarily dependent on new protein synthesis, elicited some concern as to their physiological relevance. However, more recent evidence has revealed a function for p53 outside the nucleus—as a binding partner of the anti-apoptotic/prosurvival members of the the Bcl2 superfamily such as Bcl2 and Bcl-xL (5) (see the figure, top). In this respect p53 seems to function like some of the proapoptotic members of the Bcl2 superfamily—a group of proteins defined by the presence of a region of sequence similarity called the BH3 domain—and socalled BH3-only proteins. This activity of p53 as a BH3-only protein comes as quite a surprise, because there is no obvious BH3 domain within p53. But PUMA, a validated p53-induced apoptotic protein, clearly contains a BH3 domain. This presents a conundrum: Why would p53 activate expression of PUMA, a BH3-only protein, if p53 functions as a BH3-only protein itself?

Models for the apoptotic activity of p53.

(Top) Apoptotic functions of p53 include a nuclear role as a transcription factor that activates expression of gene targets including PUMA, and a nonnuclear role in the cytoplasm and mitochondria, involving interaction with anti-apoptotic members of the Bcl2 family of proteins. Each of these may result in mitochondrial outer membrane permeabilization and cell death. A number of cytoplasmic and mitochondrial activities of p53 have been proposed (see text). In this example, mitochondrial p53 releases Bid from BclxL. In these models, the nuclear, cytoplasmic, and mitochondrial functions of p53 are not directly linked, although they may cooperate to fully activate cell death. (Bottom) The transcriptional and cytoplasmic functions of p53 are brought together by Chipuk et al. (1) in the same pathway where p53 activates the expression of PUMA, which then serves to release cytoplasmic p53 from the inhibitory interaction with Bcl-xL. The released p53 is then free to directly activate Bax. In this model p53 acts at two steps in the same chain of events, leading to mitochondrial outer membrane permeabilization and apoptosis.


The answer to this riddle appears to lie in the detail of how BH3-only proteins themselves function. The end point of the chain of events induced by BH3-only proteins is activation of the proapoptic proteins Bax and/or Bak (also members of the Bcl2 superfamily) to allow mitochondrial outer membrane permeabilization and release of apoptogenic factors such as cytochrome c. This then triggers the activation of a cascade of proteolytic enzymes called caspases and the apoptotic demise of the cell. But BH3-only proteins come in two flavors—activators and enablers (6, 7). Activator BH3-only proteins function by transiently binding to, and directly activating, Bax and Bak to start the apoptotic ball rolling at the mitochondria. The enabler BH3-only proteins take a more elaborate and indirect path to activate Bax and Bak by forming a complex with the anti-apoptotic proteins such as Bcl2 or Bcl-xL. In the current model of these pathways, the binding of enabler BH3-only proteins to the anti-apoptotic proteins by itself has no real consequences. Rather, the importance of these interactions is that they displace activator BH3-only proteins (or even Bax or Bak themselves) from an inhibitory interaction with the anti-apoptotic proteins, so releasing them to drive the initial steps of the apoptotic cascade. The beauty and simplicity of the Chipuk et al. paper lies in uniting the evidence that PUMA is an enabler (8) and p53 an activator (9) into a model in which PUMA functions to release p53 from BclxL, thereby freeing p53 to activate Bax (see the figure, bottom). In the absence of cell stress, only low levels of p53 are expressed. Thus, the nuclear level of p53 is insufficient to drive transcriptional activation of PUMA and the small amount of cytoplasmic p53 is held inactive by Bcl-xL. An unexpected stressful “jolt” to the cell, such as DNA damage or oncogene activation, results in a rapid increase of nuclear p53 and transcriptional activation of PUMA expression. PUMA then binds to Bcl-xL, hence freeing p53 to activate Bax.

Of course, there are other models of p53 function that might be considered, and alternative explanations are necessary to accommodate all the available data. A number of studies have shown that p53 moves to the mitochondria in response to stress, suggesting that translocation and binding of p53 to mitochondrial Bcl2 and Bcl-xL may also trigger apoptosis (10). In this model the mitochondrial p53 could function as an enabler BH3-only protein to release activators like Bid (see the f igure, top). Mitochondrial p53 can also show activator functions such as binding to Bak, which results in the release of Bak from Mcl1 (an anti-apoptotic protein similar to Bcl2 and Bcl-xL) (11). Direct activation of Bax by p53 appears to take place in the cytosol (9). To address the relative importance of enabler and activator functions of p53, Chipuk et al. used a Bcl-xL mutant that binds p53 but not PUMA. Although both wild-type and mutant Bcl-xL inhibited p53-mediated apoptosis, expression of PUMA could only reverse the effect of wild-type Bcl-xL. The implication is therefore that p53 has to be released from Bcl-xL by PUMA to induce apoptosis and that the binding of p53 to Bcl-xL is by itself not a proapoptotic signal. While providing elegant support for the activator function of p53, this observation does not preclude a function for p53 as an enabler and overall it seems likely that coordination of the nuclear, cytoplasmic, and mitochondrial functions of p53 will contribute to the ultimate response to stress.

A number of questions arise out of the Chipuk et al. study, including whether p53 requires PUMA, or PUMA requires p53, to induce cell death. The answer to the first question, at least in some cell types, seems to be yes. Deletion of PUMA by genetic knockout or knockdown by RNA interference strategies strongly impairs p53-dependent apoptosis in certain cell systems (3). But PUMA may not be unique in this function. A number of other BH3-only proteins, such as Noxa, are transcriptionally activated by p53 and play an essential role in the p53 apoptotic response in some cell types. So it seems likely that under certain conditions, proteins like Noxa might substitute for PUMA. Less clear is whether PUMA might require p53. Initial studies have indicated that PUMA expression is enhanced by withdrawal of serum from cells or by inducing stress in the endoplasmic reticulum, and that this induction of PUMA is p53 independent (12, 13). Furthermore, like p53, the transcription factor E2F1 can activate PUMA expression, and in this case PUMA was shown to contribute to apoptosis without requiring p53 (14). Although it is extremely exciting to consider p53 as a functional homolog of an activator BH3-only protein, other proteins, such as Bim and Bid, also display this activity. Because PUMA has an extremely high affinity for the antiapoptotic Bcl2-like proteins (15), it seems reasonable to suppose that in addition to releasing p53, PUMA will also release any bound Bim or Bid. Indeed, the study by Chipuk et al. suggests that this is possible, and may go some way to explaining the strong apoptotic activity seen following PUMA expression in some p53-null cells. But regardless of these details, which will undoubtedly be the subject of strong debate and intense research, the p53-PUMA relationship suggested by this model provides a very satisfying explanation to the quandary of why p53 should have evolved both transcriptional and cytoplasmic functions.


View Abstract

Stay Connected to Science

Navigate This Article