Sex-Specific Effects of High Fat Diet on Cognitive Impairment in a Mouse Model of VCID

Abstract

Mid-life metabolic disease (i.e. obesity, diabetes, prediabetes) causes vascular dysfunction and is a risk factor for vascular contributions to cognitive impairment and dementia (VCID), particularly in women. Using middle-aged mice, we modeled metabolic disease (obesity/prediabetes) via chronic high fat (HF) diet and modeled VCID via unilateral common carotid artery occlusion. VCID impaired spatial memory in both sexes, but episodic-like memory in females only. HF diet caused greater weight gain and glucose intolerance in middle-aged females than males. HF diet alone impaired episodic-like memory in both sexes, but spatial memory in females only. Finally, the combination of HF diet and VCID elicited cognitive impairments in all tests, in both sexes. Sex-specific correlations were found between metabolic outcomes and memory. Notably, both visceral fat and the pro-inflammatory cytokine tumor necrosis factor alpha correlated with spatial memory deficits in middle-aged females, but not males. Overall, our data show that HF diet causes greater metabolic impairment and a wider array of cognitive deficits in middle-aged females than males. The combination of HF diet with VCID elicits deficits across multiple cognitive domains in both sexes. Our data is in line with clinical data, which shows that mid-life metabolic disease increases VCID risk, particularly in females.

INTRODUCTION

Half of all dementia patients have evidence of vascular contributions to cognitive impairment and dementia (VCID), either as a single pathology or as a multi-etiology dementia (1), yet there remains a severe lack of knowledge regarding VCID. Type 2 diabetes is a major risk factor for VCID, as it is known to cause damage to the cerebral vasculature and blood brain barrier (BBB) (2-4). Furthermore, there is a sex difference in risk for VCID that varies by metabolic status. While most studies agree that in a nondiabetic population, VCID risk is slightly higher in men than in women (under the age of 90) (5), in a pooled analysis of 2.3 million individuals, it was found that diabetic women have a 19% increased risk of VCID compared to diabetic men (6). However, less is known regarding the effects of prediabetes, which is estimated to affect 34% of the US population (7), yet is widely undiagnosed. A recent meta-analysis showed that while prediabetes increased risk for both all-cause dementia and Alzheimer’s disease (AD), the risk was greatest for vascular dementia (RR 1.47) (8). Another major risk factor for dementia is mid-life obesity, which is associated with a three-fold increased risk of AD and a five-fold increased risk of vascular dementia, independent of cardiovascular disease, stroke and diabetic status (9). Central obesity increases in post-menopausal women(10). Central obesity has been associated with metabolic impairment and an increased severity of vascular risk factors (11) and may worsen VCID pathology (5). Given that the already significant prevalence of prediabetes and obesity are expected to rise, and the known sex difference in the risk of developing VCID within the diabetic population, we sought to identify sex differences in the effects of obesity with prediabetes in a mouse model of VCID.

Previously we have shown that chronic administration of a high fat (HF) diet causes cognitive impairment in single-sex studies both in the context of normal aging (12) and in a mouse model of VCID (chronic cerebral hypoperfusion induced by unilateral carotid artery occlusion) using young male mice (13). We have also recently reported that sex differences in the effects of a HF diet vary by age of diet onset: in young mice, HF-fed females are metabolically protected compared to HF-fed males, but in middle-aged mice, HF-fed females are more metabolically impaired compared to HF-fed males (14). Still, sex differences in cognitive effects have not been determined in middle-aged mice. Therefore, we tested the hypothesis that in middle aged mice, HF diet would cause greater metabolic impairment and a wider array of cognitive deficits in females relative to males in a mouse model of VCID.

MATERIALS AND METHODS

Animals and Experimental Design

All experiments were approved by the Albany Medical College Animal Care and Use Committee and in compliance with the ARRIVE guidelines. Male and female C57BL/6J mice were purchased at ~35 weeks of age (+/− 3 weeks) from Jackson Laboratories (Bar Harbor, ME, USA). Upon arrival, mice were randomly assigned to cages [group housed (3-5 per cage) at 70-72 °F, 30-70% humidity, with a 12 h light/dark cycle (7 a.m. on/7 p.m. off)]. Following one week of acclimation, during which time they were provided with Purina Lab Diet 5P76, cages of mice were randomized to treatment groups based on cage ID number and placed on either a HF diet (60% fat, energy density=5.21 Kcal/g; D12492, Research Diets, New Brunswick, NJ, USA) or a low fat (LF) control diet (10% fat, energy density=3.82 Kcal/g; D12450B, Research Diets, USA). Mice remained on their respective diets for the duration of the study. After 3 months on the diet, mice underwent a glucose tolerance test (GTT), were allowed to recover for a week, then received a right unilateral common carotid artery occlusion (UCCAO) surgery or sham surgery. After 6 months on the diet, mice underwent a second GTT, then were allowed to recover for at least 10 days before behavior testing. Vaginal cytology took place at least 4 days after the last behavior test. After the 2 weeks of vaginal cytology was completed, the following week mice were anesthetized with isoflurane and cerebral blood flow assessed via laser speckle contrast imaging (FLP1, Moor instruments, Wilmington, DE, USA). Immediately following blood flow imaging, mice were deeply anesthetized with pentobarbital and intracardially perfused with saline and tissues were collected. Mice were group housed for the duration of the study except during the nest building test. Experiments were conducted in 4 cohorts of mice (160 mice, N=20 mice per group total; N=5 per group/cohort). Out of the 160 total mice in the study, 6 died during surgery, an additional 12 died prematurely or had to be euthanized due to sickness before the end of the study (the most common cause being severe dermatitis), and an additional 2 mice were excluded due to tumors found at the time of euthanasia. This resulted in a final N of 15-20 per group for metabolic data. Due to a computer malfunction during a probe trial test day, data was lost for 7 mice for the Morris water maze (5 LF sham males, 2 HF sham males). One mouse was also excluded for failure to swim. Our NORT had two exclusion criteria: they must have explored objects for a minimum of 2 seconds and must have explored both objects (0-2 mice per group, were excluded based on these criteria). Only a subset of mice received laser speckle imaging, nest building tests, and visceral fat inflammatory gene analysis. Final N numbers for each figure are noted in figure legends. Due to differences in food color and mouse appearance, diet could not be blinded during GTT, behavior testing or blood flow imaging; however, surgery status was always blinded during in vivo procedures. Surgery, diet, and sex were always blinded during data analysis.

Glucose Tolerance Test (GTT)

At 3 and 6 months after the onset of the dietary intervention, mice were fasted for 16 hours overnight, and the next morning baseline blood glucose levels in saphenous vein blood were measured by glucometer (Verio IQ, OneTouch, Sunnyvale CA, USA). Each mouse received 2g/kg of glucose via i.p. injection and blood glucose levels were re-measured at 15, 30, 60, 90 and 120 minutes post-injection.

VCID Surgery (unilateral common carotid artery occlusion)

Right unilateral common carotid artery occlusion or sham surgery was performed after the mice had been on the diet for 3 months, as previously described (13, 15). Briefly, mice were randomly assigned to a surgical group, anesthetized with isoflurane, and body temperature maintained. After midline cervical incision, the right common carotid artery was exposed. For VCID mice, the carotid was isolated and two 6-0 silk sutures were placed under the carotid and double tied. The carotid was cauterized between the two ties. For sham mice, the membranes covering the salivary glands were separated and the carotid was exposed, but no ties were placed and the carotid was left intact. For all mice, incisions were closed with tissue adhesive, and 100μl 0.03mg/mL buprenorphine was administered via i.p. injection (2x day for 3 days). Mice were allowed to recover from anesthesia before returning to their home cage. The 1-month survival rate following surgery was 96%.

Behavior testing

Three months after surgery, mice underwent behavior testing in the following order: Anxiety-like behavior and general locomotor activity were tested with the open field (day 1), novel object recognition (NORT) (day 2), spatial learning and memory in the Morris water maze (MWM) (day 8-10 or day 10-12), and activities of daily living assessed through nest building (day 15-16). Mice were allowed to acclimate to the behavior room for one hr prior to testing in the same ambient conditions used during the test. Lights were kept dim and white noise was used to mitigate external noise during the Morris water maze. Open field, NORT, and MWM were video recorded and analyzed using automated tracking software (ANY-maze 5.1, Stoelting, Wood Dale, IL).

Open field:

Mice were placed in a square arena and allowed to explore freely for 10 min, then removed and placed in a “recovery cage” so as not to expose them to naïve cagemates. Anxiety-like behavior was measured with the amount of time spent in the center of the arena versus the edges. General locomotor behavior was assessed with the measure of total distance traveled (tracklength).

Novel Object Recognition Testing (NORT):

NORT took place as two five-minute trials—a “train” trial and a “test” trial, with an inter-trial interval of 1hr. In the train trial, mice were placed in the open field arena and allowed to explore two identical objects, one in the top right corner and one in the top left corner, each 5 cm from the wall. After the train trial, mice were placed in a recovery cage, and the open field box and objects were cleaned with 70% ethanol. In the test trial, one of the objects was returned to its original position in the arena, along with a novel object in the place of the second object. The positions of the familiar vs novel object were counterbalanced. Mice were allowed to explore again for 5 minutes. Recognition index [(time with novel object/total time with objects)*100] was computed. Exclusion criteria included failure to explore both objects and object exploration time of <2 seconds.” “For NORT, one sample t-tests were used [(recognition index vs. chance (50%)].”

Morris Water Maze (MWM):

Mice were placed in a pool 125 cm in diameter, divided conceptually into four quadrants. During trials, mice were trained to swim to a platform in order to be rescued by the researcher. The water was made opaque with white tempera paint. The temperature of the pool was maintained at 20 to 22 degrees Celsius. MWM took place in three different kinds of trials: visible, hidden, and probe. For visible and hidden trials, we used a previously validated abbreviated protocol that was created specifically for aged mice (16). During the visible trials (day 1), mice were trained to locate the escape platform (plexiglass circle), submerged below the surface of the water with the aid of visual cues (black duct tape wrapped around the edge of the platform and a pencil wrapped in duct tape sticking up out of the water). Mice underwent 5 visible trials, each separated by a 30 minute inter-trial interval. Mice were carefully placed near the wall of the pool in two alternating locations. The trial ended when the mouse located the platform. Mice were rescued, thoroughly dried and placed in a recovery cage under a heat lamp with access to food. If the mouse failed to locate the platform after 3 minutes, they were guided to the platform by placing a finger on the platform. If they still didn’t locate the platform, they were brought to the platform by being gently dragged by the tail. This was repeated until they stayed on the platform for at least 10 seconds. Throughout the visible trials, the platform was not moved. Subsequent to visible trials, mice were trained to locate a hidden platform, using extra-maze cues for spatial reference and orientation (hidden trials; day 2). Small cues made of black and white tape were placed on the walls of the pool above the water line at the borders of the quadrants. The platform was returned to the same place it had been during visible trials and remained in the same location for all hidden trials. There were five hidden trials with a 30 minute inter-trial-interval. Twenty-four hours after the final hidden trial, spatial memory retention was assessed through a three-minute probe trial (in which the platform was removed; day 3). During the probe trial, mice were placed in the water in the quadrant opposite the target quadrant. For the visible and hidden trials, latency to find the platform over the course of the trials was measured. For the probe trial, time spent in the target quadrant compared to the time spent in the three non-target quadrants was measured.

Nest Building:

Mice were individually housed for twenty-four hours and were provided with pre-weighed, intact nestlet material. After 24 hours, mice were carefully removed from the cages and group-housed in their home cage. The nests were scored by three independent scorers, blinded to condition. Nests were scored on a scale of 1-5 in 0.5 point increments adapted from a well-described protocol (17), with 1 being the lowest possible score and 5 being the highest. Scores were based on how thoroughly the mice tore up the nestlet material, and used it to form a crater-like, compact nest, confined to one corner of their cage. Scores were averaged between the three scorers.

Vaginal Cytology

Vaginal cytology (see Supplemental Figure 1 for representative images) took place at least 4 days after the last behavior test, between 11:00 AM and 1:00 PM every weekday for 2 weeks. Mice were group housed for the duration of the vaginal cytology measurements. Q-Tips soaked in PBS were used to gently swab the vaginal opening and rolled out onto a glass slide. Slides were set out to dry, and then 2 drops of crystal violet stain were applied. Slides were immediately coverslipped and put into a drawer for at least 2 minutes to protect the stain from light. Slides were examined under a microscope within minutes of staining, and estrus cycle phase was determined by analyzing the dominant composition of cells within the swab(18). Proestrus was identified by a predominance of nucleated, with some cornified epithelial cells. Estrus was identified by a predominance of cornified epithelial cells. Metestrus was determined by cornified epithelial cells as well as polymorphonuclear leukocytes. Lastly, diestrus was determined by primarily polymorphonuclear leukocytes. Mice were determined to be peri-estropausal/approaching reproductive senescence with irregular cycles (data not shown) and plasma estradiol levels were below the limit of detection (measured via Calbiotech Mouse/Rat Estradiol ELISA, catalog # ES180S-100ELISA, 3pg/ml sensitivity; data not shown).

Laser Speckle Contrast Imaging

Images of the cortical surface vasculature were obtained using a laser speckle contrast imager. Isoflurane anesthesia was administered to animals prior to imaging and exposure of the skull was available through ~2cm vertical incision of the skin and connective tissue. Mineral oil was applied to minimize skull dryness. Images were acquired for a period of 5 minutes at a rate of 1 min/frame using a full-field laser speckle imager (FLP1, Moor Instruments, Wilmington DE, USA). Images were analyzed using FLP1 Review V4.0 software (Moor Instruments). Regions of interest in the left and right frontal, parietal, temporal and Zone of Pial Anastomoses areas were drawn on each image(19). Mean flux values for each ROI were obtained. For acute changes in blood flow comparing pre- vs. immediate post-surgery, the percent difference between pre- and post-surgery regions of interest was calculated and compared to an expected difference of 0% using a 1-sample T test. In a separate cohort of mice, for long-term changes in blood flow which lacked a baseline measurement, the percent difference was calculated between each pair of ROIs (e.g. right and left temporal regions) within each image.

Quantitative RT-PCR for Adipose Inflammatory Markers

Visceral fat was extracted, flash frozen and stored at −80C. Total RNA was extracted using the Trizol method from an aliquot of about 100mg of visceral fat (Thermo Fisher scientific, cat# 15596018). RNA concentration was determined using a nanodrop and 1ug of this was converted to cDNA using High-Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Thermo Fisher scientific, cat# 4374967) according to the manufacturer’s instructions. Quantitative PCR was performed in triplicates on 50 ng of cDNA in a 10ul reaction using TaqMan probe technology on a Bio-Rad CFX-384 real time system. Taqman assays (Thermo Fisher scientific) used were: Tnfα: Mm00443258_m1; Rps17: Mm01314921_g1 and Rpl13a: Mm05910660_g1. Bio-Rad CFX Maestro 1.1 software was used to analyze the data. The relative expression levels of genes of interest was calculated using the ΔΔCq method relative to the geometric mean of two housekeeping genes: RPS17 and RPL13a (average gene normalizer M value 0.425 for both). Statistical analysis was performed on 3-5 animals/group using a 3way ANOVA followed by Sidak multiple comparison test.

Statistics

Data were analyzed with Prism 8.1 (GraphPad Software, San Diego, CA, USA). A 3-way ANOVA [surgery (sham vs. VCID) X sex (male vs. female) X diet (LF vs. HF)] with Sidak correction for multiple comparisons was used when males and females were compared. For behavior data in which performance is judged relative to chance (e.g recognition index), data are analyzed with a one-sample T-test vs. chance. When analysis was separated by sex (e.g. reproductive organ weight), 2-way ANOVAs [surgery (sham vs. VCID) X diet (LF vs. HF)] with Sidak correction for multiple comparisons were used. For data that was presented separated by sex and tracked over time (e.g. visible and hidden trials of MWM), 3-way repeated measures ANOVAs [surgery (sham vs. VCID) X diet (LF vs. HF) X trial (1-5)] were used. The robust regression and outlier removal test (Prism 8.1) was used. Data is presented as mean +/− SEM, with the exception of categorical data (nest building is presented as median +/− interquartile range).

RESULTS

Female mice on a high fat diet experience greater metabolic impairment compared to males

Chronic HF diet (60% fat) was used to model obesity with prediabetes in middle-aged mice (see Figure 1 for timeline). As we previously reported (14), in contrast to young mice, middle-aged females exhibited greater metabolic impairment compared to middle-aged males after 3 months on a HF diet (Supplemental Fig 2). After 6 months on the HF diet (3 months post-VCID/sham surgery), this sex difference persisted. Both males and females continued to gain weight in response to a HF diet (main effect of diet, p<0.0001; Figure 2A, 2B). Females experienced greater weight gain with the HF diet relative to males (sex X diet interaction; p<0.0001; Figure 2C). All HF groups showed significantly impaired glucose clearance relative to LF controls (main effect of diet, p<0.0001; Figure 2D-F). Further, females on a HF diet showed greater glucose intolerance than males on a HF diet (sex X diet interaction, p<0.0001; Figure 2F). HF mice and female mice had significantly greater percentages of subcutaneous and visceral fat than LF mice or male mice (main effects of both diet and sex, p<0.0001), and HF females also had a greater percentage of visceral fat than HF males (sex X diet interaction, p<0.0001; Figure 2G,2H). For all measures assessed, VCID surgery did not affect metabolic outcomes. Analysis of mRNA for inflammatory cytokines in the visceral fat showed that TNFalpha was higher in males than females (main effect of sex, p<0.01), and that HF diet increased TNFalpha levels in both sexes (main effect of diet, p<0.0001; Figure 2I). Taken together, these results show that HF diet caused greater metabolic impairment in females compared to males.

Figure 1. Timeline.

Male and female mice were placed on a high fat (HF; blue pellets) or low fat (LF; yellow pellets) diet at ~8.5 months of age. Glucose tolerance tests (GTT) were performed 2.75 and 5.75 months later. VCID (right unilateral common carotid artery occlusion) or sham surgery was performed 3 months after diet onset. Behavior testing (open field, novel object recognition test, Morris water maze, and nest building) were performed 3 months after surgery over a 2 week period. Mice were euthanized and tissue collected at ~15 months of age.

Figure 2. Female mice on a high fat diet experience greater metabolic impairment compared to males.

Male and female mice were placed on a HF or LF diet at ~8.5 months of age. Over the course of 6 months both sexes gained weight on a HF diet (A-B), while females gained a greater percent of weight (C) than middle-aged HF males, regardless of surgery. Mice of both sexes showed impaired glucose tolerance. HF females had significantly more impairment in glucose tolerance than HF males after 6 months on the diet (D-F) (N=15-20/group). Subcutaneous fat (G) and visceral fat (H) were increased by HF diet in both sexes, but to a greater extent in females (N=10-15/group). Visceral fat TNFalpha levels (I) were increased by HF diet in both sexes (N= 4-5/group). *p<0.05 for diet effect within same sex, **p<0.01 for diet within the same sex, ***p<0.001 for diet within same sex, ****p<0.0001 for diet within same sex, #p<0.05 for sex difference within same diet group. Data are presented as mean +/− SEM.

Unilateral carotid artery occlusion surgery causes both acute and chronic decreases in cerebral perfusion, with HF diet causing greater deficits in females only.

We have previously reported, in young male mice, that unilateral carotid artery occlusion VCID surgery results in chronic cerebral hypoperfusion, as assessed by MRI (13, 15). In the current study, we assessed both acute and chronic hypoperfusion of the cortical surface vasculature in response to this same VCID surgery using laser speckle contrast imaging. Cerebral blood flow (CBF) was measured immediately before and 5 minutes after (acute) VCID or sham surgery in a subset of 5 mice per group. Regions of interest (ROI) included frontal, parietal, and temporal cortices, and the zone of pial anastomoses (ZOA) of each hemisphere (19) (Figure 3A). CBF was significantly decreased in the right (ischemic) hemisphere (in at least one cortical region) in all VCID mice (p<0.05), with HF VCID mice of both sexes showing significant decreases in all four ROIs (p<0.05) (Figure 3B-E). This decrease in perfusion was larger in HF VCID females vs. LF VCID females (p<0.05) in all regions except the frontal cortical ROI, which showed a trend (p=0.07) toward decreased perfusion in HF vs. LF VCID females. CBF was also assessed 4 months post-surgery in a separate subset of mice (Figure 4A-E). VCID mice of both sexes showed significant decreases in CBF in the right (ischemic) temporal ROIs compared to the left, regardless of diet (p<0.05 for all VCID mice). Thus, our data show that the acute reduction in CBF post-VCID surgery is larger in HF vs. LF VCID females in several cortical regions. Additionally, VCID mice of both sexes CBF deficits in the ischemic hemisphere are maintained in the temporal region for at least 4 months.

Figure 3. Reductions in CBF immediately post-surgery are more severe in HF compared to LF VCID females.

Cerebral blood flow (CBF) was measured via laser speckle contrast imaging immediately before and 5 minutes after unilateral carotid artery occlusion surgery. Regions of interest (ROI) were drawn for the temporal (T), parietal (P), ZOA (Z), and frontal cortical regions (F), representative images are shown (A). In the right (ischemic side) temporal region, CBF was significantly decreased in HF VCID males, LF VCID females, and HF VCID females. The decrease was greater in HF VCID vs. LF VCID females (B). In the right parietal region, CBF was significantly decreased in LF VCID and HF VCID males, and HF VCID females. The decrease was greater in HF VCID vs. LF VCID females (C). In the right ZOA region, CBF was significantly decreased in HF VCID males and HF VCID females. The decrease was greater in HF VCID vs. LF VCID females (D). In the right frontal region, CBF was significantly decreased in HF VCID males and HF VCID females. The decrease showed a trend (p=0.07) towards being grater in the HF VCID vs. LF VCID females (E). *p<0.05 for diet effect within same sex, +p<0.05 decrease in CBF. ++p<0.01 decrease in CBF, +++p<0.001 decrease in CBF. Data are presented as mean +/− SEM (N= 3-5/group).

Figure 4. Reductions in CBF 4 months post-surgery persist mainly in the temporal cortical region.

CBF was measured via laser speckle contrast imaging 4 months after unilateral carotid artery occlusion surgery. Representative images are shown of ROIs in a sham and a VCID mouse (A). In right temporal region, relative CBF was significantly decreased in all VCID mice (B). In the right parietal region, there was no relative decrease in CBF in any group (B). In the right ZOA region, CBF was significantly decreased in LF VCID mice of both sexes (D). In the right frontal region, there was no decrease in CBF in any group (E). +p<0.05 decrease in CBF. ++p<0.01 decrease in CBF. Data are presented as mean +/− SEM (N=5-8/group).

VCID causes episodic-like memory impairment in females but not males, while HF diet impairs episodic-like memory in both sexes.

We have previously shown that HF diet exacerbates cognitive impairment in young male VCID mice(13); however, effects in females and in middle-aged mice had yet to be examined. Therefore, cognitive testing was performed, as well as an open field test to assess general locomotor and anxiety-like behavior that could be influencing performance on cognitive tasks. Locomotor activity (tracklength) was higher in females and LF mice [main effects of sex (p<0.01) and diet (p<0.0001); Figure 5A]. Anxiety-like behavior was higher in HF mice (main effect of diet, p<0.05), an effect which was mainly driven by higher anxiety-like behavior in HF females (sex X diet interaction, p<0.05; Figure 5B). HF mice showed greater deficits in activities of daily living (nest building test, main effect of diet, p<0.05), while females showed greater deficits than males (main effect of sex, p<0.05; Figure 5C). On average, the nests scored only a 2 (HF groups) or 2.5 (LF groups), a very poor nest in contrast to an average score of 4.5 we normally observe in young mice (data not shown), indicating middle-aged mice showed impairments. Episodic-like memory was assessed using the novel object recognition test (NORT). LF male mice, regardless of surgery, showed greater recognition of the novel object (p<0.01), while HF males did not, indicating impairment in episodic-like memory due to diet (Figure 5D). LF sham females also showed significant recognition of the novel object (p<0.01), while HF sham, LF VCID, and HF VCID females did not (Figure 5D). Neither activity level (distance traveled) nor anxiety-like behavior (% center time) in the open field correlated with NORT performance (% time with novel object; Supplemental Figure 3). Taken together, the data show an impairment in episodic memory due to HF diet in both sexes and a female-specific impairment in episodic memory due to VCID surgery.

Figure 5. Effects of diet, sex, and VCID on anxiety-like behavior, activities of daily living, and episodic-like memory.

At the end of the study behavior was assessed. In the open field, track length (locomotor activity) was decreased in HF diet mice (A) and in females there were an overall effect of diet to increase anxiety-like behavior (less time in center), however post-hoc comparisons between groups did not reach significance (B). Nest building (24hrs), an assessment of activities of daily living), was poor in all middle aged mice (C), with the median score being a two which indicates less than half of the nestlet was torn. Episodic-like memory was tested in the novel object recognition test (NORT) test. Only 3 groups showed recognition of the novel object (intact memory): LF sham males, LF VCID males, and LF sham females. All other groups showed no recognition of the novel object (D). A-B) *p<0.05 for diet effect within same sex, **p<0.01 for diet within the same sex, ****p<0.0001 for diet within same sex. D) ⁺⁺p<0.01 vs. chance (50%), ⁺⁺⁺p<0.001 vs. chance (50%), ⁺⁺⁺⁺p<0.0001 vs. chance (50%). Red line = chance (50% of time). Data are presented as mean +/− SEM for A, C, D (N=13-19/group) and median +/− interquartile range for B (N=8-14/group).

Spatial memory is impaired by VCID in both sexes, and HF diet in females only.

Spatial learning and memory were assessed with the Morris water maze (MWM). During both the visible and hidden trials (Figures 6A-B), all mice showed a reduction in latency to reach the target (main effect of trial p<0.0001 for males and visible trials in females, p<0.01 for hidden trials in females), indicating that they were all able to learn the tasks. In the hidden trials, among the males there were also main effects of diet (p<0.0001) and surgery (p<0.05), with poorer spatial learning in both the HF diet and VCID groups. In female hidden trials, there was also a main effect of diet (p<0.01), with HF females having impaired spatial learning, but no effect of surgery. “Further, a 3-way repeated measures ANOVA revealed that there was a trial X surgery X diet interaction (p<0.01), with HF VCID females performing worse than other female groups at later trials. During the probe trial, a test of spatial memory, VCID surgery reduced preference for the target quadrant in males (main effect of surgery, p<0.05; Figure 6C-D). As a result, VCID males, regardless of diet, showed no significant preference for the target quadrant (p=0.1), indicating memory impairment. Meanwhile, sham males showed a strong preference (p<0.001). This indicates that VCID surgery, but not HF diet, impaired spatial memory in males. Among the females, there were main effects of both surgery (p<0.01) and diet (p<0.05), which resulted in an additive effect of treatments whereby the LF sham females showed the strongest preference (p<0.0001), the HF sham females a moderate preference (p<0.01), the LF VCID females a mild preference (p<0.05), and the HF VCID females showed no preference. LF VCID (p<0.05) and HF VCID females performed worse (p<0.01) than the LF sham females. Swim speed did not correlate with MWM performance (% time in target quadrant; Supplemental Figure 4) in HF-fed mice, although there was a very weak but statistically significant correlation in LF-fed mice. Taken together, this data shows that VCID causes a spatial memory deficit in both sexes. However, the effect of HF diet is sex-specific, with HF impairing spatial memory in females only.

Figure 6. Spatial memory is impaired by VCID in both sexes, and HF diet in females only.

At the end of the study spatial memory was assessed via the Morris water maze (MWM). During the training trials, the males (A) and females (B) improved performance across both visible and hidden trials (3 way repeated measures ANOVA, main effect of trial). Representative heat maps are shown for the probe trial, with target quadrant in lower right (C). During the probe trial, LF sham and HF sham males and females and LF VCID females showed a preference for the target quadrant, while all other groups showed no preference, indicating a memory impairment (D). In females, the HF VCID mice also showed a lower preference than LF sham females. ⁺p<0.05 vs. chance, ⁺⁺p<0.01 vs. chance, ⁺⁺⁺p<0.001 vs. chance, ⁺⁺⁺⁺p<0.0001 vs. chance,*p<0.05 LF sham female vs. LF VCID female, **p<0.01 LF sham female vs. HF VCID female. Red line = chance (25% of time). Data are presented as mean +/− SEM (N=13-19/group).

Sex-Specific Correlations Between Metabolic and Cognitive Measures.

In order to determine the predictive relationship between our main metabolic measures and our main cognitive measures, we performed a correlation matrix (Figure 7). Males showed strong correlations for % weight gain vs. % visceral fat (r=0.72; p<0.0001) and visceral fat TNFalpha (r=0.63; p<0.01). GTT was moderately correlated with other metabolic measures [(vs. % weight gain (r=0.50; p<0.0001), vs. % visceral fat (r=0.29; p<0.05)]. Females showed even stronger correlations between % weight gained and other metabolic measures [vs. % visceral fat (r=0.94; p<0.0001), vs. GTT (r=0.72; p<0.0001), vs. visceral fat TNFalpha (r=0.70, p<0.01)]. GTT was also more strongly correlated with % visceral fat (r=0.72; p<0.0001) and was correlated with visceral fat TNFalpha (r=0.54, p<0.05). In males, CBF (temporal region 4-months post-surgery) was not correlated with metabolic measures or memory. In females, there was a trend for a moderate negative correlation between % weight gained and CBF (r=−0.33; p=0.08, data not shown). In males only, episodic-like memory in the NORT was moderately correlated with several metabolic measures [vs. %weight gain (r=−0.31; p<0.01), vs. % visceral fat (r=−0.41; p<0.01)]. Unlike males, females did not show correlations between metabolic measures and NORT performance. However, in females there was a negative correlation between spatial memory in the MWM and % visceral fat (r=−0.29; p<0.05) and visceral fat TNFalpha levels (r=−0.50, p<0.05). Taken together, these data show sex-specific correlations between metabolic measures and specific memory tests, with weight/visceral fat correlating with episodic memory deficits in males only and visceral fat/visceral fat inflammation correlating with spatial memory deficits in females only.

Figure 7. Sex-Specific Correlations Between Metabolic and Cognitive Measures.

Using a correlation matrix, we compared the relationship between percent of body weight gained during the study (Weight; N=70/sex), visceral fat pad weight as a percentage of body weight (% Vis fat; N=51-52/sex), visceral fat TNFalpha levels (TNFɑ; N= 18-19 sex), the area under the curve for the glucose tolerance test performed after 6 months on the diet (GTT; N= 69-70/sex), percent time spent with the novel object during the novel object recognition test (NORT; N=66/sex), and the amount of time spent in the target quadrant region during the probe trial of the Morris water maze (MWM; N=63-68/sex). Pearson r values are presented. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Green = positive correlation, blue = negative correlation.

DISCUSSION

The goal of this study was to determine if there are sex differences in the effect of diet-induced metabolic disease (obesity with prediabetes) on cognitive deficits in a middle-aged mouse model of VCID. In our previous study, we found that age of HF diet onset has a huge impact on sex differences in metabolic outcomes. When the diet is initiated soon after weaning, males experience greater metabolic impairment (weight gain/glucose intolerance) than females; however, sex differences in metabolic outcomes are reversed when the diet is initiated in middle-age and females are more metabolically impaired than males (14). Because of these age-dependent sex differences in diet-induced obesity, and because mid-life obesity is a risk factor for VCID in humans, it was important to use middle-aged mice in this study. Additionally, VCID pathology occurs in middle-aged/elderly humans, further highlighting the clinical relevance of using middle-aged mice. Using a chronic cerebral hypoperfusion model of VCID and a chronic high fat diet to induce obesity with prediabetes, we show that females are more susceptible to metabolic and cognitive effects of a HF diet and to VCID in the absence of metabolic disease. In metabolically healthy mice, VCID impaired spatial memory in both sexes, while VCID impaired episodic-like memory in females only. HF diet caused greater metabolic impairment (weight gain, glucose intolerance, and fat accumulation) in middle-aged females compared to males. HF diet impaired episodic-like memory in both sexes, but impaired spatial memory in females only. Further, sex-specific correlations were found between metabolic measures and specific memory tests, with metabolic impairment correlating with deficits in episodic-like memory specifically in males and with spatial memory specifically in females. Overall, we found that HF diet caused greater metabolic impairment and a wider array of cognitive deficits in females than in males, but that the combination of HF diet and VCID caused a wide array of cognitive deficits in both sexes.

Using the unilateral carotid artery occlusion model of VCID, we found that blood flow deficits were exacerbated in middle-aged HF-fed compared to LF-fed females, but not males. We previously characterized CBF deficits in the this model of VCID using arterial spin labeling magnetic resonance imaging (MRI) in young male mice and found that cerebral hypoperfusion persists long-term (13, 15) and was exacerbated by HF diet in the hippocampus(13). However, CBF had not previously been assessed in females or middle-aged males. In the current study, using laser speckle contrast imaging, we found that in middle-aged VCID mice of both sexes CBF deficits are maintained in the temporal cortex for at least 4 months. Interestingly, percent weight gain showed a strong trend toward a negative correlation with CBF in females only, suggesting that obesity might have a greater effect on CBF in middle-aged females. In support of this, reductions in CBF immediately post-VCID surgery were larger in HF vs. LF VCID females in several regions. One limitation of the current study is that CBF was only assessed on the cortical surface, and changes in blood flow to deep brain structures could not be assessed. This may be important, as our prior study using MRI showed that hippocampal CBF was reduced in HF-fed young males(13). Thus, in future studies the current findings should be validated using MRI. Further, isoflurane is not the ideal anestheic for blood flow imaging. However, since all of our reported measures are within-subjects comparisons (% blood flow in right vs. left hemisphere or % change in blood flow pre/post-surgery) this is less of a concern, as we have found that these % difference measures are relatively consistent between various anesthetics (ketamine-xylazine and isoflurane). Finally, the underlying mechanism for the diet-induced reduction in CBF was not assessed in the current study; however, others have shown that prediabetes specifically impairs endothelial function in humans and rodents (2, 20-22). HF diet has also been shown to cause cerebrovascular remodeling, including smaller arterial lumens and thicker walls, in young male rodents (23, 24). In young male rats, chronic cerebral hypoperfusion has also been shown to impair endothelial function in parenchymal arterioles (25). However, most rodent studies in the cerebral vasculature have been conducted exclusively in males, so how HF diet or VCID alters cerebrovascular endothelial function in females is poorly understood.

We observed sex differences in cognitive effects of chronic cerebral hypoperfusion, with a wider array of memory deficits in middle-aged females compared to males. Chronic cerebral hypoperfusion is known to decrease cognitive function; however, much of the prior work in this field, including our own, had been conducted exclusively in males (13, 15, 26-28) or in young rodents (13, 15, 26-28). Our current data show that the chronic cerebral hypoperfusion VCID model elicited impairments in spatial memory in middle-aged mice of both sexes, but impairments in episodic-like memory in females only. The observed cognitive effects of VCID in middle-aged males was different than previous findings in young males, in which deficits are observed in episodic-like memory (15, 26, 27) but not spatial memory (15), which suggests an interaction between VCID and aging in males. In support of this, a recent study also found that spatial memory deficits in the water maze following chronic cerebral hypoperfusion increase with age in males (29). How sex hormones may influence VCID is largely unknown, although two recent studies did show that estradiol improved memory (episodic-like (30) and spatial memory (28)) in young male rodents. Ongoing studies in our lab are assessing the role of estradiol in females.

Our study is the first to identify sex differences in the effect of high fat diet in a mouse model of VCID at middle-age. Our prior work, and that of others, has shown that chronic HF diet elicits cognitive deficits in normal mice(12, 13, 31, 32) and exacerbates cognitive deficits in mouse models of dementia (13). However, much of this work has been conducted in a single sex (12, 13, 31, 32) or in young mice (13, 15, 32). For the first time, in the current study we assessed sex differences in the cognitive effects of HF diet in middle-aged mice using a mouse model of VCID. A combination of HF and VCID produced cognitive deficits in all tests in both sexes. While HF diet alone also impaired episodic-like memory in both sexes, it only impaired spatial memory in middle-aged females. It should also be noted that HF diet affected activity levels in both sexes and anxiety levels in females in the open field, which could be contributing to some of the observed cognitive outcomes. However, neither activity (tracklength in open field, swim speed in MWM) nor anxiety-like behavior (% time in center of open field) correlated with episodic or spatial memory in our HF-fed mice, suggesting that contributions of these factors were likely minimal. However, these limitations should still be kept in mind when interpreting correlations between cognitive measures and metabolic outcomes. Additionally, future studies should assess working memory and incorporate additional tests of spatial memory to corroborate these findings. The mechanism is yet to be determined, but is unlikely related to hyperglycemia, since in this model fasting blood glucose is not significantly elevated, but more likely is a result of the increase in glucose intolerance or visceral fat in the females or other factors related to this that we did not asses (e.g. hyperinsulinemia, hyperlipidemia). In support of a role for visceral fat, in females only visceral fat and visceral fat TNFalpha levels showed a negative correlation with spatial memory in the Morris water maze. This lack of correlation between spatial memory and visceral adiposity in middle-aged males was also recently reported by others (31). We found that in males only, visceral fat showed a negative correlation with episodic-like memory in the NORT, which has also be reported by others in middle-aged males on a HF diet (31).

We found that visceral fat levels correlated with cognitive impairment in a sex-specific manner, with higher visceral adiposity correlating with worse episodic memory in males but worse spatial memory in females. Adipose tissue is being increasingly recognized for its endocrine role in regulating metabolism homeostasis (33). In a state of obesity, altered secretion of adipokines and lipokines from adipose tissue is associated with the acceleration of cardiometabolic disease (31, 33). Clinically, visceral fat is associated with a higher prevalence of white matter lesions and small lacunar infarcts, indicators of small vessel disease (34, 35). Visceral adiposity has also been linked to cerebrovascular dysfunction and cognitive impairment in male mice (31); however, direct comparisons with females are rare. Our novel sex-specific correlations between visceral fat and specific memory tasks (episodic memory in males, spatial memory in females) could be due to sex differences in systemic inflammation induced by the visceral fat. A recent study by Ahnstedt et al. comparing immune profiles of adipose tissue from middle-aged males and females showed that female adipose tissue has higher levels of CD8+ T cells which also produced more pro-inflammatory cytokines (36). Further, the middle-aged female adipose tissue has lower levels of anti-inflammatory regulatory T cells compared to male adipose tissue (36). Thus, sex differences in systemic inflammation/adipose immune cell profiles could be a key mechanism underlying the sex difference in the effects of HF diet on cognitive function. Our finding that visceral fat is correlated with spatial memory deficits in females, but episodic memory deficits in males suggests sex-specific vulnerabilities to different cognitive domains. This sex-specificity is further supported by our finding that visceral fat TNFalpha levels correlated with poorer spatial memory in females, but not males. Future studies will determine if there are sex-specific vulnerabilities to different brain regions underlying these cognitive functions (i.e. hippocampus for spatial memory, cortex for episodic memory).

The sex-specific effects discovered in the current study could have been due to organizational effects set up during development, activational effects in of sex hormones in adulthood, or sex chromosome or x-inactivation effects. In support of an activational role for estrogens, at 15 months of age, our mice were perimenopausal, with irregular cycles. There is loss of estrogen with menopause, and estrogen has many protective effects on the cerebrovasculature and the brain in general (37, 38). In humans, menopause has been linked to metabolic and vascular dysfunction (39, 40) and is a sex-specific risk factor for VCID (5). Despite estradiol production by adipose tissue (41), HF-fed mice and diabetic women are often reported to have decreased estradiol levels (42). Further, we have found that estrogen deprivation (via inhibition of estradiol production) (36) severely impairs cerebrovascular endothelial function in female mice. Acute estradiol treatment increases CBF (43), which is positively correlated with cognitive performance (44), suggesting that estradiol may protect against cognitive decline by increasing CBF. In line with this, estradiol has been shown to protect against VCID in women (45). Estradiol has also been shown to protect against cognitive impairment in young male VCID rats (28), but has not been tested in females. While estradiol has a variety of protective effects on the brain (37, 38), it is important to note that protection is likely dependent on the timing of administration in postmenopausal women (<5yrs from menopause onset), formulation (17β-estradiol), and duration (46, 47).

In line with our current results, there is clinical evidence that obesity and prediabetes both have significant links to VCID. Even though the global prevalence of obesity is higher in women, a difference that increases with age (48), research on sex differences in the effect of obesity on VCID risk is lacking. Diabetes greatly increases the risk of VCID, and this increase in risk is 19% higher for women compared to men (6). Emerging evidence indicates that metabolic disease has effects on VCID in the prediabetic state, before the onset of hyperglycemia/diabetes. A prospective cohort study (3,774 men) found a correlation between prediabetes and VCID (49). Mid-life obesity in particular has been associated with an increase in brain aging (measured as whole-brain cerebral white matter volume, cortical thickness and surface area) compared to age matched healthy controls (50). Further, both obesity and prediabetes independently have been linked to cognitive deficits (51-60), hippocampal atrophy (3, 56, 60-62), white matter damage (50, 63, 64), and reduced cerebral blood flow (CBF) (65, 66) in humans. Clinically, prediabetes is associated with decreased cerebrovascular function (20). Likewise, obesity, as an independent factor, is associated with decreased cerebrovascular reactivity in humans (67). In rodents, HF diet and the resulting obesity/prediabetes have also been shown to impair both vascular and cognitive function. For example, HF diet impairs cerebrovascular endothelial function and causes blood brain barrier (BBB) dysfunction in mice, both of which precede cognitive decline (68). Overall, there is a strong link between metabolic disease in mid-life and increased dementia risk. While evidence on sex differences in this risk is lacking, there have been some indications that they may exist. We recently showed that adult hippocampal neurogenesis, which is involved in learning and memory, is impaired by HF diet in female, but not male, mice (69). Taken together, the results of our current study, as well as clinical and other preclinical research, suggest that obesity and prediabetes increase risk of VCID and may exacerbate cognitive deficits, to a larger extent in females.

In summary, we found sex-specific effects of HF diet or VCID, with each insult causing a wider array of cognitive deficits in females. HF diet-induced obesity with prediabetes impaired episodic-like memory in both sexes, while VCID impaired it in females only. In contrast, VCID impaired spatial memory in both sexes, while HF diet impaired it in females only. When HF diet and VCID were combined the cognitive deficits were sever and apparent across all cognitive tests in both sexes. Interestingly, sex-specific correlations were found between metabolic measures and specific memory tests, with obesity/visceral fat correlating with deficits in episodic-like memory specifically in males and visceral fat correlating with spatial memory deficits specifically in females. Thus, we have identified novel sex-specific effects of obesity with prediabetes on cognitive impairment in a mouse model of VCID. Overall, HF diet caused greater metabolic impairment and a wider array of cognitive deficits in middle-aged females compared to males. This is in line with the increased risk for VCID in diabetic women (6) and suggests that obesity with prediabetes might also be a major risk factor for VCID in women.

Glossary

AD

Alzheimer’s disease

CBF

cerebral blood flow

GTT

glucose tolerance test

HF

high fat

LF

low fat

MWM

Morris water maze

NORT

novel object recognition test

UCCAO

unilateral common carotid artery occlusion

VCID

vascular contributions to cognitive impairment and dementia