Ridding the brain of toxic amyloid may require cutting off production of Aβ peptides, but another strategy that could work is to hasten the peptide’s demise. Several proteases have been shown to degrade Aβ in vivo, including neprilysin, insulin-degrading enzyme (IDE), and the metalloprotease endothelin converting enzyme (ECE). New work reveals that known plaque reducers, namely phorbol esters which activate protein kinase C and the metal chelator clioquinol (CQ), boost the activity of metalloproteases that degrade Aβ. One report, from Robert Messing and Lennart Mucke of the Gladstone Institute of Neurological Disease and University of California at San Francisco shows that overexpression of protein kinase Cε (PKCε) isoform decreases Aβ levels in mutant APP transgenic mice as a result of up-regulation of ECE in neurons. The second paper, from Ashley Bush and Colin Masters at the University of Melbourne in Australia, shows that CQ, which has anti-plaque activity in mice and has shown promising initial results in humans, can enhance extracellular Aβ degradation in cultured cells by increasing matrix metalloprotease activity. Both studies reveal novel pathways regulating Aβ proteases and open up new possibilities for the therapeutic enhancement of Aβ degradation.

In the first paper, which appears this week in the PNAS early edition, Messing and colleagues follow up on previous work showing that phorbol esters, potent activators of protein kinase C, suppress Aβ levels in cultured cells and in mouse brain. In vitro, the PKCε isoform seems to mediate this effect. To confirm if that is the case in vivo, the researchers, led by first author Doo-Sup Choi, generated transgenic mice that overexpressed human PKCε in brain by three- to fivefold, and then crossed the mice with Mucke’s H6 APP transgenic line, which carries the FAD Indiana APP mutant V707F. Double transgenics had fewer plaques as detected by thioflavin S staining and less immunoreactive Aβ. At 12-15 months of age, the PKCε-expressing animals had a 95 percent reduction in Aβ deposit area, with dramatic reductions in the extent of reactive astrocytosis and dystrophic neurites.

By several measures, APP processing was unchanged, so the researchers checked the activity of known Aβ degrading enzymes in brain. IDE and neprilysin activities were unchanged in the PKCε-APP mice, but levels of the zinc metalloprotease ECE were significantly elevated. In the hippocampus, ECE activity in PKCε/APP transgenics was roughly double that in the APP transgenic controls. To confirm that PKCε was regulating ECE, the investigators showed that phorbol ester treatment of cells transfected with PKCε boosted ECE activity.

While elevation of PKCε induced ECE and Aβ degradation, lack of PKCε did not affect levels of either protein. PKCε knockout mice maintain normal levels of ECE, and breeding those mice with the APP mice did not result in any change in plaque load. So while PKCε and ECE may not be necessary for normal regulation of Aβ, the results suggest that they might be enlisted as potential anti-amyloid agents by using strategies to increase their activity.

The second paper, which came out online April 28 as a paper in press at JBC, reports that the metal chelator clioquinol (CQ), currently in clinical testing for Alzheimer disease, can induce the degradation of Aβ in APP-expressing CHO cells. Senior investigators Bush and Masters have been investigating the mechanism of clioquinol since it was shown to reduce plaque load in APP mice (see ARF related news story). After that, a small clinical trial gave promising preliminary results in slowing cognitive decline in people with AD (see ARF related news story). Originally, the idea was that CQ, by sequestering zinc and copper ions and inhibiting them from binding to Aβ, would promote the solubilization and clearance of amyloid.

The new data challenge this idea. CQ is a lipophilic metal carrier, and lead author Anthony White shows that CQ increases metal levels in cells in vitro, and decreases their production of Aβ. Giving the cells CQ plus copper or zinc resulted in a 100-fold increase in the cellular content of the respective metals, and a 90 percent decrease in the amount of secreted Aβ peptides. The lower Aβ production was not caused by altered APP metabolism but instead was associated with a phosphoinositol-3-kinase (PI3K) and c-jun N-terminal kinase (JNK)-mediated up-regulation of the activity of two matrix metalloproteases, MMP2 and MMP3. Other MMP isoforms were not affected. Specific inhibitors of MMP2 and MMP3 blocked Aβ degradation in response to CQ-Cu. The researchers demonstrated that CQ-Cu complexes enhanced the degradation of secreted Aβ in neuronal cells by using the murine neuroblastoma cell line N2a, and human SY5Y cells with exogenously added Aβ.

The results could explain the observation by Bush in 2002 that CQ simultaneously increases CNS copper levels and lowers Aβ levels and plaque deposition in mice. If the ability of CQ-Cu to activate cell signaling pathways and boost Aβ degradation by metalloproteases can be demonstrated to occur in vivo, it will necessitate a new outlook on the role of metals and chelators like CQ in Aβ and plaque physiology.—Pat McCaffrey


  1. The paper by White et al. demonstrates that administration of the metal ligand clioquinol with copper or zinc can induce an up-regulation of matrix metalloproteases 2 and 3 in vitro in Chinese hamster ovary and Neuro2A cells. This up-regulation leads to a reduction in secreted Aβ peptide. The work identifies a new mechanism of action for clioquinol and a potential strategy for promoting Aβ clearance, namely through metal-dependent increases in MMP activity. A beneficial effect of clioquinol in reducing plaque load has been previously demonstrated by the same group of investigators using a transgenic mouse model of AD (Cherny et al., 2001). Whether the mechanism identified in the current paper also occurs in the brain remains to be shown. The study nevertheless provides additional evidence supporting strategies designed to promote Aβ degradation and clearance in AD.


    . Treatment with a copper-zinc chelator markedly and rapidly inhibits beta-amyloid accumulation in Alzheimer's disease transgenic mice. Neuron. 2001 Jun;30(3):665-76. PubMed.

  2. Metalloproteases are increasingly recognized as important mediators of Aβ metabolism. Several of these proteases, including neprilysin (NEP), insulin-degrading enzyme (IDE), and endothelin-converting enzyme (ECE), have been convincingly shown to play a role in the catabolism of the Aβ peptide under basal conditions. In addition, some early work has also demonstrated that some of the enzymes may directly play a role in disease pathogenesis, accelerating pathology in knockout mice, while delaying plaque formation in overexpressing transgenic mice. While this makes for an attractive therapeutic target for Alzheimer disease, developing drugs that stimulate proteases is not a trivial task. In this paper, Choi et al. demonstrate that the overexpression of protein kinase C ε greatly reduces plaque pathogenesis in APPInd mice. Furthermore, they demonstrate a significant increase in ECE but not the other Aβ-degrading proteases in these transgenic mice, suggesting a potential mechanism for this effect. While the stimulation of PKCε is likely to have a multitude of other effects, given our greater understanding of this signal mediator, and pre-existing drugs that manipulate it, this kinase may make a more tangible therapeutic target for Alzheimer disease than would ECE.

  3. Along with neprilysin (NEP) and insulin-degrading enzyme (IDE), another metalloprotease endothelin-converting enzyme (ECE) has also been shown to degrade Aβ in both in vitro cell culture and in vivo animal models. The paper by Choi et al. crossed transgenic mice that overexpress human PKC ε isoform with APPind (V717F) mutant transgenic mice and showed a clear reduction of amyloid plaque pathology.

    The authors observed a significant decrease in Aβ levels over a 12-18-month period (with a minimal decrease in 1-3 months), but no changes in sAPPα levels. Furthermore, they found an approximately 30-40 percent increase of ECE, but not NEP or IDE, activity in cortex and a more significant twofold increase in hippocampus in the double transgenic mice, as well as a ~30 percent increase in hy926 cells overexpressing PKCε, interestingly after phorbol treatment. The authors therefore concluded that PKCε reduces Aβ by increasing ECE activity without affecting APP processing. This result, to certain extents, supports an earlier observation by Savage et al. that acute (6-12-hour) intracortical injection of phorbol esters (PMA) to APPswe Tg mouse brain reduces the levels of both Aβ and sAPPβ without significant changes in sAPPα levels.

    For more than a decade, activation of protein kinase C has been shown independently and unequivocally, by numerous laboratories including but not limited to S. Gandy, J. Buxbamn, P. Greengard, C. Haass, E Koo, D. Selkoe, V.M. Lee, M. Racchi, F. Checler, V. Bigl, I Mook-Jung, and R. Messing’s own group, to stimulate the secretion of sAPPα. The mechanisms by which PKC increases sAPPα may include direct activation of putative α-secretase TACE/ADAM and “indirect” acceleration of APP trafficking to a compartment where α-secretase is highly active. While shrouded in some controversy, the effect of PKC activation on reducing Aβ has also been widely observed.

    The observation by Choi et al. is extremely intriguing, and may provide yet another novel mechanism by which PKC, specifically the ε isoform, increases ECE activity and hence promotes Aβ degradation. How PKC, especially its various isoforms, can indeed affect the proteolytic processing of APP remains to be carefully deliberated and clarified. One of the “easiest” explanations to reconcile this issue would be that different PKC isoforms may play different roles in regulating APP/Aβ metabolism/catabolism. Lanni and colleagues showed in 2004 that PKCε’s stimulatory effect on sAPPα is coupled to cholinergic pathways, while the involvement of the α-isoform in the cholinergic receptor-mediated regulation of APP processing is negligible. It is possible that the ability to increase ECE activity is specific to the ε isoform, a Ca2+-independent isoform (whereas the conventional α-, β, and γ-, isoforms are Ca2+-dependent). In this paper, the authors only measured the steady-state level of sAPPα in the total brain homogenate but didn’t assess the processing of APP (i.e., the production/secretion of sAPPα) in primary neurons (where the PKCε was specifically overexpressed under the Thy1.2 promoter) isolated from these mice. Since the bulk of the brain is non-neuronal cells—astrocytes, fibroblasts, etc., which are known to produce high or even higher levels of sAPPα—it is possible that any neuronal increases in sAPPα production was masked. While the increased ECE activity was convincingly shown in the paper, it remains undefined how PKCε activates ECE—can PKCε directly phosphorylate ECE or alter its trafficking and hence increase its activity? In addition, since recent in vitro and in vivo studies have demonstrated that NEP and/or IDE are responsible for the vast majority, if not all, of Aβ degradation in the brain, it is surprising to see such a drastic Aβ degradation (about 80 percent, in 12-18-month-old animals) in the absence of increased NEP/IDE activity.

    To validate the reduced Aβ levels in these animals, it would be helpful to show whether brain homogenates from the PKCε transgenic can break down Aβ peptides more efficiently than those from control mice. The pathophysiological relevance of ECE-mediated Aβ degradation would be much strengthened by future discovery of decreased activity of this enzyme due to genetic mutation, age- or disease-related alterations in gene expression, or proteolytic activity in association with increased risk for AD.


    . Turnover of amyloid beta-protein in mouse brain and acute reduction of its level by phorbol ester. J Neurosci. 1998 Mar 1;18(5):1743-52. PubMed.

    . Differential involvement of protein kinase C alpha and epsilon in the regulated secretion of soluble amyloid precursor protein. Eur J Biochem. 2004 Jul;271(14):3068-75. PubMed.

    View all comments by Huaxi Xu
  4. In their recent PNAS manuscript, Choi et al. present some very interesting findings, further suggesting a role for endothelin converting enzyme (ECE) in regulating the levels of Aβ peptides in the central nervous system. ECE is one of several potential Aβ-degrading enzymes that have been identified in brain, with other prominent candidates including neprilysin and insulin-degrading enzyme. In their present study, Choi et al. found that mice doubly transgenic for increased expression of protein kinase C ε (PKCε) and human FAD mutant AβPP exhibit decreased Aβ levels, plaque loads, and astrocytosis compared with the single transgenic human FAD mutant AβPP mice. No effects were found on AβPP levels or AβPP processing suggesting that enhanced clearance may be responsible for the reduced Aβ burden. Examination of the above-mentioned Aβ-degrading enzymes found no differences in neprilysin or insulin-degrading enzyme activities but a significant increase in ECE activity. Of note, deletion of PKCε in gene knockout mice had no effect on Aβ burden in the human FAD mutant AβPP transgenic mice. This implies that PKCε has no effect on the basal levels of ECE activity in brain.

    These findings further add to the idea that after initial production, the accumulation of Aβ in the central nervous system is largely due to inadequate clearance of the peptides from this compartment. Clearance likely involves several complementary mechanisms, including peptide degradation and transport out of the brain into the periphery. Other studies with human AβPP transgenic mouse models have similarly suggested roles for neprilysin and insulin-degrading enzyme in facilitating Aβ clearance from brain. Combined, the findings suggest that each of these mentioned, as well as other potential Aβ-degrading enzymes, contribute to maintaining Aβ levels in the central nervous system. Each may act in specific compartments within the brain or under certain circumstances, such as implicated for ECE with PKCε stimulation as shown here.

    As with any study like this, the next level of questions regards the overall physiological significance of the observations on pathological accumulation of Aβ in brain. For example, in what compartment and in what cell types does PKCε stimulate ECE expression and activity? Are other potential Aβ-degrading enzymes stimulated by PKCε? More importantly, are other Aβ clearance mechanisms enhanced by elevated PKCε expression? In this regard, it will be interesting to investigate if there is increased transport of Aβ out of the brain across the blood-brain barrier into the peripheral circulation when there is increased expression of PKCε in brain. This final point is significant since a primary role of ECE is to activate endothelin, which exerts its vasoconstrictive function at cerebral blood vessels. A first step in exploring this hypothesis would be to determine the plasma levels of human Aβ in the double transgenic mice. In any event, the work of Choi et al. further supports a role for ECE as a likely important cog on the complex wheel of Aβ clearance.

    View all comments by William Van Nostrand
  5. This is an excellent study that not only demonstrates that PKC ε overexpression can result in fairly dramatic reductions in Aβ-associated pathology but that also provides further evidence for a critical role for endothelin converting enzyme (ECE) in modulating AD-like pathology in vivo. It will be interesting to further determine the specific ECE isoform affected by PKC ε and to determine the effect of PKC ε overexpression on behavioral phenotypes and other markers in this animal model. The development of small molecular weight activators of ECE activity may have similar effects and are actively being explored for their therapeutic potential.

  6. Back in the good old days, when AD research was hyperfocused on Aβ production and aggregation, defining the mechanism of a given treatment that lowers net Aβ production was a relatively simple matter: Just check relative levels of Aβ precursor protein (APP), APP C-terminal fragments (APP-CTFs), and Aβ, and perhaps throw in a few in vitro aggregation experiments. These two papers, by showing that two established Aβ-lowering treatments act by enhancing Aβ degradation, show the errors of our past ways of thinking and offer a glimpse of the future.

    On the bright side, the sheer number of proteases implicated in Aβ degradation has dramatically increased the number of potential targets for therapeutic intervention. The sentiment expressed by an earlier commentator—that enhancing the activity of a protease is significantly more difficult than, say, blocking a secretase—was perhaps true when the list of Aβ proteases was limited to peptidases like neprilysin (NEP) and insulin-degrading enzyme (IDE). However, the latter are perhaps the exception among Aβ proteases. The activity of most proteases (plasmin and the matrix metalloproteases [MMPs] being just a couple of examples), is, in fact, exquisitely regulated by factors such as endogenous inhibitors, which can just as easily be targeted by small molecules. Scientists at Wyeth have provided the first example of the great promise of targeting protease inhibitors. At recent meetings, they have reported that inhibitors of plasminogen activator inhibitor-1, which inhibits the conversion of plasminogen to plasmin by tissue-type and urokinase-type plasminogen activator, can effectively reduce Aβ levels and reverse cognitive defects in APP transgenic mice. This early example shows that the future for AD therapies based on Aβ degradation is bright, and wide open.

    But there is some bad news with the good. Taking Aβ degradation into account raises the bar significantly in terms of defining the mechanistic basis of different treatments affecting net Aβ accumulation, as exemplified by the beautiful work of White et al. With something like 15 different proteases potentially contributing to net Aβ levels, and an unknown number that have not yet been identified, we can no longer focus simply on APP catabolites in defining mechanism. This underscores the great need for inhibitors with improved selectivity, which are particularly lacking for IDE, a point that is illustrated in the paper by White et al. To inhibit IDE, these investigators used bacitracin, a heterogeneous mixture of at least 10 cyclic dodecapeptides produced by certain strains of Bacillus subtilis. In our hands, bacitracin is not particularly potent, and it does get cleaved by IDE eventually, producing breakdown products that could inhibit other proteases non-specifically. It is also prone to batch-to-batch variability, and has biological effects other than protease inhibition. Nonetheless, bacitracin does seem to inhibit IDE reasonably specifically in short-term experiments in cells. In the experiments by White et al., bacitracin proved to be equally effective as an MMP inhibitor, yet surprisingly, they concluded that IDE was not affected based on the fact that IDE levels were unchanged. This is an odd conclusion, because changes in the levels of other proteases were not used as a criterion for the effectiveness of the corresponding inhibitors in this study. Moreover, the vast majority of IDE is present in the cytoplasm, and only a tiny minority of IDE would be expected to be in subcellular compartments relevant to Aβ degradation, so even if there were a change, it would probably be undetectable. And grinding cells or brains and looking at IDE activity would not reveal any changes occurring specifically in that tiny pool of IDE involved in Aβ degradation. Because of these issues, and because of the potential problems with bacitracin, we are left with a less-than-satisfying conclusion regarding IDE in both of the papers, at no real fault of the investigators, who simply lack the right pharmacological tools. Our lab has recently developed highly potent (low to subnanomolar IC50s) and selective IDE inhibitors, including both cell-penetrant and non-penetrant versions, which we plan to publish on soon, and we hope will aid in the molecular dissection of IDE’s function.

    In light of these two papers, and the fact that most AD research in the existing literature paid no heed to Aβ degradation, we are left to wonder whether all the conclusions are necessarily as sound as they seemed in the good old days when APP and the secretases reigned unchallenged. These new papers illustrate that it may behoove us to re-examine some of the conclusions we reached in the 1990s in the light of a more comprehensive paradigm that includes Aβ degradation.

  7. The proposal by White et al. that neuronal expression of the matrix metalloproteinases MMP2 and MMP3 is up-regulated by the metal-binding agent clioquinol (CQ) raises concerns.

    As reviewed elsewhere (1), various MMPs and notably MMP2 are up-regulated in response to injury and perturbation of neurons or other cells in various pathological settings. MMP2 and MMP3 degrade a range of extracellular matrix and other proteins. As part of an appropriately regulated “physiological” response to injury, this can be beneficial; however, MMPs are renowned for their aggressiveness and ability to cause substantial tissue damage if expression is not tightly controlled. In addition to hallmark roles in degrading the extracellular matrix, established or probable activities include altering cell adhesive properties, degrading neurotransmitter receptors, and remodeling synapses. One or more MMPs can also cause neurodegeneration and neuronal death.

    It also appears unlikely, based on the mechanisms proposed by the authors, that the reported effects of CQ would be restricted to MMP2 and MMP3 in vivo, since other cell types up-regulate other MMPs through MAPK pathways, as also reviewed elsewhere (1). The particular MMPs expressed in response to a stimulus depend on the cells being examined. The authors did not examine effects on primary astrocytes, microglia, diverse neuron types, etc.

    Indiscriminate metal chelation is also likely to exert diverse effects on other metal-dependent species in addition to the MMPs, such as the ADAMs. Except at high concentrations, metal binding agents don't usually strip ions from proteins, but they can bind ions within metalloproteins to form inactive complexes and redistribute free ions, affecting expression of metal-containing proteins (as proposed to occur by White et al.).

    However, CQ has been administered for periods up to 36 months in clinical trial patients without reported ill effects (2). Since significant long-term up-regulation of brain MMP and ADAM activity, etc., in response to CQ in these patients would result in considerable damage to brain tissue, it is improbable that CQ has the same effects in vivo over the long term that White et al. report in the short-term culture models.

    It is possible that the actual exposure of neurons to CQ in the brain is lower than in the in vitro system. Or perhaps other compensatory mechanisms involving the natural tissue inhibitors of MMPs (TIMPs) and other factors come into play in vivo to restore MMP and other metalloprotease activities to normal levels.

    Whatever the case, clearly there is considerable ignorance about the long-term effects of CQ in the brain, and it continues to be of concern that the studies of CQ have proceeded in patients before its actions have been adequately investigated in rodents or other non-human models.

    See also:

    For relevant reviews, see Matrix Metalloproteinases in the Central Nervous System. Conant K, Gottschall P (eds). Imperial College Press, London, 2005.


    . Metal-protein attenuation with iodochlorhydroxyquin (clioquinol) targeting Abeta amyloid deposition and toxicity in Alzheimer disease: a pilot phase 2 clinical trial. Arch Neurol. 2003 Dec;60(12):1685-91. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Two Ways to Attack Amyloid: Metal Chelator and Antibody
  2. Pilot Study Suggests Clioquinol Benefits AD Patients

Further Reading


  1. . Abeta-degrading enzymes: modulators of Alzheimer's disease pathogenesis and targets for therapeutic intervention. Biochem Soc Trans. 2005 Nov;33(Pt 5):1101-5. PubMed.
  2. . Endothelin-converting enzyme-1 is expressed in human cerebral cortex and protects against Alzheimer's disease. Mol Psychiatry. 2004 Dec;9(12):1122-8, 1059. PubMed.

Primary Papers

  1. . Degradation of the Alzheimer disease amyloid beta-peptide by metal-dependent up-regulation of metalloprotease activity. J Biol Chem. 2006 Jun 30;281(26):17670-80. PubMed.
  2. . PKCepsilon increases endothelin converting enzyme activity and reduces amyloid plaque pathology in transgenic mice. Proc Natl Acad Sci U S A. 2006 May 23;103(21):8215-20. Epub 2006 May 12 PubMed.