Resistance of mitochondrial p53 to dominant inhibition

Lesions that can contribute to cell transformation normally activate the homo-tetrameric transcription factor p53, primarily to stimulate genes whose products cause senescence or apoptosis [1]. In addition, p53 can provoke apoptosis directly through its interaction with key factors of the apoptotic machinery at the mitochondria and in the cytoplasm [2, 3]. As a result, ideally the transformation process is ceased. Among the p53 target genes that can suppress cell proliferation, one of the most important is p21Waf/Cip1 (CDKN1A), whereas PUMA (p53 up-regulated modulator of apoptosis) constitutes a prime candidate for a p53-responsive master gene of transcription-dependent apoptosis, at least in some tissues [4, 5]. In contrast, transcription-independent apoptosis by p53, which might have evolved to ascertain faithful tumor suppression in the face of lesions that temporarily compromise transactivation [6], involves binding of p53 to Bcl-2 family proteins outside the nucleus. Remarkably, both the transactivation of genes and the interaction with apoptosis regulators are mediated through, and rely upon, the integrity of p53's core DNA binding domain [7, 8]. Tumor-derived mutant p53 proteins are thus usually bi-dysfunctional as they are predominantly mutated within this domain.

Stresses such as DNA-damage, oncogene expression, hypoxia, and reactive oxygen can trigger the translocation of approximately 2% of p53 to mitochondria [9] in many primary and some transformed cell types [6, 9‐12]. Notably, this seems to occur fast and precede transcriptional effects in certain tissues [7, 9, 10, 13, 14]. Mitochondrial p53 is primarily present at the outer mitochondrial membrane where it may form, without the help of further factors, a permeabilizing, death-inducing complex [7, 12]. Alternatively, it may form complexes with the anti-apoptotic Bcl-2 and BclXL proteins [7, 8]. The affinities of these proteins for p53 are higher than for the pro-apoptotic BH123 proteins Bax and Bak; consequently, the latter are freed, form oligomers, and kill the cell. A further pathway may allow pro-apoptotic Bak to be released from the anti-apoptotic Mcl-1 or BclXL proteins upon their association with a distinct site of the p53 DNA binding domain, and then form oligomers and kill the cell [12]. In addition to its mitochondrial action, p53 may bind to cytosolic BclXL and liberate Bax, and may then activate cytosolic monomeric Bax to form lethal oligomers by a mechanisms involving transient interaction of Bax with p53's polyproline-rich domain [15]. Finally, p53 may act through a combined protein interaction and transactivation mechanism: The product of the p53 target gene PUMA resolves an inactive cytosolic p53/BclXL complex by binding to a distinct site on BclXL and allows the activation of Bax by free p53 [16]. Which arm of the complex death program is primarily active almost certainly is cell type and context-dependent.

Mutant p53 is present at the mitochondria regardless of apoptotic stimulus [7], suggesting that in contrast to wild-type (wt) p53, a translocation mechanism is active for the mutant proteins regardless of stress, or that the presence of mutant p53 in and by itself constitutes a death stimulus, as is the case with many other oncoproteins. Apart from exhibiting a wt p53-independent oncogenic 'gain-of-function', which at least in part seems to be mediated through inactivating interaction with other pro-apoptotic members of the p53 family (reviewed in [17]), studies of Li-Fraumeni individuals with an inherited mutated allele and of knock-in mice have clearly indicated that mutant proteins can act dominant-inhibitory, either through hetero-oligomerization with wt p53 expressed from the second allele or through the sequestration of limiting factors [18, 19]. Clearly, these interactions can compromise the transcriptional activity of wt p53.

Another potent mechanism of dominant inhibition seems to employ target gene promoter occupation by transactivation-incompetent (ΔN and ΔTA-) members of the p53 family. Interestingly, inhibition of transactivation by ΔN-p73 occupying p53 recognition motifs appears to play an important physiological role in the protection of developing sympathetic neurons from p53- and p63-provoked apoptosis (reviewed in [20]), whereas a similar mechanism based on aberrantly spliced p73 giving rise to the Δex2/3p73 isoform, can be active in human tumors (for instance, [21]). Here we began to examine to what extent dominant inhibition by the described mechanisms would affect the extra-nuclear apoptotic functions of p53.

The following studies were performed primarily in human HCT116 colon adenocarcinoma cells. HCT116 is a poorly differentiated, growth factor-insensitive cell line exhibiting microsatellite instability (MIN) caused by deficiency for the essential hMLH1 mismatch repair factor. Many forms of stress except aberrant oncogene expression can stabilize and activate the wt p53 tumor suppressor present in these cells and elicit the expected responses, including apoptosis and cell cycle arrest [22‐24]. 5-fluorouracil (5FU), the mainstay chemotherapeutic for colon cancer, induces apoptosis in a p53-dependent manner in HCT116 cells, primarily through the interference of FdUMP with RNA metabolism [23]. α-amanitin, which causes a global transcription inhibition through the initiation of RNA polymerase II degradation, can also provoke HCT116 cell apoptosis. This transactivation-independent cell death was shown to rely on p53 acting at the mitochondria [6].

When exponentially proliferating HCT116 cultures were treated with α-amanitin (10 μM) and analyzed by flow cytometry, the number of cells with a sub-2n DNA content indicative of apoptosis increased with time (see Additional file 1A). However, this increase was not as marked as the one observed by others with the same cell type [6]. In contrast, 5FU (375 μM) induced a robust apoptosis under similar conditions. In agreement with previous reports [6, 23, 25], both agents provoked apoptosis in dependence of p53 as HCT116 p53-/- cells failed to respond in this manner (see Additional file 1A). Robust cell death also ensued when HCT116 cultures were simultaneously treated with 5FU and α-amanitin. As expected, inclusion of α-amanitin in the drug cocktail blocked the strong stimulation by p53 of the p21Waf/Cip1 gene that was normally observed in the presence of 5FU at both the transcript and protein levels (see Additional file 1B). Thus, in accord with earlier findings [6], this shows that 5FU plus α-amanitin can trigger a p53-mediated cell death in HCT116 cultures in the absence of Pol II-mediated transcriptional transactivation. This cell death was apoptotic as it was accompanied by cytochrome c release from the mitochondrial intermembrane space and by caspase 3 activation (see Additional file 1C).

Apoptosis by 5FU, α-amanitin, or both combined (FA hereafter) was preceded by an increased appearance of p53 in the mitochondrial fraction prepared by biochemical cell fractionation (see Additional file 2A). Remarkably and in agreement with previous reports [7, 9], HCT116 p53-/- cells retrovirally infected to express p53 conformational mutant 175H or DNA contact mutant 273H, had mutant p53 in their mitochondrial fraction regardless of stress (see Additional file 2B). To confirm the presence of mutant p53 at the mitochondria, immune electron microscopy was performed on the cell lines in the absence of stress, and the number of gold grains at the mitochondria was counted in a blinded study. Additional file 2C shows that the number of mitochondria-associated grains was significantly higher in the cell lines that express p53 when compared to the vector-only cell line (Chi-square test: P < 0.001 for 273H and P < 0.04 for 175H). As neither mutant can interact with the Bcl-2 family of proteins [7, 8] or transactivate genes [18], no enhanced apoptosis was apparent in these cell lines (data not shown).

To study the effects of mutant p53 on the extra-nuclear, mitochondrial functions of wt p53, parental HCT116 cells (expressing wt p53) were infected, in a first set of experiments and at a multiplicity of infection of <1 colony forming unit per cell, with retroviral vectors expressing either no transgene, or mutants 175H or 273H with a C-terminal 26 amino acid residue truncation (175ΔC and 273ΔC). These truncated proteins were employed to be able to distinguish exogenous and endogenous p53 in Western immunoblots (Figure 1A), and to impair the proteins for potentially confounding gain-of-function activities associated with the C-terminus (e.g. [26, 27]). All p53 proteins (endogenous and exogenous) had arginine at amino acid position 72, the major allele among Caucasians and the p53 isoform that is most efficiently translocated to mitochondria [11]. On average the mutant p53 proteins expressed from a single gene copy per cell were produced at 1.5 to 2-fold higher levels than the wild-type in these bulk-infected cultures, and the presence of the mutant protein affected the levels of wt p53 only slightly (Figure 1A). Wt p53 and the truncated mutants were predominantly present in the nuclear fraction (not shown).

Like the full-length mutants, 175ΔC and 273ΔC were present in the mitochondrial fraction of unstressed cells (Figure 1B; compare with Additional file 2B). In contrast, wt p53 translocated to the mitochondria in these mutant p53-transduced cells only after 5FU treatment (Figure 1B). 175ΔC and 273ΔC, like the full-length mutants but unlike mutant proteins with a deletion in the oligomerization domain, were competent for homo-oligomerization, as indicated by chemical crosslinking with Bis-maleimidohexane (BMH) that crosslinks sulfhydryl groups, and with glutaraldehyde (GLD) that crosslinks amino groups (Figure 1C). Interestingly, 175H and 175ΔC consistently showed a weaker propensity to homo-oligomerize in this cell type than 273H or 273ΔC, a tendency observed with both BMH and GLD crosslinkings (Figure 1C). This became also apparent in an assay of hetero-oligomerization when the mutants were employed to co-immunoprecipitate, with p53 antibody DO-1, a full-length p53 carrying a mutated DO-1 epitope (Figure 1D). Thus, 273H and 273ΔC seem to form stable oligomers more readily than 175H and 175ΔC in HCT116 cells. Combined, these data indicate that the presence of the mutants 175ΔC and 273ΔC at approximately two-fold increased levels compared to wild-type, does not interfere with the 5FU-induced mitochondrial translocation of wt p53.

We next asked whether the mutants, at the given average levels of expression (see Figure 1A), can act dominant-inhibitory on transactivation upon 5FU treatment. Mutant 273ΔC was able to weaken the 5FU-induced stimulation by wt p53 of the p21Waf/Cip1 gene (Figure 2A). Bax, Bak, and gapdh were neither transactivated by p53 nor measurably affected by mutant p53. In accord with its incompetence for oligomerization (see above), mutant 175ΔC failed to act dominant-inhibitory in this assay (Figure 2A), suggesting that hetero-oligomerization rather than competition for limiting factors was the primary mechanism of dominant inhibition by 273ΔC. The discrepancy in dominance between 273ΔC and 175ΔC in HCT116 cells was conserved in transient transfections: 273ΔC inhibited endogenous p21Waf/Cip1 protein expression while 175ΔC did not (Figure 2B).

To confirm the inhibitory effect of 273ΔC on the induced expression of p21Waf/Cip1, chromatin immunoprecipitation (ChIP) was performed. Consistent with an inefficient promoter occupation by the hetero-oligomers, the promoter sequence of p21Waf/Cip1 was less frequently precipitated with an anti-p53 antibody from cultures expressing 273ΔC than from cultures harbouring vector only. Again, no differences were observed between the 175ΔC and control cells (Figure 2C).

We next asked whether the observed effects of mutant 273ΔC on transactivation might translate into a phenotype. The seeding of 1,000 live cells of the respective cell lines onto 10 cm dishes and the maintenance of these cultures under stress-free conditions resulted in approximately equal numbers of colonies. However, when the cells were exposed to different sub-lethal concentrations of the DNA-damaging drug etoposide, only cells that expressed mutant p53 were able to grow colonies at a threshold concentration of the drug, and again, 175ΔC proved less dominant than 273ΔC in this assay (Figure 2D). Etoposide was chosen as the stressor because it induces primarily cell cycle arrest whereas 5FU induces primarily apoptosis in HCT116 cells [23, 28]. Altogether, the data suggest that p53 mutant 273ΔC expressed from a single gene copy and produced, on average, at approximately 1.5-times the level of endogenous wt p53, can act dominant-inhibitory on transactivation and the suppression of colony formation by wt p53 in stressed HCT116 cells. 175ΔC may be impaired in these assays owing to its lower propensity to oligomerize.

Like with the parental HCT116 cells (see Additional file 2A), treatment of the cell lines HCT116-LRNL (vector-only control), HCT116-175ΔC, and HCT116-273ΔC with α-amanitin, 5FU, or both, resulted in the accumulation of endogenous wt p53 in the mitochondrial fraction (Figure 3A). In a next set of experiments, we quantified the apoptosis experienced by the three cell lines by two different means. Cell death provoked by α-amanitin, 5FU, or FA within 12 or 24 h (early apoptosis) was measured flow-cytometrically by the binding of PE-conjugated Annexin V to phosphatidyl-serine on non-fixed cells; necrotic cells were excluded by counterstaining with propidium iodide. Subsequently, DNA fragmentation as an indicator of late apoptosis was quantified by determining the numbers of cells with a sub-2n DNA content in cultures with 48 h of drug exposure. The results are summarized in Figure 3B. Neither drug was able to induce significant apoptosis within 12 h. However, by 24 h and 48 h apoptosis was present in the drug-treated cultures, and a concomitant increase of the cytoplasmic cytochrome c levels was observed in all three cell lines (Figure 3C). Importantly, neither in the cultures that were seeing 5FU alone (p53 at mitochondria, transcription intact) nor in the cultures with α-amanitin (p53 at mitochondria, transcription blocked) was a dominant-inhibitory effect of mutant p53 detectable. This pointed to the possibility that, in contrast to the transactivation of p21 (see Figure 2A), the mitochondrial apoptotic function of p53 is resistant to dominant inhibition.

To further study whether dominant inhibition might spare the extra-nuclear apoptosis by p53, we resorted to the artificial p53DD mini-protein consisting of the oligomerization domain and a small fragment from the N-terminus (aa 1–14 and 302–390) of mouse p53 [29]. p53DD has been shown to act strongly dominant-inhibitory in cells of mouse and human origin (for instance, [30‐32]). We generated HCT116 cells that contain a single copy of the p53DD gene by retroviral bulk infection and tested them for dominant inhibition and apoptosis. Figure 4A shows that p53DD is expressed and can induce stabilization of the endogenous wt p53. Figure 4B documents that the stress-induced transactivation of p21 is compromised in the HCT116-p53DD cells, in accord with dominant inhibition and with results reported previously [30‐32]. Notably, p53DD like 273H failed to inhibit transactivation-independent apoptosis in these cells (Figure 4C), lending further credence to the suggestion that the mitochondrial apoptotic function of p53 is resistant to dominant inhibition in HCT116 cells.

One explanation for the impairment of mutant p53 to interfere with wt p53 at the mitochondria might have been that the mutant cannot efficiently contaminate mitochondrial p53 oligomers through hetero-oligomerization, for instance because the mechanism that translocates wt p53 to mitochondria upon stress is selective for wt tetramers. Notably, we were unable to co-immunoprecipitate mutant and wt p53 from the mitochondrial fraction. We therefore decided to examine the state of p53 oligomerization in this fraction. Consistent with previous findings [33], the presence of Bax oligomers after apoptotic stimulus was readily detectable in total cell protein sample and the mitochondrial fraction by chemical crosslinking with BMH and subsequent immunoblot analysis, documenting that this procedure can detect oligomers in the mitochondrial fraction (Figure 5A). Similarly, p53 oligomers were detected in the total cell protein sample of HCT116 cells expressing either wt p53, 273H, or 175H. In contrast, no or only very few p53 oligomers were found in the mitochondrial fractions of either cell type under identical conditions (Figure 5B). High molecular weight bands indicative of p53 being trapped in megacomplexes were only observed in total cell extracts but not mitochondrial fractions (see Additional file 4), suggesting that a substantial p53 loss through this mechanism is unlikely.

BMH crosslinks sulfhydryl groups while glutaraldehyde (GLD) crosslinks amino groups. To confirm the absence of detectable p53 oligomers in the mitochondrial fraction, a similar study was performed with GLD. As shown in Figure 5B, GLD was able to crosslink p53 in total cell protein samples but not in the mitochondrial fraction. To test whether this was limited to HCT116 cells, we next employed human MCF7 breast adenocarcinoma cells with an intact wt p53 pathway. 5FU induced mitochondrial translocation of p53 in MCF7 cells (Figure 5C). As in HCT116 cells, the mitochondrial fraction of MCF7 cells was devoid of oligomeric p53 upon BMH of GLD crosslinking while oligomers were readily detectable in the total cell protein sample (Figure 5D). Combined, these data suggest that the majority of mitochondrial p53, be it wild-type or mutant, is monomeric. We cannot exclude that mitochondrial p53 consists of unstable oligomers, or that a small fraction of p53 is present at the mitochondria as tetramers and is trapped in megacomplexes after crosslinking. In support of the conclusion that mitochondrial p53 is mostly monomeric, p53DD capable of forming immuno-precipitable hetero-oligomers with wt p53 in HCT116-p53DD cells, was not detectable at the mitochondria whereas wt p53 was (Figure 5E).

For the reasons detailed above, we had employed C-terminally truncated versions of mutant p53 and shown that these, although capable of dominance towards wt p53-mediated transactivation of genes, failed to interfere with the transcription-independent apoptosis by p53 at the mitchondria. To study the effects of full-length mutants, expression plasmids producing HA-tagged full-length 175H, 273H and also 248W as another common tumor-associated conformational mutant were generated. All mutants were expressed to approximately equal levels upon transfection into HCT116 p53-/- cells (see Additional file 3A). In a first study we determined the ratio of mutant vs. wt p53 expression plasmids that knocked down the transactivation by wt p53 of the endogenous p21 gene (see Additional file 3B). All mutants including 175H were able to efficiently block p21 activation, indicating that 175H, unlike its truncated variant 175ΔC, was exerting dominance efficiently. This ratio of vector/wt p53 or mutant/wt p53 plasmid was then used to transfect cultures and determine the levels of transcription-independent apoptosis by p53. The results are summarized in Additional file 3C. p53-negative cultures receiving vector-only produced background levels of apoptosis (6–8%). Cultures transfected with mutant p53 plasmids alone showed apoptosis at or even below background (not shown). In contrast, wt p53 expression alone entailed significant cell death (23%). Notably, over-expression of mutant p53 to levels sufficient to inhibit p21 transactivation nonetheless allowed significant transcription-independent apoptosis to occur (14–16%). The reduction of apoptosis in cells receiving mutant p53 plasmid in addition to wt p53 plasmid may reflect anti-apoptotic function(s) of over-produced mutant p53. Combined, we interpret these results to support that the transcription-independent mitochondrial apoptosis by p53 in HCT116 cells is relatively resistant to dominant inhibition by mutant p53.

Dominant inhibition through hetero-tetramerization does not seem to be very efficient, compared with dominant inhibition through target gene promoter occupation by protein isoforms of the p53 family that lack the transactivation domain [34, 35]. For instance, the ΔTA-p73 isoforms are among the known strong transcriptional competitors of p53, both in the tumor and normal tissue developmental setting. To test whether tumor-associated Δex2/3p73 [21] can compromise mitochondrial p53-mediated apoptosis, HCT116 cells were retrovirally transduced to express this isoform (Figure 6A). Notably, although the dominant inhibition of the p53-dependent stimulation of p21 transcription by Δex2/3p73 was even stronger than the one produced by p53DD (Figure 6B), there was no inhibition of transcription-independent apoptosis (Figure 6C). In summary, our data thus suggest that the extra-nuclear apoptotic functions of p53 at the mitochondria are relatively resistant against dominant inhibition through hetero-tetramerization and promoter occupation.

p53 binds to its DNA recognition motif as a tetramer; monomers can not function as a transcription factor [36]. Multimeric transcription factors, in contrast to monomeric, can be subject to dominant inhibition not only through i) the sequestration of limiting factors by an overproduced mutant, or ii) promoter-occupation by a transactivation-defective isoform, but also through iii) contamination of the multimer by mutated monomers. Although the latter two mechanisms have been documented to be active against p53 in vitro and in vivo (reviewed in [18, 37, 38], some major conflicting observations have for some time challenged the concept of dominant-negative action against p53. Doubtlessly, the observation that of the 159 annotated human tumors which had been studied for loss-of-heterozygosity by mid-2007, and which each expressed one out of 90 distinct mutated p53 alleles, approximately 60% have lost the wild-type allele (Olivier and Hainaut in [39]), makes a case against the global importance of dominance. Similarly negative was the observation that approximately 50% of the tumors that develop in Li-Fraumeni patients with one inherited mutated full-length p53, have lost their wild-type allele [40, 41]. On the other hand, there are several important observations in strong favour of dominance. For example, while only 50% of the Li-Fraumeni tumors with a mutated full-length p53 are wt p53-deficient, this is true for almost 100% of the Li-Fraumeni tumors with an inherited p53 allele mutated to produce no protein [40, 41].

Strong support for dominant inhibition comes also from studies on transgenic mice, although in this system dominance can be somewhat more difficult to assess, mostly for two reasons. First, the two p53 mutations introduced into the mouse germ line by knock-in in two central studies [42, 43], 172H and 270H (corresponding to human 175H and 273H, respectively), display additional, wt p53-independent oncogenic functions (gain-of-function) that may superimpose dominant-negative effects. Second, whereas human p53+/- cells almost always loose their wt allele in the course of transformation, p53+/- cells from mice often retain it [44] indicating that in mice, in contrast to humans, gene dose reduction can impair critical p53 functions to a degree as to mimic, with respect to tumor development, complete p53-deficiency. Nonetheless, careful studies of the efficacy of known wt p53 activities in primary fibroblasts from 172H/+ and 270H/+ mice, compared to cells from -/+ animals, have clearly demonstrated that the mutant proteins can act dominant-negative [42, 43]. Combined, the available data clearly indicate that hetero-tetramerization can incapacitate wt p53 as a transcription factor [1, 18, 37, 38]. The degree of inhibition by a mutant depends on the mutant-to-wt protein levels, the site of mutation, and the cell type [34, 35, 39].

Here we have presented evidence suggesting that the extra-nuclear, mitochondrial death pathway of p53 protein can be resistant to dominant inhibition because mitochondrial p53 is mostly monomeric. This raises important questions. For example, can monomeric p53 trigger mitochondrial apoptosis? Moll and colleagues have convincingly demonstrated that a mitochondrially targeted monomeric p53 was not only capable of provoking apoptosis but was indeed the most efficient cell death inducer of all tested p53 proteins [9]. That group also showed that mitochondrially targeted p53 can induce apoptosis in cells expressing mutant p53 [45]. So clearly, monomeric p53 at the mitochondria can trigger apoptosis.

If the p53-mediated mitochondrial apoptosis is resistant to dominant inhibition, why are the 40% of tumors with a mutant p53 allele that have retained the wt allele (dominant inhibition active) not dying through this pathway? Several aspects may be relevant here. First, such tumors may have evolved to impair this pathway through other means, as for instance the overproduction of Bcl2/BclXL or down-regulation of Bax/Bak. Second, even in cells expressing mutant p53, the translocation of wt p53 to the mitochondria required a stress signal (see Figure 1B). Therefore, (some) p53-dependent apoptosis may well occur in these tumors despite of the inhibition of p53's nuclear transactivator function by the mutant, but this may need stress, as for example under chemotherapy, to become apparent. The HCT116 cells used in this study had to be exposed to genotoxic stress for p53 translocation to occur; the oncogenic deregulation present in these cells was insufficient to act as p53 stressor, probably primarily because HCT116 cells fail to express the oncogenic stress signal mediator p14ARF [46]. Third, Chipuk and colleagues have documented that cell types exist in which efficient mitochondrial apoptosis by p53 is dependent upon the transactivation of the p53 target gene PUMA. PUMA resolves an inactive cytosolic p53/BclXL complex and allows the activation of pro-apoptotic Bax by free cytoplasmic p53 [16]. Thus, if tumors that co-express wt and dominant-negative mutant p53 belong to this cell type, extra-nuclear apoptosis induction by p53 might be impaired despite of the resistance of mitochondrial p53 to dominant inhibition by hetero-tetramerization. Forth, some of these PUMA-sensitive tumors may have developed from cell types that are able to mobilize the p53-relatives p63 or p73 for the transactivation of PUMA. In such tumors, it will make a difference whether the expressed mutant p53 is of class I (wild-type conformation) or class II (conformation-altered) type. Class II but not class I mutants can bind and inhibit p63 and p73, which is considered to be one mechanism of 'gain-of-function' [17]. Consequently, tumors of this type with the genotype (wt p53/class II mutant), but not (wt p53/class I mutant) tumors, may exhibit a gain-of-function phenotype with respect to the impairment of mitochondrial apoptosis. Importantly, these inferences can be subject to experimental testing. In sum, the resistance of mitochondrial monomeric p53 to dominant inhibition by hetero-tetramerization should increase the selection pressure on a stressed tumor with a mutated p53 allele to loose the mitochondrial apoptosis function.

Contamination of the wt p53 tetramer with mutant p53 through hetero-tetramerization is a remarkably inefficient way to inactivate p53 [34, 35] whereas dominant inhibition through transactivation-defective (ΔN or ΔTA) isoforms of, for instance, p53 or p73, is very efficient ([21, 35], see also Figure 6). The latter mechanism can involve the formation of hetero-tetramers between wt p53 and ΔN-p53, where one truncated protein per tetramer is sufficient to incapacitate it for transactivation [35], and can involve competitive promoter-occupation by ΔN-p53 or ΔN-p73 [21, 35]. The extraordinary efficiency of this inhibitory mechanism may be one reason for the fact that it has gained important roles in several physiological processes. For example, ΔN-p73 is a very efficient and perhaps the primary inhibitor of p63-mediated apoptosis during neuronal development, and of p53-mediated apoptosis in response to adult neuron injury and neurodegeneration [20]. With this efficacy of inhibition, it is no surprise that ΔTA-isoforms of p73 have also been found in tumor cells. One might thus speculate that the ΔTA-p73-resistant apoptotic function of mitochondrial p53 has evolved, at least in some cell types, as part of a fail-safe mechanism against the powerful dominant-negative effects of these molecules.

Monomeric p53 not only escapes dominant inhibition through hetero-tetramerization but also ubiquitylation through the E3/E4 ubiquitin ligase MDM2, as MDM2 binds exclusively to tetrameric p53 [47]. On the other hand, (multi-)mono-ubiquitylation of p53 by MDM2, as opposed to poly-ubiquitylation associated with p53 degradation, can promote mitochondrial p53 translocation [48], and in accord with this, the apoptosis- and tetramerization-defective p53 mutant 337C [49] occurring in Li-Fraumeni patients fails to translocate to the mitochondria upon stress (our unpublished observation). Thus, the recent finding that cytoplasmic p53 – mono-ubiquitylated by cytoplasmic MDM2 – travels to the mitochondria without shuttling MDM2 [48], might suggest that monomerization of p53 occurs in stressed cells somewhere between this modification and p53's final destination at the mitochondria.

Mutant p53 can compromise transactivation by wild-type p53 through hetero-tetramerization, and p53/p63/p73 isoforms lacking a transactivation domain through competitive promoter-occupation as well as hetero-tetramerization. p53 has evolved to command a second, transcription-independent pathway of apoptosis induction at the mitochondria. This second pathway – although probably less efficient than full apoptosis by p53, provoked by both pathways in concert, – is resistant against dominant inhibition in HCT116 cells. Moreover, mitochondrial p53 – in contrast to nuclear – is mostly monomeric. Thus, the extra-nuclear p53-dependent apoptosis may constitute a fail-safe mechanism against dominant inhibition.