This is Part 1 of a three-part series. See also Part 2 and Part 3. Read the entire series.
28 June 2011. Founded in 1477, Uppsala University in Sweden has a long tradition of advancing the cutting edge of medicine, from its pharmacopeiac gardens tended by Carl Linnaeus of taxonomy fame, to a seventeenth-century anatomy theater hosting public operations and dissections. In this day and age, the institution aims to lead in part by hosting a conference on new technologies that are pushing toward earlier diagnosis and treatment of Alzheimer’s disease and other neurodegenerative disorders, which are looming as twenty-first-century health crises of aging Western societies. Organized by Lars Lannfelt, Martin Ingelsson, and Lars Nilsson, the 2nd International Conference on Neurodegenerative Disorders: Immunotherapy and Biomarkers took place at Uppsala University 26-27 May 2011. At about 150 attendees, the small size of the meeting belied its impact. Most participants questioned by a reporter said this meeting targets the right niche at the right time, calling it “a keeper.” With potential immunotherapeutics mushrooming, Roger Nitsch, University of Zurich, quipped that before long, Lannfelt’s catalogue of antibodies might rival Linnaeus’s one of plants.
Immunotherapy for AD started about a decade ago with Elan/Wyeth’s active immunization program targeting soluble amyloid-β (see ARF related news story), and that is still the leading target of many academic and industry labs. At the same time, this meeting showed that programs targeting other forms of Aβ (see below), tau (see Part 2), and proteins involved in other neurodegenerative diseases (see Part 3) are rapidly catching up.
One reason immunotherapy has become so attractive to drug developers is that it bypasses the need for medicinal chemistry, noted Peter Seubert, from Elan Pharmaceuticals, San Francisco. Medicinal chemists tweak small-molecule drug candidates to improve safety, pharmacodynamics, and pharmacokinetics. Their work can be time consuming and expensive. Antigens and antibodies do not need such delicate manipulations. Another draw of immunotherapy is that, contrary to conventional wisdom, a small proportion of antibodies made by the immune system or injected into the bloodstream do appear to cross the blood-brain barrier into the central nervous system—something scientists at the meeting puzzled over but seemed to accept. A case in point is bapineuzumab, a humanized monoclonal antibody that is being closely watched by observers in research and beyond. It lowered brain amyloid when given intravenously to patients enrolled in a Phase 2 trial (see ARF related news story on Rinne et al., 2010). A current Phase 3 trial powered to test for cognitive outcomes is expected to end in 2012. Though no data are officially available yet, Seubert noted that the trial has not been scuppered by vasogenic edema, as some had expected following the appearance of this side effect in some patients in Phase 2 (see ARF related news story). In April 2009, Elan dropped the highest doses of bapineuzumab from the trial in an attempt to mitigate this risk (see ARF related news story). Later that year, JANSSEN Alzheimer Immunotherapy Research & Development, a subsidiary of Johnson & Johnson, acquired all rights to Elan’s Alzheimer's immunotherapy program.
Brendon Binneman, from Pfizer Inc., Groton, Connecticut, reviewed ponezumab, this company’s humanized mouse monoclonal antibody to the C-terminal end of Aβ (see ARF related news story and clinical trials). The company is awaiting results of a Phase 2 trial begun in 2008, and will not present ponezumab data at ICAD next month in Paris, France. Binneman said Pfizer believes this antibody acts via the peripheral sink hypothesis, which posits that antibodies that mop up Aβ in the bloodstream help to coax the peptide out of the brain by tipping the blood/brain equilibrium. Binneman said the company likes to think of it as “brain dialysis.” This would make it similar to Lilly’s solanezumab, which is in Phase 3 (see ARF related news story). In Phase 1 safety trials, ponezumab had a long half-life of about 20 days in blood. Only about 0.5 percent of the infused protein made its way into the cerebrospinal fluid (CSF). After a single dose in humans, total plasma Aβ levels rise, as does CSF Aβ, which Binneman considered evidence that the peripheral sink idea might be working. The antibody appears to be safe so far, with no evidence of encephalitis or vasogenic edema. Ponezumab is modified to reduce activation of the innate and cellular immune systems, and it does not bind to Aβ well in tissue. These may be beneficial properties, said Binneman, because the antibody may be less likely to cause the type of blood vessel damage that leads to microhemorrhage or vasogenic edema.
Lars Nilsson, also from Uppsala University, described preliminary work using mAb158 to develop a PET ligand for protofibrils in the brain. To test the concept, he labeled the antibody with radioactive iodine I-125, and found this left the antibody’s immunoreactivity intact. He then tried to detect the radiolabeled antibody in mouse brain in vivo by SPECT, but found that it lacked the sensitivity of PET. Changing tack, Nilsson infused the mice with antibody, then removed the whole brain and imaged it ex vivo. After a single injection, the antibody peaked in the brain after three days but took almost a month to disperse. This is much longer than PIB, AV-45, and other amyloid ligands take to vacate the brain. Nilsson is trying to optimize the ligand by using F(ab) fragments and I-124, which has a shorter half-life. He did say that the antibody can detect protofibrils in the mouse brain at young ages before amyloid has accumulated, suggesting that if this ligand could be perfected, it might serve as an early marker of pathology and complement current plaque ligands.
Other immunotherapies aimed at different Aβ species are in various preclinical stages. Cynthia Lemere, Brigham and Women’s Hospital, Boston, and Thomas Bayer, University of Gottingen, Germany, collaborate with the German biotech company Probiodrug AG, based in Halle, Germany, to target truncated Aβ with a cyclized glutamate (pyroglutamate) at the amino end. Pyroglutamate Aβ3-42 is reputed to be highly toxic and likely to form oligomers (see ARF related news story). In Uppsala, Lemere reported on her prevention and therapeutic tests in mice of a monoclonal antibody from Probiodrug. When given intraperitoneally over 32 weeks as a prophylactic to young APP/PS1 mice, the antibody reduced cortical plaque deposits and lowered Aβ and pyrogluAβ in the hippocampus and the cerebellum. It calmed inflammatory gliosis as judged by the number of microglia sprouting the inflammatory marker CD45. As a therapy given to 23-month-old mice for seven weeks, the antibody had a less robust effect, showing only trends for decrease in pyrogluAβ and reduction in cerebellar pathology. In addition, Lemere reported on an active vaccination trial in six-month-old J20 mice. Given monthly over eight months, pyrogluAβ3-9, coupled to keyhole limpet hemocyanin as an adjuvant, elicited an immune response that generated mostly IgG1, IgG2a, and IgG2b antibodies. Immunized animals had fewer fibrillary plaques in the brain than controls, less gliosis, and yielded less guanidine hydrochloride soluble Aβ. Interestingly, mice inoculated with a control, full-length Aβ accumulated less pyrogluAβ than mice immunized with the truncated pyroglutamate antigen. Lemere said that this is most likely because the former vaccine prevents the deposition or promotes the removal of plaques that trap pyroglutamate Aβ. The point of the comparison, she said, was to test if pyrogluAβ acts as a seed for plaques. “In terms of plaque pathology, we saw reductions with both antigens. I do think pyroglutamate Aβ is a seed, but I don’t think it is the only seed,” she told ARF.
Bayer and colleagues raised antibodies against pyrogluAβ3-38 (see Wirths et al., 2010 and Wirths et al., 2010). They chose this antigen because it aggregates more slowly than pyrogluAβ3-42, and fast aggregation might complicate the immune response. The scientists identified the antibodies by their binding to human brain tissue in immunohistochemical assays. In biochemical assays, Bayer was surprised to find that one, 9D5, had an unusual binding pattern. It does not react with full-length Aβ1-42, nor with Aβ3-42 monomers, dimers, or oligomers greater than 20-mers. Under native conditions, it does seem to bind pyroglutamate Aβ oligomers in the 4- to 20-mer range, and it can block aggregation and reduce toxicity of pyrogluAβ. In three-month-old 5xFAD and APP/PS1KI mice, 9D5 detected no antigen. At six months it did detect some Aβ species, and immunoreactivity in neurons and microglia grew from there over the course of the next six months, Bayer said. Unlike generic Aβ antibodies, 9D5 did not react with plaques in brain tissue of non-demented people. When asked if this means plaques are not a reservoir of pyrogluAβ peptides, Bayer replied that plaques react strongly to other antibodies that specifically recognize pyroglutamate Aβ. In contrast, 9D5 reacts with plaques to a minor extent, showing preference for neurons, microglia, and blood vessel walls. This would fit the theory that small amounts of pyrogluAβ oligomers in the 4- to 20-mer range can form seeds for aggregation, said Bayer. He also presented a small mouse study indicating that 9D5 reduces plaque load and stabilizes cognition as judged by an elevated plus maze test. Injected once per week (10 mg/Kg) into 4.5-month-old 5xFAD mice, the antibody reduced both pyroglutamate Aβ and full-length Aβ plaques. Bayer believes this finding suggests that these small, 4- to 20-mer, pyroglutamate Aβ oligomers act as seeds for aggregation.
For his part, Lannfelt expressed concern about procedural artifacts in the study of truncated pyrogluAβ, specifically the suggestion that brain tissue that has been stored for some time contains much more pyrogluAβ than fresh brain tissue. Lemere replied that this has not been an issue in her hands. On the contrary, she said, leaving tissue too long in fixative can dramatically reduce pyrogluAβ signals, something researchers should keep in mind. Lemere told ARF that she found a distinct difference between tissue that is briefly fixed (two weeks) and routinely fixed (months to even years). In the latter, the fixative seems to mask staining for pyroglutamate Aβ, most likely because of molecular crosslinks that form when tissue sits in formaldehyde.
Lannfelt, who co-founded the Swedish biotech company BioArctic Neuroscience, is developing antibodies to protofibrillar forms of Aβ. The pharma company Eisai Inc. is now testing a humanized monoclonal antibody (BAN2401) in Phase 1. Both Lannfelt and Eisai’s Andrew Satlin summarized this program’s status at the AD/PD meeting in Barcelona last April (see ARF related news story).
In addition, Lannfelt and colleagues are using antibodies to Arctic peptides to develop assays for protofibrillar Aβ. Frida Ekholm-Pettersson, from Uppsala University, described a sandwich ELISA test using the monoclonal antibody mAb158, the mouse forerunner of BAN2401. The ELISA detects protofibrils in extract from human AD patients, but not from people who had frontotemporal dementia. Ekholm-Pettersson said she has not tested the assay on samples from people with other neurodegenerative diseases yet. To confirm that the assay does indeed detect protofibrillar species, she separated synthetic Aβ by ultracentrifugation and analyzed fractions by atomic force microscopy. The antibody detected Aβ species of median size, that is, in the second of three size fractions. The researchers are still characterizing the structures. The ELISA also detects protofibrils in human plasma, but the researchers ran into interference problems when trying to assay cerebrospinal fluid. They are still working on resolving those issues. Ekholm-Pettersson cautioned that one of the problems with this, and similar assays, is that human fluids can contain heterophilic antibodies that recognize immunoglobulins from other species. In this case, it appears that human antibodies recognize the mouse monoclonals. She depleted human fluids of IgG immunoglobulins to overcome this interference.
Tempering the general enthusiasm about immunotherapy at this conference, Dave Morgan, University of South Florida, Tampa, offered some cautionary notes. For one, he is troubled that some antibodies can have acute effects in mouse models. Since Aβ takes a long time to accumulate, Morgan said he finds it surprising that a single antibody dose can repair the brain. 3D6, the mouse precursor to bapineuzumab, for example, reverses cognitive defects in Tg2576 animals after a single dose. Lannfelt echoed this concern, saying that if accumulation takes decades, then perhaps it is best not to try to remove it too quickly. Aβ trapped in blood vessel walls can cause microhemorrhages, for instance, which may be exacerbated if therapy is too aggressive (see ARF related news story). Morgan showed that age may play a role in this. Looking at young mice with aggressive Aβ accumulation and older mice with more gradual pathology, his group found that microhemorrhages are much more prevalent in the older mice. Monocyte recruitment into the brain was also more rampant in the older mice. To some extent, this effect might be due to age-related differences in the immune system. The overriding point, Morgan said, is that age—which mice model poorly—is an important factor to keep in mind when developing immunotherapies for AD.—Tom Fagan.
This is Part 1 of a three-part series. See also Part 2 and Part 3. Read the entire series.