There are two principal ways by which the Aβ accumulation seen in the brains of people with AD can occur. One-overproduction by β- and γ-secretases-has received far more attention than the other-insufficient clearance of the offending protein. That is now changing as more researchers are focusing on the fact that the known mutations in APP and presenilin-1 and -2, all of which increase Aβ production, account for only a small fraction of overall AD cases and for none of the most common, late-onset form of the disease (LOAD). Gradually, the idea that defects in neural proteases might underlie some cases of AD is gaining currency. ApoE4, the only established genetic risk factor for LOAD, possibly impairs Aβclearance, but by itself the contribution of this polymorphism to AD onset is estimated to be no more than nine percent (Daw et al., 2000.)

A poster presented today by Wesley Farris, et al., at Brigham and Women's Hospital, Boston, presents the latest data in an emerging story on Aβ degradation. It picks up on previous research suggesting that the major protease degrading (at least soluble, monomeric) Aβ insulin-degrading enzyme (IDE) (Vekrellis et al. 2000). This metalloprotease actually cleaves numerous substrates besides insulin, including glucagon and TGF-α. To learn what IDE does in human brain, Farris et al. quantified Aβ degradation in total brain homogenates and in membrane fractions. They found that IDE accounted for 90 percent and 70 percent, respectively, of Aβ degradation.

The other significant protease in this experiment was neprilysin. (Young neprilysin knockout mice have slightly increased Aβ levels but do not form plaques, suggesting that another protease can partially compensate for the loss of neprilysin, see related news item). Farris, who works with Dennis Selkoe and colleagues, also reports that overexpressing IDE in cultured neurons lowered endogenous Aβ levels; conversely, depleting IDE increased Aβ levels. IDE knockout mice are currently being bred for analysis.

Next, the scientists analyzed Aβ degradation in cell lines established from AD families who are known to have a linkage to an area on the long arm of chromosome 10 (10q) that includes the IDE gene. This genetic connection emerged, together with a second one in a nearby region also on 10q, in a triplet of Science papers last December that implicated IDE and the protease urokinase-type plasminogen activator (uPA) as candidate genes for LOAD. (Neprilysin is located on chromosome 3.)

At this meeting, Farris et al. are reporting preliminary data on one family with strong 10q linkage in whose cells the scientists were able to detect decreased Aβ degradation.

Researchers in Rudolph Tanzi's lab at Massachusetts General Hospital, Boston, who collaborate with the Selkoe on this question, are searching for mutations in the IDE gene that would decrease its function. Also today, Lars Bertram in Tanzi's lab reported some progress in further characterizing the 10q locus, including five previously unknown polymorphisms in IDE that are currently being analyzed, but at present it remains unclear whether IDE is one of the two sought-after genes on chromosome 10. However, Farris says that even if it is not, there is now enough data to suggest that increasing IDE activity, perhaps by blocking its natural inhibitors, could one day become a promising therapeutic strategy.—Gabrielle Strobel


Farris RW et al. Insulin-degrading enzyme as the principal Aβ-degrading protease in human brain: Search for a genetic link to AD. Soc Neuroscience 2001.

Bertram L et al. Further characterization of the Alzheimer's disease locus on chromosome 10. Abstract 127.3.


  1. This news report does not mention that the serine protease at the centre of the urokinase-plasminogen activator system, namely plasmin, is selective for Aβ 42 and in cleaving Aβ 42 prevents its aggregation into β-pleated sheet structures (see Exley and Korchazhkina, 2001). It would seem that the "fog" has already begun to clear. It just went, apparently, unnoticed!


    . Plasmin cleaves Abeta42 in vitro and prevents its aggregation into beta-pleated sheet structures. Neuroreport. 2001 Sep 17;12(13):2967-70. PubMed.

  2. Aβ-degrading enzymes—many candidates, but few fit the bill

    This comment extends the news report filed directly from the meeting to provide a more detailed description of this emerging area. To date, a number of potential Aβ-degrading enzymes have been identified, including E24.11 (also known as neprilysin or NEP), insulin degrading enzyme (IDE), plasmin, matrix metalloproteinase 9, endothelial converting enzyme (ECE) and an elastase (or an elastase-like enzyme). More than 20 abstracts from 16 different groups presented data on of each of these activities (see 27.22, 56.12, 91.7, 94.2, 97.6, 98.15, 128.9, 192.11, 192.13, 322.19, 329.16, 350.4, 355.9, 428.3, 433.4, 583.11, 612.11 and 678.7).

    To date, neprilysin is the most extensively studied Aβ-degrading enzyme. It was first shown to possess this activity by Iwata et al., 2000, and its importance grew with the demonstration that degradation of both exogenous and endogenous Aβ is impaired in the brain of knockout mice in a gene dose-dependent manner (Iwata et al. 2001). At this meeting, Mark Kindy presented data extending these findings. He found that the amyloid burden (both Aβ40 and 42) in APP-transgenic mice increased after 4 months of infusion with either phosphoramidon (which inhibits neprilysin and ECE) or thiorphan (which does not inhibit ECE). Neprilysin knockout mice showed increased Aβ deposition, and neprilysin deficient neurons were more susceptible to Aβ toxicity in vitro.

    The potential importance of IDE in Aβ clearance was underscored by three poster presentations (127.3-.5), which reported genetic linkage to late-onset AD and a region on chromosome 10 that encompasses the IDE gene. Farris et al. (192.13) reported that IDE activity was decreased in lymphoblasts from an FAD kindred linked to chromosome 10. These authors also showed that unlike neprilysin, IDE appears to act on Aβ both intracellularly and at the cell surface, degrading exogenous Aβ added to the media of IDE-expressing cells while also degrading endogenous Aβ both before and after secretion.

    Two mammalian isoforms of ECE are known, ECE-1 on chromosome 1, and ECE-2 on chromosome 2. ECE-2 has an acidic pH optimum and is highly expressed in neural tissue, where it acts intracellularly. ECE-2 null mice are viable, indicating that ECE-2 is not necessary for the proteolysis of essential substrates. Nonetheless, they show increased levels of Aβ40 and 42, suggesting that ECE-2 is responsible for the turnover of at least a portion of cell-derived Aβ. ECE-1 knockout mice are embryonic lethal, indicating that ECE-1 is involved in the processing of essential substrates. Conversely, overexpressing ECE-1 in AβPP-expressing CHO cells increases clearance of Aβ.

    Carmela Abraham presented data based on inhibitor studies and N-terminal sequencing of Aβ-degrading activity, indicating that elastase 1 or an elastase-like enzyme was the principle Aβ-degrading enzyme in media conditioned by SKN-SH cells.

    There is no doubt that all of the above enzymes can degrade Aβ, but which one is the most important physiologically? The scientifically honest answer is: we don't know yet. Simple in vitro assays of Aβ degradation are of limited use; we need to assess the effects of knockout and overexpression of candidate proteases, as has been done for neprilysin. Also required is a detailed assessment of the primary structure and assembly forms of Aβ on which each protease acts. We already know that IDE degrades monomeric but not oligomeric or fibrillar Aβ, suggesting that IDE might represent the principle activity for degrading Aβ monomers whereas neprilysin might represent the major activity for removing its oligomeric forms. Other proteases may also contribute to Aβ clearance in a site- and age-specific manner.

    This information will only translate into therapy if suitable methods of upregulating activity are developed, and the pharmaceutical industry has had far more luck with inhibiting than with enhancing enzyme activity. Whatever the approach, investigators must carefully monitor the upregulation of enzymes, which are likely to have pleiotropic effects and therefore could cause serious side effects.


    . Identification of the major Abeta1-42-degrading catabolic pathway in brain parenchyma: suppression leads to biochemical and pathological deposition. Nat Med. 2000 Feb;6(2):143-50. PubMed.

    . Metabolic regulation of brain Abeta by neprilysin. Science. 2001 May 25;292(5521):1550-2. PubMed.

  3. Please see the following BMJ letter commenting on this and other related article:

    Alzheimer’s anti-amyloid vaccination and statins: two approaches, one dogma. The time for change.
    Alexei R. Koudinov, Natalia V. Koudinova
    BMJ 20 March 2002 [ Full Text ]

    View all comments by Alexei Koudinov

Make a Comment

To make a comment you must login or register.


News Citations

  1. Neprilysin Role in Amyloid Clearance

Further Reading


  1. . Evidence for genetic linkage of Alzheimer's disease to chromosome 10q. Science. 2000 Dec 22;290(5500):2302-3. PubMed.
  2. . Neurons regulate extracellular levels of amyloid beta-protein via proteolysis by insulin-degrading enzyme. J Neurosci. 2000 Mar 1;20(5):1657-65. PubMed.

Primary Papers

  1. . New frontiers in Alzheimer's disease genetics. Neuron. 2001 Oct 25;32(2):181-4. PubMed.
  2. . Clearing the brain's amyloid cobwebs. Neuron. 2001 Oct 25;32(2):177-80. PubMed.