|Home | About | Journals | Submit | Contact Us | Français|
Soluble oligomeric amyloid-β (Aβ) species are toxic to many cell types and are a putative etiological factor in Alzheimer's disease. The NINDS-Custom Collection of 1040 drugs and biologically active compounds was robotically screened for inhibitors of Aβ oligomer formation with a single-site biotinylated Aβ(1-42) oligomer assembly assay. Several quinoline-like compounds were identified with IC50's < 10 μM, including the antiprotozoal clioquinol that has been reported to have effects on metal ion metabolism. The 2-OH, 4-OH, and 6-OH quinolines do not block Aβ oligomer formation up to a concentration of 100 μM. Analogs of clioquinol have shown activity in reducing Aβ levels and improving behavioral deficits in mouse models of Aβ pathology. The inhibitory effects of clioquinol and other 8-OH quinoline derivatives on oligomer formation in vitro are unrelated to their chelating activity. Crosslinking studies suggest that clioquinol acts at the stage of trimer formation. These preliminary data may suggest that 8-OH quinolines have the potential for suppressing Aβ oligomer formation which should be considered when assessing the effects of these compounds in animal models and clinical trials.
The amyloid-β (Aβ) peptide has a high propensity to assemble into multimeric states of unusual stability. In humans with Alzheimer's disease (AD), Aβ fibrils associate into amyloid plaque structures to form the classical Alzheimer's disease pathology along with neurofibrillary tangles. Soluble oligomeric assemblies of Aβ peptide, of which some are precursors for fibril and plaque formation, are not detected by standard histological evaluation. Unlike oligomeric native proteins, the Aβ oligomers are considered misfolded structures due to their stable cross-β sheet secondary structure that renders them resistant to proteolytic degradation. Formation of Aβ oligomers follows an ordered assembly process and has been studied in some detail by Bitan  and Glabe  (Figure 1). The two extra C-terminal amino acid residues of the Aβ(1-42) peptide confer a high propensity to form stable oligomeric species at nM concentrations of Aβ(1-42). This concentration approaches the levels of Aβ oligomers found in the brain and cerebral spinal fluid (CSF) of AD subjects [11, 13] and cell culture media  that block long-term potentiation in culture. The Aβ(1-40) peptide is much less prone to form oligomers and when formed, the Aβ(1-40) oligomers differ structurally from Aβ(1-42) oligomers and are much less stable [2, 37, 39]. Mutagenesis of the C-terminal residues of Aβ(1-42) indicates a structural role for this part of the peptide in oligomer assembly . In vitro incubation of synthetic Aβ(1-42) peptide produces a series of rapidly exchanging unstable low-n oligomers culminating in a proportion of relatively stable 12-24-mers that can associate to higher order species. These soluble misfolded oligomers of the Alzheimer's Aβ peptide are significantly more toxic to neurons and other cell types than are monomers or fibrils . A number of studies have suggested that these soluble oligomers may be an etiologic agent in AD. Therapeutic approaches, such as immunization and inhibition of the secretase enzymes that produce Aβ from its precursor protein, βAPP, indirectly target oligomers by reducing levels of the monomeric peptide from which it is formed. Since Aβ monomer may have positive effects in the brain , it may be difficult to completely inhibit Aβ production or to clear oligomers from the brain once they have formed. Therefore, an alternative approach may be to directly block oligomer formation.
High affinity binding of a small molecule to a conformationally flexible peptide, such as Aβ, is difficult to achieve, although certain peptides designed to interact with β-strand edges specifically block β-sheet formation [9, 18]. An ordered peptide assembly mechanism implies the existence of multimeric intermediates that may have sufficient surface area to interact with a small molecule modulator. An empirical approach targeting low abundance intermediates of fibril formation by assaying for sub-stoichiometric inhibition has yielded inhibitors of Aβ fibrillation [24, 25, 34].
Oligomers of synthetic Aβ(1-42) peptide comprised of 10-12 monomers or more are stable to spontaneous dissociation and have been detected in vitro and in vivo [21, 28]. Further assembly into protofibrils and fibrils can occur at higher peptide concentrations [10, 12, 17, 19]. Spontaneous peptide nucleation, the sensitivity of a nucleated assembly process to container surface properties and the air-water interface, and effects of the shear forces of agitation have hampered the search for inhibitors of oligomer formation. An oligomer-conformation-specific antibody has been used in vitro to identify compounds that prevent the formation of immunoreactive oligomers from synthetic peptide . Interestingly, compounds that were either selective for the inhibition of fibril formation or for oligomer formation were identified. Other compounds inhibited both fibril and oligomer formation. Therefore, small molecules and conformation-specific antibodies  in vitro can distinguish between intermediates in both fibril and stable soluble oligomer formation.
The single-site biotin-avidin Aβ(1-42) oligomer assay  (illustrated in Figure 2) was developed to facilitate the rapid screening of small molecule compound libraries. Substituting the biotin-avidin system for antibodies reduces the impact of potential artifactual interaction of compounds with β-sheet regions of the antibodies and reduces reagent costs. The biotin-avidin interaction is also familiar to screeners, and few library compounds interfere with the binding. The sensitivity of the single-site biotin-avidin Aβ oligomer assay allows the IC50 determinations to be measured in the nanomolar range using a total of 10 nM bioAβ(1-42). At this low concentration of Aβ(1-42), the confounding of potency with compound stoichiometry at higher concentrations of peptide is now avoided. This single-site biotin-avidin assay was used to screen the NINDS-Custom Collection (NINDS-CC) of 1040 drugs and biologically active compounds (http://www.msdiscovery.com). Most of the compounds were inactive in our assay, but we have identified a series of 8-hydroxyquinolines that inhibit Aβ oligomer formation. These compounds may be of interest because clioquinol (an 8-hydroxyquinoline) and derivatives are currently being tested as therapeutics for AD based on their ability to complex metal ions [6, 30]. We find that in our in vitro system, these compounds inhibit Aβ(1-42) oligomer formation, apparently independent of metals, such as zinc. The larger and more chemically diverse ApexScreen 5040 library (TimTec) is also being assayed by the single-site method.
A detailed description of the materials and the robotic screening procedure is provided as Supplemental Material.
The NINDS-CC contains a significant number of disinfectants, detergent-like molecules, alkyl amphiphiles, hydrophobic cyclic peptides, polyketides, and other high molecular weight natural products. Since detergent-like and micellar molecules can non-selectively interfere with or even stimulate Aβ oligomer formation, any hits with these types of structures were excluded from further analysis. One particular series of small molecule inhibitors was interesting because of its homology to clioquinol, a compound shown to be active in animal models of Aβ pathology  and has been in clinical trials for AD [30, 32]. This chemical series is of the hydroxyquinoline (HQ) class shown in Table I. The highest concentration used was 100 μM. This concentration was significantly lower than the reported mM aqueous solubility of these compounds [8, 16]. The compounds were first dissolved in DMSO and then diluted into aqueous buffers to give a solution with no detectable precipitates. Clioquinol was chosen as a representative HQ oligomer assembly inhibitor for further study.
Because the biotinyl single-site assay detects stable oligomers that are > 30 kDa, size exclusion chromatography was used to verify that oligomer formation was prevented. The amount of Aβ detected in the void volume of the Sephadex G75 column was drastically reduced by clioquinol (data not shown). This indicated that there were very few high molecular weight oligomers present. Separate experiments were performed to control for the possibility that clioquinol or any of the other HQ compounds produced apparent inhibition by interacting with the assay components. The compounds did not prevent the interaction of preformed oligomers with Neutravidin™ or with Streptavidin-HRP (data not shown).
Incubation of preformed bio42 oligomers with the HQ compounds in Table I for long periods of time (Figure 3) did not significantly reduce the amount of oligomers. A possible exception may be the incubation of 6-HQ at 28 hours. The amount of preformed oligomers recovered with those inhibitors was similar to the amount of oligomers recovered when incubated in the presence of vehicle (3% DMSO). The lack of dissociation of preformed oligomers suggests that inhibition may occur early in the assembly process.
Early stages of oligomer assembly were probed by crosslinking studies. Bitan, et. al. [3, 4] had reported that synthetic peptide at nM concentrations did not accumulate the small (low-n-mer) intermediates on the pathway to stable Aβ oligomers. Hence, these small intermediates are rarely observed by Sephadex G75 chromatography because they rapidly dissociate and exchange with monomer [3, 4]. These small intermediates could be trapped with a crosslinking agent [4, 26]. Crosslinking agents with two different chemical reactivities and different lengths were employed to avoid potential interference by the compounds. The photochemical zero-length PICUP reaction exploits the free radical-catalyzed intermolecular reaction of Y10 in Aβ with an adjacent nucleophilic residue on another Aβ molecule . Tween 20 (0.1% v/v) inhibits stable oligomer formation measured by single-site assay or size exclusion chromatography . It completely inhibits crosslinking of dimer and higher multimer formation from monomer, but it does not inhibit crosslinking of preformed oligomers. Clioquinol appears to inhibit trimer formation, and has little effect on dimer crosslinking (Figure 4). This lack of effect on dimer crosslinking may suggest that the aromatic OH of clioquinol is not significantly interfering with the crosslinking reaction.
A second crosslinking agent with different chemistry corroborated the observations of clioquinol inhibition of oligomer assembly with the free radical-based PICUP reaction. The bis-aldehyde glutaraldehyde agent reacts with amino groups and multimerizes in solution to span longer distances between reactive residues . Similar results to the PICUP measurements were observed with glutaraldehyde. These findings support the interpretation that clioquinol and other hydroxyquinolines may be destabilizing the formation of trimeric and higher Aβ intermediates. Additional experiments and different techniques are required to further probe this finding since crosslinking is a covalent reaction that could introduce artifacts in the analysis.
8-Hydroxyquinoline derivatives in the NINDS-CC were effective inhibitors of in vitro bio42 oligomer formation. The fact that one of these 8-HQ compounds had already shown some activity in animal models of AD pathology and in humans [1, 32] increased the importance of investigating this class of compounds for potential anti-oligomer activity. Clioquinol originally was an antiprotozoal agent that was discontinued for that indication because of reported severe neurologic side effects, paralysis and blindness, particularly at high doses . The results from a small pilot phase II clinical trial of clioquinol in 36 AD subjects indicated a decrease relative to placebo control in the rate of decline of the most severely affected AD cases in the ADAS cognitive test . In addition, PBT-1 (clioquinol) and PBT-2 (a clioquinol analog modified to alleviate the toxicity issues) have shown some efficacy in the Tg2576 mouse model of Aβ pathology by mitigating behavioral deficits and other measures of the phenotype . A recent trial of PBT2 revealed no safety issues and showed improvement in two measures of executive function .
The mode of action of these agents is proposed to be metal ion (primarily Zn and Cu) chelation by the compound [6, 30]. Clioquinol has also been shown to bind more avidly to the metallated form of Aβ peptide . The metals hypothesis of Alzheimer's disease has recently been refined to extend beyond removal of metal ions from the Aβ peptide [6, 30, 38]. Zinc is known to aggregate proteins, in particular, Aβ . The arrangement of histidine residues 6, 13, and 14 in Aβ leads to metal ion binding to the peptide, particularly with redox active metals (copper, iron) that are implicated in oxidative chemistry [35, 36]. We excluded the likelihood that trace levels of transition metal ions effected the oligomerization of bio42 in our system since neither inclusion of 10 mM EDTA, 1 mM of 1,10-phenanthroline nor the presence of 100 μM ZnCl2 affected oligomerization or oligomer detection in the assembly or capture reactions (data not shown). The lack of effect of added zinc on oligomer formation could be due to the low (10 nM) peptide concentration. Small oligomers (<30 kDa) are also not detected in the single-site assay configuration.
In summary, our results suggest that, in addition to any effects on zinc, copper, or other metal ion-dependent mechanisms, there are intrinsic effects of the hydroxyquinolines on Aβ oligomerization that deserve further study. In addition, the responses in animal models and in the clinical trial of related compounds suggest that the observed effects of these types of compounds may also include effects on oligomer formation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.