Peripherally derived angiotensin converting enzyme-enhanced macrophages alleviate Alzheimer-related disease

Abstract

Targeted overexpression of angiotensin-converting enzyme (ACE), an amyloid-β protein degrading enzyme, to brain resident microglia and peripheral myelomonocytes (ACE10 model) substantially diminished Alzheimer's-like disease in double-transgenic APPSWE/PS1ΔE9 (AD+) mice. In this study, we explored the impact of selective and transient angiotensin-converting enzyme overexpression on macrophage behaviour and the relative contribution of bone marrow-derived ACE10 macrophages, but not microglia, in attenuating disease progression. To this end, two in vivo approaches were applied in AD+ mice: (i) ACE10/GFP+ bone marrow transplantation with head shielding; and (ii) adoptive transfer of CD115+-ACE10/GFP+ monocytes to the peripheral blood. Extensive in vitro studies were further undertaken to establish the unique ACE10-macrophage phenotype(s) in response to amyloid-β1-42 fibrils and oligomers. The combined in vivo approaches showed that increased cerebral infiltration of ACE10 as compared to wild-type monocytes (∼3-fold increase; P < 0.05) led to reductions in cerebral soluble amyloid-β1-42, vascular and parenchymal amyloid-β deposits, and astrocytosis (31%, 47-80%, and 33%, respectively; P < 0.05-0.0001). ACE10 macrophages surrounded brain and retinal amyloid-β plaques and expressed 3.2-fold higher insulin-like growth factor-1 (P < 0.01) and ∼60% lower tumour necrosis factor-α (P < 0.05). Importantly, blood enrichment with CD115+-ACE10 monocytes in symptomatic AD+ mice resulted in pronounced synaptic and cognitive preservation (P < 0.05-0.001). In vitro analysis of macrophage response to well-defined amyloid-β1-42 conformers (fibrils, prion rod-like structures, and stabilized soluble oligomers) revealed extensive resistance to amyloid-β1-42 species by ACE10 macrophages. They exhibited 2-5-fold increased surface binding to amyloid-β conformers as well as substantially more effective amyloid-β1-42 uptake, at least 8-fold higher than those of wild-type macrophages (P < 0.0001), which were associated with enhanced expression of surface scavenger receptors (i.e. CD36, scavenger receptor class A member 1, triggering receptor expressed on myeloid cells 2, CD163; P < 0.05-0.0001), endosomal processing (P < 0.05-0.0001), and ∼80% increased extracellular degradation of amyloid-β1-42 (P < 0.001). Beneficial ACE10 phenotype was reversed by the angiotensin-converting enzyme inhibitor (lisinopril) and thus was dependent on angiotensin-converting enzyme catalytic activity. Further, ACE10 macrophages presented distinct anti-inflammatory (low inducible nitric oxide synthase and lower tumour necrosis factor-α), pro-healing immune profiles (high insulin-like growth factor-1, elongated cell morphology), even following exposure to Alzheimer's-related amyloid-β1-42 oligomers. Overall, we provide the first evidence for therapeutic roles of angiotensin-converting enzyme-overexpressing macrophages in preserving synapses and cognition, attenuating neuropathology and neuroinflammation, and enhancing resistance to defined pathognomonic amyloid-β forms.

Keywords: EEA1; IGF1; TNFα; TREM2; innate immunity.

Figure 1

Beneficial effects of partial ACE¹⁰ bone marrow transplantation in APP_SWE/PS1_ΔE9 transgenic mice. (A) Schematic illustration of experimental design and timeline. At 2 months of age, three groups of double-transgenic APP_SWE/PS1_ΔE9 (AD⁺) recipient mice (n = 22) underwent an irradiation procedure, with their heads shielded by a 3-inch-thick lead block. Mice then received a lethal dose of 950 rad irradiation for 11 min (Supplementary Fig. 1A–C). Immediately after irradiation, AD⁺ recipient mice received an intravenous (i.v.) injection of 5 million bone marrow cells from one of three types of donor mice: GFP⁻AD⁺ACE^WT, GFP⁺AD⁻ACE^WT, or GFP⁺AD⁻ACE¹⁰. The resulting BM^WT::AD⁺, BM^ACE10::AD⁺, and BM^AD+::AD⁺ chimeric mice (n = 7–8 per group) had a partial (˜30–40%) bone marrow transplantation. The two groups of GFP-labelled bone marrow chimeric AD⁺ mice underwent retinal imaging at 7 months of age. Tissues were harvested for analysis from all three experimental groups when the animals were 8 months of age; the left hemisphere was used for flow cytometry, the right hindbrain for histology, and the right forebrain for protein analysis. (B) Representative fluorescence micrograph of the BM^WT::AD⁺ chimeric mouse brain showing infiltrating monocytes (GFP⁺/CD45^hi; arrows) near a 6E10⁺ amyloid-β plaque site. Scale bar = 20 μm. (C) Quantitative IHC analysis of 6E10⁺ amyloid-β plaque areas in brain cortices (Ctx) of BM^WT::AD⁺, BM^ACE10::AD⁺, and BM^AD+::AD⁺ chimeric mice (n = 5–7 mice/group). (D) Sandwich ELISA analysis of cerebral soluble amyloid-β₄₂ levels in AD⁺ chimeric mice (n = 6–8 mice/group). (E) Quantitative IHC analysis of the cortical GFAP⁺ reactive astrocyte cell count in AD⁺ chimeric mice (n = 6–8 mice/group). (F) Quantitative IHC analysis of the cortical GFAP⁺ reactive astrocyte area in AD⁺ chimeric mice (n = 7 mice/group). (G) Representative micrographs of cortical amyloid-β plaques stained with curcumin in BM^WT::AD⁺ versus BM^ACE10::AD⁺ chimeric mice. (H) Representative micrographs of cortical GFAP⁺ astrocytes in the same experimental groups. Scale bar = 50 μm (for G and H). (I) Flow cytometry analysis for cerebral infiltrating GFP⁺ monocytes in AD⁺ chimeric mice. Combinations of markers were used to identify microglia (CD11b⁺CD45^int.-lowGFP) versus infiltrating monocyte subsets (CD11b⁺CD45^hiLy6C^hiGFP⁺). (J) Representative flow cytometry map analysis showing increased infiltration of CD11b^hiCD45^hiLy6C^hiGFP⁺ ACE¹⁰ versus wild-type monocytes in AD⁺ chimeric mice. (K and L) Quantitative flow cytometry analyses of cerebral inflammatory biomarkers in chimeric AD⁺ mice following partial bone marrow transplantation. (K) Percentage of GFP⁺CD11b^hiCD45^hiLy6C^hiF4/80^low monocytes (n = 5–6 mice/group), (L) TNFα expression in infiltrating GFP⁺CD11b⁺CD45^hiLy6C^int-hiF4/80^hi Mφ (n = 4–5 mice/group). (M) Fluorescent micrographs of cortical brain regions stained against IGF1 (red) combined with Iba1 (white) CD45 (green) and nuclei (blue) in BM^WT::AD⁺ versus BM^ACE10::AD⁺ chimeric mice. Scale bars = 10 μm. Brain-infiltrating Iba1⁺CD45^hi Mo/Mφ abundantly expressed IGF1 in BM^ACE10::AD⁺ chimeric mice (bottom). (N) Quantitative IHC of % IGF1⁺ area co-localized with Mo/Mφ in BM^WT::AD⁺ versus BM^ACE10::AD⁺ chimeric mice (n = 7 mice/group). (O–Q) Representative non-invasive retinal fluorescence imaging of infiltrating GFP⁺ bone marrow cells in living AD⁺ chimeric mice. (O) A control AD⁺ mouse that underwent bone marrow transplantation of GFP⁻BM^AD+ (GFP⁻BM^AD+::AD⁺). (P) Retinal fluorescent images from a GFP⁺BM^WT::AD⁺ mouse (P’, enlarged image). (Q) Retinal fluorescent images from a GFP⁺BM^ACE10::AD⁺ mouse (arrows indicate GFP⁺ bone marrow cells). (R) Representative microscopic image obtained during the histological examination of a retina extracted from a BM^ACE10::AD⁺ mouse whose retina had been previously imaged in vivo; ex vivo staining with GFP (green), CD45 (red), 4G8 (cyan), and DAPI (blue), validating the homing of GFP⁺ACE¹⁰ monocytes to retinal amyloid-β plaques. Scale bar = 5 μm. Data from an individual mouse (filled circles for males and clear circles for females) as well as group mean ± SEM are shown. Fold increase and percentage decreases compared to control groups are shown in green. *P < 0.05, **P < 0.01. ns = non-significant, using one-way ANOVA and Bonferroni’s post-test for three group comparisons. For two-group comparisons, paired or unpaired two-tailed Student t-tests were used. Aβ = amyloid-β; FS = forward scatter; SS = side scatter.

Figure 2

Cognitive preservation and restricted pathology following adoptive transfer of a CD115⁺ ACE¹⁰ monocyte subset in AD⁺ mice transgenic mice. (A–C) Schematic representation of experimental procedure and treatment groups. (A) CD115⁺ monocytes (Mo^BM) were isolated from the bone marrow of GFP⁺ donor mice and enriched by MACS microbeads and an anti-CD115 antibody column sorting procedure. (B) Mo^BM were then intravenously (i.v.) injected into the tail vein of AD⁺ recipient mice (n = 8 mice/group, all males). (C) Additional control groups included naïve wild-type (WT) mice and AD⁺ mice injected with PBS (n = 7 mice/group, all males). (D) Schematic timeline of the in vivo preclinical experiment. Pre-symptomatic AD⁺ mice exhibiting neuropathology at 8 months of age received monthly injections of 5–6 million GFP⁺Mo^WT or GFP⁺Mo^ACE10 or PBS for 3 months (immunization regimen indicated by green arrows). At 11 months, mice underwent behavioural tests followed by tissue collection and analysis when 12 months of age. (E and F) Open field test in all AD⁺ treatment and naïve wild-type groups measuring: (E) ambulatory and (F) rearing activity. (G–K) Cognitive functions assessed by the Barnes maze test in both monocyte-treated groups as compared to the control PBS-injected group and naïve cognitively normal wild-type group (n = 6–8 mice/group). Incorrect entries (errors) for the following: (G) acquisition-training phase (Days 1–4), (H) memory retention (Day 7), (I) reversal phase (Days 8 or 9), and (J and K) memory test at Day 9. (J) Errors and (K) escape latency (s). (L) Quantitative IHC analysis of 6E10⁺ amyloid-β plaque areas in the hippocampus (HC), cingulate cortex (CC), and total brain in Mo^WT- and Mo^ACE10-treated versus PBS-injected AD⁺ mice (n = 7–8 mice/group). (M) CAA score assessed as vascular Thio-S⁺ in AD⁺ mice (n = 6–8 mice/group). Data from an individual mouse as well as group mean ± SEM are shown. (N) Quantitative IHC analysis of hippocampal pre-(VGluT1⁺) and postsynaptic (PSD95⁺) areas in Mo^ACE10-treated mice compared to PBS-injected control AD⁺ and naïve wild-type mice (n = 6–8 mice/group). Data from individual mice, lower and upper quartiles (as lower and upper horizontal lines in box), median (midline within box), and minimum/maximum values (whiskers), are shown. Percentage increase and decrease compared to control groups are shown in green. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns = non-significant, by two-way or one-way ANOVA and Bonferroni’s post-test. Significance between two-groups by unpaired two-tailed Student t-test. IR = immunoreactive.

Figure 3

Enhanced surface recognition and uptake of diverse amyloid-β₄₂ conformers by ACE¹⁰ macrophages. (A) Design and timeline of in vitro. Bone marrow was isolated from young wild-type (WT) and ACE¹⁰ mice (8–12 weeks old) and cultured for 6 days in MCSF-enriched media, generating primary macrophage cultures (Mφ^WT and Mφ^ACE10, respectively). On Day 7, Mφ were either unstimulated (Un) or stimulated by one of the three amyloid-β forms: fibrils (fAβ₄₂), prion rod-like structures (pAβ₄₂), or cross-linked oligomers (oAβ₄₂). Mφ^ACE10 were analysed and compared to control Mφ^WT for cell surface amyloid-β₄₂ binding (pre-chilled on ice; binding – 0 min) and receptor expression as well as uptake of amyloid-β₄₂ species at incremental time points from 5 min up to 60 min (incubated at 37°C; uptake: 5, 15, 30 min). Cells and media were isolated and further analysed using various methods including ICC, flow cytometry (FC), and quantitative protein analysis. (B) Forms of non-cross-linked (non-XL) and cross-linked (oAβ) amyloid-β₄₂. Oligomers as large as an octamer are observed. (C) Negative stain TEM micrographs of amyloid-β₄₂ assemblies. Left: Fibrillar amyloid-β (fAβ₄₂) displayed long fibrils of ˜10 nm diameter with a twisted morphology typically comprising two smaller filaments. Middle: Prion rod-like amyloid-β₄₂ (pAβ₄₂) displayed short (<300 nm) straight rods comprising multiple filaments (blue arrows). Narrower (˜5 nm) rod-like structures appeared to comprise twisted double filaments (black arrows). The assembly indicated by the red arrow is shown in the inset at higher magnification. Right: Cross-linked oligomers (oAβ₄₂) displayed mainly quasi-spherical assemblies (white arrows) with diameters of 5–8 nm. The assembly indicated by the red arrow is shown in the inset at higher magnification. A rare rod-like structure (yellow asterisk) was also observed. Scale bars = 100 nm or 50 nm (insets). (D) Representative fluorescence micrograph of Mφ^ACE10 prestimulated with fibril amyloid-β₄₂ on ice (at the 0-min time point) and immunolabelled for Scara1 (red), human amyloid-β (6E10; green), and nuclei (blue). Abundance of co-surface labelling of Scara1⁺ and fibril amyloid-β₄₂ (yellow spots) on the Mφ^ACE10 cell body and processes was detected, indicating large amounts of fibril amyloid-β₄₂ bound to surface Scara1. (E) Quantitative ICC analysis of a 6E10⁺ immunoreactive (IR) area for cell surface binding of the three amyloid-β₄₂ species (250 nM each) at the 0-min time point in Mφ^ACE10 versus Mφ^WT. (F) Quantitative ICC analysis of surface Scara1 expression in ACE¹⁰ versus wild-type Mφ that were either untreated or in the presence of different amyloid-β₄₂ forms at the 0-min time point. (G) Quantitative ICC analysis of TREM2 expression in ACE¹⁰ versus wild-type Mφ following prestimulation by the different forms of amyloid-β₄₂ (250 nM each). (H) Representative fluorescence micrograph of Mφ^WT and Mφ^ACE10 cells that were incubated for 5 min with 250 nM fibril amyloid-β₄₂ and later immunolabelled for Scara1 (red), human amyloid-β (6E10; green), and nuclei (blue). Amyloid-β₄₂ fibrils are seen in intracellular vesicles within these phagocytic cells (white arrow). (I) Panel of representative fluorescence micrographs of Mφ^WT and Mφ^ACE10 cells following a 15-min incubation with three different amyloid-β₄₂ forms immunolabelled for scavenger receptor type B CD36 (purple), human amyloid-β (6E10; green), and nuclei (blue). (I’) Separated single channel (cy2) for intracellular amyloid-β₄₂ uptake. (J) Quantitative ICC analysis of CD36 expression (μm²/cell number) in ACE¹⁰ versus wild-type Mφ that were either untreated or exposed to different amyloid-β₄₂ forms (250 nM) for 15 min. (K) Quantitative ICC analysis of 6E10⁺ immune-reactive area (μm²/cell number) for intracellular uptake of the three amyloid-β₄₂ species (250 nM) for 15 min by Mφ^ACE10 versus Mφ^WT. (L and M) Quantitative ICC analysis of (L) CD36 expression and (M) amyloid-β uptake by untreated versus ACE-inhibitor pretreated (1 μM lisinopril overnight in RPMI media + 0.3% BSA) Mφ^ACE10, following exposure to 100 nM fibril amyloid-β₄₂ for either 0, 5 or 30 min. (N–Q) Individual correlation analysis of the following: (N and O) surface fibril amyloid-β₄₂ and Scara1 immunoreactive area at the binding time point in (N) Mφ^WT and (O) Mφ^ACE10 (n =15 wells); (P and Q) surface fibril/oligomer amyloid-β₄₂ and CD36 expression in Mφ^ACE10 following a 15-min exposure to either fibril amyloid-β₄₂ or oligomer amyloid-β₄₂. (R) Representative peak analysis in percentage mean fluorescence intensity (MFI) of Mφ^ACE10 uptake of HiLyte Fluor 647-labelled fibril amyloid-β₄₂ at the 0-, 5-and 15-min time points. (S) Representative flow cytometry images of HiLyte Fluor 647-labelled fibril amyloid-β₄₂ MFI at the 15-min time point by F4/80⁺-Mφ^WT versus -Mφ^ACE10. (T and U) Quantitative flow cytometry analysis of HiLyte Fluor 647-labelled fibril amyloid-β₄₂ MFI uptake at the 0-, 5- and 15-min time points by (T) Mφ^WT or Mφ^ACE10 and by (U) untreated and lisinopril pretreated Mφ^ACE10. The group mean ± SEM as well as the fold reduction or increase and percentage decrease between groups are shown. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns = non-significant, by two-way or one-way ANOVA and Bonferroni’s or Sidak’s post-test, while asterisks in parentheses signify a two-group comparison using the unpaired two-tailed Student t-test.

Figure 4

Intra- and extracellular degradation of pathological amyloid-β₄₂ species by ACE¹⁰ macrophages. (A) Representative fluorescence micrograph of unstimulated (top) and fibril amyloid-β₄₂ stimulated (bottom) Mφ^WT immunolabelled for human amyloid-β (6E10; green), autophagy marker Beclin-1 (red), and nuclei (blue). Merged channel panels show that in response to fibril amyloid-β₄₂, Beclin-1 protein moves to the nuclei (white arrows). (B) Representative fluorescence micrograph panel of fibril amyloid-β₄₂ (fAβ₄₂) stimulated Mφ^WT immunolabelled for human amyloid-β (6E10; green), endosomal markers (EEA1, Lamp-1, Lamp-2, red), and nuclei (blue). (C and D) Representative fluorescence micrograph of Mφ^WT stimulated with the different fibril , prion rod-like and oligomer amyloid-β₄₂ forms for 15 min and immunolabelled for human amyloid-β (6E10; green), endosomal marker EEA1 (red), surface scavenger receptor CD36 (purple), and nuclei (blue). White arrows indicate co-localizing of intracellular amyloid-β₄₂ within EEA1⁺ vesicles. (D) Higher magnification image of co-localized fibril amyloid-β₄₂ within EEA1⁺ endosomes in Mφ^WT (yellow spots). (E and F) Representative fluorescence micrograph of Mφ^ACE10 exposed to oligomer amyloid-β₄₂ (100 nM for 15 min) and immunolabelled for human amyloid-β (6E10; green), endosomal marker EEA1 (red), surface scavenger receptor CD36 (purple) and nuclei (blue). White arrows indicate 6E10⁺ and EEA1⁺ co-localization. (F) Higher magnification image of co-localized oligomer amyloid-β₄₂ within EEA1⁺ endosomes in Mφ^ACE10 (white arrows). (G and H) Quantitative analysis of co-localized 6E10⁺ and EEA1⁺ immunoreactive area in unstimulated and fibril, prion rod-like and oligomer amyloid-β₄₂-stimulated Mφ^WT or Mφ^ACE10: (G) 6E10-IR area per macrophage and (H) count of 6E10⁺/EEA1⁺ vesicles using puncta count per macrophage. (I) Quantitative normalized analysis of percentage reduction of extracellular fibril amyloid-β₄₂ in media of Mφ^WT versus Mφ^ACE10 over 18, 20 and 24 h. (J) Quantitative ELISA analysis of total intra- and extracellular amyloid-β₄₂ levels in the media and cell fraction of Mφ^ACE10, following fibril amyloid-β₄₂ for 30 min and either untreated, lisinopril pretreated (1 μM lisinopril overnight in RPMI media + 0.3% BSA), or pre-placed in ice bath. The group mean ± SEM as well as fold increase and percentage decrease between groups are shown. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns = non-significant, by two-way or one-way ANOVA and Bonferroni’s post-test.

Figure 5

ACE overexpression in macrophages promotes a pro-healing, alternatively activated immune profile. (A) Fluorescence micrographs of inducible nitric oxide synthase (iNOS, red), scavenger receptor CD36 expression (cyan), and nuclei (blue) in Mφ^ACE10 versus Mφ^WT (merged image on right). (B–D) Quantitative analysis showing (B) iNOS, (C) arginase-1, and (D) MMP9 in the immunoreactive area of unstimulated (Un) and fibril amyloid-β₄₂-stimulated Mφ^ACE10 and Mφ^WT at the 0- and 5-min time points. (E) Representative fluorescence micrographs of Mφ^WT and Mφ^ACE10 immunolabelled for MMP-9 (red) and nuclei (blue). (F) Representative bright-field images showing a morphological difference in the cell process length of Mφ^ACE10 versus Mφ^WT with measurements of elongation (red lines). (G) Quantitative analysis of elongation factor [ratio between total cell length and soma width (left) and percentage cell elongation (right) in unstimulated Mφ^WT versus Mφ^ACE10]. (H) Elongation factor in Mφ^WT versus Mφ^ACE10 stimulated with fibril amyloid-β₄₂ for 0, 5, and 30 min. (I–L) Quantitative MSD protein assay of unstimulated (−) and stimulated with fibril amyloid-β₄₂ [100 nM for 24 h (+)] Mφ^WT versus Mφ^ACE10 for the following cytokines: (I) TNFα, (J) IL6, (K) CXCL1, and (L) IL1β. (M) Quantitative analysis of a cell viability MTT assay in Mφ^WT versus Mφ^ACE10after 24 h of exposure to 1 μM fibril amyloid-β₄₂ (n = 3 mice/genotype group, n = 6 wells/group; one-tailed Student t-test). Data from individual mouse, group mean ± SEM as well as fold increase and percentage decrease and increase between groups are shown. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns = non-significant, by two-way or one-way ANOVA and Bonferroni’s post-test, while asterisks in parentheses signify a two-group comparison using the unpaired two-tailed Student t-test.