γ-secretase is an essential enzyme for generation of amyloid β protein (Aβ), a central component of neuritic plaques in Alzheimer's Disease (AD) brain. Identification of potent γ-secretase inhibitors (GSIs) and γ-secretase modulators (GSMs) that specifically inhibit production of Aβ42, an infamous molecule thought to be responsible for AD pathogenesis, has always been major interest to AD drug development. While a few of inhibitors and modulators have reached clinical trials, most screenings have only focused on in vitro potency test and specificity test. In this issue Mitani and colleagues reported the effect of GSM-2, a second generation GSM on Aβ42 and cognitive function in vivo using an AD model mice in an article titled ‘Amelioration of cognitive deficits in plaque-bearing Alzheimer's disease model mice through selective reduction of nascent soluble Aβ42 without affecting other Aβ pools’ (Mitani et al. 2013). They discovered that GSM-2 lowered nascent Aβ42 level in vivo and improved memory deficits in both young and old age groups, indicating that GSM-2 may delay AD progression regardless of existing Aβ plaque load, and this is likely due to reduction of newly synthesized soluble Aβ42 (Fig. 1).
AD is the most common form of dementia in elderly population. The pathological hallmarks of AD are extracellular neuritic plaques, intracellular neurofibrillary tangles and significant neuronal loss in specific brain regions. The prevailing amyloid hypothesis postulates that among other factors, aberrant accumulation and deposition of Aβ is the causative event in AD pathogenesis. Aβ is derived from sequential cleavage of β-amyloid precursor protein (APP) by β-secretase (BACE1) and γ-secretase complex. After APP is cleaved by BACE1 at Asp1 site of the Aβ domain to generate the C99 fragment, γ-secretase complex cleaves C99 at different γ-sites to yield Aβ species of varying length. Evidence supporting amyloid hypothesis includes discoveries of familial Alzheimer's Disease, in which a single mutation alone in one of the three genes, APP, presenilin 1 and presenilin 2, leads to a much earlier onset of AD. Increased Aβ production is the common phenotype of the mutations in familial Alzheimer's Disease. Intriguingly, hundreds of mutations in presenilin 1 and presenilin 2 either increase the absolute level of Aβ42 in the brain or increase the relative ratio of Aβ42 over Aβ40, indicating an important role of Aβ42, the most hydrophobic and aggregation-prone species of Aβ, in AD pathogenesis.
In the past decade the amyloid hypothesis has gradually evolved into Aβ oligomer hypothesis, which blames soluble Aβ oligomers instead of Aβ plaques for AD pathogenesis. Identification of soluble Aβ oligomers as synaptotoxic species and better correlation between cognitive impairment in AD and Aβ oligomer level than Aβ plaque level support this hypothesis (Haass and Selkoe 2007; Walsh and Selkoe 2007). Regardless of whether soluble Aβ or Aβ plaque is the culprit, reduction of Aβ production, especially that of Aβ42, could be an appropriate target for clinical intervention of AD development.
In the battle with AD, researchers have been expending extensive efforts in trying to identify GSIs, hoping to intervene in the last step of Aβ production while circumventing possible side effects from disturbance of early events in the cascade (Kreft et al. 2009). First generations of GSIs have shown potent inhibition of total Aβ production. However, their non-selectivity to Notch cleavage resulted in inhibition of Notch intracellular domain (NICD) production and blockage of the essential Notch signaling pathway, leading to undesired effects such as gastrointestinal toxicity (Searfoss et al. 2003). Later efforts were put to identify second generation ‘Notch-sparing’ GSIs. Nevertheless, side effects may still arise from inhibition of other substrates of γ-secretase, from increased βCTF level associated with decreased total Aβ level, or a late rebound effect.
The concept of GSMs was first introduced when Weggen et al. (2001) demonstrated that a set of NSAIDs, whose users are associated with reduced prevalence of AD, lower Aβ42 levels without affecting Aβ40 levels. As then hundreds of GSMs have been optimized, and a few of them have entered clinical trials (Oehlrich et al. 2011). Generally, GSMs reduce Aβ42 production but do not change Aβtotal levels or affect cleavage of other substrates of γ-secretase, thus sparing possible side effects seen from GSIs.
Previous studies on GSMs have been focusing on their in vitro potency to reduce Aβ42 level and whether other substrates of γ-secretase complex are spared. Confirmation of their Aβ-lowering effect in vivo and demonstration of their rescue effect on cognitive function are essential before they can enter clinical trials. In this study, Mitani et al. (2013) explored the effects of GSM-2, a second generation GSM that decreased Aβ42 level while maintaining Aβ40 and βCTF levels, on cognitive function in an AD mouse model Tg2576. Consistent with results from their previous study (Mitani et al. 2012), after 8-day oral administration, GSM-2 rescued alternation rate in Y-maze task in mice of all age groups (10-, 14-, and 18-month old). The finding suggested that the effect of GSM-2 was not related to existing Aβ plaque burden, and it was also effective for AD at late stages.
The next question naturally raised is What Aβ species are affected by GSM-2 in vivo that is correlated with cognitive improvement? Mitani et al. measured brain soluble Aβ42 level, which is affected by GSM-2 in vitro and to be condemned according to Aβ oligomer hypothesis. Unexpectedly, only brain soluble Aβ42 in 10-month-old mice showed a significant reduction and a correlation with cognitive function. But bear in mind that brain soluble Aβ42 level after 8-day treatment is not a direct measurement of Aβ42 production during treatment. It is also related to Aβ42 level prior to treatment and clearance rate during treatment. Using an isotope-labeling technique (Bateman et al. 2006), Mitani et al. was able to quantify the amount of nascent Aβ42 produced after administration of GSM-2 in 10- and 18-month-old mice. Indeed, the de novo soluble Aβ42 levels were reduced significantly in both cases.
It has been a major topic of debate in the field of AD what Aβ species and in what multimers does Aβ exert its neurotoxic effect. This study brought up another question—if soluble Aβ42 is most responsible for cognitive impairment in AD, which pool is the true ‘bad guy’ from, the newly synthesized Aβ42 or pre-existing ones? The discrepancy between the cognitive improvement and the stable total soluble Aβ42 level in the hippocampus of 18-month-old mice may put nascent Aβ42 under suspicion. Yet it is too early to draw a conclusion. For one thing, reduction of total soluble Aβ42 level may take more than 8 days to reach statistical significance due to the heavy load of pre-existing Aβ42 in elder mice. For the other, while there is no change in terms of hippocampal Aβ42 level, local Aβ reduction in the synapses critical to memory function could be drastic, which may not be detected with total homogenate samples. It could be exciting news for AD patients in their late stages, if nascent Aβ42 is truly the culprit in AD pathogenesis. As is indicated from this study, GSM-2 may be developed into a therapeutic drug suitable for AD from early to late stage.