Taking away a samurai’s sword may prevent destruction, but it also limits his prowess. That’s one way to look at a new study on presenilins. These mighty enzymes deliver the final cut to unleash the Aβ peptides that form the amyloid plaques that gum up the brains of Alzheimer disease patients. Yet curbing their power seems to cut off other vital functions—namely, promoting learning and memory and helping neurons survive to old age—according to recent work by Jie Shen, Brigham and Women’s Hospital, Boston, and colleagues. Reported in this week’s Journal of Neuroscience, the findings also underscore the challenges of developing inhibitors of γ-secretase, the protein unit within which presenilins operate, as possible AD treatments.

The view of presenilins has widened since the membrane proteins were pigeonholed as villains on the business end of γ-secretase. Recent work has placed presenilins in a different light—as guardians of neuronal health. That research suggested new roles for presenilins in maintaining proper intracellular calcium signaling. Calcium dysregulation has garnered increased attention for its possible role in AD and other neurodegenerative disorders (for review, see Bezprozvanny, 2009). Last summer, scientists reported that presenilins interact with and activate sarco-endoplasmic reticulum Ca2+-ATPase (SERCA) pump proteins, and that SERCA activity seems to influence Aβ production (Green et al., 2008 and ARF related news story). Other work has suggested that presenilins function as ER calcium leak channels (Tu et al., 2006 and ARF related news story), and that several familial AD mutations specifically disrupt this Ca2+ leak function of presenilin 1 (Nelson et al., 2007). Some of these activities could be independent of presenilins’ γ-secretase activity.

Previous work from Shen’s group has reinforced the idea that presenilins are indispensable for cognition and neuronal integrity. Her team conditionally knocked out the two presenilin genes (PS1 and PS2) in mouse postnatal forebrain, and saw a progressive increase in memory impairment and neurodegeneration as these mice got older (Saura et al., 2004 and ARF related news story). But this study left the researchers pondering what, mechanistically, was responsible for presenilin’s essential roles in memory and neuronal survival. The current study tackles one aspect of this issue—whether these functions of presenilin depend on its γ-secretase-dependent or -independent activities.

Led by Shen and first author Katsuhiko Tabuchi, the researchers chose a similar knockout strategy. “Rather than chase after 20 published substrates, many of which are probably not physiological, we decided to first do a genetic dissection because it would be very conclusive,” Shen told ARF. Using the same promoter that drove postnatal forebrain-specific inactivation of PS1/PS2 in their previous study, her team engineered a conditional knockout (cKO) mouse whose cortical excitatory neurons lacked expression of nicastrin, one of three other γ-secretase subunits besides presenilin. (The remaining two are presenilin enhancer 2 [Pen-2] and anterior pharynx defective 1 [Aph-1].)

By and large, the nicastrin cKO mice reproduced the striking phenotype of the lab’s PS1/PS2 animals. At two months of age, both mouse strains had sharply reduced PS1 and Pen-2 levels, though Aph-1 expression had hardly changed. In addition, cortical lysates from each mouse line had normal levels of full-length amyloid precursor protein (APP) but whopping amounts of the C-terminal APP fragment, as predicted by the reduced γ-secretase-mediated cleavage of APP in those cells. These data suggest that the absence of nicastrin destabilizes most components of the γ-secretase complex, compromises its activity, and reduces its APP-cleaving ability.

The biochemical changes in the nicastrin cKO mice were associated with learning and memory deficits that showed up as early as two to three months of age, when the mice still lacked detectable changes in brain volume or cortical neuron numbers relative to the control group. At six to nine months, the nicastrin cKO animals’ performance on memory tests continued to plummet, and the researchers found evidence of neurodegeneration (white and gray matter loss in Nissl-stained sagittal brain sections) and synapse dysfunction (decreased MAP2 and synaptophysin immunoreactivity in neocortex and hippocampus). Neurodegeneration often comes with heightened inflammation, and signs of this also appeared in the nicastrin cKO mice. In Western and immunostaining analyses, their cortical lysates had higher levels of the reactive astrocyte marker GFAP (glial fibrillary acidic protein), as well as elevated Iba-1 (ionized calcium-binding adapter molecule 1), a protein specifically expressed in brain microglia, compared with control mice. The nicastrin cKO animals also had increased apoptosis (greater numbers of TUNEL-positive and caspase-3-positive cells) in the neocortex—measurable at two months of age, more severe at six months—and higher levels of hyperphosphorylated tau at six months.

All together, the data convincingly argue that presenilins support memory formation and neuronal survival through γ-secretase-dependent mechanisms, Shen said. The findings may also offer insight into ongoing discussions about the dearth of neurodegeneration in many APP-overexpressing mouse strains. “People have argued in the past that the reason APP transgenic mice don't have significant neurodegeneration is because the mice are resistant to neuronal loss,” Shen said. However, the new study shows that “the mouse brain is not very resistant to neurodegeneration if you have targeted the right gene. It highlights the importance of presenilins and nicastrin in neuronal survival,” she said. Furthermore, the authors write that their “data—including the increase in tau phosphorylation, and the widespread apoptosis—are consistent with the notion that the neurodegeneration induced by inactivation of γ-secretase subunits resembles the neurodegeneration observed in AD.”

Other scientists are not as convinced that the γ-secretase models exhibit AD-like neurodegenerative pathways. “Accumulation of amyloid has been postulated to play a critical role in AD pathogenesis and abnormal neuronal Ca2+ signaling has also been implicated as one of the pathogenic pathways involved in AD,” wrote Ilya Bezprozvanny of University of Texas Southwestern Medical Center, Dallas, in an e-mail to ARF. He noted that PS1/PS2 conditional double knockout and nicastrin conditional knockout mice do not produce amyloid, and the current study does not reveal Ca2+ signaling dysfunction in the nicastrin cKO mice. “Thus, it remains an open question whether these γ-secretase knockout mice are a faithful model for the neuronal cell death in AD,” he wrote (see full comment below).

Bart De Strooper, at K.U. Leuven in Belgium, and Bezprozvanny raised the possibility that presenilin expression levels in aging neurons of the nicastrin cKO mice could be reduced enough to affect non-proteolytic functions of presenilin. De Strooper pointed out, though, that “the Ca2+ leakage function is maintained in Aph-1 deficient cells (Tu et al., 2006), indicating that presenilin can exert that function outside of the [γ-secretase] complex.” (See full comment below.)

Recent work from De Strooper’s lab added to the emerging picture of γ-secretase as a multi-functional complex. Published in Science several months ago, that study showed that selectively knocking out the B/C isoforms of the Aph-1 component of γ-secretase rescues cognitive defects and neurodegeneration in an AD mouse model (Serneels et al., 2009 and ARF related news story). The data suggest that selective inhibition of γ-secretase components may produce the desired therapeutic benefits without the side effects from inactivating the entire complex.

Scientists seem to agree that unraveling the mechanisms underlying presenilin-mediated neuronal survival remains a top priority for future studies. “The challenge will be now to determine which signaling pathways, downstream from γ-secretase, are involved in neuronal death in the aging brain,” wrote Philippe Marambaud, Feinstein Institute for Medical Research, Manhasset, New York, in an e-mail to ARF (see full comment below). He gives a vote of confidence to signal transduction by cadherins—adhesion proteins that are expressed at mature synapses, are critical for synaptic plasticity, and are cleaved by γ-secretase in neurons upon NMDA receptor stimulation (Marambaud et al., 2003).—Esther Landhuis


Make a Comment

Comments on this content

  1. The paper by Tabuchi at al. is a fascinating study that uses genetic approach to investigate the importance of γ-secretase activity for neuronal survival. The authors generated floxed nicastrin mice and crossed them with the α-CaMKII-Cre transgenic driver line that expresses Cre recombinase specifically in the excitatory neurons of postnatal forebrain. The resulting nicastrin cKO mouse line appears to be essentially normal at two months of age. By the age of six and nine months, nicastrin cKO mice developed age-dependent memory deficits, displaying apoptotic loss of cortical neurons and active gliosis. The phenotype of aging nicastrin cKO mice is very similar to the phenotype of presenilin cDKO mice that Jie Shen’s group previously described (1). The authors conclude that 1) γ-secretase activity is essential for neuronal survival; and that 2) γ-secretase independent functions of presenilins, such as ER Ca2+ leak function (2), are not important for neuronal survival in this context.

    The experimental design is very elegant and leaves no doubt that γ-secretase function is essential for neuronal survival. Although the exact mechanism that connects γ-secretase activity with neuronal survival is not established, these findings have obvious and important implications for developing and clinically testing γ-secretase inhibitors. The conclusion about the non-essential role of ER Ca2+ leak function of presenilins is somewhat less strong. The authors assume that ER Ca2+ signaling is normal in adult neurons from nicastrin cKO mice, but did not directly test it in the paper. In the previous studies our laboratory demonstrated that ER Ca2+ signals are normal in Aph TKO MEF cells (2), so it is already known that inactivation of γ-secretase per se does not have a major effect on ER Ca2+ handling. However, it is possible that levels of presenilin expression in aging neurons from nicastrin cKO mice are sufficiently reduced to affect ER Ca2+ signaling. When cortical lysates from two-month-old mice were analyzed by Western blotting, levels of presenilin-1 were reduced approximately twofold in nicastrin cKO mice. Unfortunately, the expression levels of presenilins are not shown for six- and nine-month cortical lysates from nicastrin cKO mice.

    Another critical issue to consider is whether memory deficits and cortical neuronal loss observed in PS cDKO and nicastrin cKO mice faithfully replicate neuronal dysfunction and eventual death in AD. There is no doubt that previous (1) and present (Tabuchi at al., 2009) data from Jie Shen’s lab indicate that γ-secretase activity is important for neuronal survival. However, it remains unclear if the pathway leading to neuronal cell death in γ-secretase KO models is the same as in AD. This is a critical question that still needs to be addressed. Accumulation of amyloid has been postulated to play a critical role in AD pathogenesis (3). Abnormal neuronal Ca2+ signaling has also been implicated as one of the pathogenic pathways involved in AD (4). PS cDKO and nicastrin cKO mice do not produce amyloid and nicastrin cKO mice presumably have normal Ca2+ signaling. Thus, an open question is if neuronal cell death observed in these γ-secretase knockout mice is a faithful model for neuronal cell death in AD. In any case, the new results obtained by the authors provide very interesting and important insights into a connection between γ-secretase activity and neuronal survival. Understanding the mechanistic basis responsible for this connection will be an extremely important future task.


    . Loss of presenilin function causes impairments of memory and synaptic plasticity followed by age-dependent neurodegeneration. Neuron. 2004 Apr 8;42(1):23-36. PubMed.

    . .

    . The amyloid hypothesis of Alzheimer's disease: progress and problems on the road to therapeutics. Science. 2002 Jul 19;297(5580):353-6. PubMed.

    . .

  2. In this paper, Tabuchi et al. report a novel mouse that specifically lacks the nicastrin gene in excitatory neurons. Nicastrin cKO mice showed memory impairment and age-dependent cortical neuron loss, similar phenotypes to the PS cDKO mice. Of note, the nicastrin cKO mice showed significant memory impairment at two months of age. At this stage, the gross brain morphology was normal, suggesting that “functional” defects would have already occurred by the deletion of nicastrin gene, presumably the complete loss of γ-secretase activity in neurons. These data implicate that the neuronal γ-substrate is functionally important in learning and memory without significant synaptic loss or neuron death. Thus, unveiling the molecular mechanism whereby the cKO neurons showed functional defects is an important issue to consider the physiological role of the γ-secretase activity in the brain.

  3. I think this is a very elegant study that reveals the fundamental role of nicastrin in neuronal integrity and memory in adult mice. This work also confirms in vivo, in the adult brain, that nicastrin is essential for the stabilization and activity of the γ-secretase complex. These results are nicely in line with the notion that complete or partial loss of function of presenilins is, per se, neurotoxic.

    The challenge will be now to determine what are the signaling pathways—downstream from γ-secretase—involved in neuronal death in the aging brain. Cadherins may represent attractive candidates. Indeed, the cadherin family of cell-cell adhesion proteins is abundantly expressed at mature synapses, is critical for synaptic plasticity, and is cleaved by γ-secretase in neurons upon NMDA receptor stimulation (Marambaud et al., 2003). It is, therefore, reasonable to think that a loss of synaptic cadherin cleavage by γ-secretase may lead over time to defects in synaptic plasticity and neuronal integrity and thus may contribute to the phenotype observed in these mice.

    Another important question relates to the integrity in these mice of the CREB/CBP transcriptional pathway, which appeared to be significantly compromised in the PS cKO mice (Saura et al., 2004).


    . A CBP binding transcriptional repressor produced by the PS1/epsilon-cleavage of N-cadherin is inhibited by PS1 FAD mutations. Cell. 2003 Sep 5;114(5):635-45. PubMed.

    . Loss of presenilin function causes impairments of memory and synaptic plasticity followed by age-dependent neurodegeneration. Neuron. 2004 Apr 8;42(1):23-36. PubMed.

  4. This is a very elegant knockout study reinforcing previous work of Jie Shen published in Neuron, which showed that presenilin 1 and 2 double deficient mice display a progressive neurodegenerative disorder.

    In contrast with their previous paper in Neuron, Shen and colleagues now conclude that the neurodegeneration they see in both PS1 and 2 double KO mice, and nicastrin single KO mice is due to a γ-secretase defect, i.e., the loss of proteolytic function. This is part of an ongoing debate as to what extent postulated functions of presenilin outside the γ-secretase complex contribute to the overall phenotype of presenilin deficient mice. While I tend to believe that the neurodegeneration observed in their studies is indeed reflecting a real γ-secretase defect, the current paper is not conclusive in that regard. Indeed, knockout of nicastrin also destabilizes presenilin, and could theoretically affect functions of presenilin independent of its proteolytic function. However, I agree with Shen and colleagues that their current interpretation is the most likely one, as the major defect in nicastrin deficient cells is the loss of proteolytic activity, while some presenilin level is still maintained, which could fulfill these postulated other functions. For instance the Ca2+ leakage function is maintained in APh1 deficient cells (Tu et al., 2006) indicating that presenilin can exert that function outside of the complex.

    The big question, as the authors discuss in their manuscript, is the identification of the substrate that is responsible for the neurodegenerative phenotype. This is a very difficult question to answer, given the many different substrates, and the possibility that any of the γ-secretase substrates which accumulate in the Nct knockout mice could theoretically contribute to this phenotype.

    It is clear that γ-secretase as a drug target is not an easy one. We published recently that it is possible to knock out specifically the APh1B-γ-secretase in the brain of mice without neurodegenerative changes, and with the potential to clear Aβ peptide from the brain (Serneels et al., 2009). This indicates that a partial inhibition of the complex in brain is feasible and with acceptable side effects.


    . .

    . gamma-Secretase heterogeneity in the Aph1 subunit: relevance for Alzheimer's disease. Science. 2009 May 1;324(5927):639-42. Epub 2009 Mar 19 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Pump It Up—Presenilins Linked to ER SERCA Activity
  2. Presenilins Open Escape Hatch for ER Calcium
  3. The Senility-Presenilin Connection Turned Upside Down
  4. Double Paper Alert—Keystone Presentations Now in Press

Paper Citations

  1. . .
  2. . .
  3. . .
  4. . .
  5. . Loss of presenilin function causes impairments of memory and synaptic plasticity followed by age-dependent neurodegeneration. Neuron. 2004 Apr 8;42(1):23-36. PubMed.
  6. . gamma-Secretase heterogeneity in the Aph1 subunit: relevance for Alzheimer's disease. Science. 2009 May 1;324(5927):639-42. Epub 2009 Mar 19 PubMed.
  7. . A CBP binding transcriptional repressor produced by the PS1/epsilon-cleavage of N-cadherin is inhibited by PS1 FAD mutations. Cell. 2003 Sep 5;114(5):635-45. PubMed.

Further Reading