*2.3. Acute Subdural or Cerebral Bleeding(S)*

In both the first specimens (evaluated in mice aged 15 months) and the second specimens (evaluated in mice aged 21–23 months), no acute subdural or cerebral bleeding was found. However, in the second specimens in the aspirin group, small accumulations of erythrocytes were found in one mouse around a leptomeningeal artery with lymphocytic infiltration (Figure A2a). Hemosiderin depositions (arrow heads), representing old bleeding, were also seen around a leptomeningeal artery with vasculitis in the same mouse (Figure A2b). No accumulations of erythrocyte were found in either the control and the cilostazol group.

#### *2.4. Cerebral Micro-Hemorrhages*

The representative images of cerebral micro-hemorrhages are shown in Figure 3. In the first specimens (evaluated in mice aged 15 months), there were no differences in the number of cerebral micro-hemorrhages between the control group (median, interquartile range (IQR): 1, 0–2) and the aspirin group (1, 0.5–3.5) or the cilostazol group (0, 0–3.5) (*p* = 0.730 and *p* = 1.000, Mann–Whitney *U* test: Figure 4a). In the second specimens (evaluated in mice aged 21–23 months), there were also no differences in the number of cerebral micro-hemorrhages between the control group (3.5, 1.5–4) and the aspirin group (4, 1–7.5) or the cilostazol group (5, 2.5–8) (*p* = 0.804, and *p* = 0.277, Mann–Whitney *U* test, respectively: Figure 4b).

**Figure 3.** Representative images of cerebral micro-hemorrhage findings (a, b, Perls's Berlin blue stain with Nuclear Fast Red (Kernechtrot stain solution)); b, d, double-labeled for amyloid (brown) and hemosiderin (blue)). (**a**) Clusters of hemosiderin staining are shown in the brain parenchyma (cortex at 1.35 mm behind Bregma). (**b**) In an adjacent section to a, some of them are in contact with amyloid-β (Aβ)-positive vessels. (**c**) Localized hemosiderin shown around the vessel wall (cortex at 0.85 mm behind Bregma). (**d**) In an adjacent section to c, localized bleeding to amyloid-laden vessels is shown. Scale bars indicate 10 μm.

**Figure 4.** The number of cerebral micro-hemorrhages in the control, aspirin, and cilostazol groups evaluated at 15 months (**a**) and 21–23 months (**b**). Light-red-colored bar graphs indicate median. Boxes and bars indicate median and interquartile range, respectively. Red circles represent the number of cerebral micro-hemorrhages of each mouse. The numbers next to the red circles indicate the number of cerebral micro-hemorrhages, which are out of range of the vertical axis scale in two mice (b, Aspirin group).

#### *2.5. CAA Burden*

The representative images of CAA burden are shown in Figure 5. In the first specimens, there were no differences in the CAA score between the control group (median, IQR: 2, 1–4) and the aspirin group (6, 1–7) or the cilostazol group (2, 1–8) (*p* = 0.556 and *p* = 1.000, Mann–Whitney *U* test: Figure 6a). In the second specimens, there was no difference (*p* = 0.750, Mann–Whitney *U* test) in the CAA score

between the control group (28.5, 24.5–58) and the aspirin group (28, 23–47). Even after an outlier of the CAA score in the aspirin group (168) was excluded, non-significant results still remained (median, IQR: the aspirin group, 27, 18.5–40.5, *p* = 0.521, Mann–Whitney *U* test). In contrast, the CAA score of the cilostazol group (14, 11–31) was significantly lower in the control group (*p* = 0.046, Mann–Whitney *U* test: Figure 6b).

**Figure 5.** Representative images of cerebral amyloid angiopathy (CAA) in a Tg2576 mouse (aged 23 months old). Pan-Aβ immunostained sections show significant CAA in the cortex at 1.08 mm behind Bregma (a: arrowheads) and mild to moderate CAA (b: arrowheads) in the hippocampus at 1.33 mm behind Bregma. Arrows show senile plaques in the hippocampus (**b**). Vessel with a thin rim of amyloid in the vessel wall (c; severity grade, 1); vascular amyloid with amyloid infiltrating the surrounding neuropil (d; severity grade, 2); dysphoric amyloid with amyloid deposition within the vessel wall and with a thick and complete amyloid coat around the vessel wall (e; severity grade, 3). Scale bars indicate 100 μm (**a**,**b**) and 10 μm (**c**–**e**).

**Figure 6.** The cerebral amyloid angiopathy (CAA) scores in the control, aspirin, and cilostazol groups evaluated at 15 months (**a**) and 21–23 months (**b**). Light-yellow-colored bar graphs indicate median. Boxes and bars indicate median and interquartile range, respectively. Yellow circles represent the CAA score of each mouse.

#### *2.6. Senile Plaque*

In the first specimens, there were no differences in the percent area of senile plaque between the control group (median, IQR: 0.05, 0.03–0.40) and the aspirin group (0.24, 0.08–0.83) or the cilostazol group (0.73, 0.00–0.12) (*p* = 0.286 and *p* = 0.905, Mann–Whitney *U* test: Figure 7a). In the second specimens, there were also no differences in the percent area of senile plaque between the control group (median, IQR: 0.27, 0.19–0.54) and the aspirin group (0.28, 0.22–0.45) or the cilostazol group (0.50, 0.16–0.84) (*p* = 0.750 and *p* = 0.888, Mann–Whitney *U* test: Figure 7b).

**Figure 7.** The percent area of senile plaque in the control, aspirin, and cilostazol groups evaluated at 15 months (**a**) and 21–23 months (**b**). Light-green-colored bar graphs indicate median. Boxes and bars indicate median and interquartile range, respectively. Green circles represent the value of percent area of senile plaque of each mouse. The numbers next to the green circle (2.19) indicate percent area of senile plaque out of the range of the vertical axis scale (b, Cilostazol group).

#### **3. Discussion**

The main finding of this study is that the CAA burden in Tg2576 mice could be reduced by half with long-term cilostazol therapy, but not with aspirin. In particular, such a long-term treatment effect on the CAA burden of APP transgenic mice was firstly archived with a lower dose of cilostazol (conducted to intake 20 mg/kg/day; resulting intake 14.7 mg/kg/day) compared to a previous study (600 mg/kg/day) [12]. However, we found no long-term treatment effect of either cilostazol or aspirin on the severity of cerebral micro-hemorrhages or on the expansion of senile plaque.

PDE family proteins, mostly expressed in the brain, have attracted attention as a source of new targets for the treatment of psychiatric and neurodegenerative disorders [13,14]. Previous studies of animal models have shown that the phosphodiesterase III inhibitor, cilostazol, decreases cerebral amyloid-β accumulation [12,15]. Regarding the long-term effects of cilostazol on CAA, one study using Tg-SwDI mice mainly expressing vasculotropic Aβ demonstrated that relatively high-dose cilostazol (equivalent to 600 mg/kg/day intake in 25 g weight mouse) decreased the Aβ accumulation of the brain, resulting in improved cognitive performance [12]. Regarding the short-term effects of cilostazol, one study using C57BL/6J mice reported that the oral administration of cilostazol (20 mg/kg/day) for 6 weeks around the injection of Aβ into the cerebral ventricle almost completely prevented Aβ accumulation in the brain [15]. Thus, our finding has added one line of evidence that the efficacy of lower-dose cilostazol was replicated in a different experimental setting. Given the dose-dependent side effects (e.g., headache and dizziness) of cilostazol [16], long-term administration of the lower dose of cilostazol could be an optional strategy of the treatment for CAA burden.

The protective role of cilostazol against Aβ burden in the CAA model mice (i.e., Tg-SwDI mice or Tg2576 mice) appeared to be its vasculotropic effects, as long-term aspirin treatment did not reverse Aβ deposition. Our result of the reduced CAA score and non-reduced senile plaque suggests that the promotion of Aβ metabolism by vasculotropic cilostazol was achieved by its easy access to the perivascular area but not to the brain parenchyma. This is consistent with the finding that cilostazol poorly penetrates the blood–brain barrier [12]. The main mechanism promoting Aβ metabolism by cilostazol appears to be the increase in perivascular drainage of Aβ, followed by the decrease in degenerative changes in vascular walls with Aβ deposits [12]. Since the motive force for perivascular Aβdrainage appears to be generated by arterial pulsations [17,18], the direct action of cilostazol on the vascular smooth muscle cells to increase pulse duration time [19] and arterial elasticity [16] may have contributed to facilitating the perivascular drainage of Aβ.

None of the Aβ-targeted phase 3 clinical trials in Alzheimer's disease has shown statistically significant benefits on its pre-specified clinical endpoints. Several of these trials, however, were mis-designed in terms of patient selection, choice of agent, target engagement, and/or dose, or they had to be halted because of the off-target side effects [20]. A recent phase 2 clinical trial in patients with CAA has shown that immunotherapy using the anti-Aβ40 antibody (Ponezumab) also did not show the prespecified efficacy (improvement in cerebrovascular reactivity measured by functional magnetic resonance imaging (MRI)) [21]. The Aβ-targeted immunotherapy could cause amyloid-related imaging abnormalities (ARIA) representing vasogenic edema, micro-hemorrhages, or cortical, superficial siderosis on MRI. ARIA, which appeared to be a dose-dependent phenomenon, sometimes causes transient symptoms of headaches, confusion, and visual disturbances [22]. Thus, it might not be realistic to increase the dose of the anti-Aβ antibody to enhance the effectiveness. To explore treatment to minimize Aβ accumulation, it might to be crucial to consider not only the suppression of Aβ over-production, but also the promotion of Aβ clearance in CAA and/or Alzheimer's disease patients. A Japanese retrospective study reported that the combination therapy group (using donepezil plus cilostazol) was more effective for cognitive decline in patients with mild dementia compared to the donepezil-only group [23]. Thus, given that the balance between Aβ synthesis and clearance determines brain Aβ accumulation, a multidrug combination (e.g., low-dose anti-Aβ antibody, and low-dose cilostazol) therapy could provide a mainstream cure in the early stages of CAA and/or Alzheimer's disease [24].

Our result that aspirin had no influence on the severity of cerebral micro-hemorrhages was in line with a previous study [25]. A characteristic feature of cilostazol is that it has weaker hemorrhagic side effects than other antiplatelet drugs [26] and does not increase the bleeding time [27]. In fact, a previous study demonstrated that the short-term administration of cilostazol (30 mg/kg/day) reduced the intracranial hemorrhage volume along with sufficient inhibition of platelet aggregation in non-transgenic mice [11]. However, the present study did not suggest the expected safety benefits of cilostazol for reducing micro-hemorrhages over aspirin. In other words, this study offered evidence that low-dose cilostazol could improve the vascular amyloid burden without increasing cerebral micro-hemorrhages in a mouse model of CAA.

We also acknowledge a limitation of this study. We did not evaluate the effects of lower-dose cilostazol on cognitive performance in Tg2576 mice. Although we had discussed this issue at the planning state of the study, we decided to just focus on whether lower-dose cilostazol could reduce the pathological burden of the brain in Tg2576 mice, because of our funding limitations to develop a research environment to fully examine the cognitive function of the mice (i.e., to buy additional Tg2576 mice or appropriate experimental devices). However, our current results allow us to believe that further investigation would be meaningful to calcify whether lower-dose cilostazol could improve cognitive performance in Tg2576 mice.

#### **4. Materials and Methods**

#### *4.1. Standard Protocol Approval*

All animal procedures were performed according to the guidelines of the Animal Use and Care Committee of the Saga University (Saga, Japan). All protocols were approved by the Animal Use and Care Committee and the Genetic Recombination Experimental Committee (ethical approval code: 23-024-2), as well as Animal Research: Reporting of In Vivo Experiments guidelines [28]. The experimental data is available from the supplemental materials: Data S1 and Data S2).

### *4.2. Animals*

As a CAA animal model, we used female B6, SJL-Tg 2576Kha (APPSWE) transgenic mice (Tg2576 mice: Taonic Bioscience, Inc., NY, USA). The Tg2576 mice were housed in a room with a 12-h light/dark cycle (light on at 7:00 a.m.) with access to food and water ad libitum. A flow diagram of the study schedule and grouping is shown in Figure 1. Sixty Tg2576 mice aged 3 months were started on standard pelleted chow (vehicle). Eleven months after, mice aged 14 months were divided into three groups with adjustment for body weight: control group (mice fed with standard pelleted chow only), aspirin group (mice fed with the pelleted chow containing 0.01% aspirin), and cilostazol group (mice were fed with pelleted chow containing 0.01% cilostazol). The dose setting for each drug is described in the next section (please see 4.3, Drugs). The Tg2576 mice were randomly assigned to each group (in considering drug-associated death, we a priori planned to allocate more mice to the aspirin or the cilostazol group).

#### *4.3. Drugs*

In the brain of Tg2576 mice, Aβ deposits developed after 8 months, and dissemination of Aβ plaque progressed from 15 months to 23 months [29]. Therefore, we planned to evaluate the long-term effects of the drugs using mice aged 15 months or 21–23 months. For long-term administration of drugs, we selected oral drug intake using a pelleted chow containing each drug. The dose of cilostazol to be administered in mice varied because of the different bioavailability between humans and mice. A previous study reported that, in transgenic C57BL/6-Tg(Thy1-APPSwDutIowa) BWevn/J mice fed with pelleted chow containing 0.3% cilostazol (equivalent to 600 mg/kg/day intake in 25 g weight mouse), phosphodiesterase III inhibitor promoted the drainage of cerebrovascular Aβ [12], while a previous study demonstrated that in Aβ-injected wild-type mice, cilostazol administration of 10–20 mg/kg/day for 2 weeks exerted a protective effect against Aβ-induced cognitive deficits along with decreased Aβ accumulation [15]. To evaluate the effect of a lower dose of cilostazol on the CAA mice model, we used pelleted chow containing 0.01% cilostazol (equivalent to 20 mg/kg/day intake in 25 g weight mouse, donated by Otsuka Pharmaceutical, Tokyo, Japan) for treatment of the cilostazol group. Regarding the dose of aspirin, a previous study reported that, in spontaneous hypertensive rats, aspirin-attenuated collagen-induced platelet aggregation at 10–100 mg/kg in rats [30]. Clinically, daily intake of low-dose aspirin in humans is almost similar to cilostazol. Therefore, similar to the drug concentration of cilostazol, we used the pelleted chow containing 0.01% aspirin (equivalent to 20 mg/kg/day intake in 25 g weight mouse: donated by Otsuka Pharmaceutical, Tokyo, Japan) for treatment of the aspirin group.

#### *4.4. Measurements of Estimated Individual Food Consumption and Drug Intake by the Groups*

Food consumption per cage was measured once a week. Daily estimated individual food consumption (g/day) was calculated by the following formula: food consumption during a week per cage (g) / 7 (days) / the number of surviving mice in the cage. Daily estimated individual drug intake per weight of a mouse (mg/kg/day), which was also estimated per cage per week, was calculated by the following formula: the daily individual food consumption of the cage (g/day × 1000) × drug concentration/mean weight of a surviving mouse in the cage (kg).

#### *4.5. Histology and Immunohistochemistry*

To evaluate the serial effects of aspirin or cilostazol on Aβ deposition, pathological examinations were performed 1 month after the grouping (the first specimen, 15 months old, (*n* = 14)) and at 7–9 months after the grouping (the second specimen, 21–23 months old, (*n* = 30)). Mice were overdosed with sodium pentobarbital (50 mg/kg, intraperitoneal) and perfused transcardially with phosphate-buffered saline (PBS), followed by 4% paraformaldehyde in 0.1 M PBS. The brains were immediately removed, immersion fixed for 1 d in 4% paraformaldehyde, followed by 2 d in 10% sucrose in 0.01 M PBS, and 2 d in 30% sucrose in 0.01 M PBS. Post-fixed brains were cryoprotected, frozen, and sectioned at 25 μm with a freezing–sliding microtome [31]. Hematoxylin and eosin (H&E) and Congo red staining were done according to standard protocols [32]. Perls's Berlin Blue method was used to visualize ferric iron in hemosiderin (with Nuclear Fast Red (Kernechtrot stain solution: Lot number, 130312: Muto Pure Chemicals Co., LTD., Tokyo, Japan)) [32,33]. According to previously published protocols [31,34], the pan-Aβ stain was performed with the following primary and secondary antibodies: the primary antibody, rabbit polyclonal antibody to Aβ (1–40) (catalog number, 44–136: Invitrogen by Thermo Fisher Scientific Inc., CA, USA); the secondary antibody, biotinylated goat anti-rabbit IgG antibody (catalog number, BA-1000: Vector Laboratories, CA, USA). For confocal microscopy, double-labeling for Aβ and smooth muscle cells was achieved simultaneously using frozen sections as follows: for Aβ, the primary antibody with rabbit polyclonal antibody to Aβ (1–40) (catalog number, 44–136: Invitrogen by Thermo Fisher Scientific Inc., CA, USA); the secondary antibody with Donkey anti-Rabbit IgG antibody Cy3 conjugate (1:500; product number, AP182C: Millipore Co., CA, USA); for smooth muscle cells, monoclonal anti-alpha-smooth muscle actin conjugate FITC (product number, F3777: Sigma-Aldrich Co. LLC, MO, USA). Sections were mounted with Fluoromount (catalog number: K 024: Diagnostic BioSystems, Hague, Netherlands) and analyzed with a Confocal Laser Scanning Microscope LSM880+Airyscan Fast (Zeiss, Oberkochen, Germany).

#### *4.6. Pathological Evaluations*
