Microglia degrade Alzheimer's amyloid-beta deposits extracellularly via digestive exophagy

Abstract

How microglia digest Alzheimer's fibrillar amyloid-beta (Aβ) plaques that are too large to be phagocytosed is not well understood. Here, we show that primary microglial cells create acidic extracellular compartments, lysosomal synapses, on model plaques and digest them with exocytosed lysosomal enzymes. This mechanism, called digestive exophagy, is confirmed by electron microscopy in 5xFAD mouse brains, which shows that a lysosomal enzyme, acid phosphatase, is secreted toward the plaques in structures resembling lysosomal synapses. Signaling studies demonstrate that the PI3K-AKT pathway modulates the formation of lysosomal synapses, as inhibition of PI3K1β or AKT1/2 reduces both lysosome exocytosis and actin polymerization, both required for the formation of the compartments. Finally, we show that small fibrils of Aβ previously internalized and trafficked to lysosomes are exocytosed toward large Aβ aggregates by microglia. Thus, the release of lysosomal contents during digestive exophagy may also contribute to the spread and growth of fibrillar Aβ.

Keywords: 5xFAD; Alzheimer’s disease; CP: Neuroscience; acid phosphatase; amyloid-beta; digestive exophagy; electron microscopy; extracellular degradation; lysosomal pH; lysosome; microglia.

Figure 1.. Microglial cells establish lysosomal synapses and secrete lysosomal contents toward large Aβ_1–42 aggregates

(A–C) Microglial cells were incubated with Alexa 633-Aβ_1–42 aggregates for 1 h followed by plasma membrane labeling with Alexa 555-cholera toxin B (green), fixation, and imaging by high-resolution Airyscan confocal microscopy. (A) Single-plane image showing the plasma membrane of a microglial cell (arrowhead) partially encapsulating a portion of the aggregate in the lysosomal synapse. (A1) yz orthogonal view of the image in (A). (B and C) Imaris 3D reconstruction of the lysosomal synapse shown in (A). The Aβ_1–42 aggregate (red) is partially encapsulated within the compartment (arrowheads). In (B), the top portion of the cell was removed to show the totality of the aggregate surrounded by the cell membrane (green). (D–F) Microglial cells co-incubated with Alexa 633-Aβ_1–42 aggregates (red) were fixed and stained with Alexa 488-phalloidin (green) for 1 h, followed by confocal microscopy imaging. (D) Representative image showing that actin polymerization is increased (arrowheads) at the contact region between microglial cells and the aggregates. (E and F) Imaris 3D reconstruction of the phalloidin signal shows that actin polymerization is increased (arrowheads) at the lysosomal synapse. Closer inspection shows that the region of intense actin polymerization coincides with the presence of an extracellular compartment that surrounds a portion of an aggregate. (G) To image lysosome exocytosis, lysosomes were labeled with fluorescein-biotin-dextran. The cells were then incubated with streptavidin-Alexa 647-Aβ_1–42 aggregates for 90 min and fixed. The intracellular dextran was removed by permeabilizing the cells with Triton X-100 in the presence of excess biotin and BSA. (H) Confocal microscopy micrograph of a single plane showing that fluorescein-biotin-dextran from microglial lysosomes (green) was released and captured by the streptavidin-Alexa 647-Aβ_1–42 aggregates (red) (arrowheads in orthogonal section). (I and J) Microglial cells were co-incubated with 6-μm-diameter streptavidin-coated fluorescent purple latex beads and large streptavidin-Alexa 647-Aβ_1–42 aggregates (I), and lysosomal exocytosis of fluorescein-biotin-dextran toward beads and aggregates was quantified by measuring fluorescein signal deposited on them (J). Exocytosis toward Aβ_1–42 aggregates was significantly increased (79.53 ± 16.15, **p < 0.01) relative to streptavidin beads (1 ± 0.06). n = 9 wells imaged (107 fields acquired) for each condition, plotted as large geometrical shapes or small gray triangles, in four independent experiments. Bars indicate ±SEM. Statistical differences between mean fluorescein intensities calculated for each condition were assessed using the unpaired two-tailed Student’s t test with Welch correction. See Figure S1 for a characterization of the Aβ model aggregates used in this paper and Table S2 for statistics. Scale bars: 10 μm (A, E, F, H, and I), 3 μm (B and C), and 5 μm (D).

Figure 2.. Lysosomal synapses established by microglial cells contacting Aβ aggregates are acidic

(A) Aβ_1–42 aggregates were labeled with ApHID (green, intensity increases with increasing acidity) and Alexa 633 (red, pH-independent) and incubated on top of microglial cells for 1–2 h followed by imaging using confocal microscopy. (A1) The yz orthogonal view shows microglia-aggregate contacts that are acidified as seen in yellow-orange color (cyan arrowheads). Regions at buffer pH (~7.4) appear red, corresponding to very weak ApHID fluorescence. (B and C) (B) Aβ_1–42 aggregates labeled with fluorescein (pH-sensitive) and Alexa 633, incubated with pH-adjusted buffers for 1 h at 37°C and imaged by confocal microscopy. (C) Fluorescein and Alexa 633 channels were quantified, and the fluorescein/Alexa 633 ratio was plotted against buffer pH, yielding a curve corresponding to the titration of fluorescein, fit to a sigmoidal curve. Fluorescein/Alexa 633 intensity ratios were normalized to their respective values at pH 8.0. n = 4–6 fields imaged per buffer pH in total, acquired from two dishes in two experiments. Dotted lines indicate 95% confidence range of the fit. See Table S3 for statistics. (D–G) Fluorescein/Alexa 633 ratios were measured for lysosomal synapses using live confocal microscopy and interpolated to pH values using a calibration prepared as shown in (C). pH was color coded using ratiometric digital image analysis (D and inset expanded in E, color-coded pH scale bar to the left of D, single fluorescence plane). The dotted red lines in (E) outline two aggregates contacted by microglial cells. The yz orthogonal view of (E) shows a representative acidified region at a cell-aggregate contact site (ratiometric color-coded, F, and unprocessed fluorescence channels, G; red arrowheads point toward the acidified compartment). (H) The pH of lysosomal synapses is acidic (average pH 6.31 ± 0.03), but some regions showed lower pH values. Bafilomycin A1 treatment for 1 h prior to incubation with aggregates abolished lysosomal synapse acidification (average pH 7.23 ± 0.05; ****p < 0.0001 vs. DMSO). n = 174 and 122 lysosomal synapses quantified for DMSO-treated (green dots) and bafilomycin A-treated (red dots) microglial cells, respectively, acquired from two dishes in two independent experiments. See Table S4 for statistics. Bars indicate ±SD (C) or ±SEM (H). Statistical differences between mean lysosomal pH between conditions were assessed using the unpaired two-tailed Student’s t test. Scale bars: 10 μm (A, B, and E) and 20 μm (D). Fluorsc., fluorescein; A633, Alexa Fluor 633.

Figure 3.. Microglial cells use digestive exophagy to degrade Aβ_1–42 aggregates that cannot be phagocytosed or endocytosed

Microglial cells were incubated for 1 h on top of a streptavidin-coated glass coverslip coated with biotinylated fluorescein-Aβ_1–42 aggregates. As a result of the streptavidin-biotin interactions, the aggregates were tightly bound to the coverslip. The preparations were imaged live in a sterile humidified chamber at 37°C with 5% CO₂, 1 h after initial seeding and every 24 h thereafter, for 3 days, by confocal microscopy. A final imaging session was completed at 7 days of incubation. (A–E) Representative images showing microglial cells (delineated in the transmitted light channel) interacting with fluorescein-Aβ_1–42 aggregates (green) of different sizes over 7 days. (F–J) Representative images showing microglial cells continuously engaging a large Aβ_1–42 aggregate (≥10 μm in diameter) over a period of 7 days. (K) The time-dependent degradation of the aggregates was quantified by measuring fluorescein intensity per field normalized to Nile red bead fluorescence measured at the end of each imaging session. Microglial cells degraded one-third of the aggregated Aβ in the first 24 h (0.69 ± 0.02; ****p < 0.0001 vs. 1-h time point) and then slowly degraded the remaining aggregates over the following 7 days. Fluorescein signal for each field was normalized to its corresponding value at 1-h time point. n = 5–6 dishes imaged (75–91 fields acquired), represented by large geometrical shapes and small gray triangles, in three independent experiments. See Table S5 for statistics. (L) Quantification of fluorescein signal from large Aβ_1–42 aggregates contacted by microglial cells for a period of 7 days. Microglial cells degraded 43% of the large aggregates during the first 24 h (0.57 ± 0.06; ****p < 0.0001 vs. 1-h time point). The degradation rate slowed down following the first 24 h but continued its progression through 7 days. Fluorescein signal for each aggregate was normalized to its corresponding value at 1-h time point. n = 29 aggregates with diameter ≥10 μm were quantified in three independent experiments. See Table S6 for statistics. (M–O) As a control for aggregate stability in the absence of microglial cells, dishes with aggregates alone were incubated for 1 h (M), 24–72 h, and 7 days (N) in complete DMEM at 37°C and 5% CO₂. Fluorescence intensity did not significantly change over the incubation period (0.87 ± 0.05 at 7 days; p = 0.14 vs. 1-h time point) (O). Fluorescein signal for each field was normalized to its corresponding value at 1-h time point. n = 3 dishes imaged (47 fields acquired) in two independent experiments. See Table S7 for statistics and Figure S2 for images corresponding to intermediate 1-h, 24-h, 48-h, 72-h, and 7-day time points in absence of microglia. For all quantifications, intracellular fluorescein signal was eliminated by a mask generated using the transmitted light channel. For all plots, bars indicate ±SEM. The mean fluorescein intensity calculated for each time point was compared to a theoretical mean value of 1 assigned to the 1-h time point using the one-sample Student’s t test (K and O) or the one-sample Wilcoxon signed rank test (L). Scale bars: 10 μm.

Figure 4.. Inhibition of the PI3K-AKT pathway reduces lysosome exocytosis and actin polymerization during digestive exophagy

(A–C and G) Primary murine microglial cells with lysosomes loaded with fluorescein-biotin-dextran were treated with TGX-221 (PI3K1β inhibitor), AKT1/2i (AKT1/2 inhibitor), or control DMSO for 1 h, followed by incubation with streptavidin-Alexa 647-Aβ_1–42 aggregates for 90 min and fixation. The intracellular dextran was removed by permeabilizing the cells with Triton X-100 in the presence of excess biotin and BSA, and the preparations were imaged by confocal microscopy. Representative images show the effect of DMSO (A), TGX-221 (B), and AKT1/2i (C) on the secretion of fluorescein-biotin-dextran toward the aggregates. Side panels show fluorescein and Alexa 647 (aggregates) signal individually for the region visualized. Fluorescein co-localized with Alexa 647-Aβ_1–42 aggregates was quantified and normalized to control DMSO condition. Compared with DMSO control (1 ± 0.16), cells treated with TGX-221 exocytosed 51% less dextran (0.49 ± 0.07; **p < 0.01), and cells treated with AKT1/2i exocytosed 69% less dextran (0.31 ± 0.08; ***p < 0.001) (G). n = 8 dishes imaged (45 fields acquired) plotted as large geometrical shapes or small gray triangles, respectively, for the DMSO condition, and n = 9 dishes imaged for the TGX-221 and AKT1/2i conditions (43–45 fields acquired) in three independent experiments. See Table S8 for statistics. (D–F and H) Primary murine microglial cells were treated with TGX-221, AKT1/2i, or control DMSO for 1 h. After treatments, the cells were incubated with streptavidin-Alexa 647-Aβ_1–42 aggregates for 60 min, followed by fixation and staining with Alexa 488-phalloidin (green) to label polymerized actin. Representative images show the effect of DMSO (D), TGX-221 (E), and AKT1/2i (F) on actin polymerization in regions of contact (arrowheads) between microglial cells and aggregates. Side panels show Alexa 488 (phalloidin) and Alexa 647 (aggregates) signal individually for the region visualized. Alexa 488 fluorescence co-localized with Alexa 647-Aβ_1–42 aggregates was quantified and normalized to control DMSO condition. Compared with DMSO control (1 ± 0.09), cells treated with TGX-221 polymerized 36% less actin (0.64 ± 0.12; *p = 0.049), and cells treated with AKT1/2i polymerized 53% less actin (0.47 ± 0.11; **p < 0.01) at the compartments (H). n = 8 dishes imaged (45–46 fields acquired) for all conditions, in three independent experiments. See Table S9 for statistics. Bars indicate ±SEM. Statistical differences between fluorescein or Alexa 488 fluorescence means calculated for each time point were assessed using unpaired one-way ANOVA followed by Šídák’s multiple comparison test. Scale bars: 20 μm. Fluorsc., fluorescein; A488, Alexa Fluor 488; A647, Alexa Fluor 647.

Figure 5.. Microglia form compartments filled with extracellular space positive for acid phosphatase activity in close proximity or adjacent to Aβ deposits in 5xFAD mouse brain

(A–D) Scanning EM micrographs of brain sections from a 13-month-old 5xFAD male showing microglia contacting fibrillar Aβ plaques. A microglia cell, highlighted in green (A), contacts a fibrillar Aβ deposit and forms several pockets of extracellular space directly adjacent to the deposit (inset B, red arrowheads). When sections of the same animal were stained for the lysosomal enzyme acid phosphatase, similar pockets of extracellular space directly in contact with Aβ appeared stained with abundant electron-dense material indicative of acid phosphatase activity (C and inset expanded in D, red arrowheads). (E and F) Scanning EM micrographs of brain sections from a 15-month-old 5xFAD female stained for acid phosphatase showing a fragmented Aβ deposit being contacted by a microglial cell (E) forming pockets of extracellular space in contact with Aβ, positive for acid phosphatase activity (F, red arrowheads). (G–I) Scanning EM micrographs of 8-month-old 5xFAD female brain sections stained for acid phosphatase, showing microglia contacting Aβ deposits (G). The cells established compartments positively stained for acid phosphatase activity (G and insets expanded in H and I, red arrowheads). (J–M) Scanning EM micrographs of an 8-month-old 5xFAD male showing microglia contacting Aβ deposits surrounded by dystrophic neurites. Some sections were stained for phosphatase in absence of β-glycerophosphate to control for any unspecific staining by lead nitrate (J). The inset of (J) shows contact areas with pockets of extracellular space adjacent to fibrillar material, devoid of staining (K, red arrowheads). When matched sections of the same animal were stained for acid phosphatase in the presence of β-glycerophosphate (L and M), similar pockets of extracellular space appeared loaded with electron-dense material indicative of acid phosphatase activity (inset expanded in M, red arrowheads). The panels correspond to four different 5xFAD animals (two males and two females). Sections were prepared and imaged in four independent technical replicates. See Figures S3 and S4 for additional comparisons between 5xFAD brain regions stained for acid phosphatase in the presence or absence of β-glycerophosphate or with Tris maleate buffer alone. Green pseudo-coloring: microglia cell body (A). mg, microglia; Aβ, amyloid-beta deposit; dn, dystrophic neurite. Scale bars: 3 μm (A, C, E, G, J, and L) and 0.5 μm (B, D, F, H, I, K, and M).

Figure 6.. Microglial cells secrete intralysosomal Aβ nanofibrils toward extracellular Aβ aggregates during digestive exophagy

(A–C) Microglial cell lysosomes were loaded with fluorescein-biotin-dextran overnight followed by a 4-h chase and treatment with PMA, a PKC activator that increases lysosomal exocytosis, or an equivalent volume of DMSO for 1 h. Next, cells were co-incubated with streptavidin-coated Aβ_1–42 aggregates for 90 min in the presence of PMA or DMSO, followed by fixation and permeabilization with Triton X-100 to remove any remaining dextran. Lysosome exocytosis of fluorescein-biotin-dextran toward aggregates was quantified by measuring fluorescein signal deposited on them. (A and B) Single-plane confocal micrographs showing lysosome exocytosis of dextran by microglial cells toward Aβ_1–42-Alexa 647 aggregates (arrowheads). The cells were pretreated and maintained with either DMSO (A) or 1.6 μM PMA (B) in complete DMEM. (C) Quantification of fluorescein-dextran intensity co-localized with aggregates per cell, normalized to DMSO control, and plotted for each dish and field imaged. PMA treatment increased lysosomal dextran exocytosis toward aggregates by more than 2-fold (2.50 ± 0.36, **p < 0.01) relative to DMSO control (1 ± 0.17). n = 9 dishes imaged (41–42 fields acquired) for each condition, plotted as large geometrical shapes or small gray triangles, in three independent experiments. See Table S10 for statistics. Bars indicate ±SEM. Statistical differences between mean fluorescein intensities calculated for each condition were assessed using the unpaired two-tailed Student’s t test. (D and E) Aβ nanofibrils labeled with fluorescein were prepared by sonication of larger aggregates, yielding fibrils with lengths ranging between 10 and 100 nm (D) and 100 and 500 nm (E). (F and G) Live confocal imaging of nanofibril exocytosis by microglial cells toward large Aβ aggregates. The cells were preincubated overnight with nanofibrils (green). The vast majority of the fibrils were less than 200 nm in length, which would allow them to be endocytosed. After incubation and a brief chase in order to ensure nanofibril localization in lysosomes, loaded cells were co-incubated with extracellular Alexa 647-Aβ aggregates for 45–180 min in serum-free medium and imaged live by confocal microscopy. The overlap between exocytosed fluorescein-Aβ nanofibrils and extracellular Alexa 647-Aβ aggregates appears as orange areas (arrowheads in the corresponding xz and yz orthogonal planes for each panel). Monochrome orthogonal planes indicate fluorescein signal only, corresponding to fluorescein-nanofibrils exocytosed toward Aβ aggregates. 1.6 μM PMA treatment for 1 h led in some instances to increased nanofibril secretion toward aggregates (G) relative to DMSO-treated cells (F). In some cases, PMA treatment rearranged the intralysosomal fluorescein distribution (G). Eight or nine dishes of DMSO-treated and PMA-treated microglial cells were imaged in four independent experiments. Scale bars: 10 μm (A, B, F, and G) and 100 nm (D and E). Fluorsc., fluorescein; A647, Alexa Fluor 647.