Characterizing vascular function in mouse models of Alzheimer's disease, atherosclerosis, and mixed Alzheimer's and atherosclerosis

Abstract

Significance: Alzheimer's disease does not occur in isolation, and there are many comorbidities associated with the disease, especially diseases of the vasculature. Atherosclerosis is a known risk factor for the subsequent development of Alzheimer's disease; therefore, understanding how both diseases interact will provide a greater understanding of co-morbid disease progression and aid the development of potential new treatments.

Aim: We characterize hemodynamic responses and cognitive performance in APP/PS1 Alzheimer's mice, atherosclerosis mice, and a mixed disease group (APP/PS1 and atherosclerosis) between the ages of 9 and 12 months.

Approach: Whisker-evoked hemodynamic responses and recognition memory were assessed in awake mice, immunohistochemistry to assess amyloid pathology, and histology to characterize atherosclerotic plaque load.

Results: We observed hemodynamic deficits in atherosclerosis mice (versus Alzheimer's, mixed disease, or wild-type mice), with reduced short-duration stimulus-evoked hemodynamic responses occurring when there was no concurrent locomotion during the stimulation period. Mixed Alzheimer's and atherosclerosis models did not show differences in amyloid beta coverage in the cortex or hippocampus or atherosclerotic plaque burden in the aortic arch vs relevant Alzheimer's or atherosclerosis controls. Consistent with the subtle vascular deficits and no pathology differences, we also observed no difference in performance on the object recognition task across groups.

Conclusions: These results emphasize the importance of experimental design for characterizing vascular function across disease groups, as locomotion and stimulus duration impacted the ability to detect differences between groups. Although atherosclerosis did reduce hemodynamic responses, these were recovered in the presence of co-occurring Alzheimer's disease, which may provide targets for future studies to explore the potentially contrasting vasodilatory mechanisms these diseases impact.

Keywords: Alzheimer’s; atherosclerosis; hemodynamic; mixed disease; optical imaging spectroscopy; vasculature.

Fig. 1

Experimental timeline and representative 2D-OIS spatial images. (a) Experimental timeline for all mice included in these experiments. Subjects first underwent cognitive testing on the novel object recognition (NOR) task at 9 months, before a surgery to insert a headplate and thinned window over the whisker barrel cortex was conducted. After a recovery period (minimum 1 week), habituation to the imaging apparatus was completed (5 days) and regional hemodynamic responses were recorded in awake mice using two-dimensional optical imaging spectroscopy (2DOIS). Mice were sacrificed using transcardial perfusion, and immunohistochemical (IHC) staining was conducted to assess amyloid coverage in AD and MIX groups. Histology was also conducted on the ATH and MIX groups to assess atherosclerotic plaque burden in the aortic arch. Representative animal from each disease group (b) WT, (c) AD, (d) ATH, and (e) MIX revealing the thinned window region (row one), vessel maps revealing arteries and veins (row two), a spatial map of HbT changes to a 2-s whisker stimulation (row three) with red colors indicating an increase in HbT in response to a 2-s whisker stimulation, and the region of interest overlying the active whisker barrel region (row four, white ROI).

Fig. 2

Comparing stimulation-induced responses without the confounds of locomotion. (a) Hemodynamic time series from the artery ROI within the whisker barrel region showing total (HbT, green), oxygenated (HbO, red), and deoxygenated (HbR, blue) hemoglobin in response to a 2-s mechanical whisker stimulation in trials with no concurrent locomotion occurring between the 4 s either side of the stimulus period (dotted lines) for wild-type (WT, purple), APP/PS1 (AD, green), atherosclerosis (ATH, pink), and mixed APP/PS1 × atherosclerosis (MIX, orange) mice. (b) The maximum peak of the HbT response during the stimulation period of these “rest trials” is compared across disease groups using a linear mixed model, with a significant overall effect of disease ($p=0.008$) driven by the atherosclerosis group (pink) showing smaller 2-s stimulus-induced responses than the wildtype (purple, $p=0.02$) or mixed (orange, $p=0.03$) mice. (c) There was a significant difference between disease groups on the area under the curve of the HbR response during the stimulation period of these “rest trials” ($p=0.03$), driven by the atherosclerosis group showing smaller HbR responses than the wild-types ($p=0.04$) and AD mice ($p=0.05$). (d) There were no differences in locomotion during these “rest trials,” assessed using the area under the curve of the locomotion response during the stimulation period (right). (e) Hemodynamic responses were also compared in response to a 16-s mechanical whisker stimulation in trials with no concurrent locomotion occurring between the 4 s on either side of the stimulation period (dotted lines) for the artery ROI. There was no significant difference in the size of the (f) HbT (maximum peak, $p=0.61$) or (g) HbR (AUC, $p=0.06$) responses between WT, AD, ATH, or MIX mice, although the HbR washout was smaller at trend level due to the mixed APP/PS1 × atherosclerosis mice showing smaller responses than wild-type. (h) There were also no differences in locomotion during rest trials for 16-s stimulation events. $P$-values are taken from linear mixed-effects models with disease group inputted as a fixed-effect factor, and animal ID as the random effect (lmer package RStudio), and pairwise comparisons (with correction for multiple comparisons) conducted using the Tukey method (emmeans package RStudio). Shaded error bars represent mean ± SEM. Horizontal lines on violin plots show median and interquartile ranges. The number of trials and animals included in each group are indicated on the time series graphs.

Fig. 3

Differences across disease models in the size of 2-s stimulation responses during rest are not linked to the high number of trials. (a) There was no correlation between the trial number (1 to 30) and the size of the HbT response (max peak) for any of the disease groups (WT ($p=0.02$), purple; AD ($p=0.43$), green; ATH ($p=0.87$), pink; MIX ($p=0.12$), orange), although the WT mice showed a positive correlation whereby the size of the HbT peak increased with increasing trial number (purple, $p=0.02$). There was no significant impact of disease on the correlation between these two measures (compare slopes, $p=0.15$). However, there was an overall significant difference in the intercept (size of HbT peak) between disease groups ($p<0.0001$), with atherosclerosis mice consistently showing lower HbT peak values across all trials. (b) and (c) When HbT responses were separated as belonging to early (trials 1 to 5) or late (trials 25 to 30) trials and compared across disease and trial groups, there were no significant differences found. (d) There was also no correlation between the trial number (1 to 30) and size of the HbR response (AUC) for most of the disease groups [WT ($p=0.12$), purple; AD ($p=0.78$), green; ATH ($p=0.41$), pink], although the MIX mice did show larger HbR responses in later trials ($p=0.009$, orange). Overall, there was no significant impact of disease on the correlation between these two measures (compare slopes, $p=0.34$); however, there was an overall significant difference in the intercept (size of HbR AUC) between disease groups ($p<0.0001$), with WT mice consistently showing larger HbR AUC values across all trials. (e) and (f) When HbR responses were separated as belonging to early (trials 1 to 5) or late (trials 25 to 30) trials and compared across disease and trial groups, a significant effect of the trial was observed as HbR responses were larger in later trials. On scatterplots, individual values (dots) represent single trials, and $p$-values the result of a simple linear regression to test the correlation between variables within each disease group (legend) or the difference between groups on the slope and intercept (title). Shaded error bars represent mean ± SEM. Horizontal lines on violin plots show median and interquartile ranges. $P$-values on violin plots are taken from linear mixed-effects models with disease and trial group inputted as fixed-effect factors, animal ID as the random effect (lmer package RStudio), and pairwise comparisons (with correction for multiple comparisons) conducted using the Tukey method (emmeans package RStudio).

Fig. 4

Comparing whisker stimulus-induced responses in trials that contain concurrent locomotion. (a) Hemodynamic time series from the artery region within the whisker barrels showing total (HbT, green), oxygenated (HbO, red), and deoxygenated (HbR, blue) hemoglobin in response to a 2-s mechanical whisker stimulation in trials with concurrent locomotion occurring at the onset of the stimulus period (dotted lines) for wild-type (WT, purple), APP/PS1 (AD, green), atherosclerosis (ATH, pink), and mixed APP/PS1 × atherosclerosis (MIX, orange) mice. There was no difference between disease groups in the size of the (b) maximum peak of the HbT response ($p=0.72$) or the (c) area under the curve of the HbR response ($p=0.46$) during the stimulation period of these “locomotion trials.” (d) There were no differences in locomotion across disease groups during these “locomotion trials,” assessed using the area under the curve of the locomotion response during the stimulation period (right, $p=0.31$). (e) Arterial hemodynamic responses were also compared for a 16-s mechanical whisker stimulation in trials with concurrent locomotion occurring at the onset of the stimulation period (dotted lines). There were no significant difference in the size of the (f) HbT (maximum peak, $p=0.92$) or (g) HbR (AUC, $p=0.78$) responses between WT, AD, ATH, or MIX mice. (h) There were also no differences in locomotion during “locomotion trials” for 16-s stimulation events ($p=0.43$). $P$-values are taken from linear mixed-effects models with disease group inputted as a fixed-effect factor, animal ID as the random effect (lmer package RStudio), and pairwise comparisons (with correction for multiple comparisons) conducted using the Tukey method (emmeans package RStudio). Shaded error bars represent mean ± SEM. Horizontal lines on violin plots show median and interquartile ranges. The number of trials and animals included in each group are indicated on the time series graphs.

Fig. 5

No differences in performance on a novel object recognition task between disease groups. (a) Mice were placed in a $40\mathrm{cm}\times 40\mathrm{cm}$ arena and allowed to freely explore for 10 min, which during the training phase (left) contained two identical objects and during the testing phase (right) included a novel object to replace either the left or right familiar object from the previous training session (Figure created in BioRender Ref. 76). Video recordings were taken of mice in the arena, and they had to explore each object for a minimum of 20 s during the training phase to be included in subsequent analysis. We compared the (b) distance run (cm) and (c) velocity (cm/s) of each mouse in the training and testing arena to indicate whether mice were engaging with the task similarly between disease groups. We showed no significant difference in the amount of time or speed of travel between groups. (d) To indicate the mouse’s preference for the novel versus familiar object in the testing arena, the preference index is calculated as [cumulative duration with novel object/total exploration time (cumulative duration with familiar + novel objects)] × 100. Values over 50% indicate the mouse has distinguished between familiar and novel objects and spent more time exploring the novel item. The wild-type, atherosclerosis, and mixed groups all had values $>50\%$ indicating a preference for the novel object, and a one-way ANOVA revealed no overall difference between groups on this task. Where groups showed a normal distribution, $p$-values are taken from one-way ANOVAs with group (WT, AD, ATH, MIX) as the independent variable and task performance metric (distance, velocity, preference) as the dependent variable. As the dataset for the distance run training phase was not normally distributed (assessed using a Shapiro-Wilks test), a Kruskal-Wallis comparison was run in place of the one-way ANOVA. Shaded error bars represent mean ± SEM. Horizontal lines on violin plots show median and interquartile ranges. Individual dots on bar charts represent single mice, the wild-type group included six mice, AD group six mice, atherosclerosis group 10 mice, and mixed group six mice across all figure panels (c)–(e).

Fig. 6

No differences in pathology across disease models. (a) Representative images of the aorta from an atherosclerotic mouse (left) and a mixed disease mouse (right) from which the percent plaque burden was calculated as the density of plaques found within the area of the aortic arch (top of the image where the aorta branches). The scale bar represents 2 mm. (b) The atherosclerosis ($N=8$) and mixed disease ($N=4$) mice showed no differences in plaque burden at the aortic arch ($p=0.85$). (c) Representative images of brain slices, which include CA1 hippocampus after labeling for amyloid beta in an Alzheimer’s mouse (left) and mixed disease mouse (right). The scale bar represents $800\mu \mathrm{m}$. (d) Regions of interest were taken across the entirety of CA1 for the hippocampus, and for a rectangle overlying CA1 for cortex across all slices and the percentage of amyloid beta was calculated across the total area. There were no significant differences in amyloid coverage across our disease groups (AD $N=3$, MIX $N=6$, $p=0.11$) or brain regions ($p=0.99$).