Zhao et al. present a novel and interesting set of findings that shed light on one of the most intriguing features of the γ-secretase enzyme: its dependence on the formation of a high-molecular-weight complex. This study provides further support for the involvement of nicastrin in assembly, maturation, and stabilization of the γ-secretase complex rather than in substrate recognition or as a crucial component for catalytic activity of the complex. The results presented in this article are in accordance with our previously published study (Chavez-Gutierrez et al., 2008) and our current working hypothesis for γ-secretase.

Remarkably, this report shows that a nicastrin-less γ-secretase complex, consisting of PS1/APH1a/PEN2, displays indistinguishable catalytic properties relative to the mature γ-secretase (containing nicastrin). Moreover, this nicastrin-less complex also requires ectodomain shedding of the substrate prior to catalysis, demonstrating that nicastrin does not participate in the recognition of...  Read more

Zhao et al. present a novel and interesting set of findings that shed light on one of the most intriguing features of the γ-secretase enzyme: its dependence on the formation of a high-molecular-weight complex. This study provides further support for the involvement of nicastrin in assembly, maturation, and stabilization of the γ-secretase complex rather than in substrate recognition or as a crucial component for catalytic activity of the complex. The results presented in this article are in accordance with our previously published study (Chavez-Gutierrez et al., 2008) and our current working hypothesis for γ-secretase.

Remarkably, this report shows that a nicastrin-less γ-secretase complex, consisting of PS1/APH1a/PEN2, displays indistinguishable catalytic properties relative to the mature γ-secretase (containing nicastrin). Moreover, this nicastrin-less complex also requires ectodomain shedding of the substrate prior to catalysis, demonstrating that nicastrin does not participate in the recognition of the short substrate amino terminus. Interestingly, the signal peptide peptidase (SPP), an intramembrane protease that is most closely related to PS, does not need a cofactor protein for catalytic activity but does require prior substrate shedding, suggesting that recognition/binding of the substrate amino terminus by the γ-secretase complex may be carried out by presenilin, although the participation of Aph1 and/or Pen2 cannot be ruled out.

Additionally, given that the extremely unstable “nicastrin-less” γ-secretase complex maintains approximately 50 percent of the wild-type enzyme activity, this result strongly suggests that nicastrin is not crucial for substrate recognition/binding. Nevertheless, our data and the findings presented in this article cannot exclude another contribution of nicastrin to γ-secretase activity and/or specificity, although no biochemical data have been published to support this assumption.

The model of nicastrin as the substrate receptor for the γ-secretase complex is exclusively supported by the putative interaction between the Glu333 and the N-terminus of the substrate. In contrast, our published data has shown that mutations in the putative binding pocket, including Glu332 (Glu333 in human), affect the levels of the mature γ-secretase complex in the mutant cell lines, but they do not have an effect on the specific activity of the complex. Therefore, our results exclude the participation of the Glu332 in the activity of the complex and indicate that the observed effects could be due to a problem in the assembly or maturation (stability) of the mutant complexes. Importantly, the results presented by Zhao et al. elegantly corroborate our findings and present a “simplified” version of the γ-secretase complex that advances our understanding of its structure-function.

γ-secretase has been considered as a potential drug target for Alzheimer disease. The proposal of nicastrin as a gatekeeper of the enzyme complex presented the opportunity to approach the γ-secretase “drug targetability” from a different angle. However, the fact that two independent studies present data that do not support this model not only reopens the discussion about the function of nicastrin in the γ-secretase complex, but also adds a question mark to this drug-targeting strategy.

View all comments by Lucia Chavez-Gutierrez

Another beautiful work on the biochemistry of γ-secretase complex from the group of Bart de Strooper.

This connection to tetraspanins offers key insights into the "raft" association of γ-secretase. The raft link to APP processing is very weak and highly contended. Lipid rafts are dynamic platforms that could be, to a large extent, biochemically purified by detergent extraction methods.

Typically Triton X-100 (in concentration ranging from 0.3 percent to 1 percent) insoluble fractions of membranes represent the raft fraction. However, tetraspanin microdomains (TEM) differ from the "canonical" rafts in that these are triton soluble but are insoluble to other milder detergents such as CHAPS or BRIJ. This study clearly demonstrates a direct link between tetraspanin microdomains and γ-secretase, thereby providing evidence that much of the γ-secretase activity is probably confined to tetraspanin microdomains.

An (obvious) interesting question is, How about Notch cleavage? While γ-secretase cleavage of β-CTF seems to depend on its interaction...  Read more

Another beautiful work on the biochemistry of γ-secretase complex from the group of Bart de Strooper.

This connection to tetraspanins offers key insights into the "raft" association of γ-secretase. The raft link to APP processing is very weak and highly contended. Lipid rafts are dynamic platforms that could be, to a large extent, biochemically purified by detergent extraction methods.

Typically Triton X-100 (in concentration ranging from 0.3 percent to 1 percent) insoluble fractions of membranes represent the raft fraction. However, tetraspanin microdomains (TEM) differ from the "canonical" rafts in that these are triton soluble but are insoluble to other milder detergents such as CHAPS or BRIJ. This study clearly demonstrates a direct link between tetraspanin microdomains and γ-secretase, thereby providing evidence that much of the γ-secretase activity is probably confined to tetraspanin microdomains.

An (obvious) interesting question is, How about Notch cleavage? While γ-secretase cleavage of β-CTF seems to depend on its interaction with the tetraspanin web, it would be interesting to see if there is a lateral compartmentalization in γ-secretase (non-TEM vs. TEM γ-complexes) that could determine substrate specificity. In my opinion, this systematic interactome analysis elegantly shows that Aβ production, though primarily achieved by only three players (APP, BACE1, and γ-secretase complex), is far too complex in the cellular context.

View all comments by Lawrence Rajendran

This is an excellent study by the De Strooper lab, using a biochemical approach to identify proteins that associate with the γ-secretase complex. The work was rigorously done, with important controls to identify interactions specific to γ-secretase (e.g., parallel purification of the presenilin-like protease SPPL3). The authors confirm several proteins previously reported to interact with γ-secretase, such as TMP21, β-catenin and Rab11, further strengthening confidence in their method. Most interesting was the identification of tetraspanin proteins, which are important in many basic cellular activities, including formation of certain lipid raft-like microdomains. Indeed, γ-secretase components co-distribute with tetraspanin proteins in a sucrose density gradient.

Knockdown of certain associated tetraspanin-web proteins by RNAi led to modest but significant decreases in amyloid-β-protein levels and increases in APP C-terminal fragments that are γ-secretase substrates. Conversely, overexpression can lead to increased amyloid production....  Read more

This is an excellent study by the De Strooper lab, using a biochemical approach to identify proteins that associate with the γ-secretase complex. The work was rigorously done, with important controls to identify interactions specific to γ-secretase (e.g., parallel purification of the presenilin-like protease SPPL3). The authors confirm several proteins previously reported to interact with γ-secretase, such as TMP21, β-catenin and Rab11, further strengthening confidence in their method. Most interesting was the identification of tetraspanin proteins, which are important in many basic cellular activities, including formation of certain lipid raft-like microdomains. Indeed, γ-secretase components co-distribute with tetraspanin proteins in a sucrose density gradient.

Knockdown of certain associated tetraspanin-web proteins by RNAi led to modest but significant decreases in amyloid-β-protein levels and increases in APP C-terminal fragments that are γ-secretase substrates. Conversely, overexpression can lead to increased amyloid production.

Overall, these findings are important contributions to our understanding of the cell biology of γ-secretase, identifying partner proteins that can modulate protease activity and affect amyloid production. Going forward, it will be worthwhile to investigate possible roles of tetraspanin genes in the pathogenesis of AD or whether the levels of the encoded proteins are otherwise altered. Targeting tetraspanins to prevent or treat AD, however, is probably not advised, because these proteins affect other γ-secretase-mediated proteolytic events (syndecan-3, N-cadherin, APLP-2, and ADAM10, shown in this new report); they are probably also critical for Notch proteolysis and signaling. Inhibition of Notch signaling leads to toxic effects that must be avoided in targeting γ-secretase, whether directly or indirectly, for AD.

View all comments by Michael Wolfe

This represents a careful and significant study with clean results. It probably explains why γsecretase loses activity in Triton-X100 and helps in a more thorough functional characterization of the enzyme.

View all comments by Kumar Sambamurti

This is an excellent study from the lab of Gang Yu, the original discoverer of nicastrin as a presenilin partner needed for γ-secretase activity when he was with Peter St. George-Hyslop in Toronto. In his new lab at UT Southwestern, Yu has nailed down an essential role for nicastrin in substrate recognition. To date, virtually nothing has been known about the specific biochemical role of nicastrin in γ-secretase activity. Nicastrin (NCT) is needed for assembly and maturation of the protease complex, including presenilin endoproteolysis into N-terminal fragment (NTF) and C-terminal fragment (CTF) subunits, and the NCT transmembrane domain is critical for these events. However, all the action (substrate binding and catalysis) has so far appeared to be taking place on presenilin. Recent work from our lab (Kornilova et al., 2005) located a substrate docking site at the interface between the two presenilin subunits, at a site distinct from the active site, which is also located at the NTF/CTF interface. The implication is that...  Read more

This is an excellent study from the lab of Gang Yu, the original discoverer of nicastrin as a presenilin partner needed for γ-secretase activity when he was with Peter St. George-Hyslop in Toronto. In his new lab at UT Southwestern, Yu has nailed down an essential role for nicastrin in substrate recognition. To date, virtually nothing has been known about the specific biochemical role of nicastrin in γ-secretase activity. Nicastrin (NCT) is needed for assembly and maturation of the protease complex, including presenilin endoproteolysis into N-terminal fragment (NTF) and C-terminal fragment (CTF) subunits, and the NCT transmembrane domain is critical for these events. However, all the action (substrate binding and catalysis) has so far appeared to be taking place on presenilin. Recent work from our lab (Kornilova et al., 2005) located a substrate docking site at the interface between the two presenilin subunits, at a site distinct from the active site, which is also located at the NTF/CTF interface. The implication is that substrate docks and then squeezes between the presenilin subunits to access the internal active site, which contains water and the two catalytic aspartates (also see ARF related news story. We suggested that presenilin is the γ-secretase component that directly interacts with the substrate transmembrane domain, and nothing in the new study from the Yu lab contradicts this idea. However, we also suggested that the primary role of nicastrin, along with Aph-1 and Pen-2, is to render PS competent for proteolysis, but Yu and his colleague have clearly demonstrated that nicastrin also plays an essential role in substrate binding, specifically interacting with the substrate N-terminus.

In the new study, the evidence for a role in substrate recognition is overwhelming. First, the nicastrin ectodomain interacts directly with APP- and Notch-based membrane stubs, and expression of the nicastrin ectodomain alone interferes with γ-secretase cleavage of these substrates. Second, an aminopeptidase-like (AP) region in nicastrin is essential for interaction with substrate. This region lacks residues that would be needed for peptidase activity but apparently still retains residues (e.g., a critical aspartate) capable of substrate binding. Consistent with the requirement for the AP region, blocking the amino-terminus of the substrate in a variety of ways prevented interaction with nicastrin and γ-secretase proteolysis. Even N-formylation, a very subtle structural change, rendered the substrate uncleavable. Thus, it is clear that the nicastrin ectodomain directly interacts with the N-terminus of substrates and that this interaction is essential for proteolysis. It is interesting to note, however, that a small fluorogenic peptide was proteolyzed by AP-mutant nicastrin the same as wild-type, demonstrating that the mutant complex is still a fully active protease. Nevertheless, endogenous substrates, with their N-termini sticking out of the membrane, must pass the nicastrin gatekeeper.

Nicastrin may be a gatekeeper, but the order of substrate binding events is unclear: Although Yu and colleagues suggest a model where the nicastrin-substrate interaction leads to substrate docking on presenilin, substrate binding to nicastrin and presenilin may be virtually simultaneous or even in the reverse order (presenilin first, nicastrin second). In the context of the full γ-secretase complex, substrate may be inaccessible to the AP domain of nicastrin until docking on presenilin. In this scenario, nicastrin's role would be to draw the substrate further into the complex and help shepherd it from docking site to active site. However, these are matters for future study and do not take anything away from what is a very important advance in our understanding of γ-secretase.

View all comments by Michael Wolfe

This is an excellent paper.

View all comments by Bart De Strooper

The strength of this paper is the imaginative use of carefully crafted recombinant peptides to study a complicated process that is difficult to approach using conventional cell biological techniques. But one also has to keep in mind that no "native" molecules are actually studied under in vivo conditions. Recombinant C99 peptides are often studied as a more accessible form of the amyloid-β precursor protein (AβPP), the physiological substrate of γ-secretase, and when they are expressed in living cells, one can be reasonably confident that they are mimicking the endogenous molecules. But a consideration in this study is whether they also behave the same way in detergent extracts. Following cleavage by β-secretases, the natural C99 peptides are likely to be dimers in situ, possibly sequestered in lipid raft-like domains, and still attached to an elaborate cytoskeletal network in the adjacent cytoplasm. We have no idea how recombinant-derived peptides are arranged in detergent extracts. If they exist as small micelles, as is likely, are they sticky? Do they bind...  Read more

The strength of this paper is the imaginative use of carefully crafted recombinant peptides to study a complicated process that is difficult to approach using conventional cell biological techniques. But one also has to keep in mind that no "native" molecules are actually studied under in vivo conditions. Recombinant C99 peptides are often studied as a more accessible form of the amyloid-β precursor protein (AβPP), the physiological substrate of γ-secretase, and when they are expressed in living cells, one can be reasonably confident that they are mimicking the endogenous molecules. But a consideration in this study is whether they also behave the same way in detergent extracts. Following cleavage by β-secretases, the natural C99 peptides are likely to be dimers in situ, possibly sequestered in lipid raft-like domains, and still attached to an elaborate cytoskeletal network in the adjacent cytoplasm. We have no idea how recombinant-derived peptides are arranged in detergent extracts. If they exist as small micelles, as is likely, are they sticky? Do they bind other proteins that are present in the various cell extracts, such as APP or a presenilin, or some active factor yet unknown? One would also like to know the size and properties of the C99-nicastrin complexes; their behavior on sucrose gradients and by gel filtration might be informative, as well as some idea of the strength of their interactions.

The functional assays involving cleavage activities with various recombinant substrates are revealing, but the activities seem to be very low, requiring many hour-long incubations at 37 degrees. The use of cells derived from nicastrin (Nct) knockouts to assay the Nct mutants is another potential complication. They are described as being from embryos that lack γ-secretase, yet they have undergone some development presumably without Notch activation. Are there compensating mechanisms operating in these cells? The claim that the amino terminal amino acid of C99 is critical for Nct recognition needs more experimental support. Can this interaction be blocked by small synthetic peptides derived from the amino terminus of C99?

One has to raise these questions, since similar studies from the Wolfe lab also provide a convincing case that substrate binding and catalysis of C99 and C83 take place at specific sites on presenilin, with nicastrin relegated to a supporting role.

View all comments by Vincent Marchesi

It has been known for a while that the ectodomain of type I membrane proteins needs to be proteolytically trimmed before the proteins can be further processed by γ-secretase. This led to the speculation that γ-secretase—and more specifically, the nicastrin subunit within the complex—must somehow be able to act as a molecular ruler and measure the length of the ectodomain of the γ-secretase substrates. Just how this can happen remained unclear. The elegant and convincing work by Shah and colleagues shows that, in fact, nicastrin can act as the molecular ruler. Their appealing model proposes that nicastrin sticks out of the membrane like a crane, binds the substrate, and shifts it to its docking or processing site in the γ-secretase complex. The free N-terminus of a substrate protein seems to be the primary determinant for recognition. Thus, nicastrin should be able to bind substrates regardless of their primary sequence—as long as they have a short ectodomain and contain a transmembrane domain. This model fits well with previous data by us and...  Read more

It has been known for a while that the ectodomain of type I membrane proteins needs to be proteolytically trimmed before the proteins can be further processed by γ-secretase. This led to the speculation that γ-secretase—and more specifically, the nicastrin subunit within the complex—must somehow be able to act as a molecular ruler and measure the length of the ectodomain of the γ-secretase substrates. Just how this can happen remained unclear. The elegant and convincing work by Shah and colleagues shows that, in fact, nicastrin can act as the molecular ruler. Their appealing model proposes that nicastrin sticks out of the membrane like a crane, binds the substrate, and shifts it to its docking or processing site in the γ-secretase complex. The free N-terminus of a substrate protein seems to be the primary determinant for recognition. Thus, nicastrin should be able to bind substrates regardless of their primary sequence—as long as they have a short ectodomain and contain a transmembrane domain. This model fits well with previous data by us and others, showing that γ-secretase has a relaxed sequence specificity. In future studies, it will be exciting to study in molecular detail how the γ-substrate can move (or be pushed) for proteolytic processing into the active site of the γ-secretase complex.

View all comments by Stefan Lichtenthaler

I find this article very informative, however, it is merely educational. While I find it interesting, it is not by finding every molecular binding site and protein involved in AD pathogenesis that we will get us closer to a cure. I love molecular and cellular mechanisms of life... in the words of the great pilot D.P. Davies: "let's get on with it."

View all comments by Jacob Mack