Interview with Akihiko Takashima, RIKEN Brain Science Institute, Wako, Saitama, Japan.
ARF: What's your current thinking on the relationship between Ab and tau?
AT: Our current working hypothesis is that the normal aging process is actually the basis for the development of AD. Based on Braak and Braak's work, we know that neurofibrillary tangles (NFT) form in entorhinal cortex in normal aging. A small percentage of people develop NFTs in entorhinal cortex even as early as their 30's. You rarely see amyloid deposition at this age; that occurs later in life. But when you look at the disease state (Braak stage III and higher), it's the other way around, with NFT formation following Ab deposition. So we think that Ab is an accelerating factor for tangle formation. I think this could be an explanation for why tangles haven't been observed in APP-overexpressing mice—the natural progression of NFT formation doesn't happen so there's nothing to accelerate. On the other hand, Götz and colleagues showed that Ab injections induced NFT in their P301L tau-transgenic mouse, and APP/tau P301L cross-bred mice also develop tangles, which illustrates my point (see ARF news story.)
You can demonstrate the causal effect of Ab on tau pathology. When we took our mutant human tau mouse (V337M, Tanemura et al., 2002; Tanemura et at., 2001), and injected oligomeric or fibrillary Ab into the CA1 region of the hippocampus, we saw tangle formation in CA3 and CA4 neurons, which project to CA1. Götz's group saw similar NFT formation in amygdala when they injected Ab into CA1 and neocortex of their P301L tau transgenic. And it has to be oligomeric or fibrillary Ab; monomeric or dimeric forms don't induce tangles. It's also specific to the Ab peptide, as b-sheet fibril inducers such as polyglycine don't work.
ARF: Why don't your mice have tangles in entorhinal cortex, as AD brain does?
AT: For several reasons. One is that the transgenic mouse is, of course, an inherently artificial model. In our mouse, the PDGF or CAM kinase II promoters drive human mutant tau overexpression. These promoters generated uneven tau levels across different brain areas and expression was highest in the hippocampus. So the hippocampus was the first place to exhibit NFTs.
Then when we injected Ab, NFTs formed in neurons in adjacent hippocampal regions, whose neurons projected into the injected region. That only neurons expressing mutant tau developed NFTs in response to Ab suggests to us that neurons expressing mutant tau are vulnerable to stressors such as Ab. We think that neurons expressing mutant tau are in an "aging" state, and Ab accelerates NFT formation by activating the aging process. Mutant tau expression only gives you the potential for NFT formation, so where the tangles indeed form is highly dependent on where tau is expressed, and on where you make your injection.
ARF: So what is Ab doing?
AT: Evidence is accumulating that Ab starts by affecting the synaptic regions, reducing synaptic efficacy directly. At the same time, activation of the kinase GSK-3b by Ab causes retrograde effects in the cell body (Takashima et al., 1998). Eventually you get synaptic loss in the CA1 region, and the dendrites become smooth, without spines. What leads up to that may be that Ab interferes with the insulin-related signaling cascade related to the cell survival signal. Eventually, Ab also activates GSK-3b, leading to hyperphosphorylation of tau and of kinesin, which has been reported to be a substrate of GSK-3b (see Alzforum Live Discussion). Phosphorylated kinesin loses its ability to associate with vesicle cargo proteins, that inhibits axonal transport, which in turn leads to synaptic loss. The two processes of tau hyperphosphorylation and synaptic loss occur in parallel. Jorge Busciglio's group recently published a nice paper showing that axonal transport was inhibited in PS1 knock-in mice through activation of GSK-3b by mutant PS-1 (Pigino et al., 2003).
Because L-Ab and D-Ab cause the same neurotoxicity, we have to assume that there is no specific receptor for Ab, and that the mechanism connecting extracellular Ab to intracellular events occurs through non-specific binding.
ARF: How about the alpha7 nicotinic acetylcholine receptor?
AT: There is evidence suggesting that nicotinic agonists rescue AD, and some say that Ab does bind to the alpha7 nicotinic receptor, but I don't think it's been shown that Ab has a direct effect.
ARF: Can you elaborate on the role of GSK-3b?
AT: My colleague Koichi Ishiguro purified from the microtubule fraction of bovine brain a kinase that phosphorylated tau to its PHF state. We called it TPKI (tau protein kinase I) and thought it was a new protein, because its sequence had not been published at that time. Cloning it from a bovine brain cDNA library gave us the surprise that its sequence was identical to the GSK-3b sequence published by James Woodgett. Because GSK-3b, but not a, was present in the microtubule fraction, we focused on GSK-3b and discovered that Ab activates GSK-3b, which then phosphorylates tau, which then leads to impairment of axons and neuronal death.
We were excited when presenilin 1 was discovered as a gene causing familial AD, because that opened up the possibility that something other than Ab might cause AD. We tested 32 different PS1 mutations, and showed that there was no significant correlation between the size of the increase in Ab42 production and onset age of family members with the same mutation. So we were sure that some factor other than just Ab was involved. Marc Mercken produced a very specific PS1 monoclonal antibody, which helped us isolate human PS1-associating proteins, which we then further examined using various antibodies. We found that PS1 associates with GSK-3b and tau (Takashima et al., 1998). We also found that PS1 associates with b-catenin, and that mutant PS1 inhibits b-catenin signaling. We concluded that that mutant PS1 activates GSK-3b, which is involved in NFT formation. And since we already knew that PS1 is involved in Ab production, we naturally also assumed that activation of GSK-3b is involved in Ab production. We tested to see what would happen if we blocked GSK-3b with lithium chloride (LiCl), and found that both Ab production and tangle formation were lowered. At the present time, we assume that PS1 is associated with both Ab42 production and NFT formation, through activation of GSK-3b.
ARF: Would the GSK-3b blocker LiCl be a good treatment for AD?
AT: There's a bit of evidence out there that bipolar disorder patients who take lithium may be at a lower risk for AD. Unfortunately, we found that high concentrations of LiCl are necessary to protect against Ab and tangle formation. It would be difficult to administer it to patients at those concentrations.
ARF: Why hasn't anyone tested or developed more GSK-3b inhibitors?
AT: Because for a long time no one would believe that GSK-3b was activated by Ab! Recently, drug companies have been making a big push to develop GSK-3 blockers that won't have too many side effects. It's a difficult problem. For one thing, GSK-3b activates the degradation of b-catenin, and increased levels of b-catenin have been shown to raise the risk of cancer. But I think they're not far off. Several companies have already produced GSK-3 drugs and are testing them now in our V337M mice.
I think we have developed a good mouse model for AD (see ARF transgenic mouse directory). Because you can accelerate NFT formation by injecting Ab, you have some control over the rate of pathological progression. These mice show synaptic loss and behavioral deficits, pathological markers that can be quantified and tracked. GSK-3 inhibitors could be an ideal therapy in that it could block both Ab production and NFT formation.
ARF: Why is no one interested in GSK3-a?
AT: That's a good question. I was interested in GSK-3b because it is localized mainly in neurons, and less in glial cells whereas a is localized in both neurons and in glia. And as I said, GSK-3b is in the microtubule fraction but a is not. Actually, we never checked whether a also associates with PS1. There wouldn't seem to be much difference between the two, since structurally they're similar, and the same drugs act on both isoforms. I previously tested the effect of Ab treatment on GSK-3a, and found that a wasn't activated by Ab (Takashima et al. Neurosci Res 1998). So I concentrated on GSK-3b. Interest in GSK-3a is growing now (see news story.)
ARF: CDK5 is receiving attention as a tau kinase important to AD, also linking Ab and tau. What are your thoughts on it?
AT: Five years ago we made a transgenic mouse that overexpressed P23, an activating factor for CDK5. We didn't find any hyperphosphorylation of tau or NFT formation even though CDK5 activity was elevated to five times the normal level. CDK5 cannot form AT100 epitopes, which specifically recognize tau-CDK5's phosphorylation sites are limited to Ser 202 and 404. So I think that by itself, CDK5 is insufficient and may require contributions from other kinases.
ARF: You also study Ab aggregation. How about polyphenols as fibrillization busters?
AT: Sure, as antioxidants, they should be good busters for both Ab and tau.
ARF: Can one get a significant effect from a glass or two of wine per day?
AT: Maybe! (laughs) In any case, I drink quite regularly…
ARF: What's the state of AD research in Japan right now?
AT: I don't think there's much difference between Japan and the US or Europe in terms of the actual research being conducted. Government funding, however, is not the same. Japanese funding for basic science, even for AD research, is lower. Besides us here at BSI (RIKEN Brain Science Institute), there are only a few other major AD research groups. Private companies don't fund AD research all that much, the money all comes from the government. The government does recognize that AD is a serious problem, but it seems as though some people in government believe basic research of AD is complete, that we've solved the disease and the answer is Ab. Of course we're not finished. There are still so many questions we need to tackle, for example, understanding brain aging as a risk factor for AD.
ARF: How about treatment and care?
AT: We have special clinics for AD treatment but they need more money. We need to educate the public about aging and about AD. I think the understanding of AD in the general public could be much better than it is now, particularly in the countryside. I don't think the government is interested enough in experimenting with new care programs or treatment. This kind of policy is going to become a real problem because the Japanese population is aging so rapidly—we need money, we need to make a real investment in dealing with the problem.
ARF: Is there any good news?
AT: Well, if these GSK-3 drugs are successful…
ARF: What advice would you give a young scientist?
AT: Always think about the patients, not about your own success and glory. Remember the goal is to help patients, not the Nobel Prize. Learn what other people are doing, but do not follow what everyone else is already doing. Original thinking is so important and does not come just from reading papers. If someone has a good hypothesis for a problem in AD research, they must prove it through experiments.
ARF: What would you study if you could start over again?
AT: I'd like to study aging, to understand the natural aging process, why synaptic loss occurs, why our brains go bad—I'd really like to know that before it's too late!
I-han Chou is a science writer based in Japan.
No Available Further Reading