Introduction

Neuronal death can cause fundamental functional debility in neurological diseases1,2. Neuronal loss is found in specific brain regions in many brain disorders such as mesial temporal lobe epilepsy3,4 and Parkinson’s disease5,6. In Alzheimer’s disease (AD), neuronal and synaptic loss result in progressive atrophy in the limbic and neocortical regions, which correlates well with cognitive decline7,8,9,10,11,12. Neuronal loss in AD is a microscopically “negative” pathology13, with “positive” pathologies such as amyloid-β (Aβ) plaques and neurofibrillary tangles (NFT, briefed as “tangles” hereafter) more commonly studied14,15,16. However, identification of neuronal death programs and related molecules is important for understanding AD pathogenesis and developing new therapeutics for this disease17,18. Indeed, evidence supports that neuronal tangle pathogenesis is tightly associated with the activation of cell death pathways. Thus, enhanced macroautophagy is indicated in neurons and dystrophic neurites along with plaque and tangle pathogenesis19,20,21. Caspase-mediated apoptosis is activated in neurons undergoing tangle pathogenesis22,23,24,25. Further, multiple molecular markers involved in lysophagy26,27, mitophagy28 and necrophagy29 participate in the pathogenesis of granulovacuolar degeneration (GVD) in pretangle neurons that may further develop into mature tangles and eventually die in the form of ghost tangles30,31.

The presenilin (PS)-associated protein (PSAP), also known as mitochondrial carrier homolog-1, was discovered by screening a human brain complementary DNA library for PS1-interacting proteins, which contains the postsynaptic density protein 95, Disc-large protein and epithelial tight-junction protein zona occludens 1-like domain32. It binds to the C-terminal of PS1 and but not PS2, and mediates ɤ-secretase-dependent and independent apoptosis32,33. PSAP overexpression in cultured cells induces apoptotic death by promoting mitochondrial cytochrome c release34,35. PSAP can also mediate death receptor 6-induced apoptosis in vitro36. PSAP mRNA expression is enriched in rodent and human brains relative to the periphery32. However, its normal protein expression pattern in the central nervous system has not been characterized in either animals or human.

Given its role in cell death as established in cell biological studies, it is of interest to explore whether PSAP alteration would occur in human brain in conditions involving neuronal death or neurodegeneration. We carried out pilot experiments and observed apparently increased PSAP immunoreactivity (IR) in tangle-like neuronal profiles in aged human brains. Therefore, in the present study, we first assessed the whole brain distribution pattern of PSAP IR using pathologically-free postmortem brains from adult and some aged donors. We further explored the alteration of PSAP IR relative to phosphorylated tau (pTau) IR, tangles and cytostructural integrity using the human brains with late-stage tau and amyloid pathology. Primary age-related tauopathy (PART) refers to the occurrence of tau pathology in the absence of cerebral Aβ deposition37. We also explored PASP IR relative to pTau/tangle in human brains with PART. In addition, we explored whether PASP alteration also occurs in the brains of three commonly used transgenic AD model mice, especially the 3xTg-AD mice that develop both Aβ and pTau pathologies.

Materials and methods

Brain samples and tissue preparation

The Ethics Committee of Central South University Xiangya School of Medicine approved (#2020KT-37, 4/10/2020; #2023-KT084, 6/21/2023) the current study, in compliance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). All human brains were collected post-mortem through a willed body donation program established for medical education and research by Xiangya School of Medicine in accordance with government regulation and donation law. Donations were caried out based on written informed consent by the donor and/or the next-of kin. Donor’s clinical records prior to and/or during the last hospitalization were also obtained whenever available38. Research using postmortem human brain samples is not a human subject study (not involving live humans), therefore patient consent was not applicable to the present particular neuropathological investigation.

Brain was bisected after removal from the skull, with half fresh-frozen at -70oC and the other immersed in formalin for 2–3 weeks. The fixed half-brains were then cut coronally into 1 cm-thick slices and blocked for the preparations of cryostat and paraffin sections. Cryostat sections at 35 μm thickness were cut and stored in a cryoprotectant at -20oC before use. Paraffin sections at 4 μm-thickness were prepared using blocks from selected brain regions and cases for designated experiments. Overall histopathological assessments were carried out according to the Standard Brain Banking Protocol proposed by China Brain Bank Consortium39. The stages of amyloid and tangle pathologies for a given brain were scored according to established references15,16, with Aβ and pTau immunolabeling stained with the mouse anti-Aβ1–16 antibody 6E10 (1:4000, Signet Laboratories, Inc., Dedham, MT, USA, Cat# 39320, AB_2565327) and the mouse anti-phospho tau Ser202 and Thr205 antibody AT8 (1:4000, Thermo Fisher Scientific Inc., Waltham, MA, USA, Cat# MN1020, AB_223647) antibodies, respectively. A portion of the brain samples had been pathologically characterized in detail in our recent studies30,40,41. In the present study, fixed and frozen samples from 30 postmortem human brains were used, with the cases grouped according to donor’s ages and neuropathologies (Additional Table 1). The pathologically negative adult and aged control group included eleven brains (6 males and 5 females) with donor’s ages from 21 to 84 years old (48.7 ± 23.7 years). The PART group consisted of eight brains (4 males and 4 females) from donors died at 65–95 years (81.4 ± 9.1 years). The probable AD (pAD) and AD group consisted of 11 cases (5 males and 6 females, 74–93 years) with AD-type dementia or with cognitive assessment not available (Additional Table 1).

Cryoprotected brain sections from three transgenic mouse models of AD and wild-type mice also used in the current study. These materials were obtained and used in our earlier studies following Institutional Ethics approval (Central South University and Southern Illinois University Carbondale), which were detailed in our previous publicaitons42,43,44,45,46. All experimental methods were carried out in accordance with NIH guidelines on animal housing, maintenance and euthanasia (Sodium Pentobarbital, 50 mg/ml, i.p.), procedures, and the ARRIVE guidelines. The brain sections used in the present study were the additional sections preserved in the principal investigator (Xiao-Xin Yan)’s laboratory. Briefly, these transgenic AD models carried the familial AD (FAD) associated amyloid-β precursor protein (APP) and PS1 mutations (i.e., APP/PS1 mice) (Jackson Laboratory, Bar Harbor, ME, USA), the five FAD-related APP/PS1 mutations (i.e., 5xFAD mice) (Jackson Laboratory, Bar Harbor, ME, USA), and three mutations in APP/PS1/tau genes (i.e., 3xTg-AD mice, a colony propagated from breeding pairs provided by Dr. Frank LaFerla at University of California at Irvine)42,43,44,45,46. Specifically, coronal or sagittal mouse brain sections used for immunohistochemistry were from the APP/PS1 mice at 4, 6–7 and 12–14 months of age (n = 4/age)45, the 5xFAD mice at 2, 4 and 8 months of age (n = 4/age)44, and the 3xTg-AD mice at 1–3, 6–8, 18–22 months of age (n = 4/age)43.

Immunohistochemistry

To comparatively assess PSAP IR between the brains with and without AD-type neuropathology, we obtained a set of cryoprotected sections from each identified brain passing representative neuroanatomical structures, including the frontal, parietal and occipital lobes, insula/striatum/thalamus, cerebellar cortex and deep nuclei, middle brain, pons, and medulla oblongata). Sections from 3 to 4 brains were stained immunohistochemically for PSAP. This antibody is a polyclonal rabbit antibody (Ab1) generated against a synthetic peptide corresponding to amino acid residues 21 to 35 of the N-terminus of PSAP, which detects protein bands migrated at ~ 39 and ~ 35 kDa in immunoblot34,36. The antibody was diluted at 1:1000 for immunohistochemistry and 1:500 for immunofluorescence, with the sections incubated overnight at 4 oC. The sections were subsequently reacted with a pan-specific secondary antibody (biotinylated horse anti-mouse, rabbit and goat IgG) (1:400, Cat#BA-1300, AB_2336188, Vector Laboratories, Burlingame, CA, USA) for 2 h and further with avidin-biotin complex kit (1:400, Cat#PK-6100, AB_2336819, Vector Laboratories) for 1 h at room temperature. The immunoreaction product was visualized with diaminobenzidine as the detecting chromogen. Adjacent sections from the PART, pAD/AD cases were also processed in the current study with other primary antibodies, including mouse anti-Aβ (6E10, 1:4000), mouse anti-pTau (AT8, 1:4000), mouse anti-casein kinase I isoform δ (CK1δ; 1:1000, Cat#sc-55553, AB_831049, Santa Cruz Biotechnology, Inc., Dallas, Texas, USA) and goat anti-sortilin (1:1000, Cat#AF3154, AB_2286389, R&D Systems, Inc., Minneapolis, MN, USA)30. Rabbit anti-Aβ (D12B2, 1:2000, Cat#50-204-9523, Cell Signaling Technology, Inc., Danvers, MA, USA) and anti-pTau (1:2000, Cat#T6819, AB_261745, Sigma-Aldrich, Inc., St. Louis, MO, USA) antibodies were used for immunolabeling in mouse brain sections, which were stained by including a few AD human brain sections as positive control. Immunolabeled sections were dehydrated with ascending ethanol, cleared with xylene, and coverslipped with a mounting medium, with some sections also counterstained with hematoxylin before dehydration.

Immunofluorescence staining

Double immunofluorescence was carried out with the rabbit anti-PSAP antibody and another antibody raised in mice or goat (i.e., 6E10, AT8, CK1δ and sortilin) with an overnight incubation at 4oC, followed by 4,6-diamidino-2-phenylindole (DAPI) nuclear counterstain. For triple fluorescent experiments, the sections were incubated with the rabbit anti-PSAP and mouse anti-pTau (AT8), followed by the Amylo-Glo fluorescent histological stain (described below). The immunofluorescent signals were visualized with the Alexa Fluor® 488 and Alexa Fluor®594 conjugated donkey anti-mouse (Cat# A-21203; RRID: AB_141633), anti-rabbit (Cat# A-21206, RRID: AB_2535792) and anti-goat (Cat# A-11058; RRID: AB_2534105) secondary antibodies (1:100, Invitrogen, Carlsbad, CA, USA). All sections were treated in 0.1% Sudan black to block autofluorescence before they were coverslipped with VECTASHIELD® antifading mounting media (Vector Laboratories).

Amylo-Glo fluorescent staining

The ready-to-use kit of Amylo-Glo RTD Amyloid Plaque Stain Reagent was purchased from Biosensis Pty. Ltd. (#TR-300, Thebarton, SA, Australia). This fluorescent dye visualizes both amyloid plaques and tangles in human brain sections as bright blue stain in the ultraviolet channel47. Briefly, following the double immunofluorescent labeling, the sections were rinsed in 0.1 M phosphate-buffed saline and immersed in 10X diluted Amylo-Glo reagent solution according to manufacturer’s instruction, with the fluorescent signal monitored under an Olympus B53 microscope. The sections were differentiated in 50% ethanol shortly, rinsed in saline, and treated in 0.1% Sudan black before coverslipped with VECTASHIELD® antifading mounting media.

Western blot assay

Immunoblot experiments were aimed to probe the elevation of PSAP in partnership with tauopathy in the cerebral neocortex. Thus, we identified eight pAD/AD brains with tau pathology ≥ Braak stage IV and Aβ pathology ≥ Thal phase 2, and eight brains of adult cases without tau and amyloid pathology (Additional Table 1). The tissue block was cut at the orbit gyrus from the frontal lobe slice of the frozen hemibrain passing the anterior end of anterior horn of the lateral ventricle. Samples were homogenized in 10% (w/v) radioimmuno-precipitation (RIPA) buffer containing a cocktail of proteinase inhibitors (Beyotime, China, #P0013B). The cortical lysates were centrifuged at 16,000 x g for 15 min at 4oC, with the supernatants collected and assayed for protein concentration (Bio-Rad Laboratories, Hercules, CA, USA). Extracts containing 30 µg total protein were run in 8–15% sodium dodecyl sulfate and polyacrylamide gel electrophoresis (SDS-PAGE) and electrotransferred onto Trans-Blot pure nitrocellulose membranes. Whole-size membranes were immunoblotted with the rabbit anti-PSAP (1:400, 10% SDS-PAGE), mouse anti-pTau AT8 (1:1000, 10% SDS-PAGE) and rabbit anti-Aβ (D12B2, 15% SDS-PAGE) antibodies42, respectively. During the revision of the manuscript, immunoblotting was additionally performed with the Aβ antibody 6E10 (1;1000, 12% SDS-PAGE) and a mouse monoclonal tau antibody that reportedly detects total tau (TAU-5, 1:1000, ab80579, Abcam, 12% SDS-PAGE), according to reviewer’s recommendation. The bound proteins were visualized using horseradish peroxidase (HRP)-conjugated goat anti-rabbit (1:6000, Cat# #1706515, AB_11125142, Bio-Rad) and anti-mouse (1:5000, Cat#1706516, AB_2921252, Bio-Rad) IgGs, and the PierceTM ECL-Plus Western Blotting Substrate detection kit (Cat#32132, AB_2080115, Thermo Fisher Scientific). Immunoblots were image-documented in the UVP ChemStudio/PLUS device (Analytik-Jena/UVP, Upland, CA, USA). After the above first round of blotting, the membranes were washed in a Restore™ Western stripping buffer (Cat#21059, Thermo Fisher Scientific), followed by reblotting of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1:5000, MA1-16757, AB_568547, Thermo Fisher Scientific) or β-actin (1:5000, AB_2536844, Thermo Fisher Scientific) antibody raised in mouse, as internal loading control (Additional file 1).

Image acquisition

Bright-field micrographs were scan-imaged on a Motic-Olympus microscope (Motic China Group Co. Ltd., Wuhan, Hubei, China). The images were examined with the Motic viewer (Motic China Group Co. Ltd.) to assess the labeling across anatomical regions from low to high resolutions, with the areas of interest extracted at desired magnifications and exported for figure preparation. Fluorescently stained paraffin Sect. (4 μm in thickness) were scan-imaged with a Keyence or a 3DHISTECH imaging system using the 20× magnification objective and a Z-stack setting of 1 μm scanning depth. The resulting images were examined from low to high magnifications, with areas of interest cropped at desired magnifications using the KV-X800 Analyzer (KV-800, Keyence Corporation, Osaka, Japan) or the Caseviewer interface (3DHISTECH Ltd, Budapest, Hungary), and used for figure preparation.

Quantitative image analysis

Quantitative analyses were carried out to estimate the extent of colocalization of PSAP IR relative to other markers in double and triple fluorescent labeling preparations. Single-, double- and triple- (in the case of double immunofluorescence together with Amylo-Glo stain) labeled somata were counted on-screen at 20X magnification. Cell counting was conducted in 20 non-overlapping screen fields while moving across the parahippocampal gyrus (PHG) and from the subicular to CA1 areas, respectively. The screen fields were set to cover the cortical layers in the PHG and across the stratum pyramidale in the subicular to CA1 regions. While examining the labeled neurons in a given field, single-, double-, or triple-labeled somata were determined using overlapped as well as single-channel fluorescent viewing options, with the number of each type of somata recorded into an Excel spreadsheet. After counting and recording, the total numbers of singly labeled and colocalized cells and their fractions (%) were calculated for each brain. The optic densities of immunoblotting bands were measured using the NIH ImageJ2, with the values of PSAP and pTau normalized to GAPDH, and Aβ to β-actin. All figures were prepared using the Photoshop software (2024) (Adobe, San Jose, CA, USA) by assembling selected low and high magnification histological images and graphs generated through image and data analyses, whenever appropriate.

Statistical analysis

The mean ± standard derivation (SD) and the proportional values (%) thereof in the brain groups were graphed. The normalized optic density data of immunoblotted PSAP, pTau and Aβ levels were statistically compared between the sample groups using two-tailed unpaired Student’s t-test, or One-way analysis of variance (ANOVA) together with Bonferroni post hoc test, with P < 0.05 set for the cutoff level for significant intergroup difference (GraphPad Prism 10 for Windows, Boston, Massachusetts USA, www.graphpad.com).

Results

PSAP IR in human brains free of AD-type neuropathology

The present study is the first to explore PSAP expression and alteration in the human brain. Therefore, we first attempted to determine the normal distribution pattern of PSAP IR in the entire brain. For this aim, PSAP IR was assessed in immunohistochemically labeled (diaminobenzidine immunolabeling) sections of the cerebrum and major subcortical anatomical regions. All these brains were confirmed to be absent of AD-type neuropathologies (Additional Table 1). Representative micrographs from a 29-year-old adult case are shown as main and supplemental figures (Fig. 1, and Additional Figs. 1, 2, 3, 4, 5 and 6).

Fig. 1
figure 1

Presenilin-associated protein (PSAP) immunoreactivity. (A) shows the low magnification view, with the framed areas enlarged sequentially as indicated. PSAP IR appears neuropil-like in the cerebral cortex and cellular layers of the hippocampal formation. (B-B3, C-C3) Relatively large-sized pyramidal neurons show the highest IR, especially in those located in deep layers III and V. (D-D3; E-E3) The pyramidal neurons in the layer II islands of the entorhinal cortex and the subicular and hippocampal subregions as well as the hilar mossy cells in the dentate gyrus are also clearly labeled. CA1-CA3: Ammon’s horn subregions; DG: dentate gyrus; FG: fusiform gyrus; GCL: granule cell layer: Hi: hilus; ITG: inferior temporal gyrus; I-VI: cortical layers I to VI; ML: molecular layer; MTG: middle temporal gyrus; PHG: parahippocampal gyrus; Pro-S: prosubiculum; Pre-S: presubiculum; Sub: subiculum; s.o.: stratum oriens; s.p.: stratum pyramidale; s.r.: stratum radiatum, WM: white matter. Scale bars are as indicated.

Full size image
Fig. 6
figure 2

Double immunofluorescent characterization of presenilin-associated protein (PSAP), and casein kinase 1δ (CK1δ) in pAD/AD human brains. (A) The low power view of double-labeled temporal lobe paraffin section with DAPI counterstain, with framed areas enlarged as other panel sets (B-E). The PSAP-enriched neuronal somata (pointed by arrows) and dystrophic neurites (open arrowheads) rarely colocalize with CK1δ IR, whereas CK1δ labeled granulovacuolar degeneration (GVD) bodies (pointed by white arrowheads) are present in other neuronal somata with light PSAP IR. (F) The numbers of PSAP positive neurons with and without CK1δ labeled GVD bodies quantified in the parahippocampal neocortex and subicular to CA1 pyramidal layers from four brains, with the ratios indicated (mean ± SD, n = 4, two-tailed unpaired t-test). CA1-CA3: Ammon’s horn subregions; DG: dentate gyrus; PHG: parahippocampal gyrus; Pro-S: prosubiculum; Pre-S: presubiculum; Sub: subiculum. Antibodies and fluorescence channels, and scale bars, are as indicated in the image panels.

Full size image

Across the cerebral neocortex, PSAP IR appeared largely as a weak neuropil-like labeling over the gray matter, with some neuronal somata exhibiting a higher intensity above the background, as seen in the temporal (Fig. 1A-C3), frontal (Additional Fig. 1A-C2), occipital (Additional Fig. 1D-F), precentral (Additional Fig. 2A-B2) and postcentral (Additional Fig. 2C-D2) cortical regions. The labeled cellular profiles were mostly seen in layers III to VI and appeared to be mostly the principal neurons according to their pyramidal somal shape and relatively large somal size. The immunoreactive product was primarily located in the somata and proximal dendrites, with the axons sometimes visible near the base of the soma (Fig. 1B-B3, C-C3, Additional Fig. 1B-B2, C-C3, E, F and 2B-B2, D-D2). In the entorhinal cortex, a large number of labeled principal neurons occurred in the layer II cell islands, while other pyramidal-like neurons were labeled in layer III (Fig. 1D-D3). In the hippocampal formation, PSAP-labeled pyramidal neurons were seen across the subicular and CA1-3 subregions (Fig. 1A, and E-E3). The granule cells in the dentate gyrus exhibited much lighter IR relative to the hippocampal pyramidal neurons (Fig. 1E2).

Fig. 2
figure 3

Presenilin-associated protein (PSAP) IR in the temporal lobe regions in a probable Alzheimer’s disease (pAD) human brain. (A) Low magnification view, with the framed areas enlarged sequentially as indicated. PSAP IR is apparently enhanced in a subpopulation of pyramidal neurons in the temporal neocortex (B-B2, C-C2) and subicular and CA subregions (D, D1), and mossy cells in the dentate gyrus (D, D2). These darkly labeled neurons show a tangle-like morphological appearance. Heavily labeled neuritic processes (as pointed by arrows) are also present in the cortical and hippocampal regions. CA1-CA3: Ammon’s horn subregions; DG: dentate gyrus; FG: fusiform gyrus; GCL: granule cell layer: Hi: hilus; ITG: inferior temporal gyrus; I-VI: cortical layers I to VI; ML: molecular layer; MTG: middle temporal gyrus; PHG: parahippocampal gyrus; Pro-S: prosubiculum; Pre-S: presubiculum; Sub: subiculum; s.o.: stratum oriens; s.p.: stratum pyramidale; s.r.: stratum radiatum, WM: white matter. Scale bars are as indicated.

Full size image

In the section passing the insula, basal ganglia and thalamus (Additional Fig. 3), PSAP IR was observed in the cortical gray matter, claustra and putamen, with little labeling in the white matter, internal capsule, external capsule and globular pallidum. Across the thalamic and subthalamic regions, areas with relatively high and low labeling intensity were present (Additional Fig. 3A). Again, relatively large-sized neuronal somata exhibited PSAP IR higher than background, which were located in the deep cortical layers, claustrum (Additional Fig. 3B, and B1) and globular pallidum (Additional Fig. 3C, C2). The putamen exhibited a higher reactivity than the globular pallidum, although few labeled somata could be identified (Additional Fig. 3C, C1). Lightly to moderately labeled neuronal somata with short processes were present in major thalamic and subthalamic nuclei, many appeared multipolar in shape (Additional Fig. 3D, D1, E, E1, F, and F1).

Fig. 3
figure 4

Comparative assessment of presenilin-associated protein (PSAP) and phosphorylated tau (pTau) immunolabeling in paraffin sections and their protein elevations in pAD/AD human cortical lysates. (A-A4, B-B4) Micrograph panels show PSAP (A-A4) and pTau (B-B4) immunolabeling with hematoxylin counterstain in adjacent temporal lobe sections from an AD case. PSAP and pTau labeled somata (arrows) and neurites (hallowed arrows) show similar morphological features. The nuclei in the somata with heavy tangle-like products appear to be lost, dislocated near the cell border, or shrunken (A1-A4, B1-B4). PSAP-labeled neurites are mostly elongated, while pTau-labeled neurites appear elongated (hallowed arrows) as well as globular (arrowheads). (C) show the immunoblotted PSAP (assayed with 10% SDS-PAGE gel) and pTau (AT8, assayed with 10% SDS-PAGE gel). (D) show the bands of total tau (TAU-5, assayed with 12% SDS-PAGE), and Aβ products blotted with the antibody D12B2 (assayed with 15% SDS-PAGE gel. (E) The full-length protein and clearage fragment of the β-amyloid precursor protein (APP) and Aβ productions blotted with the 6E10 antibody (assayed 12% SDS-PAGE gel). Immunoblots from a set of cortical lysates from four pAD/AD cases and four neuropathology-free controls (CTL) are presented as indicated. Original western blot images were shown in Additional file 1. (F) The bar/dot graphs show the densitometric data and statistics from the two groups as indicated (mean ± SD, n = 8/group, two-tailed unpaired t-test). Arrowheads point to the putative monomer Aβ productions. CA1-CA3: Ammon’s horn subregions; DG: dentate gyrus; FG: fusiform gyrus; GCL: granule cell layer: Hi: hilus; PHG: parahippocampal gyrus; Pro-S: prosubiculum; Pre-S: presubiculum; Sub: subiculum; s.p.: stratum pyramidale; s.r.: stratum radiatum, WM: white matter.

Full size image

In the midbrain, lightly labeled cells were seen in the superior colliculus, periaqueductal gray, reticular formation, red nucleus and substantia nigra at low magnification, with the neurons in the dorsal raphe well displayed (Additional Fig. 4A). At high magnifications, labeled somata could be identified in the deep layers of the superior colliculus and the central gray (Additional Fig. 4B-B2). Neurons in the dorsal raphe, red nucleus and substantia nigra pars compacta showed moderate PSAP IR in their somata and proximal dendrites (Additional Fig. 4C, D, D1 and D2). In the transverse section of the upper pons (Additional Fig. 5), labeled neurons were present largely in the locus coeruleus (Additional Fig. 5B, and B1). A large population of labeled neurons occurred in the ventral part of the pons between the unlabeled fiber tracts (Additional Fig. 5C, C1, D, and D1).

Fig. 4
figure 5

Double immunofluorescent characterization of presenilin-associated protein (PSAP) and phosphorylated tau (pTau) (AT8 antibody) in pAD/AD human brain sections. (A) The low power view of immunofluorescently stained temporal lobe paraffin section with DAPI counterstain (from an AD case), with framed areas enlarged as other panel sets (B-E). Fluorescent markers are as indicated. PSAP and pTau immunofluorescence are partially colocalized in neuronal somata and dystrophic neurites (open arrows). Thus, neuronal somata singly labeled for PSAP (pointed by arrows), singly labeled for pTau (pointed by empty arrows), and double-labeled (pointed by arrowheads), are present in the same microscopic field. A diminished DAPI labeling is seen among many PSAP labeled neurons. (F) The numbers of the single- and double-labeled neurons based on quantification in the parahippocampal neocortex from four brains, with cell type ratios indicated (mean ± SD, n = 4, one-way ANOVA together with Bonferroni post hoc test). CA1-CA3: Ammon’s horn subregions; DG: dentate gyrus; Ent: entorhinal cortex; PHG: parahippocampal gyrus; Pro-S: prosubiculum; Pre-S: presubiculum; Sub: Subiculum; WM: white matter. Antibodies and fluorescence channels, and scale bars, are as indicated in the image panels.

Full size image
Fig. 5
figure 6

Triple fluorescent characterization of presenilin-associated protein (PSAP), phosphorylated tau (pTau) and Amylo-Glo labeling in pAD/AD human brain sections. (A) The low power view of triple-labeled temporal lobe paraffin section, with framed areas enlarged. (B) Neuritic plaques (hallowed triangles) in the molecular layer of the dentate gyrus. A great majority of the dystrophic neurites appear triple-labeled, with the Amylo-Glo also staining extracellular amyloid deposits. (C, D) Varied colocalization patterns of PSAP, pTau and Amylo-Glo among somal profiles, which can be identified as five groups: PSAP/AT8/Amylo-Glo triple-labeled (pointed by white arrows), PSAP/Amylo-Glo double-labeled (pointed by white arrowheads), AT8/Amylo-Glo double- labeled (hallowed white arrows), Amylo-Glo single labeled (hollowed arrowheads) and AT8 single labeled (two-tailed arrows). (D) The numbers of the five types of neuronal somata were quantified in four brains, with the calculated cell type ratios indicated (mean ± SD, n = 4, one-way ANOVA together with Bonferroni post hoc test). CA1-CA3: Ammon’s horn subregions; DG: dentate gyrus; Ent: entorhinal cortex; GCL: granule cell layer; Hi: hilus; ML: molecular layer; PHG: parahippocampal gyrus; Pro-S: prosubiculum; Pre-S: presubiculum; Sub: subiculum. Antibodies and fluorescence channels, and scale bars, are as indicated in the image panels.

Full size image

In the cerebellum, PSAP IR occurred in the cortical gray matter and deep nuclei, with labeled Purkinje cells and cells in the dentate nucleus visible at low magnification (Additional Fig. 6A). Thus, the Purkinje cell somata and dendritic trees were well labeled. The labeling in granular cell layer was weak, with no stained somata recognizable (Additional Fig. 6B, and B1). Labeled neurons in the dentate nucleus occurred in the mantle part, and were multipolar in shape, with their dendrites mixed into a relatively high local neuropil labeling (Additional Fig. 6A, C, and C1).

Enhanced PSAP IR in pAD and AD human brains

We examined PSAP immunolabeled sections from pAD and AD cases with both cerebral Aβ and pTau pathologies (Additional Table 1). All cortical regions, hippocampal formation, amygdala and some subcortical structures were assessed. A major finding was the enhanced PSAP labeling of neuronal and neuritic profiles morphologically comparable to pTau-positive profiles. Representative images of this aberrant PSAP labeling are included as main and supplemental figures, with Aβ and pTau labeling in adjacent sections provided for pattern comparison.

As observed in the temporal lobe section from a pAD case (Additional Table 1), a large amount of neuronal somata and neuritic profiles exhibited apparently darker PSAP IR relative to other labeled neurons in the neocortical and limbic cortical regions, and the hippocampal subregions (Fig. 2). The heavily labeled neuronal somata showed a tangle-like appearance, characterized by distorted somal shape, uneven somal labeling, and dendritic truncation or sudden ending (Fig. 2A, B-B2, C-C2, and D-D2). Labeled neuritic profiles were present in the same microscopic field, including the relatively thin neuropil threads and relatively thick dystrophic neurites arranged in clusters (Fig. 2B2, C1, D1, and D2). In the section passing the insula (Additional Fig. 7A, and B-B2), heavily labeled neuronal somata and neurites were seen largely in the insular cortex, while some neuronal somata in the claustrum and striatum showed increased PSAP IR (Additional Fig. 7A, and C-C2). In the occipital lobe, somal and neuritic profiles with enhanced PSAP IR were observed more frequently in area 18 than area 17 (Additional Fig. 7D, E-E2 and F-F3). These profiles were distributed more commonly in the infragranular than supragranular layers. The extent of tauopathy in this case was estimated to be Braak stage IV (Additional Fig. 8), while the score of Aβ deposition was Thal phase 5 (Additional Fig. 9).

Fig. 7
figure 7

Double immunofluorescent characterization of presenilin-associated protein (PSAP) and sortilin in pAD/AD human brains. (A) Section map, with enlarged areas show that the majority of PSAP-enriched neuronal somata (pointed by arrows) are not colocalized with sortilin immunoreactivity across the regions (B-E), while a few of cells colocalized somata observed (pointed by arrowhead). No sortilin colocalization is seen in PSAP-labeled dystrophic neurites (C, open arrows). The DAPI nuclear stain in the PSAP enriched neurons appears reduced or lost. (F) The numbers of PSAP-positive neurons with and without sortilin labeling counted in the parahippocampal neocortex and subicular to CA1 pyramidal layers, with the ratios calculated (mean ± SD, n = 4, two-tailed unpaired t-test). CA1-CA3: Ammon’s horn subregions; DG: dentate gyrus; Ent: entorhinal cortex; PHG: parahippocampal gyrus; Pro-S: prosubiculum; Pre-S: presubiculum; Sub: subiculum. Antibodies and fluorescence channels, and scale bars, are as indicated in the image panels.

Full size image
Fig. 8
figure 8

Comparative assessment of presenilin-associated protein (PSAP), phosphorylated tau (pTau) and β-amyloid (Aβ) IR in 3xTg-AD and C57BL/6J mouse forebrain. Light PSAP immunolabeling is present in the cerebral cortex and hippocampal formation, localized to the somata and dendrites of pyramidal neurons, which is similar between the 3xTg-AD (A-A3) and C57BL/6J (D-D2) brain sections. pTau IR is distinct in cortical and hippocampal pyramidal neurons in the transgenics (B-B3), which is absent in the C57BL/6J mouse brain section (E-E2). Aβ deposition is seen only in the transgenics, most evident in the subicular and CA1 areas (C-C3; F-F1). CA1 and CA2: Ammon’s horn subregions; DG: dentate gyrus; III-VI: cortical layers; WM: white matter. Scale bars are indicated.

Full size image

In another brain with tauopathy scored as Braak stage V (Additional Fig. 10) and Aβ deposition as Thal phase 4 (Additional Fig. 11), a subpopulation of neurons in the neocortex and entorhinal cortex showed increased PSAP IR (Additional Fig. 12A-D2). Again, the former appeared to be morphologically distorted with a tangle-like appearance (Additional Fig. 12B-B2, and C-C3). Lightly labeled neurons were also present in the cerebral cortex and hippocampal formation, exhibiting a normal-appearing morphology (Additional Fig. 12B-B2, C1-C3, and D-D2).

In the immunolabeled paraffin section with hematoxylin counterstain from an AD case (Fig. 3), the nuclei of the PSAP-enriched neurons were not visible or appeared to be broken or dislocated (Fig. 3A-A4). The PSAP immunoreaction product often appeared to be condensed, unevenly packed or arranged as fibrillary threads. Of note, some deformed cellular profiles exhibited otherwise faint PSAP IR but lacked the hematoxylin-stained nuclei (Fig. 3A2-3, indicated by arrowheads). The PSAP-labeled neurites were mostly thread-like, but not appeared as swollen spheroids. In pTau immunolabeled paraffin sections with hematoxylin counterstain (Fig. 3B), many labeled neuronal somata also showed a loss or dislocation of the nucleus, the presence of fibrillary threads, and shrinkage of somata and proximal dendrites (Fig. 3B1-4). pTau immunoreactive neurites appeared mostly as neuropil threads and elongated dystrophic neurites, However, some pTau-positive neurites appeared as swollen spheroids (Fig. 3B1, and B2).

Elevation of PSAP and pTau protein levels in pAD/AD neocortical lysates

Western blot was carried out to verify an elevation of the PSAP protein along with pTau using the orbit gyrus samples from pathologically evaluated brains with (pAD/AD cases, n = 8, 4 males, 4 females, 84.7 ± 7.9 years) and without (adult and aged cases, n = 8, 5 males, 3 females, 51.1 ± 20.2 years) Aβ and pTau pathologies (Additional Table 1, and Additional file 1). Consistent with previous reports34,36, two major peptide bands migrated at ~ 39 and ~ 35 kDa were distinctly detected in the human cortical lysates. The densities of both bands were heavier in the lysates of the pAD/AD cases relative to negative pathological controls (Fig. 3C). The AT8 antibody blotted bands migrated ~ 130 kDa and 55–70 kDa in the pAD/AD samples, whereas the signal was minimal in the control samples (Fig. 3C). During the revision of the manuscript, it was recommended to also blot the levels of total tau. The total tau blotted by the TAU-5 antibody migrated at 40–70 kDa, with the overall levels appeared to be largely comparable between the pAD/AD and control cases (Fig. 3D). We assessed the levels of Aβ products using a rabbit antibody D12B2, and additionally with the 6E10 antibody during the revision of the manuscript. The D12B2 antibody visualized putative Aβ monomer (4–6 kDa), oligomers and aggregates (10–40 kDa) (Fig. 3D), while the 6E10 antibody detected multiple bands likely representing the full-length APP (~ 100 kDa), putative APP C-terminal fragments and their truncated forms, as well as Aβ and their aggregates of varying molecular sizes (Mosser et al., 2020; Zhang et al., 2024) (Fig. 3E), with the Aβ product (~ 10 kDa) better seen by an overexposure of immunoblot membrane (low panel). Quantitatively, the levels of ~ 39 kDa PSAP product were significantly elevated in the pAD/AD (92.8 ± 32.5%, same format below) relative to control (61.6 ± 25.0%) groups (P < 0.0494, two-tailed unpaired t-test), as were the ~ 35 kDa product (63.1 ± 26.5% vs. 21.7 ± 9.2%, P = 0.0009) (Fig. 3F). The levels of the putative pTau products migrated at ~ 55–70 kDa (AT8 antibody) were elevated in the pAD/AD (251.1 ± 136.1%) relative to the control (45.3 ± 18.1%) sample groups (P = 0.0008; Fig. 3F). The levels of total tau did not reach different between pAD/AD (98.7 ± 31.9%) relative to the control (119.1 ± 12.2%) sample groups (P = 0.114; Fig. 3F). Further, the levels of Aβ monomer (4–6 kDa), oligomers and aggregates (10–40 kDa) blotted with the D12B2 antibody were higher in the pAD/AD (87.5 ± 12.8%) relative to control (26.2 ± 9.8%) groups (P < 0.0001; Fig. 3F).

Double immunofluorescent analysis for PSAP colocalization with pTau and Aβ

In PSAP/pTau double-labeled temporal lobe paraffin sections with DAPI counterstain, a subset of somal and neuritic profiles exhibited colocalization (Fig. 4A). The intensities of PSAP and pTau IR varied among the co-labeled neurons (Fig. 4B-E). We counted all encountered single- and double-labeled somata in 20 screen fields by moving across the PHG and the subicular and CA1 subregions and calculated the percentage rates of the cell types. The rates obtained from the above two regions were comparable, with the data from the PHG presented in Fig. 4. Thus, the PSAP and pTau single-labeled neurons were 30 ± 2.8% and 42 ± 2.2%, respectively, with 28 ± 2.8% co-labeled for PSAP/pTau, while no statistically significant difference excited between the above subpopulations (Fig. 4F). In PSAP/6E10 double immunofluorescence, PSAP-labeled neuritic processes occurred around, but were not colocalized with, the Aβ deposits (Additional Fig. 13). Notably, the DAPI nuclear stain was often reduced or essentially invisible in the PSAP heavily labeled somal profiles, which was noticed in both the PSAP/pTau and PSAP /6E10 double immunofluorescent sections (Fig. 4C-E, and Additional Fig. 13A-D).

Triple fluorescent characterization of PSAP, pTau and tangle colocalization

We further carried out PSAP/AT8 double immunofluorescence along with Amylo-Glo stain, by taking the advantage that this fluorescent dye can stain tangles as well as amyloid deposits. A concurrent staining of amyloid plaques and tangles by Amylo-Glo was clearly seen in the temporal lobe sections from the pAD/AD brains (Fig. 5, and Additional Fig. 14). The fluorescent immunolabeling and Amylo-Glo labeling visualized a row of neuritic plaques across the molecular layer of the dentate gyrus in some cases (Fig. 5A, and B), while labeled plaques and cellular profiles existed in other cortical and hippocampal regions (Fig. 5A-D, and Additional Fig. 14). The amyloid deposits (pointed by hollowed triangles) appeared as bright-blue extracellular amorphous product. Dystrophic neurites occurred around the amyloid deposits, but their relative amounts or areas occupied showed substantial variability when comparing individual plaques or neuritic clusters (Fig. 5B, and Additional Fig. 14A-D). The neuritic clusters were largely co-labeled for PSAP, AT8 and Amylo-Glo (Fig. 5C).

The extent of PSAP/AT8/Amylo-Glo colocalization was differential among the labeled somal profiles. Thus, the labeled somata could be distinguished into five groups: PSAP/AT8/Amylo-Glo triple-labeled (pointed by white arrows), PSAP/Amylo-Glo double-labeled (pointed by white arrowheads), AT8/Amylo-Glo double-labeled (hallowed white arrows), AT8 single-labeled (hallowed white arrowheads), and Amylo-Glo single-labeled (two-tailed white arrows) (Fig. 5A, C, D, and Additional Fig. 14A-F). Based on quantification in the parahippocampal cortex from four pAD/AD brains, 18 ± 5.5% of the somata were triple-labeled for PSAP/AT8/Amylo-Glo, 10 ± 2.2% of the somata were double-labeled for PSAP/Amylo-Glo, and 21 ± 2.5% double-labeled for AT8/Amylo-Glo, while 29 ± 3.5% and 22 ± 11.2% exhibited single labeling for Amylo-Glo and AT8, respectively (Fig. 5E). No statistical difference was found between the above subgroups.

Double immunofluorescent analysis for PSAP with CK1δ and sortilin

GVD bodies are membrane-rich inclusions originally defined in aged human hippocampal pyramidal neurons30. CK1δ is a well-characterized marker for the GVD bodies, which occur primarily in pretangle neurons and reduce with the advance of tangle development30,31,48. In double immunofluorescence, the PSAP-enriched neurons were rarely found to contain CK1δ-labeled GVD bodies (Fig. 6A-E). Based on quantitation in the parahippocampal cortex and the subicular to CA1 regions, about 96% of PSAP-enriched neurons were not co-labeled for CK1δ IR in both regions (Fig. 6F).

Sortilin is normally expressed in cell membrane, endosomes and trans-Golgi network49,50. Intraneuronal sortilin aggregation can form GVD-like inclusions in aged and AD human brains, with sortilin IR in neurons reduced along with the increase of pTau accumulation30. In double immunofluorescence, the PSAP-enriched cellular profiles exhibited reduced or lost sortilin IR (Fig. 7A-E). Based on cell count, about 90 ± 1.8% and 85 ± 5.0% of the PSAP-enriched neurons were not colocalized with sortilin IR in the PHG and subiculum/CA1, respectively (Fig. 7F).

Assessment of PSAP IR in human brains with PART

PART is frequently observed in the brains of aged individuals37,49. Examination of human brains with PART could help establish the relevance of PASP upregulation to tau pathogenesis independent of the Aβ pathology in the brain. Therefore we examined PSAP IR in sections from PART cases (Additional Table 1), with representative data included as supplemental information (Additional Figs. 15, 16, 17, 18 and 19). In the temporal lobe paraffin section from a case with tauopathy at Braak stage III (Additional Fig. 15, and 16), pyramidal-like neurons in the temporal cortex and hippocampal formation exhibited increased PSAP IR, with some exhibited a tangle-like appearance (Additional Fig. 17). In PSAP/pTau double immunofluorescence, the PSAP-enriched neurons were colocalized with pTau IR, and showed a reduced DAPI stain (Additional Fig. 18). The differential somal colocalization patterns described above in the pAD/AD cases were also found in the PSAP/pTau/Amylo-Glo triple labeling in the sections from the PART cases (Additional Fig. 19A-F). Notably, no Amylo-Glo stained amyloid deposition was found in the triple-labeled sections (Additional Fig. 19B-F).

Assessment of PSAP IR in transgenic mouse models of AD

As indicated in the Method section, cryoprotected sections from APP/PS1, 3XTg-AD and 5XFAD transgenics at representative age points before and after the development of cerebral amyloid pathology were immunohistochemically stained for PSAP. The original sets of sections from these mice were pathologically evaluated in detail in our previous studies43,44,45. Overall, no enhanced PSAP IR was observed in the sections of these transgenic mouse brains (data not shown to avoid redundancy). For example, we show a set of representative data here from 18-month-old 3XTg-AD transgenics and adult C57BL/6J mice.

Thus, in both the 3XTg-AD and C57BL/6J mice, light PSAP IR was observed in the cerebral cortex and hippocampal formation, with the labeling seen in the somata and proximal dendrites of the hippocampal pyramidal neurons (Fig. 8A-A3, and Fig. 8D-D2). pTau IR was clearly present in the somata and apical dendrites of the layer V pyramidal neurons (Fig. 8B-B3). The somata, dendrites and axonal processes of the CA1 pyramidal neurons were heavily labeled for pTau, with the mossy fiber terminals in CA3 and the hilus of the dentate gyrus also labeled (Fig. 8B1-B3). The rabbit anti-Aβ antibody labeled the somata and dendrites of layer V and hippocampal pyramidal neurons, with extensive Aβ deposition in the subiculum (Fig. 8C-C3). No specific pTau nor Aβ IR was seen in the sections from the C57BL/6J mice (Fig. 8E-E2, and F-F2).

Discussion

Many proapoptotic proteins are constitutively expressed in the brain and thought to execute non-canonical biological functions51,52,53. Thus, immunolabeling of the caspase family including caspase-2, 3, 6, 7, 8 occurs in normal-appearing neurons in the human brain23,24,54,55,56, which was suggested to regulate neuroplasticity by cleaving substrate proteins involved in cytoskeletal remodeling, neuritic growth, regeneration and degeneration, and cell-cell communication57,58,59,60,61. Some BCL-2 family proteins are intrinsically expressed in neurons in the human brain62,63,64,65, considered to play nonapoptotic biological roles, such as regulation of mitochondrial, endoplasmic and nuclear hemostasis and whole-cell metabolism66. Moreover, several members of the calpain family are ubiquitously in neurons, which may modulate synaptic plasticity67,68,69,70.

In the present study, PSAP IR was observed in neuronal profiles in pathologically-free adult and aged human brains. In the cerebrum, the labeled neurons appeared to be largely the pyramidal neurons. The labeled neuronal somata in the putamen, thalamus and hypothalamus were relatively large in size and mostly multipolar in shape. In the brainstem, PSAP IR was observed in neuronal profiles in the relay nuclei, including the substantia nigra, red nucleus, dorsal raphe nucleus, locus coeruleus and pontine nucleus. In the cerebellum, PSAP IR occurred in the somata and dendrites of Purkinje cells, with some large-sized neurons in the dentate nucleus also labeled. Overall, PSAP expression was not found in glia-like cells. Thus, the constitutive neuronal expression of PSAP implies that this protein may also play some nonapoptotic roles in human brain.

Tangles are a histopathological term defining the principal intraneuronal lesion of AD, which can also occur in aging human brain13,15,16,71,72. The discovery of pTau and development of pTau antibodies advanced the understanding of neuronal tangle formation. Thus, tangles are developed in neurons in a continuum from pretangle to mature tangle, and further to ghost tangle, stages. Structural and functional deficits would occur in the affected cells, especially from the mature to ghost tangle stage, given that ghost tangles only exhibit some cellular outline but have lost much of the nuclear and perikaryal structures in microscopy. As such, ghost tangles are also described as being extracellular. Ghost tangles would be eventually dissolved by tissue clearance mechanisms, resulting in microscopical and macroscopical atrophy. The in vivo timeline of tangle pathogenesis among individual neurons in different brain regions or individual human subjects could not be addressed by cross-sectional postmortem studies, but has been considered to be variable73,74. Nonetheless, pretangle, mature and ghost tangles can coexist in a single microscopic field in a human brain section, depicting a cross-sectional landscape of pathological evolution. pTau antibodies mostly detect pretangle and mature tangle, as well as neurofibrillary neurites and neuropil threads. On the other hand, various histological stains can visualize mature and ghost tangles, which include silver, hematoxylin and eosin, Congo red and fluorescent amyloid tracer stains72. It should be noted that a large number of GVD antibody markers can label pTau positive pretangle neurons, with a trend of loss of the GVD bodies seen as the affected neurons evolve from mature to ghost tangle stage30.

In the present study, we used western blot to detect the elevation of PSAP along with pTau in cortical lysates using frozen samples obtained from the orbit gyrus, wherein the tauopathy is evident by Braak stage IV. Frozen tissues from brains with and without microscopically confirmed tauopathy were assayed for comparison. Because there is an age-related increase in the occurrence of tauopathy in humans, the donor’s ages were not matched between the two sample groups. Thus, the observed intergroup difference in PSAP and pTau protein levels reflected an age-related parallel elevation of the protein products in the cortical samples.

In immunohistochemistry, neuronal and neuritic profiles with increased PSAP-IR were found in pAD/AD as well as PART brains, indicating PSAP upregulation along with tau but not necessarily with Aβ pathogenesis. The double (PSAP/pTau) and triple (PSAP/pTau/Amylo-Glo) fluorescent characterizations indicated that PSAP upregulation was associated with pTau and tangle pathology. Specifically in the triple labeling, the increased PSAP IR was colocalized with mature and ghost tangles (PSAP/Amylo-Glo double- and PSAP/pTau/Amylo-Glo triple-labeled). The pTau antibody (AT8) labeled pretangle neurons (AT8 single-labeled), mature tangles and some ghost tangles (pTau/Amylo-Glo and PSAP/pTau/Amylo-Glo). As with some other tangle-binding probes75,76, the Amylo-Glo dye also singly labeled ghost tangles not colocalized with pTau or PSAP in the sections from the pAD/AD cases. This finding implicates a loss of PSAP and pTau (or the pTau antigenic epitopes) in the progress from mature to ghost tangles. The reduction of PSAP and pTau appeared to be not parallel, such that PSAP/Amylo-Glo and pTau/Amylo-Glo double-labeled cells were found in the sections. The association of increased PSAP with tangle maturation to ghost tangle formation was also supported by other double labeling characterizations in the current study. Specifically, pretangle neurons containing CK1δ immunolabeled GVD bodies rarely exhibited increased PSAP IR. The diminished sortilin (normally located to membrane, endosomal and Golgi components) in the PSAP-enriched neurons indicated a loss of intraneuronal organelles30.

As noted in the Introduction, many proapoptotic proteins are activated in pretangle and tangle-bearing neurons and dystrophic neurites in AD human brains. Some of these previously reported proapoptotic proteins appear to act synergically with PASP based on our earlier cell biological findings. For examples, Bax appears to form a complex with PSAP in the outer membrane of mitochondria33; Bax upregulation is reported in tau-positive tangles in human brain77,79. Caspases appear to act downstream to or interplay with PSAP33,35. Multiple caspase members are shown to be activated in tangle-bearing neurons and dystrophic neurites, including caspase-380, caspase-624,25, caspase-881 and caspase-982. Thus, our current findings suggest that PASP elevation is a part of the proapoptotic programs activated along with tangle pathogenesis in neuronal and neuritic structures. Given that PSAP is a PS1-binding mitochondrial proapoptotic protein, the current findings may be particularly of interest in the context of PS1/γ-secretase and mitochondrial abnormality in AD pathogenesis83. PS1 mutations cause familial AD, therefore may promote AD pathogenesis by mediating Aβ overproduction84. However, a growing body of evidence suggests that presenilin dysfunctions may also relate to mitochondrial deficits including disruption of calcium homeostasis85,86,87, and can causes neuronal degeneration in an Aβ-independent manner88. In this regard, it is notable that the N-terminal and C-terminal fragments of PS1 could be detected microscopically in dystrophic neurites of senile plaques and tangle-bearing neurons89,90,91.

Although intraneuronal pTau accumulation appears to mark a degenerative fate of the affected neurons72, it remains uncertain as to whether the death of these neurons is caused by tau toxicity, disrupted proteinous hemostasis, aberrant pro-/anti-apoptotic signaling, or a combination thereof30,92. Because the increased PSAP IR occurred in partnership with tau/tangle pathogenesis in human brain, there raised a question as to whether pTau accumulation induces PSAP upregulation. To gain a preliminary assessment on this matter, we examined PSAP expression in transgenic mouse models of AD. No enhanced PSAP IR was seen in the brain sections from APP/PS1, 3XTg-AD and 5XFAD transgenic mice examined. Specifically, only light PSAP IR was observed in neuronal somata in the cerebral cortex and hippocampal formation in 18-month-old 3XTg-AD mice despite a substantial presence of both Aβ and pTau pathologies in the brain. Therefore, Aβ, pTau, or together, does not appear to be sufficient to induce microscopically detectable PSAP alteration in the rodent brain. In this regard, it is notable that limited neuronal death has been considered an issue of incompleteness of transgenic rodents in terms of modeling human AD by many investigators93,94,95,96.

Finally, it should be noted that AD is a complex disease with numerous neuropathological changes and multi-domain cognitive deficits. The definition of AD is evolving from a clinicopathological to a biological entity with the advance in the identification of fluid and brain imaging biomarkers14,34,97,98,99. The morphological characterizations in the current study are focused and pathological in essence, with the results revealing a clear microscopical link between PSAP and pTau accumulation including the formation of ghost tangles. The molecular mechanism underlying PSAP elevation remains to be investigated in the future. In addition, whether PSAP is elevated in the cerebrospinal fluid and blood in correlation with AD-type cognitive deficits among preclinical and AD subjects is worth exploring.

Conclusion

The proapoptotic PSAP is constitutively expressed in adult human brain largely among pyramidal/projection neurons in the cerebrum and subcortical structures. Enhanced PSAP immunolabeling occurred in human brains in partnership with neuronal and neuritic tangle formation. The results suggest that PSAP plays a neurobiological role in human brain, with its enhanced expression potentially participating in a proapoptotic activation possibly potentiating tangle-associated neuronal degeneration.