Wednesday, October 21, 2009

Air Capture Update – Is GRT’s Air Capture Device Viable?

Back in the July this blog wrote a detailed post about why the development of certain technological strategies to work in consort with emission reduction would be necessary to ward off significant climate change.

That post can be found here:

One of the more promising technologies was air capture, which involves capturing already emitted atmospheric CO2 instead of focusing on newly emitted CO2 for power plants. However, neither the system proposed by David Keith or Frank Zeman appeared to be cost effective and the system that Global Research Technologies (GRT) has proposed under the guidance of Klaus Lackner is still quite mysterious when it comes to the kinetics and costs.

Interestingly perhaps some light can now be shed on the GRT system in that in a recent interview with the Columbia University school newspaper, the Columbia Daily Spectator, (note that this interview has been recently taken down and appears to be no longer available) Mr. Lackner is reported to have said regarding the GRT device: “And even with current ‘dirty’ electricity, these machines are 80 percent efficient: Only 20 percent of what they sequester is cancelled out by their electricity use.”1 By knowing the efficiency of the device a cost per net ton of CO2 removed can be calculated. Also note that it is highly probable that Mr. Lackner misspoke in that statement in that the CO2 is not technically sequestered, but simply absorbed by the air capture device. This post is assuming that long-term sequestration of the absorbed CO2 typically within underground rock formations is not included in the efficiency calculation.

First, it is appropriate to assume that the current electrical grid, which is comprised of a balanced mix of electricity sources, is being used to power the device. As is typical 2007 grid percentages will be utilized in the analysis, for 2008 and 2009 data are not accurate assessments due to the recession. In 2007 coal made up 48.97% of the grid and produces approximately 1 ton of CO2 per MW-h generated;2 natural gas made up 20.08% of the grid and produces approximately 0.4 tons of CO2 per MW-h generated (when burned at maximum efficiency);2 petroleum made up 1.58% of the grid and produces approximately 0.75 tons of CO2 per MW-h;2 the rest of the grid can be assumed to be made up of trace/zero emission sources (nuclear, hydroelectric, wind, biomass and solar). Taking these elements into account it can be determined that electricity provided from the grid on average will produce 0.8594 kW-h of electricity/lb of CO2 emitted into the atmosphere. Knowing that the efficiency of the system is 80% the total amount of kW-h of electricity required to capture 1 gross ton of CO2 can be calculated at 0.17188 kW-h/lb of CO2.

Second, assume the same price on electricity per kW-h as was previously assumed in the sequestering post (9.1 cents). Knowing the system capture energy requirement, efficiency and the cost of electricity, a cost per ton of CO2 can be calculated → 0.091 dollars/kW-h * 0.17188 kW-h/lb of CO2 captured * 2000 lb/ton = $31.28 per gross ton CO2 captured. Of course that value is only the gross value, efficiency must be inserted into the equation to calculate the net value of $39.10 per net ton of CO2 capture. Based on these calculations it appears that GRT’s device is much better than the devices proposed by both David Keith and Frank Zeman.

However, although it seems better there are a variety of issues that remain regarding this value. First, just from a physics standpoint the above calculation seems very suspect. If accurate the GRT system only requires about 60 kJ to capture 1 mole of CO2. Without actual scientific documentation, heck even with it, it is extremely difficult to take that value seriously. Also the calculated value is already significantly lower than the $100 per ton of CO2 value that is frequently cited by GRT for the operation of their device, which does not make any sense.3

Second, regarding the above issue, there still appear to be no scientific publications on the device that GRT proposes, thus there is no genuine information regarding its abilities and capacities outside of the sound-bite blurbs that GRT representatives, most notably Mr. Lackner, give to the press like the above efficiency number. This lack of empirical evidence regarding how efficient and effective the device is leads to continued questioning and concern about the actual abilities of the GRT device. If the resin used in the device is proprietary and prototypes have already been tested where efficiency numbers can be declared then it naturally seems suspicious that no publications regarding the device have surfaced, at least where one could find them through conventional means.

Third, there is still a lack of information pertaining to the amount of water used per net ton of CO2 absorbed. The water question is critical because as discussed in the sequestering post, it is highly probable that the world will face significant water shortages in the near future and having to use millions to billions of gallons of water in air capture for a reduction of just 1 ppm could very well sink air capture as a viable remediation technology. Based on the limited information regarding the functionality of the GRT absorption resin it definitely appears to require water to function properly; however, there is no information regarding how much is required.

Fourth, the passive transport system is an attribute of the GRT device that may cause serious viability problems. Based on press releases and discussions with the press,3 it appears that the device as it is currently structured can absorb 1 ton of CO2 per day or 365 tons of CO2 per year if average day capacity is maintained throughout the year. This value is dreadfully small, for it would require approximately 21,370,000 devices to remove only 1 ppm from the atmosphere in a given year. Although it is reasonable to assume that the absorption value will increase with scale-up it is unlikely that the overall value using passive transport will exceed an average of 3 tons per day (1095 tons per year or 7,123,334 devices for 1 ppm per year).

Currently initial capital construction costs are estimated at $200,0001 (which means it will probably end up costing more). In the future Mr. Lackner believes that these costs will scale down to $20,000,1 but that seems unlikely; realistically $100,000 is probably as cheap as these devices will go. Splitting the difference half at $200,000 and half at $100,000, it would currently cost 1.0685 trillion dollars [in just initial capital costs not including maintenance or operational costs (note that operational costs are the real killer)] to construct the number of devices necessary to achieve a 1 ppm per year reduction at a 3 tons of CO2 per day absorption rate.

Overall it seems highly unlikely that anyone or government would invest such capital in a device that generated such small returns, especially without any viable long-term market for the capture CO2 (as discussed in the sequestering post). Mr. Lackner likes to use automobile production as an example to justify his future cost-estimates and the will of the public to invest in air capture, but such reasoning is seriously flawed because of the aforementioned lack of a long-term market for captured CO2. It is troubling that Mr. Lackner believes that the soft drink, oil and sanding equipment industries will provide viable markets that will be able purchase gigatons of CO2 annually.

Therefore, an active transport system needs to be utilized, even though such an addition would increase energy required to absorb CO2, which would increase the cost to absorb CO2, but such a cost increase is necessary if significant climate change is to be avoided. Recall that the primary advantage of air capture over simply planting trees is speed, an attribute the GRT system does not maximize. Without an active transport system, which literally sucks air into the system, it almost seems more rational to simply plant a bunch of trees than bother building an air capture network.

The more information available the better when it comes to evaluating where the government and private investors should put their money. Air capture might be the only real hope for humanity when it comes to avoid detrimental climate change. Unfortunately there appear to be a significant number of problems with the system proposed by GRT even despite the lack of available information. Note that the technical literature section of their website is still coming soon.

1. Columbia Newspaper Article; [Although the article no longer appears to be available a copy of text was made before it was taken down and is posted below]

2. “Electric Power Industry 2007: Year in Review.” Table ES1. Summary Statistics for the United States, 1996 through 2007. Energy Information Administration. May 2008.

3. Various Press-Releases and News Articles from GRT Wesite:


Columbia geophysics professor Klaus Lackner is building a device that can take carbon right out of the air. The machine, which is often described as looking like “a goal-post with Venetian blinds,” counteracts climate change in the same way trees do: by collecting carbon to keep it out of the atmosphere. But his devices are 100 times better at it and could address climate change without disrupting infrastructure, industry, or productivity. Lackner claims that “with the best technology currently available, I can get about a ton a day, maybe a little more, with a unit that can fit in a standard shipping container.”

These air capture devices could also be set up anywhere—from Antarctica to the Sahara. It doesn’t matter because, as Lackner points out: “You can have a car in San Francisco producing CO2 and a collector in the desert of Australia taking it back. It’s all the same CO2.” Once the machines are set up and plugged in (they do require some electricity) they are essentially passive, just collecting carbon dioxide as air flows across their “blinds,” which are strips of a special material developed by Lackner and his colleagues for grabbing CO2. And even with current “dirty” electricity, these machines are 80 percent efficient: Only 20 percent of what they sequester is cancelled out by their electricity use.

After they collect the carbon, Lackner and his team can go in one of two directions. They can either incorporate the CO2 into stable carbonates and bury it in the ground, or they can combine it with water to make synthesis gas: the precursor to synthetic gasoline and oil. This solution is especially promising because it could create a closed cycle of energy in which carbon dioxide is captured and re-emitted with no net increase in harmful greenhouse gasses.

Of course, one machine tree isn’t enough. We would need a million of them to mitigate climate change and up to 10 million to return to pre-industrial carbon levels. Surprisingly, this is feasible. As with cars, building just one air capture device would be extremely expensive, but mass-producing them in a streamlined system is 10 times cheaper. “To put that in perspective,” Lackner says, “the world produced 73 million cars in 2006 … so several million air capture units is not unreasonable.” “And,” he adds, “we clearly have enough room to park 10 million cars.”

Lackner is currently raising money to construct a full-scale prototype of his air capture device, which he estimates will cost about $200,000. Though this seems a high price, Lackner is confident that it will eventually fall to car-level: about $20,000. “We could actually collect carbon without ruining the world economy or changing how things are done,” he says. “With 3 or 4 factories, we’d be in the game.”

The air capture industry could also scale up just as the automobile industry did—starting out with small niche markets and higher prices and then, as manufacturing becomes more efficient, lowering costs and expanding into wider and wider markets. There are, in fact, a number of industries, including soft drink production and sanding equipment manufacturing, that must purchase large quantities of concentrated carbon dioxide to produce their goods. “That gives us a leg up,” says Lackner. “We can start with a few tons of CO2 in one location, and as we gain experience we can build more units that are cheaper, and our potential market will get larger.”

Air capture is unlike other climate solutions in that it addresses the problem directly and won’t require an overhaul of infrastructure or policy. Switching to electric cars, for example, would require a completely new and highly convenient refueling system, and would have a huge impact on trade and trade relations. Because Lackner’s devices are modular and can be installed anywhere, they don’t require that we develop a new transportation system or change the way we do things. Also, air capture is an ideal fit for the cap-and-trade system already in place because it collects carbon in a precisely quantifiable way.

When asked about his efforts to generate political interest for his work, Lackner laughs, “Well, we’re talking about it right?” It takes time to convince people that a technology is real and viable, he says, and it takes time to convince people it has policy implications. “It [climate change] also has to hurt before people will act, and we could well be too late by that time.”

Still, Lackner sees a bright future for climate and energy research. “Business as usual manifestly does not work, but we desperately need the energy,” he says. “So to me, this is one of the most exciting fields to be involved in.” A solution to climate change is about more than just science—it’s about economics and politics as well. “This field will not succeed,” he says, “unless it combines the restraints of all fields into one way to think about the problem.”

Monday, October 19, 2009

A Way to Win the War on Drugs?

To date it would be difficult to effectively argue that the war on drugs has been a success and unfortunately it does not appear that any success will be forthcoming at any point in the near future. Direct information regarding spending on combating use and distribution of illegal drugs is somewhat sketchy, but it seems that in 2002 the federal government spent 13.6 billion dollars in drug enforcement which can be adjusted to 15.2 billion dollars gross in 2006 dollars, when taking inflation into account (13.8 billion net due to drug use based fines).1 State and local governments spent a combined 30.3 billion dollars in 2006.1 The reason the war on drugs has failed for so long is that the primary focus was and currently still is on the supply side. It should be rather obvious that as long as profit margins for drug distribution and sales remain high it will be nearly impossible to eliminate even a significant percentage of drug houses and fields. Such a sentiment is especially true in third world countries where the choice between growing drugs or a drug precursor, which has a huge market vs. growing a food crop, which may not have viable market, is really no choice at all. Also the rise in the number of individuals that elect to grow marijuana plants in the United States for profit demonstrates that as long as a market exists with a high enough revenue potential, individuals will fill that market even if it violates the law. When money is involved morality tends to vanish rather quickly.

Basic economics offers the solution to the reducing the revenue for the sale of illegal drugs, reduce the demand for a product and either the supply or the price of the product will have to fall in kind. There have been noble, but naïve attempts in the past to reduce demand through pre-teen and teenage education programs like Drug Abuse Resistance Education (DARE), which were designed to teach children early in life that there was little benefit to consuming illegal drugs and demonstrate techniques that could be used to extricate oneself from a situation involving drug use. Unfortunately these programs have not worked as well as intended and most do not even exist anymore due to lack of funding and/or lack of success. Although it would be unrealistic to expect these programs to completely derail drug use among those individuals exposed to them, it is fair to ask the question why these programs were not as successful as anticipated?

Two reasons immediately come to mind to potentially explain why neutralization/prevention programs do not appear to work at effectively. First, the general rebellious nature of drug use can easily eclipse the perception of a low combined percentage of severity and certainty of both biological/psychological and legal consequence. This perception is a difficult idea to combat because for a lot of teenagers making decisions that defy authority can be viewed not only as a means of self-expression, albeit not a very unique one, but also as a means of stress relief from the rigors of the daily grind. Without direct demonstration to the consequences of drug use, it seems unlikely that any passive level of education will change the predisposed mindset that an individual has when entering a drug awareness/education program.

One strategy that has been suggested to confront this motivation and also end the perceived futility of the war on drugs is to eliminate the illegality of those drugs generally defined as experimental with less intensive addictive qualities either through decriminalization (distribution is still a crime, possession is not) or complete legalization. In theory legalization aids the war on drugs in two ways: first, legalization should diminish the aspect of rebellion in the use of these drugs because now they are no longer taboo following the reasoning that some individuals are attracted to drug use because it is against the law. Second, allowing big pharmaceutical companies to enter the marketplace will lead to a huge gut of new supply, which will drive prices downward significantly making it difficult for individual players to profit forcing them to leave the market. Even so-called ‘drug lords’ would find it hard to continue to deal drugs because of the hypothesized price collapse unless they diversify their industry.

Unfortunately there are some reservations about beginning a policy of legalization. Regardless of the level of addictive components available in any currently illegal drug there is a potential for addiction at an unhealthy level. In addition it would be reasonable to assume that if certain drugs were made legal then there would be more individuals that would become casual users of those newly legalized drugs increasing the probability of drug abusers. It can be argued that there are some individuals that are deterred from using 'recreational' drugs due to their illegality and the associated legal consequences. Also there are individuals that may not use these types of drugs because they are not widely available. For some it is difficult to proposition an interest in purchase if one does not know the nuances in which to do it and the illegality of the action itself reduces the probability of inquiry without a high probability of success. With an increase in the number of drug abusers one may see a slight to moderate increase in low-level crime such as robbery and vandalism. Finally there is the lingering question of any potential increase in medical costs or procedures due to the increased use of the newly legalized drugs. The change in medical costs is the biggest question mark in any legalization policy because it may transfer a bulk, if not proportionally more, strain from the currently strained court system to the already strained medical care system resulting in legalization costing more capital than drug enforcement.

Of course the counter-argument could be made regarding alcohol and cigarettes, but any direct comparison between these two elements without considering their current status in society would be inherently flawed. The restriction of alcohol has already occurred once in our society and was not effective, largely because it made a legal product of wide use illegal; obviously no illegal drug shares such a circumstance. Also illegal drugs have a much higher probability of fostering genuine addiction than alcohol in which a genetic predisposition is almost always required. Cigarettes have already started to be displaced from public places, which is appropriate due to the potential health risk to non-smokers. So support for the legalization of drugs cannot look at examples provided by alcohol or cigarettes in a general fashion. However, in the end legalization is a difficult question with multiple viewpoints, although both proponents and opponents of legalization think otherwise frequently offering illogical and simplistic arguments, and this particular blog post is not the venue to further elaborate on the pros and cons of those viewpoints.

The second reason education and prevention programs do not work is that for some individuals the ephemeral psychotropic effects of drugs are perceived to outweigh any potential economic or health detriments. Under such a circumstance, exclusion tactics such as those encouraged by DARE, do not have a high probability of success because simple cost/benefit arguments would have little influence on dissuading those individuals from using drugs. Instead one would need to utilize substitution tactics, finding alternative strategies to generate similar euphoric feelings. For example if one can introduce non-drug based activities to an individual that produce a sufficient level of enjoyment at a lower resource expenditure then cost/benefit arguments become more effective. Basically instead of denying the individual a pleasurable experience the detrimental one is substituted with a less/non-detrimental one. Unfortunately these methods are difficult to apply on a consistent basis.

Finally another positive strategy for reducing demand is expansion and administration of drug treatment programs as a requirement for those incarcerated for drug offenses. Due to the fact that drug use is not typically a single time event, simply incarcerating an individual for use of illegal drugs and not giving that individual an opportunity or assistance to attempt to change the behavior is not a logical course of action. However, opportunities for treatment should not be limited to inmates or those just recently released, but also must be made available to anyone that wants to participate. Initiating widespread treatment programs for illegal drug use should not only reduce demand for these drugs, but also reduce the total cost of the war on drugs. Unfortunately the public seems more interested in punishing non-violent drug users than assisting them even if treatment is almost always more cost-effective.

Despite the positive attributes associated with these alternative options there has been resistance to their widespread deployment. With this lack of alternative option application to reduce demand there is reason to understand why the war on drugs continues to attack the supply side of the drug question. However, another option is becoming more and more viable for reducing the demand side of the drug use curve. If the neurotoxic influence of a drug can be eliminated then there is no opportunity to induce any change in neuronal firing patterns which leads to addiction.

Most drugs that facilitate addiction do so via similar neuronal influence. For example cocaine affects both dopamine and glutamate transmission to generate addiction (GABA transmission could also influence addiction via its inhibitory influence on glutamate). The dopamine angle has widely been acknowledged due to the rewarding and pleasurable effects of cocaine. The dopaminergic pathway influenced by the pharmacological action of cocaine functions through enhanced dopamine transmission from neurons projecting from the prefrontal cortex, nucleus accumbens and amygdala.2,3 The anterior cingulate and ventral orbital cortex appear to be the regions in the prefrontal cortex most commonly associated with addiction and the amygdala is the primary component responsible for drug related relapse.4,5,6

Whereas dopamine enhancement is primarily responsible for short-term influences in cocaine use, glutamate enhancement is primarily responsible for long-term influences in cocaine use and behavior.7,8 This influence is understandable when one considers that long-term potentiation and long-term depression are primarily driven by changes in glutamate concentration and its interaction with AMPA and NMDA receptors. Most long-term addictive behavior for cocaine abusers seem to flow between the interaction of the core compartment of the nucleus accumbens and the amygdala where both the projecting and receiving neurons between those regions are glutamatergic.4 The prefrontal cortex also plays a role in the gluamatergic transmission between these two locations. There is evidence that suggests increased blood flow into the amygdala during a drug-induced craving.5 Antagonist blocking of AMPA receptors in the nucleus accumbens or GABA receptor agonists in the prefrontal cortex significantly reduce the probability of relapse.3,9,10

Although there are therapeutic targets available for reducing the probability of addiction from cocaine use and corresponding treatments like disulfiram and amantadine, those targets also play important roles in normal brain function and their over or under-activation can produce side effects that can lead the patient to cease treatment after only a short period of time.

Based on the neuronal mechanism behind addiction another strategy for reducing/eliminating a drug’s neurotoxic influence is to bind the drug molecule with an antibody, which would change its structure and disallow it from passing through the blood brain barrier to bind with the appropriate neuronal receptor. Currently most of the advancement in developing a vaccine for illegal drug use has occurred for cocaine largely driven by the widespread use of cocaine, the lack of a pharmacological treatment and the magnitude of its ill-health effects relative to other illegal drugs.

The basis behind the current cocaine vaccine uses succinylnorcocaine (SNC) molecules covalently linked to a carrier protein derived from the cholera B toxin (rCTB) strengthened by an aluminum hydroxide adjuvant.11,12 The idea is similar to other vaccines in that in response to the injection the body will successfully induce the development of specific IgG antibodies that can bind to cocaine in the future preventing cocaine molecules from crossing the blood brain barrier. Without the ability to cross the blood brain barrier the bound cocaine is left exposed allowing the various cholesterases in the body to interact with the bound molecule and break it down to inactive metabolites, which are later excreted from the body with other waste products. A useful attribute to a vaccine treatment is that the serum antibodies that are generated do not appear to be able to travel across the blood brain barrier either, which reduces any detrimental or unwarranted psychological effects from the immune response.

Phase I and II trials have already demonstrated that higher concentrations of vaccine increase the probability of less cocaine use12 as well as sufficient ability to neutralize the ability of cocaine to activate its neuronal dopaminergic pathway fostering addiction.13 However, there are questions regarding the inability of the vaccine to consistently generate a significant amount of antibodies to negate the ability of cocaine to maintain addiction.13 Also there are questions regarding the lifespan of the antibodies. Although lifespan issues can be dealt with booster shots every 2-3 months, it is currently unknown how long treatment would be required before an individual is no longer addicted to cocaine, thus it is difficult to gauge the total cost of such a treatment.

Another concern is that some individuals may increase their cocaine use after receiving the vaccine. The reason for this behavior change is that although the new antibodies are binding the cocaine, the addictive neuronal pathway that induces the individual to take cocaine is not neutralized in any way. In some context for these patients the vaccine generates an infinite tolerance level where no matter how much cocaine is taken no neuronal reaction is generated. Therefore, if these vaccines were used as a treatment method it would be possible that in the early stages of treatment some participants would demonstrate a significant increase in cocaine use. Fortunately because the cocaine is not having any biological effect due to the antibody there is almost no chance for the individual to overdose. Instead the only real detriment from this behavior is significant loss of the capital required to purchase the cocaine and multiple instances of possible legal ramifications for doing so.

One issue of exploration that does not receive much attention, if any at all, is the prospect of using a cocaine vaccine as a preventative measure instead of a therapeutic one. Of course with the current version of the vaccine the cocaine antibodies do not have a long enough residence time to warrant its use as a preventative measure, but it stands to reason that the lifespan issue will improve in the future. Improvement of antibody lifespan would make a drug vaccine eligible for administration in consort with other childhood vaccines. For example it makes sense that non-cocaine using parents would be interested in eliminating the possibility that their children become cocaine addicts in the future. Administration of the vaccine before the individual can form the neuronal addiction pathways for cocaine would significantly increase the probability that the individual would not become addicted to cocaine even if consumed later in life. Without the addiction pathways and with no euphoria generated from the consumption of the cocaine there would be no rational reason for the individual to begin consuming cocaine. The potential profitability of such a vaccine in a preventative context should be high, which would suggest that it presents a quality investment opportunity for a pharmaceutical company. In time it stands to reason that multiple drug vaccines could be created which would neutralize the influence of those drugs and in the end these vaccines could be the game-changer that ends the war on drugs in victory.

1. Miron, Jeffrey. "The Budgetary Implications of Drug Prohibition.” 2008.

2. Berridge, K, and Robinson, T. “What is the role of dopamine in reward: hedonic impact, reward learning, or incentive salience?” Brain Res Rev. 1998. 28: 309-369.

3. Kalivas, P. “Glutamate systems in cocaine addiction.” Current Opinion in Pharmacology. 2004. 4: 23–29.

4. Kalivas, P, and McFarland, K. “Brain circuitry and the reinstatement of cocaine-seeking behavior.” Psychopharmacology. 2003. 168: 44-56.

5. Goldstein, R, Volkow, N. “Drug addiction and its underlying neurobiological basis: neuroimaging evidence for the involvement of the frontal cortex.” Am. J. Psychiatry. 2002. 159: 1642-1652.

6. Kantak, K, et, Al. “Dissociable effects of lidocaine inactivation of the rostral and caudal basolateral amygdala on the maintenance and reinstatement of cocaine-seeking behavior in rates.” J Neurosci. 2002. 22: 1126-1136.

7. Winder, D, et, Al. “Synaptic plasticity in drug reward circuitry.” Curr. Mol. Med. 2002. 2: 667-676.

8. Tzschentke, T, and Schmidt W. “Glutamatergic mechanisms in addiction.” Mol Psychiatry. 2003. 8: 373-382.

9. McFarland, K, and Kalivas, P. “The circuitry mediating cocaine-induced reinstatement of drug-seeking behavior.” J Neurosci. 2001. 21:8655-8663.

10. McLaughlin, J, and See, R. “Reversible inactivation of the dorsomedial prefrontal cortext attenuates conditioned reinstatement of cocaine-seeking behavior in rats.” Psychopharmacol. 2003. 168: 57-65.

11. Holmgren, J, et, Al. “Strategies for the induction of immune response at mucosal surfaces making use of cholera toxin B subunit as immunogen, carrier, and adjuvant.” Am J Trop Med Hyg. 1994. 50: 42-54.

12. Martell, Bridget, et, Al. “Vaccine pharmacotherapy for the treatment of cocaine dependence.” Biol Psychiatry. 2005. 58: 158-164.

13. Martell, Bridget, et, Al. “Cocaine Vaccine for the Treatment of Cocaine Dependence in Methadone-Maintained Patients.” Arch Gen Psychiatry. 2009. 66(10): 1116-1123.

Friday, October 9, 2009

Deciphering Alzheimer’s disease – Searching for Successful Therapies

Alzheimer’s disease (AD) is primarily characterized by neuronal damage and loss over specific areas in the brain. The neocortex and entorhinal area lose the large glutaminergic pyramidal neurons; in the hippocampus pyramidal cells are damaged in the CA1 and CA2 regions; the nucleus basalis, medial septal nucleus and diagonal band of the Broca area all lose cholinergic neurons.1 Studies have demonstrated that this neuronal loss creates degeneration in signaling in the temporal and parietal lobes as well as the cingulated gyrus and possibly the frontal cortex.1 This damage to the brain results in indiscriminate memory loss (both long and short-term) as well as dementia eventually leading to death.2 Unfortunately there is no viable treatment, although some drug combinations can diminish magnitude of symptoms, nor is there a diagnostic regiment that can definitively confirm Alzheimer’s sans a brain biopsy.1,3

Three major theories have been postulated to explain the cause of AD. The oldest theory, the specific and significant reduction in acetylcholine concentration, a generally excitatory neurotransmitter, loses more and more popularity as time progresses due to known treatments that should enhance acetylcholine production and retention not curing cognitive degradation. Realistically a better idea regarding acetylcholine would be to focus on inhibition of acetylcholine receptors rather than a direct lack of acetylcholine.4 Currently the most popular theory for AD is the amyloid hypothesis where amyloid beta (Aβ) deposits in ‘senile’ plaques and soluble oligomers cause AD. This theory is commonly referred to as the ‘amyloid cascade hypothesis’.5 The third theory, the tau theory, suggests that disease progression depends on the rapid phosphorylation of a mutated tau protein which then combines with other tau proteins creating large neurofibrillary tangles (NFTs) in neuronal cell bodies. These tangles then somehow destabilize microtubules leading to the systematic breakdown of neuronal processes.6

A significant factor explaining the popularity of the amyloid hypothesis is the location of the beta amyloid precursor protein (APP) gene on chromosome 21. A major reason that this location is deemed important is that those suffering from trisomy 21 (Down Syndrome) almost always suffer from AD (frequently early onset AD).7,8 Also excess amyloid plaques (dense insoluble deposits of beta amyloid peptide and cellular material) buildup in the brain shortly before or just after the occurrence of AD symptoms.1 In addition transgenic mice carrying a mutant APP gene develop fibrillar amyloid plaques and similar AD symptomology.9 Other research target non-plaque derived Aβ oligomers because they are believed to bind to prion protein receptor to induce AD type physiology.10,11

Based on the two above valid prevailing theories, AD can be classified as either proteopathy or tauopathy. Proteopathy describes a disease or condition that results from abnormal protein structures due to protein misfolding. In the case of AD the misfolded proteins are beta amyloid and tau. Beta amyloid is a fragment from the larger trasmembrane protein APP. Tauopathy describes diseases that are a result of pathological aggregation of the tau protein [a microtubule associated protein (MAP)] and typically lead to the formation of NTFs.

There are a number of genetic indicators/risk factors that are believed to play a role in the development and/or progression of AD: 1. mutations in the APP gene on chromosome 21;1 2. mutations in the presenilin 1 gene on chromosome 14;1 3. mutations in the presenilin 2 gene on chromosome 1;1 4. alleles for apolipoprotein E (ApoE) positioned on the proximal long arm of chromosome 19;7,8,1 5. the potential mutation in the alpha-2 macroglobulin gene on chromosome 12;1 6. CLU (ApoJ) gene on chromosome 8;12,13 7. complement receptor 1 (CR1) gene on chromosome 1; 12,13 8. PICALM gene. 12,13 The first 3 genes are associated with early onset AD (40-59), the next 2 genes are associated with late onset AD (60+) and the last 3 have only been recently characterized as playing a role in AD and have yet to be confirmed with only one or both onset cases of AD.

APP is a transmembrane protein, which has three principal isoforms, 695, 751 and 770, each which contains the 4 kDa Aβ peptide and is synthesized in the rough endoplasmic reticulum and glycosylated in the Golgi apparatus.1 APP is typically found in dendrites, cell bodies and axons, which allows for effective Aβ concentration dispersement. Endopeptidase α-secretase cleaves within the Aβ region, eliminating any opportunity to form an Aβ peptide. If APP is not cleaved by α-secretase then APP can be incorporated into endocytic compartments for cleavage by β-secretase and/or γ-secretase. β-secretase cleaves APP at the N terminus of the Aβ peptide sequence and γ-secretase cleaves at the C terminus.1 γ-secretase can cleave at multiples sites creating multiple length Aβ peptides (typically 40, 42 and 43).1

Genetic mutations in APP seem to increase the probability for cleavage of higher number Aβ peptides. Amyloid comprises large fibrils and a b-sheet secondary structure – characterized by Congo red or thioflavin S staining.11 Under normal conditions a vast majority of the formed Aβ is Aβ1-40; however, in AD larger quantities of Aβ1-42 and Aβ1-43 are synthesized which nucleate more rapidly into amyloid plaques.1 This action may explain why most people of an advanced age have some amyloid plaques (the natural synthesis of Aβ1-42 and Aβ1-43 just in very small quantities), but in AD there are so many more amyloid plaques because of the significant increase in Aβ1-42 and Aβ1-43 synthesis.

One of the earliest theories pertaining to AD pathology involved the loss of cholinergic neurons (neurons that release acetylcholine). The reason disruption in the acetylcholine pathway was suggested as a rational for AD pathology was the symptomology of AD focusing on the loss of cognitive functions, especially memory and learning, functions in which acetylcholine plays a critical role. In effort to combat these losses early treatments focused on administering acetylcholine precursors and muscarinergic agonists, but neither strategy worked very well. One reason to explain the unsatisfactory results is that early in AD before any significant neuronal loss Aβ1-42 has the potential to bind to acetylcholine receptors and act as a reversible direct inhibitor against post-synaptic acetylcholine binding.4

Therefore, a portion of the disruption of cognitive function in the early stages of AD may be the direct result of this inhibitory effect instead of the loss of cholinergic neurons. This change in acetylcholine functionality may be why the influence of acetylcholine precursors is less prominent than that of cholinesterase inhibitors. The rate of synaptic release of acetylcholine is still under a specific level of neuronal control as well as the resultant released concentration of acetylcholine from synaptic vesicles which may not necessarily result in an increase in acetylcholine residing in the synaptic cleft whereas a cholinesterase inhibitor has more influence on acetylcholine concentrations in the synaptic cleft. This differing influence is probably why reducing the activity of the cholinesterase with a cholinesterase inhibitor reduces symptoms more effectively.

This pathology is further supported by the various deficits in acetylcholine neurotransmission both due to the loss of cholinergic neurons in later stages of AD and a reduced rate of neurotransmitter release.14 Of the two different types of acetylcholine receptors, muscarinic and nicotinic, Aβ1-42 is known to bind with a high affinity to both α-7 nicotinic receptors and non-α-7 nicotinic receptors,4,15,16 and it does not appear that there is any significant inhibition of muscarinic receptors.16

Binding to the nicotinic receptors reduces current amplitude by 39% +- 3% when using caged carbachol as a binding agent, topping out at a Aβ1-42 concentration of 500 nM, although inhibition was demonstrated at concentrations as low as 100 nM and 59% +- 7% under pressure application.16 It is believed that most of the disperity between these inhibition values is due to a more rapid densensitization. With these concentrations early strategies for dealing with the symptoms of AD focused on outcompeting Aβ1-42 by treating patients with cholinesterase inhibitors to reduce the degradation rate of acetylcholine in the synaptic cleft. Unfortunately these methods do not work over the long-term nor do they seem to effectively treat any of the underlying causes of neuronal death brought on by AD.

One reason for their lack of effectiveness could be that typical concentrations of Aβ1-42 in AD are believed to range from 10-50 nM, although such an estimate may be on the low side due to a suspected non-uniform distribution; these concentrations may not be significantly large enough where increasing the concentration of acetylcholine would drive a significant change in neuronal firing.16,17 Once cholinergic neurons begin to die, no amount of cholinesterase inhibitor will help because the source of acetylcholine is no longer able to produce acetylcholine. Therefore, it is reasonable to suggest that anti-cholinesterase drugs will only be useful in treatment of early to mid stages of AD progression.

A concern that does not appear to be entertained in the pathology of AD is that the application of cholinesterase inhibitors may actually be detrimental in the long-term. While cholinesterase inhibitors demonstrate the potential to increase cognitive abilities in the short-term their use could increase the overall speed of AD progression. The reason for such a counterintuitive statement is that β-secretase activity/influence on APP appears to have an association with neuronal activity in that the more depolarized the cell for the longer period of time the higher probability of β-secretase interaction with APP.18 This increased action seems to be brought on by a greater frequency of endocytosis of surface APP which closes the proximity between the APP and the β-secretase in endosomal recycling increasing the probability of interaction.18

Recall that Aβ peptides, especially Aβ1-42, bind to nicotinic acetylcholine receptors and reduce neuronal activity.15,16 Therefore, β/γ-secretase based Aβ peptides have a principle negative feedback effect as Aβ peptides seem to serve a role as, somewhat ironically, an excitotoxicity inhibitor.19 However, the application of cholinesterase inhibitors has a positive/excitatory effect on neuronal activity which leads to the increase in γ-secretase activity which in turn increases the synthesis of γ-secretase based peptides.18 Thus, it appears reasonable to suggest that cholinesterase inhibitors may decrease the lifespan of those suffering from AD due to their influences on neuronal activity and the overall resultant concentration of Aβ1-42 and other γ-secretase based Aβ peptides in the brain if indeed Aβ peptides are responsible in some part for neuronal death, which is difficult to dispute.

Recently an association between the prion protein and Aβ1-42 oligomers was identified with the prion protein acting as the receptor for the Aβ1-42 oligomer.20 The prion protein tested was of normal conformation (PrPc) not the pathogenic conformation (PrPsc), thus there is no distinction regarding whether or not Aβ1-42 can bind to PrPsc. The prion protein-Aβ1-42 complex seems to have an inhibitory effect on long-term potentiation (LTP) in the hippocampus while influencing the CA1 and CA3 regions.20 However, the cellular pathway that induces this LTP inhibition function was not fully identified. The specific region of Aβ1-42 binding appears to be the charged region of the prion protein between residues 95 and 110.20

There are a couple of questions with this finding in that the prion protein-Aβ1-42 complex did not induce any conformational changes in or interact with GluR1–4 receptors and NR-2B and -2D containing receptors for heterologous X. laevis oocyte system. Also the study concluded that the binding affinity between Aβ1-42 and α-7 nicotinic acetylcholine receptor was almost non-existent which is in direct contrast to other studies.4,15,16 Despite these questions some believe that this result is the turning point in AD treatment and if one can successfully block the Aβ1-42-prion interaction significant progress will be made in finding a cure for AD. Unfortunately such a philosophy heavily simplifies the relationship between Aβ1-42 and PrPc to the point where blocking the interaction may be disastrous.

Between PrPc and PrPsc, PrPsc has received a majority of the attention due to its believed role in neurodegenerative diseases. The divergence in study may also be a reaction to the difficulty of evaluating PrPc pathways because PrP-null mice, unlike most other null gene mice, have not been very clean-cut in highlighting a sensory pathway of action.21,22 On its own PrPc is a glycoprotein with two N-linked oligosaccaride chains and most are localized on the cell surface attached to the lipid bilayer via a C-terminal, glycosyl-phosphatidylinositol (GPI) anchor.23,24 Lipid rafts also seem to play a role in hosting cell-surface PrPc.25

Although there are not many clear roles for PrPc in normal neuronal function, there seems to be reason to believe that PrPc can acts in an apoptosis resistant pathway due to its ability to interact with apoptosis inducer Bax.26,27,28 The ability to interact with Bax reduces the probability of cellular death when PrPc is activated. The protective influence of PrPc was further demonstrated when deleting the residue sequences 32-121 or 32-134 resulted in progressive neurodegenerative illness in mice when lacking both gene copies of endogenous PrP (Prn-p), but not when lacking only a single allele.29 The most likely candidate for the interaction behavior between PrPc and Bax appears to be the direct interaction between the cytoplasmic portion of PrPc and Bax either through direct contact or a secondary messenger type system, with the secondary messenger system being more probable.27,28

PrPc action may also play a role in the prevention of damage due to oxidative stress as mice that lack both Prn-p suffer a higher probability of neuronal damage and/or death from oxidative stress.30,31,32 The protective effect of PrPc relative to damage induced by oxidative stress is somewhat controversial, but is thought to occur primarily through the function of superoxide dismutase (SOD) either directly (the PrPc in specific situations undertakes behavior/action similar to SOD)33 or indirectly by up-regulating other SOD proteins like Cu-Zn SOD.34 Currently the latter option of indirect action seems to be more probable due to inconsistencies in PrPc copper binding affinities.35

If one ties the protective effects of PrPc together with the repolarization influence of Aβ peptide, the formation of the Aβ-PrPc complex may not actually be a negative, but a positive biological action. Initially such a statement may seem foolish as Aβ1-42 binding to PrPc appears to demonstrate inhibition of LTP and induction of long-term depression (LTD).20 However, taking a step back, under normal biological (non-AD) conditions the influence of the Aβ1-42-PrPc complex would not be long-term because of the very low natural concentration of Aβ1-42 peptide. Instead production of that peptide would be increased when a cell was overexcited and possibly facing excitotoxicity and not only interact with PrPc to not only reduce the depolarization duration and rate through some secondary pathway influence on NMDA and/or AMPA receptors, but also activate defenses against apoptosis due to any excitotoxicity because of the over-activation. After a specific period of time, the Aβ1-42 dissociates somehow from the PrPc and the inhibitory activity stops.

Unfortunately in AD, the extracellular concentration of Aβ1-42 is dramatically increased which significantly increases the probability that PrPc remains active in the Aβ1-42-PrPc complex, which continues the inhibitory effects. These inhibitory effects may still slow down the progression of AD because instead of neuronal death being induced by excitotoxicity it is induced by a slower LTD derived axonal and dendritic retraction. On a side note although copper binding is prevalent in PrPc, it does not seem to influence Aβ1-42 binding.20

With all that has been said, the most interesting potential action very well may be the fact that antibody-induced cross-linking of PrPc on a neuroectodermal cell line stimulated non-receptor tyrosine kinase fyn.10 This stimulation of fyn required an interaction between PrPc and caveolin and later resulted in the stimulation of NADPH oxidase and extracellular-regulated kinases (ERKs).36 The reason activation of fyn is interesting will be explained later. Overall at the moment until an actual pathway for pathogenesis can be uncovered with the normal conformational prion protein, it unclear how useful targeting the prion protein would be at treating AD, if even useful at all for if it does act in a more protectionist manner over detrimental then blocking its action may actually increase the rate of progression in AD patients.

Despite a lot of support in the scientific community, which may be in the process of eroding depending on who one talks to, there are significant questions regarding the influence of the amyloid plaques in AD and the role these plaques play in the progression of the disease. Multiple amyloid plaque degradation treatments have been experimented with and none have generated enough positive statistically relevant results to be included in mainstream treatments. The failure in a Phase III trial of Flurizan (tarenflurbil) after a successful Phase II trial was somewhat shocking and disappointing to the medical community.

In culture Aβ1-42 protofibrils that eventually become plaques have a tendency to kill cells through application of oxidative stress and can induce a greater frequency of excitatory post-synaptic potentials,37,38 but there is no definitive evidence that plaques themselves actually have a neurotoxic influence. Also there appears to be no proportional ratio between the number of plaques and the level of neurological disfunction, in addition to lingering questions regarding proximity of plaques to neuronal damage.39,40,41 In fact some studies have shown that neuronal damage occurs outside of or in absence of plaque formation.11,42,43 Note that there are three different classifications of plaque: diffuse, fibrillar and dense-cored where diffuse plaques lack an identifiable or distinguishable morphology, fibrillar plaques have a central mass of β-amyloid with compact spoke-like extensions and dense-cored plaques have a compacted central mass surrounded by an outer sphere of β-amyloid.43 Early in the progression of AD the majority of the plaques are diffuse whereas the ratio shifts to favor fibrillar and dense-cored plaques as the disease reaches later stages.44 This change is probably most influenced by the increasing concentration of Aβ peptides creating an increased opportunity for peptide aggregation.

A serious concern is that researchers may have simply assumed a negative role for the plaques because of their association with AD and neuronal death. For example the plaques are the bystander covered in blood in a room with a recently murdered individual. The police arrive and naturally assume that this bystander is the murderer instead of considering that he happened upon the scene and maybe even tried to help the dying individual. Could the plaques be markers of a counter-response to the overproduction of Aβ1-42 instead of a detrimental element generated by Aβ1-42 or something else entirely? Such a conclusion would explain the failure of plaque degradation treatments because if the plaques were positive or neutral then destroying them does nothing to help the patient, thus there would be no statistical difference between these drugs and placebos.

Initially the statement that the amyloid plaques are not negative seems foolhardy largely because it is believed that Aβ1-42 fragments makeup a significant portion of a plaque, which are neurotoxic and there are various pieces of empirical evidence that seem to support a negative role for plaques. However, there are two different rationalities that can be applied to explain the relationship between Aβ1-42 fragments and plaques. First, when Aβ1-42 fragments aggregate into a plaque they loses their toxicity because they are no longer able to bind to a specific receptor (nicotinic acetylcholine, etc.) initiating a toxic influence. This explanation implies that Aβ1-42 has one of three destinies when secreted from a neuron: aggregate to an oligomer and bind to a receptor, self-aggregate with other Aβ1-42 to form fibrils or plaques or be destroyed or removed via something like a microglia or other clearance method.

Second, what if the amyloid plaques are not solely comprised of Aβ1-42, but are comprised of both Aβ1-42 and Aβ1-43. Aβ1-42 and Aβ1-43 are remarkably similar compounds both are insoluble and create fibrils rather easily. There is no evidence to demonstrate that Aβ1-42 and Aβ1-43 cannot bind together creating plaques probably because it was not viewed as an important point of study (understandably). Both Aβ1-42 and Aβ1-43 can form plaques independently. The action of Aβ1-43 may in fact reduce the neurotoxicity of Aβ1-42 by hastening its ammelgamation into a plaque. Overall the first option seems more viable than the second option as it is difficult to believe that a neuron would self-trigger Aβ1-43 cleavage.

The explanation for why plaques appear toxic in certain research when they really aren’t could be explained by addressing plaque stability. There is no reason to assume that plaques are 100% stable, once formed they do not go through any further change, thus small portions of the plaques could break off into Aβ1-42 dimers, trimers and oligomers which are toxic and proceed to induce the cascade that influences neuronal death. The real question regarding any negative influence of plaques is does their size induce any accelerated microglia or other inflammatory responses or is any resultant inflammatory response a reaction to nearby oligomers?

If Aβ1-42 is a critical component in the onset and advancement of AD then it stands to reason that the administration of Aβ1-42 antibodies would limit the neurotoxic influence of Aβ1-42. Various studies have confirmed this position where peripheral antibodies for Aβ1-40 and Aβ1-42 were applied in transgenic mice models and non-human primates resulting in a reduction in neuritic dystrophy, synaptic degeneration and early tau tanglement.45,46,47,48,49 The chief method through which antibodies work is by binding the Aβ peptide which eliminates its ability to bind to nicotinic acetylcholine receptors or other receptor targets and later trigger phagocytosis or another form of clearance destroying the Aβ peptide.42 However, there is an interesting concern in that the increased phagocytosis/cytokine release may also increase microglia activation which may neutralize the positive effects of the Aβ antibody due to collateral damage generated from the microglia.50,51

This concern of excess immune response may very well be true because unfortunately the use of Aβ antibodies in an actual therapeutic environment has not been very successful largely due to an increased T-cell autoimmune response against the Aβ sometimes resulting in aseptic meningoencephalitis,52,53 inconsistent results where certain populations of patients improve and others do not and little reduction in tau-based late NTFs.52,54 New strategies are being investigated to eliminate this T-cell response, which if successful could make the administration of Aβ antibodies a possible therapeutic treatment once again.

Another theory relating to neuronal death in AD focuses not on Aβ1-42 as the principle actor in the damage leading to death, but more as an instigator that leads to overreaction by microglia, which actually release the toxins that drive neuronal death. Microglia are the principle macrophage in the brain due to the fact that most antibodies and other immune system components once fully differentiated cannot penetrate the blood-brain barrier (BBB). There are typically four types of activation states for microglia: ameboid (principle scavenger role), ramified (inactive/resting/central body motionless probing with branch processes), activated non-phagocytic (partially active, secretes cytotoxic factors and recruitment molecules, uptake of MHC class proteins and proliferation) and activated phagocytic (fully active, secretes cytotoxic and pro-inflammatory factors, antigen presenting and available to phagocytose).

There is evidence that demonstrates microglia have the capacity to bind to the N-terminus of either soluble Aβ1-40 or Aβ1-42, which activates them to at least an activated non-phagocytic state resulting in the active secretion of neurotoxins in effort to clear away the binding target.55 Unfortunately based on the generic close proximity of the Aβ1-40 or Aβ1-42 to the neurons these deleterious agents produced by the microglia not only act against the Aβ1-40/42, but also the neurons. One of the chief agents from microglia thought to induce neuronal death is a neurotoxic version of phenolic amine that binds to NMDA receptors and could begin an excitotoxicity cascade similar to that of glutamate.55

In addition microglia have been shown to infiltrate amyloid plaques in AD.56 However, this infiltration is interesting in the context that one of the chief histological features of AD is amyloid plaques surrounded by microglia and astrocytes. In these AD environments the amyloid plaques are still intact which leads to the question, are microglia able to effectively clear away amyloid plaques? This question also ties into the issue regarding the fact that there are studies that identify significant plaque formation in areas away from significant neuronal death without close proximity to plaques. Note that only dense/core plaques, not diffuse plaques seem to attract and have the potential to activate microglia.55 If microglia did destroy plaques with a high degree of specificity then it would stand to reason that these plaques would eventually be destroyed because once a neuron died the excess secretion of Aβ peptides would significantly decrease.

There appears to be three possible explanations for these lingering plaques. First, microglia are not activated or have a very low activation potential due to fibril/plaque based Aβ peptides; instead only soluble Aβ1-40 or Aβ1-42 have the requisite N-terminus to facilitate binding and high probable activation. Therefore, the microglia that are incorporated into the plaques are those that are binding to the unstable portions of the plaque that break-off to become soluble peptides. Second, the microglia have become inactive due to reaching their phagocytosis limit (basically these incorporated microglia cells have taken their gitter form) or lose the necessary receptors to induce phagocytosis. However, the loss of thes types of receptors, like Toll-like receptors, does not make very much sense. Third, the microglia are supplementing the elimination of the Aβ peptides and their associated plaques by secreting their own Aβ peptides,57 creating a dynamic equilibrium between the clearance of Aβ peptide and its secretion.58

If the third option is correct, then the relationship between microglia and Aβ peptides becomes complicated. The enhanced release of additional Aβ peptides from microglia could either increase the rate of AD progression due to increasing the overall concentration of neurotoxic agents (especially Aβ1-42) in the brain or could decrease the rate of AD progression by increasing the rate of Aβ1-42 plaque formation which significantly limits/eliminates their neurotoxic influence. Or a third option exists in that this release creates an dynamic equilibrium between Aβ peptide synthesis and clearance doing little to help those suffering from AD and in fact possibly being a net detriment due to the occurrence of any ‘collateral damage’ neuronal death because of the release of cytotoxic factors when binding Aβ peptides. In addition to microglia, localized astrocyte populations seem to increase in the presence of Aβ1-42, but neuronal death does not seem to be increased or decreased via the action of astrocytes.55 Instead of clearing away Aβ peptides and plaques astrocytes seem to play the role of barrier formation by forming a wall between the plaques and neuropils.59

It is also reasonable to suggest that microglia would not be able to bind to an Aβ1-42 oligomer that is bound to a receptor due to the fact that it is highly probable that the N-terminus binding site is concealed. Therefore, for microglia to be responsible for significant neuronal death it appears that they would have to bind to a soluble Aβ1-40 or Aβ1-42 to activate and release of their degradation agents in reasonably close proximity, not necessary in direct contact, to the neuron. There is no reason to suspect that phagocytosis would induce neuronal death. Due to the liklihood of such a situation, it is difficult to theoretically view microglia as a chief element in neuronal death in AD, that is not to say that there is no microglia activation, but probably not enough to warrant its role in neuronal death as significant.

However, reality seems to differ from the above hypothesis in that when transgenic mice and AD patients are treated with anti-inflammatory medication there appears to be a reduction in neuronal death.60,61,62 If microglia influence is apparently so difficult to induce via interaction with Aβ1-40 or Aβ1-42, why do anti-inflammatory treatments reduce neuronal death in the short-term? Perhaps the answer lies in the fibril plaques that are created as AD advances. Previously it was suggested that these plaques have no inherent toxicity despite various studies that seem to indicate the contrary. To explain this alleged contradiction it was reasoned here that plaques are in a quasi-dynamic equilibrium state where small oligomers are continually being added and subtracted from the plaque. If this were the case then it would go a long way to explaining why neuronal cell death is significantly reduced because the anti-inflammatory agent is preventing microglia from activating due to interaction with the oligomers that are breaking off from the plaque. It may be reasonable to suggest that plaque-based (fibril) Aβ1-42 is about only 1/5th as toxic as Aβ1-42 oligomers.63 Such a result makes sense if one considers small oligomers breaking off from the plaque to either interact with microglia or other receptor targets vs. the same amount of Aβ1-42 available to interact when in a non-fibrillar state when concentrations are initially equal.

Most of the studies touting the benefits of anti-inflammatory agents are short-term. The reason neuronal death is reduced in the short-term, but not in the long-term is because microglia could very well act as one of the faster pathways when inducing neuronal death. This belief seems to make sense when considering the methods of neuronal death involved in AD. Instead of having to wait on the destruction or excitatory collapse of nearby neurons to induce significant excess glutamate release or anticipate a calcium secondary messenger system over-activation, which triggers hyperphosphorylation of a MAP protein (tau) finally resulting in axonal collapse/retraction, the microglia toxicity acts immediately on NMDA receptors to generate a cascade failure relatively quickly.

Also it is highly probable that the microglia can act over a wider range of immediate influence than glutamate or tau based death, thus not only are neurons killed faster, but more could die in shorter period of time. Unfortunately preventing microglia activation through anti-inflammatory agents or other means is not a cure for AD because they are only addressing one potential neuronal death pathway, other pathways are not neutralized and the neurons that are salvaged due to the anti-inflammatory agent will probably be eventually killed later. Basically microglia action can be regarded as a fast secondary means of neuronal death.

A bright spot in significant microglia induced death may be its action against NMDA receptors. If the neurotoxic activity demonstrated by microglia does indeed influence NMDA receptors and resultant calcium influx then anti-inflammatory agents may not be a necessary element to reducing the influence of microglia on neuronal death. A drug that will be discussed later, memantine, may very well serve a double beneficial purpose in the treatment of AD due to its antagonistic action against NMDA receptors and channel opening. Thus, instead of having to get rid of initial microglia action, its overall neurotoxicity against neurons can be neutralized.

Unfortunately phenolic amine and its action against NMDA receptors may not be the only cytokine derived from microglia that plays a significant role in AD. Microglia can also release interleukin-1(IL-1) α and β, primarily β, when interacting with APP and Aβ1-42.64 IL-1β can bind to surface receptors and activate p38 mitogen-activated protein kinase (p38-MAPK) which seems to have the ability to both reduce the concentration of synaptophysin and phosphorylate tau.64 This phosphorylation could be a step in the hyperphosphorylation of tau and the formation of paired helical filaments (PHFs) and NTFs that are a trademark of AD, because P38-MAPK phosphorylates tau at five sites that are phosphorylated in PHFs.65 Inhibition of IL-1β using an anti-IL-1β antibody or blocking the IL-1 receptor with IL-1ra reduced neuronal tau phosphorylation and increased the concentration of synaptophysin when exposed to APP-activated primary microglia vs. control samples.64 This pathway could demonstrate a secondary means of microglia derived neuronal death. In addition it may muddy the waters with regard to inhibiting the hyperphosphorylation of tau by introducing valid evidence to support the action of another kinase.64,66,67

In the microglia based p38 MAPK pathway, hyperphosphorylation of tau is only one issue; the loss of synaptophysin may also play a role in cognitive degradation. Synaptophysin is an integral membrane protein that is typically phosporylated by tyrosine kinases and is thought to regulate synaptic vesicle release.1 A reduction in synaptic vesicle release would reduce the total concentration of neurotransmitter released which in turn will more than likely reduce depolarization of neighboring neurons reducing signaling.

The story on anti-inflammatory drugs does not end with reducing the activity of microglia. Some anti-inflammatory drugs, most notably non-steroidal anti-inflammatory drugs (NSAIDs), have the ability to inhibit γ-secretase, which reduces the synthesis of Aβ1-42 and Aβ1-43.68 Unfortunately it is difficult to evaluate how influential γ-secretase and β-secretase inhibitors would be as therapeutic strategies because it seems reasonable to suggest that as AD progresses further to mid and late stages the less useful these inhibitors would be in alleviating symptoms due to the large concentrations of Aβ peptide already synthesized. Improvement in AD early diagnostic procedures would go a long way to improving the prospects of secretase inhibitors as treatment agents. Also there is a concern that γ-secretase inhibitors would interfere with other cleavage targets performed by γ-secretase like Notch 1.68

Another method that has been explored to reduce Aβ1-42 concentration is stimulation of the M1 muscarinic acetylcholine-receptor which has been shown to increase the activation of α-secretase, which eliminates the ability of γ-secretase and β-secretase to create Aβ peptides.69,70 Unfortunately M1 activation enhancement has not been extensively tested, so their actual therapeutic value in the long-term is still unknown.

MAP tau is thought to play an important role in neuronal differentiation and axonal development as well as axonal maintenance.71,72 The primary attention on its role in these functions focuses on its ability to influence microtubule assembly and stability. In fact because the phosphorylation of tau alters its ability to bind to microtubules, there are many that believe changes in the phosphorylation rate of tau plays a significant role in the neuronal death witnessed in AD due to a reduced microtubule stability.73,74 The general theory seems to be that under normal conditions tau is a normal elongated protein that aids in promoting microtubule assembly, stability to the microtubular ‘roadway’ and bundles microtubules in the marginal band allowing synaptic vesicles and organelles to move freely and efficiently from the neuronal cell body to the synaptic bouton. However, in the case of AD a hyperphosphorylated tau becomes destabilized and no longer binds to microtubules causing the axon to become destabilized and the axon retracting leading to abnormalities and shortfalls in the delivery of vesicles and organelles impairing communication between neurons.

Support for this theory is largely derived from the two primary characteristics of tau in AD, the abnormally high presence of several phosphorylated serines and threonines 75,76 and the presence of NFTs which are comprised of PHFs and straight filaments of which hyperphosphorylated taus are a principle component.1,77 There is also the probability that the loss of tau solubility is brought on by hyperphosphorylation and increases the probability of NFT/PHF formation. Due to this behavior of tau in AD, one of the more common treatments is to break up NFTs/PHFs with the hopes that it will reduce the probability of neuronal death. The immediate problem with this strategy is if the treatment strategy does not involve dephosphorylating the tau or preventing the phosphorylation in the first place then breaking up the NFT/PHFs will probably do little to stem neuronal death because the phosphorylation itself is what drives the neuronal death, the fact that tau eventually forms NFT/PHFs is just a secondary symptom.

Due to its relationship with microtubules, the axon and possibly other cytoskeletal proteins, neutralizing the mechanism behind the hyperphosphorylation of tau looks to be a promising strategy in the treatment of AD and the reduction of neuronal death. Currently there are two major strategies that are being utilized to address this issue, down-regulation or inhibition of tau phosphorylating kinases or up-regulation of dephosphorylating protein phosphatases. Between these strategies most of the focus have been on kinases GSK-3β and cdk5/p25 and phosphatase (PP)-2A.

There is evidence to suggest that cdk5/p25 plays a role in the development of AD.78,79 Cdk5 is a Cyclin-dependent kinase with an associated regulatory subunit at p35. Proteolytic cleavage of p35 generates p25, which frequently results in abnormal Cdk5 activation, especially in AD.78 Overexpression of p25 results in the hyperphosphorylation of endogenous tau and the eventual formation of NFTs.78 Inhibition of Cdk5 or p25 reduces the amount of tau hyperphosphorylation and neuronal death, but does not eliminate tau aggregates or NFTs. Therefore, it appears that cdk5/p25 has some form of catalytic effect on the hyperphosphorylation of tau. Unfortunately there may not be anything that can be done regarding cdk5 in a long-term treatment regiment because inhibition of cdk5 activity tends to also inhibit fast anterograde axonal transport and the redistribution of cellular proteins.80 Fortunately this lack of inhibition is not a game-breaker because the principle kinases that are thought to be responsible for hyperhosphorylation can still be inhibited.

The aforementioned glycogen synthase kinase-3 beta (GSK-3β) is thought to be a principle actor in hyperphosphorylation of tau. Mammalian GSK-3 has two isoforms, α and β, and on average is constitutively more active in neurons than other kinases.81 Inhibition of GSK-3β via multiple inhibitors demonstrates a reduction in tau aggregation levels as well as a reduced level of neuronal death while overexpression of GSK-3β results in hyperphosphorylation of tau and NFTs.82,83 The inhibition of GSK-3β occurs through two pathways, the inhibitor either competes with magnesium to limit activation or phosphorylates the serine9 residue, which aids inhibition.82

However, there are some concerns with GSK-3β being the driving kinase behind tau hyperphosphorylation. First, in single tau transgenic mice, an increase in GSK-3β activity appeared to reduce neuropathology and motor impairments, basically doing the exact opposite of what would be rationally expected with an increase in tau phosphorylation.84 Second, reduction in tau phosphorylation using a GSK-3β inhibitor does not neutralize all of the phosphorylation of serines on tau. For example treatment with lithium, a known inhibitor of GSK-3β, reduces the level of phosphorylation at Ser202 and Ser 396/404, but not at Ser 262 or Ser422.82 The lack of phosphorylation preventation at Ser422 is of note because Ser422 is commonly regarded as a phosphorylation site that is specific for disease, including AD.85,86 Although it can be argued that Ser202 andSer396/404 are more important in the facilitation of tau-based NFTs.87 Third, there is the question of the natural constitutively activation of GSK-3β and why this activity does not induce more spontaneous NFTs? Maybe it does, but the rate of generation is not large enough to induce any significant changes to microtubule organization? Maybe tau needs to be pre-treated in some fashion to place it in closer proximity to GSK-3β before excessive phosphorylation? Maybe there is associated phosphatase activity that neutralizes natural GSK-3β influence? Fourth, while tau aggregation levels are reduced when GSK-3β is inhibited, the total number of NFTs that form in the transgenic mice are not necessarily reduced.82

At one point in time MAP kinase ERK2 was also viewed as a potential agent in the hyperphosphorylation of tau due to increased co-distribution with the neurofibrillary changes in Alzheimer’s disease and the fact that it appeared to phosphorylate all tau relevant serine-theronine residues at the maximal stoichiometry 88,89 However, it seems more probable that this co-distribution is reflective of a secondary pathway not attributable to hyperphosphorylation tau progression because in studies where MAP kinase ERK2 activity was stimulated or inhibited no significant change in tau progression occurred in kind.90,91,92

Phosphatase PP-2A seems to act as an inhibitor of hyperphosphorylation as when PP-2A is inhibited or down-regulated there is significant tau hyperphosphorylation at Ser202/Thr205 and Ser422.93,94 Clearly it makes sense that in the case of hyperphosphorylation there would be a decrease in respective phosphatase activity. However, there do not appear to be any identifying studies with regard to the interaction between tyrosine kinase fyn and phosphatase PP-2A. It may make more sense that the conformational change that occurs when tau is phosphorylated at tyrosine18 by fyn prevents phosphatase PP-2A from dephosphorylating tau at Ser202/Thr 205 and Ser422 rather than a decreased rate of dephosphorylation due to down-regulation of phosphatase PP-2A.

It is also believed that one of the elements responsible for phosphorylation of tau could be a src family tyrosine kinase, called fyn.75,95,96 Bolstering this claim is that when cells are co-transfected with only fyn or tau AΒ1-42 toxicity is statistically eliminated when accounting for neuronal death.75,97 However, this conclusion leads to an interesting question. Tyrosine kinase fyn phosphorylates tau at the tyrosine18 residue, not at any serine or threonine.72 If this phosphorylation site is accurate, then tau may not induce neuronal death exactly through the aforementioned method of axonal regression due to microtubule destabilization. When the tyrosine18 residue is phosphorylated there is no apparent reduction in probability of tau binding microtubules.75 The ability of tau to bind microtubules seems more dependent on whether or not the serines and threonines are phosphorylated. Due to this action the pathology of neuronal death due to hyperphosphorylation of tau may be more reminiscent of how tau behaves during neuronal development instead of neuronal maintenance.

During neuronal development a subpopulation of tau persists in the distal portion of the axon and the growth cone.98,99 and aids in its outgrowth. In addition various src-family non-receptor tyrosine kinases also exist in the growth cone 100 meaning that it is likely that fyn is among those tyrosine kinases. Tau localization is disrupted when exposed to tyrosine phosphatase inhibitors101; therefore, a fyn and tau interaction could drive tau action in neuronal development. Specific action of tau is related to altering the actin based growth cone to facilitate dynamic microtubule incursion. Then tau binds to the new microtubules in order to organize them to drive the forward advance of the growth cone.102 This action could identify a new mode of action to induce neuronal death in that the breakdown of microtubules is not due to lack of tau binding, but instead is the result of tau programmed regression due to new growth behavior.

For example assume that in early development most of the phosphorylation of tau is induced by fyn on the tyrosine18 residue which drives neuronal differentiation and axonal development whereas after development phosphorylation of tau is induced by other tyrosine kinases that focus on serine and threonine residues which aids in microtubule stabilization; it could be possible that no further phosphorylation by fyn takes place for critical functions after development. Whether or not this is true is dependent on the direct function of fyn in the PrPc neuroprotective function.

In the case of AD, the action of Aβ1-42 triggers a renewal in phosphorylation of tau by fyn creating a sensory trigger that causes tau to breakdown the current axon and attempt to rebuild a new axon, similar to the first axon created during development. However, before this new axon can be developed, tau forms NFTs with other taus ceasing the process. Unfortunately there is only speculation regarding how Aβ1-42 eventually activates fyn. Another question is why doesn’t tyrosine kinase fyn in pyramidal cells in the hippocampus hyperphosphoryalze tau during LTP? A possible answer to the latter question is that the rate of activation under normal LTP conditions is not long enough to induce sufficient phosphorylazation or fyn is not the first step in the fyn activation pathway.

Of the four AD drugs that are currently used for therapeutic purposes the most successful appears to be memantine. Memantine’s success is believed to be derived from its slow inhibition of Ca2+ influx after long-term activation of NMDA receptors and its associated ion channel.103 In the brain there are three classes of ionotropic glutamate dependent ion channels where glutamate must bind to a receptor to trigger opening: NMDA, AMPA and kainite. The most important of these three receptors is NMDA due to its elevated permeability for Ca2+. If a NMDA-based channel is open for too long, then the concentration of Ca2+ that enters the neuron will be too large and will disrupt Ca2+ homeostasis of the neurons leading to a secondary messenger system cascade that will more than likely lead to cell death. This death can come from many different avenues, oxidative damage, proteolytic processes, mitochondria driven apoptosis, etc. Due to the threat of this detrimental possibility NMDA receptors have a magnesium ion that normally blocks the ion channel, which requires the neuron in question to have some level of depolarization before the magnesium ion is repelled enough to clear the channel. This required depolarization typically occurs through the AMPA and kainite channels that open in response to glutamate and allow sodium to flow into the neuron.

Under normal circumstances activation of NMDA receptors and Ca2+ influx is strictly controlled mostly through LTP feedback functions (open for only a few milliseconds at most). However, AD upstream neuronal death due to Aβ1-42 toxicity can lead to the release of large quantities of previously isolated intracellular glutamate that can hyper-activate NMDA and AMPA receptors beginning the neuronal death cascade largely in the CA1 and CA2 regions of the hippocampus. Previous NMDA antagonist treatments were designed to completely block NMDA function in effort to prevent neuronal death, but because NMDA is required for normal learning and memory as well as certain brainstem functions like wakefulness, these 100% block antagonists had severe side effects.103 Memantine works because it acts as uncompetitive antagonist, where its blocking effectiveness increases with channel activation time.103 Basically if the channel is open for a short period of time, very little inhibition occurs vs. if it is open for a long period of time, a significant amount of inhibition occurs. Overall memantine does not stop AD and progressive neuronal death, but does reduce neuronal death on some level with few side effects.

Previously it was hypothesized that Aβ1-42 acted as an inhibitor against the NMDA family of receptors leading to loss of LTP in AD patients and the advancement of LTD in the affected neurons reasoned from a loss of dendrite density.104 However, this result seems to be somewhat confusing in that NMDA activation inhibition would be detrimental because of the success of memantine. Such confusion is understandable and the nature of this confusion will be addressed later.

Another avenue for neuronal death in AD that has gained traction in recent years is mitochondrial induced apoptosis. Recall from freshman biology that the mitochondria is the ‘powerhouse’ of the cell where a majority of the energy production reactions occur (TCA Cycle, electron transport, etc.) that create a majority of the ATP and other energy storage molecules. However, in addition to its duties in providing energy, the mitochondria also possesses a wide variety of signaling molecules that induce cellular apoptosis like Apoptosis Inducing Factor (AIF), Smac/DIABLO and cytochrome C.1 The chief family of proteins that are responsible for driving the mitochondria apoptotic pathway is the bcl-2 proteins. Some of the bcl-2 proteins induce apoptosis (Bax 1, Bak, Bik and Bad) and others resist apoptosis (bcl-2, bcl-W and bcl-XL).1

Under normal circumstances apoptosis inducing factors are normally scattered through the cytosol in effort to detect any cellular stress/damage. If the protein detects a form of stress (triggered via phosphorylation or some other pathway), then that apoptosis inducing factor migrates to the surface of the mitochondria to interact with an apoptosis resisting factor. If enough inducing factors bind to resisting factors transport pores form in the outer mitochondrial membrane. These pores allow cytochrome C and other more isolated apoptosis inducing factors from the intermembrane space (the area between the outer and inner membranes of the mitochondria) to interact with Apaf-1 forming the apoptosome (pro-caspase 9 + cytochrome C + Apaf-1) and finally activating caspase 9 induced apoptosis.1

In AD it appears possible that most of the initial steps in apoptosis are circumvented by intracellular Aβ1-42 binding to Cyclophilin D (CypD) which forms the necessary mitochondrial permeability transport pores to release cytochrome C and start the apoptosis cascade.105,106 However, there is a question in that it is believed that CypD interacts with AΒ1-42 within the intermembrane space.105 So how does Aβ1-42 pass through the outer mitochondrial membrane to reach the intermembrane space and interact with CypD? Is there an intermediate transmembrane protein that facilitates the interaction without requiring a pore? Also how does the Aβ1-42 get into the cytoplasm in the first place to even have the possibility of passing through the outer mitochondrial membrane?

Addressing the second question first, when the various secretases cleave APP the resultant Aβ peptide fragments are typically secreted out of the neuron or to lysosomes. However, there is the possibility that certain concentrations of Aβ1-42 escape this mechanism leaving them in the cytoplasm until cleaned out by other means possibly driven by APOE4 mutations. While lingering in the cytoplasm the Aβ1-42 could migrate towards the mitochondria. Although the probability exists that Aβ1-42 could be in the appropriate proximity to the mitochondria it is still unclear how it could penetrate the outer membrane. Another means of cytoplasmic concentration increase could involve Aβ1-42 from the extracellular space (largely derived from another neuron or any microglia) passing into a neuron through the endocytic pathway. In fact in late-onset AD early endosomes increase in volume up to approximately 2 times.107 Such an increase could very well increase the input of Aβ1-42 into the neuron and increase the probability that newly internalized Aβ1-42 interacts with CypD leading to apoptosis.

The type of APOE gene that the individual possesses influences endosome size. Apolipoprotein E4 (APOE4) genes eventually produce endosomes that are 1.5 times the size of endosomes derived from APOE2 or APOE3 genes.107 However, as late-onset AD progresses the difference in volume between APOEx and APOE4 endosomes shrinks because for some unknown reason APOE4 endosomes shrink.107 There is reason to believe that this increase in endosome volume occurs years before the legitimate onset of AD107, which may provide a new diagnostic mechanism to identify the probability that an individual will suffer from AD in the future. If an effective method for categorizing endosomes, especially those derived from cortical pyramidal neurons, in vivo can be developed it could dramatically improve the effectiveness of therapeutic treatments that use β-secretase and/or γ-secretase.

There are a couple of avenues of exploration for such a diagnostic methodology. The optimal way, based on certainty and non-invasiveness, would be to design a radioligand that could bind exclusively to early endosomes in vivo and observe any significant size or pattern changes through some form of brain scan like PET. A second idea would be to attempt free-flow zone electrophoresis. Unfortunately such a technique would probably require either a spinal tap or a form of brain bioposy. For a diagnostic procedure that has little single-shot prediction power (if endosomes are not enlarged there is no reason to assume that they will not become enlarged in the future) such a process may not be beneficial overall.

Another aspect of the endocytic pathway is its management and interaction with cholesterol. After cholesterol is synthesized in the endoplasmic reticulum it is processed in the Golgi and shipped to the extracellular matrix via a secretory vesicle. Also LDL is shipped from the extracellular matrix into an endosome where the apoprotein B portion of the LDL is dissolved in a lysosome and the remaining cholesterol is freed and released into the cytoplasm.1 APP cleavage rates appear to be influenced by intracellular free cholesterol levels where higher levels lead to higher rates of cleavage and lower levels generate lower rates of cleavage.108,109 However, there is conflicting data regarding whether or not the administration of statins improves cognitive function. Some retrospective epidemiological studies indicated that statins reduced the probability of developing dementia, but other studies failed to demonstrate any statistically significant protective effects associated with the loss of cognitive functions.110,111 Clearly there are elements that need to be better identified before prescribing statins as a form of treatment for AD.

One of those elements that deserve further investigation may be the LDL receptor-related protein (LRP). LRP is a member of the LDL receptor family and is known to bind and mediate endocytosis of soluble APP (APP cleaved by α-secreatase) and cell surface APP (APP yet to be cleaved by any secreatase) if the APP isoform contains a Kunitz proteinase inhibitor (KPI) domain.112,113,114 The importance of LRP is that it appears to have an influence on whether α-secreatase or β-secreatase/γ-secreatase is the dominant form of action on APP. Blocking LRP with RAP increased both the amount of cell surface APP levels (largely due to an increase in APP synthesis) and increased α-secreatase processing due to the existence of more soluble APP.114 Also after treatment with RAP secretion of Aβ peptides dropped significantly. Technically it cannot be fully concluded that RAP reduces Aβ peptide synthesis due to β-secreatase/γ-secreatase as blocking LRP could simply reduce the amount of Aβ peptide secreted, which would result in much higher intracellular Aβ peptide concentrations, but such a result is unlikely.

There are three possible methodologies with which blocking LRP interferes with the endocytic pathway that favors Aβ peptide synthesis. First, LRP association blocks α-secreatase action before formation of an endosome guaranteeing that β-secreatase/γ-secreatase will have an opportunity to cleave. Second, LRP association induces a conformation change in the APP better exposing the β-secreatase/γ-secreatase cleavage site increasing the probability of a successful cut. Third, LRP association increases the total time within an endosome vs. other endocytic pathway associated proteins, which could then increase the probability that the APP interacts with β-secreatase/γ-secreatase.

The relationship between LRP, APP and AD progression is also strengthened by the fact that silent polymorphism in the LRP gene is associated with increased risk for AD.115 This polymorphism probably influences Aβ peptide concentrations in two different ways. First, it shifts the interaction of APP in favor of β-secreatase/γ-secreatase over α-secreatase. Second, LRP is one of the major receptors used in the transport of Aβ peptides across the BBB. Based on this information it seems reasonable to suggest that increasing the number of LRP for a given cell will increase the APP interaction and the number of synthesized Aβ peptides. This proportional relationship between Aβ peptide synthesis and LRP expression/availability could explain the link between the previous studies that associates a higher cholesterol level with a greater probability of developing AD. The higher the cholesterol level leads to greater LDL and probably LRP expression, which would lead to greater synthesis rate and concentration of AD peptides. If correct this mechanism not only explains a portion of diet association with the development of AD, it also explains why statins have a sketchy history alleviating AD sympotoms.

The principal function of statins is to reduce the overall level of cholesterol synthesis in the liver by inhibiting 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase. Reducing cholesterol synthesis in a given cell will cause the cell to increase its expression of LDL receptors in effort to balance the difference in cholesterol levels between the cytoplasm and extracellular matrix. This up-regulation could also result in the up-regulation of LRP, which as discussed increases the probability of Aβ peptide synthesis. However, if that were the case it would be more probable that statins would hasten the progression of AD instead of possibly slow its progression.

The situation with statins differs from that with high levels of cholesterol because for high cholesterol patients a dynamic equilibrium of LDL receptor expression is attained and typically only increases as extracellular cholesterol levels increase. With statins, the inhibition of HMG-CoA reductase creates a greater LDL receptor expression rate than seen in high cholesterol patients, remember that a vast majority of statin users initially have high cholesterol, in the initial administration of the statins. As time passes this expression rate decreases as a new dynamic equilibrium of LDL receptor expression is reached due to the action of the statin; this new dynamic equilibrium of expression is lower than that of the high cholesterol no-statin patient. Therefore, statins probably do have a small therapeutic role in the treatment of AD symptoms, but this positive effect probably only occurs in high cholesterol patients over a significant time frame. If the above behavior is correct, short-term studies would probably conclude that statins did little to nothing to reduce AD progression or symptoms. Also the effect of statins will be muted because LRP expression will not be eliminated, just reduced.

An increase in β-secretase and/or γ-secretase activity is not the only means responsible for increasing the concentration of Aβ peptides in the brain. There is also a question of the breakdown of the typical clearance mechanisms largely situated near the periphery along the BBB. If clearance mechanisms begin to function improperly even normal synthesis levels of AB peptides can quickly become excessive and increase the probability of detrimental influence in the brain.

Although there are multiple proteins that are responsible for Aβ peptide clearance, the three principle proteins thought to exert the most influence are receptor for advanced end glycation products (RAGE), apoE and LRP.115,116 As previously discussed LRP expression and its resultant interaction with APP is thought to alter various secretase production rates, but LRP also plays a role in the transport of Aβ peptides across the BBB expelling Aβ peptides from the brain into the circulating blood.116 RAGE is a multi-ligand receptor in the immunoglobulin (IgG) superfamily and is the counter agent to LRP as it ferries Aβ peptides from the blood to the brain.116 Over-expression of RAGE or under-expression of LRP can increase the concentration of Aβ peptides in the brain. Fortunately there are self-correcting mechanisms where it appears that if RAGE or LRP have unbalanced interactions with Aβ peptides it facilitates the up-regulation of the under-binding receptor.116 Although disruptions in the RAGE-LRP relationship could increase Aβ peptide concentrations, it seems more probable that the clearance mechanism malfunction responsible for increasing Aβ peptide concentrations in the brain is APOE, which was previously discussed.

In addition to possibly inducing apoptosis, Aβ1-42 is also thought to have a positive effect on the production of superoxides, which generate cell damage and inhibits aconitase, which is an enzyme used in the TCA cycle, significantly reducing the ability of the cell to produce energy.63 Although these studies do identify that increased levels of SOD and deferoxamine (an iron chelator) reduce the total level of toxicity of Aβ1-42,63 the role of other neuronal death elements like tau phosphorylation, microglia recruitment and NMDA hyperactivity are not identified, so it is unclear whether mitochondrial based toxicity is a primary neuronal death factor or a secondary one induced much later in the process after neuronal death is a foregone conclusion, kind of like the linebacker that jumps on the pile long after the running back has been tackled.

For example the latter rationality would offer an explanation to how Aβ1-42 could pass through the outer membrane in that a hyperphosphorylated tau could be a trigger for an apoptotic inducer to bind to an apoptotic resistor opening a pore in the membrane and allowing the Aβ1-42 to pass through. However, if that were the case then blocking CypD would not terminate neuronal death, but simply somewhat reduce the speed of neuronal death.

An element that may play a significant role in AD pathology that has been on the backburner until recently is apolipoprotein J (APOJ) or clusterin. The verification of the APOJ gene as a potential link to the progression of AD has refocused attention on its role in AD.12,13 Clusterin is a disulfide-linked heterodimeric glycoprotein117 that plays a role in many biological and pathologic processes such as phagocyte recruitment, tissue reconstruction, cytolysis inhibition and apoptosis.118 Most of the research surrounding clusterin involves its role in cancer and classification as a small quasi-heat shock protein. In cancer clusterin is acts as a stress-associated cytoprotective chaperone118, which is up-regulated during activation of the apoptotic pathway and is able to bind and inhibit activated Bax to reduce the probability of mitochondria pore formation, cytochrome C release and resultant apoptosis.119

Clusterin also plays a role in the AD, although that role is not exactly clear. Similar to cancer, clusterin synthesis is increased during the progression of AD. However, it is unclear whether or not this increases is beneficial or detrimental to the survival of neurons. Originally it was observed that clusterin bound Aβ peptides to prevent further association with other Aβ peptides to neutralize fibrillization and plaque formation.120 Also a Aβ peptide bound to clusterin has a much higher potential of interacting with megalin receptors on glial cells increasing extracellular clearance through endocytosis. Despite these protection methods, naturally produced clusterin does not appear to be able to prevent AD progression.

One reason for the inability of clusterin to derail AD is that there is contradictory information in that in some scenarios clusterin seems to enhance the level of oxidative stress and neuronal death not prevent neuronal death.121,122 Also clusterin increases the probability for the formation of Aβ-derived diffusible ligands (ADDLs),63 which have a higher probability of surface receptor binding over fibril Aβ peptides.

So does clusterin have both a protective and a destructive role in AD? One way to explain this apparent contradiction is characterize clusterin as an element with a type of catalytic effect on Aβ1-4x. At lower concentrations clusterin enhances the aggregation of Aβ peptides hastening their transition from monomers to oligomers, which have the ability to bind to cell surface receptors. At higher concentrations clusterin is able to bind for a longer range of time facilitating greater aggregation into plaques or removal via glial cells lowering the probability of Aβ peptide interaction with surface receptors. However, if this is the case then the binding affinity of clusterin to the Aβ peptides cannot be very significant. Despite this concern such a methodology seems to make sense in that fact that clusterin prevents rapid self-aggregation of Aβ peptides, especially 42 and 43, with preference for a slower aggregation process with clusterin as a centerpiece.

If the above conclusion is correct then clusterin is a valid therapeutic target for treatment of AD through two different pathways. First, an inhibitor of clusterin can be applied in effort to drastically reduce the rate of aggregation of Aβ peptides reducing the probability that they interact with nicotinic and other receptors. Second, stimulation of clusterin can be attempted to supersaturate the extracellular matrix with clusterin which would eliminate the ability of Aβ peptides to bind to receptors due to conformational change brought on by clusterin binding and eventually result in the clearance of both the clusterin and the Aβ peptide commonly via glial cells.

As previously discussed some have concluded that Aβ somehow directly induces LTD via interaction with some portion of a NMDA receptor, which presents a confusing and contradictive result because NMDA antagonists have been noted to improve AD symptoms. In rudimentary terms LTD can be viewed as basically a biological driven NMDA antagonist as dephosphorylation of AMPA receptors prevents them from opening in response to glutamate which in turn prevents them from depolarizing the neuron which removes the blocking magnesium ion in the NMDA receptor channels. If clinically applied NMDA antagonists like memantine improve AD symptoms, it stands to reason that the ‘natural’ method of applying a NMDA antagonist would do the same instead of increase the severity of AD.

The confusion from this issue can be resolved when considering two separate factors. First, a vast majority of AD experimentation, understandably, occurs in a vacuum where condition or pathway is not measured against another pathway. This form of isolation largely results in the single question of: does eliminating or enhancing factor x lead to neuronal damage/death? The failure to consider other pathways results in a failure to consider the second factor, speed of neuronal death.

There can be no logical argument that LTD is a detrimental outcome, but when all other options that result in neuronal death are considered, LTD is a much better outcome. The following example better illustrates the above reasoning. Suppose that an individual had to select one of the following five options: be shot in the head, have the throat slit, be given radiation poisoning, be injected with HIV or receive a subdural hematoma. Although none of the presented options are desirable, the best option would be to be injected with HIV because although there is a high probability of death, the quality of life after making the choice is significantly better than any of the remaining choices.

This circumstance may describe Aβ driven LTD, which can be classified as a condition that will probably lead to death, but will take a longer time to reach that outcome over other detrimental pathways influenced by Aβ peptides in AD. Basically it is slowest means of neuronal death.

Overall the biggest problem with AD may be all the results from all of the knockout experiments or isolation experiments seem to generate contradictory views on what elements in the pathology are important. Various studies knockout component x and seem to demonstrate that without that component neuronal death due to AD either no longer occurs or is significantly reduced. Unfortunately these types of results drive the ‘silver bullet’ mindset or drug development in that if a drug can be developed to neutralize component x then instant AD cure. However, these experiments rarely touch on the other elements that could induce neuronal death. Basically all of these experiments seem to take place in a vacuum where only component x is important. Such is not the case in reality for AD appears remarkably complex.

The best way to illustrate this apparent disconnect between AD drug testing and biological reality is with the following example. Suppose that a criminal is being put to death, but the individuals responsible for deciding upon the method of death cannot agree to any particular method, so instead of selecting a single method they elect to apply all five methods they have been debating. Therefore, the criminal is lead out to the killing ground and set on fire, shot in the heart, electrocuted, injected with poison and has his throat slit. Now reset the situation and suppose that just before the man is executed an anti-death penalty advocate comes running forward and puts a metal collar around the criminal’s neck believing that it will save his life. Unfortunately for the advocate, no such luck as even though the collar does protect the criminal from getting his throat slit, it does not protect him from getting set on fire, shot in the heart, electrocuted or injected with poison.

Clearly any bystander would view such a situation with confusion, for how could the advocate believe that preventing only one method of death would spare the criminal? The same logic can be applied to AD drug testing. No wonder most AD drug investigations fail when reaching Phase III testing; investigating to see if drug x can extend lifespan and mental conditions by neutralizing death condition y does not seem to be productive if death condition z is not also lessened/neutralized by drug x. It can be argued that electrocuting the criminal does make it easier to set him aflame, but setting him on fire is not dependent on electrocuting him. Therefore, it makes more sense that when testing for possible AD drugs that can limit neuronal death and progression of the condition that a drug cocktail be used instead of a single drug. Attacking AD on multiple fronts is the only guaranteed way to develop an effective and successful treatment.

So the big question is ‘where to attack’? To know where to attack, one must have a general idea of the pathology of AD. Early onset AD is driven primarily by mutations in either the APP, presenilin 1, presenilin 2 which accelerate the rate at which Aβ1-42 is cleaved from APP. Due to the existence of APP in dendrites, cell bodies and axons the additional Aβ1-42 concentrations can become diffuse throughout the local region of the mutation. Under normal conditions the vast majority of Aβ1-40, Aβ1-42 and any other Aβ peptides are neutralized. In this situation concentrations of Aβ1-42 are small enough that there is no significant neuronal damage. Recall that Aβ1-42 is able to interact with clusterin or other monomers of Aβ1-42 outside of the cell and begin to form dimers, trimers and larger oligomers. There is reason to believe that self-Aβ1-42 aggregation is more rapid than clusterin induced Aβ1-42 aggregation. This rapid form of self-Aβ1-42 aggregation may be more beneficial then detrimental because plaques probably limit the neurotoxicity of Aβ1-42. If clusterin concentrations are saturating then it is likely that most Aβ1-42 will be neutralized. If not, then there is reason to believe that clusterin could aid in the formation of deleterious Aβ1-42 oligomers.

One result from the formation of these Aβ1-42 oligomers is that they are able to bind to nicotinic acetylcholine receptors (preferably the α-7 family) providing a sufficient level of direct inhibition. This inhibition forces the channel to stay open longer to generate a depolarized state in the neuron. The additional time open hastens the already inherent rapid desensitization processes in nicotinic receptors.123

The desensitization process for nicotinic ACh receptors involves the activation of protein kinase A (PKA), protein kinase C (PKC) or tyrosine kinase. PKA phosphorylates the gamma and the delta subuints of the receptor. PKC phosphorylates the alpha and delta subunits of the receptor. Tyrosine kinase phosphorylates the beta, gamma and delta subunits of the receptor.1 These three elements inactivate nicotinic acetylcholine receptors. Although it initially takes time, the desensitization process initiated in part by a family of tyrosine kinases could be responsible for the hyperphosphorylation of tau. Eventually more and more tau are hyperphosphorylated causing a retraction in the axon due to the tau either losing the ability to stabilize the microtubules or tau reverting back to its development role of axonal development. If the latter option is correct then it is probable that the tau proteins that are driving this new axonal development tangle with each other forming NFTs before the new axon can be rebuilt.

There is little reason to question that interrupted axonal regrowth or significant axonal destabilization would trigger an apoptotic reaction in the cell. It is also possible that the formation of NFTs could also induce an apoptotic response through normal mitochondria apoptotic influences (inducer binds resistor). Apoptosis destroys the neuron causing large concentrations of intracellular glutamate to leak out of the cell into the extracellular matrix and various nearby synaptic clefts. This non-signalled glutamate is of significant concentration that it is highly probable that it interacts with AMPA and NMDA receptors on surrounding neurons, most notably in pyramidal cells in the CA1 and CA2 regions of the hippocampus.

Binding to the AMPA receptors initially depolarizes the cell removing the magnesiums blocking the NMDA receptor channels allowing for the dramatic increase in the influx of calcium. Due to the excessive concentrations of glutmate remaining in the synaptic cleft, due to the sizable intracellular concentrations of glutamate in normal neurons, the depolarization of the given neuron continues for an extended period far beyond normal excitation. The calcium driven secondary messenger system activates various kinases in the neurons in the hippocampus affected by the released glutamate including tyrosine kinase fyn and p38 MAPK, which phosphorylates tau and initiates the first series of steps in the apoptotic pathway.

This activation of fyn and p38 MAPK implies that glutaminergic neurons in the hippocampus that die in AD may have their apoptotic pathway activated via multiple different means. Also such a situation would explain the existence of NFTs in the hippocampus where cholinergic neurons are not as plentiful as other regions. Significant neuronal death in the hippocampus would also interfere with LTP and thus interfere with memory, learning and other cognitive abilities, hallmark symptoms of AD, and similar to inhibition of cholinergic neurons. This dual detrimental effect would compound any learning problems that arise from the partial inhibition of the acetylcholine receptors and their channels.

There is also the possibility of long term inhibition of NMDA receptors through long term depression of AMPA receptors if the neuron is not killed via apoptosis or Aβ1-42 somehow binds to the NMDA receptor. These processes (the excess calcium influx or Aβ1-42 binding) lead to the activation of calcineurin. Calcineurin dephosphorylates inhibitor-1, increasing the activity of phosphatase-1 resulting in dephosphorylation of the AMPA receptors.1 This dephosphorylation closes the AMPA channels heavily reducing the influx concentration of Na+, leading to the repolarization of neuron and the return of the magnesium channel blocker which would stop in the influx of calcium through the NMDA gated channels. This process would prevent LTP inducing cognitive damage, but it is unclear if it would eventually kill the neuron (although it is likely). Also phosphatase-1 may play a role in dephosphorylating tau which would reduce the probability of neuronal death via microtubule destabilization.

It is also highly probable that internal Aβ1-42 driven apoptosis plays a role in neuronal death, although the magnitude of that role is unknown, albeit at the moment it seems probable that its role is secondary and limited. The reason that this pathway of neuronal death is thought to be limited is that first in either early or late onset a genetic mutation is responsible for the necessary increase in Aβ1-42 concentration to generate the required concentrations for a high enough probability that Aβ1-42 will remain in the neuron and achieve the proximity required to act with CypD to induce apoptosis. Therefore, in this scenario there would need to be a large number of genetic mutations throughout specific regions of the brain, for internal Aβ1-42 driven apoptosis to be a primary means of cell death, which is unlikely. Second, it is highly unlikely that even if certain oligomer structures of Aβ1-42 could enter the neuron through an open ion channel or an endosome, that most of these oligomers would be available to do so as most of these oligomers would either be bound into a plaque or bound to a receptor site. There is the possibility that AD patients will high cholesteral levels could have a higher probability of suffering from internal Aβ1-42 driven apoptosis due to increased levels of Aβ1-42 production due to greater LRP expression.

Microglia derived neuronal death is the trickiest of all pathways to classify. The significance of microglia interaction and neuronal death is that it seems probable that it follows a somewhat inversely proportional rate of influence vs. the progression of the AD. In early and mid-stages of AD microglia derived death could be a primary means of neuronal death due to its fast action both in breadth and toxicity damage. However, as AD progresses the role of microglia derived death more than likely becomes less significant due to higher concentrations of Aβ1-42 in the extracellular matrix, glutamate release due to neuronal death and possible inactivation of microglia cells due to loss of receptors. So in the early stages of AD the primary means of neuronal death can be attribute to microglia and Aβ1-42 binding. As AD progresses NMDA excitotoxicity due to excess glutamate more than likely begins to outpace microglia influenced NMDA excitotoxicity.

The role of Aβ1-42 binding to prions may be the most interesting. There is reason to believe that formation of the Aβ1-42-prion complex somehow influences inhibition of NMDA channel activation through the calcineurin pathway. Such a result is interesting because the Aβ1-42-prion complex may actually extend the lifespan of the AD patient by reducing the severity of any excitotoxicity influence on hippocampal neurons in favor of the more muted LTD response that will typically kill the neuron, but do so at a slower rate and with less collateral damage. Therefore, the Aβ1-42-prion complex may actually be ‘biologically’ therapeutic relative to other Aβ1-42 pathways vs. its popularly detrimental reputation.

The figure below illustrates the various pathways towards neuronal death in AD. Note that there is little difference between early onset and late onset AD. The chief difference is that late onset involves a mutation in the APOE4 gene which influences the endocytic pathway in a cell whereas early onset typically involves a mutation in the APP, pre1 or pre2 gene.

So if the general pathogensis of the AD is properly described and illustrated above, what are possible treatments? Plaque busters like flurizan do not appear to have any role in AD treatment because destroying plaques that do not appear to have toxic affects can be nothing but detrimental (potentially releasing new non-fibril Aβ1-42 oligomers). Inhibiting β and/or γ-secretase could serve a useful purpose limiting the total amount neuronal damage, but it seems reasonable to suggest that the usefulness of β and/or γ-secretase inhibitors is inversely proportional to the progression of AD. Realistically β and/or γ-secretase inhibitors only appear to be really useful in the interim stages of AD as later excitotoxicity reduce its influence.

Currently inhibiting the CypD/Aβ1-42 interaction would probably only slightly reduce neuronal death because it seems to play a secondary pathway for neuronal death despite its speed. Anti-inflammatory drugs would work well early in AD preventing microglia collateral damage, but their effectiveness should fade as AD progresses; however, there is reason to hypothesize that the influence of anti-inflammatory drugs would fade at a slower rate than β or γ-secretase inhibitors.

It would be difficult to recommend the continuation of cholinesterase inhibitors if in fact Aβ has a repolarizing/inhibitory effect as reducing the inhibitory effect associated to Aβ binding nicotinic receptors seems to increase the probability of Aβ peptide synthesis which would exasperate the progression of AD for the sake of limited short-term gains.

Until the T-cell issues are effectively neutralized the application of Aβ antibodies appear to be too dangerous and unpredictable to be used as an effective therapeutic option. Inhibition of clusterin, which plays a role in the formation of Aβ1-4x oligomers from monomers, may be an interesting inhibitory target as Aβ1-4x monomers do not appear to be able to bind to intra and extracellular receptor sites.

Inhibition of tau hyperphosphorylation seems to be a potential important treatment strategy, but also the most complicated. There are at least 3 different kinases (fyn, GSK-3β and p38 MAPK) that have empirical backing for playing some role in tau phosphorylation. Unfortunately these studies typically only targeted one of the three kinases and did not account for either of the other two. Therefore, it is difficult to identify if there is any hierarchy within these kinases where only one needs to be blocked to prevent hyperphosphorylation of tau or if more than one needs to be blocked to fully prevent tau. The best option may be looking at fyn first because fyn appears to play a more limited role in normal homeostasis than GSK-3β or p38 MAPK. If tau phosphorylation on trysone18 can be blocked it may prevent the N terminus of tau from interacting with the proline-rich area creating a conformation change that could reduce the probability of phosphorylation at other sites.

Overall there appear to be a variety of different neuronal death elements that occur during AD, thus there are multiple avenues of therapeutic attack, but such attacks can be short-circuited if considerations are not made for the other mechanisms of neuronal death. Therefore, drug trials cannot longer be limited to single drug at a time, but instead must use multiple drug cocktails in addition to single trials to increase the probability of successfully treating AD. In short when it comes to AD drug companies need to stop thinking about unilateral control of AD treatments and develop Phase II and Phase III testing partnerships for 25-50% of an AD treatment is much better than 100% of failure.

1. Kendel, Eric, Schwartz, James, Jessell, Thomas. Principles of Neural Science. 4th Edition. McGraw-Hill 2000.

2. Waldemar, G, et, Al. “Recommendations for the diagnosis and management of Alzheimer's disease and other disorders associated with dementia: EFNS guideline.” Eur J Neurol. 2007. 14(1): 1–26.

3. Bäckman, L, et, Al. “Multiple cognitive deficits during the transition to Alzheimer's disease.” J Intern Med. 2004. 256(3): 195-204.

4. Wang, H, et, Al. “β-Amyloid 1-42 binds to α-7 nicotinic acetylcholine receptor with high affinity.” J Biol Chem. 2000. 275: 5626-5632.

5. Hardy, J.A. and Higgins, G.A. “Alzheimer’s disease: the amyloid cascade hypothesis.” Science. 1992. 256: 184–185.

6. Iqbal, K, et, Al. “Tau pathology in Alzheimer disease and other tauopathies.” Biochim Biophys Acta. 2005. 1739(2-3): 198–210.

7. Nistor, M, et, Al. “Alpha and beta-secretase activity as a function of age and beta-amyloid in Down syndrome and normal brain.” Neurobiol Aging. 2007. 28(10): 1493-1506.

8. Lott, I, and Head, E. “Alzheimer disease and Down syndrome: factors in pathogenesis.” Neurobiol Aging. 2005. 26(3): 383-89.

9. Polvikoski, T, et, Al. “Apolipoprotein E, dementia, and cortical deposition of beta-amyloid protein.” N Engl J Med. 1995. 333(19): 1242–47.

10. Mouillet-Richard, S, et, Al. “Signal transduction through prion protein.” Science. 2000. 289: 1925-1928.

11. Klein, William, Krafft, Grant, Finch, Caleb. “Targeting small Ab oligomers: the solution to an Alzheimer’s disease conundrum?” TRENDS in Neurosciences. 2001. 24(4): 219-223.

12. Lambert, J, et, Al. “Genome-wide association study identifies variants at CLU and CR1 associated with Alzheimer’s disease.” Nature Genetics. Published online: 6 September 2009.

13. Harold, Denise, et, Al. “Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer's disease.” Nature Genetics. Published online: 6 September 2009. doi:10.1038/ng.440.

14. Selkoe, D. “Translating cell biology into therapeutic advances in Alzheimer’s disease.” Nature. 1999. 399: A23-31.

15. Wang, H, et, Al. “Amyloid peptide Aβ(1-42) binds selectively and with picomolar affinity to α-7 nicotinic acetylcholine receptors. J. Neurochem. 2000. 75: 1155-1161.

16. Pettit, D, Shao, Z, Yakel, J. “β-Amyloid1-42 Peptide Directly Modulates Nicotinic Receptors in the Rat Hippocampal Slice.” The Journal of Neuroscience. 2001. 21: 1-5.

17. Mucke, L, et, Al. “High-level neuronal expression of Aβ1-42 in wild-type human amyloid protein precursor transgenic mice: synaptotoxicity without plaque formation.” J Neurosci. 2000. 20: 4050-4058.

18. Kamenetz, F, et, Al. “APP processing and synaptic function.” Neuron. 2003. 37: 925-937.

19. Steinbach, J, et, Al. “Hypersensitivity to seizures in beta-amyloid pre-cursor protein deficient mice.” Cell Death Differ. 1999. 5: 858-866.

20. Lauren, Juha, et, Al. “Cellular Prion Protein Mediates Impairment of Synaptic Plasticity by Amyloid-β Oligomers.” Nature. 2009. 457(7233): 1128–1132.

21. Büeler, H, et, Al. “Normal development and behavior of mice lacking the neuronal cell-surface PrP protein.” Nature. 1992. 356: 577–582.

22. Manson, J, et, Al. “129/Ola mice carrying a null mutation in PrP that abolishes mRNA production are developmentally normal.” Mol Neurobiol. 1994. 8: 121–127.

23. Stahl, N, et, Al. “Scrapie prion protein contains a phosphatidylinositol glycolipid.” Cell. 1987. 51: 229–249.

24. Westergard, Laura, Christensen, Heather, Harris, David. “The cellular prion protein (PrPC): its physiological function and role in disease.” Biochim Biophys Acta. 2007. 1772(6): 629–644.

25. Gorodinsky, A, and Harri,s D. “Glycolipid-anchored proteins in neuroblastoma cells form detergentresistant complexes without caveolin.” J Cell Biol. 1995. 129: 619–627.

26. Bounhar, Y, et, Al. “Prion protein protects human neurons against Bax-mediated
apoptosis.” J Biol Chem. 2001. 276: 39145–39149.

27. Roucou, X, et, Al. “Cytosolic prion protein is not toxic and protects against Bax-mediated cell death in human primary neurons.” J Biol Chem. 2003. 278: 40877–40881.

28. Roucou, X, and LeBlanc A. “Cellular prion protein neuroprotective function: implications in prion diseases.” J Mol Med. 2005. 83: 3–11.

29. Shmerling, D, et, Al. “Expression of amino-terminally truncated PrP in the mouse leading to ataxia and specific cerebellar lesions.” Cell. 1998. 93: 203–214.

30. Brown, D, et, Al. “Prion protein-deficient cells show altered response to oxidative stress due to decreased SOD-1 activity.” Exp Neurol. 1997. 146: 104–112.

31. Brown, D, Nicholas, R, Canevari, L. “Lack of prion protein expression results in a neuronal phenotype sensitive to stress.” J Neurosci Res. 2002. 67: 211–224.

32. Spudich, A, et, Al. “Aggravation of ischemic brain injury by prion protein deficiency: Role of ERK-1/-2 and STAT-1.” Neurobiol Dis. 2005. 20: 442–449.

33. Brown, D, et, Al. “Normal prion protein has an activity like that of superoxide dismutase.” Biochem J. 1999. 344: 1–5.

34. Klamt, F, et, Al. “Imbalance of antioxidant defense in mice lacking cellular prion protein.” Free Radic Biol Med. 2001. 30: 1137–1144.

35. Rae, T, et, Al. “Undetectable intracellular free copper: the requirement of a copper chaperone for superoxide dismutase.” Science. 1999. 284: 805–808.

36. Schneider, B, et, Al. “NADPH oxidase and extracellular regulated kinases 1/2 are targets of prion protein signaling in neuronal and nonneuronal cells.” PNAS. 2003. 100: 13326–13331.

37. Walsh, D.M, et Al. “Amyloid beta-protein fibrillogenesis. Structure and biological activity of protofibrillar intermediates.” J. Biol. Chem. 1999. 274: 25945–25952.

38. Hartley, D.M, et Al. “Protofibrillar intermediates of amyloid beta-protein induce acute electrophysiological changes and progressive neurotoxicity in cortical neurons.” J. Neurosci. 1999. 19: 8876–8884.

39. Terry, R.D, et, Al. “The neuropathology of Alzheimer disease and the structural basis of its cognitive alterations.” Alzheimer Disease. 1999. 187–206.

40. Einstein, G, Buranosky, R, Crain B. “Dendritic pathology of granule cells in Alzheimer’s disease is unrelated to neuritic plaques.” J. Neurosci. 1994. 14: 5077-5088.

41. Katzman, R, et, Al. “Clinical, pathological and neurochemical changes in dementia: a subgroup with preserved mental status and numerous neocortical plaque.” Ann. Neurol. 1988. 23: 138-144.

42. Mucke, L, et Al. “High-level neuronal expression of Ab 1–42 in wild-type human amyloid protein precursor transgenic mice: synaptotoxicity without plaque formation.” J. Neurosci. 2000. 20: 4050–4058

43. Giulian, Dana, et, Al. “Specific Domains of b-Amyloid from Alzheimer Plaque Elicit Neuron Killing in Human Microglia.” The Journal of Neuroscience. 1996. 16(19): 6021–6037.

44. Dickson, T, and Vickers, J. “The morphological phenotype of B-amyloid plaques and associated neuritic changes in Alzheimer’s disease.” Neuroscience. 2001. 105(1): 99-107.

45. Schenk, D, et, Al. “Immunization with amyloid-beta attenuates Alzheimer-disease-like pathology in the PDAPP mouse.” Nature. 1999. 400: 173-177.

46. Lombardo J, et, Al. “Amyloid-beta antibody treatment leads to rapid normalization of plaque-induced neuritic alterations.” J. Neurosci. 2003. 23: 10879-10883.

47. Oddo, S, et, Al. “Abeta immunoterapy leads to clearance of early, but not late, hyperphosphorylated tau aggregates via the proteasome.” Neuron. 2004. 43: 321-332.

48. Brendza, R, et, Al. “Anti-Abeta antibody treatment promotes the rapid recovery of amyloid associated neuritic dystrophy in PDAPP transgenic mice.” J. Clin. Invest. 2005. 115: 428-433.

49. DeMattos, R, et, Al. “Peripheral anti-Abeta antibody alters CNS and plasma Abeta clearance and decreases brain Abeta burden in a mouse model of Alzheimer’s disease. PNAS. 2001. 98: 8850-8855.

50. Bard, F, et, Al. “Epitope and isotype specificities of antibodies to beta-amyloid peptide for protection against Alzheimer’s disease-like neuropathology.” PNAS. 2003. 100: 2023-2028.

51. Wilcock, D, et, Al. “Intracranially administered anti-Abeta antibodies reduce beta-amyloid deposition by mechanisms both independent of and associated with microglial activation.” J. Neurosci. 2003. 23: 3745-3751.

52. Ferrer, I, et, Al. “Neuropathology and pathogenesis of encephalitis following anyloid-beta immunization in Alzheimer’s disease.” Brain Pathol. 2004. 14: 11-20.

53. Bayer A, et, Al. “Evaluation of the safety and immunogenicity of synthetic Abeta42 (AN1792) in patients with AD.” Neurology. 2005. 64: 94-101.

54. Gilman, S, et, Al. “Clinical effects of Abeta immunization (AN1792) in patients with AD in an interrupted trail.” Neurology. 2005. 64: 1553-1562.

55. Giulian, D, et, Al. “Senile plaques stimulate microglia to release a neurotoxin found in Alzheimer brain.” Neurochem Int. 1995. 27: 119-137.

56. Scali, C, et, Al. “Beta(1–40) amyloid peptide injection into the nucleus basalis of rats induces microglia reaction and enhances cortical gamma-aminobutyric acid release in vivo.” Brain Res. 1999. 831: 319-321.

57. Wegiel, J, et, Al. “The role of microglial cells and astrocytes in fibrillar plaque evolution in transgenic APP(SW) mice.” Neurobiol Aging. 2001. 22: 49–61.

58. Gordon, M, et, Al. “Time course of the development of Alzheimer-like pathology in the doubly transgenic PS1þAPP mouse.” Exp Neurol. 2002 173: 183–195.

59. Weldon, D, et, Al. “Fibrillar beta-amyloid induces microglial phagocytosis, expression of inducible nitric oxide synthase, and loss of a select population of neurons in the rat CNS in vivo.” J Neurosci. 1998. 18: 2161–2173.

60. Breitner, J, Gau, B, Welsh, K. “Inverse association of anti-inflammatory treatments and Alzheimer’s disease: initial results of a co-twin control study.” Neurology. 1990. 44: 227-232.

61. Lucca, U, et, Al. “Non-steroidal anti-inflammatory drug use in Alzheimer’s disease.” Biol. Psychiatry. 1994. 36: 854-856.

62. Stephan, A, Laroche, S, Davis, S. “Learning deficits and dysfunctional synaptic plasticity induced by aggregated amyloid deposits in the dentate gyrus are rescued by chronic treatment with indomethacin.” Eur. J. Neurosci. 2003. 17: 1921-1927.

63. Longo, Valter, et, Al. “Reversible Inactivation of Superoxide-Sensitive Aconitase in Ab1–42-Treated Neuronal Cell Lines.” Journal of Neurochemistry. 2000. 75(5): 1977-1985.

64. Li, Yuekui, et, Al. “Interleukin-1 Mediates Pathological Effects of Microglia on Tau Phosphorylation and on Synaptophysin Synthesis in Cortical Neurons through a p38-MAPK Pathway.” The Journal of Neuroscience. 2003. 23(5): 1605-1611.

65. Reynolds, C, et, Al. “Phosphorylation sites on tau identified by nanoelectrospray mass spectrometry: differences in vitro between the mitogen-activated protein kinases ERK2, c-Jun N-terminal kinase and P38, and glycogen synthase kinase-3beta.” J Neurochem. 2000. 74: 1587–1595.

66. Sheng, J, et, Al. “Interleukin-1 promotion of MAPK-p38 overexpression in experimental animals and in Alzheimer’s disease: potential significance for tau protein phosphorylation.” Neurochem Int. 2001. 39: 341–348.

67. Griffin, W, and Mrak, R. “Interleukin-1 in the genesis and progression of and risk for development of neuronal degeneration in Alzheimer’s disease.” J Leukoc Biol. 2002. 72: 233–238.

68. De Strooper, B. “Aph-1, Pen-2 and nicastrin with presenilin generate an active gamma-secreatse complex.” Neuron. 2003. 38: 9-12.

69. Fisher, A. “Therapeutic strategies in Alzheimer’s disease: M1 muscarinic agonists.” Jpn J. Pharmacol. 2000. 84: 101-112.

70. Caccamo, A, et, Al. “M1 receptors play a central role in modulating AD-like pathology in transgenic mice.” Neuron. 2006. 49: 671-682.

71. Mandell J, and Banker, G. “A spatial gradient of tau protein phosphorylation in nascent axons.” J Neurosci. 1996. 16: 5727–5740.

72. Kosik, K. “Tau: structure and function.” Brain Microtubule Associated Proteins. 1997. pp. 43-52.

73. Goedert, M, et, Al. “Moleculear dissection of the paired helical filament.” Neurobiol. Aging. 1995. 16: 325-334.

74. Billingsley, M, and Kincaid, R. “Regulated phosphorylation and dephosphorylation of tau protein – effects on microtubule interaction, intracellular trafficking and neurodegeneration. Biochem J. 1997. 323: 577-591.

75. Lee, Gloria, et, Al. “Phosphorylation of Tau by Fyn: Implications for Alzheimer’s Disease.” The Journal of Neuroscience. 2004. 24(9):2304 –2312.

76. Grace, E, and Busciglio, J. “Aberrant activation of focal adhesion proteins mediates fibrillar amyloid beta-induced neuronal dystrophy.” J Neurosci. 2003. 23: 493-502.

77. Rapoport, M, et, Al. “Tau is essential to beta-amyloid-induced neurotoxicity.” PNAS. 2002. 99: 6364-6369.

78. Noble, W, et, Al. “Cdk5 is a key factor in tau aggregation and tangle formation in vivo.” Neuron 2003: 38: 555–65.

79. Town, T, et, Al. “p35/Cdk5 pathway mediates soluble amyloid-beta peptide-induced tau
phosphorylation in vitro.” J. Neurosci. Res. 2002. 69: 362–372.

80. Ratner, N, Bloom, G, Brady, S. “A role for cyclindependent kinase(s) in the modulation of fast anterograde axonal transport: effects defined by olomoucine and the APC tumor suppressor protein.” J. Neurosci. 1998. 18: 7717–7726.

81. Hanger, D, et, Al. “Glycogen synthase kinase-3 induces Alzheimer's disease-like phosphorylation of tau: generation of paired helical filament epitopes and neuronal localisation of the kinase.” Neurosci Lett. 1992. 147(1): 58–62.

82. Noble, W, et, Al. “Inhibition of glycogen synthase kinase-3 by lithium correlates with reduced tauopathy and degeneration in vivo.” PNAS. 2005. 102: 6990–5.

83. Lucas JJ, et, Al. “Decreased nuclear beta-catenin, tau hyperphosphorylation and neurodegeneration in GSK-3beta conditional transgenic mice.” EMBO J. 2001. 20: 27–39.

84. Spittaels, K, et, Al. “Glycogen synthase kinase-3beta phosphorylates protein tau and rescues the axonopathy in the central nervous system of human four-repeat tau transgenic mice.” J Biol Chem. 2000. 275: 41340–9.

85. Hasegawa, M, et, Al. “Characterization of mAb AP422, a novel phosphorylation-dependent monoclonal antibody against tau protein.” FEBS Lett. 1996. 384: 25–30.

86. Goedert, M, et, Al. “Phosphorylation of microtubule-associated protein tau by stress-activated protein kinases.” FEBS Lett. 1997. 409: 57–62.

87. Necula, M, and Kuret, J. “Pseudophosphorylation and Glycation of Tau Protein Enhance but Do Not Trigger Fibrillization in Vitro.” J. Biol. Chem. 2004. 279: 49694-49703.

88. Pei, J, et, Al. “Up-regulation of mitogen-activated protein kinases ERK1/2 and MEK1/2 is associated with the progression of neurofibrillary degeneration in Alzheimer’s disease.” Brain Res Mol Brain Res. 2002. 109: 45–55.

89. Le Corre, S, et, Al. “An inhibitor of tau hyperphosphorylation prevents severe motor impairments in tau transgenic mice.” PNAS. 2006. 103: 9673–8.

90. Latimer, D, et, Al. “Stimulation of MAP kinase by v-raf transformation of fibroblasts fails to induce hyperphosphorylation of transfected tau.” FEBS Lett. 1995. 365: 42–6.

91. Ho, D, Shayan, H, Murphy, T. “Okadaic acid induces hyperphosphorylation of tau independently of mitogen-activated protein kinase activation.” J Neurochem. 1997. 68: 106–11.

92. Giasson, B, et, Al. “The environmental toxin arsenite induces tau hyperphosphorylation.” Biochemistry. 2002. 41: 15376–87.

93. Kins, S, et, Al. “Reduced protein phosphatase 2A activity induces hyperphosphorylation and altered compartmentalization of tau in transgenic mice. J Biol Chem. 2001. 276: 38193–200.

94. Gong, C, et, Al. “Phosphorylation of microtubule-associated protein tau is regulated by protein phosphatase 2A in mammalian brain. Implications for neurofibrillary degeneration in Alzheimer’s disease.” J Biol Chem. 2000. 275: 5535–44.

95. Shirazi, S, and Wood, J. “The protein tyrosine kinase, fyn, in Alzheimer’s disease pathology.” Neuroreport. 1993. 4: 435-437.

96. Lee, Gloria, et, Al. “Tau interacts with src-family non-receptor tyrosine kinases.” Journal of Cell Science. 1998. 111: 3167-3177.

97. Lambert, M, et, Al. “Diffusible, non-fibrillar ligands derived form Abeta1-42 are potent central nervous system neurotoxins.” PNAS. 1998. 95: 6448-6453.

98. Mandell, J, and Banker, G. “The microtubule cytoskeleton and the development of neuronal polarity.” Neurobiol. Aging. 1995. 16: 299-237.

99. Black, M, et, Al. “Tau is enriched on dynamic microtubules in the distal region of growing axons.” J. Neurosci. 1996. 16: 3601-3619.

100. Bixby, J, Jhabvala, P. “Tyrosine phosphorylation in early embryonic growth cones.” J. Neurosci. 1993. 13: 3421-3432.

101. Mandell, J, and Banker, G. “A spatial gradient of tau protein phosphorylation in nascent axons.” J. Neurosci. 1996. 16: 5727-5740.

102. Gordon-Weeks, P. “MAPs in growth cones.” Brain Microtubule Associated Proteins. 1997. 53-72.

103. Lipton, Stuart. “Paradigm shift in neuroprotection by NMDA receptor blockade: Memantine and beyond.” Nature Reviews Drug Discovery. 2006. doi:10.1038/nrd1963.

104. Shankar, Ganesh, et, Al. “Natural Oligomers of the Alzheimer Amyloid-ß Protein Induce Reversible Synapse Loss by Modulating an NMDA-Type Glutamate Receptor-Dependent Signaling Pathway.”

105. Du, Heng, et, Al. “Cyclophilin D deficiency attenuates mitochondrial and neuronal perturbation and ameliorates learning and memory in Alzheimer’s disease.” Nature Medicine. 2008. 14: 1097-1105.

106. Du, Heng, and Yan, Shirley. “Mitochondrial permeability transition pore in Alzheimer's disease: Cyclophilin D and amyloid beta.” 2009. doi:10.1016/j.bbadis.2009.07.005

107. Cataldo, Anne, et, Al. “Endocytic Pathway Abnormalities Precede Amyloid β Deposition in Sporadic Alzheimer’s Disease and Down Syndrome: Differential Effects of APOE Genotype and Presenilin Mutations.” American Journal of Pathology. 2000. 157(1): 277-286.

108. Simons, M, et, Al. “Cholesterol depletion inhibitions the generation of beta-amyloid in hippocampal neurons.” PNAS. 1998. 95: 6460-6464.

109. Cordy, J, et, Al. “Exclusively targeting beta-secretase to lipid rafts by GPI-anchor addition up-regulates beta-site processing of the amyloid precursor protein.” PNAS. 2003. 100: 11735-11740.

110. Jick, H, et, Al. “Statins and the risk of dementia.” Lancet. 2000. 356: 1627-1631.

111. Li, G, et, Al. “Statin therapy and risk of dementia in the elderly: a community-based prospective cohort study.” Neurology. 2004. 63: 1624-1628.

112. Kounnas, M, et, Al. “LDL receptor-related protein, a multifunctional ApoE receptor, binds secreted beta-amyloid precursor protein and mediates its degradation.” Cell. 1995. 82: 331-340.

113. Knauer, M, et, A. “Cell Surface APP751 Forms Complexes with Protease Nexin 2 Ligands and is Internalized via the Low Density Lipoprotein Receptor-Related Protein (LRP).” Brain Res. 1996. 740: 6-14.

114. Ulery, Paula, et, Al. “Modulation of β-amyloid precursor protein processing by the low density lipoprotein receptor-related protein (LRP).” The Journal of Biological Chemistry. 2000. 275(10): 7410-7415.

115. Yan, S, et al. “RAGE and amyloid-beta peptide neurotoxicity in Alzheimer’s disease.” Nature. 1996. 382 (6593): 685–691.

116. Deane, Rashid, Wu, Zhenhua, Zlokovic, Berislav. “Rage (yin) versus LRP (Yang) balance regulates alzheimer amyloid {beta}-peptide clearance through transport across the blood-brain barrier.” Stroke: Journal of the American Heart Association. 2004. 35: 2628-2631.

117. Blaschuk, O, Burdzy, K, Fritz, I. “Purification and characterization of a cell-aggregating factor (clusterin), the major glycoprotein in ram rete testis fluid.” J Biol Chem. 1983. 258:7714–20.

118. So, Alan, et, Al. “Knockdown of the cytoprotective chaperone, clusterin, chemosensitizes human breast cancer cells both in vitro and in vivo.” Mol Cancer Ther. 2005. 4(12): 1837-1849.

119. Zhang, H, et, Al. “Clusterin inhibits apoptosis by interacting with activated Bax.” Nat Cell Biol. 2005: 7: 909–15.

120. Boggs, Leonard, et, Al. “Clusterin (Apo J) protects against in vitro amyloid β (1-40) neurotoxicity.” Journal of Neurochemistry. 1996. 67: 1324-1327.

121. Oda, T, et, Al. “Purification and characterization of brain clusterin.” Biochem. Biophys. Res. Commun. 1994. 204: 1131-1136.

122. Oda, T, et, Al. “Clusterin (ApoJ) alters the aggregation of amyloid β-peptide (Aβ1-42) and forms slowly sedimenting Aβ complexes that cause oxidative stress.” Exp. Neurol. 1995. 136: 22-31.

123. Ji, D, and Dani, J. “Inhibition and disinhibition of pyramidal neurons by activation of nicotinic receptors on hippocampal interneurons.” J. Neurophysiol. 2000. 83: 2682-2690.