The work involving experimental animals was conducted under the protocol approved by the Institutional Animal Care and Use Committee at the Medical College of Georgia, Augusta University. All experimental animal procedures performed in this study were in compliance with the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes. Experiments were carried out in nine to 10-month-old male and female APP/PS1 mice, which are double transgenic mice expressing a chimeric mouse/human amyloid precursor protein (Mo/HuAPP695swe) and a mutant human presenilin 1 (PS1-dE9), both directed to CNS neurons (MMRRC strain: 034832-JAX, The Jackson Laboratory; Jankowsky et al., 2004). The mice are on the C57BL/6J genetic background. The mice were housed in the animal care facility and accessed rodent chow and tap water ad libitum with a 12-h light:dark cycle.

Animal behavioral tests were conducted in the small animal behavior core lab in Augusta University.

Recombinant adeno-associated virus 9 (AAV9) was constructed and purchased from VectorBuilder. Mouse ADAM17-mCherry-AAV9 and eGFP-AAV9 were used in the following experiments. Mouse ADAM17-mCherry-AAV9, or eGFP-AAV9 controls, were used to overexpress ADAM17 in the APP/PS1 mice through intravenous tail-vein injections (particle number: 1 × 10¹¹ of AAV9 diluted in 200 μL sterile PBS, total volume of 300 μL injected). Mice were assessed for functional and morphological endpoints 3 months after injections.

The method used for mice brain microvessel isolation was performed as described previously (Paraiso et al., 2020). Briefly, the brain tissue was placed in a glass tube containing 8 mL of ice-cold HBSS solution (14025-076, Gibco) and homogenized. The brain homogenate was mixed with an equal volume of 40% Ficoll (17030010, Cytiva) to a final concentration of 20% Ficoll in HBSS solution. The tubes were shaken before centrifugation and the resulting homogenate was centrifuged at 5,800g for 20 mins at 4°C. The pellet was gently washed with 5 mL of 1% BSA/HBSS (BP1600, Fisher BioReagents). It was then filtered with 300 μm nylon mesh and the flow through was washed with 1% BSA/HBSS. The mixture was filtered through a 30 μm strainer, with a speed of 1 drop every 2 s. The brain microvessels were captured on top of the strainer. The 30 μm strainer was inverted and rinsed with 10 mL of 1% BSA/HBSS, to wash the vessels down. The wash-through was centrifuged at 5,800g for 20 mins at 4°C. The supernatant was discarded and the vessels pellet was saved for further analysis.

Whole brain samples and the isolated brain microvasculature were homogenized in radio-immunoprecipitation assay (RIPA, R0278, Sigma) buffer mixed with 1% protease inhibitor cocktail (P8340, Sigma). Protein concentration was measured by Bradford assay. Equal amounts of protein were loaded for gel electrophoresis. After blotting, membranes (Hybond-P, GE Healthcare) were probed with a rabbit polyclonal anti-ADAM17 antibody (1:1000, ab2051, Abcam), followed by incubation with a HRP linked secondary antibody (anti-rabbit IgG, 7074S, Cell Signaling). Enhanced chemiluminescence was visualized autoradiographically by ChemiDoc XRS+ (170-5061, BioRad). Protein expression was normalized to β-actin (1:1000, 8H10D10, Cell Signaling).

Mice, under deep anesthesia, were decapitated and the brain was removed. The bottom section of the brain was placed in ice-cold artificial cerebral spinal fluid (ACSF) solution, equilibrated with a gaseous mixture of 10% O₂-5% CO₂ – balanced nitrogen, at a pH of 7.4. Mice were euthanized by exsanguination and removal of the aorta; other vital organs were saved and used for further assessments. With the use of microsurgical instruments and an operating microscope, the posterior cerebral artery was isolated and transferred into an organ chamber containing two glass micropipettes, filled with ACSF solution. The artery was cannulated at both ends and the micropipettes were connected with silicone tubing to a hydrostatic reservoir to set the intraluminal pressure to 50 mmHg. Artery preparations were incubated for a 1-h period; spontaneous arterial tone developed in response to 50 mmHg pressure during the incubation period, if spontaneous tone did not develop during the incubation period, U46619 (~10^-8 M) was applied to the arteries to induce an approximately 25% tone. Arteries that did not develop a spontaneous tone or could not be preconstricted were excluded from the analyses. After the incubation period, diameter changes were measured with a videocaliper (Colorado; Bagi and Koller, 2003; Erdei et al., 2006) in response to cumulative concentrations of the endothelium-dependent agonist, acetylcholine (ACh, 10^-9–10^-6 M, Sigma), and to the direct vascular smooth muscle acting, nitric oxide (NO) donor, sodium nitroprusside (SNP, 10^-9–10^-5 M, Sigma).

At the end of the experiments, the internal and outer artery diameter was measured in response to incremental increases in intraluminal pressure from 10 to 70 mmHg, in 20 mmHg increments, in the absence of extracellular calcium (calcium-free ACSF solution). The normalized artery diameter was calculated (as measured at 10 mmHg intraluminal pressure). The wall stress (s) was calculated as follows: s = P × D/2WT, where P is pressure and 1 mmHg = 1334 dyne/cm², and as described (Ohanian et al., 2014). The incremental elastic modulus (Einc) was calculated as follows: Einc = (Δp/Δro)2rix2ro/(ro2−ri2), where ri is the inner, ro is the outer radius, Δro is the change in outer radius in response to intraluminal pressure change of Δp, as described (Mericli et al., 2004).

Mice were anesthetized using isoflurane (1-3%) and were perfused transcardially with 20 mL of room temperature 0.01 M PBS followed by 20 mL of cold 4% formaldehyde (sc-281692, Santa Cruz Biotechnology). Then, the mouse brains were dissected and post-fixed in 4% formaldehyde overnight at 4°C. Brains were cryoprotected with 30% sucrose in 0.01 M PBS for a minimum of 72 h. The brains were cut into 40 μm thick cryosections and saved in a cryoprotectant solution (50 mM PBS, ethylene glycol, and glycerol) at −20°C. For immunohistochemical staining, brain slices located at +1.32, −0.82 (cortex), and −1.64 −2.92 (hippocampus and cortex) were used. After being removed from the cryoprotectant solution, brain slices were washed with 0.01 M PBS three times at room temperature, 45 mins per wash. Brain slices were permeabilized with 0.01 M PBS containing 0.5% Triton X-100 (T8532, Sigma) for 10 mins at room temperature. Brain slices were blocked with 0.01 M PBS containing 10% normal horse serum for 1 h at room temperature. Primary antibodies were applied in 0.01 M PBS containing 0.3% Triton X-100 for 48 h at 4°C. To identify β-amyloid plaques, the primary chicken anti-amyloid beta antibody was used (1:1000, AP31802PU-N, Origene) followed by incubation with secondary antibody (1: 3000, anti-chicken, AlexaFluor^®555) for 4 h at room temperature. DAPI (H-2000, Vector Laboratories) was used for nuclear staining. Structured illumination microscopy (SIM-Apotome, AxioImagerM2, CarlZeiss) was used for immunofluorescent detection. The analysis of positive β-amyloid staining was quantified using ImageJ.

For cerebral vascular network reconstruction, the thickly cut brain slice sections (40 μm) described in the previous section were immuno-labeled with Tomato-Lectin DyLight 594 Antibody (Vector Laboratories, DL-1177, 1:150, overnight, 4°C). DAPI was used for nuclear staining. Structured illumination microscopy (SIM-Apotome, AxioImagerM2, CarlZeiss) was used for immunofluorescent detection. Images were taken using z-stack imaging, with images taken every 0.5 μm, using 20X magnification for vascular reconstruction. Per animal, three fields of view were taken using z-stack imaging and averaged. After obtaining z-stack images via immunofluorescence microscopy, image stacks were uploaded into Vesselucida360^® software for unbiased reconstruction and consequent analysis. Vessels were reconstructed using the Automatic Rayburst Crawl tracing option with vessel tracing set at a sensitivity of 70. Refinement of tracing seeds was between 1 and 3 based on the sample. User-guided manual tracing was used following automatic tracing, with the Rayburst Crawl tracing option, followed by manual tracing of any remaining vessels. Reconstructions were exported to the Vesselucida Explorer^® software for automatic quantification of vessel network parameters.

For Mass Spectrometry, whole brain samples were homogenized in radio-immunoprecipitation assay (RIPA, Sigma) buffer mixed with 1% protease inhibitor cocktail (Sigma). Afterwards, 50 μg of protein per sample was sent to the Proteomics Core Laboratory at the Medical College of Georgia for protein analysis via liquid chromatography mass spectrometry analysis (LC-MS/MS). Samples were digested for 16 h with trypsin at 37°C, trifluoroacetic acid was added to a final concentration of 0.1% (v/v) to stop the digestion. Samples were centrifuged at 15000g for 5 mins and supernatants were used for the LC-MS analysis. Digested peptide samples were analyzed on an Orbitrap Fusion tribrid mass spectrometer (Thermo Scientific), with an Ultimate 3000 nano-UPLC system (Thermo Scientific) connected. Peptide samples were first trapped on a Pepmap100 C18 peptide trap (5 μm, 0.3 × 5 mm). Cleaned peptides were further separated on a Pepman 100 RSLC C18 column (2.0 μm, 75 μm × 150 mm) at 40°C, using a gradient of 2% to 40% acetonitrile with 0.1% formic acid over 120 min at a flow rate of 300 nL/min. Eluted peptides were introduced into the Orbitrap Fusion MS via nano-electrospray ionization source at a temperature of 300 °C and a spray voltage of 2000 V. The peptides were then analyzed in the MS using data-dependent acquisition in positive mode with the Orbitrap MS analyzer for precursor scans at 120,000 FWHM from 400 to 1600 m/z (with quad isolation) and the ion-trap MS analyzer for MS/MS scans at top-speed mode (3-s cycle time), with dynamic exclusion settings (repeat count 1 and exclusion duration 15 s). Higher-energy C-trap dissociation (HCD) was used to fragment the precursor peptides with a normalized energy level of 30%. Raw MS data were processed via the Proteome Discoverer software (ver 1.4) and submitted for SequestHT algorithm against the SwissProt mouse protein database. SequestHT search parameters were set as 10 ppm precursor and 0.6 Da product ion mass tolerance, with static carbidomethylation (+57.021 Da) for cysteine and dynamic oxidation for methionine (+15.995 Da). The percolator peptide spectrum matching (PSM) validator algorithm was used for PSM validation. Proteins unable to be distinguished based on the database search results were grouped to satisfy the principles of parsimony. Proteomic data was analyzed for further analysis if the ΣPSM value ≥2 per group analyzed (WT, eGFP, ADAM17) using unpaired student t-test between two group comparisons. Principal component analysis was run using AtlAnalyze software. Statistically significant (p < 0.05) proteins identified were used in heat-map analysis (-log₁₀(p-value) vs WT) and for further pathway analysis. Gene Ontology pathway analysis with statistically significant proteins was performed (Panther Classification System). For GO pathway analysis the following parameters were used: Analysis Type: PANTHER Overrepresentation Test; Annotation Version: GO Ontology database; Reference List: Mus musculus (all genes in database); Annotation Data Set Used: GO Biological Process complete, GO Molecular Function complete, GO Cellular Component complete; Test Type: Fisher’s Exact; Correction: Calculated False Discovery Rate (FDR). GO pathways identified with an FDR < 0.05 were used for further analysis. GO pathway analysis was visualized in scatterplot format using multidimensional scaling (MSD) whereby semantically similar GO pathways can be visualized in close proximity (Revigo).

All statistical analyses were performed using GraphPad Prism Software. Data were drawn to analyze after being tested for normality using Kolmogorov-Smirnov test. When meeting normality data comparisons between groups were analyzed by two-way repeated-measures ANOVA followed by Sidak’s multiple comparisons test, or with a two-tailed, unpaired Student t-test, as appropriate. A non-parametric Kruskal-Wallis test was used when normality assumptions were not met. Data are expressed either as mean±SEM or box-and-whisker plots, in which the minimum, the 25th percentile, the median, the 75th percentile, and the maximum are presented, and + indicates the mean of values. P < 0.05 was considered statistically significant.

To confirm impaired short term spatial memory and learning function in the APP/PS1 AD mouse model, 9–10-month-old, female and male, APP/PS1 and WT mice were subjected to a Morris Water maze (MWM) test. We found that the individual latency curves of APP/PS1 and WT mice were similar on day 1 and 2 in the learning phase, indicating that the APP/PS1 mice did not differ in swimming ability, or with motivation to escape from the pool. From day 3 to 5, however, acquisition trials revealed a progressive and statistically significant decrease in latencies in APP/PS1 mice compared to WT mice (Figure 1A). During the probe trial, APP/PS1 mice spent longer times to reach the platform area, spent less time in the platform area upon finding it, and had a trend of less platform line crossings (Figures 1B–D). There were no differences between WT and APP/PS1 mice for the percentage of spontaneous alternations or in velocity (locomotion; Figures 1E,F).

ADAM17 has been implicated in various cardiovascular pathologies, yet there have been only limited number of studies elucidating the role of vascular ADAM17 in neurogenerative diseases. Here, we found that while the protein expression of ADAM17 was similar in WT and APP/PS1 mouse whole brain lysates (Figure 2A), the expression of ADAM17 was remarkably reduced in purified cerebral microvessel preparations obtained from the APP/PS1 mice, when compared to those of WT mice (Figures 2B,C).

In order to delineate the role of vascular ADAM17 in impaired short-term memory in the APP/PS1 mice, we used a systemic, AAV9-mediated genetic delivery approach to increase ADAM17 expression. We found that 3 months after the delivery of the ADAM17-AAV9 construct, the protein expression of ADAM17 in cerebral microvessels of APP/PS1 mice was augmented compared to eGFP-AAV9 injected mice and that the expression level was similar to that measured in the cerebral microvessels of age-matched WT mice (Figures 2D,E).

We then determined the effect of ADAM17 re-expression on short-term memory and cognitive dysfunction in the APP/PS1 mice by performing multiple, independent behavioral tests. WT mice and APP/PS1 mice receiving eGFP-AAV9 served as controls. First, the ADAM17-AAV9 or eGFP-AAV9 injected mice were repeatedly subjected to the Morris Water Maze (MWM) test. We found that, 3 months after ADAM17 re-expression, APP/PS1 mice spent shorter times to reach the platform area (A-40) and spent longer times in the platform area, while APP/PS1 mice that received the eGFP-AAV9 injection displayed increased latency curves, similar to the level of APP/PS1 mice before the AAV injections (Figures 3A,B).

Mice in the experimental groups were additionally subjected to spontaneous Y-maze and Novel Objective Recognition (NOR) tests. In the spontaneous Y-maze, the eGFP-AAV9 injected APP/PS1 mice had a reduced number of arm-entries compared with the WT and ADAM17-AAV9 injected APP/PS1 mice (Figure 3C). There was no difference observed in the percent of spontaneous alternations between the three groups (Figure 3D). In the NOR test, compared to the age-matched WT mice, the APP/PS1 eGFP-AAV9 injected mice had a reduced delta novel-familiar score (Figure 4A), recognition index (novel/familiar, N/F) (Figure 4B) and discrimination index (d2 ratio) (Figure 4C). These parameters were similar in WT and the ADAM17-AAV9 injected APP/PS1 mice (Figures 4A–C). There was no significant difference in the total exploration time among the experimental groups (Figure 4D). Taken together, these observations indicates that re-expression of ADAM17 in APP/PS1 mice improved some of the indices of short-term memory and cognitive function to a level comparable to the aged-matched WT control mice.

Brain sections of experimental mice were immunostained for amyloid beta (Aβ) for semi-quantitative assessment of cortical amyloidosis. When compared to age-matched WT mice, we found that there was a significant increase in the number of Aβ plaques in the APP/PS1 mouse cortex receiving the control eGFP-AAV9. Whereas we found no significant effect, i.e. decrease in the number of Aβ plaques, in the APP/PS1 mice injected with the ADAM17-AAV9 construct (Figures 5A,B).

We assessed cerebral artery vasodilator function and vascular biomechanics in WT, as well as eGFP-AAV9 and ADAM17-AAV9 injected APP/PS1 mice using isolated and pressurized cerebral arteries. We found that the endothelium-dependent vasodilation in response to acetylcholine (ACh) was significantly reduced in eGFP overexpressed APP/PS1 mice, when compared to WT controls (Figure 6A). ADAM17 re-expressed APP/PS1 mice showed an improved vasodilatory response to acetylcholine (ACh) compared to the eGFP overexpressed APP/PS1 mice; no significant difference was observed between the ADAM17 re-expressed APP/PS1 mice and WT mice (Figure 6A). There were no differences in the nitric oxide donor, sodium nitroprusside (SNP)-induced, vascular smooth muscle acting vasodilation in the isolated cerebral arteries of WT, eGFP, or ADAM17 overexpressed APP/PS1 mice (Figure 6B).

We also found that cerebral arteries of APP/PS1 (eGFP-AAV9) mice displayed a reduced passive diameter (in calcium free PSS and normalized to 10 mmHg) to incremental increases to intraluminal pressure (from 10 to 70 mmHg), when compared to WT controls (Figure 6C). These changes were accompanied by increases in artery wall circumferential stress and elastic modulus, as well as elastic modulus wall stress relationships in the APP/PS1 (eGFP-AAV9) mice (Figures 6D–F). Delivery and re-expression of ADAM17 in the APP/PS1 mice did not significantly affect these biomechanical properties of the cerebral arteries (Figures 6D–F). Taken together, these results indicate that re-expression of ADAM17 was associated with selective improvement of endothelial-dependent vasodilator function in the cerebral arteries of APP/PS1 mice.

In order to quantify vascularization and microvessel density of the cerebral cortex, the vascular tree was immunofluorescently labelled and in thick sections (40 μm), z-stack images were used for 3-dimensional reconstruction followed by unbiased quantification of microvascular networks (Figure 7A). The microvascular networks in the WT, eGFP control, and ADAM17 re-expressed APP/PS1 groups showed no differences in the measured parameters, including, total vessel length, surface area, vessel volume, or the number of branching nodes (Figures 7B–E).

We performed mass spectrometry analysis to identify proteins that were differentially expressed in the brain of the APP/PS1 mice, which were potentially affected by the AAV-mediated re-expression of ADAM17. Liquid chromatography mass spectroscopy (LC-MS) was performed in triplicate per group. First, we wanted to identify the general trend in the proteomic profile between WT, APP/PS1 + eGFP, and APP/PS1 + ADAM17. Via a principal component analysis (PCA), we saw a differential trend in the protein profile in APP/PS1 + eGFP mice versus the WT control mice. Whereby in APP/PS1 mice with ADAM17 re-expression, we saw an overall trend in the protein profile shifted towards the WT mice (Figure 8A). To further elucidate microvascular proteomic changes due to ADAM17 re-expression, we analyzed individual proteins identified via LC-MS, with a cut-off of a ΣPSM (peptide-spectrum match) of 2 per group (Figure 8B). Through these proteomic analyses, multiple proteins were identified as being significantly downregulated in the APP/PS1 + eGFP mice, that were augmented via ADAM17 re-expression, including numerous proteins that have been studied in relation to neurodegenerative diseases, such as Septin, Ankyrin-2 (Ank2), and Moesin (Msn), among others (Figure 8C).

Furthermore, Gene Ontology (GO) pathway analyses (Panther^®) were completed in order to identify significant differential pathways that ADAM17 may impose its beneficial effects through. Using the significant identified proteins, biological processes, molecular functions, and cellular components were identified using an FDR of <0.05. The top 15 significant pathways per analysis are listed with the corresponding enrichment scores (Figure 9A). Moreover, these GO biological processes identified were mapped (Revigo^®) to identify connected pathways, showing significant differences occur in pathways connected to amyloid precursor protein as expected, as well as in many metabolic pathways and in cellular organizational pathways (Figure 9B).

There are several limitations to our study. It should be noted that our results provide no direct evidence for the mechanistic, causative link between impaired/improved vasodilation and reduced/restored ADAM17 expression in cerebral microvessels, which has yet to be elucidated and will be part of future studies. We did not measure overall, basal, or stimulated cerebral blood flow changes in experimental groups, which could also impact and determine abnormal cognitive functioning and the role of ADAM17 in the APP/PS1 mice. In addition, possible mechanisms independent from that of a microvascular mechanism in leading to impaired or improved cognitive functioning, which can also be related to ADAM17 in APP/PS1 mice, cannot be excluded. For example, we cannot exclude the role of astrocytes, microglia, or neuronal mechanisms, which we did not evaluate in this study. In this regard, it should be noted that we used systemic delivery and re-expression of ADAM17 in the APP/PS1 mice. Although we measured changes in ADAM17 protein expression in purified microvascular preparations, the expressional changes in other aforementioned cell types could have an impact on AD related neuropathological changes and memory and cognitive function in the APP/PS1 mice, which has yet to be delineated in future studies.

In conclusion, our study identifies a reduced expression of microvascular ADAM17 as a novel mechanism by which microvascular dysfunction occurs in an AD mouse model and by which it could contribute to the development of cognitive decline and memory deficits. We propose that targeting and potentially restoring normal activation of ADAM17 to improve microvascular endothelial function could be an effective approach, as it appears to improve some of the key molecular abnormalities previously implicated in AD pathogenesis.