Introduction
Alzheimer’s disease (AD) is the most common form of aging-associated dementia and is clinically diagnosed by cognitive deficits [
1]. Over the past three decades of AD research, the most widely studied “amyloid cascade hypothesis” enters on the sequential pathological changes that abnormal accumulation of β-amyloid peptides (Aβ) is the prominent early event, leading to neuronal abnormalities, development of senile (neuritic) plaques, changes in tau phosphorylation and the emergence of various cognitive symptoms [
2]. Comparative studies on cognition in AD patients, as well as the presence of AD biomarkers and neuroimaging studies, demonstrate a pre-clinical phase of 10–20 years that precedes the manifestation of symptomatic AD [
3‐
5]. Specifically, the toxic form of oligomeric Aβ during the pre-clinical phase induces local toxicity that advances both intra- and extra-neuronal pathologies such as neurofibrillary tangles, dystrophic neurites (DNs), mitochondrial degeneration, and glial dysfunction [
3,
6,
7]. The accumulation of diversified proteins and organelles in DNs surrounding amyloid plaques correlates with the dysfunction of various neuronal processes during disease progression [
8‐
15]. Because the pre-clinical stage precedes symptomatic disease onset by a decade or more [
16‐
18], it is critical to understand exactly how and when Aβ exerts harmful effects during the development of amyloid plaques in AD brains.
One critical pathological feature in AD brains is the presence of DNs, which are recognized as swollen bulbous or ring-like neuritic processes [
19‐
21]. DNs can be detected by antibodies specific to autophagy proteins, tubular ER proteins, ubiquitin, neurofilament, phosphorylated tau, and amyloid precursor protein (APP). Recently, we have shown that proteins important for several essential cellular homeostasis networks, such as the autophagy-endosomal system, ER tubulation, and ubiquitin proteasome machinery, are abnormally accumulated in DNs at different time points in three distinguishable layers during Aβ plaque growth [
11]. An early autophagy protein ATG9A is found in the 1st layer of DNs at the initial stage of Aβ plaque development. This is followed by the accumulation of ER tubules, as detected by anti-reticulon-3 (RTN3), and clustered mitochondria in the 2nd layer. The 3rd and outer-most layer of DNs contains the late autophagy/endosomal proteins such as RAB7 and LC3.
Lysosomes are major cellular degradative organelles that contain ~ 25 membrane proteins and more than 60 soluble proteins in lumen [
22]. The majority of lysosomal luminal proteins are acid hydrolases, which are actively involved in the degradation and recycling of several classes of macromolecules such as proteins, lipids, polysaccharides, and nucleic acids, delivered through autophagy, endocytosis, or other cellular trafficking pathways [
22,
23]. Two luminal proteins, lysosomal luminal membrane proteins 1 and 2 (LAMP1 and LAMP2) are routinely used to label endo-lysosomal vesicles and are entrapped in DNs [
15,
24]. Since autophagy and endo-lysosomes are integrated cellular homeostasis processes, it is our interest to understand how lysosomes participate in the formation of DNs in AD brains. In this study, we monitored lysosomal accumulation in DNs during amyloid plaque growth by examining brains from three different AD mouse models: APP knock-in (APP
NL-G-F), 5xFAD and APP/PS1ΔE9 (PA) mice. LAMP1 and the lysosomal activator/lysosomal restricted proteins, saposins (SAPs), were chosen to define the lysosomal organelles. Our results show that cathepsin-deficient, LAMP1
+- and SAP-C
+-primary lysosomes were massively accumulated in the 1st layer of DNs during the initial stage of plaque formation. Notably, the intensity of DN-like lysosomal clusters declined during plaque growth, and diminished more in older AD mouse brains, in which LAMP1 was found mainly in the active lysosomes from microglia, specifically the disease-associated microglia (DAM). Similarly, SAP-C
+-DNs were also diminishing during plaque growth and were nearly absent at the late stage of AD mouse and AD patients’ brains. Our data suggest that continuing intracellular or extracellular Aβ insults cause early accumulation of lysosomes in DNs and sequential impairments in lysosomal structure and functions in neurons, as reflected by the gradual loss of immuno-reactivity of lysosomal proteins in AD mouse brains. The impaired lysosomal systems may viciously facilitate growth of amyloid plaques. Preserving lysosomal functions early is therefore a potential therapeutic strategy for AD intervention.
Materials and methods
Mouse strains and AD postmortem brain tissues
APP
NL-G-F mice were obtained from the RIKEN Center for Brain Science, Wako, Japan [
25]. 5xFAD and Tg-APPsw/PSEN1DE9 (PA) mice were purchased from Jackson Laboratory (stock # 34840 and 004462, respectively). All mice in the study were maintained and used according to protocols approved by the Institutional Animal Care and Use Committee of the University of Connecticut. AD postmortem brain samples were obtained from the NIH NeuroBioBank (Harvard Brain Tissue Resource Center; Human Brain and spinal Fluid Resource Center, Los Angeles; Mount Sinai NBTR Tissue Distribution; University of Maryland Brain and Tissue Bank; University of Miami Brain Endowment Bank).
Immunohistochemistry and immunofluorescent confocal microscopy
A standard method of immunohistochemistry and immuno-confocal experiments was performed as previously described [
26]. Briefly, mouse brains were dissected and fixed with 4% paraformaldehyde fixation for 12 h and immersed in 20% sucrose overnight at 4 °C. The fixed brain tissue was then sectioned in the sagittal plane at a 14 μm thickness using a cryostat after O.C.T. compound embedding. Brain sections were stored at − 20 °C. After three washes with phosphate-buffered saline (PBS), the sections were permeabilized with 0.3% Triton X-100 for 30 min and rinsed in PBS three times to remove detergent. For 3,3′-diaminobenzidine (DAB) staining, 0.3% H2O2 was added with 0.3% Triton X solution. After being rinsed in PBS two times to remove detergent, antigen retrieval was then performed by heating in a microwave in 0.05 M citrate-buffered saline, pH 6.0, for 2-3 min. The sections were then blocked with 5% normal goat serum and incubated with individual primary antibodies: 6E10 (Covance Research Products Inc. Cat# SIG-39330-200, RRID: AB_662804), ATG9A (Abcam Cat# ab108338, RRID:AB_10863880), β-Actin (Sigma Aldrich Cat# A2228, RRID:AB_476697), ATG9A (Abcam Cat# ab108338 RRID:AB_10863880), β-Galactosidase
(Thermo Fisher Scientific Cat# A-11132, RRID:AB_221539), β-Glucosidase (Santa Cruz Cat# sc-166,407, RRID:AB_2109068), Calnexin (Sigma Aldrich Cat# C4731, RRID:AB476845), Cathepsin B (Santa Cruz Cat# sc-377,299, RRID:AB_10842446), Cathepsin D (Santa Cruz Cat# sc-377,299, RRID:AB_2827539), EEA1(Millipore Cat# 07–1820, RRID:AB_10615480), GFAP (Thermo Fisher Scientific Cat# 13–0300, RRID:AB_2532994), GFAP/SMI22 (Sigma-Aldrich Cat# G3893, RRID:AB_477010), Iba1 (Wako Cat# 019–19,741, RRID:AB_839504), LAMP1(Abcam Cat# ab24170-rabbit, RRID:AB_775978), LAMP1(Abcam Cat# ab25245-rat, RRID:AB_449893), LAMP2 (Abcam Cat# ab25339, RRID:AB_470455), LPL (Abcam Cat# ab93898, RRID:AB_10562464), Neurofilament L (Millipore Cat# AB9568, RRID:AB_11213875), RAB4A (Santa Cruz Cat#sc-517,263, RRID:AB_2177555), RAB4B (Santa Cruz Cat# sc-271,982), RAB5 (Abcam Cat# ab109534, RRID:AB_10865740), RAB6A (Santa Cruz Cat# sc-81,913, RRID:AB_1128894), RAB7 (Abcam Cat# ab137029, RRID:AB_2629474), RAB9A (Santa Cruz Cat# sc-53,145), RAB11(Abcam Cat# ab95375, RRID:AB_10688715), RTN3 monoclonal (recently developed in the Yan lab), PSAP/SAPs (Proteintech Cat# 10801–1-AP, RRID:AB_2172462), Saposin (Santa Cruz cat# sc-100,584, RRID:AB_1128802), SAP-C (Santa Cruz Cat# sc-374,118, RRID:AB_10915437), SMI31 (Covance Research Products Inc. Cat# SMI-31R-100, RRID:AB_10122491), TGN46 (Abcam Cat# ab2809, RRID:AB_2203290), and Ubiquitin (Sigma-Aldrich Cat# U0508, RRID:AB_477599). After overnight incubation at 4 °C, sections were washed with PBS three times and incubated with secondary antibodies conjugated with Alexa Fluor 488 or Alexa Fluor 568 or Alexa Fluor 633 (Molecular Probes) for 2 h at room temperature. For thioflavin-S (Thio-S) staining, sections were washed after the 2nd antibody treatment and 0.01% Thio-S solution (in PBS) was added and incubated for 20 mins. The sections were finally washed with PBS 3 times and mounted with Vectasheild mounting medium. For DAB staining, sections were subsequently reacted with the specified primary antibody and corresponding biotinylated secondary antibodies (1:200) and developed according to the protocol from the Vector ABC kit (1:400; Vector Laboratories). Images were examined and captured with a Keyence BZ-X810 fluorescence or a Ziess LSM800 confocal microscope.
Western blotting
Snap-frozen mouse brain cortices were homogenized on ice in RIPA buffer containing complete protease inhibitors (Roche Diagnostics Cat# 06538304001, Mannheim, Germany). The homogenates were rotated for 30 min at 4 °C to ensure extraction of membrane proteins. After centrifugation at 15000 Å ~ g for 120 min, supernatants were collected and protein concentrations were measured with the bicinchoninic acid protein assay reagent (Thermo Scientific, Grand Island, NY, USA). Equal amounts of protein samples were resolved on 4–12% NuPage Bis-Tris gels purchased from Invitrogen. Following incubation with the indicated primary antibody, an appropriate horseradish peroxidase-conjugated secondary antibody was added. Immunoreactivity was detected by chemiluminescence using Super Signal West PICO reagent (Thermo Scientific). Image-J software was used to quantify the mean gray value for a fixed area of each protein band [
27]. The original full-blot images can be found in “
Additional File_Full gel blots”.
Three-dimensional electron microscopy (3D EM)
A 10-month-old 5xFAD mouse was anesthetized by injecting pentobarbital interperitoneally and perfused intracardially with ~ 100 ml of 0.1 m sodium cacodylate buffer containing 4% paraformaldehyde and 2.5% glutaraldehyde, pH 7.4. The left and right hippocampus of the extracted brain were fixed overnight with 0.1 m sodium cacodylate buffer containing 4% paraformaldehyde and 2.5% glutaraldehyde. After washing 3 times with 0.1 M sodium cacodylate, the samples were placed in 0.1% tannic acid for 30 min and then washed with sodium cacodylate solution. Next, the samples were sequentially processed by 2 h incubation with 2% osmium tetroxide/potassium ferrocyanide on ice, followed by a 20 min treatment with freshly-prepared 1% thiocarbohydrazide at 60 °C, and then a 1 h incubation with 2% osmium tetroxide solution on a rotor at room temperature. The samples were then placed in new vials, washed 3 times with distilled water, and placed in saturated uranyl acetate solution overnight at 4 °C. The following day, samples were washed 3 times with distilled water and finally stained with lead aspartate solution (prepared by dissolving 0.066 g of lead nitrate in 10 ml 0.03 m aspartic acid, pH 5.5) at 60 °C for 30 min and washed again with distilled water 3 times. The samples were then dehydrated by dipping the samples twice (5 min each) in a gradient series of freshly-prepared solutions of 50, 75, 85, 95, and 100% ethanol, and finally by placing in anhydrous acetone for 10 min at room temperature. The samples were then transferred to freshly-made 50% Epon resin in propylene prepared by mixing 5 ml propylene to 5 ml of 100% Epon resin formulated as 10 ml EMBed-812, 8 ml dodecenyl succinic anhydride, 4 ml methyl-5-norbornene-2,3-dicarboxylic anhydride, and 0.4 ml 2,4,6-tri (dimethylaminomethyl)-phenol. After incubating for 2 h at room temperature in 50% Epon, the samples were transferred to 100% Epon and rotated for 90 min. Finally, the samples were placed into molds and fresh Epon resin was poured and kept at 60 °C for 2 days for polymerization and solidification.
A sample block of 0.5 × 0.5 × 0.5 mm size for each hippocampal tissue was prepared by trimming the resin using a razor blade. Each sample block was mounted on a pin that was set on the stage in a Zeiss Sigma VP scanning electron microscope equipped with a Gatan 3View in-chamber ultramicrotome and a low kilovolt backscattered electron detector. The diamond knife of the microscope was set to make ∼500 sections at a thickness of 70 nm. Images were generated at 2.0–2.25 kV under standard vacuum conditions using an aperture set at 30 μm and captured at 5 nm/pixel resolution (5000× magnification).
3D reconstruction of microglia
Registered image sets were analyzed using Reconstruct software [
28]. The properties of each image set were entered manually as 70 nm slice thickness and 5 nm pixel size. Areas for amyloid plaques, microglia, and DNs were traced by manual tracing. Each trace with the same name is considered as one object by the software. After tracing each morphology (e.g., amyloid plaque, microglia, DNs, and dense body), a 3D model for each object was generated.
Cell cultures and Aβ treatment
Mouse neuroblastoma (N2a) cells were cultured in Dulbecco’s Modified Eagles medium (high glucose) and supplemented with 10%(v/v) fetal bovine serum (FBS, GIBCO Cat # 26140–079) and 1% antibiotics (GIBCO Cat # 15240–062) at 37 °C under 5% CO2. The primary neurons from cortex and hippocampus of embryonic stage 18 (E18) wild-type mice were cultured separately in neurobasal medium, supplemented with B21 and Glutamex, for 7 days. Human β-amyloid (Aβ)42 peptide (Millipore Sigma Cat# AG968-1MG) was dissolved in 1% NH4OH at a concentration of 440 μM, sonicated for 2 min in a cold-water bath, and immediately stored at − 80 °C after aliquoting as stock. The peptide was then dissolved at the desired concentration in cell culture medium. An equal volume of 1% NH4OH was included as a control (buffer or mock). The cells were incubated with Aβ42 peptide for 16 h and the cells were then subjected to either immuno-confocal microscopy or Western blotting.
Cell fractionation/enrichment of lysosomes
Fractionation of cellular organelles from WT and 5xFAD mouse brains were performed using lysosome isolation kit (Millipore Sigma Cat# LYSISO1) according to the modified protocol [
29,
30]. Briefly, whole brain tissue (without cerebellum) in extraction buffer (with protease inhibitor) were homogenized with IKA T8 at 8000–9500 rpm for 5 s × 3 times on ice. The homogenate was centrifuge at 1000 x g for 10 min at 4 °C. The supernatant was centrifuged at 5000 x g for 10 min to precipitate mitochondria. The precipitates were washed with extraction buffer and dissolved in minimum volume of buffer. The supernatant was centrifuged at 20,000 x g for 2 h at 4 °C to precipitate lysosomes. The precipitate was washed and dissolved in minimum amount of buffer. After measuring protein concentration for each sample in mitochondria, lysosomes and cytoplasmic fractions, equal amount of protein was subjected to SDS-PAGE and Western blotting.
Discussion
We have recently shown that DNs are enriched with heterogenous vesicles, growing sequentially in three layers in the surrounding amyloid plaque core: ATG9A-containing pre-autophagosomes are present in the inner layer (1st layer), RTN3-conaining clustered tubular ER is present in the middle layer (2nd layer), and multi-vesicles and mature autophagosomes labeled by RAB7 and LC3 are present in the outer layer (3rd layer) [
11]. The lysosome-like multi-vesicle bodies have been previously shown to be accumulated in DNs and in glia cells [
15,
31,
48,
49], but it was not clear when LAMP1-containing lysosomes are enriched in DNs. In this study, we demonstrate that DNs enriched lysosomes, containing LAMP1 and SAP-C, as early as amyloid plaques begin to form. With the growth of amyloid plaques, axonal degeneration near plaques is also increasing and functional lysosomes in axons are reduced, and this may viciously facilitate the growth of amyloid plaques. Thus, preventing lysosomal damage in axons is likely to be an effective intervention strategy to reduce amyloid toxicity.
Lysosomes are highly dynamic subcellular organelles originating from the Golgi. They acquire hydrolases through a series of dynamic events during lysosomal biogenesis [
50,
51]. In neurons, lysosomes are distributed not only in the soma, but also in axons and dendrites [
38,
52‐
54]. LAMP1, a commonly used lysosomal marker, is localized on the membranes of endosomes after synthesis and is transferred to lysosomal organelles via the endo-lysosomal pathway. It has been shown that a significant pool of lysosomes, lacking degradation capacity due to devoid of certain hydrolases such as cathepsin-B, −D and glucocerebrosidase, is more enriched in axons and dendrites, while hydrolase-containing lysosomes are more restricted to neuronal soma [
34,
55]. SAPs (SAP-A to -D) are small amphipathic glycoproteins that are processed from their precursor, PSAP, mostly in lysosomes, and act as activators of glucocerebrosidase (GCase) or galactocerebrosidase during lysosomal hydrolysis of glycosphingolipids [
56,
57]. The functions of each SAP are specific to their associated hydrolases, and deficiency of one may not be compensated by another [
58,
59]. Mutations in PSAP may cause deficiencies in full PSAP or in a specific SAP, which leads to an excessive accumulation of lysosomal sphingolipids in certain organs, specifically in spleen, liver, bone, and bone marrow, and are associated with numerous pathological conditions collectively termed as lipid storage disorders [
56,
60,
61]. SAP-C is particularly required for activating GCase and is present in a large portion (~ 64%) of peripheral LAMP1
+ vesicles [
36]. Our results showed that hydrolase-deficient but SAP-C-enriched LAMP1
+ primary lysosomes were accumulated in the 1st layer of DNs during the initial plaque formation stage (Figs.
2 and Fig.
3A-O). It is expected that the accumulation of numerous vesicles or multi-vesicle bodies in DNs surrounding amyloid plaques are likely proportional to a gradual impairment of axonal trafficking due to continuous Aβ accumulation in AD brains. This is because the hydrolase-enriched LAMP1
+ lysosomes, viewed as degradative lysosomes and predominantly localized in neuronal soma, are anterogradely transported to distal axons to maintain local degradation capacity [
38] and the autophagosomes formed at distal neurites can be retrogradely transported to somata for degradation [
55,
62]. The early trapping of such lysosomes in DNs likely impairs this trafficking balance, various lysosomal functions including GCase and lipid metabolism.
However, at the initial plaque-forming stage, axonal trafficking may only be restrictively impaired to areas containing ATG9A
+ vesicles or LAMP
+/SAPs
+ primary lysosomes, as active lysosomal transport along axons appears to be functional based on visible LAMP1-labelled lysosomes distributed along axons (Fig.
5A, supplemental Fig.
3, left panel). At this early stage, endo-lysosomes at distal axons can still be partially maintained and are transported from neuronal soma. At later stages of plaque growth, when amyloid loads are sufficiently high to block the trafficking of degradative lysosomes from soma, LAMP1
+ lysosomes are rarely visible in neurofilament-positive axons (Fig.
5A, supplemental Fig.
3, right panel). Under these conditions, local degradation capability is likely diminished and non-degraded autophagosomes and endosomal vesicles at the axonal terminals form an outer layer of DNs surrounding plaques [
11]. Hence, gradual deposition of Aβ during plaque growth appears to disrupt the balance between the primary and functional lysosomal pools in neurites by interrupting their dynamic trafficking.
Both protein levels and CSF levels of PSAP are elevated in AD brains [
63,
64]. An enrichment of SAP-C in lysosomes induce lysosomal membrane permeability (LMP) and LMP-mediated apoptotic cell death [
41‐
43]. We noted a significant reduction of mature cathepsin D levels in autolysosomes/mature lysosomes, correlated with more SAP-C in 2-month-old 5xFAD (Fig.
8A-C). While cathepsin D is the main protease for processing of PSAP into SAPs [
65], an enhanced SAP-C, but reduced cathepsin D, in autolysosome-enriched fractions likely an event resulted from LMP-mediated release of hydrolases such as cathepsin D. SAP-C enrichment also likely causes the damage of lysosomal maturation as demonstrated by an increased level of a lysosomal damage sensor protein, galactin-3 in LFs of 5-month-old 5xFAD compared to WT (supplemental Fig.
6A-B). Thus, in addition to Aβ-induced disruption of lysosomal trafficking [
31,
66], enhanced SAP-C protein levels and SAP-C accumulation in lysosomes potentially lead to LMP [
67‐
70] and induction of LMP-mediated apoptosis [
71‐
73], which eventually could lead to neuronal degeneration in AD brains. Further studies will be conducted to attest this postulation.
We noted that SAP-C
+-DNs disappeared at older AD mouse brains and late stage of AD progression in human brain (Fig.
3 and Fig.
4). We speculated that lysosomal exocytosis in response to elevated calcium (Ca
2+) levels during amyloid plaque growth could lead to a selective reduction in SAP- and LAMP1-labelled DNs in older AD mouse brains. Lysosomal exocytosis is a Ca
2+-regulated process in which lysosomes are docked at and fuse with the plasma membrane through LAMP1, in which it releases the luminal contents out of cells [
74,
75]. Dysregulation of Ca
2+ signaling, perhaps related to the perturbed regulation of ER Ca
2+ signaling, the impaired ability of mitochondria for Ca
2+ buffering, and elevated levels of intracellular Ca
2+ has been shown in AD brains [
76‐
78]. The tubular domains of ER and mitochondria play a critical role in intracellular Ca
2+ balancing [
79,
80]. Thus, subsequent accumulation of dysfunctional tubular ER and clustered mitochondria in the 2nd layer of DNs, at relatively later time points [
11,
12], may contribute to elevated intracellular Ca
2+ levels in neurons; this, in turn, would stimulate the exocytosis of accumulated LAMP1- and SAP-labeled primary lysosomes, resulting in a gradual reduction in LAMP1
+/SAPs
+ DNs at later time points (i.e., 9-months; Figs.
3A-R) when tubular ER/mitochondria-mediated DNs (i.e., RIDNs) are extensively accumulated in surrounding Aβ plaques (Supplemental Fig.
2A). Future in vitro and in vivo studies measuring Ca
2+ signaling and lysosomal exocytosis in cultured neurons and in AD brain may reveal the underlying mechanisms.
Intriguingly, our study also revealed that immunohistochemically-detectable LAMP1 near mature amyloid plaques appeared to be in DAM, rather than in DNs. We showed that LAMP1 was mainly detected in DAM after plaque formation is stabilized in AD mouse brains as well as in AD postmortem brains. At the same time, LAMP1 immunoreactivity in intact axons was decreased, likely due to severe axonal degeneration as well as loss of LAMP1-immuno-epitopes of damaged autolysosomes in AD DNs. Hence, caution should be used when interpreting LAMP1 in DNs at late plaque-forming stages. The detection of both LAMP1 and hydrolases in DAM may indicate that phagocytic DAM perhaps retain functional lysosomes at later stages of AD development, or even during clinical manifestation [
3,
4]. Further functional assays will be needed in order to clarify the state of degradative lysosomes in DAM.
In summary, using three different AD mouse models and AD postmortem brain samples, we have shown that accumulated LAMP1
+ DNs at the early plaque formation stage are mostly SAP-enriched, but hydrolase-deficient primary lysosomes, dissociate from plaques after plaque formation eventually reaching a plateau. Our data suggest that Aβ deposition will impair lysosomal biogenesis at the initial plaque-forming stage. Enhanced lysosomal accumulation of SAP-C could be a contributing factor to lysosomal dysfunction in AD brains by inducing leakages of lysosomal hydrolases in DNs. Dysfunctional lysosomes will also facilitate tau pathologies through tau secretion and propagation [
81]. Hence, preventing lysosomal damage and stimulating lysosomal biogenesis at early pre-clinical phases is likely an alternative AD therapeutic strategy.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.