The tumor suppressor and apoptotic transcription factor p53 has been linked to Alzheimer disease in numerous ways, but perhaps the most direct link is revealed in the June 7 Journal of Neuroscience. Frederic Checler and colleagues report that APP intracellular domains (AICDs) activate the p53 promoter and increase p53 mRNA and protein activity. The findings help explain observations of increased p53 in Alzheimer disease (AD) brain and suggest that p53-driven apoptosis may play a significant role in AD pathology.

First author Cristine Alves da Costa and colleagues at Checler’s lab at Nice-Sophia-Antipolis University, Valbonne, France, together with Peter St. George-Hyslop and colleagues at the University of Toronto and Nadège Girardot at the Pitié-Salpêtrière Hospital, Paris, used immunoreactivity and luciferase reporter assays to measure levels of p53 and activation of the p53 promoter and p53 target genes. Their first hint that APP processing might be tied to p53 came from studies of presenilin (PS)-negative fibroblasts. The authors found that p53 levels are significantly lower in cells devoid of both PS1 and PS2 than in wild-type. The luciferase p53 promoter assay confirmed that the loss of the transcription factor was due, at least partly, to poor transcription of the p53 gene.

Next, Alves da Costa and colleagues found that γ-secretase inhibitors DFK167 and L685458 also lowered p53 levels, supporting the idea that γ-secretase-dependent signaling may be linked to activation of p53. Of course, the intramembrane protease has many substrates, including APP and Notch, but when the authors transfected cells with APP intracellular domains (AICDs), they found that p53 activity and transactivation of p53 target genes were increased. In this experiment the authors used AICD50 and AICD59, cleavage products of γ and ε cleavage of APP, respectively. They also found that transfecting fibroblasts with the AICDs led to an increase in staurosporine-stimulated caspase-3 activity. Because the caspase is a key regulator of apoptosis and is activated in response to p53, the findings outline a direct signaling pathway from APP to p53 to activation of apoptosis. In support of this, the authors found that PS2-induced caspase-3 activation failed in the absence of p53 and that p53 levels were reduced almost 50 percent in APP-negative fibroblasts.

But how might the intracellular domains of APP stimulate the p53 promoter? Growing evidence links the intracellular domain of APP to transcriptional activation with several AICD partners, including the proteins Fe65 and Tip60 that form a trimeric complex with AICD (see ARF related news story), suspected of playing a role. Indeed, when Alves da Costa and colleagues co-transfected these two proteins with AICD59, the APP intracellular domain was stabilized (as previously reported—see Kimberly et al., 2001), and caspase-3 activity increased by about threefold. The combined evidence suggests that AICD can activate p53 transcription and subsequently activate caspase-3, though what lies directly downstream of AICD in this signaling pathway remains to be determined.

Most of this work was done in cultured fibroblasts, but whether these relationships might hold up in neurons is unclear. However, there are indications that they might. The authors did find higher levels of immunoreactive p53 in both sporadic and familial AD brain samples, confirming previous similar reports (see, for example, de la Monte et al., 1997). More specifically, they found that the number of neurons staining positive for the transcription factor was twofold higher in AD and FAD samples compared to controls. The increase in p53-positive neurons in idiopathic AD samples is difficult to link to APP processing, because as the authors state, sporadic AD is not generally regarded as being due to altered γ-secretase activity. Instead, the authors suggest that p53 increases in sporadic AD may result from poor degradation of AICDs. In fact, they report that insulin-degrading enzyme (IDE) levels are lower in AD compared to both FAD and normal brain tissue; in addition to Aβ, IDE has also been shown to degrade AICD (see Edbauer et al., 2002).

In the FAD samples it may be easier to reconcile the increased p53 levels with changes in APP processing because when Alves da Costa and colleagues transfected cells with FAD PS1 mutants, they noticed that p53 activity was increased between two- and fourfold. It should be noted, however, that there is increasing evidence that many FAD PS mutations result in loss of γ-secretase function (see Bentahir et al., 2006 and related ARF Forum Discussion debating whether such mutations result in gain or loss of function). So how could loss of PS1 activity result in higher p53? The answer to that conundrum might come from studying the relationship between the two presenilins and p53 signaling. The authors found that the two PS isoforms have opposite effects on p53. Overexpression of PS1 decreased p53 expression and activity, while overexpression of PS2 increased p53. Interestingly, p53 may suppress expression of PS1 (see Roperch et al., 1998), suggesting that there is some cross-talk between the two presenilins. That cross-talk “could ultimately control the levels of γ-secretase-mediated AICD formation,” write the authors. In this regard, the increase in p53 elicited by PS1 mutations could be due to a compensatory increase in PS2 activity, which would fit with the loss-of-function theory for PS1 FAD mutations.

In a final caveat, the authors hammer another nail in the coffin of any potential γ-secretase-linked AD therapy. “The present report additionally suggests that chronic treatment with γ-secretase inhibitors could affect p53 drastically and, potentially therefore, ultimately lead to tumorigenicity,” the authors write.—Tom Fagan


  1. To keep apoptosis in check, it may be best to avoid the ups and downs of γ-secretase. Zebrafish lacking the γ-secretase component Pen-2 induce p53-dependent apoptosis throughout the body (Campbell et al., 2006), while mice lacking the Aph-1A component demonstrate increased apoptosis in the neuroepithelium and surrounding mesoderm (Serneels et al., 2005). Now, Alves da Costa et al. have reported that a loss of γ-secretase activity, or its substrates APP and APLP2, reduces the activity of p53. Their report has clearly demonstrated an involvement of γ-secretase in mediating p53-dependent cell death, and the regulation of these pathways is extremely complicated. Previous studies from the same group and others have shown that AD mutant presenilin causes cell death or an increased sensitivity to inducers of apoptosis in cultured cells. The current study has provided further evidence linking mutant presenilin to p53-dependent apoptosis.

    Whether γ-secretase inhibitors could affect p53 function and lead to tumorigenicity is an interesting topic. Alves da Costa et al. have shown that overexpression of the AICD of APP in cultured cells triggers p53-dependent apoptosis, and an earlier report from Jie Shen's group has demonstrated p53-dependent neuronal apoptosis induced by Notch activation (Yang et al., 2004). Since there are more than a dozen γ-secretase substrates that have been identified, apparently AICD and NICD are not the only ICDs derived from γ-secretase cleavage of its substrates. We do not know whether other ICDs could promote/inhibit p53-dependent apoptosis; therefore, some classes of γ-secretase inhibitors may possibly be in a position to prevent tumorigenesis.


    . Zebrafish lacking Alzheimer presenilin enhancer 2 (Pen-2) demonstrate excessive p53-dependent apoptosis and neuronal loss. J Neurochem. 2006 Mar;96(5):1423-40. PubMed.

    . Differential contribution of the three Aph1 genes to gamma-secretase activity in vivo. Proc Natl Acad Sci U S A. 2005 Feb 1;102(5):1719-24. PubMed.

    . Notch activation induces apoptosis in neural progenitor cells through a p53-dependent pathway. Dev Biol. 2004 May 1;269(1):81-94. PubMed.

  2. Didn't you ever think it was odd that there are 155 disease-causing mutations in PS1 but only 10 in PS2? To me, that smacks of a mechanistic difference between the two gene products. And the restriction to 10 PS2 mutations is more consistent with the stringent requirements of a gain-of-function than are the 155 willy-nilly mutations in PS1. So, statistically, it would seem to make sense that PS2 activity has some negative impact on health (e.g., p53 induction) and PS1 a positive one (e.g., p53 suppression).

  3. It's a very interesting question proposed by Tom Fagan. What is the immediate downstream target of AICD involved in the apoptotic pathway? I have suggested that the AICD target (KAI1) CD82 may be involved at an early stage (see comment [1]). It's of interest that Marreiros et al. report that KAI1 promoter activity is dependent on p53, junB and AP2; however, the Duriez group reports data that KAI1 expression is not significantly modulated by p53. Would the AICD be expected to influence expression of AP2 and junB? If not, might you then expect increased p53?

    See also:

    Comment by Mary Reid on Stellwagen et al., 2006


    . KAI1 promoter activity is dependent on p53, junB and AP2: evidence for a possible mechanism underlying loss of KAI1 expression in cancer cells. Oncogene. 2005 Jan 20;24(4):637-49. PubMed.

    . Absence of p53-dependent induction of the metastatic suppressor KAI1 gene after DNA damage. Oncogene. 2000 May 11;19(20):2461-4. PubMed.

  4. I would also like to propose that another p53 target gene which may be involved in AD is brain-specific angiogenesis inhibitor 1(BAI1) and that AICD may be a player in the antiangiogenic pathway. Bescond and Rahmani (1) find that DYRK1A interacts with a brain specific protein, phytanoyl-CoA alpha-hydroxylase-associated protein 1 which is reported to be a binding partner of BAI1.(2) DSCR1 is also overexpressed in AD and the Minani group (3) describes the secondary inhibition of NF-AT signalling as a natural brake in the angiogenic process. Might the beneficial effects of statins be due to the stimulation of angiogenesis?


    . Dual-specificity tyrosine-phosphorylated and regulated kinase 1A (DYRK1A) interacts with the phytanoyl-CoA alpha-hydroxylase associated protein 1 (PAHX-AP1), a brain specific protein. Int J Biochem Cell Biol. 2005 Apr;37(4):775-83. PubMed.

    . Characterization of mouse brain-specific angiogenesis inhibitor 1 (BAI1) and phytanoyl-CoA alpha-hydroxylase-associated protein 1, a novel BAI1-binding protein. Brain Res Mol Brain Res. 2001 Mar 5;87(2):223-37. PubMed.

    . Vascular endothelial growth factor- and thrombin-induced termination factor, Down syndrome critical region-1, attenuates endothelial cell proliferation and angiogenesis. J Biol Chem. 2004 Nov 26;279(48):50537-54. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Long-elusive Function for APP Cleavage Product Comes into View: It's Gene Transcription

Webinar Citations

  1. Gain or Loss of Function—Time to Shake up Assumptions on γ-Secretase in Alzheimer Disease?

Paper Citations

  1. . The intracellular domain of the beta-amyloid precursor protein is stabilized by Fe65 and translocates to the nucleus in a notch-like manner. J Biol Chem. 2001 Oct 26;276(43):40288-92. PubMed.
  2. . Correlates of p53- and Fas (CD95)-mediated apoptosis in Alzheimer's disease. J Neurol Sci. 1997 Nov 6;152(1):73-83. PubMed.
  3. . Insulin-degrading enzyme rapidly removes the beta-amyloid precursor protein intracellular domain (AICD). J Biol Chem. 2002 Apr 19;277(16):13389-93. PubMed.
  4. . Presenilin clinical mutations can affect gamma-secretase activity by different mechanisms. J Neurochem. 2006 Feb;96(3):732-42. PubMed.
  5. . Inhibition of presenilin 1 expression is promoted by p53 and p21WAF-1 and results in apoptosis and tumor suppression. Nat Med. 1998 Jul;4(7):835-8. PubMed.

Further Reading

Primary Papers

  1. . Presenilin-dependent gamma-secretase-mediated control of p53-associated cell death in Alzheimer's disease. J Neurosci. 2006 Jun 7;26(23):6377-85. PubMed.