28 October 2006. On Friday, October 13, the day before the kick-off of the Society for Neuroscience meeting in Atlanta, Lary Walker of Emory University, Atlanta, Georgia, and Richard Kostrzewa of East Tennessee State University in Johnson City, hosted a satellite symposium at the Yerkes National Primate Research Center on the Emory campus. Sponsored by the Neurotoxicity Society, the symposium brought together a distinguished slate of speakers who considered the central, yet unsolved question of Alzheimer disease, and neurodegeneration in general—how exactly do neurons die? Two talks highlighted novel players in AD. One, by Gilles Guillemin from the University of New South Wales, Sydney, Australia, focused on the production of the excitotoxic tryptophan metabolite quinolinic acid by activated microglia and macrophages. The other came from Efrat Levy, Nathan Kline Institute in Orangeburg, New York, on the neuroprotective actions of the protease inhibitor cystatin C. Bill Klein, Northwestern University, Chicago, talked about the hottest topic in Aβ toxicity—oligomers. In addition, Todd Golde, Mayo Clinic, Jacksonville, Florida, presented his work on the use of adeno-associated viral vectors. He has used these to generate quick in-vivo animal models of neuron death in order to investigate the proteins and pathways that precipitate neurodegeneration.
There are over a hundred naturally occurring AAV variants showing a range of tissue tropism suitable for gene delivery, but most of them have not been studied in the CNS, Golde said. Work done largely by Ronald Klein at Louisiana State University Health Sciences Center in Shreveport showed that injecting AAV8- or AAV1-encoding green fluorescent protein into the rat hippocampus could drive long-term, high-level expression of GFP mostly in neurons (Klein et al., 2005). Comparison of different AAV serotypes showed that some gave better dispersal through the brain tissue, while others drove higher expression. When Golde and colleagues looked in the mouse, the serotypes behaved differently than in rats, suggesting that a trial-and-error approach to evaluate multiple serotypes is necessary to optimize gene delivery in different species.
Though AAV-mediated delivery of genes can save time and money compared to traditional transgenic construction, the resulting animals are far more than a poor person’s transgenics. Viral delivery makes it possible to look quickly at the effects of nearly any protein solely in the brain and in different species, including mice, rats, or even guinea pigs. Golde showed that AAV can be used to make a rat model of β amyloid expression by introducing the Aβ1-40 or Aβ1-42 peptide fused to the Bri protein, which has a furin cleavage site (see ARF related news story). In collaboration with Trish Lawlor and Matt During at the University of Auckland, New Zealand, and Ross Bland at Neurologix Inc., New York, Golde found that the Aβ42 fusion causes plaque formation and behavioral impairments in animals. The approach can even be used in guinea pigs, an attractive model for several reasons: guinea pig Aβ sequence is identical to the human peptide; the animals handle cholesterol and lipoproteins more like humans; and they have a more active inflammatory response compared to mice or rats.
In mice, Golde described the production of “somatic brain transgenics,” where recombinant AAV virions were injected freehand into the ventricles of newborn cryo anesthetized mouse pups. This procedure was originally described by John Wolfe and colleagues (Passini and Wolfe, 2001) but has not been widely applied. The procedure takes 15 minutes per litter, and with AAV1, yields lifelong transduction of the cortex and hippocampus. With the AAV9 serotype, they saw almost all neurons in the brain transfected using a GFP reporter.
Golde reported that in one study, using the vectors to introduce single-chain antibodies to Aβ40, Aβ42, or a pan-specific antibody, he found all the proteins attenuated plaque deposition in CRND8 mice as effectively as the passive immunization they previously reported with antibodies of the same specificity (see ARF related news story). But the AAV study was cheaper and faster, being completed in 5 months at a fraction of the cost of a breeding study. The second study, which was largely carried out by Andrew Nyborg in Golde’s lab, involved expressing a mutant non-cleavable form of pro-NGF. Previously, other researchers had shown that pro-NGF induces neuron death via the SorLa family receptor sortilin and the p75NGF receptor (Nykjaer et al., 2004). Introduction of a non-cleavable pro-NGF caused build-up of the pro-peptide over the first weeks of life, and early death of the mice, which was not seen with either sortilin overexpression or introduction of wild-type pro-NGF. The brain tissue had increased astrocytosis, microgliosis, tau phosphorylation, and apoptotic cells. The entire study took 3 months, Golde said. Compared to transgenics, the method is cheap, fast, reproducible, flexible (vectors can accommodate shRNAs, for example), and great for proof-of-concept studies. Right now, the technique works best on young mice, and Golde says they want to get to where they can transduce the whole brain in adults as well.
Moving on to other neurotoxic pathways, Guillemin, a neuroimmunologist working with Bruce Brew and Karen Cullen in Sydney, gave a primer on tryptophan oxidation and outlined some data connecting this inflammation-related pathway to Alzheimer disease. The bottom line is that the induced catabolism of tryptophan to form the excitotoxin quinolinic acid may be one reason why inflammation is dangerous to neurons.
The metabolism of the essential amino acid tryptophan is initiated during immune responses by the cytokine-induced expression of the enzyme indoleamine 2,3-dioxygenase (IDO). This enzyme catalyzes the rate-limiting first step in the catabolism of Trp down the kynurenine pathway. The resulting depletion of cellular tryptophan inhibits the replication of intracellular pathogens, and also plays a role in immune tolerance by inhibiting T cell activation. One downstream product of the pathway is quinolinic acid (QA), an excitotoxin that binds to NMDA receptors. QA also increases glutamate release by neurons, blocks glutamate uptake by astrocytes, and can induce apoptosis in astrocytes. IDO protein and QA are both increased in AD brain (Guillemin et al., 2005). In vitro treatment with Aβ42, but not 40, induced IDO expression in macrophages, and to a lesser extent human fetal microglia. This suggests that macrophages and microglia around plaques might be producing quinolinic acid, and Guillemin showed that in moderate stages of AD, microdissected amyloid plaques which were surrounded by activated macrophages and microglia contained a lot of quinolinic acid, but in later disease, where inflammation was less apparent, there was less.
QA is not produced by neurons, but immunostaining showed it to be present in neurons from AD brains. How does it get there? To answer this question, the researchers applied QA to brain tissue, and saw uptake into vacuoles in neurons. They also presented data that QA induces γ-secretase activity, tau phosphorylation and apoptosis in cultured neurons. QA also activates astroglia, which could amplify IDO expression by making more cytokines.
These results suggest that QA could play a role in inflammation-mediated neurodegeneration. Specific kynurenine pathway inhibitors are now in clinical trials for stroke. If the QA hypothesis is correct, these might be expected to be of some use in AD.
From neurotoxicity to neuroprotection. Efrat Levy, Nathan Kline Institute, Orangeburg, New York, showed new data supporting the role of the endogenous protease inhibitor cystatin C as a neuroprotective agent in AD. By way of background, cystatin C inhibits the cathepsin cysteine proteases, which play a role in many diseases. In AD, cathepsin B in particular has been implicated as a β-secretase and more recently, as an Aβ-degrading enzyme (see ARF related news story).
Cystatin C first came to attention when it was recognized that a mutated form of the protein causes cerebral amyloid angiopathy. But another genetic variant, the CST3 polymorphism has been associated with late-onset AD, and works in conjunction with ApoE4 to increase risk for the disease. The polymorphism is associated with decreased secretion of the protease inhibitor, and decreased levels of it in the CSF.
Previously, Levy showed that cystatin C binds to soluble Aβ and to the Aβ region of APP, and exerts an anti-aggregation activity. It is found in plaques, perhaps carried there by soluble Aβ. Consistent with an anti-plaque activity, Levy’s new results showed that crossing a cystatin C transgenic mouse with the Tg2576 or APP23 mouse caused a reduction in plaque load at 18 or 12 months, respectively. The cystatin C transgenic mice displayed no changes in Aβ production, but there was a reduction in fibrillar Aβ in the crossed mice.
Cystatin C expression is known to increase in response to injury—but is it neuroprotective? Adding cystatin C (0.2-0.4 μM) directly to N2a cells in culture protected them from serum deprivation. The same conditions also protected cells from toxicity of preformed Aβ fibrils. The protection was not due to dissolution of the fibrils, since incubation of preformed Aβ fibrils with cystatin C did not dissolve the fibrils, but did prevent their toxicity. Finally, Levy showed that cystatin C also protected the cells against toxicity of Aβ oligomers.
Recent data on cathepsin B knockout mice (see ARF related news story) suggested that cathepsin B protects from amyloid toxicity by degrading fibrillar Aβ, but this new data shows that cathepsin B inhibitor is protective. How can these results be reconciled? Levy says the investigators are currently studying how cystatin C levels affect cathepsin B and D activity in the cells and in animals. Levy thinks the neuroprotective effects of cystatin C may not involve cathepsins directly, but may instead involve the induction of protective autophagy in response to the stress of serum deprivation.
Bill Klein, Northwestern University, Chicago, presented some new data on the action of Aβ oligomers, or ADDLs, on neurons. He showed that memantine, the NMDA receptor antagonist and memory enhancer, blocks the ability of Aβ oligomers to increase reactive oxygen species in neurons. Antibodies to the NMDA receptor N-terminus, but not its C-terminus, had the same effect. The results focus attention squarely on the NMDA receptor as a mediator of the toxicity of Aβ oligomers, but the identity of any protein receptor for these Aβ assemblies remains unknown. Klein said they have isolated a large receptor complex, but have not identified the ADDL binding site.—Pat McCaffrey.