Modulation of SPARC/Hevin Proteins in Alzheimer’s Disease Brain Injury

Subjects: Case selection and neuropathologic assessment

Tissue samples from AD and control cases were obtained from the Massachusetts Alzheimer’s Disease Research Center (MADRC) at Massachusetts General Hospital. Brain tissues from selected donor autopsies performed at Massachusetts General Hospital are accompanied by the reports generated by the Neuropathology Core of the MADRC. These reports include neuropathology diagnosis of AD-associated changes and descriptive clinical information on the severity of neurological signs and symptoms of AD-associated cognitive decline. The study subjects and their next of kin gave written informed consent for the brain donation, and the Massachusetts General Hospital Institutional Review Board approved the study protocol. The subjects fulfilled the National Institute of Neurological and Communicative Disorders and Stroke, Alzheimer’s Disease and Related Disorders Association criteria for probable AD and the National Institute on Aging-Reagan criteria for high likelihood of AD. For this study, only AD brains with Braak stage 6 were chosen from the brain bank. Sections were collected from frozen, non-fixed tissue blocks of temporal cortex and were selected according to tissue quality. AD status was assessed using immunohistochemistry and immunofluorescence (detection of senile plaques and NFTs). In total, 12 AD and 12 control cases were investigated, both of which included 6 female brains and 6 male brains. For AD, the ages ranged between 39–77 years old and mean age of death was 62 (Table 1). Despite the broad age range, the selected patients were not considered early onset cases as Braak staging and histopathological assessment showed clear signs of progressed diseased and protein deposits in isocortical areas. Likewise, the control samples were chosen in the same region and had to be from cases where death was caused by non-neurological and non-systemic diseases (Table 1). The mean age at death for the control cases was 72 years (age range 47–95). Variation in age at death in AD and the control cases was determined by the Mann-Whitney U test, showing no significant differences.

Confocal microscopy and double-immunofluorescence staining (for SPARC/Hevin and neurons, astrocytes, microglia/macrophages/dendritic cells)

Frozen tissue blocks from the temporal cortex were sectioned at 8–10 micron thickness using a cryostat microtome. Tissue slides were fixed in 75% ethanol and blocked with 3% BSA and 5% normal goat serum at 37°C for 15 min. Immunofluorescence staining was achieved by incubating the sections with primary antibodies against intracellular and extracellular SPARC (Santa Cruz Biotechnology) or Hevin (Santa Cruz Biotechnology), followed by labeling these proteins using secondary Alexa Fluor antibodies (Life Technologies). After washing the slides with PBS, sections were co-stained for neuronal and inflammatory cells using markers against NeuN, GFAP, Iba-1, CD11b, and CD83 (primary and secondary antibody combination cf. Table 2). Eventually, slides were mounted using Vectashield mounting medium (Vector Laboratories). Tissue slides were imaged on the Olympus BX51 microscope (Olympus) and/or the Leica True Confocal Scanner (Leica Microsystems). Lipofuscin autofluorescence, which is typically seen in immunofluorescence images of brain tissue from elderly individuals, is visible in both the control and AD tissue. Lipofuscin is highly autofluorescent and difficult to avoid during image acquisition due to its broad emission spectrum (360–647 nm). To ensure that positive signal from protein staining is specific and not lipofuscin autofluorescence, images were also taken under UV excitation. Ultimately, overlaying all channels allowed to discern true positive staining signal from lipofuscin-induced autofluorescence. Through the Olympus microscope, lipofuscin appeared in red, whereas through the confocal scanner it was visible in yellow.

Confocal microscopy and immunofluorescence staining (for Aβ and SPARC/Hevin protein)

Tissue sections were prepared and incubated with primary antibodies (dilution 1:50) against either SPARC or Hevin and secondary antibodies (Alexa Fluor 488, dilution 1:200). After incubation with the Alexa Fluor 488 antibody, the tissue was counterstained with 0.1% -filtered thioflavin-S (Sigma-Aldrich) and incubated at room temperature for 2-3 min. Thioflavin S was dissolved in distilled water and filtered via a 0.22 micron sterile filter. Finally, slides were mounted using Vectashield mounting medium (Vector Laboratories) and imaged on the Olympus BX51 microscope at 60X magnification. SPARC and Hevin (appearing in green), and thioflavin-S staining (blue in the UV channel) were imaged to verify whether SPARC or Hevin expression occurred near Aβ plaques.

Immunofluorescence/-peroxidase staining (for Aβ, SPARC and Iba-1)

Frozen tissue sections were fixed and blocked before the primary mouse anti-human SPARC antibody (Santa Cruz Biotechnology) was applied at 1:50 and incubated for 60 min at 37°C. Likewise, after washing with PBS, the secondary goat anti-mouse Alexa Fluor 488 (Life Technologies) antibody was added 1:200 and incubated for 45 min. Following mounting with PBS, images were taken with a dual filter using the Olympus microscope. Thereafter, Aβ plaques were stained with thioflavin S. After treatment with 80% ethanol and distilled water, slides were mounted with PBS and microscopically examined using a UV filter. After removal of the coverslips, 1% H₂O₂ was added to the sections for 5 min. Following a rinse with PBS, slides were incubated with rabbit anti-Iba-1 antibody at 1:50 at 37°C for 45 min. Subsequent to wash steps with PBS, goat anti-rabbit secondary antibodies (Life Technologies) were applied and slides were incubated likewise. Slides were then washed in distilled water and calorimetrically developed using a DAB staining kit solution (Vector Laboratories). Afterward, slides were dehydrated and cleared through a graded ethanol series (50%–100%) into xylene before they were mounted using Vectashield mounting medium (Vector Laboratories). Eventually, Iba-1 positive microglia were located in the sections and white-light photographed (Olympus BX51).

Laser capture microdissection (LCM)

Frozen tissue sections were lightly fixed in 70% ethanol for 40 seconds. Next, the sections were incubated in 95% and 70% ethanol (each for 2 min), followed by a quick rinse in distilled water before being stained with 0.1% -filtered thioflavin T (Sigma-Aldrich) for 5 min. After brief washes in 80% ethanol and water, the sections were incubated in a toluidine blue solution (Applied Biosystems) for 1 min, followed by dehydration in increasingly concentrated ethanol and xylene solutions prior to microdissection. All incubations were carried out at room temperature. Pre-selected control and plaques-bearing tissue were sectioned onto glass microscope slides (by LCM) and stained via the described methods. Following dehydration through the graded ethanol series, slides were placed in xylene for 5 min, air-dried, and placed into a desiccation chamber prior to LCM. Complete dehydration is a prerequisite for LCM using the Arcturus Pixcell IIe platform. For LCM, the Arcturus Pixcell IIe system was set to deliver a 750 ms duration laser pulse of 65 mW so that a 15-micron diameter spot was produced on macro collecting caps (Arcturus CapSure, Applied Biosystems). Each cap was used to collect small plaques-bearing areas from diseased tissue and small plaques-free areas from control tissue. For qPCR, total RNA was extracted and isolated using the PicoPure RNA Isolation Kit (Arcturus, Applied Biosystems) and reverse transcribed using SuperScript III RT (Invitrogen).

Real-time qPCR

Initially, qPCR analysis of the cDNA product was carried out using a standard panel of housekeeping gene primers including β-actin, tubulin, GAPDH, and 18 S ribosomal RNA. For purposes of assuring the integrity of LCM RNA preparations, we compared newly acquired, plaque regional LCM samples to the PCR activity (cycle number, primer-dimer formation, melt curve analysis) of a standardized control brain cDNA library isolated via LCM. We utilized a real time PCR system (Bio-Rad iCycler) and employed SYBR green mastermix (Bio-Rad Laboratories, Inc.) and 25μl reaction volumes. Real-time qPCR amplification was for 42 cycles: 30 s at 95°C, 60 s at 55°C, and 90 s at 72°C.

For analysis, triplicate samples of cDNA from each LCM sample were probed with human sequence primer sets directed against SPARC (forward: 5-AGCACCCCATT-GACGGGTA-3; reverse: 5-GGTCACAGGTCTCGAAAAAGC-3) and Hevin (forward: 5-GACCAACAGGGAAAACCTCA-3; reverse: 5-TGCAGGCTCCAAAATAATCC-3). Resulting values were normalized to the GAPDH gene (forward: 5-GAGTCAACGGATTTGGTCGT-3; reverse: 5-TGGAAGATGGTGATGGGATT-3). Fold change values were calculated by using control cortex values as reference calibrators and by using the Livak method. A serial dilution experiment resulted in average PCR efficiencies of 94% (R² value: 0.99) for SPARC primers and 91% (R² value: 0.99) for Hevin primers.

SPARC and Hevin expression in astrocytes and neurons

Animal studies on the expression of SPARC and Hevin contributed to a better understanding of the protein function and helped determine the protein-expressing cell type. As for SPARC, astrocytes and microglia have been suggested as expressers [28, 29], but neuronal expression was excluded [30]. In contrast, Hevin expression was attributed to astrocytes and neurons, but not to microglia [28, 31]. Based on the aforementioned findings, we double-stained tissue sections for SPARC/Hevin and the astrocytic marker GFAP. Figure 1a shows SPARC (green) and astrocytes (red) in AD Braak 6 and control cases. Generally, we did not observe any noticeable variances in gene and protein expression related to differences in age, or sex within the groups. Despite the presence of an activated astrocyte (arrowhead Fig. 1a B) in AD tissue (image series A–C), SPARC expression was not associated with astrocytes. When scanning the entire tissue section under the microscope, SPARC appeared only at certain loci of the cortex samples and showed elevated amounts when compared to the controls. Similar results were found for Hevin (Fig. 1b A–C), with the notable exceptions that the amount of protein decreased and that expression was detected throughout the tissue. Hevin is localized in blood vessels and in close proximity to neurons, and is detectable by the presence of lipofuscin, which is secreted by neurons (arrowheads in Fig. 1b C). The open arrowhead in Fig. 1b C gives proof that astrocytes are not Hevin positive. Control tissue did not feature any association of SPARC (Fig. 1a D–F) or Hevin (Fig. 1b D–F) with astrocytes. Interestingly, by only studying the AD tissue, Hevin levels were lower than in control cases, suggesting a downregulation during disease. Regarding expression by neurons (marked with anti-NeuN), neither SPARC nor Hevin showed any association. Hence, as was presented for mice, we confirmed the incapability of neurons to express SPARC in human tissue (data not shown). Against expectations, Hevin immunoreactivity (arrowheads) with neurons was disproved for both AD (Fig. 2A–C) and control tissue (Fig. 2D–F). In summary, both SPARC and Hevin are not expressed by astrocytes or neurons. SPARC is found elevated only in AD cases and there is indication that Hevin may be downregulated in the diseased state. This altered SPARC/Hevin protein expression may contribute to synaptic dysfunction in AD brain as SPARC and Hevin are negative and positive regulators of synaptogenesis, respectively [19, 20].

OPEN IN VIEWER

Fig.1

a) Immunostaining of SPARC and astrocytes in AD (images A–C) and control tissue (images D–F). Series A–C depicts increased AD-associated SPARC expression as compared to control tissue. The extracellular distribution of SPARC (arrowheads in A) is not associated with reactive AD astrocytic GFAP expression (arrowhead in B). Images D–F show that extracellular SPARC (arrowhead in D) is not associated with quiescent control astrocytes (arrowheads in E). Lipofuscin fluoresces in both the red and green channels and appears yellow in the merged images (arrowhead in G and F). b) Immunostaining of Hevin and astrocytes in AD (images A–C) and control tissue (images D–F). Hevin is expressed to a lesser extent in AD samples and is clearly not associated with astrocytic activity, cf. arrowheads in image C. Instead, astrocytes seem to gather around neurons (arrowhead, image C) and blood vessels (open arrowhead, image C). Control tissue has distinct Hevin positive cells that are likely to be neurons (arrowhead, image F) and other cell types (open arrowhead, image F). Yellow structures in C and F are lipofuscin in degenerate neurons or extracellular deposits.

OPEN IN VIEWER

Fig.2

Immunostaining of Hevin and neurons in AD (images A–C) and control tissue (images D–F). Despite the proposal that Hevin is expressed by neurons, all of the images show cell types (arrowheads) other than neurons.

SPARC and Hevin expression in immune cells

Microglia are the resident immune cell population of the brain and were thought to be the exclusive population of the local innate immune system. Today, there is increasing evidence that other blood-derived immune cells, such as macrophages and DCs might extravasate into diseased or injured tissue. Microglia are Iba-1- and CD11b-positive. Although CD11b (or type 3 complement receptor, CR3) is present on activated microglia [32], it is in addition a marker for myeloid-derived macrophages and migrating DCs [33]. High-level expression of CD11b facilitates increased cellular adhesiveness to vascular endothelium, thus initiating the extravasation of myeloid cells into the inflamed tissue. To detect activated mature DCs with the ability of priming naive T-cells, a suitable marker is CD83 [34]. This surface molecule helps differentiate between immature DCs (iDCs) and mature DCs (mDCs). As CD83 + cells are also CCR7 + (a chemotactic receptor), there is indication that mDCs travel to the lymphatic system where they induce a primary immune response by antigen-presentation to T- and B-cells. Figure 3a A demonstrates SPARC+/CD11b + immunoreactivity in AD tissue, adumbrating activated hypertrophied microglia or activated macrophages expressing SPARC and taking care of damaged neurons. The detection of lipofuscin is indicative of the presence of (former) neurons or neuronal debris. SPARC+/CD83 + cells (Fig. 3a B) were found in AD tissues where they were distributed to a lower amount as compared to CD11b + cells. SPARC was positively colocalized to CD83, indicating that mDCs possess the property of SPARC expression (Fig. 3b). In control cases (not shown), CD83 + cells were restricted to expression in vasculature, suggesting an intact BBB function. Moreover, only the resting microglia showed CD11b positivity in controls, which is an indication of their regular tissue-scavenging function.

OPEN IN VIEWER

Fig.3

a) Immunostaining of SPARC positive cells in AD tissue. A: CD11b positive macrophages colocalized and associated with SPARC expression. B: CD83 positive cell expressing SPARC. b) Colocalization of SPARC and CD83, demonstrating the ability of dendritic cells to induce SPARC expression. This image shows the diseased state. Controls (not shown) do not exhibit such expression patterns.

Double staining against Hevin and CD11b/CD83 resulted in images that showed a clear association of Hevin with immune cells. Hevin+/CD11b + cells were detected throughout AD tissue; however, they were detected in a much lower abundance than SPARC+/CD11b+. In accordance with published data, microglia lacked the ability to produce Hevin [28, 31]. Based on the relatively big cell size observed in our studies, Hevin+/CD11b + cells are proposed to be activated macrophages (Fig. 4a). Moreover, there are lipofuscin vesicles in close proximity, indicating a pre-existence of neurons. Arrowheads in Fig. 4b A highlight the densely distributed Hevin that is associated with the CD11b immunoreactivity and may be involved in tissue remodeling processes. This finding was further confirmed in Fig. 4b B that features lipofuscin vesicles (from pre-existing neurons) in close proximity to Hevin+/CD11b + cells (arrowheads). Figure 4b B shows CD11b+cells (indicated by asterisks) that actively express Hevin around pathologic areas (arrowhead), and this expression confirms the tissue remodeling characteristics. Compared to AD tissue, the control tissue featured only slightly increased amounts of Hevin, which, was mainly expressed in the grey but not the white matter. Interestingly, Hevin expression was not limited to diseased areas but was also detected in oddly shaped patterns, as demonstrated in Fig. 4c. Hevin+/CD11b + cells were located within those same structures and expressed Hevin extensively, which was highlighted by the arrowheads. As mentioned, UV exposure helped to differentiate lipofuscin from a positive signal, as presented in Fig. 4c B. When comparing the images of CD11b and UV/lipofuscin it becomes clear that the yellow color in Fig. 4c A is a result of overlapping Hevin and CD11b signal. Control cases did not have those staining patterns. In addition, AD tissue revealed the presence of paired immunoreactive cells, one being Hevin+, and the other being CD11b + (Fig. 4d A, B). Intriguingly, these cells seemed to interact strongly, as the cell-to-cell interactions appeared in yellow color (Fig. 4d B). Lipofuscin was detected around the cells but without association to either CD11b or Hevin expression. As for DCs, Hevin showed very little immunoreactivity with CD83 (Fig. 4e) and reacted in only a few AD cases. Since most AD samples were free of CD83 + cells, we concluded, despite the indication of positive Hevin expression by CD83 + cells, that mDCs are not the major cell population expressing Hevin.

OPEN IN VIEWER

Fig.4

a) CD11b positive cell expressing Hevin in AD tissue. Colocalization is indicated by color overlay, highlighted here by white arrowheads. Based on the cell-size this cell could be a macrophage. b) Hevin expression in AD tissue. A: Hevin is expressed in areas where lipofuscin is detectable – an indication of active macrophages taking care of dying neurons and neuronal debris. B: But Hevin is also detected around plaques (arrowhead). CD11b positive microglia (asterisks) are located in pathological areas. We suggest that Hevin induces alternative activation of microglia for tissue repair. c) A: Hevin positive structures contain CD11b positive cells that, at the same time, are strongly Hevin positive, indicated by arrowheads. B: Lipofuscin is not associated with Hevin positivity. Arrowheads highlight a Hevin+/CD11b + cell that is free of lipofuscin as the yellow signal disappears upon UV exposure. d) Cells that show different expression of specific cell markers. C demonstrates Hevin immunoreactivity, whereas D is positive for CD11b. B is an enlarged view at the cells from A. The yellow-appearing color suggests a strong interaction of both cells. e) Activated dendritic cell in AD tissue. CD83 immunoreactivity gave proof of the presence of DCs in the brain. Hevin expression as shown in B, however, cannot be correlated to DCs. Hevin positivity in respect to CD83 was neglected although image A appears to show Hevin positivity in the overlay. Overlooking the brain sections, there was hardly any Hevin immunoreactivity.

As mentioned before, Hevin expression is decreased in AD patients compared to the controls. However, when expressed, Hevin is found around pathologic areas. Here, Hevin is proposed to be expressed mainly by macrophages and to possess immune-modulating properties, especially in the conception of injury repair and tissue remodeling. SPARC, by contrast, is expressed by microglia and macrophages, but is not restricted to either cell type. mDCs can also express SPARC, suggesting that its role is in acute inflammation and modulation of the immune response. The high protein levels during disease confirm the results from the immunostaining. It could be possible that there are distinct interactions between SPARC and Hevin; however, it remains difficult to interpret interactions only based on co-localization found in micrographs.

SPARC is colocalized to pathologic hallmarks of AD

Believing that SPARC regulates microgliosis and modulates the immune response, we were interested to examine if SPARC was colocalized to the hallmarks of AD. We stained for SPARC and AD-specific plaques (using thioflavin S) and discovered that SPARC expression occurred in close proximity to AD-related protein aggregates (Fig. 5a). SPARC signal was detected next to or surrounding the plaques (Fig. 5a A–C), or right in the middle of the plaques (Fig. 5a D), suggesting that SPARC + cells are able to locate the protein deposits and modulate the immune response. The immunostainings we have performed (cf. Fig. 3a, b) showed CD11b and CD83 positivity, suggesting that SPARC expression may stem from microglia/macrophages, or mDCs, respectively. Here, we propose microglia as a key population for SPARC expression. We believe that the expression of SPARC follows a temporal manner and that other activated immune cells follow with expression upon respective stimulus. In order to corroborate SPARC’s contribution to inflammation, we applied immunofluorescence/-peroxidase staining. Figure 5b shows immunoreactive SPARC (Fig. 5b A) around microglia (confirmed in Fig. 5b C) that are in the middle of AD-related senile plaques (visualized by thioflavin S via UV-filter, Fig. 5b B).

OPEN IN VIEWER

Fig.5

a) SPARC is colocalized to amyloid plaques. SPARC expression (green) is increased around pathological protein deposits (ThioS positive structures) (A–D). To differentiate between true ThioS signal and lipofuscin autofluorescence, the SPARC channel (appearing in green) has been overlaid with the red channel (unstained, but lipofuscin autofluorescence). Image B shows dying neurons covered in plaques and surrounded by SPARC. Image D depicts immune cells accumulating in the plaques area. b) SPARC is expressed by microglia found in close proximity to pathological protein aggregates. A: Positive SPARC staining, shown in green and auto-fluorescent lipofuscin in red. B: White-light image of Iba-1 positive microglia, immunoperoxidase stain. C: Thioflavin S staining against amyloid plaques. D: Merged image of A and C at the exact same location as in B, allowing for the conclusion that microglia express SPARC and that there is an association between SPARC expression and the pathological hallmarks of Alzheimer’s disease.

Gene expression differences using real-time qPCR

To examine the differences in gene expression during AD, real-time qPCR was applied. Figure 6a shows the results obtained from four identical experiments based on eight patients. Although statistically not significant, this data confirmed previous findings that SPARC expression is increased during AD. Since Hevin is downregulated in AD, the fold change was presented as a negative bar. Because we were interested in SPARC’s contribution to neuroinflammation, we analyzed the expression changes by microdissecting tissue from different loci of the grey matter. Presented in fold change, Fig. 6b shows the alteration in gene expression. The results were juxtaposed to the same control group as from whole tissue lysates. Control tissue is considered free of plaques, i.e., non-plaques, and should ideally be identical to the baseline. However, LCM and subsequent RNA extraction can lead to a loss of RNA integrity and quality. Comparing AD plaques-containing tissue with AD plaques-free tissue, a significant alteration in SPARC expression was detected. The real-time qPCR results confirmed the colocalization of SPARC to senile plaques and its pivotal role in the immunological changes that are present in AD brain.

OPEN IN VIEWER

Fig.6

a) qPCR results for fold-change SPARC and fold change Hevin, resulted from whole lysate tissue of AD patients. AD-specific values were normalized to endogenous genes and control values. Although statistically insignificant, SPARC appears to be up- regulated and Hevin downregulated in AD. b) SPARC is upregulated and colocalized to pathological plaques in AD patients. After microdissection (via LCM) and RNA isolation, plaques areas reveal the most potent changes. qPCR results from LCM-cut tissue is compared to values from whole lysate treatment (right bars).