Orally available compound prevents deficits in memory caused by the Alzheimer amyloid- oligomers.код для вставкиСкачать
Orally Available Compound Prevents Deficits in Memory Caused by the Alzheimer Amyloid-␤ Oligomers Matthew Townsend, PhD,1* James P. Cleary, PhD,2,3 Tapan Mehta, BS,1 Jacki Hofmeister, VT,2 Sylvain Lesne, PhD,4 Eugene O’Hare, PhD,5 Dominic M. Walsh, PhD,6 and Dennis J. Selkoe, MD1 Objective: Despite progress in defining a pathogenic role for amyloid ␤ protein (A␤) in Alzheimer’s disease, orally bioavailable compounds that prevent its effects on hippocampal synaptic plasticity and cognitive function have not yet emerged. A particularly attractive therapeutic strategy is to selectively neutralize small, soluble A␤ oligomers that have recently been shown to mediate synaptic dysfunction. Methods: Using electrophysiological, biochemical, and behavioral assays, we studied how scyllo-inositol (AZD-103; molecular weight, 180) neutralizes the acutely toxic effects of A␤ on synaptic function and memory recall. Results: Scyllo-inositol, but not its stereoisomer, chiro-inositol, dose-dependently rescued long-term potentiation in mouse hippocampus from the inhibitory effects of soluble oligomers of cell-derived human A␤. Cerebroventricular injection into rats of the soluble A␤ oligomers interfered with learned performance on a complex lever-pressing task, but administration of scyllo-inositol via the drinking water fully prevented oligomer-induced errors. Interpretation: A small, orally available natural product penetrates into the brain in vivo to rescue the memory impairment produced by soluble A␤ oligomers through a mechanism that restores hippocampal synaptic plasticity. Ann Neurol 2006;60:668 – 676 The large, insoluble deposits (amyloid plaques) of aggregated amyloid ␤-protein (A␤) that occur abundantly in the limbic and association cortices of Alzheimer’s disease (AD) patients are in equilibrium with small, diffusible oligomers of the peptide that appear capable of interfering with hippocampal synaptic function and memory.1 Secreted forms of soluble A␤ oligomers can be generated by cultured cells that express human ␤-amyloid precursor protein (APP).2 These low-n oligomers are present in the conditioned medium (CM) at low-nanomolar to subnanomolar concentrations similar to those of A␤ in human cerebrospinal fluid (CSF) and brain.3,4 Radiosequencing, immunoprecipitation with various A␤-specific antibodies, and separation by size-exclusion chromatography under nondenaturing conditions have all established that these dimers, trimers, and tetramers are composed of N- and C-terminally heterogeneous human A␤ peptides, including the A␤1-40 and A␤1-42 species that occur in human brain and extracellular fluids.2,4 – 6 In normal adult rats, the oligomers have been found to potently inhibit hippocampal long-term potentiation (LTP) in vivo,7 and this synaptic effect can be prevented by A␤ antibodies in vivo via passive infusion or active A␤ vaccination.8 Moreover, intracerebroventricular (ICV) injection of the secreted oligomers interferes potently but transiently with the memory of a complex task in rats.9 Given the enormous personal and societal burdens that AD inflicts, agents that interfere with the adverse effects of A␤ on neuronal function are being sought intensively. McLaurin and colleagues10,11 identified myo-inositol, the head group of the naturally occurring glycolipid phosphatidylinositol, and certain ste- From the 1Center for Neurologic Diseases, Harvard Medical School and Brigham and Women’s Hospital, Boston, MA; 2Geriatric Research, Education and Clinical Center, Minneapolis Veterans Affairs Medical Center; Departments of 3Psychology and 4Neurology, University of Minnesota, Minneapolis, MN; 5School of Psychology, Queen’s University, Belfast, United Kingdom; 6Laboratory for Neurodegenerative Research, Conway Institute of Biomedical and Biomolecular Research, University College Dublin, Dublin, Republic of Ireland. M.T. and J.P.C. contributed equally to this work. *Current address: Merck Research Laboratories—Boston, Boston, MA 02115 This article includes supplementary materials available via the Internet at http://www.interscience.wiley.com/jpages/0364-5134/ suppmat Published online Dec 22, 2006 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.21051 Address correspondence to Dr Selkoe, Center for Neurologic Diseases, Harvard Medical School and Brigham and Women’s Hospital, Boston, MA 02115. E-mail: email@example.com Received Aug 7, 2006, and in revised form Oct 30. Accepted for publication Nov 6. 668 Published 2006 by Wiley-Liss, Inc., through Wiley Subscription Services reoisomers as small-molecule inhibitors of synthetic A␤42 aggregation and neurotoxicity in vitro. These investigators showed that, among inositol stereoisomers, myo-inositol, scyllo-inositol (also called scyllocyclohexanehexol or AZD-103), and epi-inositol, but not chiro-inositol, interacted with A␤42 peptide in vitro to promote its conformational change from random coil to ␤-sheet structure and stabilized it in small, nonfibrillar complexes.11 These stabilized complexes were significantly less toxic to NGFdifferentiated PC-12 cells and primary human neuronal cultures than were untreated A␤42 or chiroinositol–treated A␤42.11 The authors observed that the active inositol stereoisomers attenuated neurotoxicity, at least in part by interfering with the ability of A␤42 to interact with the neuronal membrane.11 They recently reported that scyllo-inositol reduces plaque burden in TgCRND8 mice and improves performance on memory tasks such as the Morris water maze.12 Many features of scyllo-inositol, including its small size, neutral pH, stability, stereo-selectivity, and favorable toxicity profile, make it a highly attractive candidate to test for therapeutic benefit in AD. Here, we demonstrate that this small, orally available compound can successfully neutralize the inhibitory ef- Fig 1. Cyclohexanehexols rescue long-term potentiation (LTP) from the inhibitory effects of soluble amyloid ␤ protein (A␤) oligomers. (A) Perfusion of wild-type mouse hippocampal slices with conditioned media (CM) collected from control CHO cells had no effect on LTP. The CM was applied to a recirculating perfusion for 20 minutes before delivering 4 highfrequency stimuli (arrows; 100Hz over 1 second). In contrast, the CM of 7PA2 cells readily inhibited LTP at about 1 hour after HFS and thereafter (7PA2 CM ⫺ LTP at 60 minutes ⫽ 133.2 ⫾ 5.4 standard error of the mean [SEM] [n ⫽ 26] and CHO⫺ CM LTP at 60 minutes ⫽ 213.4 ⫾ 12.0 [n ⫽ 20]; Student’s t test, p ⬍ 0.01). (B) A 15 -minute preincubation of 7PA2 CM with 1.25 M AZD103 (scyllo-cyclohexanehexol) before perfusion over brain slices was sufficient to rescue LTP (LTP at 60 minutes ⫽ 205.5 ⫾ 14.1; n ⫽ 11). Two enantiomers of AZD-103 were similarly tested. Epi-cyclohexanehexol also restored LTP to control levels (LTP at 60 minutes ⫽ 184.2 ⫾ 13.8; n ⫽ 7), whereas chiro-cyclohexanehexol was not significantly different from 7PA2 alone (127.9 ⫾ 6.3; n ⫽ 8) (analysis of variance [ANOVA] Tukey’s post hoc test, p ⬍ 0.05 for epi vs chiro and p ⬍ 0.01 for scyllo vs chiro). (C) Basal synaptic transmission was unaffected by AZD-103 (scyllo alone, no HFS), and LTP induced in the presence of CHO⫺ CM was similarly unaltered by AZD-103 (LTP at 60 minutes ⫽ 219.2 ⫾ 19.3; n ⫽ 7). The addition of 1.25 M AZD-103 to the perfusion 20 minutes after the addition of 7PA2 CM (postperfusion) did not rescue the inhibition of LTP (LTP at 60 minutes ⫽ 138.3 ⫾ 8.3; n ⫽ 5; ANOVA Tukey’s post hoc test, p ⬎ 0.05). fects of pre-existing A␤ oligomers on hippocampal LTP and rescue memory function in awake, behaving animals. Results Scyllo-inositol Prevents the Inhibition of Long-term Potentiation by Soluble Amyloid ␤ Protein Oligomers Field potential recordings were made in the CA1 region of wild-type mouse hippocampal slices. As in our previous work,5,7 a brief (20-minute) treatment of hippocampal neurons with A␤-rich CM from CHO cells stably expressing human APPV717F (7PA2 cells) caused a strong and statistically significant inhibition of LTP measured at 60 minutes after high-frequency stimulation, compared with CM from the untransfected parental line (designated CHO⫺) (Fig 1A). Lowmolecular-weight oligomers of A␤, particularly trimers, ‹ Townsend et al: A␤ Neutralized by AZD-103 669 isolated from the 7PA2 CM by nondenaturing size exclusion chromatography (SEC) have been shown to be sufficient to mediate this inhibition of LTP, whereas the simultaneously isolated A␤ monomer is without effect.5,13 To determine whether scyllo-inositol could neutralize the effect of soluble A␤ oligomers on this form of hippocampal synaptic plasticity, we preincubated 7PA2 CM in vitro for 15 minutes with 1.25 M scylloinositol before perfusion over brain slices. This low micromolar concentration of scyllo-inositol fully rescued LTP from the inhibitory effects of 7PA2 CM (see Fig 1B). Two enantiomers of scyllo-inositol (see Suppl. Fig. 3) were tested to establish whether the rescue of LTP was stereospecific within this family of compounds. Epi-inositol (epi-cyclohexanehexol), which protects neurons from synthetic A␤42 cytotoxicity,11 also rescued LTP, whereas chiro-inositol (chirocyclohexanehexol), which is inactive in assays of synthetic A␤42 neurotoxicity,11 was without effect (see Fig 1B). Thus, two cyclohexanehexol enantiomers previously shown to stabilize synthetic A␤ as a small, nontoxic conformer effectively neutralized the inhibitory effects of secreted, soluble A␤ oligomers on LTP. Because scyllo-inositol (AZD-103) proved to be more effective at protecting PC-12 cells than epi-inositol,11 all further studies were performed with the former compound. We performed additional controls to test AZD-103 for nonspecific effects on LTP. AZD-103 alone did not alter baseline synaptic transmission (no high-frequency stimulation), nor did it affect the normal LTP induced in the presence of control CHO⫺ CM (see Fig 1C). A 20-minute application of 1.25 M AZD-103 after perfusing slices with 7PA2 CM did not confer any protection on LTP compared with 7PA2 alone (see Fig 1C). These results demonstrate that although AZD103 can neutralize the acute synaptic effects of A␤ before the soluble oligomers encounter hippocampal tissue, it did not reverse the effects of A␤ under these experimental conditions. Dose and Time Dependence of the AZD-103 Effects The rescue of LTP by AZD-103 was found to be dose dependent. The final concentration of AZD-103 added to 7PA2 CM (“postconditioning”) was varied between 0.125 and 5 M and then tested in LTP experiments, as done previously (Fig 2A). The 1.25 M dose was found to be the lowest tested concentration that significantly improved LTP. The 7PA2 CM prepared for this set of experiments was slightly more potent than that shown in Figure 1, and thus should be compared with the internal controls within this experiment (see Fig 2A). A curve fit applied to the data set suggested an inhibitory concentration of 50% (IC50) of approximately 1.0 M. 670 Annals of Neurology Vol 60 No 6 December 2006 Fig 2. Dose- and time-dependent responses of AZD-103 on hippocampal long-term potentiation (LTP). (A) The concentration of AZD-103 added to 7PA2 conditioned media (CM) was varied between 0.125 and 5.0M before testing on LTP. In these postconditioning experiments (diamonds), 1.25M AZD-103 was the lowest concentration that provided significant protection against the inhibitory effects of A␤ on LTP at 60 minutes after HFS (5.0M AZD-103 LTP at 60 minutes ⫽ 201.3 ⫾ 12.9, n ⫽ 5; 1.25M AZD-103 LTP at 60 minutes ⫽ 183.4 ⫾ 10.4, n ⫽ 9; 0.5M AZD-103 LTP at 60 minutes ⫽ 134.1 ⫾ 6.0, n ⫽ 13; and 0.125M AZD-103 LTP at 60 minutes ⫽ 120.8 ⫾ 10.6, n ⫽ 4; analysis of variance [ANOVA] Tukey’s post hoc test, p ⬍ 0.05). Based on these data, the inhibitory concentration 50% (IC50) is predicted to be approximately 1M under these experimental conditions. In preconditioning experiments (circles), AZD-103 was added directly to the 7PA2 cells during the conditioning period, resulting in almost complete rescue of LTP at 0.125M (0.5M AZD-103 LTP at 60 minutes ⫽ 191.9 ⫾ 12.6, n ⫽ 7; and 0.125M AZD-103 LTP at 60 minutes ⫽ 184.7 ⫾ 10.7, n ⫽ 6; ANOVA Tukey’s post hoc test, p ⬍ 0.05). The controls for this experiment are: 1.25M AZD-103 ⫹ CHO⫺ (⫻), LTP at 60 minutes ⫽ 201.0 ⫾ 13.2, n ⫽ 7; 7PA2 alone (‚), LTP at 60 minutes ⫽ 112.8 ⫾ 12.4, n ⫽ 4; 7PA2 no HFS, LTP at 60 minutes ⫽ 88.4 ⫾ 7.7, n ⫽ 3 䊐). (B) The duration of the AZD-103 (0.5M) coincubation with 7PA2 CM was varied between 30 minutes and 4 hours (light gray bar, 30 minutes; gray bar, 2 hours; dark gray bar, 4 hours). (Compare to 15 min co-incubation at this dose shown in Fig. 2A.) No improvement in the small partial rescue of LTP at this low dose occurred with longer coincubation (ANOVA Tukey’s post hoc test, p ⬎ 0.05). We next determined whether application of AZD103 directly to 7PA2 cells during their production and secretion of the oligomers (referred to as “preconditioning”) would similarly rescue the effects of the secreted A␤. AZD-103 was added to 7PA2 cells at the beginning of the 15-hour conditioning period, after which the CM was collected (prepared as described in Methods) and perfused over brain slices as in Figure 1. Intriguingly, when applied directly to the A␤-secreting 7PA2 cells, AZD-103 proved more effective at rescuing LTP, achieving almost full rescue at the lowest concentrations tested (0.125M) (see Fig 2A, circles). This could not be attributed to toxicity to the 7PA2 cells, because they continued to robustly secrete a range of proteins (including A␤) into the medium at the same levels as untreated cells (see later). Therefore, AZD-103 not only neutralized pre-existing A␤ oligomers, but provided an additional benefit when applied directly to cells during oligomer generation and secretion. A time course was established to determine whether coincubating 7PA2 CM with AZD-103 for longer periods would affect the rescue of LTP. A dose of 0.5M AZD-103, which had a modest effect after a 15-minute postconditioning incubation with 7PA2 CM (see Fig 2A), was chosen to test whether extending the coincubation from 30 minutes to 4 hours would improve the effectiveness of the compound. As shown in Figure 2B, there was no additional benefit of incubations longer than 15 minutes. AZD-103 Binds to but Does Not Depolymerize CellDerived, Low-n Amyloid ␤ Protein Oligomers AZD-103 has been found to bind to and alter the conformation of pure, synthetic A␤ in vitro11 and to inhibit endogenous A␤ aggregation in vivo.12 To test for depolymerization of the oligomers by AZD-103, we fractionated the 7PA2 CM using SEC. As previously reported, this technique effectively separates soluble A␤ species by molecular weight.5 The cell-secreted oligomers were highly stable during overnight incubations at 37°C and did not spontaneously aggregate or disaggregate.4 Treating the oligomer-containing SEC fractions with AZD-103 (5M) for 14 hours did not significantly alter the pattern or intensity of the A␤ species (Fig 3A). We conclude that AZD-103 does not decrease the stability of low-n oligomers. Although AZD-103 did not alter the electrophoretic pattern of the SEC-fractionated oligomers on sodium dodecyl sulfate polyacrylamide gel electrophoresis, the compound did directly neutralize the synaptic effects of these isolated, low-n oligomers. SEC fractions containing A␤ dimers, trimers, and tetramers were pretreated with AZD-103 and tested in LTP experiments. Whereas untreated SEC fractions that contained these oligomers readily inhibited LTP as expected, pretreating equivalent SEC fractions with 1.25M AZD-103 fully restored normal LTP (see Fig 3B), confirming that AZD-103 specifically neutralizes the inhibitory effects of isolated, low-n oligomers of human A␤ on this form of synaptic plasticity. Because AZD-103 did not destabilize low-n oligomers yet did neutralize their effects on LTP, we hypothesized that perhaps AZD-103 binds to A␤ oligomers in their native conformation and masks one or more epitopes that are essential for their biological effects. If so, we reasoned that AZD-103 may compete with anti-A␤ antibodies for binding to A␤ in immunoprecipitation experiments. Although multiple antibodies to distinct epitopes within A␤ were tested (R1282, 6E10, m266, 4G8, and 12F4; see Supplemental Figs 1A and D), we observed no significant difference in the immunoprecipitation profile of 7PA2 CM pretreated with AZD-103 compared with that of untreated 7PA2 CM, as exemplified with R1282 in Figure 3C. Even high AZD-103 concentrations (100M) did not alter the ability to immunoprecipitate the A␤ species and detect them by Western blotting (see Supplemental Fig 1A). We conclude that AZD-103 neither destabilizes A␤ oligomers nor interferes with epitope recognition by multiple anti-A␤ antibodies. Because submicromolar concentrations of AZD-103 were sufficient to rescue LTP when applied directly to the 7PA2 cells (but not when applied to the 7PA2 CM post-conditioning), we investigated whether this difference could be attributed to changes in the pattern of oligomers found in the CM. However, no differences were observed in the oligomer/monomer patterns after direct treatment of the 7PA2 cells with AZD-103 (“preconditioning”) compared with the addition of AZD-103 to the collected CM (“postconditioning”) (see Fig 3C). Thus, the available data suggest that although AZD-103 does not interfere with A␤ synthesis or oligomerization in our system, it does have a more potent ability to bind and neutralize the oligomers when AZD-103 is present at the time of oligomer formation. Notably, these results also indicate that the health of the 7PA2 cells was not compromised by incubation with AZD-103. Epoxy-activated Sepharose has been shown to bind to myo-inositols similar to AZD-103.14 We therefore coupled AZD-103 or its inactive stereoisomer, chiroinositol, to this resin to determine whether AZD-103 would pull down A␤ species from 7PA2 CM (see Fig 3D and Supplemental Fig 1E). The chiro/epoxy resin preferentially pulled down monomeric A␤, whereas the AZD-103/epoxy resin preferentially pulled down A␤ trimers (n ⫽ 5) relative to beads alone. Notably, only the active stereoisomer, AZD-103, bound trimeric A␤ above background levels, and we have shown the trimer to be a particularly potent inhibitor of LTP.13 These results demonstrate that AZD-103 binds directly Townsend et al: A␤ Neutralized by AZD-103 671 Fig 3. AZD-103 neutralizes the synaptotoxic activity of isolated low-n amyloid ␤ protein (A␤) oligomers but does not alter their stability. (A) Size exclusion chromatography (SEC) was used to separate A␤ species as described previously.5 The fractions were incubated for 14 hours with or without 5.0 M AZD-103 at 37°C, and then subjected to Western blotting. Treatment with AZD103 did not cause a significant change in the size distribution of any A␤ species (blot is representative of three independent SEC runs). (B) Nevertheless, AZD-103 neutralized the synaptotoxic effects of SEC fractions 7 and 8 in LTP experiments. SEC fractions were preincubated with AZD-103 at 37°C for 20 minutes before delivering a HFS (as in Fig 1) (SEC frac 7 and 8 LTP at 60 minutes ⫽ 131.1 ⫾ 14.3, n ⫽ 7; SEC frac 7 and 8 ⫹ 1.25M AZD-103 LTP at 60 minutes ⫽ 222.0 ⫾ 26.47, n ⫽ 5; Student’s t test, p ⬍ 0.01). (C) Similar to the SEC results in the western blots in (A), immunoprecipitation of A␤ from whole 7PA2 conditioned media (CM) showed no significant effects of AZD-103 on A␤ oligomers. AZD-103 was applied directly to 7PA2 cells during the conditioning period (pre-cond) or else to the CM (post-cond) and immunoprecipitated with a polyclonal anti-A␤ antibody (R1282). There was no clear dose-dependent effect of AZD-103 as detected by blotting with anti-A␤ monodoxal antibody 6E10. (D) AZD-103, coupled to epoxy-agarose beads, precipitated A␤ oligomers from 7PA2 CM. Low levels of A␤ adhered nonspecifically to the beads alone. Nevertheless, AZD-103/epoxy-agarose preferentially bound trimeric A␤ (n ⫽ 5), which we have reported to be a potent inhibitor of LTP, whereas the inactive isomer chiro-inositol preferentially bound monomeric A␤. The standard pattern of A␤ bands in the 7PA2 cm is shown after R1282 IP in the right lane. An additional experiment using stronger blocking conditions is shown in Supplemental Figure 1. MW ⫽ molecular weight. to and neutralizes secreted A␤ trimers that inhibit hippocampal LTP. AZD-103 Prevents Reference Memory Errors Induced by Amyloid ␤ Protein Oligomers In Vivo We previously reported that the SEC-isolated A␤ oligomers from 7PA2 CM significantly increase errors during recall of a complex learned behavior in normal adult rats, whereas the A␤ monomer fraction from the same chromatography run and at greater concentrations has no effect.9 Similar results were also obtained with whole 7PA2 CM compared with control CHO⫺ CM.9 We therefore used the same learned task, the Alternating Lever Cyclic Ratio (ALCR) assay, to determine whether AZD-103 could mitigate the detrimental effects of soluble A␤ oligomers on cognitive behavior in normal adult rats. ALCR, a test of delayed alternation and complex reference memory, has proved sensi- 672 Annals of Neurology Vol 60 No 6 December 2006 tive to direct experimental administration of A␤ in several previous studies.9,15,16 In brief, rats learn a complex sequence of lever-pressing requirements to earn food reinforcement in a two-lever experimental chamber. The exact number of lever presses required for each reward changes by first increasing in 7 unequal steps from 2 lever presses per food pellet up to 56 presses per pellet, and then decreasing back to 2 required presses per reward (ie, the sequence 2, 6, 12, 20, 30, 42, 56, 56, 42, 30, 20, 12, 6, 2 presses). Subjects must also alternate between levers after each food reward. This “cycle” of response requirements repeats six times during each daily assessment session. Each failure to alternate levers and each premature switch from a correct lever choice is recorded as a “Switching Error.” If the subject perseverates in responding on an incorrect lever choice, these are recorded as “Perseveration Errors.” To assess the effect of AZD-103 on soluble oligomer-induced errors, we trained 35 rats under ALCR until relatively low baseline error rates demonstrated mastery of the complex reference memory rules associated with the task. Each rat then received a single permanent indwelling cannula placed in either the right or left lateral ventricle and was allowed to recover for 5 days.9 During the next week, baseline error rates were established under ICV injections of saline (0.9%) or “sham” injections during which the entire injection procedure was performed but no injectant was administered. All ICV injections were 20l over a 5-minute period given to awake, freely moving rats 2 hours before the ALCR behavioral testing session. As a positive control to test the sensitivity of the ALCR paradigm to disruption of reference memory and learned behavior, we administered the prototypical anticholinergic amnestic agent, scopolamine. Compared with baseline error rates (mean of three previous ALCR sessions), ICV scopolamine injections significantly increased Switching Errors to 400% ( p ⬍ 0.001) of baseline rates and increased Perseveration Errors to 611% ( p ⬍ 0.001) of baseline rates (all tests ⫽ Bonferroni-corrected Student’s t test). When 7PA2 CM containing soluble A␤ oligomers and monomers was injected ICV (20l) (see Supplemental Fig 2), mean Switching Error rates increased significantly to 120% of their baseline levels ( p ⫽ 0.044), and mean Perseveration Error rates increased to 135% ( p ⫽ 0.026) (Fig 4A). Injections of A␤-free CHO⫺ CM did not increase either type of error, as described previously.9 To determine whether AZD-103 could mitigate the error increases caused by A␤ oligomers, we incubated either 7PA2 CM or CHO⫺ CM with 5.0M AZD-103 at 37°C for 2 hours before ICV injection. Preincubation of 7PA2 CM with AZD-103 completely rescued both types of ALCR errors from the detrimental effect of the ICVadministered oligomeric A␤ (see Fig 4A). AZD-103 incubated with CHO⫺ CM had no effect on error rates (see Fig 4A). Compounds that can be administered orally and readily penetrate to brain sites of action offer significant clinical advantages. To assess the effectiveness of oral AZD-103 administration in preventing cognitive deficits produced by soluble A␤ oligomers, we administered AZD-103 to rats in their drinking water. Three ascending doses of 30, 100, and 300mg/kg/day were given for at least 3 days at each dose level (5 days for the initial dose of 30mg/kg/day). To achieve these daily oral doses, we adjusted concentrations of AZD103 based on average daily drinking water consumption. Rats received sham or 0.9% saline ICV injections and were tested daily under ALCR during the entire oral dosing regimen. After at least 3 days on each AZD-103 oral dose, we administered 7PA2 CM ICV as described earlier and assessed the rats’ performance Fig 4. ICV or oral AZD-103 returned 7PA2 conditioned media (CM)–induced error increases to baseline levels. (A) AZD-103 (5.0 M) was incubated with 7PA2 CM for 2 hours before intracerebroventricular (ICV) injection. When compared with the mean error rate from the three previous sessions, 7PA2 CM alone resulted in significantly increased Switching/Approach Errors (blue bars; 120%, corrected p ⫽ 0.044) and Perseveration Errors (red bars; 135%, corrected p ⫽ 0.026). After AZD-103 pre-incubation with 7PA2 CM, error rates returned to baseline levels (dotted lines). AZD incubated with CHO⫺ CM, (containing no A␤ oligomers), did not significantly affect either type of error. (B) When AZD103 was given in drinking water at doses of 30, 100, or 300 mg/kg/day, Switching Errors were reduced from 130% of baseline after ICV 7PA2 CM alone (p ⫽ 0.01) to levels (112, 100, and 113%, respectively) that were not significantly different than baseline (dotted lines). Perseveration Errors under oral AZD were reduced from 169% of baseline error rate under 7PA2 CM alone to error rates not different from baseline levels (124, 120, and 102% at the same respective doses). under ALCR. Thereafter, the effects of 7PA2 CM were reassessed under ALCR after a 5-day washout period during which normal tap water was reinstated for drinking. As expected, A␤ oligomers in 7PA2 CM significantly increased Switching Errors to 130% of the baseline error rate ( p ⫽ 0.01, Bonferroni t test) and increased Perseveration Errors to 169% of baseline ( p ⫽ 0.004) (see Fig 4B). However, under all three oral doses of AZD-103, both types of errors were restored essentially to baseline levels (see Fig 4B). No dose of Townsend et al: A␤ Neutralized by AZD-103 673 oral AZD-103 administered alone (without 7PA2 CM) had a significant effect on Switching Errors or Perseveration Errors (data not shown). Importantly, no AZD-103 dose either affected how quickly rats completed the required cycles or reduced the number of food reinforcers they received (data not shown). When 7PA2 CM was injected ICV into the same rats after the 5-day washout period (ie, on normal drinking water), the rats again showed significantly elevated error rates under the ALCR test (data not shown). CSF collected from the fourth ventricle of age-matched rats treated exactly as described earlier under the oral dosing regimen showed mean AZD-103 CSF levels of 3.86, 6.32, and 10.08g/ml (21.4, 35.1, and 56.0M) after the respective oral doses of 30, 100 and 300mg/ kg/day. Histological examination of the rat brain hemispheres contralateral to the cannula placement was entirely normal after either oral or ICV administration of AZD-103 (data not shown). Discussion Here, we report that a small (molecular weight, 180) organic molecule previously found to alter the conformation of synthetic A␤42 and inhibit its aggregation into neurotoxic assemblies10 –12 can penetrate the brain after oral administration and can fully prevent the adverse effects of secreted oligomers of human A␤ on both cognitive performance and hippocampal LTP. AZD-103 significantly improved LTP at lowmicromolar to submicromolar concentrations when added to 7PA2 CM containing pre-existing A␤ oligomers (postconditioning) and proved even more effective when incubated with the 7PA2 cells during the period of oligomer formation and secretion (preconditioning). The molecular specificity of its effects is supported by the finding that the epi-enantiomer of AZD-103 was almost as potent as AZD-103 itself, whereas the chiro-enantiomer was without activity in the LTP assay. That AZD-103 specifically neutralizes low-n A␤ oligomers was established using SEC to separate these oligomers from monomers and the much larger soluble fragments of secreted APPs. We found that AZD-103 coupled to epoxy-agarose can pull down A␤ trimers, which we have shown to inhibit LTP. We therefore hypothesize that AZD-103 interacts directly with A␤ to prevent its subsequent binding to neuronal target molecules in the hippocampus that mediate its toxicity. Far larger quantities of purified cell-derived A␤ oligomers would be required to test whether AZD-103 alters the conformation of A␤ oligomers using biophysical methods such as circular dichroism or nuclear magnetic resonance, which is not currently feasible. We have previously reported that learned behavior and reference memory under the ALCR test are adversely affected by physiologically relevant levels of sol- 674 Annals of Neurology Vol 60 No 6 December 2006 uble A␤ oligomers that are present in 7PA2 CM. We now establish that when AZD-103 and 7PA2 CM are incubated together before ICV injection, A␤ oligomer– induced errors in behavioral performance and reference memory are entirely prevented. These cognitive results mirror the ability of AZD-103 at similar concentrations to block the synaptic effects of A␤ oligomers on hippocampal LTP. Oral administration of AZD-103 in the drinking water at dose levels that could readily be administered to humans blocked A␤-induced cognitive errors at all doses tested. Washout of the AZD-103 reversed the protection and allowed A␤-induced errors to recur. AZD-103 levels found in CSF from the fourth ventricle of treated rats varied directly with the dose of orally administered AZD-103, demonstrating the absorptive bioavailability and brain penetration of the compound. At even the lowest dose tested (30mg/kg/day), there was an essentially complete rescue of the oligomerinduced deficits in cognitive performance. Although a daily dose of 30mg/kg/day was sufficient to block the adverse behavioral effects of A␤ oligomers, appreciably greater doses of AZD-103 (300mg/kg/day) had no discernible adverse effects on behavior under ALCR or any clinical or neuropathological effects throughout the study. Consistent with these observations, AZD-103 by itself was not found to have any effects on hippocampal synaptic function. These observations suggest that the safe dose range of AZD-103 is significantly greater than its effective dose. After the completion of our studies, McLaurin and colleagues12 recently reported beneficial effects of sustained oral administration of AZD-103 on cerebral A␤ burden, A␤-associated cytopathology, and Morris water maze performance in APP transgenic mice. Our study corroborates and extends these findings and provides further insights into mechanism. First, we show that AZD-103 does not depolymerize synaptotoxic low-n oligomers A␤, but rather appears to bind and neutralize them, fully preventing their biological effects on hippocampal LTP. Second, our experimental approach allows one to specifically ascribe the rescue of both hippocampal LTP and the memory of a novel learned behavior distinct from the water maze to a biochemically defined species of soluble A␤, low n-oligomers, in the absence of monomers, protofibrils, or fibrils. Third, we find that the compound is most potent in neutralizing the synpatotoxicity of low-n oligomers if it is applied to cells during the generation of the oligomers, but it also successfully neutralizes preexisting oligomers. Fourth, we show that AZD-103 does not prevent the loss of synaptic plasticity if it is preapplied to hippocampal slices before the addition of oligomers, suggesting that the compound needs to bind A␤ before A␤ encounters its neuronal targets. Fifth, we show that the compound has acute, rapid effects on hippocampal synapses and can provide short-term relief of oligomer-compromised memory in vivo, in addition to the effects seen with chronic administration by McLaurin and colleagues.12 Sixth, we establish a dosedependent increase in brain levels of AZD-103 after oral administration and rapid reversibility of benefit after cessation of dosing. Seventh, we confirm the lack of any detectable neurotoxicity in vivo. Eighth, although the compound can decrease high-molecular-weight oligomers in brain and enhance trimer levels,12 our data indicate that although AZD-103 does not further depolymerize trimers and related low-n oligomers, it prevents their synaptotoxic effects on LTP and on at least one form of cognition. In conclusion, we report that scyllo-cyclohexanehexol (AZD-103) has many of the properties sought in a disease-modifying therapeutic for AD. It is a small, orally bioavailable compound with enantiomeric specificity. It addresses a well-documented target, the soluble assemblies of secreted A␤ that have been extensively confirmed as interfering with hippocampal synaptic function in a variety of animal models of AD (reviewed in Walsh and Selkoe17). Although side effects unrelated to its mechanism of action cannot be excluded, initial in vivo toxicity data appear benign. Moreover, the compound is soluble, can be readily administered orally, and penetrates into the brain in quantities sufficient to prevent cognitive deficits produced by levels of A␤ similar to those that occur in human CSF. Taken together, our findings provide evidence that AZD-103 neutralizes the neurotoxic activity of soluble A␤ oligomers. Experimental Procedures Cyclohexanehexols were provided by Transition Therapeutics (Toronto, Ontario, Canada). The structures of these molecules are provided in Supplemental Figure 3A. CYCLOHEXANEHEXOLS. 7PA2 cell CM113 was collected and prepared as described previously.5 A〉 PREPARATION. ANTIBODIES. Antibodies directed against A␤ included monoclonals 6E10 and 4G8 (Signet), 2G3, 21F12, m266, and 3D6 (Elan) and the rabbit polyclonal R1282. Supplemental Figure 1D details the regions of A␤ that are recognized by these antibodies. IMMUNOPRECIPITATION. Precipitation of A␤ from 8ml CM prepared as described earlier was performed as documented previously.5 Immunoprecipitations with epoxy-agarose were performed using a protocol published by Pharmacia (Gaithersburg, MD). In brief, 5M AZD-103 (or its chiro-enantiomer) were coupled to 0.25gm rehydrated epoxy-agarose resin in 50mM phosphate buffer, pH 9.0. Excess ligand was sequentially washed off with water, 0.1M bicarbonate, pH 8.0, and 0.1M acetate buffer, pH 4.0. The beads were then blocked with 1M ethanolamine; washed with 0.1M acetate, 0.5M NaCl, and 0.1M borate; and stored in 0.5 M NaCl. To reduce nonspecific binding in subsequent experiments, we also blocked beads with 0.5% bovine serum albumin (see Supplemental Fig 1). SIZE EXCLUSION CHROMATOGRAPHY. SEC was performed as Walsh and colleagues5 described. Samples were electrophoresed on 10 to 20% Tricine gels (Invitrogen, La Jolla, CA, or BioRad, Richmond, CA), and the proteins transferred to 0.2m Optitran nitrocellulose. The membranes were boiled in phosphate-buffered saline (to enhance the exposure of A␤ epitopes) and blocked for 1 hour in 50% Odyssey blocking buffer diluted in phosphate-buffered saline. Blots were probed with the monoclonal antibody 6E10. Immunoreactive bands were detected and quantified using a Licor Odyssey imaging system. WESTERN BLOTS. Field potential recordings were made from coronal sections of postnatal days 16 to 28 male and female Swiss Webster mice, as described previously,13 in compliance with Harvard University’s Animal Resources and Comparative Medicine policies for use of laboratory animals. Field potential recordings were made at room temperature. Electrodes were specifically placed just below the surface of the slice to maximize the exposure to circulating A␤. The intensity of the stimulus was set to 20 to 30% of the maximum evoked excitatory postsynaptic potential or until a population spike was elicited. Slices were perfused for 20 minutes in ACSF to establish a steady baseline. During this interval, 1ml 15⫻ concentrated CM was thawed, mixed with AZD-103 (for postconditioning experiments), and incubated at 37°C for 15 minutes. This mixture was then added to 14ml circulating media, giving a final concentration of 1.25M AZD-103, unless otherwise indicated. This 1⫻ CM/ACSF/AZD103 solution was recirculated over the slice at 2.5 to 3ml/min whereas being continuously aerated with 95% oxygen. LTP was induced 20 minutes later by delivering four 100Hz stimuli every 5 minutes. The slope of the EPSP was monitored for 1 hour after the last highfrequency stimulation. ELECTROPHYSIOLOGY. BEHAVIORAL TESTING, SURGERY, AND DRUG ADMINISTRATION. The ALCR procedure was used to assess cog- nitive performance and reference memory. This task involves progressively increasing and decreasing response requirements as described in Results. The two types of errors generated under this procedure have proved sensitive to ICV injections of A␤ oligomers.9 Townsend et al: A␤ Neutralized by AZD-103 675 Under general anesthesia, a single permanent indwelling cannula was aimed at either the right or left lateral ventricle and permanently affixed to the skull. Rats were allowed to recover for 5 days, and then stable response and error rates on ALCR were reestablished. Cannula placement was confirmed by histological examination of the brain after the last experimental session. Behavioral testing begun 2 hours after ICV injection of A␤ oligomers. All ICV injections were 20l and were slowly administered over a 5-minute period in freely moving animals. For ex vivo assessments, AZD103, at a concentration of 5.0M, was pre-incubated with either 7PA2 CM or CHO⫺ CM for 2 hours at 37°C. For oral dosing, AZD-103 was mixed in the rats’ drinking water and was available ad libitum for at least 3 days before behavioral testing after ICV A␤ oligomer administration. Daily doses were determined by adjusting the concentration of AZD-103 in drinking water based on each rat’s average daily water consumption (minus spillage) to achieve doses of 30, 100, and 300mg/kg/day AZD-103. Doses of oral AZD-103 were studied in an ascending order over sequential independent experiments. Typical comparisons were done using Student’s t test or analysis of variance with Tukey– Kramer post hoc tests. For the behavioral studies, analysis was done using a Bonferroni test. Error bars indicate the standard error of the mean. STATISTICS. This research was supported by the NIH (AG027443, D.J.S.; T32 NS07484, M.T.). We thank J. McLaurin and P. St George-Hyslop for advice on this project. J. Copeman at Transition Therapeutics provided us with inositol reagents and contributed to the critical review of the manuscript. Transition Therapeutics also contributed small amounts of supplemental research funds to both J.P.C.’s and D.J.S.’s laboratories to defray some of the costs of these experiments, as did K. Ashe. S. Narayanan and M. LaVoie made valuable contributions to experimental design. References 1. Selkoe DJ. Alzheimer’s disease is a synaptic failure. Science 2002;298:789 –791. 2. Podlisny MB, Ostaszewski BL, Squazzo SL, et al. Aggregation of secreted amyloid beta-protein into sodium dodecyl sulfatestable oligomers in cell culture. J Biol Chem 1995;270: 9564 –9570. 676 Annals of Neurology Vol 60 No 6 December 2006 3. Kawarabayashi T, Shoji M, Younkin LH, et al. Dimeric amyloid beta protein rapidly accumulates in lipid rafts followed by apolipoprotein E and phosphorylated tau accumulation in the Tg2576 mouse model of Alzheimer’s disease. J Neurosci 2004; 24:3801–3809. 4. Walsh DM, Tseng BP, Rydel RE, et al. The oligomerization of amyloid beta-protein begins intracellularly in cells derived from human brain. Biochemistry 2000;39:10831–10839. 5. Walsh DM, Townsend M, Podlisny MB, et al. Certain inhibitors of synthetic amyloid beta-peptide (Abeta) fibrillogenesis block oligomerization of natural Abeta and thereby rescue longterm potentiation. J Neurosci 2005;25:2455–2462. 6. Podlisny MB, Walsh DM, Amarante P, et al. Oligomerization of endogenous and synthetic amyloid beta-protein at nanomolar levels in cell culture and stabilization of monomer by Congo red. Biochemistry 1998;37:3602–3611. 7. Walsh DM, Klyubin I, Fadeeva JV, et al. Naturally secreted oligomers of amyloid beta protein potently inhibit hippocampal long-term potentiation in vivo. Nature 2002;416:535–539. 8. Klyubin I, Walsh DM, Lemere CA, et al. Amyloid beta protein immunotherapy neutralizes Abeta oligomers that disrupt synaptic plasticity in vivo. Nat Med 2005;11:556 –561. 9. Cleary JP, Walsh DM, Hofmeister JJ, et al. Natural oligomers of the amyloid-beta protein specifically disrupt cognitive function. Nat Neurosci 2005;8:79 – 84. 10. McLaurin J, Franklin T, Chakrabartty A, Fraser PE. Phosphatidylinositol and inositol involvement in Alzheimer amyloidbeta fibril growth and arrest. J Mol Biol 1998;278:183–194. 11. McLaurin J, Golomb R, Jurewicz A, et al. Inositol stereoisomers stabilize an oligomeric aggregate of Alzheimer amyloid beta peptide and inhibit abeta-induced toxicity. J Biol Chem 2000;275:18495–18502. 12. McLaurin J, Kierstead ME, Brown ME, et al. Cyclohexanehexol inhibitors of Abeta aggregation prevent and reverse Alzheimer phenotype in a mouse model. Nat Med 2006;11:11. 13. Townsend M, Shankar GM, Mehta T, et al. Effects of secreted oligomers of amyloid beta-protein on hippocampal synaptic plasticity: a potent role for trimers. J Physiol 2006;572: 477– 492. 14. Fazldeen H, Breitenbach M. A new sepharose derivative containing covalently bound myo-inositol: its structure and application. Monatshefte fur Chemie 1979;110:823– 830. 15. O’Hare E, Levine AS, Semotuk MT, et al. Utilization of a novel model of food reinforced behavior involving neuropeptide Y, insulin, 2-deoxy-d-glucose and naloxone. Behav Pharmacol 1996;7:742–753. 16. Richardson RL, Kim EM, Shephard RA, et al. Behavioural and histopathological analyses of ibuprofen treatment on the effect of aggregated Abeta(1-42) injections in the rat. Brain Res 2002; 954:1–10. 17. Walsh DM, Selkoe DJ. Deciphering the molecular basis of memory failure in Alzheimer’s disease. Neuron 2004;44: 181–193.