This is Part 1 of a two-part series. See also Part 2.
9 December 2009. Pyroglutamate Aβ has been making a return to center stage in AD research after languishing in relative obscurity for a decade following its initial discovery. Researchers from the German biotech company Probiodrug AG in Halle rekindled interest in this post-translational modification of the peptide implicated in Alzheimer disease. Outsiders to the field of AD research, these researchers claimed at conferences that a two-step process of truncation of Aβ’s N-terminus, followed by enzymatic cyclization of its new end, generates a particularly incendiary form of Aβ that makes for a better drug target than does the full-length version itself. Meeting sessions sponsored by the company featured mostly its own scientists, leaving the field at large skeptical at first. But pyroGlu Aβ’s flame is spreading. At recent conferences and increasingly in the literature, a growing number of academic groups have begun presenting their independent studies. These academic groups are confirming some of the company’s ideas, questioning others, and pushing the topic forward in the process. See below for a summary of the main points thus far. (For previous Alzforum stories on pyroGluAβ, see ARF related AD/PD 2007 Salzburg story; ARF related Keystone 2008 story; ARF related AD/PD 2009 Prague story.)
At the ICAD conference held last July in Vienna, Austria, academic research presentations began with a talk by Cynthia Lemere of Brigham and Women’s Hospital in Boston. In order to establish how common this form of Aβ truly is in brain, Lemere obtained monoclonal antibodies the Probiodrug scientists had raised against Aβ cyclized at the 3 position, i.e., pyrogluAβ3-42, and tested the antibody’s performance across a wide swath of AD-relevant material including human, two different species of non-human primate, and at least six different lines of transgenic mouse. This work recapitulated and extended earlier work by Takaomi Saido at RIKEN in Saitama, Japan, who had used a polyclonal antibody in his early work that put pyroGluAβ on the map (Saido et al., 1995).
“These new monoclonals are extremely specific, and they confirm Takaomi’s work,” Lemere told Alzforum. Comparing the new pyroGluAβ antibodies to a standard Aβ antibody, Lemere first showed that it stained all plaques in the brains of 12 of 12 AD brains examined to date. The pyroGluAβ antibody also stained all plaques in all Down syndrome brains examined so far, as well as small amounts of diffuse plaque found in seven of 10 aged controls. PyroGluAβ was apparent in human AD cortex and hippocampus. “In humans, just about every amyloid plaque was positive for pyroglutamate Aβ,” Lemere said.
Lemere next examined the brains of vervet monkeys from a colony kept on the eastern Caribbean island of St. Kitts. Starting at age 15, these animals develop cerebral Aβ plaques in the parenchyma and blood vessels; these deposits, too, were heavily labeled with the new pyroGluAβ antibodies. A different primate model showed similar results. Cottontop tamarins are a small native species living in Colombia’s rain forests. They are endangered and not sacrificed for research, but some archival tissue is on hand for study at the New England Primate Research Center in Southborough, Massachusetts, Lemere said. Sections from most of the 18 brains available in this way had Aβ plaques and some CAA starting at around 12 years of age; about half had pyroGluAβ.
At the Society for Neuroscience Conference last October in Chicago, Rebecca Rosen of Atlanta’s Emory University told this reporter that she has obtained similar results as Lemere. Using a commercial anti-pyroGluAβ antibody from IBL Japan, Rosen first confirmed with Western blots that the antibody did not recognize Aβ1-40 or 1-42, and then tested it on human and non-human primate brain. PyroGluAβ staining came up intensely on Aβ plaques and vascular amyloid in cortical section from AD brain, from aged chimpanzees, rhesus macaques, and squirrel monkeys. In 2006, Rosen presented a poster at SfN in Atlanta, Georgia, reporting that she and her colleagues had detected pyroGluAβ3-42 in cortical tissue extracts of both human and monkey temporal and occipital cortex using MALDI-TOF mass spectrometry.
A separate, international collaboration of scientists found much the same. In the October 13 Journal of Neural Transmission, the groups of Thomas Bayer at Germany’s University of Goettingen, Lars Lannfelt and Martin Ingelsson of the Sweden’s Uppsala University; Gerd Multhaup at Free University of Berlin; Paul Lucasson at University of Amsterdam; and David Brody of Washington University, St. Louis, Missouri, reported results of their own two new monoclonal antibodies against pyroGluAβ. These antibodies heavily stained amyloid plaques in all of 14 samples of sporadic AD, as well as sections of familial AD caused by the Arctic and by the Swedish APP mutations and by a PS1 mutation (Wirths et al., 2009). PyroGluAβ staining is abundant in brains of people from a large Colombian pedigree with a different PS1 mutation, as well, Lemere told this reporter. When asked at recent conferences, other scientists said they harbored little doubt that this modified form of Aβ constitutes a significant component of deposited amyloid in AD. On this point, the new data confirm published studies.
Both the Lemere lab and the international collaboration noticed pyroGluAβ-positive plaques in a fraction of cognitively normal human controls, though fewer than in AD brains. This suggests, as has other research before (e.g., Vanderstichele, 2005), that pyroGluAβ appears in early stages of Alzheimer disease, but it also differs from observations in a recent publication (Schilling et al., 2008). Active discussion also arises from the question of whether pyroGluAβ acts as the seed for plaque formation or not. Many scientists agree with the conclusion that pyroGluAβ aggregates more readily and is more stable and toxic than full-length Aβ. But whether it precipitates amyloid deposition in the development of human disease is still unclear. Based on data emerging this year, academic groups tend to argue that the pyroGlu form is unlikely to be the initial seed. Rather, they say, deposition might start with full-length Aβ; perhaps even harmlessly enough for a while, at least in some people. Later on, possibly fanned by neuroinflammation that upregulates pyroGlu’s generating enzyme glutaminyl cyclase in local brain areas, cyclization could occur on existing plaques and render them more toxic, Lemere speculated. This hypothesis is difficult to test directly in humans. One way to look at it indirectly would be to see if people who age cognitively intact have predominantly full-length Aβ, if any, in their brains.
The question of what comes first can be addressed in mouse models by staining brain sections at different time points across the mice’s lifespan. At ICAD, the Lemere lab reported initial data of exactly such a study, and added further data at SfN. In early summer, Lemere approached six colleagues to request sections of their respective transgenic mouse lines. “All responded promptly, showing not just generosity but also growing interest in pyroGluAβ in the research community,” Lemere said. So far, data from 10 widely used transgenic strains are in, including, for example, the mThy-1-hAPP751 (Rockenstein et al., 2001), TgSwDI: APP (Davis et al., 2004), PSAPP (Holcomb et al., 1998), 5XFAD-APP/PS1 (Oakley et al., 2006), 3XTg-AD (Oddo et al., 2003), and J20APP (Mucke et al., 2000; Chin et al., 2005; Palop et al., 2005; Aucoin et al., 2005; Patel et al., 2005; Chin et al., 2004; Moolman et al., 2004; Seabrook et al., 2004; Palop et al., 2003). As expected, results varied somewhat along with the known variation of amyloid deposition between and even within a given transgenic model; however, all had in common that as the mice aged, full-length Aβ deposition showed up first, followed by pyroGluAβ, in a subset of amyloid plaques. Whether it was at two months, three months, or six months of age that Aβ plaques first appeared, whether in parenchyma or blood vessels, they were always full-length Aβ plaques. As the mice grew older, some of these plaques became positive for pyroGluAβ, as well. It was never the other way around, Lemere said. In those mouse models that feature neuronal loss, this loss generally occurred around the time pyroGluAβ became abundant, Lemere noted.
In the past six months, Jeffrey Frost in the lab performed extensive single and double immunofluorescence labeling comparing full-length and pyroGluAβ species. “Using this method, we see much more pyroGluAβ in the mouse models, even at younger ages. However, it is not apparent in every plaque or amyloid-bearing blood vessel. Instead, in the mice, it tends to be associated with compacted, thioflavin S-positive fibrillar plaques and vessels. In humans, it is observed in both compacted and diffuse Aβ deposits. We have experiments underway to help determine if pyroGluAβ is, in fact, necessary for plaque deposition,” Lemere wrote to ARF.
The international group led by Bayer further extended this finding with one particular mouse line. Using an APP/PS1 knock-in mouse initially made by Laurent Pradier at Aventis (Casas et al., 2004; ARF related news story), these scientists found that as the mice aged, the number of pyroGluAβ plaques kept increasing over time, whereas that of full-length Aβ plaques even decreased somewhat. The interpretation here would be that, as disease progresses, the N-terminus of Aβ in plaques gradually becomes chewed off and the exposed glutamate cyclized. This would generate more and more of the stable pyroglutamate form at the expense of the full-length form, implying that deposited amyloid undergoes continuous rearrangement over the course of years. A recent paper correlating amyloid pathology and dementia drew attention for showing that this link weakens in the oldest old; however, the study assessed plaques only with full-length Aβ antibodies, not anti-pyroGluAβ antibodies (Savva et al., 2009).
Last but not least, here’s one question where the pyroGluAβ field is quite unsettled: Does pyroGluAβ play an important role inside neurons? Academic groups have no broadly overlapping data yet to suggest as much. The Lemere lab, in surveying a range of different mouse models, found no significant intraneuronal pyroGluAβ in any of them, including the 5XFAD-APP/PS1 line that demonstrably accumulates Aβ42 inside neurons (Oakley et al., 2006; ARF related SfN story). On the other hand, Bayer noted evidence linking intraneuronal aggregation of pyroGluAβ to neuron loss in the APP/PS1KI model (Breyhan et al., 2009), as well as in a separate model that expresses only transgenic Aβ3-42 (not APP) in neurons (Wirths et al., 2009; see comment below). This issue generated discussion at ICAD, but no emerging consensus as yet.—Gabrielle Strobel.
This is Part 1 of a two-part series. See also Part 2.