Endosomal pH favors shedding of membrane‐inserted amyloid‐β peptide

Abstract

Amyloid‐β peptides (Aβs) are generated in a membrane‐embedded state by sequential processing of amyloid precursor protein (APP). Although shedding of membrane‐embedded Aβ is essential for its secretion and neurotoxicity, the mechanism behind shedding regulation is not fully elucidated. Thus, we devised a Langmuir film balance‐based assay to uncover this mechanism. We found that Aβ shedding was enhanced under acidic pH conditions and in lipid compositions resembling raft microdomains, which are directly related to the microenvironment of Aβ generation. Furthermore, Aβ shedding efficiency was determined by the length of the C‐terminal membrane‐spanning region, whereas pH responsiveness appears to depend on the N‐terminal ectodomain. These findings indicate that Aβ shedding may be directly coupled to its generation and represents an unrecognized control mechanism regulating the fate of membrane‐embedded products of APP processing.

Abbreviations

Aβ

amyloid‐β peptide

AD

Alzheimer's disease

APP

amyloid precursor protein

CD

circular dichroism

Introduction

Alzheimer's disease (AD), which is the most common type of dementia worldwide.1, 2, 3, 4 One prominent consequence of dysregulated APP processing is the overproduction of amyloid‐β peptide (Aβ). Aβ production depends on sequential cleavage of APP by β‐secretase within the ectodomain and γ‐secretase within the membrane‐spanning region, which generates membrane‐embedded peptides of 36–43 amino acids in length. Following this, membrane‐embedded Aβ is secreted, mostly in the soluble state,5, 6, 7, 8 constituting the major component of extracellular amyloid plaques found in the brains of AD patients.1, 2, 3, 4 Moreover, levels of secreted Aβ in cerebrospinal fluid and plasma are sensitive biomarkers that predict brain amyloid burden.9

Although the toxicity of Aβ in AD was questioned,10 the recent success of the Aβ‐ Dysregulated processing of amyloid precursor protein (APP) is central to initiating targeting antibody aducanumab in early‐phase AD treatment supports the etiological significance of Aβ.11 Secreted Aβ is considered a key mediator in evoking neurotoxicity largely due to its intrinsic capability to aggregate and interact with membranes.12, 13 Therefore, uncovering the factors that contribute to the shedding of membrane‐embedded Aβ into the fluid phase is important. In the present study, we attempted to elucidate these factors with a Langmuir film balance connected to a syringe pump (Fig. 1). The Langmuir film balance is widely used to investigate protein–membrane interactions by measuring changes in the surface pressure of lipid monolayers.14, 15 Insertion of a test protein into the hydrophobic core of a lipid monolayer will increase surface pressure, whereas the shedding of an already inserted protein will decrease surface pressure. This setup allows us to separately monitor membrane insertion of Aβ and its subsequent shedding in a single experiment. Our results suggest that Aβ shedding occurs primarily in lipid raft microdomains of endosomes immediately following Aβ generation.

Figure 1.

Experimental design. Lipid monolayers were first formed on the subphase in the trough to an initial surface pressure of π0. Aβ was then injected into the subphase for insertion into the preformed lipid monolayers. After 1500 s of insertion, Aβ‐containing buffers were washed out with Aβ‐free buffers using a connected syringe pump. πa and πb denote surface pressure at the beginning and at the end of the washout, respectively.

Results

Experimental design

To separate the membrane insertion and shedding events of Aβ, we first injected Aβ into 3 mL subphase for 1500 s to allow a near saturable insertion into the preformed lipid monolayer with a fixed area (3.14 cm²), which increased surface pressure from π₀ to π_a. Aβ sheds under π_a, which is an approximate physiological membrane pressure of 32 mN/m.16, 17 π_a below physiological membrane pressure (approximately 26 mN/m) is a very important control as it is not easy for Aβ to shed at lower membrane pressure because the side pressure of lipid molecules is not noticeable. Subsequently, Aβ was washed out with 20 mL peptide‐free buffer using a connected syringe pump for 1200 s, decreasing surface pressure from π_a to π_b (Fig. 1). Control experiments indicated that buffer washout per se had little effect on the surface pressure of monolayers without inserted Aβ (Figs. 2, 3, 4, 5). We focused our analysis of Aβ shedding within the first 1200 s of buffer washout to minimize confounding effects caused by extended incubation. Therefore, the extent of Aβ shedding was evaluated by (π_a‐π_b)/(π_a‐π₀).

Figure 2.

Shedding of Aβ40 from lipid monolayers. (A, C, E) Surface pressure versus time curves of Aβ40 interacting with (A) lipid raft, (C) DPPC, and (E) DMPC monolayers. (B, D, F) Relative shedding of Aβ40 from (B) lipid raft, (D) DPPC, and (F) DMPC monolayers. Shedding experiments where πa was larger than 30 mN/m are around physiological membrane pressure 16, 17, whereas those with πa less than 30 mN/m are below physiological membrane pressure. pH 7.4 → pH 7.4 denotes washout using a pH 7.4 buffer. pH 7.4 → pH 5.5 denotes washout using a pH 5.5 buffer. The extent of shedding was calculated as (πa‐πb)/(πa‐π0). For ease of comparison, all calculated values were normalized to that of Aβ40 shedding from lipid raft monolayers at physiological membrane pressure after washing out with a pH 7.4 buffer. The experimental conditions of Aβ40, lipid raft, physiological membrane pressure, and physiological pH were regarded as the standard [(πa‐πb/πa‐π0)% = 13.57%, which was set as coefficient “1”]. Other groups of (πa‐πb/πa‐π0)% were quantified as a relative value (relative membrane shedding). Results (n > 3) are provided as the mean ± S.E.M.

Figure 3.

Shedding of Aβ36 from lipid raft monolayers. (A, C, E) Surface pressure versus time curves of Aβ36 interacting with (A) lipid raft, (C) DPPC, and (E) DMPC monolayers. (B, D, F) Relative shedding of Aβ36 from (B) lipid raft, (D) DPPC, and (F) DMPC monolayers. Shedding experiments where πa was larger than 30 mN/m are around physiological membrane pressure, whereas those with πa less than 30 mN/m are below physiological membrane pressure. pH 7.4 → pH 7.4 denotes washout using a pH 7.4 buffer. pH 7.4 → pH 5.5 denotes washout using a pH 5.5 buffer. The extent of shedding was calculated as (πa‐πb)/(πa‐π0). For ease of comparison, all calculated values were normalized to that of Aβ40 shedding from lipid raft monolayers at physiological membrane pressure after washing out with a pH 7.4 buffer.

Figure 4.

Shedding of Aβ42 from lipid raft monolayers. (A, C, E) Surface pressure versus time curves of Aβ42 interacting with (A) lipid raft, (C) DPPC, and (E) DMPC monolayers. (B, D, F) Relative shedding of Aβ42 from (B) lipid raft, (D) DPPC, and (F) DMPC monolayers. Shedding experiments where πa was larger than 30 mN/m are around physiological membrane pressure, whereas those with πa less than 30 mN/m are below physiological membrane pressure. pH 7.4 → pH 7.4 denotes washout using a pH 7.4 buffer. pH 7.4 → pH 5.5 denotes washout using a pH 5.5 buffer. The extent of shedding was calculated as (πa‐πb)/(πa‐π0). For ease of comparison, all calculated values were normalized to that of Aβ40 shedding from lipid raft monolayers at physiological membrane pressure after washing out with a pH 7.4 buffer.

Figure 5.

Shedding of Aβ17–42 from lipid monolayers. (A, C, E) Surface pressure versus time curves of Aβ17–42 interacting with (A) lipid raft, (C) DPPC, and (E) DMPC monolayers. (B, D, F) Relative shedding of Aβ17‐42 from (B) lipid raft, (D) DPPC, and (F) DMPC monolayers. Shedding experiments where πa was larger than 30 mN/m are around physiological membrane pressure, whereas those with πa less than 30 mN/m are below physiological membrane pressure. pH 7.4 → pH 7.4 denotes washout using a pH 7.4 buffer. pH 7.4 → pH 5.5 denotes washout using a pH 5.5 buffer. The extent of shedding was calculated as (πa‐πb)/(πa‐π0). For ease of comparison, all calculated values were normalized to that of Aβ40 shedding from lipid raft monolayers at physiological membrane pressure after washing out with a pH 7.4 buffer.

Acidic pH favors Aβ shedding from lipid rafts

The most abundant Aβ isoform produced in the brain is Aβ40, which is 40 amino acids in length. Moreover, cholesterol‐enriched lipid rafts have been proposed as the major sites of Aβ generation,2, 18 and we have also shown that Aβ40 preferentially inserts into lipid rafts.19 Therefore—as expected—addition of Aβ40 into a neutral pH subphase induced a substantial increase in the surface pressure of monolayers with lipid raft compositions. Washout of Aβ40 with neutral pH buffers resulted in only a moderate decrease in surface pressure (~13%), suggesting rather limited shedding [Fig. 2(A,B)]. Notably, however, shedding was enhanced by 2.5‐fold upon washout with pH 5.5 buffers. This acidic pH‐enhanced shedding was also observed at a surface pressure lower than that of biological membranes and for monolayers with different compositions [Fig. 2(C–F)]. Nevertheless, lipid raft‐inserted Aβ40 appeared to be the most primed for shedding. Thus, our results suggest that Aβ shedding occurs primarily from lipid rafts in acidic endosomes (and possibly the trans‐Golgi network), which are also major sites of Aβ generation.2, 18

Length of the C‐terminal membrane‐spanning region determines Aβ shedding efficiency

Four kinds of Aβs with different C‐terminal lengths are summarized in Figure 6 and their shedding was examined after washing out with acidic buffer. It can be seen that the Aβ C‐terminal plays an important role in protein shedding under approximate physiological membrane pressure, because Aβ36 is more serious than Aβ40 and Aβ42. In addition, Aβ36 sheds easier under physiological membrane pressure than under lower membrane pressures, which supports the stability and reliability of the system.

Figure 6.

Shedding of the four Aβs from lipid monolayers. (A, B, C) Surface pressure versus time curves of Aβ36, Aβ40, Aβ42, and Aβ17–42 interacting with (A) lipid raft, (B) DPPC, and (C) DMPC monolayers. Shedding experiments where πa was larger than 30 mN/m are around physiological membrane pressure, whereas those with πa less than 30 mN/m are below physiological membrane pressure. pH7.4 → pH 5.5 denotes washout using a pH 5.5 buffer. (D) For ease of comparison, relative shedding values of the four peptides were aggregated. All calculated values were normalized to that of Aβ40 shedding from lipid raft monolayers at physiological membrane pressure after washing out with a pH7.4 buffer.

Most Aβ isoforms differ only in the length of their C‐terminal membrane‐spanning region.2, 18 Aβ36, with a shorter membrane‐spanning region, was much more prone to shedding than Aβ40 (Fig. 3). In contrast, the more amyloidogenic isoform Aβ42 with a longer membrane‐spanning region showed impaired shedding (Fig. 4). Because the strength of membrane insertion is generally positively correlated with the length of the membrane‐spanning region, these results indicate that the tighter the Aβ inserts, the less likely shedding will occur.

The N‐terminal ectodomain determines pH responsiveness of Aβ shedding

Despite different efficiencies, Aβ36, Aβ40, and Aβ42 shedding was enhanced under acidic pH conditions (Figs. 2, 3, 4, 5, 6). As the three isoforms have identical N‐terminal ectodomains, this suggests that the pH responsiveness of Aβ shedding may depend on the N‐terminal ectodomain. Besides Aβ, APP can be alternatively processed on plasma membranes (at neutral pH) outside lipid rafts2, 18 to generate the nonamyloidogenic product Aβ17–42 (also termed p3), which harbors the intact membrane‐spanning region but lacks amino acids 1–16 of the N‐terminal ectodomain. In line with our abovementioned speculation, Aβ17–42 shedding was limited, likely due to the intact membrane‐spanning region, and was only marginally enhanced by acidic pH for both raft and nonraft monolayers (Fig. 5).

Acidic pH reduces Aβ insertion in membranes

If low pH values enhance membrane‐inserted Aβ shedding from the phospholipid membrane, then acidic conditions may not be suitable for Aβ insertion into membranes compared with that of neutral conditions. For this reason, we used lipid rafts and nonraft phospholipids to study Aβ42 insertion into monolayers (Fig. 7). Under physiological conditions (pH 7.4), the critical insertion pressure of lipid rafts reached 34 mN/m while that of 1,2‐dipalmitoyl‐sn‐glycero‐3‐phosphocholine (DPPC) monolayers was approximately 32 mN/m. Meanwhile, the critical insertion pressure of lipid rafts reached 29 mN/m under pH 5.5 and was approximately 28 mN/m for DPPC monolayers. These findings indicate that Aβ membrane insertion is difficult under acidic conditions.

Figure 7.

Aβ insertion into lipid raft and non‐lipid raft monolayers at different pH values. (A, C) Aβ40 monomer (600 nM) was injected into the subphase beneath (A) lipid raft or (C) DPPC monolayers with an initial surface pressure of 21 mN/m, after which time curves of surface pressure increase (Δπ) were recorded. (B, D) Surface pressure change (Δπ)‐initial surface pressure (πi) plots of Aβ interaction with (B) lipid raft and (D) DPPC monolayers. Critical insertion pressure (πc) values of Aβ for lipid raft and DPPC monolayers at the intersection point of the X‐axis. Because the πc at pH 5.5 is lower than 30 mN/m, which is the physiological lateral pressure of the cell membrane, Aβ is unable to directly insert into membrane bilayer, unlike under pH 7.4 conditions.

To further elucidate why acidic conditions are not suitable for Aβ insertion into monolayers, we employed circular dichroism (CD) spectroscopy to detect the secondary structure of Aβ in solution (Fig. 8). Aβ40 and Aβ42 are prone to aggregation and have complex secondary structures.20 In this experiment, we expected that pH 5.5 would change the secondary structure of Aβ. All CD curves were tested by a secondary structure calculation tool as described previously.21, 22 The results show that both Aβ40 and Aβ42 possessed a high content of α‐helical structures at pH 7.4. At pH 5.5, this abundance of α‐helices was lost. Therefore, it is clear that the acidic environment has a great influence on the secondary structure of Aβ40 and Aβ42. This may be a critical factor in the shedding of Aβ from membranes under acidic environments. It is likely that the disordered structure adopted by Aβ under acidic conditions is not suitable for membrane insertion. The α‐helix conformation is then a suitable protein structure for membrane insertion, which is consistent with a previous report.23 Therefore, our results indicate that the solution is more suitable for protein retention than the membrane under acidic conditions.

Figure 8.

Subphase pH changes the secondary structure of Aβ. (A) Secondary structure curves of Aβ42 under pH 7.4 and pH 5.5 conditions. Mouse Aβ42 was used as negative control. (B) Secondary structure curves of Aβ40 under pH 7.4 and pH 5.5 conditions. From pH 7.4 to pH 5.5, the percentage of α‐helix conformations was lost.

Discussion

The pathogenic and diagnostic significance of secreted Aβ has been well recognized.1, 2, 3, 4, 9 However, it remains unclear how Aβ, generated as a membrane‐embedded peptide, is secreted into the fluid phase. To address this question, an assay with sufficient sensitivity to monitor shedding events is required. Herein, we examined Aβ shedding by monitoring the decrease in surface pressure of monolayer‐inserted Aβ in real time with a Langmuir film balance assay. Based on the experimental setup, there are two major caveats that would affect result interpretation. The first caveat is that Aβ inserted into monolayers from the fluid phase may not recapitulate the membrane‐embedded state upon Aβ generation. Nevertheless, we have shown that membrane insertion of fluid‐phase Aβ depends on the C‐terminal membrane‐spanning region24 and is associated with an increased tendency toward an α‐helical conformation.19 This suggests that membrane‐inserted Aβ may resemble that of Aβ generated with a membrane‐spanning α‐helix prior to shedding. Alternatively, utilizing monolayers prepared directly by mixing Aβ and lipids is possible. However, such a setup would require precise control of the amounts of lipids and peptides applied in each experiment for accurate quantification and comparison, which proved difficult.

The second caveat is that a decrease in surface pressure may be the result of conformation changes within the lipid monolayer24 rather than a full release or shedding of Aβ. However, conformation changes usually require extended reaction times with moderate changes in surface pressure.24 In this regard, the substantial decrease in surface pressure within a short time frame, particularly in case of lipid raft monolayers with inserted Aβ, favors the interpretation of shedding. Moreover, the extent of Aβ shedding may in fact be underestimated in our analysis. This is because our analysis did not extend beyond 1200 s (after which a continuous decrease in pressure was still observed) and Aβ shedding became less prominent when monolayer surface pressure decreased below that of physiological membranes [Fig. 2(A,B)].

Taking all this into consideration, our results revealed that Aβ shedding occurs more readily in lipid rafts. Indeed, a previous study reported that GM1 and/or cholesterol are required for the aggregation of Aβ integrated into liposomes, indicating Aβ shedding from lipid rafts.25 We further demonstrated that Aβ shedding from lipid rafts can be markedly enhanced by an acidic environment, which is frequently found in endosomes. This enhanced shedding may be due to protonation that intensifies electrostatic repulsion between the N‐terminal ectodomain of Aβ and the head‐groups of lipids at the monolayer–subphase interface, leading to destabilization of the membrane‐inserted state of Aβ. Interestingly, endosome lipid rafts are thought to be the major site for Aβ generation,2, 18 which also possess conditions that favor Aβ shedding. This would suggest an intimate coupling between the generation and shedding of Aβ, with shedding occurring immediately following generation. Thus, if the acidic environment supports Aβ shedding, then the possibility of Aβ reinsertion into the membrane is relatively small. We verified this notion by using lipid raft and nonraft phospholipid monolayers, which indicated that it was not easy for Aβ to insert into the membrane under acidic conditions (Fig. 7). Furthermore, we examined the secondary protein structure of Aβ by CD spectroscopy and found that Aβ40 and Aβ42 conformation under pH 7.4 conditions was mainly α‐helical, which then changed to a disordered structure at pH 5.5 that was not suitable for membrane insertion (Fig. 8). α‐helices are maintained by intra‐chain hydrogen bonds and low pH levels alter these bonds, which supports the promotion of Aβ shedding but not reinsertion under acidic pH.

It should be explained that the PH of endosome is likely to be a range. For example, the acidic environment of the lysosomal lumen pH is 4.5–5.0.26 Trans‐Golgi pH in skin fibroblasts is 6.17 ± 0.02.27 Trans‐Golgi pH in HeLa cells is ~6.5828 and in lumen cell is 6.4.29 We chose pH 5.5 as the representative value of endosome pH. We have not only experimented at pH 5.5, but also at pH 6.5. These experiments include monolayer insertion, CD and Aβ shedding. The relative value of Aβ shedding under pH 6.5 condition is between pH 5.5 and pH 7.4. These data did not change the conclusion that endosomal pH favors shedding of membrane‐inserted Aβs.

Our results further suggest that shedding can be a determinant of the fate of membrane‐embedded products of APP processing; those that are not easily shed from membranes may be more prone to degradation in lysosomes. Assuming a comparable mechanism of metabolism, secretion levels of different processing products would be primarily determined by the rate of generation and the ease of shedding. This may at least partly account for the observation that the difference between cellular production of Aβ40 and Aβ4230 is much less pronounced than the difference between their secretion into conditioned media30 and body fluids.9, 31, 32 Furthermore, it also implies that the nonamyloidogenic product p3, due to its limited shedding, is more likely to be degraded rather than secreted following endocytosis. Further investigations are warranted to validate these assumptions at the cellular level.

Experimental Procedures

Reagents

Aβ36, Aβ40, Aβ42, Aβ17–42, and mouse Aβ42 peptides were purchased from rPeptide (Watkinsville, GA). 1,1,1,3,3,3‐Hexafluoro‐2‐propanol (HFIP) and dimethyl sulfoxide (DMSO) were purchased from Sigma‐Aldrich (St. Louis, MO). The lipids DPPC, 1,2‐dimyristoyl‐sn‐glycero‐3‐phosphocholine, GM1 ganglioside (brain, ovine‐ammonium salt), 1,2‐dioleoyl‐sn‐glycero‐3‐phosphocholine (DOPC), cholesterol, and sphingomyelin (brain, porcine) were obtained from Avanti Polar Lipids (Alabaster, AL). All lipids were dissolved in chloroform/methanol (75/25 vol %) to 1 mg/mL. Lipid rafts were fabricated by mixing DOPC, sphingomyelin, cholesterol, and GM1 with a molar ratio of 32:32:31:5.

Preparation of monomeric Aβ

Monomeric Aβ was prepared as described previously by us.24, 33 Briefly, lyophilized Aβ was first dissolved in HFIP to 1 mM by vigorous vortexing for 2 min followed by overnight incubation at 4°C with slow shaking. The peptide solution was then aliquoted, evaporated with a N₂ stream, and stored at −80°C until further use. Prior to the experiments, the peptide aliquots were dissolved in DMSO to 1 mM by brief vortexing for 1 min and then water batch sonicated for 1 min.

Assay for determining Aβ shedding from lipid monolayers

Monolayer experiments were conducted with a μTrough‐S microbalance (Kibron, Helsinki, Finland). Lipids were spread onto the subphase (10 mM Tris, 140 mM NaCl, and pH 7.4; 3 mL) in the trough with a fixed area of 3.14 cm² to form a monolayer. The monolayer was equilibrated for 1000 s to reach a stable initial surface pressure (π₀). Subsequently, Aβ was injected through the side hole into the subphase to 612 nM for 1500 s. The insertion of Aβ into the monolayer increased the surface pressure to π_a. The Aβ‐containing subphase was then washed out with 20 mL peptide‐free buffer (10 mM Tris, 140 mM NaCl, and pH 7.4; or 200 mM Na₂HPO₄, 100 mM citric acid, and pH 5.5) at a rate of 1 mL/min using a syringe pump connected to the trough. At the end of the washout, the surface pressure of the monolayer decreased to π_b. The experiments were performed at a temperature of 23.5 ± 0.5°C with continuous stirring.

CD spectroscopy

CD spectra were recorded on an Applied Photophysics Chirascan (Surrey, UK) at 25°C. Spectra were obtained from 190 to 260 nm, with a 1‐nm step, 1‐nm bandwidth, and 10‐s collection time per step. The effects of pH on peptide conformation were determined by adding Aβ solutions to lipids suspended in PBS (pH 7.4) as well as in 200 mM Na₂HPO₄ and 100 mM citric acid (pH 5.5). Final peptide concentration was 0.45 μg/μL. CD spectra were examined immediately after addition of Aβ. All secondary structure curves of Aβ were tested by an online secondary structure calculation tool as previously described.21, 22

Supporting information

Appendix S1: Supplementary materials for review