The authors might like to note our recent research on Aβ degradation by plasmin, (Exley & Korchazhkina, 2001). I can also add that in research in press in JAD, we have identified a cleavage site for plasmin in Aβ25-35, and we have shown that the activity of plasmin was inhibited by aluminum. One other point, made in our paper in press with JAD, concerns the suggestion by Tucker et al., 2000, that plasmin will degrade aggregated Aβ. We have found no evidence to support this. Plasmin only acts upon monomeric Aβ . The only way that it could reduce the amount of aggregated Aβ is through equilibrium effects. Thus, removing monomeric Aβ will cause...  Read more

The authors might like to note our recent research on Aβ degradation by plasmin, (Exley & Korchazhkina, 2001). I can also add that in research in press in JAD, we have identified a cleavage site for plasmin in Aβ25-35, and we have shown that the activity of plasmin was inhibited by aluminum. One other point, made in our paper in press with JAD, concerns the suggestion by Tucker et al., 2000, that plasmin will degrade aggregated Aβ. We have found no evidence to support this. Plasmin only acts upon monomeric Aβ . The only way that it could reduce the amount of aggregated Aβ is through equilibrium effects. Thus, removing monomeric Aβ will cause aggregated Aβ (perhaps small oligomers as opposed to full-blown aggregated peptide) to "dissolve" (only very slowly) to replace the degraded monomeric fraction, which is in equilibrium with the aggregated forms.

View all comments by Chris Exley

Although our approach that led us to identify neprilysin as a candidate for the major Aβ-degrading enzyme does not fully fulfill the above criteria, our reverse genetic data, also confirmed by the group of Steve Younkin, indeed indicate that neprilysin deficiency increases the steady-state Aβ levels in the brain (and also in plasma, unpublished data) to the...  Read more

Although our approach that led us to identify neprilysin as a candidate for the major Aβ-degrading enzyme does not fully fulfill the above criteria, our reverse genetic data, also confirmed by the group of Steve Younkin, indeed indicate that neprilysin deficiency increases the steady-state Aβ levels in the brain (and also in plasma, unpublished data) to the extent comparable to or even greater than that caused by most of familial early-onset AD-causing mutations (Iwata et al., Science, 2001). Most importantly, the Aβ levels were neprilysin dose-dependent, indicating that even partial reduction of neprilysin activity could cause Aβ accumulation in the brain.

Consistently, we have found that neprilysin expression and activity decreases upon aging, using neprilysin-KO mice as controls, particularly in the terminal zones of the mossy fiber and perforant path (Fukami et al., J. Neurosci. Res., 2002, and Iwata et al., J. Neurosci. Res., 2002). This observation suggests that aging-dependent decline of neprilysin expression/activity is a natural process and is, in our view, even programmed (and will elevate local synaptic Aβ levels in the areas particularly important in the memory formation; the perforant path projects from entorhinal cortex). Therefore, we are not too much concerned about the human genetic evidence that would support or deny the possible involvement of neprilysin in AD pathogenesis although we have been hearing about the presence of both the risk and anti-risk SNPs in the neprilysin gene, which may, incidentally, cause apparent LOD scores to look smaller than they really are.

It seems to us that the question is not whether neprilysin activity declines and causes Aβ accumulation upon aging, but rather when (at what age) it starts and how rapidly it declines to the threshold that would initiate the pathological processes even in terms of human genetics. Besides, if almost all the human beings are supposed to develop AD if we lived up to the ages of 120-140, the effect of any genetic risk factor will become smaller as the ages of the subjects become older. It may thus be more substantial to look for risk factors in relatively young subjects or for anti-risk factors in very very aged people (> 100 years old, for instance) without AD to prove or disprove the possible involvement of Aβ-degrading enzyme candidates that include IDE, ECE, ACE, tPA, etc.. Incidentally, we have had tPA-KO mice almost three years and observed no increases in brain Aβ levels at 8 weeks or no pathological Aβ deposition at 16 months of age.

In any case, I believe that the research community is still in the process of proposing as many potentially beneficial strategies as possible. The ultimate proof of the relevance of any hypothesis or even of any experimental data will be the success in clinical terms in a practical manner. The year of 2006 will be a centennial year since the initial scientific description of AD by Dr. Alzheimer. I hope we will have something more substantial that really works for the prevention and therapy of the disease by then than now.

View all comments by Takaomi Saido

Recently, therefore, we investigated whether IDE release from the cultured cells is specific or not (results were reported at the 5th International Conference on Progress in Alzheimer¹s and Parkinson¹s disease (2001) Kyoto, Japan. Abstract 386). The media conditioned by BV2 microglial cells, as well as by primary rat brain cultures, human neuroblastoma cell lines, C6 rat glioma cells and CHO Cells, were examined for the presence of IDE by Western blot analysis and by assaying degradation of 125I-insulin, synthetic ¹²⁵I-A β and endogenous [35S]Aβ. Only three of these cell lines, BV2, C6 glioma and CHO, released IDE into the medium. At the same time, these cell...  Read more

Recently, therefore, we investigated whether IDE release from the cultured cells is specific or not (results were reported at the 5th International Conference on Progress in Alzheimer¹s and Parkinson¹s disease (2001) Kyoto, Japan. Abstract 386). The media conditioned by BV2 microglial cells, as well as by primary rat brain cultures, human neuroblastoma cell lines, C6 rat glioma cells and CHO Cells, were examined for the presence of IDE by Western blot analysis and by assaying degradation of 125I-insulin, synthetic ¹²⁵I-A β and endogenous [35S]Aβ. Only three of these cell lines, BV2, C6 glioma and CHO, released IDE into the medium. At the same time, these cell lines also released into the medium lactate dehydrogenase (LDH), a protein principally residing in cytoplasm. Importantly, the percentage of insulin-degrading activity secreted from the cells at no time points exceeded that of LDH activity, indicating that IDE release was due to plasma membrane damage.

We conclude that IDE secretion is an artifact of cell culturing, and therefore this enzyme is unlikely to be responsible for clearance of extracellular Aβ in vivo. Consistent with it, inhibition of IDE fails to protect from degradation Aβ injected into rat brain (Iwata et al., 2000). Well, the possibility still remains that IDE could degrade Aβ prior to its secretion from cells and/ora cytosolic pool of Aβ.

It seems possible that IDE activity against Aβ can be diminished during aging as a result of accumulation of additional IDE substrates. I proposed recently that IDE, in addition to Aβ, could be responsible for degradation of amyloid-forming peptides in general ( Kurochkin, 1998; Kurochkin, 2001). Age-dependent accumulation of damaged proteins, particularly oxidized proteins, is well-documented. In this connection, we demonstrated that oxidatively damaged, but not the native form of, lysozyme prevents cross-linking of ¹²⁵I-Aβ to IDE (Kurochkin & Goto, 1994).

View all comments by Igor Kurochkin

Reply by Malcolm Leissring
Dr. Kurochkin's correctly notes that IDE does not contain a signal peptide. Surprisingly, this is true for a large number of zinc metalloproteases, including neurolysin (EC 3.4.24. 15), thimet oligopeptidase (3.4.24.16), and the IDE relative N-Arginine dibasic convertase (3.4.24.61). Like IDE, many of these proteases are also cell surface-associated in some cell types but not others.

Despite lacking a classical signal peptide, it is well-established that these proteases act on extracellular substrates (e.g., insulin, in the case of IDE). How, then do they get out of the cell? Well, this issue is currently unresolved. However, it has receently been shown that other so-called leaderless proteins, such as IL-1beta, are secreted through a mechanism called "microvesicle shedding" that involves the activation of purinergic receptors. Activation of this pathway in vitro leads to the release of IL-1beta, but also of LDH and other cytosolic markers. Perhaps, then, IDE is released by certain cell types together with other cytosolic contents....  Read more

Reply by Malcolm Leissring
Dr. Kurochkin's correctly notes that IDE does not contain a signal peptide. Surprisingly, this is true for a large number of zinc metalloproteases, including neurolysin (EC 3.4.24. 15), thimet oligopeptidase (3.4.24.16), and the IDE relative N-Arginine dibasic convertase (3.4.24.61). Like IDE, many of these proteases are also cell surface-associated in some cell types but not others.

Despite lacking a classical signal peptide, it is well-established that these proteases act on extracellular substrates (e.g., insulin, in the case of IDE). How, then do they get out of the cell? Well, this issue is currently unresolved. However, it has receently been shown that other so-called leaderless proteins, such as IL-1beta, are secreted through a mechanism called "microvesicle shedding" that involves the activation of purinergic receptors. Activation of this pathway in vitro leads to the release of IL-1beta, but also of LDH and other cytosolic markers. Perhaps, then, IDE is released by certain cell types together with other cytosolic contents.

I look forward to more discussion on this very interesting question.

View all comments by Malcolm Leissring

However, it is by no means sufficient to exclude that other proteases may perform a similar function either alone or in tandem with neprilysin in Alzheimer's disease. Professor Saido should not be surprised that his tissue plasminogen activator-deficient mice are not showing any Aβ-related events. In fact, a cursory glance at the literature on such mice, e.g. anything from Strickland’s group, would indicate that tPA deficiency is linked with neuroprotection as opposed to neurodegeneration.

The evidence is different for the urokinase plasminogen activated system. Both tPA and uPA...  Read more

However, it is by no means sufficient to exclude that other proteases may perform a similar function either alone or in tandem with neprilysin in Alzheimer's disease. Professor Saido should not be surprised that his tissue plasminogen activator-deficient mice are not showing any Aβ-related events. In fact, a cursory glance at the literature on such mice, e.g. anything from Strickland’s group, would indicate that tPA deficiency is linked with neuroprotection as opposed to neurodegeneration.

The evidence is different for the urokinase plasminogen activated system. Both tPA and uPA are capable of converting plasminogen to plasmin, though they may have entirely different roles in the brain. One might even see a physiological balance between the two systems? tPA-deficient mice are still producing plasmin, whereas in the AD brain the plasmin levels (possibly activity as opposed to absolute amounts) are significantly lower than in the normal brain. It is surely too early to discount plasmin as one of a suite of proteases implicated in Aβ degradation in vivo?

View all comments by Chris Exley

Although Dr. Exley's interpretation of these data as reflecting an aggregate/non-aggregate equilibrium is possible, the most parsimonious interpretation would appear to be that plasmin can degrade Aß aggregates. A lack of initial reproducibility does not mean that someone's data is wrong. In this regard, it may be appropriate to note that...  Read more

Although Dr. Exley's interpretation of these data as reflecting an aggregate/non-aggregate equilibrium is possible, the most parsimonious interpretation would appear to be that plasmin can degrade Aß aggregates. A lack of initial reproducibility does not mean that someone's data is wrong. In this regard, it may be appropriate to note that plasmin can auto-degrade. We have noted differences in plasmin- specific activity between plasmin lots that can sometimes be substantial. As we describe in our manuscript, we normalize each plasmin lot relative to a colorimetric plasmin substrate standard. Since plasmin degrades Aß aggregates more slowly, perhaps the initial difficulty in reproducing our result was due to a plasmin lot with low activity. See you at noon!

View all comments by Steven Estus

The Aβ degradation issue needs to be classified into at least two categories: "physiological degradation" that determines the normal steady-state levels and "pathological degradation" that may be mobilized to reduce Aβ levels after deposition starts. My previous statement dealt only with the first category although neprilysin is likely to be involved in the 2nd category degradation as well through altering the dynamic balances between monomers, oligomers and polymers. I also need to add that I do believe that the 2nd category degradation is as important as the 1st one and that there is no reason to deny the possible involvement of plasmin particularly in the 2nd category. In fact, we had hypothesized that the plasmin system and MMP...  Read more

The Aβ degradation issue needs to be classified into at least two categories: "physiological degradation" that determines the normal steady-state levels and "pathological degradation" that may be mobilized to reduce Aβ levels after deposition starts. My previous statement dealt only with the first category although neprilysin is likely to be involved in the 2nd category degradation as well through altering the dynamic balances between monomers, oligomers and polymers. I also need to add that I do believe that the 2nd category degradation is as important as the 1st one and that there is no reason to deny the possible involvement of plasmin particularly in the 2nd category. In fact, we had hypothesized that the plasmin system and MMP system might be the candidates that could dispose of polymerized Aβ because they the major proteinases capable of degrading fibrillized proteins and we then identified a novel brain-specific MMP, MT-5 MMP, (Sekine-Aizawa, Eur. J. Neurosci., 2001). Although our anti-MT5-MMP antibodies do stain senile plaques in AD patients. We later found that NFT is even more strongly stained and this was confirmed by Dr. Marion Maat-Schieman in HCHWA-D brains and also by Takeshi Iwatsubo in sporadic AD brains (personal communications). Because MT-5 is a type 1 membrane-spanning protein, we do not know yet how important this proteinase is in AD pathogenesis. This project is presently suspended in our laboratory.

Therefore, we have no intention of denying the role of the plasmin system or any other proteases particularly in the 2nd category degradation. Besides, Berislav Zlokovic, David Holtzman, and others have shown that there exists a dynamic balance between the brain and the circulatory system through the transport via BBB or CFS, we may need to view the Aβ degradation in a more systemic manner. Interestingly, Steve Younkin and colleagues have shown in the Stockholm Meeting that, in uPA-KO mice, plasma Aβ levels but not brain Aβ levels are significantly elevated. The uPA gene is one of the candidate genes on chromosome 10 locus associated with the risk of AD development. In this regard, I humbly suggest that Dennis and Rudy have plasma Aβ levels measured in IDE-KO mice, which showed no increase of Aβ42 and 10-20% increase of Aβ40 in the brain if what I heard at the Stockholm Meeting is correct. Besides, there still is an argument whether IDE is really the major insulin-degrading enzyme. I would like to see whether insulin levels are significantly increased in the IDE-KO mice.

Another issue that we have to keep in the mind is the difference between mice and humans. For instance, human brains are more than 1,000 times larger in size. This size issue makes me predict that degradation within the brain rather than transport out of the brain may be even more important in humans than in mice, but this is apparemtly my personal hopeful thought rather than anything conclusive with scientific and logical basis.

Let me apologize again for the oversimplification and for not being polite enough (or for being unpleasantly sarcastic). You and we are allies of scientists to fight one of our real common and mightiest enemies, Alzheimer's disease. Please stay away from anything potentially dangerous especially at this time of the year.

P.S. I am right now editing a book on Aβ metabolism, which will be open to public through Internet first and then as printed books in a few months. The authors include such stars as Martin Citron, Ditier Hartman (from Bart de Strooper's lab), Mike Wolfe, Todd Golde, Chris Eckman, Maho Morishima and Yasuo Ihara, Cindy Lemere, Berislav Zlokovic and David Holtzman. Unfortunately, I did not succeed in convincing such important people as Dennis Selkoe, Dale Schenk, Sam Sisodia/Gopal Thinakaran, amd Christian Haass to join us, but it is understandable because they have been asked to write reviews too many times. Anyway, please take look at the book when it comes out at Eurekah website and purchase the book if you think it is good enough.

View all comments by Takaomi Saido