Alzheimer’s disease (AD) is the most common form of dementia and places a huge burden on individual sufferers, their families, society, and economies. There is currently no cure, and therapies to ameliorate symptoms or slow progression are of limited efficacy. The discovery of alternative therapies is, therefore, a health and social priority. However, the pathological mechanisms underlying AD remain poorly understood. Currently, the most commonly cited pathological paradigm for AD is the deposition of beta-amyloid, and a key target for AD research in recent years has been the role of secretase pathways in the cleavage of amyloid precursor protein (APP) to form beta-amyloid. Beta-amyloid derives from the concerted action of β-secretase and γ-secretase on APP; whilst in contrast, α-secretase-mediated cleavage of APP forms the soluble neuroprotective fragment sAPPα, and therefore precludes formation and deposition of beta amyloid. Low levels of sAPPα are found in cerebrospinal fluid (CSF) of patients with AD, suggesting a relative reduction in α-secretase activity. As a result, recent research has focused on blocking the "harmful" beta and gamma pathways or, alternatively, potentiating the "beneficial" alpha pathways.

This article reviews three recent papers that seek to elucidate the possible mechanisms underlying AD and explore potential targets for intervention. The first paper examines the effect of an inhibitor of the kinase PDK1 on prion proteins and beta-amyloid in prion disease and AD, respectively. The second paper demonstrates that disruption of the unfolded protein response can arrest disease progression in prion-infected mice. Finally, we review the results of a phase 3 trial of the gamma secretase inhibitor semagacestat which, although a promising target for intervention, proves disappointing.

PDK1 decreases TACE-mediated α-secretase activity and promotes disease progression in prion and Alzheimer’s diseases

This paper examines the role of a protease belonging to the ADAM (a disintegrin and metalloprotease) family, called tumor necrosis factor-α (TNF-α) converting enzyme (TACE, or ADAM17), in stimulating α-secretase–mediated cleavage of both cellular prion protein (PrPC) and APP. The kinase PDK1 regulates TACE availability at the cellular membrane and so influences the processing of beta amyloid and PrPC.

Central nervous system–derived cells from mice infected with mouse-adapted prions were examined, and using TNFR1 shedding as a marker of TACE activity, researchers noted that prion infection triggered an approximately three-fold to four-fold increase in the amount of TNFR1 at the cell surface. This rendered prion-infected cells highly vulnerable to soluble TNF-α (sTNF-α)–associated toxicity,with cell death and neuronal dysfunction being induced at significantly smaller doses of sTNF-α than those required for similar effects in non-prion infected cells. The authors then analysed TACE localisation; in uninfected cells, TACE is normally present at the cell surface, but in prion-infected cells it was found intracellularly, co-distributed with caveolin-1 (Cav-1), and at almost undetectable levels at the plasma membrane. Therefore, prion infection promotes internalisation of TACE through mechanisms that probably involve Cav-1–enriched microvesicles. The authors identify that the kinase PDK1 regulates TACE availability at the cellular membrane, and that the pharmacological blockade of this kinase abolishes TACE hyperphosphorylation and triggers the relocation of biologically active TACE at the membrane. Finally, in prion-infected mice, intraperitoneal injection of the PKD1 inhibitor BX912 significantly delayed mortality (193.8 ± 2.1 versus 166.0 ± 1.8 day, n = 10, P < 0.0001), reduced impairments in motor function, attenuated prion-induced neuronal loss in the cerebellum, and reduced PrPSc deposition.

Recent data suggesting that PrPC is involved in beta-amyloid neurotoxicity in AD prompted the authors to examine whether the accumulation of beta-amyloid would increase PDK1 activity. They found that PDK1 activity was increased by approximately 150 % in the hippocampal neurons of transgenic mice with beta-amyloid deposition (evidenced by PET imaging) compared with controls. siRNA-mediated silencing of PrPC in hippocampal neurons reversed the increase in PDK1 activity, suggesting that beta-amyloid-induced increases in PDK1 activity depend on PrPC. They went on to demonstrate that pharmacological inhibition and siRNA-mediated silencing of PDK1 had positive effects on markers of AD. Injection of BX912 into 200-day-old transgenic mice reduced the number of mice with amyloid plaques at 275 days; interestingly, the effects on amyloid pathology persisted until day 330, after which the mice were killed. It also decreased CSF markers of AD pathology (sAPPβ, Aβ40 and Aβ42) and, perhaps most importantly, showed significant effects on memory and cognitive impairments.

Comments and discussion. In summary, this paper shows that in both PrPSc-infected and AD neurons, activation of PDK1 reduces cell surface TACE-mediated α-secretase activity by triggering internalization of TACE and that PDK1 inhibition attenuates prion and AD progression. Although this paper reveals promising findings and suggests that PDK1 may be a useful target for interventions in both AD and prion disease, a key problem is the PDK1 inhibitor (BX912) toxicity. Furthermore, it is unclear whether inhibition of PDK1 could have other deleterious effects given its role in regulation of cellular processes such as cell survival, differentiation, growth, and protein expression. Nevertheless, this work does raise the possibility of an alternative potential therapeutic target in AD, and although the use of PDK1 inhibitors may still be some way off, this paper provides important clues in understanding the mechanisms underlying these two important and devastating neurodegenerative diseases.

Pietri M et al. Nat Med. 2013 Sep;19(9):1124-31

Oral treatment targeting the unfolded protein response prevents neurodegeneration and clinical disease in prion-infected mice

All cells require correctly folded proteins to function efficiently. Several neurodegenerative diseases, including Alzheimer’s disease and prion diseases, are associated with the accumulation and aggregation of misfolded proteins in the brain. These unfolded or misfolded proteins cause cellular stress that is detected by sensors in the lumen of the endoplasmic reticulum. In turn, this initiates a signalling cascade known as the unfolded protein response (UPR). The authors of this study demonstrate that oral treatment with a specific inhibitor of the kinase PERK (protein kinase RNA–like endoplasmic reticulum kinase), a key mediator of the UPR pathway, prevents UPR-mediated suppression of translation and halts development of clinical prion disease in mice.

Increased PERK-P and eIF2a-P are found in the brains of patients with AD, PD, ALS, and prion disease. Prior work by the authors has demonstrated that in prion disease–infected mice, misfolded PrP caused sustained overactivation of the PERK/eIF2a branch of the UPR, causing a decline in levels of key proteins. They also noted that genetic manipulation of the pathway upstream and downstream of eIF2a-P in prion-diseased mice reduced eIF2a-P and restored vital translation rates, allowing recovery of synaptic protein levels, resulting in marked localised neuroprotection and increased survival of mice with prion disease. In this study, the authors aimed to replicate those results using a selective PERK inhibitor (GSK2606414). A dose of 50 mg/kg BD was used, and good brain penetration was verified using liquid chromatography and mass spectrometry. The drug was given to two groups of mice intracerebrally inoculated with prions at 4 weeks; the first group was treated from 7 weeks after inoculation, where synaptic loss was already occurring but before memory and behavioural deficits occurred (n = 20), and the second was treated at 9 weeks, when spongiform prion pathology and behavioural change were established (n = 9). There was also a control group infected with prions but given the vehicle alone, and a group uninfected with prion disease but given GSK2606414.

By 12 weeks after inoculation, all GSK2606414-treated animals in both groups (from 7 to 9 weeks after inoculation) were free of diagnostic signs of prion disease, whereas all controls were terminally sick. The results also showed a reversal of decline in burrowing behaviour (a characteristic of early prion disease in mice) treated at 9 weeks. Histopathologically, the authors noted neuroprotection consistent with lack of clinical symptoms. The neuronal ribbon of hippocampal regions CA1–4 was protected and CA1 pyramidal neuron counts from treated mice at early and later time points were equivalent to non-prion-infected mouse brains, whereas they had declined to <30 % of control values in untreated mice. Astrocyte number and activation were also reduced in treated animals, and there was minimal spongiform degeneration.

Comments and discussion. This paper supports PERK inhibition as a target for drug discovery in the treatment of prion disorders. One of the main findings is that this appears to occur downstream and independently of the central pathogenic process of prion replication. Two of the key problems with many neurodegenerative diseases are that histopathological changes can occur many years prior to clinical manifestations and that there are no definitive screening or diagnostic tests. To identify a potential therapeutic target downstream of these changes that may arrest or even reverse clinical changes could prove important.

Once again, a substantial issue was of drug side effects, most notably impairment of pancreatic function, with hyperglycaemia and weight loss in the mice. However, as the authors note, the mice were otherwise well and the level of hyperglycaemia did not fall within the murine diabetic range. It would also have been interesting to see whether the mice who exhibited early indications of prion disease but did not meet the diagnostic definition would have gone on to develop the disease at a later stage; however, the authors were unable to perform survival studies due to regulations on animal welfare regarding body mass. Whether the results of this study can be extrapolated to other neurodegenerative diseases remains to be seen, but this paper certainly provides an interesting starting point for exploration of the wider role of the UPR.

Moreno JA et al, Sci Transl Med. 2013 Oct 9;5(206):206ra138.

Semagacestat for treatment of Alzheimer’s disease

This paper describes a Phase 3, double-blind, placebo-controlled trial of the gamma secretase inhibitor semagacestat. Semagacestat functionally inhibits γ-secretase, rather than competitively inhibiting the enzyme active sites, and in vitro trials had shown a 25 % reduction in beta-amyloid synthesis in humans and reduced beta-amyloid in the brain, CSF, and plasma of mice treated with the drug.

In this Phase 3 trial, 1,537 patients with "probable" Alzheimer’s disease underwent randomisation to receive 100 mg of semagacestat, 140 mg of semagacestat, or placebo daily. There were no significant differences between the three groups in terms of baseline characteristics. Participants were aged 55 or over, and probable AD was defined using the National Institute of Neurological and Communicative Diseases and Stroke–Alzheimer’s Disease and Related Disorders Association (NINCDS–ADRDA) criteria. Patients with depression, defined by a geriatric depression score of six or less were excluded. Patients taking memantine or cholinesterase inhibitors were included.

Cognitive testing occurred at baseline and throughout the 76-week trial period using a number of measures; the primary outcome measures were the cognitive subscale of the Alzheimer’s Disease Assessment Scale for cognition (ADAS-cog) and the Alzheimer’s Disease Cooperative Study-Activities of Daily Living (ADCS-ADL) scale. Biological markers were also assessed including apolipoprotein E genotyping, lymphocyte phenotyping, and plasma levels of Aβ. In a subset of patients, CSF Aβ and tau volumetric MRI and PET with 18F-florbetapir for visualization of Aβ were also performed. A mixed-model repeated-measures analysis was used to compare model-adjusted least squares means at 76 weeks.

The trial was terminated before completion due to a high occurrence of adverse events. The ADAS-cog scores worsened in all three groups, with mean changes of 6.4 points in the placebo group, 7.5 points in the group receiving 100 mg of the study drug, and 7.8 points in the group receiving 140 mg (P = 0.15 and P = 0.07, respectively, for the comparison with placebo). The ADCSADL scores also worsened in all groups, with mean changes at week 76 of −9.0 points in the placebo group, −10.5 points in the 100-mg group, and −12.6 points in the 140-mg group (P = 0.14 and P < 0.001, respectively, for the comparison with placebo). Laboratory abnormalities included reduced levels of lymphocytes, T cells, immunoglobulins, albumin, total protein, and uric acid but elevated levels of eosinophils, monocytes, and cholesterol; urine pH was also elevated. The treatment groups also had more skin cancers and infections, discontinuations due to adverse events, and serious adverse events (P < 0.001 for all comparisons with placebo).

In terms of biological markers, plasma levels of Aβ40 and Aβ42 were significantly reduced in both treatment groups, but CSF Aβ40, Aβ42, and tau were not significantly changed. Phospho-tau in the CSF, however, was significantly reduced in the treatment groups. Volumetric MRI of the entorhinal cortex and hippocampus showed inconsistent results. There were no significant differences on PET scanning.

Comments and discussion. The results of this trial are disappointing. It is difficult to disentangle whether the failure of this trial lies with the formulation of semagacestat itself, the notion of gamma-secretase inhibition as a therapeutic target, or even the very assertion that the primary driver of pathology in AD is the deposition of beta-amyloid. It seems possible that if deposition of beta-amyloid is the primary pathology, then a trial of a drug whose primary target is APP processing is of limited use in those with mild-to-moderate AD in whom beta-amyloid deposits are already well established.

The authors suggest that the lower levels of beta-amyloid in plasma but not in CSF suggest the target (APP processing) was engaged but peripherally, rather than centrally, and that the lack of changes on volumetric MRI and PET suggest that semagacestat did not have a directly detrimental effect on the brain. They also note that the reason for the high level of adverse events is the inhibition of Notch, and that semagacestat may be more selective for Notch than for APP processing by γ-secretase; in animal models, reductions in γ-secretase activity have been shown to increase skin cancers and affect lymphopoiesis. Other γ-secretase inhibitors are currently under development, and perhaps a more selective version will prove to be effective in the future.

Doody RS et al. N Engl J Med. 2013 Oct 24;369(17):1661.