. Trans-dominant negative effects of pathogenic PSEN1 mutations on γ-secretase activity and Aβ production. J Neurosci. 2013 Jul 10;33(28):11606-17. PubMed.


Please login to recommend the paper.


  1. This new study from the Shen lab seeks to resolve the continuing controversy over the mechanism by which presenilin mutations cause autosomal-dominant, early onset familial AD (FAD). Toward this end, the experiments are designed to test a hypothesis first raised 10 years ago, that the mutant presenilins can affect wild-type presenilin so that the latter produces a greater proportion of Aβ42 to Aβ40. In that 10-year-old report (Shroeter EH et al., 2003), FAD mutations were installed into catalytically dead (Asp-mutant) presenilin-1 (PS1), to observe Aβ produced only from wild-type PS1. Despite this elegant design, statistically significant changes in wild-type PS1-generated Aβ42/Aβ40 by mutant PS1 were not seen.

    In the present study, effects of mutant PS1 on wild-type were investigated in the absence of the Asp mutation, expressing wild-type and mutant PS1 together into presenilin-null cells. On their own, the PS1 mutants showed effects on endoproteolysis of the PS1 N-terminal fragment (NTF)(which occurs upon assembly of PS1 with other γ-secretase components), on APP and Notch intracellular domain (ICD) production, as well as on Aβ40 and Aβ42 production. The conclusion that mutant PS1 can affect wild-type PS1 is based heavily on non-additive effects at these levels and on the assumption that wild-type and mutant PS1 are not simply competing for interaction with the other components of the γ-secretase complex. That assumption is based on a linear dose-response of PS1 NTF formation at a range of plasmid amounts. Whether this is true for the PS1 mutants or when wild-type and mutant PS1 are expressed together is not clear.

    Most problematic, however, is the evidence for a physical interaction between wild-type and mutant PS1. Wild-type and mutant PS1 appear capable of being co-immunoprecipitated (co-IP’d). However, the level of CHAPSO detergent used to solubilize isolated cell membranes was almost certainly above the critical micelle concentration (CMC). Thus, the observed co-IP would be due to simply pulling down micelles containing multiple presenilin/γ-secretase complexes. The CHAPSO concentration is described oddly as “4g/g protein”; it is impossible from this information to discern the actual solution concentration. In any event, this is a common error in the study of the γ-secretase complex, which has led to estimates of the size of the complex being as high as 2 MDa. Six years ago, our lab established unambiguously that such co-IPs do not occur below the CMC and that the IP’d γ-secretase complex is proteolytically active with only one PS1 per complex and with one of each of the other 3 components as well (Sato et al., 2007). The size of the purified, active complex, ~230 kDa, is consistent with this stoichiometry (Osenkowski et al., 2009).


    . A presenilin dimer at the core of the gamma-secretase enzyme: insights from parallel analysis of Notch 1 and APP proteolysis. Proc Natl Acad Sci U S A. 2003 Oct 28;100(22):13075-80. PubMed.

    . Active gamma-secretase complexes contain only one of each component. J Biol Chem. 2007 Nov 23;282(47):33985-93. PubMed.

    . Cryoelectron microscopy structure of purified gamma-secretase at 12 A resolution. J Mol Biol. 2009 Jan 16;385(2):642-52. PubMed.

  2. While we appreciate Michael Wolfe’s interest in our study, the validity of his commentary is undermined by a number of inaccuracies and misconceptions. Firstly, in objecting to our demonstration of co-immunoprecipitation (co-IP) of wild-type presenilin 1 (PS1) with mutant PS1, Wolfe curiously advocates biochemical analysis of the γ-secretase complex using detergent concentrations below the critical micellar concentration (CMC). To justify this approach, he cites a prior study from his group (Sato et al., 2007) that he claims “established unambiguously that such co-IPs do not occur below the CMC.” We believe this interpretation of his group’s findings is incorrect. Contrary to the claim that their analysis was performed with detergent concentrations below the CMC (see Sato et al., 2007), Wolfe and his colleagues performed co-IPs using a digitonin concentration that is 10-100 times higher than the CMC for this detergent (Matsumoto et al., 1978; Moore et al., 1992). Thus, the co-IP experiments in this study were actually performed on γ-secretase solubilized in digitonin micelles.

    Secondly, using detergent concentrations below the CMC means working with membrane proteins that are not solubilized. For this reason, it is standard practice to conduct biochemical investigations of membrane proteins using detergent concentrations above the CMC. Membrane protein solubilization typically occurs at or near the CMC, and detergent concentrations sufficiently above the CMC yield micelles containing a single functional protein. In contrast, detergent concentrations below the CMC yield large membrane fragments containing many proteins with detergent monomers bound at the periphery. Thus, adventitious co-IP of γ-secretase complexes due to colocalization in the same lipid/detergent structure would be a greater concern below the CMC, rather than above it.

    A third issue relates to detergent properties. Compared with nonionic detergents such as digitonin, zwitterionic bile salts such as CHAPSO are more effective at solubilizing proteins in their functional state while limiting non-specific aggregation. In addition, the higher CMC and lower aggregation number of CHAPSO relative to digitonin greatly reduce the likelihood that protein-detergent micelles will contain multiple proteins. Wolfe also questions our description of CHAPSO concentration in terms of the detergent-protein ratio. However, prior studies have shown that the detergent-protein ratio is a more important determinant of the efficiency of membrane protein solubilization than the detergent concentration itself (Hjelmeland, 1990; Thomas and McNamee, 1990). For the typical working range of protein concentrations, protein solubilization generally occurs at a detergent-protein ratio ≥ 1, which corresponds to a solution concentration near or above the CMC for most detergents.

    Fourth, Wolfe questions our demonstration that mutant PS1 can affect wild-type PS1 activity by raising the possibility that mutant PS1 may compete with wild-type PS1 for interaction with other components of γ-secretase. This possibility was addressed in our study, and we showed that such competition does not occur at the limiting levels of PS1 expression used in our analysis (Heilig et al., 2013). Notably, mutant PS1 could inhibit the activity of co-expressed wild-type PS1 activity without interfering with its expression or endoproteolysis. Moreover, beyond relying solely on non-additive effects, our study directly demonstrates the dominant-negative action of clinical PS1 mutations by showing that mutant PS1 can reduce the activity of co-expressed wild-type PS1 below the level displayed by wild-type PS1 alone. In such cases, the contribution of mutant PS1 to total γ-secretase activity is effectively worse than nothing (i.e. more deleterious than a null mutation). Collectively, the effects of PS1 mutations reported in our study satisfy the traditional criteria for dominant-negative or “antimorphic” mutations, and provide strong evidence that mutant PS1 can interfere with the catalytic activity of wild-type PS1.

    Our finding that mutant and wild-type PS1 can physically interact is consistent with other published data, including the higher-order assemblies of γ-secretase observed using various biochemical approaches. In their investigation of possible orthosteric interactions between PS1 molecules that could form a dimeric catalytic site, Schroeter et al. (2003) presented evidence for cross-linking between wild-type PS1 N-terminal fragments. Moreover, recent crystallographic analysis of an archaeal PS1 homolog revealed a tetrameric structure (Li et al., 2013). The most likely reason why Wolfe’s group detected no PS1 multimerization is that their analysis was performed under conditions that disrupted physiological interactions between γ-secretase monomers.


    . Trans-dominant negative effects of pathogenic PSEN1 mutations on γ-secretase activity and Aβ production. J Neurosci. 2013 Jul 10;33(28):11606-17. PubMed.

    . Solubilization of native membrane proteins. Methods Enzymol. 1990;182:253-64. PubMed.

    . Structure of a presenilin family intramembrane aspartate protease. Nature. 2013 Jan 3;493(7430):56-61. PubMed.

    . Effect of digitonin concentration on regeneration of cattle rhodopsin. Biochim Biophys Acta. 1978 Feb 9;501(2):257-68. PubMed.

    . A simple, highly sensitive assay for measurement of digitonin during receptor solubilization. J Neurosci Methods. 1992 Jul;43(2-3):153-6. PubMed.

    . Active gamma-secretase complexes contain only one of each component. J Biol Chem. 2007 Nov 23;282(47):33985-93. PubMed.

    . A presenilin dimer at the core of the gamma-secretase enzyme: insights from parallel analysis of Notch 1 and APP proteolysis. Proc Natl Acad Sci U S A. 2003 Oct 28;100(22):13075-80. PubMed.

    . Purification of membrane proteins. Methods Enzymol. 1990;182:499-520. PubMed.

  3. Ray Kelleher's response to my comment on the Heilig et al. study reveals a serious misunderstanding about how detergents are employed in the study of membrane proteins. The major problem is the conflation of the detergent's role in membrane solubilization and its role in membrane protein function. While it is true that detergents must be above the critical micellar concentration (CMC) to effectively solubilize membranes and their incorporated proteins, detergent concentrations below the CMC are often required for functional studies. Indeed, the first reported γ-secretase enzyme assay (Li et al., 2000) involved solubilization in 1 percent CHAPSO (above the CMC of ~0.5 percen) but with subsequent dilution to 0.25 percent CHAPSO for the functional assay of Aβ production from recombinant substrate. This assay has been used by us and others for more than 13 years now and continues to be the standard assay for proteolytic function of the γ-secretase complex.

    Furthermore, co-immunoprecipitation (co-IP) conditions also include solutions used to wash the coimmunoprecipitated material. Kelleher is correct that in our earlier work (Sato et al., 2007) we performed co-IP of γ-secretase complexes in digitonin above its CMC, but we washed twice in 0.25 percent CHAPSO, below the CMC, and carried out analysis of proteolytic function of the co-IP'd complexes in 0.25 percent CHAPSO. We found that differentially tagged PS1 molecules were not co-IP'd but that the γ-secretase complexes were nevertheless proteolytically active. In the report by Heilig et al., no such functional analysis of the co-IP'd material was carried out. From Kelleher's description above, and the experimental section of the report, it appears that both the co-IP and wash conditions were far above the CMC for CHAPSO, so it is not surprising that both tagged forms of PS1 were observed, since both forms would be present in the micelles. There is no evidence that each micelle would contain only one functional protein. Moreover, if assays for Aβ production from recombinant substrate using the co-IP'd material had been carried out in the very high CHAPSO concentration, no proteolytic activity would have been observed (see Okochi et al., 2013 for the latest demonstration of this).

    Finally, the statement that zwitterionic detergents, such as CHAPSO, are better than nonionic detergents, such as digitonin, for solubilizing membrane proteins is a generalization. The reality is that choice of detergent is determined empirically on a case-by-case basis, as what works well with one membrane protein may not work well for another (Arachea et al., 2012). Solubilization and functional analyses of the γ-secretase complex have been carried out by many laboratories in many parts of the world for many years now, and these provide investigators with valuable groundwork that can be used for further investigations.


    . Detergent selection for enhanced extraction of membrane proteins. Protein Expr Purif. 2012 Nov;86(1):12-20. PubMed.

    . Presenilin 1 is linked with gamma-secretase activity in the detergent solubilized state. Proc Natl Acad Sci U S A. 2000 May 23;97(11):6138-43. PubMed.

    . γ-secretase modulators and presenilin 1 mutants act differently on presenilin/γ-secretase function to cleave Aβ42 and Aβ43. Cell Rep. 2013 Jan 31;3(1):42-51. PubMed.

    . Active gamma-secretase complexes contain only one of each component. J Biol Chem. 2007 Nov 23;282(47):33985-93. PubMed.

Make a Comment

To make a comment you must login or register.