Treating the Lesions, Not the Disease

In this issue of The American Journal of Pathology, Caccamo and colleagues1 demonstrate that treatment of the triple transgenic mouse model of Alzheimer’s disease (AD) with lithium, a weak inhibitor of glycogen synthase kinase (GSK), attenuates tau pathology but not amyloid-β nor working memory. This work follows closely on the heels of another therapeutic strategy in these mice, where reductions of soluble amyloid-β and tau, but not soluble amyloid-β alone, attenuated cognitive decline.2 Taken together, these articles likely provide important insights into potential treatment strategies for AD.

The most conspicuous microscopic changes in the brains of individuals with AD are senile plaques composed of amyloid-β and neurofibrillary tangles composed of tau protein. Although arguments have been made that the diagnostic requirement3 of clinical dementia coupled with neuropathological senile plaques and neurofibrillary tangles is a somewhat artificial construct that forces a correlation between disease and pathology,4 the majority of investigators remain convinced that either amyloid-β or tau causes AD. In this regard, for the most part, amyloid-β has been clearly ahead in terms of the popular vote.5 Amyloid-β is toxic in vitro,6 and mutations in the amyloid-β protein precursor7,8 are associated with familial forms of the disease. These findings coalesced into the Amyloid Hypothesis9 that, to this day, remains the leading hypothesis for disease pathogenesis.10 However, the development of transgenic mice with supraphysiological loads of amyloid-β, but little evidence of neurodegeneration,11,12,13 led many to question whether amyloid-β was sufficient to cause AD.14 This lull in the amyloid-β field was coincident with a major tau-related finding—namely, that mutations in tau were associated with familial forms of neurodegeneration,15 albeit FTDP-17 and not AD. Currently, therefore, which lesion to therapeutically target, amyloid-β or tau, has been keenly debated with the fragmented view of the disease leading, naturally, to a fragmented approach to treating the disease.

At present, therapeutic approaches are mainly divided along “party” lines,5 either looking at reducing amyloid-β or blocking tau phosphorylation. Regarding the latter, upstream kinases that phosphorylate tau have received the most attention, in particular GSK. GSK-3β can phosphorylate tau both in vivo and in vitro and is found at increased levels in the AD brain, as well as in normal neurons and those containing neurofibrillary tangles.16,17 The appearance of GSK-3β is associated with amyloid-β, presenilins, and the acetylcholine pathway, making it a seemingly attractive target with which to attack multiple pathways involved in the disease. However, it is worth noting, especially in a chronic disease such as AD, where cause and consequence are omnipresent, that GSK-3β has been attributed to other factors involved in the disease, such as oxidative stress18,19 and oxidative stress-induced neuronal cell pathology.20

Transgenic models overexpressing amyloid-β but lacking neuronal degeneration11,12,13 were followed by the development of transgenic models with abnormal tau phosphorylation in the brain.21 The logical and consequent step was the triple transgenic model,22 which develops both amyloid-β plaques and tau aggregation, closely mimicking the human condition. This mouse therefore offered the promise to be useful not only for testing potential therapeutic targets but also in settling the argument about whether amyloid-β or tau is the primary pathogenic protein.

Lithium is an effective, albeit weak, inhibitor of GSK and has been used therapeutically for a variety of disorders.23 Relevant to AD, lithium acts, presumably via GSK inhibition, to modulate tau phosphorylation and has previously been shown to prevent hippocampal tau pathology in a transgenic mouse model overexpressing GSK-3β and in the FTDP-17 mutant tau model when administered early in the disease.24 Of relevance to the current work by Caccamo,1 lithium administration at later stages, although reducing tau phosphorylation, was not able to reverse tau aggregation.23 Likewise, using a molecular approach, transgene shutdown in Tet/GSK-3 mice results in both normalized GSK activity and reduced levels of phosphorylated tau, which consequently prevents neuronal death and cognitive decline.25 These findings, along with others,24 recently led the Alzheimer Cooperative Disease Study Center to enter lithium into clinical trails for the treatment of AD. Although the results from using lithium in triple transgenic mice1 might forecast pessimism for the outcome in human patients, there are a number of important aspects that provide food for thought, and it is therefore premature to write off lithium in AD patients based on lack of efficacy in a mouse model.

First, although close to 50 different therapeutic strategies have proven effective in treating single amyloid precursor protein transgenic mice, none have proven effective in human clinical trials thus far. Although this does not mean that none will prove effective, it does indicate that these transgenic mouse models are not predictive of which treatments will work. Whether the triple transgenic animal is a more reliable barometer for therapeutics remains to be seen.

Second, in triple transgenic mice, it seems that both amyloid-β and tau are important, and individually targeting one, but not the other, is insufficient to attenuate cognitive decline.1,2 On the other hand, targeting both pathologies2 is effective. However, is it valid to translate such findings to the human condition? Simply stated, the fact that forced overexpression of tau and amyloid-β in vulnerable regions of the brain leads to cognitive dysfunction and that reversing these leads to improved cognitive function is as artificial a system as one can think of—ie, preventing the insult prevents the phenotype. Although this works wonderfully in animal models, where we know the (transgenic) insult, this lesion-centric approach will only translate effectively into human AD if the lesions are the cause of the disease. However, if the lesions are a consequence of the disease, we should not expect similarly spectacular results.26,27

Third, in conjunction with the findings of Engel et al,25 Caccamo’s work1 presented in this issue reiterates the fact that early intervention is key. The current work uses lithium administered to the triple transgenic mice aged at 15 months, after pathological structures have been well developed. Although tau phosphorylation is reduced, cognition deficits were not attenuated.

In summary, the triple transgenic model may prove to be a very useful tool in understanding the development of AD. Biochemically, the abnormal proteins mimic those found in the human brain. Morphologically, the amyloid-β plaques develop similarly to AD, and the phosphorylated tau accumulates in the brain regions and cell types as in the human counterpart.22 Cognitive deficits can be reliably measured. Moreover, like their human counterparts, prevention may be far easier to attain than treatment. However, knowing whether AD is caused by amyloid-β and/or tau is critical in translating any findings into humans. If amyloid-β and/or tau are downstream markers of disease, we learn naught.