Budding RNAi Therapies, APP Protein Interaction Map Impress at Meeting

Early last month, 118 scientists braved a snowstorm on the Eastern seaboard to gather at Cold Spring Harbor Laboratory on Long Island, New York, for a conference on therapeutic opportunities in neurodegenerative diseases. Inclement weather grounded about 30 participants in their hometowns, including a few speakers. Even so, the two-day meeting, co-organized by Sam Gandy, Marcy McDonald, and Harry LeVine, featured a packed and stimulating lineup of talks and poster presentations. Here are some highlights.

Presenilin Complex Grows Again
Gerard Drewes of Cellzome AG in Heidelberg, Germany, a biotech company on the campus of the European Molecular Biology Laboratory, reported results of a proteomics project aimed at identifying new therapeutic targets in the pathways surrounding APP processing. Working with Adele Rowley of Cellzome’s London site, Drewes tagged proteins in human neuroblastoma cells, purified interacting proteins with a method called tandem affinity purification (TAP), and then identified the components of complexes with mass spectrometry. This form of proteomics establishes maps of protein-protein interactions that are called “interactomes” as they draw from previously cloned cDNA libraries of open reading frames (“orfeomes”).

Drewes showed an elaborate network of about 225 protein-protein interactions, which he said represents a near-complete TAP/MS analysis of all proteins implicated in APP processing. Besides the obvious core components, such as the secretases and nicastrin, this map also contains the novel γ-secretase components Aph-1 and Pen-2, which the TAP/MS approach independently identified at about the same time that other labs found them (see news story 617). The map also contains more peripheral players, for example, the catenins and cadherins, as well as numerous new proteins. Some of the newly identified components might be drug targets, he added. In total, the study confirmed 39 of the 49 APP-related protein-protein interactions described in the literature. It also found new interactions between known players, for example, between BACE1 and nicastrin, Drewes said.

The scientists grouped the protein-protein interactions into first-pass functional areas such as dimer formation, novel proteases, trafficking/glycosylation, or cholesterol-regulated interactions. Drewes described comigration studies and other experiments to validate that certain proteins indeed do interact and to understand their role in Aβ production and trafficking. For example, RNA interference experiments (where one lowers translation of a particular protein in the cell by removing its mRNA) indicated that without two new complex components, Aβ generation in the cell decreases.

Drewes also presented some data showing that the new presenilin complex components Aph-1 and Pen-2 are involved in assembling the proteolytic complex in the Golgi apparatus. Presenilin and nicastrin shuttle between the endoplasmic reticulum and the Golgi, Drewes suggested, but assemble in the Golgi when interacting with these new components. This scenario might help resolve what some researchers call the spatial paradox, namely that gamma-secretase is mainly found in earlier membrane compartments in the cell than APP, he said.

Philip Wong of Johns Hopkins University in Baltimore, Maryland, presented new data on the role of nicastrin, another poorly understood member of the g-secretase complex. Wong reported that mice homozygous for nicastrin deletion die at embryonic day 9.5/10.5 with patterning defects similar to those seen in notch or PS1/2 double-knockout embryos. This indicates that nicastrin is essential for notch cleavage. Cultured fibroblasts from these mice did not produce Aβ, and fibroblasts from heterozygous nicastrin knockouts produced about half as much Aβ as wild-type. Wong et al. also asked what nicastrin’s role might be in assembling g-secretase subunits into a functional complex. By complementing nicastrin function in nicastrin-/- cells with various mutants, they identified a conserved domain (DYIGS) in nicastrin that appears to facilitate assembly.

Cholesterol Connection is Branching Out
As cholesterol is becoming a more widely accepted player in Alzheimer’s pathogenesis, several laboratories are branching out to investigate additional ways to modulate its levels beyond the use of statins, which inhibit cholesterol synthesis, appear to lower brain Aβ levels, and are moving into clinical trials for AD. These groups are also probing the mechanism by which cholesterol metabolism might affect APP processing. Two presenters described initial results of these efforts.

Dora Kovacs of Massachusetts General Hospital in Charlestown, Massachusetts, reported follow-up data of her prior work showing that inhibiting the enzyme ACAT lowers Aβ generation in several types of cultured cells (see ARF related news story).
Kovacs used the ACAT inhibitor CP113,818 by Pfizer (a related compound is in phase 3 trials for peripheral vascular disease) to run proof-of-principle studies on whether inhibiting ACAT could reduce brain Aβ in mice. She implanted a 21-day continuous-release pellet in PS1/APP double-transgenic mice and found that brain cholesteryl ester levels decreased by 86 percent after 21 days. (Blood cholesterol was down by 29 percent and liver cholesteryl esters were down by 93 percent.) This quick response of brain cholesterol indicates that this inhibitor enters the brain, since brain cholesterol reacts to changes in the peripheral cholesterol pool much more slowly.
Preliminary experiments hint that decreases in Aβ generation accompany reduced cholesteryl ester levels. Kovacs found insoluble Aβ40 levels to be down after 21 days of ACAT inhibitor treatment but added that she needs to substantiate this finding with more animals and by adding behavioral tests.
Cell-culture experiments aimed at understanding how decreased ACAT activity could affect Aβ generation indicate that normally, a novel, alternate APP cleavage event creates a 55 kDa, N-terminally truncated protein. This cleavage prevents Aβ generation and is regulated by sterols. It is also sensitive to ACAT, and ACAT inhibition may enhance it, Kovacs suggested.

Saravanapavan (Paru) Parvathy, working with Suzana Petanceska and Larry Refolo at NYU in Orangeburg, as well as Gandy and Michelle Ehrlich at Thomas Jefferson University in Philadelphia, is working out signal transduction pathways of statins in neurons in the hope of finding other ways of activating the non-amyloidogenic cleavage of APP by α-secretases. While trying to figure out which pathway might mediate an a-secretase-activating effect of statins that was previously reported, Parvathy et al. discovered that an experimental PI-3-kinase inhibitor, LY294002, enhances this effect in neuroblastoma cells. The compound does so at much lower concentrations than are necessary to inhibit PI-3-kinase, indicating this might not be the pathway at play here. The effect also appears to be independent of protein kinase C and MAP kinase phosphorylation. In theory, a statin-enhancer might potentiate the effect of a statin, a desirable goal especially if the statin dose needed to sufficiently reduce Aβ turns out to exceed doses considered safe for long-term use. Protein kinase inhibitors have been studied extensively as cancer therapeutics but tend to be toxic. This concern does not apply here, as PI-3 kinase is a lipid kinase, Gandy pointed out. Wortmannin, another PI3-kinase inhibitor, reduced brain Aβ in APP-transgenic mice (see Haugabook, 2001; see also Alzforum discussion).

Getting Rid of Aβ-Put in Perspective
Don Frail of Pharmacia Corporation in Kalamazoo, Michigan, stimulated discussion by laying out some of the challenges of testing the amyloid hypothesis in the clinic. For example, one of the research priorities toward better clinical studies lies in understanding how the different pools of Aβ in serum, the CSF, and brain change in response to drugs inhibiting Aβ production. How are these pools connected What are the steady-state levels between soluble and aggregated pools of Aβ in brain, and does the presence of plaque affect this relationship In animal models, does transgene expression affect these steady-state levels, given that they express both mouse and human Aβ Numerous studies of the effect of treatment on Aβ pools exist, but they don’t allow a complete comparison because most did not measure all three pools of Aβ, Frail said.

Pharmacia’s own studies (Lanz et al., Soc. Neuroscience, 2002, Abstract 483.6) of the effect of the γ-secretase inhibitor DAPT indicated that a three-hour treatment reduced total brain Aβ levels in young APP-transgenic animals that did not yet have plaques, but not in older mice that had plaques. However, the same treatment decreased CSF levels of Aβ in young as well as in old mice, suggesting that CSF levels reflect the soluble Aβ fraction, not the plaque-Aβ pool. Translated to humans, this would suggest that CSF Aβ would measure efficacy of an agent (like a secretase inhibitor) that would primarily reduce the soluble pool of Aβ. This could provide evidence of a desired effect in shorter time frames than are currently used for AD trials. To this end, imaging methods and other surrogate markers to measure the effect of experimental therapies must be standardized and validated in consensus with FDA, Frail urged.

Frail also noted that the “peripheral sink” hypothesis, which is gaining momentum in the field as an attractive way to remove brain Aβ indirectly, with Aβ-antibodies in the blood, still needs to prove that brain Aβ levels actually drop acutely when Aβ is bound to antibody in the periphery. Research by Berislav Zlokovic and others presented at last month’s Neuroscience meeting in Orlando indicates that Aβ is actively transported across the blood-brain barrier by RAGE and LRP-1, raising questions about how such transporters would respond to a peripheral sink of Aβ. Ron DeMattos reviewed the current state of knowledge on this approach, which so far indicates that in mice, peripheral Aβ-binding to antibody can indicate the level of amyloidosis in the brain (see ARF related news story).

Karen Duff presented new data on a variation of the peripheral sink hypothesis. She reasoned that if binding Aβ in the periphery “draws” it out of the brain, the binding agent need not necessarily be an antibody. This might open up a new drug-development avenue that avoids some of the risks of immunotherapies, she said. To test the idea, she used two molecules that have been reported to bind Aβ in the low micromolar range. Gelsolin is an actin-binding protein too large to enter the brain; GM1 ganglioside can enter in small amounts but does not lower Aβ levels there when injected directly in a control experiment, Duff said. The NKI team, led by Yasuji Matsuoka, treated PS/APP double-transgenic mice at nine weeks of age, when their amyloidosis begins. Both gelsolin and GM1 significantly reduced levels of brain Aβ extracted with formic acid, but were more effective at removing diffuse Aβ than thioflavin-S positive, fibrillar Aβ, Duff reported. GM1 did not work when injected into older animals that have abundant plaques. Duff added that the effect of these Aβ-binding agents on Aβ-plasma levels were not the same as for antibody treatment and emphasized that the dynamics of Aβ clearance in the periphery must be better understood to understand this relationship (see related ARF news story). However, she said, the data suggest that the strategy of finding small-compound drugs that bind Aβ is worth pursuing in trying to develop an Alzheimer’s treatment that works from outside the brain. The team’s data appeared in yesterday’s Journal of Neuroscience, see Matsuoka et al., 2003.

Vaccination Tests Move into Monkeys
The failure of Elan’s phase II trial of an Aβ42 vaccine (see ARF discussion) signaled the end of the first, heady-some say rushed-phase of AD immunotherapy development, when researchers had felt bold enough to jump from experiments in mice to humans with little study in between. Now, however, scientists are conducting more detailed research of the approach in higher mammals, time-consuming work that traditionally precedes human trials of most vaccines. Cynthia Lemere of Brigham and Women’s Hospital in Boston, Massachusetts, presented initial results of her vaccination studies with a colony of aging Caribbean Vervet monkeys that live in St. Kitts. Imported from Africa in the seventeenth century, these animals live up to 30 years in captivity and develop thioS-positive amyloid plaques in the brain’s parenchyma and on blood vessels. Their Aβ sequence is identical to that of humans, and ELISAs to assay Aβ levels in blood and CSF exist, Lemere said.

To measure humoral and cellular responses to Aβ immunization, Lemere’s collaborators injected five monkeys 15 to 30 years old five times over 90 days with 1 mg of Aβ40/42, and boosted once another 10 weeks later. The animals produced antibodies by day 42, which reached high titers by day 100. Their antibodies recognize an epitope between Aβ’s amino acids 1 and 7. The antibodies recognized monomers and oligomers, but not fibrils, nor APP or APP-CTF, Lemere reported. She has no data yet on a decrease in brain Aβ.

Preliminary measurements of Aβ levels in the CSF indicate that Aβ40 decreased 60 percent by day 100. Lemere et al. saw a transient rise in Aβ in plasma around day 22, followed by a decrease. (Similarly, a separate research collaboration led by Gandy and including Lemere and others elsewhere, reported a rise in plasma Aβ in vaccinated rhesus monkeys.) Aβ began to appear in urine at day 42, indicating there may be some clearance. T cell proliferation assays in three immunized and three control monkeys did not show different reactions to Aβ. Lemere et al. conducted this experiment to address the question whether the meningioencephalitis seen in 17 of 300 patients in the human trial was due to a T cell-based autoimmunue reaction. While she was surprised not to see T cell activation, she noted that her group so far has only done this experiment on five monkeys, an inconclusive number given that about five percent of the patients had the side effects.

This colony is interesting to study in other ways, as well, Lemere added. She has not found tangles yet, but has not analyzed sufficient numbers of animals to draw conclusions. The natural history of cognitive decline, if any exists, has not yet been studied in this colony, though a literature on how to assess dementia in monkeys does exist. While Lemere said that she has seen abundant neuritic dystrophy, she has not yet conducted a detailed study on neuronal loss.

RNAi and Viruses: New Therapies of the Future
Two scientists impressed the audience with presentations exploring the feasibility of a new treatment approach based on small inhibitory RNAs. The last issue of Science celebrates these siRNAs as the “breakthrough of the year” for their ability to target specific messenger RNAs for destruction, in a process called RNA interference, or RNAi. Some scientists in this growing research field are exploring the versatile roles these siRNAs play in the genetics and evolution of the cell, while others are investigating therapeutic applications of RNAi in damping down the activity of disease-related genes. Zuoshang Xu and Phillip Zamore of the University of Massachusetts Medical School in Worcester are trying to gauge their potential use in those neurodegenerative diseases that arise from gain-of-function mutations. A small number of cases of amyotrophic lateral sclerosis (ALS), for example, are caused by such mutations in the SOD1 gene and human mutant SOD1-transgenic mice are the leading model for drug discovery in this disease. Xu presented initial data from his study asking whether an RNAi-based strategy could selectively suppress expression of mutant SOD1 while leaving expression of the normal SOD1 unaffected. His lab tested various constructs and found that siRNAs homologous to the mutant SOD1 but differing from wildtype SOD1 at a single nucleotide position were able to do just that, not only in cultured cells but also in mice.

Could one devise such an approach for FAD presenilin mutations, tau mutations causing tauopathies, or even just hyperphosphorylated tau in AD (see, for example, Fath et al., 2002) Victor Miller, working with Henry Paulson and others at the University of Iowa College of Medicine in Iowa City, is planning to test such hypotheses. Like Wu, Miller reasoned that one wants to target a dominantly acting disease gene and leave expression of its normal cousin intact to preserve its protein’s function. He devised siRNAs that specifically silence expression of a missense tau mutation that causes frontotemporal dementia with Parkinsonism, FTDP-17. In these studies, Miller et al. also tested constructs that were able to reduce wild-type tau. This might be therapeutic in AD, although this remains entirely speculative at this point, Miller added.

In other diseases, however, one cannot directly target the offending mutation; this includes the expanded CAG repeats encoding polyglutamine tracts in diseases such as Huntington’s. To circumvent this problem, Miller targeted siRNAs to a nearby single nucleotide polymorphism (SNP) that segregates with the mutation causing spinocerebellar ataxia 3 (SCA3) and managed to selectively abolish expression of the mutant gene using both liposomal and viral delivery methods. Miller et al. worked with multiple cell lines including differentiated PC12 neurons and are currently preparing to test safety and efficacy in mouse models. While human RNAi AD therapies remain a pie in the sky at this point, the approach bears watching. The major limitation right now lies in how best to deliver these nucleic acids into the brain, said Miller.

On drug delivery to the brain, Beka Solomon presented previously published work on using filamentous phages to deliver cargo (in her case, therapeutic antibodies) to the brain via the nose (see Frenkel & Solomon, 2002).

Two other presentations updated the audience on the progress of using lentiviral gene delivery systems. Inder Verma at the Salk Institute and others have pioneered the use of these as an alternative gene therapy vehicle that promised more efficient and stable expression of the therapeutic gene than other vectors. In the past five years, a number of groups have refined vectors derived from lentiviruses that appear to enable efficient, long-lasting gene transfer into the CNS. Eliezer Masliah of the University of California, San Diego, described studies using lentiviral vectors to decrease amyloid deposition in transgenic animal models of AD and a-synuclein deposition in models of Parkinsonism. Working in collaboration with Verma and Fred Gage, also at the Salk, Masliah picked up extensive prior research by Takaomi Saido on neprilysin’s role in degrading small Aβ aggregates. Hypothesizing that waning neprilysin activity with age or polymorphisms leading to weak neprilysin activity might account for some cases of LOAD, the researchers outfitted a lentiviral vector with the neprilysin gene and injected it into one hemisphere of two different strains of APP-transgenic mice around the time they began developing plaques. They saw robust neprilysin transgene expression for up to a year, Masliah said. Transduced mice of both strains had fewer and smaller plaques in frontal cortex and hippocampus than did vector-injected mice, as well as a reduction in soluble Aβ.
Masliah also reported that, unlike the controls, the neprilysin-transduced mice had near-normal synaptic density. The group is currently analyzing the water-maze deficits of their mice.
Switching over to α-synuclein, Masliah said that he and many others believe this protein’s accumulation plays a significant role in Alzheimer’s, as well. He pointed to prior work suggesting that an interaction with β-synuclein prevents α-synuclein’s aggregation (see ARF related news story), and to observations in humans suggesting that the ratio between α- and β-synuclein expression is linked to whether a person develops disease. Masliah et al. packaged β-synuclein into a lentiviral vector to try it out as an α-synuclein aggregation breaker in transgenic mice. Indeed, he said, this strategy reduced α-synuclein aggregation, as well as a dopaminergic deficit, in the cortex and hippocampus of α-synuclein-transgenic mice.

Liang-Fong Wong, working with Nicholas Mazarakis and others at Oxford Biomedica in Oxford, England, described efforts at modifying viral vectors with specific glycoproteins in order to enable their retrograde axonal transport from the injection site in the CNS. With this approach, the scientists saw stable transfection of up to one year in neurons surrounding and projecting to the injection site in the striatum and spinal cord of adult rats. Injection of the vectors into the muscle of neonatal rats also resulted in retrograde transport to motor neurons in the spinal cord.

Finally, a talk on prion diseases got high marks for novelty, so this report includes it even though it is not directly relevant to Alzheimer’s. Marcela Karpuj, who works with Stanley Prusiner at University of California, San Francisco, reported that oligomers of a nuclease-resistant DNA derivative called phosphothioate DNA eliminated almost completely the infectious version of the prion protein from neuroblastoma cells previously infected with PrPSc homogenate. The effect was independent of the oligo’s sequence but specifically removed the PrPSc protein and its normal cellular counterpart PrPC while leaving other protein production untouched. Karpuj postulated that the DNA binds PrP at the cell membrane, becomes internalized, and then is degraded together with PrP in lysosomes. Cells so treated for seven weeks became resistant to re-infection with PrPSc. Last July, German scientists reported that injecting oligodeoxynucleotides into prion-infected mice extended their survival by an unknown mechanism (see Sethi et al., 2002). Karpuj suggested that perhaps such an agent could be delivered to treat prion diseases, but also noted that the small numbers of humans suffering from prion diseases may preclude a large effort to develop innovative drugs in this area. Even so, her study is the “first example of a DNA, regardless of its sequence, degrading a specific protein,” Karpuj said, and as such is of biological interest.

To view commentaries, primary articles and linked stories, go to the original posting on Alzforum.org here.

Copyright © 1996–2017 Biomedical Research Forum, LLC. All Rights Reserved.

Share this: