Scientists have known for some time that amyloid plaques alone correlate poorly with the severity of AD. However, the presence of pyroGluAβ, which lurks in plaques in both sporadic and familial AD, does correlate with disease severity, said Hans-Ulrich Demuth of Probiodrug. This modified Aβ species comes about when the first two N-terminal amino acids of Aβ42 are chopped off, and an enzyme called glutaminyl cyclase (QC) ties together the ends of the exposed glutamate residue, making pyroglutamate. This creates a highly stable, neurotoxic form of Aβ that aggregates faster than conventional full-length Aβ (see Piccini et al., 2005 and ARF related news stories: AD/PD 2007 Salzburg story; Keystone 2008 story; AD/PD 2009 Prague story; and Society for Neuroscience 2009 Chicago story). Recently, researchers led by Thomas Bayer at the University of Göttingen, Germany, developed an antibody that recognizes oligomeric forms of pyroglutamateAβ. They reported that this antibody binds to plaques in the brains of familial AD patients, and to a lesser extent to tissue from sporadic cases. Bayer and colleagues also reported that these oligomers are decreased in plasma taken from AD patients (see Wirths et al., 2010). Although there is broad agreement that pyroGluAβ is present in AD, whether it seeds aggregation remains somewhat controversial. PyroGluAβ production depends on QC, and this enzyme is increased in AD brains, making QC a potential target for therapeutic strategies.

Drug Companies Get Into the Pyro Act
Pharmaceutical and other biotech companies are now turning their attention to pyroGluAβ. Researchers from Merck & Co., based in West Point, Pennsylvania, developed a sandwich ELISA for pyroGluAβ, described in a poster by first author Guoxin Wu. This is a highly sensitive assay in which one antibody is used to capture the protein of interest and another is used to detect it. They tested it on APP/PS1 mice, as well as on human normal and AD brains, and verified the results using surface enhanced laser desorption/ionization time-of-flight mass spectrometry. The ELISA assay recovered nearly all the pyroGluAβ spiked into buffer, with a detection limit of 3 pM at the low end, the researchers said. With this method, the scientists determined that pyroGluAβ is elevated 8.5-fold in AD brains compared to controls. This contrasts with a 2.6-fold increase in all other forms of Aβ in Alzheimer’s brains.

Researchers at 21^st Century Biochemicals in Marlborough, Massachusetts, described on a poster polyclonal antibodies they generated to be specific for two forms of pyroGluAβ, pE3Aβ and the smaller, truncated pE11Aβ. First author Eric Berg demonstrated specificity by Western blot and ELISA. The researchers then used these antibodies to stain AD and normal elderly brains. They found pyroGluAβ formed the center of plaques, with full-length Aβ at the periphery, suggesting that pyroglutamate forms of Aβ seed deposits.

Mouse Pyrotechnics? Modeling QC Role
For their part, the Probiodrug crowd added new data as well. Several researchers presented data from mouse models that appeared to strengthen the case for QC as a therapeutic target. Bayer, who consults for Probiodrug, crossed 5XFAD mice with mice that express human QC under the control of the neuron-specific Thy1 promoter. Normal 5XFAD mice already produce much more pyroGluAβ than do other AD models, but in the presence of human QC, the levels went up higher still, Bayer said. These mice developed worse motor and memory problems than the 5XFAD mice. Bayer and colleagues also generated 5XFAD mice with endogenous QC knocked out. The knockouts showed a significant rescue of wild-type behavior in the elevated plus maze, demonstrating a role for endogenous QC in the pathology of AD mice.

Similarly, Stephan Schilling of Probiodrug described a cross of an AD mouse that expresses human APP containing the Swedish and London mutations (APP-SL) with the mouse expressing neuron-specific human QC. By nine months of age, the amount of pyroGluAβ in the double transgenics was two- to fourfold higher than in APP-SL mice, Schilling said. At six months of age, double transgenics showed worse memory and behavior problems in Morris water maze and fear conditioning tests than did APP-SL mice, and a greater activation of microglia. When researchers gave a QC inhibitor starting at three months, however, they saw less pyroGluAβ at eight months.

PyroGluAβ Processing
Anca Alexandru, from the company’s Munich site, used mouse models to see whether different methods of inducing pyroGluAβ made a difference. She generated a transgenic mouse (TBA2.1) that expresses a secretory truncated Aβ missing two N-terminal amino acids. By two or three months old, these mice developed extracellular pyroGluAβ deposits, neuroinflammation, and severe neurodegeneration, Alexandru said, losing 35 percent of hippocampal neurons. These mice showed extreme motor problems as early as four to six weeks, making behavioral testing difficult. Alexandru and colleagues generated another mouse strain (APP-NLQ-10) that expresses full-length mutant human APP with the faster-cyclizing glutamine at position 3 of Aβ (normal Aβ has glutamate at that position). These mice modeled a later-onset, slower-progressing disease than TBA2.1, Alexandru said. At two to three months, they had significant intracellular Aβ and pyroGluAβ deposits, but no inflammation or degeneration in their brains. By 15 months, these mice carried a heavy load of pyroGluAβ and plaques and showed astrogliosis. The mice did not have motor problems, but demonstrated cognitive difficulties at a late age. Together, the results indicate that the subcellular localization of pyroGluAβ determines the type and severity of neuropathology, Alexandru said.

Vivian Hook of the University of California in San Diego also investigated pyroGluAβ secretion. She used primary cultures of neuronal-like chromaffin cells to show that various forms of Aβ, including pyroGluAβ, colocalize with secreted neuropeptides and neurotransmitters (such as galanin, somatostatin, dopamine, and epinephrine) in regulated secretory vesicles. These vesicles also contain full-length APP, β- and γ-secretases, and QC, indicating that the components necessary to process Aβ are present in these organelles.

Previous work has shown that QC not only injures neurons by producing pyroGluAβ, but also has a role in promoting inflammation. The enzyme acts on monocyte chemoattractant protein (MCP-1, also known as CCL2), making a pyroglutamate derivative that is stable and resists proteases. MCP-1 activates microglial migration and neuroinflammation, is upregulated in early AD, and has been found to contribute to degeneration in AD mouse models. Holger Cynis of Probiodrug described the discovery of an isoenzyme of QC that does not get secreted, but is confined to the Golgi apparatus inside cells. To dissect its role, Cynis and colleagues compared inflammation in QC knockout mice and isoQC knockout mice. In QC knockouts, they saw no difference in cyclized MCP-1 or monocyte counts, but in isoQC knockouts, both were sharply reduced. This suggests that isoQC controls the inflammatory response, while QC produces pyroGluAβ. This would appear to make QC a more attractive drug target.

Dissecting Molecular Mechanisms
One of the burning questions about pyroGluAβ is how it inflicts its harm. A talk by Justin Nussbaum, who works in the laboratory of George Bloom at the University of Virginia in Charlottesville, provided clues to the mechanisms of pyroGluAβ toxicity. The laboratory receives research support from Probiodrug. Nussbaum treated primary cortical neuron cultures with synthetic pyroGluAβ and conventional Aβ42, and found that pyroGluAβ poisoned cells at a fivefold lower concentration than did Aβ42. When Nussbaum mixed pyroGluAβ with a 19-fold excess of Aβ42 during oligomer formation, however, he achieved a mixture that was 10- to 50-fold more deadly to neurons than either peptide alone. Mixed oligomers of Aβ continued to promote the formation of toxic Aβ species even in the absence of pyroGluAβ, Nussbaum said, suggesting a prion-like propagation mechanism. Intriguingly, the toxicity of Aβ mixtures depended on the presence of tau, as neuronal cultures from tau knockout mice remained healthy at Aβ concentrations that killed normal neurons within 24 hours. This jibes with previous AD research showing that tau acts downstream of Aβ (see ARF related news story on Götz et al., 2001 and Lewis et al., 2000 and ARF related news story on Roberson et al., 2007).

Edging Toward Therapy
If pyroGluAβ does seed oligomers and Aβ protofibrils, then preventing pyroGluAβ production would be a logical therapeutic intervention. Probiodrug is pursuing a small molecule drug-discovery program for inhibitors of QC, Demuth said. They have synthesized and tested more than 1,300 compounds for their ability to specifically bind and inhibit QC and isoQC, and are identifying promising candidates. Demuth said he hopes for Phase 1 trials next year.

Researchers led by Cynthia Lemere at Brigham and Women’s Hospital are collaborating with researchers at Probiodrug to look at using the immune system to inhibit pyroGluAβ formation. First author Jeffrey Frost presented a poster describing a passive immunization study in two-year-old APP/PS1ΔE9 mice. Weekly injections of 200 micrograms of a monoclonal anti-pyroGluAβ antibody for seven weeks slashed total Aβ by 25 percent and fibrillar Aβ deposits by 50 percent in the hippocampus. Similar results were obtained in 3xTg AD mice, as described in a poster by first author Qiaoqiao Shi. The researchers are now testing prevention by means of passive immunization in younger AD mice, as well as examining the effects of active immunization in J20 mice. They plan to compare the effects of vaccination with pyroGluAβ versus Aβ40 and 42.—Madolyn Bowman Rogers.