23 November 2005. Oligomeric forms of amyloid-β (and also early pretangle forms of tau) captured much of the buzz at the 35th Annual Meeting of the Society for Neuroscience, held last week in Washington, D.C. Despite some labs' preoccupation with oligomers these days, other groups steadfastly maintain that having a head full of plaques is a dangerous thing in its own right. Immunotherapy appears able to remove plaques and is inching its way through a notoriously fickle clinical trials process. β and γ-secretase remain leading targets for small-molecule drugs, yet whether inhibitors for these enzymes will be able to remove mature plaques is an open question. New evidence shows them to be quite entrenched, at least in mice, and new approaches for removing them may need exploring.
A collaborative research team led by Joanna Jankowsky at California Institute of Technology in Pasadena and David Borchelt’s former group at Johns Hopkins School of Medicine in Baltimore, Maryland, addressed this question genetically. (Borchelt recently moved to the University of Florida, Gainesville.) These scientists wanted a fresh approach to measure in an animal model what happens when Aβ production is suppressed once plaques are in place. After all, people most likely already have plaques when they come to therapy. (PET imaging with Pittsburgh Compound B [Mintun, 2005] is confirming that people with even very mild Alzheimer disease have abundant fibrillar amyloid pathology in their frontal and temporal cortex. Presentations on longitudinal human PIB studies at the SfN conference began reporting that MCI patients who have fibrillar amyloid progress to diagnosis. Researchers even noted cases of patients who appeared normal clinically and neuropsychologically when they entered, but had abundant PIB binding and then declined to an AD diagnosis a few years later.)
To ask what closing the Aβ spigot will do in mice modeling this stage, Jankowsky and colleagues developed a new model that overexpresses the Swedish and Indiana mutations of APP driven by the CamKII promoter with the TET-off system. This allows them to end APP expression at will. They did so at 6 months, when the mice had extensive amyloid plaque pathology, and then kept APP silent for 3 months and 6 months, respectively, before analyzing the mice. Jankowsky showed that amyloid pathology indeed did not progress any further. At the same time, however, the existing pathology stayed stubbornly put: Thioflavin S-positive cored plaques, as well as diffuse amyloid, remained unchanged, and biochemical analysis indicated that both insoluble and soluble Aβ levels remained largely stable. Gliosis stayed in place, as did dystrophic neurites. The study appeared online in PLoS Medicine during the conference (see Jankowsky et al., 2005).
Beyond these measures, Jankowski et al. have not, to date, analyzed neurodegeneration in their new mice. Neither did they present data about behavioral or synaptic deficits. This could address the question of how damaging the residual pathology is, and whether plaques and newly generated, soluble forms of Aβ might be toxic in quite separate ways, as a growing number of researchers are beginning to show.
Jankowsky’s data suggesting that mature brain amyloid is quite entrenched contrasts with previous work reporting amyloid clearance by immunotherapy in mice and humans, and lentiviral delivery in mice of siRNA against BACE (Singer et al., 2005). Her results raise the question of how much secretase inhibitors alone can help. Scientists favor secretase inhibitors because they lie upstream of a pathogenic cascade. Some believe that the different pools of amyloid are connected by dynamic equilibria, so that depleting one form will slowly draw down the others, as well, and the brain will eventually clear them. Jankowsky’s data, however, suggest that clearance of existing amyloid will require separate measures beyond secretase inhibition, for example, immunotherapeutic stimulation of glial cells or activation of Aβ-degrading enzymes.
Can Enzymes Safely Degrade Plaques?
The Aβ-degrading enzymes that are known so far generally remove monomeric or oligomeric forms of Aβ, not the tightly aggregated fibrils that make up dense plaques. At the SfN conference, researchers from Washington University, St. Louis, Missouri, introduced a new player that may nibble away at these stable structures. In side-by-side posters, Xiaoyan Hu and Kejie Yin, working with Jin-Moo Lee and other collaborators at Washington University School of Medicine in St. Louis, in essence propose that when astrocytes gear up to remove plaques, they release the potent enzyme matrix metalloproteinase 9 (MMP-9), which might attack plaques in the human AD brain. MMP-9 belongs to a large family of enzymes that remodel tissues by digesting extracellular matrix components; in the brain they have been implicated in plasticity and regeneration. The Washington University researchers picked up earlier work showing that astrocytes participate in clearing Aβ (see Koistinaho et al., 2004). This study had focused on the role of ApoE in this process. Lee’s group then set out to identify the active proteases. Yin and colleagues first noticed that the activated astrocytes that congregate around plaques in APP/PS1 transgenic mouse models expressed MMP-2 and MMP-9. At the SfN conference, they reported that conditioned medium from cultured astrocytes degrades not only synthetic Aβ into characteristic fragments characterized by mass spectrometry, but also fibrillar Aβ made in vitro. When incubated with slices of APP/PS1 transgenic mice, this astrocyte medium reduced the Aβ load later determined by ELISA.
Hu then compared MMP-9 to neprilysin, IDE, or ECE (but not ACE). Of these enzymes, all degraded soluble Aβ, but only MMP-9 degraded fibrillar Aβ. It did so at sites in the C-terminal hydrophobic region of Aβ that are thought to be important for β-pleated sheet formation. Hu’s poster showed electron microscopy images of Aβ fibrils incubated with these enzymes. The fibrils exposed to MMP-9 looked sparser and showed what appeared to be degraded material. Moreover, Hu incubated fresh brain slices of APP/PS1 mice with each of these proteases and, again, only MMP-9 reduced subsequent thioflavin S staining. Antibodies against MMP-9 lit up astrocytes surrounding plaques in the brains of old APP/PS1 and APPsw mice, as well as compact plaques themselves.
It is important to note that even if further work were to show that MMP-9 indeed helps degrade plaques in AD brain, turbo charging this enzyme does not automatically follow as a therapeutic strategy. MMP-9 has been implicated in the pathogenesis of inflammatory, infectious, and cancerous diseases in many organs. In the brain, it has been implicated in brain tumors such as astrocytomas and is part of an inflammatory cascade. Studies on neuroprotection models of stroke and other conditions suggest that, at least in those instances, MMP-9 needs restraining, not further activation. Further research is needed to sort out how best to handle this voracious enzyme.—Gabrielle Strobel.