Cholesterol is an Important Factor Affecting the Membrane Insertion of β -Amyloid Peptide (A β 1-40) Which May Potentially Inhibit the Fibril Formation

β and neuronal Here we show that A β is able to insert into lipid bilayer. The membrane insertion ability of A β is critically controlled by the ratio of cholesterol to phospholipids. In low concentration of cholesterol A β prefers to stay in membrane-surface region mainly in β -sheet structure. In contrast, as the ratio of cholesterol to phospholipids is above 30mol%, A β can spontaneously insert into lipid bilayer by its C-terminal. During membrane insertion A β generates about 60% α -helix and removes almost all β -sheet structure. The fibril formation experiments show that such membrane insertion can reduce fibril formation. Our findings reveal a possible pathway that A β prevents itself from aggregation and fibril formation by membrane insertion.


Introduction
The formation of extracellular amyloid plaques is one of the characteristics of AD. The core component of plaques is Aβ, which is the proteolytic product of the larger transmembrane amyloid precursor protein (APP) (1,2). Aβ contains 39 to 42 amino acid residues with a molecular mass of approximate 4KDa. Aβ is an amphiphilic peptide with a hydrophilic N-terminal domain (residues 1 to 28) and a hydrophobic C-terminal (residues 29 to 40 (42)), the latter corresponding to a part of the transmembrane domain of APP.
Aβ is a normal constituent of neuronal and non-neuronal cells (3,4). It can be detected in cerebrospinal fluid (CSF) at subnanomolar concentrations in normal individuals. Such a concentration of Aβ has its own physiological functions, for example, increasing tyrosine phosphorylation, increasing the activity of phosphoinositol-3-kinase, inducing the rapid change of cellular calcium and extracellular protein kinase C etc. (5). It was reported that in cultured hippocampal neurons, Aβ in low concentration (10 -11~1 0 -10 M) is neurotrophic to undifferentiated, immature hippocampal neurons (6,7). As a proteolytic fragment of APP, Aβ can be secreted by membrane-anchored APP or by reinternalized APP (8)(9)(10). Also, Aβ can be degraded either via low-density lipoprotein receptor-related protein (LRP) mediated endocytosis into primary neurons and astrocytes (11)(12)(13)(14)(15) or via scavenger receptor mediated uptake of aggregates of Aβ into microglial cells (16). The hydrolytic enzymes in lysosomes then can degrade Aβ. Thus, whether Aβ is the primary effector of the disease is questioned.
Several studies on conformation show that Aβ in the core of amyloid plaques adopts an antiparallel by guest on March 21, 2020 http://www.jbc.org/ Downloaded from Effects of cholesterol on membrane insertion of Aβ 4 β-sheet (17). So Aβ in the form of β-sheet may be in favor of aggregating into fibril. In addition, in vitro studies with cell cultures have demonstrated that fibrillar Aβ is toxic to neurons while monomeric Aβ is not (18)(19)(20). Therefore the factors inducing Aβ to generate β-sheet may contribute to the pathogenesis of AD.
Aβ's neurotoxicity exhibits in many fields. One of the potential mechanisms for inducing the neurotoxicity of Aβ is direct interaction with the membranes. It has been reported that Aβ is able to form ionic pores (21); Aβ can destroy the structure of brain membranes (22); Aβ may stimulate free radical production by interfering with the regulation of calcium homeostasis and cell enzymatic activity (23); Aβ can alter the physical-chemical properties of neuronal membranes, including membrane fluidity, membrane lipid dynamic, and the activity of various membrane-bound proteins (24,25).
In the present work how lipid membrane affects the behavior of Aβ is emphasized. In particular the membrane insertion of Aβ , the major species normally secreted from cells (26), is studied by lipid monolayer and vesicle systems. At first, monolayer technique was performed to detect the membrane insertion ability of Aβ. Secondly, combination of MALDI-TOF mass spectrometry (MS) with enzymatic hydrolysis was carried out to confirm the membrane insertion of Aβ into phospholipid vesicles. And thirdly, circular dichroism (CD) spectroscopy was employed to study the conformational change of Aβ upon insertion. At last, electron microscopy (EM) work tested that membrane insertion of Aβ could reduce the formation of fibril. Our results show that Aβ(1-40) is able to insert into lipid bilayer, and there is a close correlation between membrane insertion of Aβ and its secondary structure, which is critically dependent upon the ratio of cholesterol to phospholipids. Our findings suggest the possibility that Aβ may prevent itself from aggregating by membrane insertion. Aβ  and , were purchased from AnaSpec Co. USA, whose purity (>95%) was analyzed by HPLC and checked by MALDI-TOF MS. And unless stated, Aβ refers to Aβ  in the following text.
Papain and CCA used in hydrolysis were purchased from Sigma Chemical Co. All the other chemicals used were of analytical grade and manufactured in China. Usually the subphase buffer was 50mM Tris-HCl containing 25mM NaCl with pH7.4.

Monolayer surface pressure measurements
The monolayer surface pressure (π), defined as the change of the surface tension after spreading a monolayer on the water surface, was measured with a NIMA 9000 (England) microbalance. The peptide insertion was determined by a circular Teflon trough with a volume of 4 ml and surface area of 10cm 2 . A filter paper of 1.0cm width was employed as the Wihelmy plate. The surface pressure measurements were made with the plate in a fixed-height position, and the data were automatically collected and recorded by computer.
In general, the experiments were conducted as follows. Firstly, the circular trough was filled with 4 ml of buffer. Then the phospholipid monolayers were prepared by carefully spreading the lipid solution (dissolved in a solvent of chloroform/methanol=3:1 v/v with the concentration of 1.0mg/ml) on to the buffer surface. After the surface pressure stabilized at a constant desired value − initial surface pressure (π i ), Aβ was injected into the subphase through a side sample hole. The pressure change was monitored until the by guest on March 21, 2020 http://www.jbc.org/ Downloaded from Effects of cholesterol on membrane insertion of Aβ 6 surface pressure increase (∆π) had reached a maximal value, usually within 2 hr. The water used was deionized, and the subphase buffer contained 50mM Tris-HCl, pH7.4 and 25mM NaCl. During the measurement, the subphase was continuously stirred with a magnetic bar. The temperature was controlled at 24.0±1.0.

Preparation of phospholipid vesicles
Small unilamellar vesicles (SUVs) were prepared as follows: lipids of the desired composition were mixed in chloroform/methanol (3:1, v/v) and dried under a stream of nitrogen. Residual solvents were removed under high vacuum for 5-7 hr. The lipid films were then resuspended and sonicated in the desired buffer to near optical clarity by using a probe sonicator. The metal debris from the titanium tip were gotten rid of by the centrifugation. The concentration of phospholipid was determined by phosphate analysis (27).

Hydrolysis of Aβ
The enzyme used to hydrolyze Aβ was papain, isolated from the latex of Carica papaya. Papain consists of a single polypeptide chain with 212 residues and the molecular weight is about 23 KDa (28).
Papain's specific hydrolytic sites are Glu-X, Gly-X, Tyr-X, His-X, Lys-X and Arg-X. For the convenient measurements of MS the enzymatic hydrolysis of Aβ was carried out in low ion strength buffer (10mM Tris-HAc, pH7.4) at 37in Eppendorf tubes. In a typical experiment, before hydrolysis, vesicles sonicated in the same buffer reacted with Aβ and at the same time, Aβ in the identical volume of buffer only (without vesicles) was prepared as controls.

MALDI-TOF MS
The instrument used here is BIFLEX III MALDI-TOF Mass Spectrometer made by Bruker Daltonics Co. A pulsed nitrogen laser operating at 337nm is used to generate the MALDI (matrix-assisted laser desorption/ionization) ions.
The matrix used in these experiments was α-cyano-4-hydroxycinnamic acid (CCA). CCA was dissolved in a solution of a 1:1 mixture of acetonitrile and water containing 0.1% trifluoroacetic acid for sample preparation. A 0.5 µl aliquot of CCA mixture was added into an Eppendorf tube, and then mixed with 0.5 µl of reaction solution of peptide and papain. The final peptide-matrix mixture in the Eppendorf tube was deposited on a stainless steel probe tip and allowed to dry at room temperature.

Circular dichroism (CD) spectroscopy
CD measurements were carried out on a Jasco J-715 spectropolarimeter. Samples were scanned at least ten times at the rate of 200 nm/min with a 0.5nm step, 1nm bandwidth and then averaged. The path length of the quartz cell was 2mm. In the experiments, a blank run made with the vesicles or buffer alone was carefully subtracted from the experimental spectra for correction. The 200-250nm spectra were used for analysis and calculation because in this wavelength range, the vesicles scattering had little effect on the CD spectra. All spectra were smoothed and converted to the mean residue ellipticity, [θ] in deg * cm 2 /dmol, by using mean residue molecular weight of 110. And the secondary structure of the peptide was estimated from spectral simulations based on reference CD spectra of Yang (29).
In a general experiment, a desired amount of vesicle solution, determined by the lipid/protein ratio of by guest on March 21, 2020 http://www.jbc.org/ Downloaded from Effects of cholesterol on membrane insertion of Aβ 8 100, was added to an Eppendorf tube, and then Aβ (1mg/ml) was injected to acquire the final concentration of 0.1 mg/ml. After incubated for 90 min (providing enough time for the interaction between Aβ and vesicles), the mixture was measured by the spectropolarimeter.

Electron microscopy (EM)
For the vesicle containing sample preparation, vesicle solution was firstly added into an Eppendorf tube and then a certain amount of protein solution was injected to acquire the final protein concentration of 0.2 mg/ml. Stored at 37 for 2 days the vesicle suspensions were applied to carbon-coated copper grids, dried, negatively stained with 2% (w/v) uranyl acetate and visualized in a Philips CM120 transmission electron microscopy operated at 120kV. As a control the same procedure was performed for the sample containing only peptide without vesicles.

I Monolayer experimental results
In the experiments we have applied a model, which assumes that the surface pressure will increase only when the protein inserts into the monolayer and the surface pressure, as reported earlier (30), do not increase if the protein only interacts with the phospholipid head group. Thus, when peptide molecules are injected into the subphase, the corresponding change in surface pressure (∆π) can be interpreted as the result of the peptide inserting into the lipid monolayer. ∆π can be obtained as a function of various π i for each sampleÈand then the plot of ∆π versus π i yields a straight line with negative slope, which intersects the abscissa at a limiting surface pressure. The limiting surface pressure is defined as the critical insertion pressure (π c ) of Aβ for the corresponding lipid monolayer, which is used as a quantitative measure to evaluate the insertion ability of the peptide to the phospholipid monolayer.
In order to acquire appropriate conditions of the experiments, firstly the surface activity of Aβ was detected by measuring the self-penetration of Aβ into the air-water interface without spread lipid monolayer. The peptide solution was injected into the subphase to a final concentration of 200, 400, 500, 600 and 800 nM respectively, and the surface pressure was then measured and plotted versus the reaction time. The results (data not shown) show that Aβ could significantly increase the surface pressure, which indicate its strong surface activity. The maximum ∆π induced by the self-penetration of Aβ was found to be 13.7 mN/m and the minimum concentration of Aβ to reach such a maximum was 500nM. Thus, we kept π i of the lipid monolayers spread onto the subphase surface at or above 15.0 mN/m and the concentration of Aβ injected into the subphase at 500 nM in the following experiments. The further experiments, as shown in Figure 2, were performed for PC monolayers containing a certain amount of sphingomyelin (SPM) and cerebroside, imitating the content in brain membrane. From the ∆π∼π i curves of Figure 2, it can be seen that only addition of SPM or cerebroside has little influence on the insertion ability of Aβ. In the presence of 33% cholesterol, however, for both mixtures, PC/SPM and PC/cerebroside, the π c is obviously shifted to higher surface pressure, 32-33 mN/m. Such behavior provides evidence that it is cholesterol that contributes to the most increase of π c . In addition, the correlation between π c and mole fraction of cholesterol in PC/SPM mixtures was also measured (data not shown). The result showed a similar behavior as that exhibited in Figure 1 The above results indicate that the insertion ability of Aβ(1-40) is critically dependent upon the ratio of cholesterol to phospholipids. It has been established that the biological membrane pressure is 31-34 mN/m (31). The packing density of lipid monolayer with a surface pressure in this region can be assumed to be comparable to that of lipid bilayer (32,33). Therefore, Aβ  should be able to insert into the lipid bilayer in which the content of cholesterol is above 30%.
The same insertion experiments were performed for Aβ , and we found that Aβ(1-28) had no surface activity and thus could not induce the increase of surface pressure for different phospholipid monolayers (data not shown). This may be due to the hydrophilic property of Aβ(1-28) which prevents the peptide from inserting into the monolayers.

II MALDI-TOF MS results
In order to confirm whether Aβ(1-40) could insert into lipid bilayer, papain as a protease was used to hydrolyze Aβ after it reacted with lipid vesicles. And the hydrolysis products were analyzed by MALDI-TOF MS. The MS spectrum, which was used to determine the hydrolysis fragments (indicated by m/z), obtained for free Aβ (without vesicles) was analyzed at first. The spectrum is shown in Figure 3 Table 1. From Table 1 we can see that papain treatments only result in a partial hydrolysis of potential sites of cleavage, which may be due to some second structure of Aβ preventing more cleavage. In the spectra only the peaks with m/z value higher than 1000 are exhibited because the substances in matrix (such as CCA and its contaminants giving peaks at low m/z) can cause by guest on March 21, 2020 http://www.jbc.org/ 12 interference at low m/z. Some non-papain-cleavage fragments in Table 1 may be caused by the broken peptide since some peptide bonds may be readily broken during flying in the MS experiments.

Effects of cholesterol on membrane insertion of Aβ
The MALDI-TOF MS spectra of the hydrolysis products for Aβ reacting with DMPC and 33% cholesterol containing DMPC vesicles are shown in Figure 4. For reaction with DMPC vesicles the MS peaks are nearly identical with that of free Aβ. This result provides evidence that papain can approach to nearly all cleavage sites as for free Aβ, indicating that Aβ does not insert into such lipid vesicles. In contrast, after reaction with 33% cholesterol containing DMPC vesicles, as shown in figure 4B, the characteristic peaks of the MALDI-TOF MS spectrum of Aβ have a considerable change. Several fragments that could be obtained from free Aβ are disappeared here, such as the peaks of m/z2932, 3672 and 4072. By comparing the difference in the spectra between them in Figure 4, we can find that the inaccessible cleavage sites are Gly 33 -Leu 34 and Gly 37 -Gly 38 . Both two sites are located in the C-terminal region of Aβ. This phenomenon can be explained as their locus in the membrane-insertion part of Aβ, which protects them from being cleaved by papain. These results lead to conclusion that Aβ indeed inserts into the rich-cholesterol containing vesicles by its C-terminal domain.
The same experiments were performed also for Aβ , and no such effect was found (data not shown). This may be evidence that no membrane insertion occurs for Aβ(1-28).

III Circular dichroism (CD) results
Whether there is a conformational change of Aβ during membrane insertion is a crucial question, by guest on March 21, 2020 http://www.jbc.org/ 13 since Aβ's conformation may relate with its aggregation (17). Here the CD spectroscopy was employed to measure the secondary structure change. Before CD measurement a pre-experiment was performed to determine the critical molar ratio for Aβ binding, which is defined as the minimal ratio of lipid/Aβ for maximal binding of Aβ to lipid vesicles. In the present case the critical molar ratio of lipid/Aβ of about 50 was obtained. Thus in the following measurements we choose 100 as the actual lipid/Aβ ratio to assure enough of lipid. Figure 5 shows three CD spectra acquired from lipid-free Aβ, Aβ reacted with DMPC and with 33% cholesterol containing DMPC vesicles respectively. The CD spectrum of lipid-free Aβ is typical of a peptide containing significant random-coil content. The computer fit results show that lipid-free Aβ consists of 48.9% random-coil, 23.5% β-sheet and 1.7% α-helix. The results are consistent with the reports of Terzi E. et al. (1,34). Addition of DMPC and 33% cholesterol containing DMPC vesicles leads to a remarkable alteration in the CD curves as also shown in Figure 5, indicating that Aβ undergoes a significant conformational change. According to the computer fit results one can see that in the case of addition of DMPC vesicles, both contents of β-sheet and α-helix have about 7% increase, i.e., β-sheet increased to 31.2% and α-helix to 9.5%. In contrast, when 33% cholesterol containing DMPC vesicle was added, the structure of Aβ altered drastically. The β-sheet structure decreased to zero and the α-helix increased remarkably to 58.8%. Such results indicate that membrane insertion of Aβ eliminates its β-sheet structure and induces its α-helices.

Effects of cholesterol on membrane insertion of Aβ
In addition, the effect of pH on Aβ secondary structure was tested by means of CD. Firstly, Aβ in solution with different pH was measured and the result (data not shown) showed that pH had a certain by guest on March 21, 2020 http://www.jbc.org/ 14 influence on Aβ secondary structure, which was similar to that of MaLaurin's work (35). Then the conformations of Aβ after reacting with cholesterol-rich vesicles at different pH were measured (data not shown). The results showed that little influence could be observed, indicating that the final state of membrane inserted Aβ was not sensitive to pH, which is not like the case Aβ in solution.

Effects of cholesterol on membrane insertion of Aβ
The above measurements were performed also for Aβ , and the results (data not shown) show that there is almost no conformational change induced by lipid vesicles, which again is an evidence for the hydrophilic property of Aβ .

IV Electron microscopy (EM) results
The effect of phospholipid vesicles on the formation of Aβ

Discussions
Cholesterol is an integral component of all eukaryotic cell membranes and is essential for normal cellular functions (36,37). Within the cell cholesterol is not uniformly distributed. Plasma membrane contains the highest levels of cholesterol (nearly 90% of total cellular cholesterol) (38). And it was demonstrated that the cholesterol/phospholipid ratio of plasma membranes is about 0.52~0.70 (mol/mol) (39)(40)(41). Even in a certain membrane, cholesterol is distributed asymmetrically --enriched in cytofacial leaflet (42,43). It was noticed that cholesterol in the cytofacial leaflet of brain synaptic plasma membrane is about 87% of total plasma membrane cholesterol when young, but the distribution of cholesterol tends to be homogeneous in the cytofacial and exofacial leaflet during the aging process (44). Also cholesterol of plasma membranes in nerve system increases with age (45). The cholesterol ratio of brain plasma membrane, one can estimate, is at least 30mol% when aged. Thus, the currently used model membrane system simulates the normal aged physiological condition to a certain extend.
Two model membrane systems were employed in the current work, i.e., monolayer and vesicle (closed bilayer). Lipid monolayer is a unique system to distinguish the functional role of a particular component in the membrane mixtures. According to the monolayer results obtained from Figures 1 and 2, it is cholesterol when at high but physiologically reasonable level could obviously enhance the insertion ability of Aβ while other brain membrane components such as SPM and cerebroside not. A big inspiration drawn from the monolayer experiments is that the insertion ability of Aβ can be high enough, in the present case π c > 31 mN/m, for it to insert into lipid bilayer. This result has been proved by the MS analysis of hydrolytic products with vesicle system. The MS measurements provided direct evidence (as shown in Figure 4) that by guest on March 21, 2020 http://www.jbc.org/ 16 Aβ can really insert into the lipid vesicles containing rich cholesterol while can not without cholesterol.

Effects of cholesterol on membrane insertion of Aβ
The MS measurements also determined that the membrane insertion of Aβ is by its C-terminal. This is quite reasonable since hydrophobic domain locates there.
The membrane induced conformational change is investigated in the present work by CD technique with vesicle system. This should be a very important problem since the behavior of membrane insertion of Aβ may be related with its neurotoxicity by its conformational change. Several laboratories have reported their CD work on Aβ. Different experimental conditions may induce different Aβ conformations, i.e., (i) α-helix in trifluoroethanol (TFE), sodium dodecyl sulfate (SDS) micelles or induced by gangloiside containing vesicles (34,35,(46)(47)(48)(49), and also in our experiments we use ethanol-water system to simulate the membrane condition, and the CD results indicate that α-helix increases along with the increase of ethanol's content (data not shown), which coincide with the previous reports very well; (ii) essentially random-coil structure with β-turns in aqueous solution at low peptide concentrations; and (iii) β-structured aggregates in solution or in contact with lipid membranes (2,34). So far as our knowledge, systems containing cholesterol that assuredly is the common component of membranes are not employed yet in these published papers.
From our CD results shown in Figure 5, one can realize that Aβ solved in aqueous solution adopts a conformation mainly in random-coil; while when reacted with DMPC vesicles it contains more β-sheet, then when reacted with rich-cholesterol containing vesicles it becomes mainly in α-helix. One interesting phenomenon is that a certain conformational change occurs even when Aβ reacts with DMPC vesicles.
From the monolayer surface pressure measurements, as mentioned before, the π c for PC monolayer is self-penetration of Aβ shows that Aβ is a peptide with a strong surface activity. This nature of Aβ makes itself prefer to state in an amphiphilic environment, i.e., membrane surface region. The physical chemistry property in membrane surface region is different from that in the inner bilayer and in the bulk, which has been widely reported. That such difference may induce protein conformational change has also been noticed (50). Therefore, though Aβ can not insert into PC bilayer, a certain conformation change may also happen when it reaches to membrane surface region. This is the possible case when Aβ reacts with DMPC vesicles. Quite important point is that under such case Aβ adopts considerable structure in β-sheet, suggesting that the aggregation of Aβ may occur around the surface region of membranes since β-sheet may be a factor inducing Aβ aggregation. This coincides with the previous direct observation by using confocal laser scanning microscope and transmission electron microscopy (2,(51)(52)(53).
Another crucial point we obtained is that Aβ adopts mainly α-helix after reacted with rich-cholesterol containing PC vesicles. This effect could be beneficial to reduce the formation of aggregation by depressing the β-sheet conformation, suggesting a possible pathway that Aβ aggregation that ultimately induces the formation of plaques may be prevented by its membrane insertion. Electron micrographs shown in figure 6 provided a direct test performed in vitro to support this hypothesis: addition of rich-cholesterol containing vesicles could effectively depress the formation of Aβ    18 According to the above results we can know the existence forms of Aβ: (i) existing in the aqueous solution and adopting a random-coil conformation; (ii) existing in the surface region of vesicles and adopting a conformation containing more β-sheet, such as interaction with DMPC vesicles; (iii) inserting into rich-cholesterol containing membranes and adopting an α-helix conformation. Altogether, our data show that the existence forms of Aβ is obviously dependent on the ratio of cholesterol to phospholipid of membranes, which on one side reflects the fact that membrane cholesterol distribution could be an important event in the Aβ-related disease.

Effects of cholesterol on membrane insertion of Aβ
Aβ generated to perform its physiological function can be degraded through endocytosis, such as via LRP or via scavenger receptor. But when aged, one of important changes for Aβ metabolism is that LRP will reduce approximately 45% (54). The decrease of LRP will partly block the degradation pathway of Aβ, which would increase the extracellular content of Aβ. If the content of Aβ can't be reduced, it would aggregate and form plaques. The increase of cholesterol, also a change with aging, may be a compensatory factor reducing extracellular Aβ by membrane insertion (according to our results). In the case for AD subjects, however, a significant difference is present: brain membranes isolated from them show dramatically decreases in membrane cholesterol content (the ratio of cholesterol to phospholipids decreases about 30%) (55,56). Under such condition, Aβ may be not able to insert into membrane to avoid from aggregation.    Table 1.