. Statins reduce amyloid-beta production through inhibition of protein isoprenylation. J Biol Chem. 2007 Sep 14;282(37):26832-44. PubMed.


Please login to recommend the paper.


Make a Comment

To make a comment you must login or register.

Comments on Primary Papers and News

  1. Statins on the Brain
    This paper by Gary Landreth’s group casts important new light on the regulation of APP processing by statins and the actions of statins on small GTPases. Previous data demonstrate that statins inhibit APP processing. The mechanism is thought to be mediated by isoprenylation. When HMG-CoA reductase is inhibited by more than 90 percent, the amount of substrate flowing through the cholesterol synthetic pathway decreases to a point where isoprenylation also decreases because of decreases in geranylgeranyl pyrophosphate and farnesylfarnesyl pyrophosphate. Because small GTPases such as Rac and Rho need isoprenylation to associate with membranes, their activity declines.

    Landreth’s team has now investigated this issue in more detail. They make the striking observation that the sensitivity of GTPases to statins differs depending on the GTPase. Cdc42 is highly sensitive, Rac and Rab1b GTPases are moderately sensitive, and Rab4/5b GTPases are relatively insensitive. The mechanism underlying these differences might lie in differential sensitivity or distribution of geranylgeranyl and farnesylfarnesyl transferases (1). Clearly, there is much to be investigated. The differential sensitivity is important for two reasons. The first point to note is that simvastatin reaches a concentration in the brain of about 300-500 nM (this is a new figure provided by Gibson Wood’s team) (2). Rac and Rab1b both respond to 200 nM simvastatin, but Rab4 and Rab5b do not. The second point to note is that Rho GTPases regulate APP processing; previous reports suggest that Rab1b specifically regulates APP processing, which suggests that it should be possible to inhibit APP processing with statins at doses that might not impair other cellular functions (3,4).

    These data open up a number of new avenues for understanding the regulation of APP and should provide new and useful tools for modulating APP processing. I can’t resist putting in one clinical note about this issue, though. I recently published a study indicating that simvastatin is associated with similar reductions in both dementia and Parkinson disease (5). If we assume that the reductions are real and occur through similar mechanisms in both diseases, then the data suggest that the mechanism is not mediated by inhibiting Aβ accumulation because Aβ has not been implicated in the pathophysiology of Parkinson disease. These results are consistent with findings by several investigators indicating that statin use is not associated with a reduction in Aβ levels in CSF. So, statins might be very useful in the arsenal of medications used to manage neurodegenerative disease, but whether the mechanism of action involves changes in APP metabolism is subject to debate.


    . Differential prenylation of proteins as a function of mevalonate concentration in CHO cells. Arch Biochem Biophys. 1993 Mar;301(2):210-5. PubMed.

    . Chronic administration of statins alters multiple gene expression patterns in mouse cerebral cortex. J Pharmacol Exp Ther. 2005 Feb;312(2):786-93. PubMed.

    . Retention of the Alzheimer's amyloid precursor fragment C99 in the endoplasmic reticulum prevents formation of amyloid beta-peptide. J Biol Chem. 2001 Jun 8;276(23):20267-79. PubMed.

    . Statins cause intracellular accumulation of amyloid precursor protein, beta-secretase-cleaved fragments, and amyloid beta-peptide via an isoprenoid-dependent mechanism. J Biol Chem. 2005 May 13;280(19):18755-70. PubMed.

    . Simvastatin is associated with a reduced incidence of dementia and Parkinson's disease. BMC Med. 2007;5:20. PubMed.

  2. Too Much ROCK and (Rho)ll gives Rac a Bad Rap
    The importance of the itinerary that APP takes around the cell, and its residence time at each stop along the way, has been evident since the protein was first shown to be concentrated in the trans-Golgi network (TGN); yet, only a small fraction of APP is processed in the TGN. The localization of the secretases is largely post-TGN, with α on the plasma membrane (PM), and β in endosomes primarily but also with a bit in the TGN. Hence, sorting at the TGN is crucial (reviewed in Small and Gandy, 2006) in specifying the fraction of APP that colocalizes with α and is cleaved in a non-amyloidogenic fashion between Aβ residues 16 and 17 to generate sAPPα. TGN sorting also determines how much APP encounters β-secretase, either by a direct pathway via TGN clathrin-coated vesicles (TGN CCVs) to the endocytic pathway, or else the APP molecules on the PM that escape α can be internalized by PM CCVs and enter the endocytic pathway thus.

    We have recently provided evidence that extended residence of APP β-CTFs in the TGN might be a bad thing (Gandy et al., 2007), proposing therein that endogenous TGN isoprenoids might cause excess Aβ42 to be generated, consistent with the Koo/Golde model (Kukar et al., 2005). Intriguingly, this dovetails well with the increasing implication that dietary factors (perhaps via isoprenoids) may play roles in modulating Alzheimer’s risk (Qin et al., 2006). In experiments designed at dissecting statin action in N2a cells, Steve Pedrini, a former postdoc in our group, reported that various isoprenoid pathway-related substances (farnesyl pyrophosphate, geranylgeranyl pyrophosphate, a synthetic farnesyl transferase inhibitor) seemed to converge upon Rho kinase (ROCK) to send APP to the PM for a date with α-secretase. Constitutively active ROCK inhibited isoprenoid-stimulated α secretion and dominant negative ROCK appeared to relieve a tonic physiological inhibition of α secretion. We surmised, but never demonstrated, that concomitant diminution of Aβ release would follow, as is often the case in “regulated cleavage” situations.

    In their new JBC paper, Gary Landreth and colleagues (Ostrowski et al., 2007) home in not on Rho but on Rab1b, suggesting that targeting of APP to lysosomes via Rab1b is an important isoprenoid-modulated mechanism that controls Aβ generation. They go on to show that some of these reactions are cell type-specific. As if this weren’t already complex enough, recall that Frank Lezoualc’h and colleagues have implicated Rac and Rap (Maillet et al., 2003). Of course, what we and they want to know is how all this operates in brain in vivo. For that, we’ll have to wait until next year to hear from the ADCS’s progression-slowing clinical trial of simvastatin, headed by Mary Sano. Establishing (or demolishing) the proposed role of statins in slowing or preventing dementia will be a watershed event in determining how much effort the industry invests in dissecting the complex and challenging effects of isoprenoids on APP sorting and, more to the point, on its metabolism to Aβ.

    View all comments by Samuel Gandy
  3. Gary Landreth’s group (Ostrowski et al.) has published a very interesting and thorough study demonstrating that the isoprenylation status of specific Ras-related GTPases can have a significant impact on APP metabolism and Aβ genesis. Through a set of elegant and well-planned experiments the group has determined that statin inhibition of GTPase isoprenylation causes a decrease in Aβ secretion. Interestingly, statins appear to mediate this decrease through different mechanisms in different cell types.

    GTPase biology is highly complex, and the Rab GTPase superfamily alone is composed of over 150 members. Not only is it likely that the GTPase complement is specific to cell type, but the situation is further complicated by the fact that each GTPase may interact with multiple effector molecules, and in specific cases a cooperative binding mechanism for certain effector molecules (e.g., ROCK) may exist (reviewed in Cole and Vassar, 2006). Indeed, Ostrowski et al. illustrate this complexity by demonstrating that the statin-induced decrease in Aβ secretion is mediated through mechanisms involving different GTPases in specific cell types. In neuroblastoma cells, the statin-induced impairment of Rab isoprenylation leads to accumulation of APP within the cell, an effect that appears to limit Aβ production. However, in neuroglioma cells, statins appear to impact Aβ levels primarily through an effect mediated by inhibition of Rho activity and the subsequent lysosomal degradation of APP CTFs.

    In agreement with the observations of Ostrowski et al., our previous statin study using APPswe expressing HEK293 cells demonstrated that inhibition of protein geranylgeranylation by statins led to APP and CTF accumulation in cells (Cole et al., 2005). However, as Ostrowski et al. discuss, in contrast to their observation of decreased Aβ secretion following treatment, we demonstrated that statin-induced inhibition of isoprenylation led to elevated intracellular Aβ levels, with no apparent effect on Aβ secretion. Similar to their conclusions, we suggest that these differences are possibly due to three key experimental differences: 1) cell type differences; 2) differences in wild-type APP verses APP harboring the Swedish mutation; 3) the concentration and duration of statin treatment. As Ostrowski et al. report, the major effects of statins on Aβ metabolism are mediated by Rabs in neuroblastoma cells and Rhos in neuroglioma cells. Therefore, it is possible, if not likely, that similar differences in GTPase makeup account for the disparate effects of statins on HEK293 cells versus neuronally derived cells, at least in part. Our previous study showed that under low isoprenoid conditions the trafficking of APP is altered such that APP accumulates within the secretory pathway. The Swedish mutation of APP (APPswe) enhances cleavage by the β-secretase, BACE1. Given that under low isoprenoid conditions increased APPswe levels are localized to compartments such as the TGN, where BACE1 is active, enhanced cleavage of APP and the generation of C99 and APPsβ will occur. As γ-secretase is also localized to this compartment, this could lead to the increased conversion of C99 to Aβ and the observed intracellular accumulation of Aβ (Cole et al., 2005). It is perhaps the case that two pools of Aβ exist, at least within HEK293 cells: one destined for secretion, which is unaffected by reduced isoprenoids levels, and the other an isoprenoid-dependent intracellular pool. We speculate that the production of these two pools of Aβ could be due to differences in the metabolism of APPswe versus wild-type APP. Finally, in contrast to Ostrowski’s study, we routinely drug-treated for 48 hours, so differences in the exposure time to statins could account for some observed differences in APP metabolism.

    Indeed, a significant merit of the current study is the use of low, physiologically relevant, nM statin concentrations. Importantly, at such concentrations, statins specifically affect only a subset of GTPases, and Ostrowski et al. demonstrate that the statin-induced effects are cell type-dependent. While neuroblastoma cells may be useful to provide an indicator of what may be occurring in the CNS, it should be noted that, as they are not differentiated, they may well not truly mirror the responses that would be observed in a mature, neuronal population. Indeed, while there is no doubt that the current paper has detailed important information regarding statin-mediated effects on APP metabolism in both neuroblastoma and neuroglioma cell types, it should be noted that the increase in APP observed in these two cell lines was apparently not observed by the authors in primary neuronal cultures maintained under similar conditions (Fig. 3D). Interestingly, under our conditions, we observed APP accumulation in statin-treated Tg2576 primary neuron cultures. Once again, the reason for this remains to be determined but may be related to the expression of APPswe in these transgenic neuron cultures and/or differences in statin exposure.

    This study by Ostrowski et al. is one of a very small number of studies that have attempted to elucidate the underlying mechanisms by which statins may impart beneficial effects in reducing AD risk. However, the actual benefit of statin therapy in reducing AD risk still hangs in the balance. While it has been previously indicated that long-term statin treatment may significantly reduce AD risk (Jick et al., 2000; Wolozin et al., 2000), several later studies failed to replicate these data (Li et al., 2004; Zandi et al., 2005; Rockwood, 2006). Nevertheless, a recent finding indicated that Atorvastatin therapy appeared to improve the cognitive ability in patients living with mild to moderate AD (Sparks et al., 2005). Due to lipophilic differences, statins differ in their central bioavailability. Interestingly, early retrospective clinical trial data indicated that both hydrophilic and hydrophobic statins might lower AD risk (Wolozin et al., 2000). If further, large-scale studies do support the potential therapeutic benefit of statin therapy in reducing AD risk, but if AD risk abatement is independent of statin blood-brain barrier penetration, then it must also be considered that statins may impart their effects via a peripherally mediated mechanism rather than a direct central effect on neurons and/or glia (reviewed in Cole and Vassar, 2006). Given the association between cardiovascular disease (and associated risk factors) and AD, perhaps the therapeutic effects of statins on the cardiovascular system may account for the potential protective effect of statins on AD through an indirect mechanism.

    View all comments by Robert Vassar
  4. I would like to point out an elegant paper from the laboratory of Kumar Sambamurti that adds to this growing body of work examining isoprenylation and Aβ. Zhou and colleagues have published an article in FASEB Journal showing that geranylgeranyl isoprenoids selectively stimulate γ-secretase and modulate the amount of active complex.

    View all comments by Benjamin Wolozin