The search is on for factors that interact with the complex protein cutter at the upstream end of the amyloid cascade, that is, γ-secretase, and the enzyme phospholipase D1 (PLD1) could be one of them, according to back-to-back papers released in PNAS online this week. Paul Greengard at Rockefeller University, New York, and colleagues report that the lipase, which hydrolyzes phosphatidylcholine to generate phosphatidic acid, independently affects two processes, the trafficking of amyloid-β precursor protein (AβPP)-loaded vesicles through the trans-Golgi network, and the presenilin-catalyzed cleavage of amyloid-β (Aβ) from AβPP. The findings suggest that the phospholipase could be a target for therapeutics.

Though the work was a collaboration among many labs in the U.S., Rockefeller’s Dongming Cai was lead author on both papers. In the first, Cai and colleagues report how the phospholipase restores AβPP trafficking in neurons harboring presenilin (PS) mutants that cause familial AD. They found that when PLD1 is overexpressed in neuroblastoma cells, it rescues vesicle budding that has been impaired by the expression of PS1ΔE9, a presenilin mutant that causes overproduction of Aβ42 (see related ARF mutation data pages). The rescue seems to depend on the enzymatic activity of the lipase because overexpression of PLD1 increased budding of AβPP vesicles from the tans-Golgi network by over twofold, whereas overexpression of catalytically inactive lipase did not. The authors also found that the PLD1 inhibitor 1-butanol prevents increases in AβPP vesicle budding that occur in PS1 knockout fibroblasts. This not only supports the idea that the catalytic activity of PLD1 is crucial for its effects on budding, it also suggests that this action of PLD1 is independent of presenilin.

Yet there is another side to the story. In the second paper, the researchers report that PLD1 does interact with presenilin, after all: It apparently inhibits the protease, reducing production of Aβ. This action, by contrast, does not depend on the catalytic activity of PLD1. The conclusions are based on the following data. First, Cai used coimmunoprecipitation experiments to show that PLD1 and PS1 interact in embryonic stem cells. The authors narrowed down the site of interaction to the C-terminal end of PS because antibodies to a loop in this region blocked the interactions. Next, the authors determined that overexpression of PLD1 reduces intracellular and secreted Aβ by nearly half. However, unlike in the trafficking experiments, catalytically inactive PLD1 was just as effective here, reducing production of Aβ by similar amounts. The inactive lipase also bound to the C-terminal of PS1.

Taken together, the two papers show that PLD1 has mechanistically distinct actions on vesicle budding and PS1 activity. How the lipase affects the former is unclear, though given the dependence on catalytic activity, it may well be related to levels of phosphatidylcholine or phosphatidic acid. And, while PLD1 can accelerate vesicle budding in PS-negative cells, the lipase’s impact on vesicle trafficking may depend to some degree on the protease, because Cai also found that PLD1 activity was significantly reduced in cells expressing certain PS FAD mutations.

As for the effect of PLD1 on AβPP processing, this seems to be due to binding of the lipase to presenilin, causing disruption of the γ-secretase complex. The authors found that overexpression of PLD1 caused dissociation of the γ-secretase components—PEN2, APH1, and nicastrin—from PS1.

What effect PLD1 might have in vivo, and in particular on the pathology of AD, if any, is uncertain because most of the reported experiments depended on overexpression of PLD1 in cultured cells. When the authors used RNAi to knock down PLD1 protein by around 70 percent, they observed almost a threefold jump in intracellular Aβ, suggesting that relatively modest losses of PLD1 activity could have a significant effect on AβPP processing. In addition, they found that transfecting PLD1 into cortical neurons expressing the M146V PS1 mutation restored neurite outgrowth. When one considers that there may be a vicious cycle brought on by FAD PS mutations—first inactivating PLD1, which then fails to inhibit γ-secretase, which then produces more Aβ—then maintaining PLD1 activity might go some way toward ameliorating AD pathology.—Tom Fagan


  1. Aβ generation strongly depends on lipids. First of all, APP is a membrane protein, defining its most proximate neighboring molecules; second, substrate turnover of the secretases is regulated by membrane lipid composition; and third, Aβ peptides are signaling molecules involved in cholesterol and sphingolipid homeostasis. Now the Greengard lab adds a new stone to this mosaic. PLD1, a phospholipase, apparently binds to PS1 and absence of PLD1 increases Aβ generation. The story is complex because at least two independent pathways are involved. The first pathway modifies assembly or stability of γ-secretase and is independent of PLD1 enzymatic activity; the other one strictly depends the phospholipase activity, altering APP trafficking in the presence of PS1 and overexpressed PLD1. Moreover, it changes neurite growth, but only in the presence of PS-FAD.

    Interestingly, PLD1 affects Aβ generation as much as it affects Notch cleavage. Gopal Thinakaran recently reported that NICD generation in adult cells, unlike Aβ, is produced outside of rafts. Does this indicate a role of PLD1 for embryonic processing of APP, which appears to take place outside of the raft? Clearly it will be important to see whether this interaction (and the altered trafficking/neurite sprouting) can be found with adult wild-type mice. Alternatively—and the γ-secretase components data may suggest this—PLD1 could act as an (anti?) cofactor during assembly of the γ-secretase complex. In such a case, PLD1 should modify total cellular γ-secretase activity with no selective impact on specific γ-secretase substrates.

    It is clear from these publications that PLD1 has functional interactions with PS and protein trafficking, including trafficking of APP. Neurite outgrowth is only affected when PS-FAD is overexpressed, and the authors conclude that this is due to altered APP trafficking. But is this really the only interpretation? The effect of PLD1 on APP trafficking was not assayed in the wild-type situation, and there are many other possible pathways, for example, altered PLD1 localization. Complicating interpretation further is that PS-FAD causes altered membrane fluidity, which could easily impair vesicle budding or trafficking for some cargo proteins. Does PLD1 "restore" this situation because it corrects membrane fluidity, or is there a close functional and mechanistic relation, as the interaction with PS may suggest? The multiple links PLD1 offers to APP biology are truly fascinating, and it will be exciting to see how this story develops over time.

  2. Several lines of evidence suggest that presenilins (PS) could contribute to both AβPP processing and trafficking to the membrane, but whether these two functions were related and intimately linked to the proposed catalytic activity of presenilins remained a matter of question. In these two back-to-back papers, the groups of Dr. Paul Greengard and Dr. Huaxi Xu interestingly suggest that phospholipase D1 (PLD1) could interact physically with PS, promote AβPP trafficking, and modulate Aβ production by apparently distinct mechanisms.

    First, the group convincingly demonstrates that endogenous PS1 physically interacts with PLD1 but not with other PLD members, and binds to this phospholipase via its cytoplasmic loop domain. Apparently, PS1 recruits PLD1 in the Golgi/TGN, since PLD1 distributes within both cytosolic and Golgi/TGN compartments in wild-type ES cells, while PS1 deficiency triggers diffuse and only cytosolic localization of PLD1. Interestingly, PLD1 overexpression reduced the levels of both secreted and intracellular Aβ and increased βCTF, while PLD1 reduction by antisense approach led to the opposite phenotype, that is, increase in Aβ levels and redution in βCTF-like immunoreactivity.

    By which mechanism could PLD1 trigger Aβ reduction? Greengard and colleagues suggest that this could occur via the disruption of the PS-dependent γ-secretase complex by PLD1. First, they show that PLD1 interacts physically with PS1 but not with Pen-2, another member of the γ-secretase complex. By using the anti-Pen-2 immunoprecipitation approach, it is shown that PLD1 overexpression reduces the interaction of Pen-2 with the other components of the complex, PS, Aph-1, and nicastrin.

    These data are interesting since the disruption of the complex is far from complete, but sufficient to trigger a drastic decrease of Aβ production. It had been suggested that perhaps a limited reduction of γ-secretase activity could lead to significant Aβ reduction without eliciting the deleterious effects on the production of other γ-secretase-derived products. Particularly, PLD1 overexpression was a smart way to examine whether one could partly diminish Aβ-production without affecting the Notch pathway. However, Greengard et al. unfortunately show that partial disruption of the γ-secretase complex also leads to the inhibition of NICD, the γ-secretase-derived product of Notch.

    The influence of PLD1 on AβPP processing is independent of its catalytic activity. Thus, catalytically inactive PLD1 (K898R) reduces secreted and intracellular Aβ production and apparently disrupts γ-secretase assembly, although to a lesser extent. These data are interesting and somewhat puzzling. Although Pen-2 immunoprecipitation of cells expressing PLD1K898 led to reduced immunoreactivities of the various components compared to control cells, the extent of inhibition appears variable depending on the component examined. While PLD1 and its mutant similarly reduced PS1 NTF and Aph-1 immunoreactivities, PS1-CTF and nicastrin expression were differentially affected. These data may be due to the fact that the stochiometry of the components of the γ-secretase complex is not 1/1/1/1, and that unlike wild-type PLD1, mutant PLD1 might have interacted differently with some of the components, either free or inside the complex.

    It remains that while catalytically inactive PLD1 reduces Aβ, this mutant was unable to affect AβPP trafficking. Therefore, this suggests that PLD1 could harbor two distinct functions. First, the promotion of AβPP trafficking, and second, the inhibition of γ-secretase activity.

    As do most interesting papers, these raise new questions. First, it was usually admitted that FAD mutations in PS generally lead to similar effects on both AβPP trafficking and γ-secretase processing. Here we are facing a new molecule that displays opposite effects on trafficking of AβPP and Aβ production. Is the increased trafficking to the membrane shortening transit in the Golgi/TGN and thereby, Aβ production? This would mean that one of the main sites of Aβ production indeed occurs intracellularly and not at the plasma membrane. Second, if PLD1 and its mutant both display "γ-secretase disruption" and trigger Aβ reduction while only wild-type PLD1 affects AβPP trafficking, what is the role of the PLD catalytic site? Since both processing and trafficking are affected when PLD1 is overexpressed, that is, when γ-secretase is partly disrupted, does that mean that only a fraction of the γ-secretase complex is necessary for underlying these two functions? Alternatively, is there a γ-secretase subcomplex specifically targeted by PLD1 (that would explain the rather selective effect of mutant PLD1 on certain γ-secretase components in the "disruption/immunoprecipitation" experiments)? Is the catalytic PLD1 interacting with a subcomplex of γ-secretase that participates only in processing?

    With cell biology papers, the question stands as to whether the observed cellular phenotype could account for in-vivo physiological mechanisms. In this context, it would be interesting to examine deeply the influence of PLD1 or its mutant in cells overexpressing PS-FAD mutants distinct from ΔE9, particularly because the PS1/PLD1 physical interaction was demonstrated on endogenous PS1, that is, intact protein harboring the integral cytoplasmic loop while the ΔE9-PS1 deletion truncates PS1 from a part of its cytoplasmic domain. Whether ΔE9-PS1 physically interacts with PLD1 has not been included in the article, although functionality of the system argues in favor of such an interaction. Finally, it would be of most interest to examine the effect of PLD1 inhibitors such as 1-Butanol in transgenic mice that overproduce Aβ to examine whether this inhibitor (if it’s not toxic per se) could accelerate amyloidogenesis and Aβ deposits.

    Overall, these two papers are very interesting and raise fundamental questions about the roles of presenilins, which, undoubtely, will be adressed very soon.

Make a Comment

To make a comment you must login or register.


Other Citations

  1. ARF mutation data pages

Further Reading


  1. . Regulation of cholesterol and sphingomyelin metabolism by amyloid-beta and presenilin. Nat Cell Biol. 2005 Nov;7(11):1118-23. PubMed.

Primary Papers

  1. . Phospholipase D1 corrects impaired betaAPP trafficking and neurite outgrowth in familial Alzheimer's disease-linked presenilin-1 mutant neurons. Proc Natl Acad Sci U S A. 2006 Feb 7;103(6):1936-40. PubMed.
  2. . Presenilin-1 uses phospholipase D1 as a negative regulator of beta-amyloid formation. Proc Natl Acad Sci U S A. 2006 Feb 7;103(6):1941-6. PubMed.