Inositol Stereoisomers Stabilize an Oligomeric Aggregate of Alzheimer Amyloid β Peptide and Inhibit Aβ-induced Toxicity

Inositol has 8 stereoisomers, four of which are physiologically active. myo-Inositol is the most abundant isomer in the brain and more recently shown that epi- andscyllo-inositol are also present. myo-Inositol complexes with Aβ42 in vitro to form a small stable micelle. The ability of inositol stereoisomers to interact with and stabilize small Aβ complexes was addressed. Circular dichroism spectroscopy demonstrated that epi- and scyllo- but not chiro-inositol were able to induce a structural transition from random to β-structure in Aβ42. Alternatively, none of the stereoisomers were able to induce a structural transition in Aβ40. Electron microscopy demonstrated that inositol stabilizes small aggregates of Aβ42. We demonstrate that inositol-Aβ interactions result in a complex that is non-toxic to nerve growth factor-differentiated PC-12 cells and primary human neuronal cultures. The attenuation of toxicity is the result of Aβ-inositol interaction, as inositol uptake inhibitors had no effect on neuronal survival. The use of inositol stereoisomers allowed us to elucidate an important structure-activity relationship between Aβ and inositol. Inositol stereoisomers are naturally occurring molecules that readily cross the blood-brain barrier and may represent a viable treatment for AD through the complexation of Aβ and attenuation of Aβ neurotoxic effects.

Inositol has 8 stereoisomers, four of which are physiologically active. myo-Inositol is the most abundant isomer in the brain and more recently shown that epi-and scyllo-inositol are also present. myo-Inositol complexes with A␤42 in vitro to form a small stable micelle. The ability of inositol stereoisomers to interact with and stabilize small A␤ complexes was addressed. Circular dichroism spectroscopy demonstrated that epi-and scyllo-but not chiro-inositol were able to induce a structural transition from random to ␤-structure in A␤42. Alternatively, none of the stereoisomers were able to induce a structural transition in A␤40. Electron microscopy demonstrated that inositol stabilizes small aggregates of A␤42. We demonstrate that inositol-A␤ interactions result in a complex that is non-toxic to nerve growth factor-differentiated PC-12 cells and primary human neuronal cultures. The attenuation of toxicity is the result of A␤-inositol interaction, as inositol uptake inhibitors had no effect on neuronal survival. The use of inositol stereoisomers allowed us to elucidate an important structure-activity relationship between A␤ and inositol. Inositol stereoisomers are naturally occurring molecules that readily cross the blood-brain barrier and may represent a viable treatment for AD through the complexation of A␤ and attenuation of A␤ neurotoxic effects.
Alzheimer's disease is characterized neuropathologically by amyloid deposits, neurofibrillary tangles, and selective neuronal loss. The major component of the amyloid deposits is amyloid-␤ (A␤), a 39 -43 residue peptide. Soluble forms of A␤ generated from cleavage of amyloid precursor protein are normal products of metabolism (1,2). The importance of residues 1-42 (A␤42) in Alzheimer's disease was highlighted in the discovery that mutations in codon 717 of the amyloid precursor protein gene, presenilin 1 and presenilin 2 genes result in an increased production of A␤42 over A␤1-40 (A␤40; Refs. [3][4][5]. These results in conjunction with the presence of A␤42 in both mature plaques and diffuse amyloid (6) lead to the hypothesis that this more amyloidogenic species may be the critical element in plaque formation. This hypothesis was supported by the fact that A␤42 deposition precedes that of A␤40 in Down's syndrome (7,8), in PS1 mutations (9) and in hereditary cerebral hemorrhage with amyloidosis (10).
Many in vitro studies have demonstrated that A␤ can be neurotoxic or enhance the susceptibility of neurons to excitotoxic, metabolic, or oxidative insults (11)(12)(13)(14). Initially it was thought that only the fibrillar form of A␤ was toxic to neurons (15)(16)(17)(18) but more thorough characterization of A␤ structures demonstrated that dimers and small aggregates of A␤ are also neurotoxic (19,20). These data suggested that prevention of A␤ oligomerization would be a likely strategy to prevent AD-related neurodegeneration. Several studies have demonstrated that in vitro A␤-induced neurotoxicity can be ablated by compounds that can increase neuronal resistance by targeting cellular pathways involved in apoptosis (21), block downstream pathways after A␤ induction of destructive routes (14,22,23), or compounds that block A␤ oligomerization and ultimately fibril formation (16, 24 -26). The site at which A␤ acts to induce neurotoxicity has yet to be elucidated but its toxic effects have been blocked by a variety of disparate agents.
Docking of A␤-fibrils to neuronal and glial cell membranes may be an early and intervenable step during the progression of AD. 1 Formation of amyloid plaques, as well as, neurotoxicity and inflammation may be direct or indirect consequences of the interaction of A␤ with molecules containing sugar moieties. Previous studies have demonstrated that A␤ interaction with glycosaminoglycans results in aggregation of A␤ possibly adding to their insolubility and plaque persistence (27)(28)(29). Glycosaminoglycans have also been implicated in neuronal toxicity (30) and microglial activation (31,32). Alternatively, interaction with glycolipids such as gangliosides results in the stabilization and prevention of A␤ fibril formation, as well as, the site of A␤ production (33)(34)(35)(36)(37). The family of phosphatidylinositols, on the other hand, results in acceleration of fibril formation (38). The headgroup of phosphatidylinositol is myo-inositol a naturally occurring simple sugar involved in lipid biosynthesis, signal transduction, and osmolarity control.
We have demonstrated that myo-inositol stabilizes a small micelle of A␤42 (38). The interaction of A␤ with small sulfated compounds, antibiotics, and glycosaminoglycans has been shown to vary as the charge distribution across the compound is varied (12,38,39). myo-Inositol has 8 stereoisomers that alter the distribution of hydroxyl groups across the surfaces of the sugar ring. In the present study, we examined the ability of four inositol isomers ( Fig. 1) to stabilize small aggregates of A␤40 and A␤42. The resultant A␤-inositol complexes were subsequently examined for their ability to modulate A␤-induced toxicity of nerve growth factor (NGF)-differentiated PC-12 cells and primary human fetal neuronal cultures.

MATERIALS AND METHODS
Inositol stereoisomers: myo-, epi-, and scyllo-inositol were purchased from Sigma, chiro-inositol from Wako Chemicals (Richmond, VA). PC-12 cells were from ATCC. NGF was purchased from Alamone Laboratories (Israel). Competitive inhibitors used in this study, phloridzin and D-glucose, were purchased from Sigma.

A␤ Peptides
A␤40 and A␤42 were synthesized by solid phase Fmoc chemistry by the Hospital for Sick Children's Biotechnology Center (Toronto, Ontario). Peptides were purified by reverse phase high performance liquid chromatography on a C18 Bondapak column. Peptides were initially dissolved in 0.5 ml of 100% trifluoroacetic acid (Aldrich, Milwakee, WI), diluted in distilled H 2 O and immediately lyophilized. Peptides were then dissolved in 40% trifluoroethanol (Aldrich) in H 2 O and stored at Ϫ20°C until use. Alternatively, the lyophilized peptides were dissolved in distilled H 2 O at 10 mg/ml concentration and used immediately.

Circular Dichroism
CD spectra were recorded on a Jasco Circular Dichroism Spectrometer Model J-715 (Easton, MO) at 25°C. Spectra were obtained from 200 to 260 nm, with a 0.5-nm step, 1-nm band width. Peptide:inositol ratios were varied from 1:1 to 1:20 (w/w) with a final peptide concentration of 10 M. The effects of the inositols on peptide conformation were determined by adding an aliquot of stock peptide solutions to inositol suspended in PBS, 50 mM phosphate buffer or dH 2 O. The contribution of inositols to the CD signal was removed by subtracting the inositol only spectra. A␤ peptide conformations were determined in 40% trifluoroethanol/H 2 O and in buffer under the same conditions.

Electron Microscopy
Peptides were incubated with the inositol stereoisomers at a 1:1 ratio (w/w) in 50 mM phosphate buffer (pH 7.0) or in dH 2 O. For negative staining, carbon-coated pioloform grids were floated on aqueous solutions of peptides (100 g/ml). After grids were blotted and air dried, the samples were stained with 1% (w/v) phosphotungstic acid (pH 7.0). The peptide assemblies were observed in a Hitachi H-7000 operated with an accelerating voltage of 75 kV.

Primary Human Neuronal Cultures
Neural cells are derived from human fetal central nervous system (cerebral hemispheres) tissue obtained at 12-16 weeks gestation as described previously (40). Cultures were obtained using MRC (Canada) approved guidelines. The cultures are prepared by dissociation of the fetal central nervous system tissue with 0.05% trypsin and 50 g/ml DNase, passing the tissue through a 125-m nylon mesh screen, and then through a 70-m screen. After washing with PBS, the cells are suspended in minimal essential medium supplemented with 5% fetal bovine serum, 0.1% glucose, and 1 mM sodium pyruvate and placed onto poly-L-lysine-coated 96-well dishes. Cultures are treated on day 4 with 1 mM 5-fluorodeoxyuridine to deplete astrocytes. The treatment is repeated twice over a 2-week period.

Toxicity Assays
PC-12 cells were plated at 500 cells per well in a 96-well plate and suspended in 30 ng/ml NGF diluted in N 2 /Dulbecco's modified Eagle's medium (Life Technologies, Inc.). Cells were differentiated over 5-7 days to a final cell number of 10,000 -15,000 per well. A␤ was either used directly or aged 3 days at room temperature to induce fibrillogenesis. The A␤ solutions were then incubated with various inositols for 20 -24 h at room temperature. A␤ with and without inositols was added to cultures at a final A␤ concentration of 0.1 g/l and incubated for 24 or 72 h at 37°C. Toxicity was assayed using the sulfhydryl rhodamine B (SRB) assay and the lactate dehydrogenase assay (LDH).
SRB Assay-Cells were fixed with trichloroacetic acid at a final concentration of 10%. Plates were washed with H 2 O and air-dried. Protein was stained with 0.4% SRB (Molecular Probes Inc) in 1% acetic acid for 30 min (41). Plates were washed with 1% acetic acid and air-dried. The dye was extracted in unbuffered 10 mM Tris and absorbance was assayed at 550 nm on a Bio-Rad Benchmark microtiter plate reader.
LDH Assay-Prior to addition of A␤ and inositols, fetal calf serum was added to NGF-differentiated PC-12 cells to a final concentration of 1% in order to stabilize LDH in the supernatant. Supernatants from the A␤-treated cultures were removed and analyzed for LDH release using a commercial kit (Sigma). Results are expressed as B-B units/ml.

Proliferation Assay
The proliferative properties of NGF-differentiated PC-12 cells were determined using a [methyl-3 H]thymidine incorporation assay. Briefly, cells were differentiated with NGF for 5 days at 37°C. In order to determine the basal, A␤ and inositol-induced proliferation 1 mCi of [methyl-3 H]thymidine (NEN Dupont, Mississauga, ON) was added to each well and incubated for 18 h. Cells were then harvested onto glass fiber filters and radioactivity determined as counts/min per well by liquid scintillation counting on a Beckman ␤-counter.

Inositol Inhibitor Studies
NGF-differentiated PC-12 cells were cultured in glucose-free media for inositol competition assays. 1 mM D-Glucose was added to PC-12 cells immediately prior to the addition of A␤/inositol mixtures. Similarly, 100 M phloridzin was added to NGF-differentiated PC-12 cells in the presence and absence of A␤/inositol mixtures. Cells were then incubated for 24 h before toxicity was measured using both the SRB and LDH assays.

Immunofluorescence Studies
PC-12 cells were plated onto poly-L-lysine-coated glass coverslips at 1000 cells per slip and differentiated in 30 ng/ml NGF in N 2 /Dulbecco's modified Eagle's medium for 5 days. The presence of A␤ on the cell surface of NGF-differentiated PC-12 cells was examined between 30 min and 3 h of incubation in the presence of A␤ with and without inositols. A␤ was visualized using A␤-specific antibodies, 6E10 and 4G8 (Senetek, St. Louis, MA) followed by goat anti-mouse Ig conjugated to Cy3 (Dako, Cerpinteria, CA). Cells were post-fixed in 2% paraformaldehyde prior to fluorescence visualization.

Structural Characteristics of A␤-Inositol
Complexes-In order to determine the interaction of the inositol stereoisomers ( Fig. 1) with A␤, we investigated the effect of these compounds in the random coil to ␤-structure transition necessary for A␤micelle and fibril formation. Examination of A␤ structure immediately upon incubation with inositol will lend evidence for the effect of inositols on fibril nucleation. Previously, we have shown that A␤42-myo-inositol interactions result in an immediate conformational change from random coil to ␤-sheet, whereas, A␤40 does not undergo this transition (38). The structural details of A␤40 and A␤42-inositol isomer interactions were investigated by CD spectroscopy (Fig. 2). A␤40 and A␤42 stored in 40% trifluoroethanol displayed CD spectra indicative of partially ␣-helical structures as previously reported (34,42). At a concentration of 10 M, both A␤40 and A␤42 became unstructured after dilution in PBS (pH 7.0) (Fig. 2, A and D).
When incubated in the presence of epi-, scyllo-, and chiroinositol, A␤40 remained random as seen with myo-inositol ( Fig.  2; Ref. 38). Variation of the charge density on the inositol ring did not induce a conformational switch in A␤40. We have previously shown that A␤40-GAG interactions are inhibited in buffers containing phosphate counterions (29). Therefore, we examined the interactions of A␤40 with inositol in dH 2 O (pH 6.8). None of the stereoisomers were able to induce a conformational transition in A␤40, suggesting that charge shielding was not the limiting factor in A␤40-inositol interactions. Increasing the concentration of myo-inositol to 1 M induced a ␤-structural transition in A␤40. At this concentration, myoinositol functions as a chemical chaperone to stabilize the peptide through osmotic remediation and not peptide binding (42,43).
In contrast, A␤42 was immediately induced to form ␤-structure in the presence of both epi-inositol (Fig. 2D) and scylloinositol ( Fig. 2E) but not in the presence of chiro-inositol at peptide:inositol ratio of 1:20 (w/w) (Fig. 2F). The stereoisomers differ from myo-inositol in the number of hydroxyl groups extending on either side of the inositol ring (Fig. 1). myo-Inositol has four hydroxyl groups on one surface of the ring and two hydroxyl groups on the other. If we consider only the more highly charged surface of the ring, then epi-inositol increases the surface charge by one hydroxyl group, whereas, scylloinositol decreases the hydroxyl groups from 4 to 3 (Fig. 1). chiro-Inositol also decreases the number of hydroxyl groups to 3 but unlike scyllo-inositol, the spacing of the charge density is uneven (Fig. 1). The structure-activity relationship between inositol stereoisomers and A␤ elucidate the necessity for hydroxyl groups with the same orientation at either positions 1, 3, and 5 or 2, 4, and 6 of the inositol ring to stabilize the A␤42inositol complex. The concentration dependence of A␤42-inositol structural transition was examined to determine the stoichiometry of this interaction. Both epi-and scyllo-inositol were able to induce a transition from random to ␤-structure in A␤42 at a 1:1 ratio (by weight). This corresponds to one molecule of A␤42 to 25 molecules of inositol. Although inositol is in excess amounts with respect to A␤42, this is not disparate with inositol concentrations in the central nervous system of young adults.
Effect of Inositols on A␤ Fibril Structure-Our CD studies show that inositols induce the structural transition necessary for fibrillogenesis but this may not correlate with increased fibril growth. The characteristics of A␤40 and A␤42 fibrils in the presence and absence of inositol stereoisomers were examined by electron microscopy. Unseeded samples of both A␤40 and A␤42 were incubated in the presence of epi-, scyllo-, chiroinositol, and alone for 96 h. Negative stain electron microscopy demonstrated that when A␤40 was incubated in buffer alone, it formed fibrils of varying lengths with some apparent intertwining of fibrils (Fig. 3A). When A␤40 was incubated in the presence of epi-inositol (Fig. 3B), scyllo-inositol, or chiro-inositol the fibrils formed were indistinguishable from those of A␤40 alone.
Negative stain electron microscopy analysis of A␤42 demonstrated that when A␤42 is incubated in buffer, fibrils were of varying lengths (Fig. 4A). In the presence of chiro-inositol, the A␤42 fibers were indistinguishable from A␤42 alone (Fig. 4B). In contrast, no fibrils could be detected in the presence of epi-inositol (Fig. 4C) and scyllo-inositol (Fig. 4D) demonstrating an activity similar to myo-inositol (38). The fine thread-like structures detected in the A␤42-epi-and scyllo-inositol samples were present in the inositol solutions alone and therefore not A␤ fibrils. These results demonstrate that although epi-and scyllo-inositol are able to induce ␤-structure in A␤42, the A␤42-inositol complex does not progress to form fibrils. Similar to that reported for myo-inositol, A␤42 can form stable ␤-structured, non-fibrillar complexes with epi-and scyllo-inositol.
A␤-induced Neurotoxicity-A␤-induced toxicity of neuronal cell lines and primary cultures is well established (15,17,44). Not only is the A␤-fibril toxic to neuronal populations but smaller A␤ aggregates and dimers also induce toxicity (19,20,45). The ability of inositol stereoisomers to induce ␤-structure and stabilize a small A␤ complex appears to have all the requirements so far described as necessary for A␤-induced toxicity. Light microscopy demonstrated that addition of aggregated A␤40 or A␤42 to NGF-differentiated PC-12 cells resulted in decreased cell number and retraction of neurites (Fig. 5B) as compared with PC-12 cells alone (Fig. 5A). NGF-differentiated PC-12 cells retained their morphology when incubated in the presence of A␤42-myo-inositol complexes (1:20; Fig. 5C). A␤ has been shown to kill neurons through both apoptotic and necrotic pathways, therefore A␤-induced toxicity in the presence of inositol stereoisomers was examined using the SRB and LDH assays. The SRB assay measures total cell death whereas the LDH assay measures cell death that is associated with membrane damage. The SRB assay demonstrated that A␤40 treatment resulted in a 61% cell survival that was not significantly changed by preincubation of A␤40 with myo-inositol (Table I), whereas, A␤42 resulted in a 55% cell survival which increased to 80% by preincubation with myo-inositol (Table I). Finally supernatants from the toxicity assays were assessed for the release of LDH, A␤40-induced LDH release was decreased slightly by preincubation with myo-inositol, whereas A␤42myo-inositol decreased the amount of LDH released to levels of myo-inositol treatment alone (data not shown). These results suggest that incubation of A␤42 with myo-inositol either stabilizes a small non-toxic oligomer, blocks A␤-neuronal cell surface interactions, or alternatively that myo-inositol alone blocks A␤-induced toxicity.
It was surprising to find that when A␤40 was preincubated in the presence of epi-and scyllo-inositol, these mixtures increased the cell survival of PC-12 cells from 56 to 93% and 83%, respectively ( Table I). Preincubation of A␤42 with epi-and scyllo-inositol increased cell survival from 54 to 89% and 83%, respectively. Preincubation of A␤40 with various concentra- . When incubated in the presence of chiro-inositol fibers were indistinguishable from A␤42 alone but were less abundant (B). Similar to that seen with myo-inositol, no fibers could be detected when A␤42 was incubated in the presence of epiinositol (C) and scyllo-inositol (D). Scale represents 50 nm.
tions of epi-and scyllo-inositol resulted in attenuation of toxicity as low as 1:1 A␤:inositol ratio (by weight; Table I). Similar results were seen when A␤42 was preincubated with the stereoisomers (Table I). In contrast, chiro-inositol preincubation did not rescue PC-12 cells from either A␤40-or A␤42-induced toxicity with neuritic dystrophy similar to A␤ treatment alone ( Table I). The ability of epi-and scyllo-inositol to attenuate toxicity when preincubated at low concentrations suggests that the A␤-inositol interaction is specific.
In order to more closely mimic in vivo conditions, we examined the ability of the inositol stereoisomers to protect primary human fetal neuronal cultures from A␤-induced toxicity. Primary human cultures contain a small population of astrocytes and microglia, both of which are present in brain milieu. Both cell types have been proposed to either enhance neuronal toxicity in the presence of A␤ through the production of cytokines and neurotoxins (31,46,47) or attenuate toxicity by removal of A␤ from the extracellular milieu (48 -50). Preincubation of A␤42 but not A␤40 with myo-inositol (Fig. 6, A and B), epiinositol, and scyllo-inositol (data not shown) at a 1:20 ratio (by weight) protected the primary neuronal cultures from death. Attenuation of toxicity was also detected at 72 h after addition of A␤-inositol complexes. LDH assay confirmed the decreased toxicity upon preincubation of A␤42 with inositols (data not shown). Ratios of as low as 1:1 A␤42 to myo-inositol resulted in an increase in the cell survival of primary human neuronal cultures (Fig. 6A). LDH release was decreased in the presence of increasing concentrations of myo-inositol for both PC-12 cells and primary human neuronal cultures (data not shown). The decreased amount of myo-inositol necessary to induce cell survival in the primary cultures may result from the presence of astrocytes and microglia both of which contain receptors for A␤ and inositol removal.
Mechanism of Inositol Attenuation of Neurotoxicity-The CD studies suggest that the mechanism by which the inositol stereoisomers protect neuronal populations from A␤-induced toxicity is through binding of peptide, thereby sequestering it from cellular interactions or by stabilizing a non-toxic oligomer. A  a Paired t test indicates p Ͻ 0.01 when compared to the absence of inositol treatments.
b Paired t-test indicates p Ͻ 0.05 when compared to the absence of inositol treatments. A and B). A␤40 and A␤42 were incubated in increasing ratios of peptide to inositol and the subsequent cell survival was calculated with respect to primary cultures alone. Values are expressed as mean Ϯ S.D. for at least three experiments. Paired t test demonstrated that myo-inositol treatment was significantly different for A␤42 (p Ͻ 0.01) but not for A␤40.

FIG. 6. Concentration dependence of inositol attenuation of A␤-induced toxicity was investigated on primary human neuronal cultures (
contributing factor may be that inositol competes for A␤-binding sites on primary neurons or that inositol stimulates second messenger systems and therefore enhances neuronal survival. Preincubation of NGF-differentiated PC-12 cells with inositols for up to 12 h prior to A␤ treatment was investigated to examine the ability of inositol to compete for A␤-binding sites (Table  II). The pretreatment of PC-12 cells with all four stereoisomers of inositol had no significant effect on PC-12 survival (Table II). A␤40 was added to inositol pretreated PC-12 cells and at higher inositol concentrations a slight attenuation could be detected in comparison to untreated cells. In contrast, when A␤42 was added to myo-, epi-, or scyllo-inositol pretreated PC-12 cells the survival was increased even at the lowest concentrations. Pretreatment of cells with chiro-inositol had no effect on the A␤40 or A␤42-induced toxicity (Table II). These results suggest that inositol treatment attenuates toxicity at least in part by blocking A␤-cell interactions. Proliferation assays in the presence and absence of inositols failed to show any significant difference in the amount of [methyl-3 H]thymidine incorporation over 18 h demonstrating that the PC-12 cells do not de-differentiate (data not shown).
To examine the effect of inositol stereoisomers on A␤-neuronal interactions, we used immunofluorescence localization of A␤. In order to examine cell surface binding in the absence of internalization, A␤ accumulation on the surface of PC-12 cells was examined over 3 h and at 4°C. A␤42 accumulation on the cell body and processes of PC-12 cells was evident at 30 min (Fig. 7A). In contrast, in the presence of myo-inositol (Fig. 7B), epi-, and scyllo-inositol (Fig. 7C), A␤42 accumulation was decreased. No difference could be detected in the amount of A␤42 accumulation in the presence of chiro-inositol, although the immunofluorescence had a more punctate appearance (Fig.  7D). A␤40 accumulated on the cell surface of PC-12 cells when incubated alone (Fig. 7E) which was partially blocked in the presence of epi-and scyllo-inositol (Fig. 7F). These results demonstrate that A␤-inositol interactions decrease the interaction of A␤ with neuronal membranes which may contribute to attenuation of toxicity.
In order to rule out any direct effects of inositol on neuronal cell survival, we included inositol transport inhibitors in our assay system. Inositol is taken up by cells through both passive diffusion and active transport (51,52). Prior to inhibition assays, PC-12 cells were incubated in glucose-free media. Previous studies have demonstrated that 1 mM D-glucose inhibits passive diffusion of inositol into cells (53). The addition of 1 mM D-glucose to cells resulted in enhanced survival of PC-12 cells, and in turn resulted in a slightly decreased A␤-induced toxicity (Table III). D-Glucose was not able to compete with myo-, epi-, and scyllo-inositol attenuation of toxicity, suggesting that inositol alone does not contribute to attenuation of toxicity. Active transport of inositol has been shown to be a Na ϩ -dependent, stereoisomer-specific, saturable mechanism that is active in the brain and at the blood-brain barrier (51,54). A known inhibitor of active uptake is phloridzin (52,53). Phloridzin had no effect on the cell survival of PC-12 cells that were treated with A␤40/42 in the presence or absence of all inositol stereoisomers in the range of 0 -100 M (Table III). These results suggest that inositol uptake through the Na ϩ -dependent transporter does not contribute to the attenuation of A␤ toxicity.

DISCUSSION
Factors that alter amyloid aggregation or fibril formation may contribute to AD pathology. Molecules associated with neuritic plaques have been proposed to either enhance or decrease both plaque formation and neuronal loss. The ability of inositol stereoisomers to induce non-fibrillar ␤-structure in A␤42 is a striking phenomena. It has been proposed that molecules with the appropriate pattern of polar and non-polar surfaces, including hydrogen donors and acceptors, may interact with A␤ to form either a template for fibril growth or for inhibition (55). Inositol stereoisomers vary the charge distribution across the surfaces of the sugar ring, in effect varying the pattern of available hydrogen donors or acceptors, which may explain the differences in myo-, epi-, scyllo-, and chiro-inositol's ability to inhibit A␤42 fibrillogenesis. Many molecules that bind A␤ in vitro and in vivo have been tested for their ability to effect fibrillogenesis. Congo red has been shown to have variable ability to inhibit A␤40 and A␤42 fibrillogenesis with inhibition seen for A␤40 only (16). Alternatively, A␤ interaction with ApoJ results in the formation of slowly sedimenting oligomers of A␤42 but not A␤40 (45). These results demonstrate some inherent differences in the interaction with A␤40 and  A␤42, which results in variations in the formation of aggregates and fibrils.
Clusterin, ␣ 2 -macroglobulin, and glycosaminoglycans have all been shown to attenuate A␤-induced toxicity presumably by binding A␤ and thereby preventing interaction with the cell (24 -26). The interaction of A␤ with inositol stereoisomers is reminiscent of these molecules in that inositol prevents A␤ interactions with the cell membrane. It was previously demonstrated that A␤ dimers are only neurotoxic in the presence of microglial cells (19) and that soluble oligomers of A␤-clusterin are also toxic to neurons but are sufficient on their own (45). Our results suggested the formation of a small complex of A␤-inositol that was non-toxic in both clonal cell lines and mixed human cultures. This suggested that this small complex was unable to induce activation of microglia and subsequent loss of neurons as previously reported (19,31). The interaction of A␤ with inositol may allow for more efficient clearance of the complex than A␤ oligomers or fibrils alone.
Inositol has been shown to be dysregulated in both AD and Down's syndrome. The uptake of myo-inositol was shown to be enhanced in Down's syndrome fibroblasts (56) and Trisomy 16 mice (57). These results were later shown to be the result of increased number of myo-inositol transporter, which is present on human chromosome 21 and mouse chromosome 16. It is also of interest that large amounts of A␤42 are present in Down's syndrome central nervous system prior to the deposition of plaques (7,8). It would be interesting to postulate that the presence of high cerebral myo-inositol in young Down's syndrome patients without dementia (58) and the ability to tolerate an increased A␤42 load might be due to A␤-inositol interactions. The stability of a non-toxic A␤42 complex would allow the high A␤ load without detrimental effects. In AD, it is well established that phosphoinositide levels are reduced (59) thereby effecting signal transduction. It is unclear whether inositol levels are increased or decreased. Our data suggest that inositol treatment for AD patients may help to prevent A␤-deposition and A␤-induced toxicity. The use of inositol stereoisomers may represent a therapeutic benefit over myo-inositol, since these isomers are present in very low concentrations in the brain, are incorporated poorly into phosphoinositides but have similar mechanisms of uptake (54,60).

TABLE III
Inhibitor studies NGF-differentiated PC-12 cells were incubated in the presence of A␤-inositol complexes and either diffusion or active transport inhibitors. Diffusion inhibitor cell survival was calculated with respect to PC-12 cells in the presence of the D-glucose and the absence of A␤inositol complexes. Active transport inhibition was calculated with respect to cell survival in the presence of A␤-inositol complexes but the absence of phloridzin. Cell survival was determined using the SRB assay. Data is reported as the mean Ϯ S.D. of three independent experiments. % cell survival