I am quite impressed by this paper, which dives very deep and is of high quality. The observations raise two important issues:

1. Given the apparent central role of the presenilins in memory, and given the fact that loss of presenilins causes neurodegeneration, is it possible that presenilin dysfunction at least partially contributes to the neurodegenerative process in some familial forms of Alzheimer's? In my opinion, we indeed still have to learn a lot about the fundamental processes of neurodegeneration in Alzheimer's disease, and this paper contributes significantly to that aim.

2. What are the implications of the findings for drug development programs trying to target presenilin/γ-secretase? The second issue is not the main message of this paper, but obviously it is a question that will be raised by many researchers and managers in companies. My opinion is that a genetic knockout and a pharmacological modulation of a protein are two very different situations. For example, the HMGCoA reductase knockout gives a very early lethal phenotype—still statins are one of the...  Read more

I am quite impressed by this paper, which dives very deep and is of high quality. The observations raise two important issues:

1. Given the apparent central role of the presenilins in memory, and given the fact that loss of presenilins causes neurodegeneration, is it possible that presenilin dysfunction at least partially contributes to the neurodegenerative process in some familial forms of Alzheimer's? In my opinion, we indeed still have to learn a lot about the fundamental processes of neurodegeneration in Alzheimer's disease, and this paper contributes significantly to that aim.

2. What are the implications of the findings for drug development programs trying to target presenilin/γ-secretase? The second issue is not the main message of this paper, but obviously it is a question that will be raised by many researchers and managers in companies. My opinion is that a genetic knockout and a pharmacological modulation of a protein are two very different situations. For example, the HMGCoA reductase knockout gives a very early lethal phenotype—still statins are one of the most widely used drugs in the world. Thus, I would say that it is absolutely necessary to continue with the γ-secretase inhibitor programs, especially with programs that try to develop "modulators" of the protease (like the NSIADs.) (Editor's note: see ARF related news story). It is important to find compounds that discriminate between APP processing and Notch processing. Once those are found, only clinical trials will allow us to make real decisions about the viability of these drugs. We have few options in Alzheimer's disease treatment; thus, it would be unwise to drop too quickly any potential approach for treatment.

View all comments by Bart De Strooper

This is a key paper for the actual and future understanding of the pathogenesis of Alzheimer's disease. With very interesting, complete and provocative findings, the paper shows that in adult brain the total lack of presenilin function leads to early functional alterations (LTP and memory failure) that some months later are followed by morphological and structural changes (loss of neurons). The paper also highlights the importance of the study of AD models from an experimental interdisciplinary approach in a longitudinal way.

View all comments by Diego Forero

We are very intrigued by the two recent papers (Feng et al., 2004; Saura et al., 2004) showing neurodegeneration and tau hyperphosphorylation in PS null mouse brains. Although at first sight the non-amyloid neuropathology doesn’t appear to be directly relevant to AD, these studies nevertheless clearly challenge the currently widely accepted view that AD-related PS mutations are gain of function mutations (as measured by amyloid-β deposition/secretion).

In this context, it is of interest that we have recently published a paper (Dermaut et al., 2004) reporting a novel PS1 mutation in a patient with a pure tauopathy (Pick’s disease, a subtype of FTD) but without any detectable amyloid deposits. Moreover, preliminary evidence on the molecular nature of this mutation suggests that it might act as a partial loss-of-function allele due to aberrant exon splicing, which would be in agreement with the loss of function/dominant negative mechanism that has been proposed for another PS1 mutation (insArg352) associated with FTD, though not pathologically confirmed (Amtul et al., 2002;...  Read more

We are very intrigued by the two recent papers (Feng et al., 2004; Saura et al., 2004) showing neurodegeneration and tau hyperphosphorylation in PS null mouse brains. Although at first sight the non-amyloid neuropathology doesn’t appear to be directly relevant to AD, these studies nevertheless clearly challenge the currently widely accepted view that AD-related PS mutations are gain of function mutations (as measured by amyloid-β deposition/secretion).

In this context, it is of interest that we have recently published a paper (Dermaut et al., 2004) reporting a novel PS1 mutation in a patient with a pure tauopathy (Pick’s disease, a subtype of FTD) but without any detectable amyloid deposits. Moreover, preliminary evidence on the molecular nature of this mutation suggests that it might act as a partial loss-of-function allele due to aberrant exon splicing, which would be in agreement with the loss of function/dominant negative mechanism that has been proposed for another PS1 mutation (insArg352) associated with FTD, though not pathologically confirmed (Amtul et al., 2002; Tang-Wai et al., 2002). In addition, we have shown that lowered neuronal expression of PS is genetic risk factor for early-onset AD (Theuns et al., 2003), Although still highly speculative, these studies together appear to raise the exciting possibility that, in humans, partial loss of PS function could result in primarily tau-mediated neurodegenerative pathways.

References:
Amtul, Z., Lewis, P. A., Piper, S., Crook, R., Baker, M., Findlay, K., Singleton, A., Hogg, M., Younkin, L., Younkin, S. G., et al. (2002). A presenilin 1 mutation associated with familial frontotemporal dementia inhibits γ-secretase cleavage of APP and notch. Neurobiol Dis 9, 269-273. Abstract

Dermaut, B., Kumar-Singh, S., Engelborghs, S., Theuns, J., Rademakers, R., Saerens, J., Pickut, B. A., Peeters, K., van den Broeck, M., Vennekens, K., et al. (2004). A novel presenilin 1 mutation associated with Pick's disease but not β-amyloid plaques. Ann Neurol 55, 617-626. Abstract

Feng, R., Wang, H., Wang, J., Shrom, D., Zeng, X., and Tsien, J. Z. (2004). Forebrain degeneration and ventricle enlargement caused by double knockout of Alzheimer's presenilin-1 and presenilin-2. Proc Natl Acad Sci U S A. 2004 May 25;101(21):8162-7. Abstract

Saura, C. A., Choi, S. Y., Beglopoulos, V., Malkani, S., Zhang, D., Shankaranarayana Rao, B. S., Chattarji, S., Kelleher, R. J., 3rd, Kandel, E. R., Duff, K., et al. (2004). Loss of presenilin function causes impairments of memory and synaptic plasticity followed by age-dependent neurodegeneration. Neuron 42, 23-36. Abstract

Tang-Wai, D., Lewis, P., Boeve, B., Hutton, M., Golde, T., Baker, M., Hardy, J., Michels, V., Ivnik, R., Jack, C., and Petersen, R. (2002). Familial frontotemporal dementia associated with a novel presenilin-1 mutation. Dement Geriatr Cogn Disord 14, 13-21. Abstract

Theuns, J., Remacle, J., Killick, R., Corsmit, E., Vennekens, K., Huylebroeck, D., Cruts, M., and Van Broeckhoven, C. (2003). Alzheimer-associated C allele of the promoter polymorphism -22C>T causes a critical neuron-specific decrease of presenilin 1 expression. Hum Mol Genet 12, 869-877. Abstract

View all comments by Bart Dermaut

This study confirms some of our results. It reproduces the neurodegeneration phenotype we saw in an independently generated presenilin double conditional KO. Feng et al. report an increase in apoptosis; we tested for this but did not see it in our mice. For more, see Carlos Saura's remarks below.

View all comments by Jie Shen

This work largely confirms the earlier conditional KO study by Saura et al., 2004. There is a dramatic reduction in cortical tissue and a massive hydrocephalic-like appearance in animals that lack both presenilins. I wonder how significant and relevant this is to presenilin mutations in people. I also wonder what would happen if other key integral membrane proteins were similarly deleted in such animals. The wild-type animals are not adequate controls. I agree with Bart De Strooper that neither study should discourage attempts to modify secretase activities for potential treatments.

View all comments by Vincent Marchesi

Feng et al. describe cortical degeneration in the forebrain of mice lacking the Alzheimer’s disease genes presenilin (PS) 1 and 2. The mice help to define essential roles of presenilins in the maintenance of brain function. These results confirm previous observations published recently, which showed hippocampal and cortical neuronal degeneration caused by loss of presenilin function in the mouse forebrain (Saura et al., 2004). At 10 months of age, the double PS1/PS2 conditional knockout (dKO) mice by Feng et al. showed reduced body weight and increased locomotor activity in the open-field arena, which the authors attribute to the extensive degeneration of the cortex (thinner cortical layers and enlarged lateral ventricles) occurring in these mice. As previously described, mice lacking both presenilins in the forebrain exhibit signs of neuronal degeneration such as reduced neuronal markers (NeuN), shrinkage of neuronal processes and increased phosphorylation of tau.

The authors attribute the massive neuronal...  Read more

Feng et al. describe cortical degeneration in the forebrain of mice lacking the Alzheimer’s disease genes presenilin (PS) 1 and 2. The mice help to define essential roles of presenilins in the maintenance of brain function. These results confirm previous observations published recently, which showed hippocampal and cortical neuronal degeneration caused by loss of presenilin function in the mouse forebrain (Saura et al., 2004). At 10 months of age, the double PS1/PS2 conditional knockout (dKO) mice by Feng et al. showed reduced body weight and increased locomotor activity in the open-field arena, which the authors attribute to the extensive degeneration of the cortex (thinner cortical layers and enlarged lateral ventricles) occurring in these mice. As previously described, mice lacking both presenilins in the forebrain exhibit signs of neuronal degeneration such as reduced neuronal markers (NeuN), shrinkage of neuronal processes and increased phosphorylation of tau.

The authors attribute the massive neuronal death to apoptosis, however, report no data to quantify the neuron loss and/or apoptotic cells in the dKO mice. This is an important issue because cells expressing FAD-linked PS have been previously shown to be more sensitive to apoptotic insults. In fact, we were unable to detect apoptosis by TUNEL or cleaved caspase-3 staining in our PS dKO mice. Therefore, even if some apoptosis occurs and cannot be detected by classical methods, it is unlikely to account for the massive neuron loss (~25 percent) seen in the PS dKO mice at nine months of age. This may explain why, in the present investigation, the majority of neurons with abnormal nucleic morphology are not TUNEL-positive. Furthermore, it is difficult to reconcile the observations that apoptotic events (cleavage of caspase-3) and upregulation of anti-apoptotic factors (Bcl-2, XIAP and CIAP) occur simultaneously in the dKO mice.

The molecular mechanism(s) that link PS to these or other survival pathways is not explored in the present work and awaits further investigation. In my opinion, the investigation of molecular mechanisms of neuronal degeneration regulated by presenilins will probably represent valuable avenues for AD therapeutics.

View all comments by Carlos A. Saura

By Akihiko Takashima In this paper, Feng and colleagues showed that neurodegeneration occurred in mice deficient in both presenilins. Overlapping results have been reported in Neuron by Jie Shen’s lab. Their results also showed that aged PS1/2 double knockout mice exhibited brain atrophy and memory deficits. These results suggest that presenilin is required to maintain neurons during aging. Presenilin may be indeed a pre-senile factor, and in this sense, these presenilins studies are significant contributions to our understanding of brain aging.

While brain aging is a prerequisite for the development of AD, the precise marker of the aging remains unknown. However, based on Braak’s study, I would suggest that this aging starts when neurofibrillary tangles (NFTs) form in the entorhinal cortex. AD results when Abeta accelerates the rate of NFT formation, and NFTs spread into the limbic and isocortical regions, resulting in AD. We previously observed that increases in the level of A1-42 do not correlate...  Read more

By Akihiko Takashima In this paper, Feng and colleagues showed that neurodegeneration occurred in mice deficient in both presenilins. Overlapping results have been reported in Neuron by Jie Shen’s lab. Their results also showed that aged PS1/2 double knockout mice exhibited brain atrophy and memory deficits. These results suggest that presenilin is required to maintain neurons during aging. Presenilin may be indeed a pre-senile factor, and in this sense, these presenilins studies are significant contributions to our understanding of brain aging.

While brain aging is a prerequisite for the development of AD, the precise marker of the aging remains unknown. However, based on Braak’s study, I would suggest that this aging starts when neurofibrillary tangles (NFTs) form in the entorhinal cortex. AD results when Abeta accelerates the rate of NFT formation, and NFTs spread into the limbic and isocortical regions, resulting in AD. We previously observed that increases in the level of A1-42 do not correlate with the onset age of FAD with PS1 mutation (Murayama et al., 1997); these results suggest that NFT-like tau accumulation may be another determining factor for the onset age of AD. This would also explain why some PS1 mutations accelerate NFT formation and neuronal loss without accelerating A deposition (Gomez-Isla et al., 1999). Thus, in instances where an FAD mutation in presenilin causes a loss of function this accelerates NFT formation, and A1-42 production accelerates it further, so the cause of the early onset of AD may be through PS1 mutation.

Neuronal loss occurs accompanied by tau hyperphosphorylation in PS1/2- deficient mice. This suggests that hyperphosphorylation of tau itself, or a mechanism associated with it, facilitates neuronal death during brain aging. That underlying mechanism might also include presenilins, however presenilins may not be directly involved in tau hyperphosphorylation or neuronal apoptosis as both phenomena occur at a later stage. This finding may be the key that reveals the connection between brain aging and AD.

View all comments by Akihiko Takashima

Following quickly on the heels of the excellent Neuron paper by Carlos Saura and colleagues is this complementary work by Ruiben Feng et al. Both groups of investigators crossed a knock-out PS2 KO line, made by traditional transgenic methods (and revealing no overt structural or functional aberrations) with a conditional and spatial (forebrain) ko of PS1. The doubly homozygous mice, lacking any expression of these presenilins, developed severe neurodegenerative changes as they aged. There were some differences in the genetic backgrounds of these lines (B6/CBA for Feng et al. and B6/129 for Saura et al.)

The new Feng et al. paper emphasizes the striking ventricular enlargement—reminiscent of what one sees clinically in patients with normal pressure hydrocephalus (although their mice apparently did not show the gait disturbances and urinary incontinence characteristic of that mysterious but relatively common geriatric problem). They also emphasize an active apoptotic pathway, a phenotype not described in the...  Read more

Following quickly on the heels of the excellent Neuron paper by Carlos Saura and colleagues is this complementary work by Ruiben Feng et al. Both groups of investigators crossed a knock-out PS2 KO line, made by traditional transgenic methods (and revealing no overt structural or functional aberrations) with a conditional and spatial (forebrain) ko of PS1. The doubly homozygous mice, lacking any expression of these presenilins, developed severe neurodegenerative changes as they aged. There were some differences in the genetic backgrounds of these lines (B6/CBA for Feng et al. and B6/129 for Saura et al.)

The new Feng et al. paper emphasizes the striking ventricular enlargement—reminiscent of what one sees clinically in patients with normal pressure hydrocephalus (although their mice apparently did not show the gait disturbances and urinary incontinence characteristic of that mysterious but relatively common geriatric problem). They also emphasize an active apoptotic pathway, a phenotype not described in the Saura et al. paper.

Both groups believe that their findings support the hypothesis of a loss of function as the basis of AD associated with mutations at the presenilin loci. But I believe that such a conclusion is premature. First of all, the findings, although consistent with several important AD phenotypes, are not specific for AD. Second, if a loss of function were the predominant basis for these rare autosomal dominant forms of AD, one might expect to find some autosomal recessive pedigrees within populations with high frequencies of consanguinity. I spent five months of field work in India trying to find such a pedigree, to no avail, although I must admit that ascertainment was extremely difficult.

Conditional knockouts are the wave of the future, but I hope we shall see more conditional knockdowns, as different levels of “tweaking” relevant pathways may be more relevant to the differential expression of a disorder that takes many decades to emerge in our species. A wonderful example of the striking impact of modest tweaking is found in the exciting paper by Rolf Postina, Anja Schroeder et al. in the current Journal of Clinical Investigation , which reminds us of the therapeutic potential of modest enhancements of the α-secretase cleavage of the beta amyloid precursor protein.

View all comments by George M. Martin

This is an excellant paper showing the effects of knocking out PS-1 and PS-2. The major differenct it seems between the Feng el al. paper and Saura et al., is the question of how the neurodegeneration is taking place? Feng el al., suggest an apoptotic pathway, which seemed to be not evident in the Saura study. To address this contradiction, we suggest using a site-directed caspase-cleavage antibody to tau to test in both of these animal models. Our lab (in collaboration with Carl Cotman's group) have synthesized such an antibody and shown that it labels neurons in a triple transgenic mouse model developed by Frank LaFerla's group. Because this antibody recognizes a product of caspase-cleavage (tau) that is stable and accumulates over time, it is an excellant and extremely sensitive marker to indicate caspase activation. A similar antibody has been developed based on a study by Gamblin et al., 2003. The Gamblin antibody to caspase-cleaved tau is actually commerically available but it is a monoclonal antibody, so...  Read more

This is an excellant paper showing the effects of knocking out PS-1 and PS-2. The major differenct it seems between the Feng el al. paper and Saura et al., is the question of how the neurodegeneration is taking place? Feng el al., suggest an apoptotic pathway, which seemed to be not evident in the Saura study. To address this contradiction, we suggest using a site-directed caspase-cleavage antibody to tau to test in both of these animal models. Our lab (in collaboration with Carl Cotman's group) have synthesized such an antibody and shown that it labels neurons in a triple transgenic mouse model developed by Frank LaFerla's group. Because this antibody recognizes a product of caspase-cleavage (tau) that is stable and accumulates over time, it is an excellant and extremely sensitive marker to indicate caspase activation. A similar antibody has been developed based on a study by Gamblin et al., 2003. The Gamblin antibody to caspase-cleaved tau is actually commerically available but it is a monoclonal antibody, so it may not be suitable to examine for caspase activation in mouse models.

References:
Rohn, T.T., Rissman, R.A., Davis, M.C., Kim, Y., Cotman, C.W. and Head, E. (2002). Caspase-9 activation and caspase cleavage of tau in the Alzheimer’s disease brain. Neurobiol Dis 11:341-354. Abstract

Rohn, T.T., Rissman, R.A., Head, E. and Cotman, C.W. (2002). Caspase Activation in the Alzheimer’s Disease Brain: Tortuous and Torturous. Drug News & Perspectives 15(9): 549-557. Abstract

View all comments by Troy Rohn

Presenilin (PS) 1 and 2 rose to prominence in the mid 90’s, when mutations in these genes were shown to be associated with sub-forms of Alzheimer’s disease characterized by the early onset of symptoms and their dominant inheritance. Many years of intense research have revealed that PS1 and PS2 proteins constitute the catalytic core of γ-secretase, a multi-subunit protease responsible for the intramembrane cleavage of several integral membrane proteins. The most relevant physiological substrates of the γ-complex are the Notch receptors, which participate in a multitude of signalling pathways crucial during development. Indeed, the combined genetic ablation of both PS1 and PS2 in mice results in a phenotype that closely resembles the Notch-1 phenotype. The presenilins/γ-secretase also cleave the amyloid precursor protein (APP) leading to the release of a small peptide, Aβ, which accumulates in brains of AD patients and is supposed to be the toxic entity in the disease. All the familial Alzheimer’s disease (FAD)-mutations in the PS genes selectively increase the generation of a...  Read more

Presenilin (PS) 1 and 2 rose to prominence in the mid 90’s, when mutations in these genes were shown to be associated with sub-forms of Alzheimer’s disease characterized by the early onset of symptoms and their dominant inheritance. Many years of intense research have revealed that PS1 and PS2 proteins constitute the catalytic core of γ-secretase, a multi-subunit protease responsible for the intramembrane cleavage of several integral membrane proteins. The most relevant physiological substrates of the γ-complex are the Notch receptors, which participate in a multitude of signalling pathways crucial during development. Indeed, the combined genetic ablation of both PS1 and PS2 in mice results in a phenotype that closely resembles the Notch-1 phenotype. The presenilins/γ-secretase also cleave the amyloid precursor protein (APP) leading to the release of a small peptide, Aβ, which accumulates in brains of AD patients and is supposed to be the toxic entity in the disease. All the familial Alzheimer’s disease (FAD)-mutations in the PS genes selectively increase the generation of a slightly longer Aβ peptide (42 amino acids instead of 40) that is more amyloidogenic. Yet, how these FAD mutations cause the disease and the normal physiological functions of presenilins, especially in brain, are still issues of investigation. (presenilins are gamma-secretases, catalytic subunits, cleave notch andThis paper by Reuben Feng et al. sheds some light on these issues.

To investigate the function of PS in the adult CNS, the authors generated a conditional knock out line in which the PS1 gene is specifically ablated in the forebrain of adult mice, therefore overcoming the limitation of the early embryonic lethality of conventional PS1 null mice. In a previous publication they showed that these animals are normal overall, except for a deficiency in enrichment-induced neurogenesis in the dentate gyrus (1). To analyse the consequences of deleting both presenilins in adult brain, they now crossed the PS1 conditional KO mice with conventional PS2-deficient mice that by themselves have no obvious phenotype. Importantly, forebrain-specific ablation of PS function was sufficient to cause neuronal atrophy, which translates into a drastic shrinkage of the cortex and significant enlargements of ventricles, severe astrogliosis, and elevated levels of tau phosphorylation.

Brain degeneration was accompanied by overt abnormal behaviors. The authors performed all the analysis on 10-month-old mice, that is, four months after the knockout of PS1 reaches completion and the age when the animals start to exhibit reduced body weight. Although it is clear from this work that the coordinate functions of PS1 and PS2 are essential for the maintenance of the adult brain structure and function, the age-dependent changes that occur in the degenerating brains and the molecular pathways impaired by the absence of presenilins are not analyzed in depth.

A recent study by Carlos Saura et al., did address these questions (2). These authors also generated a related conditional DKO mouse model to study PS function in the adult CNS. They could show that loss of PS function resulted in defects in long-term potentiation (LTP) and destabilization of synaptic connections prior to the onset of detectable neuropathological changes. Mild memory impairments were seen as early as two months. Analysis of mice at 6, 9, and 16 months showed an age-dependent synaptic, dendritic, and neuronal degeneration with astrogliosis and tau phosphorylation first being evident at the age of nine months.

Saura et al. also characterized the molecular mechanisms that underline these effects. For example, lack of presenilins selectively impaired the delivery to synapses of the NMDAR subunits NR1 and NR2. Saura et al. could show that NMDARs interact physically with PS1, suggesting that PS1 plays a role in the trafficking and localization of NMDARs to synaptosomes [Editor’s note: see also ARF Live Discussion on PS and axonal transport.] Synaptic and dendritic localization of CaMKII, a downstream effector of NMDARs in LTP induction, were also reduced. Moreover, presenilin deficiency resulted in decreased levels of CBP, probably because of reduced Notch signalling, which subsequently led to a reduction in the expression of genes regulated by CREB/CBP. And CRE-dependent gene expression has been shown to play a role in the consolidation of long-term forms of synaptic plasticity and memory. The hyper phosphorylation of tau that occurs at later stages and likely contributes to the ongoing neurodegeneration, correlated with increased levels of p25, an activator of the tau kinase Cdk5. [Editor’s note: see also ARF related news story.]

Taken together, the results have at least two important implications for AD. First, they raise the possibility that a partial loss of PS function would contribute, together with elevated Aβ42 levels, to neurodegeneration in FAD cases. This would contradict the general view that FAD-linked mutations in PS represent a gain of toxic function, which has been suggested based on the fact that they selectively increase Aβ42 cleavage normally at the expenses of Aβ40. A second implication from these studies concerns the consideration of presenilins as viable drug targets. Inhibition of presenilin function as therapeutic strategy in AD may, as suggested by Saura et al., accelerate rather than attenuate the development of memory loss and neurodegeneration. However, it should be taken into account that the single PS1 knock out led to significant reduction in Aβ with no major consequences for the mice’s well-being. It should also be noticed that severe phenotypes associated with genetic ablation of certain genes do not necessarily predict the value of this gene as a therapeutic target. One such example is HMG-coA, which causes severe phenotypes in mice when the gene is ablated; yet millions of people worldwide take inhibitors of this enzyme to reduce their cholesterol levels, without major side effects (3).

In conclusion, the current papers provide further insight into the function of presenilins in the CNS. Whether this really should influence drug programs trying to develop therapies against γ-secretase is, however, doubtful.

References:
1. Feng, R., Rampon, C., Tang, Y. P., Shrom, D., Jin, J., Kyin, M., Sopher, B., Miller, M. W., Ware, C. B., Martin, G. M., Kim, S. H., Langdon, R. B., Sisodia, S. S., and Tsien, J. Z. (2001) Neuron 32, 911-926. Abstract

2. Saura, C. A., Choi, S. Y., Beglopoulos, V., Malkani, S., Zhang, D., Shankaranarayana Rao, B. S., Chattarji, S., Kelleher, R. J., 3rd, Kandel, E. R., Duff, K., Kirkwood, A., and Shen, J. (2004) Neuron 42, 23-36. Abstract

3. Marjaux, E., Hartmann, D., and De Strooper, B. (2004) Neuron 42, 189-192. Abstract

View all comments by Diana Dominguez