6 December 2010. Shoring up insulin responses is a tried-and-true strategy for helping people manage type 2 diabetes (T2D), but could it also prevent or delay Alzheimer’s disease? Strong epidemiological evidence suggests diabetes is a risk factor for AD, but capitalizing on that connection has not been plain sailing. While there is some early hint that intranasal insulin might offer some protection against dementia, sensitizing agents such as pio- and rosiglitazone failed to pass muster in larger clinical trials for dementia (ARF related news story). Could it be time for a new tack? At the 40th annual meeting of the Society for Neuroscience (SfN), held 13-17 November in San Diego, the promise of glucagon-like peptide agonists seemed to breathe new life into insulin signaling as a therapeutic target.
Because of new human and animal model data, it appears the field might see several clinical trials testing this drug class in the coming year, according to researchers. “These agonists have remarkable properties,” said Konrad Talbot, University of Pennsylvania, Philadelphia. “They do not affect cells that have normal insulin signaling, but they can boost synthesis of the insulin receptor, insulin receptor substrate-1 (IRS-1), and glucose transporter 4, and they enhance the response of cells to insulin stimulation. Boosting insulin responsiveness would help correct a major neuronal insulin signaling deficit we have found in the hippocampus of AD cases,” he told ARF. Whether that approach will be successful remains to be seen, but in an SfN nanosymposium organized by Talbot, speakers showed that glucagon-like protein 1 (GLP-1) agonists, which are approved for treating diabetes, protect against Aβ toxicity and rescue learning and memory deficits in model mice.
Just what are those insulin signaling deficits that crop up in AD? Insulin binds to cell surface receptors, kicking off a signaling cascade that relies on a number of kinases, commonly including phosphoinositol-3-kinase (PI3K) and Akt (see ARF related news story). At the SfN meeting, Talbot and Hoau-Yan Wang, City University of New York Medical School, reported on their collaborative work to examine the activation status of this pathway in the brain at baseline, and following insulin stimulation. The researchers used immunohistochemistry to look for phosphorylated epitopes in early postmortem tissue taken from normal controls, people with MCI, and people with AD. Tissue was obtained from the University of Pennsylvania and from the Religious Orders Study cohort at Rush University, Chicago, Illinois.
Talbot reported that in 24 age-matched samples from UPenn, AD cases displayed significant decreases in basal activation of the insulin receptor and increases in basal activation of downstream signaling molecules (Akt, mTOR, GSK-3β, and PKC λ/ζ). The most prominent AD feature was serine phosphorylation of insulin receptor substrate 1 (IRS-1), which helps propagate signals from the insulin receptor to downstream kinases. The phosphorylation, of serines 312, 636, and especially 616, was mostly extranuclear, which is unusual. Serine phosphorylation of these sites is inhibitory, Talbot explained, and in keeping with this, he saw reduced activation of IRS-1. The levels of S616 phosphorylation were also markedly elevated in MCI patients but not in tissue from normal controls or people with other forms of dementia. “This might be particularly interesting because it could be some sort of early marker of dysfunctional insulin signaling,” said Talbot. He reported similar findings using tissue samples taken from 30 normal, 29 MCI, and 31 AD cases from Rush.
Furthermore, the density of neurons containing extranuclear, serine-phosphorylated IRS-1 correlated with elevated oligomeric plaque load and deficits in global cognition, working memory, and especially episodic memory, that last having a tight correlation R value of 0.65, “which is phenomenal for postmortem tissue,” said Talbot, given the variability that can exist among samples. After the researchers adjusted for a variety of cofactors, including age, sex, years of education, density of neurofibrillary tangles, and oligomeric plaque load, the number of cells with inactivated IRS-1 still correlated with cognition, which suggests that dysfunctional insulin signaling is related to whatever is causing memory problems, said Talbot. In contrast, correlations between oligomeric plaque load and cognition disappeared when corrected for serine-phosphorylated IRS-1. “These results are consistent with evidence that oligomeric Aβ induces serine phosphorylation of IRS-1, and that this in turn disrupts insulin signaling at synapses, thereby resulting in cognitive deficits,” said Talbot.
Talbot collaborated with Wang, an expert in ex-vivo analysis, to study the ability of brain tissue to respond to insulin. They took early postmortem hippocampal tissue samples from six age-matched pairs—normal and AD—incubated them with low concentrations (1 and 10 nM) of insulin (to stimulate insulin but not insulin-like growth factor receptors), then looked for downstream effects. This ex-vivo analysis revealed a blunted insulin response in AD tissue. Tyrosine phosphorylation (activation) of IRS-1 was impaired. The AD tissue samples mounted a lackluster activation of a variety of downstream kinases, as judged by phospho-epitope analysis. Phosphorylation of Akt (serine 473), mTOR (serine 2448), and ERK2 (tyrosine 204) were all significantly lower than in normal tissue treated with insulin. Curiously though, baseline levels of these phospho-epitopes were higher in AD tissue, which could reflect an attempt to compensate for impaired insulin signaling. “This is the first direct evidence of insulin resistance in the brain of AD cases and suggests that basal levels of activated downstream insulin signaling molecules reflect inadequate compensatory responses to such upstream resistance,” said Talbot.
What causes dysfunctional insulin signaling in AD, and could correcting it help? A hint to the former came from Fernanda De Felice from the Federal University of Rio de Janeiro, Brazil. When working with William Klein at Northwestern University, De Felice found that Aβ oligomers reduce cell surface insulin receptors on neurons, and vice versa (see ARF related news story). Now, she has taken that one step further, looking at the downstream effects of Aβ on insulin signaling in hippocampal neuron cultures. At SfN, she reported that synthetic Aβ oligomers increase IRS-1 serine phosphorylation and decrease IRS-1 tyrosine phosphorylation, disabling the insulin signaling pathway. Aβ oligomers boosted phosphorylation of IRS-1 serines 636 and 639. These post-translational modifications are also found in muscle and adipose tissue in diabetes patients, said De Felice. In the periphery, IRS-1 inactivation is mediated by JNK and tumor necrosis factor α (TNFα); and De Felice presented evidence that those pathways might be involved in blunting insulin signaling in hippocampal neurons as well. Introducing a dominant-negative JNK, or an inhibitory TNFα antibody, prevented IRS-1 serine phosphorylation in neurons incubated with Aβ oligomers.
De Felice further reported that both insulin and exendin-4, an agonist of glucagon-like peptide (GLP-1), block the Aβ oligomer-induced increase in IRS-1 serine phosphorylation in cultured neurons. She also found that exendin-4 had the same effect when given to 13-month-old APP/PS mice, suggesting that the GLP-1 agonist might be a potential AD therapeutic.
Christian Holscher, University of Ulster, Coleraine, U.K., followed on that same theme in his talk. Holscher briefly reviewed some of the properties of GLP-1 and its mimics. These molecules increase insulin sensitivity by facilitating insulin release and re-sensitizing insulin receptors. They act in the brain, where neurons, particularly pyramidal neurons of the cortex and hippocampus, express GLP-1 and its receptors. The peptide has neurotrophic properties, boosting synaptic transmission and neuronal progenitors.
There are hints that GLP-1 analogues might protect against AD. Earlier this year, Holscher reported that a protease-resistant form (ValineGLP-1) not only crosses the blood-brain barrier and increases long-term potentiation (LTP) in the hippocampus, but also reduces the number of dense core plaques in APP/PS transgenic mice (see Gengler et al., 2010). At the SfN meeting, Holscher reported that the GLP-1 analogue liraglutide, which is FDA approved for treatment of diabetes, has similar effects. An eight-week treatment restored novel object memory as well as spatial memory in nine-month-old APP/PS1 mice. Liraglutide also dramatically boosted LTP in anesthetized animals, said Holscher. Paired pulse facilitation, another correlate of learning, also improved. This is an indicator of altered GABAergic signaling, said Holscher, which would dovetail with expression of the GLP-1 receptor in pyramidal cells, as these are GABAergic. Interestingly, the agonist had little or no effect on LTP or PPF in wild-type controls, suggesting that the drug only works when normal LTP/PPF have gone awry.
Perhaps the most surprising effect of these agonists is on amyloid plaques. After the eight-week liraglutide treatment, plaque density was half that in untreated animals and the number of dense core plaques fell by two-thirds. Levels of soluble Aβ in the brain were down by about a third.
Previous results from Nigel Greig and colleagues at the National Institutes of Health, Baltimore, Maryland, suggest that GLP-1 has a similar effect (see Perry et al., 2003) and that exendin-4 reduced plaque load in 3xTG mice treated with streptozotocin to induce diabetes (see Li et al., 2010). During question time, Holscher said he did not know how these agonists influence Aβ load, but told ARF that exendin-4 reduces APP synthesis in cultured neurons (also see Perry et al., 2003). Whether this also happens in vivo needs to be tested, he said.
GLP-1 agonists are approved for diabetes, and trials testing exendin-4 in Parkinson’s disease patients are underway in London. Could trials in AD soon follow? “I find the prospect incredibly exciting,” said Talbot. “I have a vested interest,” he admitted, “but the effect on plaques that Christian Holscher reported was stunning, and it seems to me that [these drugs] are ready for clinical testing.”
Talbot told ARF that the NIH is planning an exendin-4 trial for AD. Across the pond, Holscher has been in discussions with colleagues at Hammersmith Hospital, London, to put forward a plan for a clinical trial of liraglutide. Novo Nordisk, which makes liraglutide under the trade name Victoza®, seems interested in the proposal, said Holscher, but whether it will sponsor such a trial is not clear at present. “With the agreement of my colleague Steven Arnold, who heads geriatric psychiatry and is associate director of the Alzheimer’s Disease Center here at the University of Pennsylvania, I am working with Christian to encourage Novo Nordisk to support Victoza trials in the U.S. and Europe,” Talbot told ARF. He noted that since exendin-4 and liraglutide have different pharmacological properties, trials of both are warranted.—Tom Fagan.