High‐caloric intake rescues early symptomatic AD‐induced hippocampal neurovascular coupling deficits

Abstract

INTRODUCTION

Alzheimer's disease (AD) involves progressive hippocampal dysfunction and atrophy. Obesity, common in AD patients, is a known dementia risk factor. Studying their interaction is difficult in humans due to AD's slow progression. Experimental AD models comorbid with obesity are needed for translational insights. This study examined the effects of a high‐carbohydrate, high‐fat (HCHF) diet in 12‐month‐old TgF344‐AD rats.

METHODS

Nontransgenic (nTg) and TgAD rats received CHOW or CHOW and HCHF diet items from 9 to 12 months of age. Hippocampal neurovascular function was assessed using pseudo continuous arterial spin labeling (pCASL)‐MRI during forepaw stimulation. Neuronal activity was recorded with Neuropixels probes.

RESULTS

CHOW‐fed TgAD rats showed reduced hippocampal cerebral blood flow (CBF), CBF changes spread, and neuronal power responses to somatosensory stimulation; all of these deficits were improved on the HCHF diet.

DISCUSSION

This approach provides a sensitive, task‐free assay of hippocampal neurovascular coupling. The transiently improved neurovascular and electrophysiological metrics in HCHF‐fed TgAD rats may be a manifestation of metabolically dysregulated AD brain benefitting from increased metabolite availability.

Highlights

• Pseudo‐continuous arterial spin labeling (pCASL) magnetic resonance imaging (MRI) ‐based characterization of hippocampal functional hyperemia.

• Hippocampal functional hyperemia is attenuated in symptomatic Alzheimer's disease (AD) pathology.

• A high‐carbohydrate, high‐fat (HCHF) diet transiently restores hippocampal functional hyperemia in symptomatic AD pathology.

1. INTRODUCTION

Alzheimer's disease (AD) is characterized by the spatiotemporally specific accumulation of amyloid and tau pathologies. ¹ , ² , ³ As the disease progresses, both neuronal and cerebrovascular changes emerge. Hippocampal neuronal function declines, leading to dysfunction and loss, ⁴ , ⁵ , ⁶ , ⁷ while functional hyperemia is attenuated, resulting in altered perfusion to activated regions. ⁸ , ⁹ , ¹⁰ , ¹¹ These neurodegenerative changes manifest in alterations in brain oscillatory activity, including increased theta and delta band power and decreased alpha and beta band power, ¹² , ¹³ , ¹⁴ , ¹⁵ which are associated with cognitive decline. ¹⁶

AD pathology rarely occurs in isolation (e.g., in only 97 out of 1079 patients ¹⁷ ), with significant cerebrovascular alterations being particularly prevalent not only in AD ¹⁸ but also in mild cognitive impairment (MCI). ¹⁹ Indeed, neuroimaging studies have demonstrated that cerebrovascular dysregulation often precedes beta amyloid deposition, metabolic dysfunction, and brain atrophy, ¹⁸ suggesting that vascular factors play a critical role in disease progression. Nonetheless, investigating the interaction between AD pathology and cerebrovascular comorbidities, particularly from a functional rather than structural perspective, remains challenging. The slow progression of AD, along with the heterogeneity of the patient population and the high prevalence of mixed pathologies, complicates efforts to delineate the contributions of vascular dysfunction to cognitive decline. ²⁰ , ²¹ , ²²

Obesity is a highly prevalent comorbidity of AD, with 40%–60% of dementia patients classified as overweight or obese. ²³ , ²⁴ Obesity induces significant vascular damage, so experimental models that integrate AD pathology and vascular comorbidities are needed, in combination with sensitive and translational imaging assays, to investigate the interaction between the AD pathology and obesity. The goal of the present study was to evaluate its sensitivity to AD pathology comorbid with obesity. We hypothesized that hippocampal functional hyperemia in response to somatosensory stimulation may serve as a sensitive neuroimaging assay of AD‐pathology–related neurovascular compromise.

The TgF344‐AD rat model (TgAD) recapitulates key aspects of AD pathology, including amyloidogenesis, tau pathology, frank neuronal loss, and progressive cognitive decline. ²⁵ , ²⁶ , ²⁷ To induce obesity, TgAD rats and their non‐transgenic (nTg) littermates were given ad lib access to a high‐carbohydrate, high‐fat (HCHF) diet comprised of grocery store‐purchased, energy‐dense, and highly palatable food items to imitate an obesogenic human diet. ²⁸ , ²⁹ , ³⁰ The interaction of AD pathology and obesity on hippocampal neurovascular function was then assessed using pseudo‐continuous arterial spin labeling (pCASL) MRI during forepaw stimulation, while hippocampal neuronal function was examined using high‐sensitivity electrophysiological recordings.

Hippocampal functional hyperemia was measured using pCASL‐MRI in response to somatosensory stimulation, a highly translational assay. Although it is well established that somatosensory signals reach the hippocampus via the perforant pathway originating from layer II entorhinal neurons, ²⁸ , ³¹ the corresponding hippocampal hyperemic response to somatosensory stimulation remains largely understudied. The somatosensory stimulation paradigm is unaffected by effort and attention variability, mitigating major confounds of the tests of cognitive function that are typically deployed in AD patients, which are influenced by both attention ³² and effort level, ³³ particularly in elderly populations. ³⁴ , ³⁵ , ³⁶ We here measured the hippocampal functional hyperemia to somatosensory stimulation to assess hippocampal function, of high interest given the pronounced susceptibility of the hippocampus to AD pathologies and its key role in AD‐elicited cognitive decline. Our findings demonstrate significant hippocampal vascular and neuronal dysfunction in the symptomatic stage of AD that was prevented by 3‐month‐long HCHF food item consumption. In addition, we have also established a noninvasive MRI assay of hippocampal neurovascular function for studying sequelae of AD pathology and its cerebrovascular comorbidities that is of high translational potential, given limited aging/neurodegeneration‐dependent effects on somatosensation and ease of delivering somatosensory stimuli.

2. METHODS

2.1. Animals

A total of 48 rats were used in this study (37 for MRI, 11 for electrophysiological recordings) and were assigned into four groups. The nTg and TgAD rat colonies at Sunnybrook were given ad libitum standard chow alone (CHOW group); or ad libitum chow and high carbohydrate and high fat diet (HCHF group) for three months prior to MRI, with ad libitum access to water alone (CHOW group) or water and a 12% w/v sucrose solution (HCHF group). The HCHF group thus received CHOW and water, supplemented with three highly palatable items (listed in Table 1) and a 12% w/v sucrose solution. The HCHF diet was administered three times a week for 3 months following earlier work on the evolution of diet effects in rats. ²⁹ , ³⁰ , ³⁷ The summary of animal numbers per experiment and the number of attritions are provided in Table 1. Experimental protocols were approved by the Animal Care Committee of the Sunnybrook Health Sciences Center, which adheres to the guidelines and policies of the Canadian Council on Animal Care and meets the Provincial Statute of Ontario, Animals for Research Act, and Federal Health of Animals Act. The TgF344‐AD rats colony was started following the donation of breeders by Dr. Terrence Town; these animals overexpress Swedish mutant human APP (APPswe: APP KM670/671NL) and delta exon 9 mutant of human PS1 (PS1ΔE9). ²⁵ The rats were kept in a reverse light cycle (12‐h dark:light, light from 7pm to 7am).

TABLE 1.

Summary of rats used for the study by experiment, diet, genotype, and sex.

			Genotype (sex)
Data/MRI protocol	Diet	Total N	nTg (M,F)	TgAD (M,F)	Attrition	Reason
Total for MRI imaging: 37
Resting state	CHOW	15	5 (3,2)	8 (4,4)	2	— a , b
	HCHF	22	10 (7,3)	11 (5,6)	1	— a , b
Functional hyperemia	CHOW	15	3 (2,1)	7 (3,4)	5	— a , b
	HCHF	22	9 (7,2)	9 (5,4)	4	— a
Total for electrophysiology: 11
Electrophysiology	CHOW	4	2 (0,2)	2 (0,2)
	HCHF	7	4 (3,1)	3 (2,1)

^a Severe motion (motion > 1 voxel).

^b Acquisition problem (set up/hardware/software).

Abbreviations: F, female; HCHF, high‐carbohydrate, high‐fat; M, male; MRI, magnetic resonance imaging, nTg, non‐transgenic; TgAD, TgF344‐AD rat model.

RESEARCH IN CONTEXT

1. Systematic review: Review of literature using PubMed revealed a gap in understanding the varied effects of obesity induced by high‐caloric diets in models of Alzheimer's disease (AD) and the need for a sensitive assay of hippocampal function in anesthetized animals.

2. Interpretation: Our findings demonstrate the utility of pseudo‐continuous arterial spin labeling (pCASL)‐MRI to measure cerebral blood flow (CBF) responses to somatosensation and to characterize the interaction of early AD pathology and high carbohydrate high fat (HCHF) diet.

3. Future directions: Future studies will focus on identifying metabolic pathway changes both systemically and centrally, as well as distinguishing components of the HCHF diet for AD animal models that drive its beneficial effects in early‐stage AD. Of particular interest will be the changes in GLUT1 and GLUT3 transporters in AD pathology comorbid with obesity. This understanding

2.2. In vivo MRI experiment

The rats were anesthetized with isoflurane (5% induction, 1%–1.5% maintenance) to minimize motion artifacts and stabilize baseline hemodynamics, while supporting extended image acquisition time, and affording simple delivery and rapid onset/offset kinetics. Once anesthetized, the animal was transferred to the MRI scanner. During experiment procedures, the rats were ventilated with 30% O₂ and 70% N₂ controlled by a programmable gas mixer (GSM‐3, CWE Inc., Ardmore, PA, USA). The systemic state was monitored throughout the session using a pulse oximeter recording heart rate, respiration rate, and pO2 (MouseOx Plus, Starr Life Sciences Corp., Oakmont, PA, USA). Respiratory rhythm was monitored by a pneumatic pillow (Model 1025, SA Instruments Inc., Stony Brook, NY, USA). Core body temperature was maintained at 37 ± 1°C using a rectal temperature probe and water‐circulating heating pad (SA Instruments Inc., Stony Brook, NY, USA). Somatosensory stimulation was achieved with subcutaneous MRI‐compatible electrodes (SWE‐30G/5T/2 subdermal wire electrode, IVES EEG Solutions, Newburyport, MA, USA) placed in each forepaw. The electrical pulse sequence was generated by the BioPac MP150 System (STM100C and STMISOC, Biopac Systems Inc., Goleta, CA, USA). Stimulation output and pulse oximeter physiology were recorded on the Acknowledge Data Acquisition and Analysis Software.

Rats were scanned in a 7T horizontal bore MRI scanner (BioSpec 70/30 USR with BGA‐12SHP gradients running ParaVision 6.0.1, Bruker BioSpin, Billerica, MA, USA). A 20‐mm diameter loop receive coil was placed over the animal's head and an 86‐mm diameter volume coil was used for radiofrequency (RF) excitation. Ear and incisor bars were used to secure the rat's head and positioned inside the magnet such that the center of the brain volume of interest was at the isocenter. Structural scan comprised a multislice 2D T2‐weighted rapid acquisition with refocused echoes (RARE) sequence (repetition time/echo time [TR/TE]_eff = 2500/44 ms, RARE factor = 8, field of view [FOV] = 20 × 20 mm², in‐plane resolution = 0.2 × 0.2 mm², slice thickness = 1 mm, averages = 2). ParaVision's Map Shim was run to minimize B₀ inhomogeneities. The experimental design is shown in Figure 1.

FIGURE 1.

Experimental design. Animals were scanned in magnetic resonance imaging (MRI) or underwent electrophysiological recordings at 12 months of age after being fed either CHOW alone or CHOW and high‐carbohydrate, high‐fat (HCHF) food items for 3 months, starting at 9 months of age. Pseudo‐continuous arterial spin labeling (pCASL) was used to measure resting perfusion and functional hyperemia in response to bilateral electric forepaw stimulation. Neuropixels 1.0 electrodes were lowered along the S1FL region of the cortex and the hippocampus, to record the neuronal activity in both somatosensory cortex and subfields of the hippocampus (CA2, CA3, and dentate gyrus [DG] regions).

2.3. Functional hyperemia acquisition and analysis

The pCASL acquisition protocol and analysis followed our previous work. ³⁰ , ³⁸ Unbalanced labeling with 5000 RF pulses was employed, each lasting 400‐µs with 4µT, B₁ amplitude, G_max/G_mean of 45/10 mT/m and 800 µs inter‐pulse interval, with the inter‐pulse phases optimized to maximize arterial spin labeling (ASL) contrast. ³⁹ The imaging slice was positioned axially at −5 mm anterior–posterior (AP), maximizing the cross‐sectional area of the hippocampal region; the labeling plane was placed 19 mm caudal from the imaging plane. To measure the relative perfusion change in response to bilateral electric forepaw stimulation, a single post‐labeling delay (PLD) pCASL was used ⁴⁰ with 𝝉 = 2 s (labeling duration), and w = 0.35 s (PLD). 264 repetitions of single‐slice label‐control echo planar imaging (EPI) pairs were acquired (TR/TE = 2500/11 ms, 0.32 mm in‐plane resolution, 1 mm slice thickness) during 4 min of baseline alternating with three 2‐min periods of bilateral forepaw electric stimulation (10 mA, 0.3 ms pulses delivered at 3 Hz). The 4‐min baseline acquisition was used to estimate resting cerebral blood flow (CBF) using the general kinetic model as described previously, ⁴¹ , ⁴² with 𝛌 = 0.9 mL/g (blood–brain water partition coefficient), ⁴³ R_1a = 0.45 s⁻¹ (longitudinal relaxation time of blood), ⁴⁴ and 𝝰 = 0.85.

The pCASL images were analyzed as described previously. ³⁰ , ³⁸ , ⁴⁵ At first, the pCASL images were masked (Aedes, Matlab) using registered MRI data in Fischer‐344 rat brain atlas ⁴⁶ using ANTsPy (Version 2.5.2), ⁴⁷ motion corrected (2dImReg, AFNI, Analysis of Functional NeuroImages), and low‐pass filtered (3dBlurToFWHM, AFNI; 0.55 mm FWHM). To generate activation maps showing which voxels were activated during stimulation, the pCASL data were fitted to a generalized linear model (3dDeconvolve, AFNI), producing least squares estimates of the linear regression coefficients with corresponding statistics, thresholding at a false discovery rate of q < 0.05 and minimum cluster size of 4 ³⁰ , ³⁸ for each generated region of interest (ROI). Using the registered MRI images in Fischer‐344 rat brain atlas ⁴⁶ space, the activation concordance map (overlap count map) was generated from stacking the activation masks of each rat, normalized by the number of subjects for each group. Using the hippocampal mask, subject‐wise‐average hippocampal ΔCBF (CBF—Mean of resting CBF) time courses were generated for each group.

2.4. Electrophysiological recordings and analysis of spontaneous and evoked activities

Eleven 12‐month‐old rats (4 CHOW, 7 HCHF) were utilized for electrophysiological recordings, after 3 months of diet exposure. Following induction with 5% isoflurane, animals were transferred to a stereotactic frame, where anesthesia was maintained at 2%–2.5% isoflurane. Physiological monitoring of heart rate, respiration rate, and pO2, were monitored as described in the MRI Methods section (MouseOx Plus, Starr Life Sciences Corp., Oakmont, PA, USA). Core body temperature was maintained at 37 ± 1°C via a feedback system using a rectal probe and a heat pad (TC‐1000, CWE Inc., Ardmore, PA, USA). BioPac MP150 System and Acknowledge Data Acquisition and Analysis Software (Biopac Systems Inc., Goleta, CA, USA) were used to acquire the animal physiology throughout the experiment. The skull was exposed and thinned using a 1 mm drill bit, over the right somatosensory forepaw area (S1FL), approximately AP +2.0 to −4.0 mm and ML +0.5–+5.5 mm. Once the ideal location, avoiding major pial vessels, was located, targeting AP −1.8 mm and ML +4.2 mm, an approximate 1.5 mm × 1.5 mm opening was created. At the completion of the surgical preparation, the rat was positioned in the Faraday cage, and the silicon probe Neuopixels 1.0 was inserted at 2 µm/s using a four‐axis manipulator (uMp‐4, Sensapex), following a trajectory through the S1FL region of the cortex and the CA2, CA3, and dentate gyrus (DG) region of the hippocampus. Forepaw stimulation was generated with subcutaneous electrodes placed in the left forepaw, contralateral to the cranial opening, as described in the In vivo experiments section. The forepaw stimulation consisted of 120 s of a 3 Hz pulse sequence (0.3 ms pulses at 10 mA), followed by 180 s of no stimulation. The pulse sequence was generated by the BioPac MP150 System and repeated five times (blocks). Neuropixels 1.0 allowed for the acquisition of extracellular recordings during electric forepaw stimulation and during the ten minutes of spontaneous baseline activity before the beginning of forepaw stimulation. The dual‐band electrophysiological signal generated by the Neuropixels probes was acquired via a PXIe acquisition module (IMEC) and visualized and stored using Open Ephys software in two different streams (local field potential [LFP]: 0.5–500 Hz, and AP: 0.3–10 kHz) sampled, respectively, at 2.5 and 30 kHz. The raw AP data were then further processed in spike interface ⁴⁸ and, using the preprocessing module, a phase‐shift correction, bandpass (0.5–6 kHz), and common‐median referencing were applied before the data were further processed in Kilosort 4 for spike sorting. ⁴⁹ Quality metrics were then applied to individual clusters where the LFP data were similarly preprocessed in the spike interface, where, following phase‐shift correction, a bandpass filter (0.5–300 Hz) and a downsampling step (1.2 kHz) were applied. The LFP data were further processed in MATLAB, where only a subset of channels, in the proximity of areas of stronger electrical activity (identified by current source density analysis), were used for power and phase‐amplitude‐frequency coupling (PAC) analysis. Channels were assigned to either S1FL or the hippocampus based on their depth from the brain surface and the anatomical map based on the rat brain atlas provided in Pinpoint. ⁵⁰ LFP power was estimated in each condition (baseline, forepaw stimulation) using MATLAB's built‐in function “spectrogram” over 2 s window time using a Hann window with 50% overlap to reduce spectral leakage. Power was then expressed in the dB scale by computing the logarithm. The normalized change in power (Mean power during stimulation—Mean power at baseline) across six bands of interest (delta: 1–4 Hz, theta: 4–8 Hz, alpha: 8–12 Hz, beta: 12–30 Hz, low gamma: 30–80 Hz, and high gamma: 80–120 Hz) was computed from the power spectra at baseline and during stimulation. PAC was computed to estimate network synchronization by calculating the Modulation Index (MI), ⁵¹ which quantified the dependency of the gamma band amplitude to the phase of lower frequencies. Comodulograms were created using 1 Hz frequency resolution (phase component) and 3 Hz resolution (amplitude component) as described by Tort. ⁵¹ The average MI was calculated as the mean of the values in phase‐amplitude coupling to theta, low‐ and high‐gamma bands (e.g., theta‐low gamma and theta‐high gamma). The change in MI (Mean MI during stimulation—Mean MI at baseline) was computed from MI at baseline and during stimulation.

2.5. Immunofluorescence and immunohistochemistry

Upon completion of diet exposure and/or MRI imaging, rats were transcardially perfused under 5% isoflurane with phosphate buffered saline (PBS)‐0.1% heparin, followed by PBS‐4% paraformaldehyde, before brains were extracted, post‐fixed overnight in PBS‐4% paraformaldehyde, and cryopreserved in PBS‐30% sucrose for microtome sectioning. For each brain, three 40‐µm sections evenly spaced between −3.0 mm AP and −4.5 mm AP were sampled.

For immunofluorescence, all sections were blocked with 0.5% Triton X‐100 and 0.5% bovine serum albumin (BSA) in PBS for 1 h, followed by overnight incubation with the primary antibodies/dyes diluted in blocking buffer. The primary staining used tomato lectin‐Texas Red (1:250, Vector #TL‐1176‐1), which was detected by a secondary antibody anti‐rabbit IgG‐Alexa 488 (1:250, ThermoFisher #A21206) for 2 h at room temperature. Sections were then mounted for imaging.

Zeiss Observer.Z1. was used to acquire fluorescence images, which were analyzed using Fiji. For analysis of capillary diameter, 100 vessels (< 10 µm in diameter) were included per rat. The diameter of each capillary was estimated by dividing the cross‐sectional area by vessel length.

2.6. Statistical analysis

A linear mixed‐effects model was used, which included effects of diet, genotype, and sex and their interactions with resting CBF, extent of the brain showing significant CBF changes in response to forepaw stimulation, and the magnitude of CBF responses to forepaw stimulation. The model was implemented in the MixedLM class from the statsmodel library in Python. The mixed‐effects model employed was used to analyze relationships between the dependent variable (e.g., ratio of activated voxels) and a combination of independent variables (e.g., diet, genotype) and interactions between these independent variables. For example: resting_CBF = β₀diet + β₁genotype + β₂rat_id + ϵ where β₀ , β₁ , were the fixed effect coefficients that represents the average change in response variable (resting_CBF) associated with one‐unit change in corresponding predictor (e.g., diet, genotype) while holding other variables constant; β₂ was a random effect coefficient; rat_id was the random effect grouping variable, and ϵ was the error term. This formula specified the fixed predictors (diet, genotype) of the independent variable (e.g., resting CBF) and incorporated the random effect of the grouping variable “rat_id” allowing the model to estimate variability between rats. For electrophysiology data modeling, the channels were nested within subjects. Subsequently, pairwise comparisons were done using Tukey's honestly significant difference (HSD) test via the statsmodels.stats.multicomp.pairwise_tukeyhsd function (Version 0.13.5) in Python ⁵² or Mann–Whitney U test using scipy.stats module in Python (Version 1.7.3.) ⁵³ for normally and not normally distributed data sets, respectively. Bonferroni correction was used in channel‐wise electrophysiological recording data using the multipletests function from statsmodel package in Python. ⁵²

3. RESULTS

3.1. HCHF‐fed cohorts gained significantly more weight than the CHOW‐fed cohorts

Rats were fed an HCHF diet for 3 months before MRI and electrophysiological recordings at 12 months of age (Figure 1). We imaged a total of 37 rats over 260 h of MR imaging and 120 man‐hours of feeding/weighing. Rats were included in the analysis if their imaging data were free from severe motion (defined as movement exceeding one voxel) and if no acquisition‐related technical issues (e.g., setup errors, software malfunctions, or hardware failures) were present. During the 3 months of HCHF feeding (food items listed in  Table S1), the 9‐month‐old nTg and TgF344‐AD rats consumed significantly more calories per week, on average (Figure 2A | HCHF‐fed rats = 570 ± 110 kcal, CHOW‐fed rats = 292 ± 66 kcal, p < 0.001), as well as more carbohydrates (Figure 2A | HCHF‐fed rats = 361± 83 kcal, CHOW‐fed rats = 223 ± 50 kcal, p < 0.01), fats (Figure 2B | HCHF‐fed rats = 209 ± 32 kcal, CHOW‐fed rats = 69 ± 16 kcal, p < 0.001), and sodium (Figure 2C | HCHF‐fed rats = 550 ± 120 mg, CHOW‐fed rats = 384 ± 87 mg, p < 0.01) when compared to CHOW‐fed rats. The significantly higher calorie intake of HCHF‐fed rats was primarily driven by their consumption of sugar water. In addition, HCHF‐fed rats consumed more HCHF food items (92% of total calories) relative to the standard CHOW pellets (8% of total calories) (Figure 2D). The total calories from fats were higher in HCHF‐fed rats when compared to CHOW‐fed rats (Figure 2E). HCHF feeding resulted in significant increase in weight, in both female nTg (HCHF‐fed: 384 ± 32 g, CHOW‐fed: 289 ± 4 g, p < 0.001) and TgAD (HCHF‐fed: 383 ± 25 g, CHOW‐fed: 320 ± 13 g, p < 0.01) rats; as well as in male nTg (HCHF‐fed: 605 ± 55 g, CHOW‐fed: 471± 10 g, p < 0.01) and TgAD (HCHF‐fed: 629± 49 g, CHOW‐fed: 428± 99 g, p < 0.001) rats, when compared to their respective CHOW‐fed littermates (Figure 2F).

FIGURE 2.

High‐carbohydrate, high‐fat (HCHF) consumption led to significant weight gain. (A) The average total calories, (B) calories from carbohydrates, (C) calories from fats and (D) amount of sodium consumed by HCHF‐fed rats were greater when compared to CHOW‐fed rats. In general, HCHF‐fed rats consumed more HCHF food items than CHOW pellets. (E) The total calories from fat was greater in HCHF‐fed rats when compared to CHOW‐fed rats. (F) Across all cohorts, three months of HCHF feeding resulted in an increase in weight in both female (by 41%) and male (by 20%) rats. *p < 0.05, **p < 0.01. HCHF, high‐carbohydrate, high‐fat.

3.2. Utility of somatosensory stimulation for assessment of hippocampal functional hyperemia

Electrophysiological recordings and pCASL MRI imaging were utilized to measure changes in neuronal activity and functional hyperemia in the hippocampus of anesthetized CHOW‐fed nTg rats during electric forepaw stimulation. To assess neuronal response to forepaw stimulation, we quantified stimulation‐induced changes in hippocampal LFP power. Representative hippocampal LFP spectrogram at baseline, during forepaw stimulation, and post stimulation is shown in Figure 3A. LME analysis showed that power was significantly different during stimulation when compared to baseline levels across different bands (power during stimulation: theta, p < 0.001; alpha, p = 0.017; low gamma, p < 0.001; high gamma, p < 0.001). In addition, the post‐stimulation power was also significantly different from baseline (power post‐stimulation: theta, p < 0.001; alpha, p < 0.001; low gamma, p < 0.001; high gamma, p < 0.001). Time course of mean power at different bands (theta, alpha, low‐ and high‐gamma) at baseline, during stimulation, and post stimulation periods is shown in Figure 3B. At the start of stimulation, power exhibited a sharp and transient increase, followed by a gradual decline over time across the frequency spectrum, with the theta band decrease being below baseline level. After stimulation ended, power overshot across all frequency bands before gradually returning to baseline levels.

FIGURE 3.

Hippocampal neuronal power and functional hyperemia of non‐transgenic (nTg) rats in response to electric forepaw stimulation. (A) Representative spectrogram computed from hippocampal local field potential (LFP) at baseline, during forepaw stimulation, and post‐stimulation. The power was computed as 20log(V), where V is the recorded voltage (in the range of ± 0.002 V). (B) The time course average of power change (Power—Mean power at baseline) was computed for each frequency band across each channel and blocks. The average power transiently increased during and after forepaw stimulation. (C) Activation map in a representative subject. (D) Maps of coincidence of activations across subjects, in atlas space, where higher values indicate voxel activation in a greater number of subjects. (E) Time course of across‐subject mean cerebral blood flow (CBF) changes in the hippocampus, with the black bar indicating the forepaw stimulation period. For electrophysiology data, n = 2 (channels = 29, blocks = 5) were CHOW‐fed nTg rats. For MRI data, n = 3 were CHOW‐fed nTg rats.

The pCASL signal was averaged voxelwise within the hippocampus region to estimate cerebrovascular reactivity, that is, the increase in cerebral blood flow in response to forepaw stimulation. A representative activation map in response to bilateral forepaw stimulation of CHOW‐fed nTg cohort is shown in Figure 3C. Coincidence of activations across subjects (Figure 3D), where higher values represent voxels that were activated in more subjects, showed consistent activation during stimulation, not just in the cortex, but also in the hippocampus (white outline in the concordance map). Time course of across‐subject mean blood flow change (ΔCBF) in the hippocampus at baseline, during stimulation, and after stimulation is shown in Figure 3E. During stimulation, ΔCBF increased and remained elevated before gradually returning to baseline after stimulation ended. These findings demonstrate the sensitivity of our protocol for measuring hippocampal functional hyperemia during forepaw stimulation.

3.3. Attenuation of hippocampal evoked responses and hippocampal hyperemia in the presence of AD transgene

Having demonstrated that pCASL MRI can be used to measure functional hyperemia in the hippocampus in response to electric forepaw stimulation, this protocol was utilized to evaluate the neurovascular effects of AD pathology. Representative hippocampal power spectrograms at baseline, during forepaw stimulation, and post stimulation of CHOW‐fed nTg versus TgAD rats are shown in Figure 4A. LME analysis showed that nTg and TgAD rats exhibited power changes during forepaw stimulation relative to pre‐stimulation baseline (power during stimulation vs. baseline: theta, p < 0.001; alpha, p = 0.021; low gamma, p < 0.001; high gamma, p < 0.001) and post‐stimulation relative to pre‐stimulation baseline (power post‐stimulation vs. baseline: theta, p < 0.001; alpha, p < 0.001; low‐gamma, p < 0.001; high‐gamma, p < 0.001). Time courses of mean power of theta and low gamma bands at baseline, during stimulation, and post‐stimulation periods of CHOW‐fed nTg and TgAD rats are shown in Figure 4B. On stimulus onset, power rose sharply, then gradually declined over the stimulation period, then rose after stimulus offset before gradually returning to baseline levels. Similar power response evolution was observed in TgAD rats, but the changes in power were attenuated when compared to those of nTg rats. LME analysis showed significant variability in baseline power and power changes across recording channels in all frequency bands (theta, alpha, low‐gamma, and high‐gamma; all p < 0.001), highlighting the importance of modeling channel‐wise differences. To assess the level of hippocampal network activity, the channel‐wise baseline power (Figure 4C) and power change (Power during stimulation—Mean power at baseline, Figure 4D) were calculated in each rat. CHOW‐fed TgAD rats showed a trend of lower baseline alpha power (−86.8± 3.4 dB) and elevated baseline high gamma power (−111.1± 1.9 dB), when compared to CHOW‐fed nTg rats (alpha: −85.2 ± 4.4 dB, p = 0.058; high‐gamma: −112.3 ± 3.0 dB, p < 0.001) (Figure 4C). Upon stimulation, CHOW‐fed TgAD rats showed attenuated power changes (theta: −3.7 ± 2.4 dB, alpha: −1.0 ± 1.5 dB, low gamma: 0.61 ± 0.82 dB, high gamma: 2.6 ± 1.4 dB) when compared to nTg rats (theta: −1.4 ± 1.9 dB, p < 0.001; alpha: 0.4 ± 1.9 dB, p < 0.001; low gamma: 1.04 ± 0.93 dB, p < 0.05; high gamma: 1.7 ± 1.5 dB, p < 0.001) (Figure 4D).

FIGURE 4.

Attenuated hippocampal power and functional hyperemia in symptomatic rats. (A) Representative electrode recordings of spontaneous activity in the hippocampus at baseline and during forepaw stimulation of CHOW‐fed non‐transgenic (nTg) and TgF344‐AD rat model (TgAD) rats. The power, in dB, was computed as 20log(V), where V is the voltage measured in Volts (for a typical local field potential [LFP] signal of ± 0.0002 V, the power in dB is negative). (B) Average channel‐wise electrophysiological recording of theta and low gamma powers at baseline, during stimulation, and post stimulation showing increased neuronal activity in both nTg and TgAD rats in response stimulation. The period of forepaw stimulation is indicated by a horizontal black bar. (C) Baseline theta, alpha, low‐ and high‐gamma powers. Baseline high‐gamma power of TgAD rats was higher when compared to nTg rats. (D) TgAD rats showed a significant reduction in power in response to stimulation (stimulation – baseline) across theta, alpha, low‐ and high‐gamma bands. (E) Representative maps of CBF changes, and (F) maps of coincidence of activations across subjects of CHOW‐fed nTg and TgAD rats. (G) Subject wise‐average hippocampal ΔCBF time courses normalized by the mean resting cerebral blood flow (CBF); the periods of bilateral forepaw stimulation are indicated by horizontal black bars. (H) The resting CBF of CHOW‐fed cohorts, derived from the baseline acquisition, were indistinguishable from each other. (I) The hippocampal CBF change in response to forepaw stimulation was attenuated in CHOW‐fed TgAD (7 ± 14 mL/100 g/min) rats relative to CHOW‐fed nTg (49 ± 21 mL/100 g/min) rats. (J) The volume of hippocampal activation was attenuated in CHOW‐fed TgAD (0.03 ± 0.08 mm³/mm³) rats compared to CHOW‐fed nTg (0.34 ± 0.14 mm³/mm³, p < 0.001) rats. For electrophysiology data, n = 4 [2 nTg (channels = 29, blocks = 5), 2 TgAD (channels = 29, blocks = 5)] were CHOW‐fed rats. For MRI data, n = 10 (3 nTg, 7 TgAD) were CHOW‐fed rats. *p < 0.05, **p < 0.01, ***p < 0.001.

Baseline LFPs in the primary somatosensory cortex (S1FL) were also recorded (Figure S1A). In contrast to the attenuated power observed in hippocampal alpha band of CHOW‐fed TgAD rats compared to CHOW‐fed nTg rats, the power (theta, alpha, low‐gamma, and high‐gamma) in S1FL were larger in TgAD rats when compared to nTg rats (theta | nTg: −84.1 ± 2.1, TgAD: −80.9 ± 3.5, p < 0.001; alpha | nTg: −88.2 ± 2.3, TgAD: −85.5 ± 2.2, p < 0.001; low‐gamma | nTg: −106.0 ± 1.5, TgAD: −102.9 ± 1.9, p < 0.001; high‐gamma | nTg: −112.7 ± 1.7, TgAD: −110.5 ± 1.5, p < 0.001) (Figure S1B). In response to stimulation, unlike in the hippocampus where CHOW‐fed TgAD showed weaker power changes than CHOW‐fed nTg, CHOW‐fed TgAD and nTg rats showed comparable changes in power (theta, alpha, low gamma) (Figure S1).

The hippocampal spiking response to forepaw stimulation of CHOW‐fed cohorts was also measured (Figure S2). The positive spiking responses of TgAD rats (1.2 ± 1.0) were comparable to nTg rats (1.5 ± 1.5, p = 0.43) (Figure S2A). The magnitude of negative spiking responses in TgAD rats (−0.48 ± 0.19) was greater (by 34% ± 19%) when compared to that of nTg rats (−0.36 ± 0.14, p < 0.01). In S1FL, the magnitude of negative spiking responses of CHOW‐fed TgAD rats was lower when compared to CHOW‐fed nTg rats (nTg: −0.45 ± 0.13, TgAD: −0.42 ± 0.11, p < 0.01) (Figure S2B).

Representative activation maps, concordance of activation maps, and mean ΔCBF time course in response to bilateral forepaw stimulation of CHOW‐fed nTg and TgAD cohorts are shown in Figure 4E–G, respectively. The average hippocampal CBF change (Figure 4I) and spread of activation (Figure 4J) in response to forepaw stimulation of CHOW‐fed TgAD rats (ΔCBF: 7 ± 14 mL/100 g/min, activation volume: 0.03 ± 0.08 mm³/mm³) were attenuated when compared to that of CHOW‐fed nTg rats (ΔCBF: 49± 21 mL/100 g/min, p = 0.029; activation volume: 0.34 ± 0.14 mm³/mm³, p < 0.001). Hippocampal resting perfusion was not distinguishable between CHOW‐fed nTg and TgAD cohorts (Figure 4H).

In summary, AD pathology was associated with reduction of hippocampal power change (theta, alpha, low‐ and high‐gamma) and an attenuation of functional hyperemia (CBF change and activation volume) in response to forepaw stimulation. In percent terms, the functional hyperemia was attenuated more than what would be expected from the level of attenuation of the neuronal activation.

3.4. Enhancement of hippocampal evoked responses with HCHF in AD pathology

After establishing the sensitivity of our protocol to AD pathology, we used it to assess the interaction between AD pathology and obesity. We have speculated that an HCHF diet would lead to exacerbation of the AD phenotype, but, surprisingly, found it to have protective effects. To evaluate the effects of diet on neuronal function in the presence of AD pathology, evoked LFPs in the hippocampus were recorded and changes in power analyzed across the frequency spectrum. Representative power spectra (Figure 5A) and mean power change time courses (theta and low‐gamma, Figure 5B) of CHOW‐ and HCHF‐fed cohorts were computed. HCHF‐fed rats exhibited a sharp and transient power increase following stimulation onset, followed by a gradual decline (Figure 5B). After stimulation, neuronal power in HCHF‐fed rats gradually declined toward the baseline, whereas CHOW‐fed rats exhibited a power overshoot (Figure 5B). LME analysis showed significant variability in baseline power and power changes across recording channels in all frequency bands (theta, alpha, low‐gamma, and high‐gamma; all p < 0.001), highlighting the importance of modeling channel‐level differences. LME modeling indicated that baseline power was comparable across the genotypes, with a trend toward a significant effect of diet (HCHF vs. CHOW: theta, p = 0.087; alpha, p = 0.085; low‐gamma, p = 0.082; high‐gamma, p = 0.085). In terms of pairwise comparisons, hippocampal baseline theta, alpha, and low gamma powers of HCHF‐fed TgAD rats were potentiated compared to those of CHOW‐fed TgAD rats (theta | HCHF: −80.4 ± 4.9 dB, CHOW: −82.7 ± 3.6 dB, p < 0.001; alpha | HCHF: −83.2 ± 5.2 dB, CHOW: −86.8 ± 3.4 dB, p < 0.001; low gamma | HCHF: −101.5 ± 4.1 dB, CHOW: −103.6 ± 2.9 dB, p < 0.001) (Figure 5C). However, LME modeling indicated that the HCHF diet affected the power changes in response to stimulation of nTg and TgAD rats differently (diet*genotype: theta, p = 0.003; alpha, p < 0.001; low‐gamma, p < 0.001; high‐gamma, p < 0.001). TgAD rats exposed to the HCHF diet exhibited enhanced power changes in response to forepaw stimulation as shown in Figure 5D (theta: 1.4 ± 3.0 dB, alpha: 2.6 ± 2.3 dB, low‐gamma: 2.4 ± 1.7 dB, high‐gamma: 2.6± 1.4 dB) compared to CHOW‐fed TgAD rats (theta: −3.7± 2.4 dB, p < 0.001; alpha: −1.0 ± 1.5 dB, p < 0.001; low‐gamma: 0.61 ± 0.82 dB, p < 0.001; high‐gamma: 0.62 ± 0.67 dB, p < 0.01). In summary, the HCHF diet resulted in potentiation of both baseline (alpha, low‐ and high‐gamma in nTg rats and theta, alpha, and low‐gamma in TgAD rats) and stimulation‐evoked hippocampal neuronal activity (in theta, alpha, low‐ and high‐gamma for both nTg and TgAD rats). Additionally, HCHF‐fed TgAD rats exhibited greater baseline and stimulation‐evoked hippocampal activity compared to CHOW‐fed nTg rats.

FIGURE 5.

High‐carbohydrate, high‐fat (HCHF) resolves hippocampal power and functional hyperemia deficits in symptomatic Alzheimer's disease (AD). (A) Representative electrode recordings of spontaneous activity in the hippocampus at baseline, during and post forepaw stimulation of CHOW‐fed and HCHF‐fed non‐transgenic (nTg) and TgF344‐AD rat model (TgAD) rats. The raw power was computed as 20log(V). (B) Average channel‐wise electrophysiological recording of theta and low gamma power showing change in neuronal activity in both HCHF‐fed nTg and TgAD rats during and post stimulation. (C) At baseline, CHOW‐fed TgAD rats had a trend of lower alpha and higher high‐gamma powers when compared to nTg rats. With HCHF diet, TgAD rats had potentiated theta, alpha, and low‐gamma baseline powers when compared to CHOW‐fed TgAD rats. (D) Change in power (Mean power during stimulation—Mean power at baseline) time course averaged across channels during stimulation at theta, alpha, low‐ and high‐gamma bands. HCHF‐fed TgAD rats exhibited elevated power change (theta, alpha, low‐ and high‐gamma) when compared to CHOW‐fed TgAD rats. (E) Representative activation maps of cerebral blood flow (CBF) changes, and (F) maps of coincidence of activations across subjects. (G) Subject‐wise average hippocampal ΔCBF time courses normalized by the mean resting CBF. (H) The resting CBF of CHOW‐ and HCHF‐fed cohorts, derived from the baseline acquisition, were indistinguishable from each other. (I) The hippocampal CBF change in response to forepaw stimulation of TgAD rats was enhanced with HCHF diet (HCHF: 60 ± 26 mL/100 g/min; CHOW: 7 ± 14 mL/100 g/min) (diet‐genotype, p = 0.001). (J) The volume of hippocampal activation of TgAD rats was also enhanced with HCHF diet (HCHF: 0.25 ± 0.15 mm³/mm³; CHOW: 0.03 ± 0.08 mm³/mm³, p < 0.00) (diet‐genotype, p < 0.001). (K) Correlation plot of CBF change and power change in response to stimulation showed HCHF‐fed TgAD rats had enhanced both CBF and power changes when compared to CHOW‐fed TgAD rats. For electrophysiology data, n = 11, where n = 4 [2 nTg (channels = 29, blocks = 5), 2 TgAD (channels = 29, blocks = 5)] were CHOW‐fed and n = 7 [4 nTg (channels = 29, blocks = 7), 3 TgAD (channels = 26, blocks = 5)] are HCHF‐fed. For MRI data, n = 28, where n = 10 (3 nTg, 7 TgAD) were CHOW‐fed and n = 18 (9 nTg, 9 TgAD) are HCHF‐fed. *p < 0.05, **p < 0.01, ***p < 0.001.

HCHF feeding resulted in elevation of S1FL baseline alpha, low‐ and high‐gamma powers in nTg animals (Figure S1B: green bars) and attenuation in low‐ and high‐gamma powers in TgAD animals (Figure S1B: orange bars). Further, baseline S1FL neuronal power was elevated in HCHF‐fed TgAD versus nTg animals (Figure S1B: theta, alpha, low gamma), but high gamma was slightly decreased (Figure S1B: high gamma). HCHF feeding resulted in elevation of stimulation‐induced power response in both TgAD and nTg rats (Figure S1C: theta, alpha, low gamma, high gamma: green and orange horizontal bars), and the HCHF‐fed TgAD rats had elevated S1FL neuronal power responses across all bands investigated. In toto, the HCHF diet enhanced the stimulation‐evoked neuronal activity in the S1FL cortex, but increased both the baseline and stimulation‐evoked neuronal activity in the hippocampus. However, in CHOW‐fed rats, S1FL power changes did not differ between genotypes, whereas hippocampal power changes did show a genotype‐dependent difference (Figure 4D).

To determine whether the HCHF‐induced improvements in hippocampal blood flow and oscillatory power reflected enhanced neuronal output rather than subthreshold activity or network synchrony alone, we measured stimulus‐evoked spiking responses (Figure S2). HCHF differentially affected the positive spiking responses in nTg and TgAD rats (diet*genotype, p < 0.001) (Figure S2A). The positive spiking responses of TgAD rats (1.2 ± 1.0) were not affected by HCHF diet (TgAD: 1.1 ± 1.1; p = 0.19) (Figure S2A). However, HCHF‐fed nTg rats (0.81 ± 0.81) showed attenuated (by 56% ± 65%) spiking response when compared to CHOW‐fed nTg rats (1.5 ± 1.5, p < 0.05) (Figure S2A). The magnitude of negative spiking responses was attenuated in both nTg (CHOW: −0.36 ± 0.14, HCHF: −0.17 ± 0.10; p < 0.001) and TgAD (CHOW: −0.48 ± 0.19, HCHF: −0.38 ± 0.16; p < 0.001) rats when fed with HCHF (diet, p < 0.001). Overall, HCHF feeding reduced positive hippocampal spiking responses in nTg rats but had no effect on TgAD rats. Negative spiking responses were greater in TgAD rats but were attenuated by HCHF in both genotypes.

In contrast to the hippocampus, spiking response in S1FL of TgAD rats (Figure S2B) was not affected by HCHF diet (whereas the magnitude of hippocampal negative spiking responses in TgAD rats was attenuated with HCHF feeding). The positive spiking responses of HCHF‐fed TgAD rats in S1FL were lower when compared to those of HCHF‐fed nTg rats (nTg: 2.2 ± 1.7, TgAD: 0.93 ± 0.72, p < 0.001) (Figure S2B). HCHF‐fed nTg rats had a potentiated magnitude of negative spiking responses when compared to CHOW‐fed nTg rats (HCHF‐nTg: −0.32 ± 0.16, CHOW‐nTg: −0.45 ± 0.13, p < 0.001), while there was no diet effect in TgAD rats (Figure S2B). In S1FL, the positive spiking responses of HCHF‐fed TgAD rats were lower than those of HCHF‐fed nTg rats. HCHF attenuated the magnitude of negative spiking responses in nTg rats but had no effect on TgAD rats.

3.5. HCHF feeding enhanced the hippocampal theta‐gamma coupling in symptomatic AD rats

To understand how the HCHF diet impacts neuronal network coordination in AD‐affected hippocampal networks, we measured the relationship between the slow oscillation phases (i.e., theta frequency) and amplitude of faster oscillations (i.e., gamma frequency). Phase–amplitude coupling (PAC) is key for information processing and has been proposed to be a predictor of cognitive function. ⁵¹ , ⁵⁴ , ⁵⁵ The MI, used as an index of PAC, was visualized via comodulograms (Figure S3A), computed in theta‐low gamma and theta‐high gamma frequency combinations at baseline and during stimulation. The average MIs of theta‐low/high gamma were computed at baseline (Figure S3B) and stimulation‐induced MI change (MI during stimulation—Mean MI at baseline) (Figure S3C).

LME analysis showed significant variability in baseline MI and MI changes during stimulation across recording channels (theta‐low gamma, and theta‐high gamma; all p < 0.001), highlighting the importance of modeling channel‐level differences. LME modeling indicated that the HCHF diet affected the baseline MI of nTg and TgAD rats differently (diet*genotype: theta‐low gamma, p = 0.003; theta‐high gamma, p < 0.001). Pairwise comparison at baseline, CHOW‐fed TgAD rats showed higher hippocampal MI compared to nTg rats (theta‐low gamma | TgAD: −7.71 ± 0.83, nTg: −7.99 ± 0.66, p < 0.001; theta‐high gamma | TgAD: −7.31 ± 0.77, nTg: −7.51 ± 0.60, p < 0.001). During forepaw stimulation, TgAD rats also showed a greater change in MI (theta‐low gamma: 0.57 ± 0.73 and theta‐high gamma: 0.50 ± 0.73) than did the compared to nTg rats (theta‐low gamma: 0.23 ± 0.54, p < 0.001 and theta‐high gamma: 0.35 ± 0.51, p < 0.001). Remarkably, HCHF‐fed TgAD rats had increased baseline MI (theta‐low gamma: −7.28 ± 0.99, and theta‐high gamma: −6.99 ± 0.87) compared to CHOW‐fed TgAD rats (theta‐low gamma: −7.71 ± 0.83, p < 0.001; and theta‐high gamma: −7.31 ± 0.77, p < 0.001) and to CHOW‐fed nTg rats (theta‐low gamma: −7.99 ± 0.66, p < 0.001; and theta‐high gamma: −7.51 ± 0.60, p < 0.001). However, HCHF‐fed nTg rats had attenuated baseline MI (theta‐low gamma: −8.21 ± 0.49 and theta‐high gamma: −7.91 ± 0.49) compared to CHOW‐fed nTg rats (theta‐low gamma: −7.99 ± 0.66, p < 0.001 and theta‐high gamma: −7.51 ± 0.60, p < 0.001).

During forepaw stimulation, LME analysis showed no significant effect of genotype (theta‐low gamma, p = 0.467; theta‐high gamma, p = 0.360), diet (theta‐low gamma, p = 0.237; theta‐high gamma, p = 0.242), and genotype‐diet interaction (theta‐low gamma, p = 0.115; theta‐high gamma, p = 0.126) in MI change during stimulation. After pairwise comparison, both HCHF‐fed nTg (theta‐low gamma: ‐0.01 ± 0.51 and theta‐high gamma: 0.04 ± 0.53) and TgAD rats (theta‐low gamma: 0.51 ± 0.26 and theta‐high gamma: −0.14 ± 1.02) had lower change in MI when compared to their CHOW‐fed counterparts (nTg | theta‐low gamma: 0.23 ± 0.54, p < 0.001 and theta‐high gamma: 0.25 ± 0.51, p < 0.001; TgAD | theta‐low gamma: 0.57 ± 0.73, p < 0.001 and theta‐high gamma: 0.50 ± 0.73, p < 0.001). Note that the change in MI of HCHF‐fed TgAD rats was lower compared to HCHF‐fed nTg rats and CHOW‐fed nTg rats.

Whereas the effects of diet on hippocampal MI were genotype‐dependent, they were genotype‐independent in the S1FL. HCHF‐fed rats (both nTg and TgAD) had lower S1FL MI (theta‐low gamma: by 19% in nTg and 10% in TgAD; theta‐high gamma: by 18% in nTg and 10% in TgAD) when compared to their CHOW‐fed counterparts (theta‐low gamma | CHOW‐nTg: −6.19 ± 0.74, HCHF‐nTg: −7.39 ± 0.72, p < 0.001; CHOW‐TgAD: −5.73 ± 0.80, HCHF‐TgAD: −6.3 ± 1.2, p < 0.001 and theta‐high gamma | CHOW‐nTg: −6.11 ± 0.67, HCHF‐nTg: −7.23 ± 0.61, p < 0.001; CHOW‐TgAD: −5.81 ± 0.80, HCHF‐TgAD: −6.4 ± 1.1, p < 0.001) (Figure S4B).

3.6. Rescue of hippocampal functional hyperemia with HCHF diet in symptomatic AD pathology

Hippocampal functional hyperemia was utilized to characterize the interactive effect of AD transgene and obesity on brain hemodynamics. HCHF diet differentially affected the hippocampal CBF change (diet‐genotype, p = 0.001) and volume of activation (diet‐genotype, p < 0.001) in response to bilateral forepaw electric stimulation of the nTg versus TgAD cohorts. On visual inspection, representative activation maps (Figure 5E) showed fewer activated voxels in HCHF‐fed nTg versus CHOW‐fed nTg but increased activation spread in HCHF‐fed TgAD rats when compared to the CHOW‐fed TgAD. Activation concordance maps (Figure 5F) demonstrated the subject‐wise consistency of activations in each cohort. The functional hyperemia (CBF change and spread of activation) during forepaw stimulation was quantified, as shown in Figure 5I,J. In the TgAD rats, the hippocampal CBF change (Figure 5I) and volume of activation (. J ) in response to forepaw stimulation were enhanced with HCHF diet (ΔCBF | CHOW: 7 ± 14 mL/100 g/min, HCHF: 60 ± 26 mL/100 g/min, p < 0.05; volume of activation | CHOW: 0.03 ± 0.08 mm³/mm³, HCHF: 0.25 ± 0.15 mm³/mm³, p < 0.001). The attenuated CBF change in response to stimulation of CHOW‐fed TgAD rats when compared to CHOW‐fed nTg rats and enhancement with HCHF diet was also observed globally (diet‐genotype, p = 0.020) and in the cortex (diet‐genotype, p = 0.005). However, a significant pairwise difference between the CBF response to stimulation between CHOW‐ and HCHF‐fed TgAD rats was detected only in the whole imaging slice (Figure S5). There was no significant effect of either diet or genotype on resting perfusion. On pairwise comparison, the hippocampal resting perfusion among the groups was indistinguishable (Figure 5H). However, segregating by sex, HCHF‐fed rats had lower resting CBF when compared to CHOW‐fed rats, mainly because of the high resting perfusion of CHOW‐fed male nTg rats (Figure S6) (diet‐sex, p < 0.001). Moreover, lower resting CBF globally (diet, p < 0.001) and in the cortex (diet, p < 0.001) were measured in HCHF‐fed cohorts when compared with CHOW‐fed cohorts (Figure S5). Overall, hippocampal functional hyperemia was sensitive to the effects of both the AD transgene and HCHF diet. TgAD rats showed reduced functional hyperemia, but HCHF feeding restored it to normal levels, matching those of CHOW‐fed nTg rats.

The correlation plot of hippocampal CBF change (on MRI) and power change (on electrophysiological recordings) in response to stimulation revealed that HCHF‐fed TgAD rats exhibited enhanced CBF and power changes compared to CHOW‐fed TgAD rats. Furthermore, HCHF‐fed TgAD rats showed functional hyperemia similar to CHOW‐fed nTg rats, but with greater power changes.

3.7. Attenuation of hippocampal capillary diameter of HCHF‐fed nTg rats

We investigated the structural changes in the cerebrovasculature that may have contributed to the difference in cerebrovascular reactivity. The brain's vascular bed is predominantly made up of brain capillaries; thus, representative images of hippocampal capillaries of CHOW‐ and HCHF‐fed nTg and TgAD 12‐month‐old rats are shown in Figure S7A, with capillary diameter quantification in Figure S7B. Hippocampal capillary diameter of 12‐month‐old CHOW‐fed nTg rats (6.62 ± 0.36 µm) was larger than that of all other groups, with the hippocampal capillary diameter of nTg and TgAD rats fed with HCHF was indistinguishable. Therefore, the morphological changes did not explain the aforementioned functional contrasts, whereas the observed amyloid‐ and HCHF‐induced constrictions were anticipated. ⁵⁶ , ⁵⁷ , ⁵⁸

4. DISCUSSION

This study evaluated the use of PCASL‐MRI for measuring hippocampal functional deficits. Assessing the hippocampal functional hyperemia with somatosensory stimuli is advantageous in both experimental models and in the clinical setting in that somatosensory stimulation can be easily administered and its processing is not contingent on the subject's level of attention or cooperativity. In contrast, memory tasks, conventionally employed to assess hippocampal functioning, can be influenced by a variety of factors beyond hippocampal function (e.g., attention ³² and level of effort ³³ ), especially in the elderly ³⁴ , ³⁵ , ³⁶ . In applying this assay in our experimental model of AD pathology comorbid with obesity, we found it sensitive to the AD transgene and obesity. In particular, TgAD rats had attenuated hippocampal functional hyperemia (CBF change and volume of activation) when compared to nTg. HCHF diet resolved hippocampal functional hyperemia impairment in TgAD rats, though it had no effect on capillary constriction. We also found that the AD pathology‐associated attenuation in neuronal power response was mitigated by exposure to the HCHF diet. With regard to neuronal network synchronization at rest, the HCHF diet increased the already elevated hippocampal theta‐low/high gamma MI in TgAD and attenuated the elevated theta‐low/high gamma MI changes to forepaw stimulation in TgAD. These findings suggest that the HCHF diet differentially affected AD and normal aging, conferring benefits to the metabolically challenged AD brain while being damaging to the healthy aging brain.

4.1. Functional hyperemia assessment in the hippocampus

To examine the hippocampal vascular functioning, we measured the blood flow during periods of increased neural demand induced by bilateral forepaw stimulation. In CHOW‐fed cohorts, TgF344‐AD rats had attenuated functional hyperemia (by 86% in ΔCBF magnitude and 90% in volume of activation) relative to that of nTg rats. The difference between genotypes demonstrated the sensitivity of hippocampal functional hyperemic assay, in contrast to the preserved functional hyperemia of the primary somatosensory cortex that we reported on in the 12‐month‐old TgF344‐AD rats. ³⁰ However, changes in other sensory regions not captured in our imaging protocol could potentially affect the hippocampal responses, highlighting the need for future studies to fully rule out their contribution. In 9‐month‐old Tg‐F344AD rats, ³⁸ we showed the detrimental effect of early AD pathology on the hippocampus. AD‐pathology elicited compromise in functional hyperemia has also been observed in the Tg2576 mouse model of AD ⁵⁹ , ⁶⁰ as well as in patients, ⁸ , ⁹ , ⁶¹ parallels the dysfunction of hippocampal functional hyperemia we observed in symptomatic AD pathology.

When fed with an HCHF diet, TgAD rats showed enhancement of both hippocampal ΔCBF, by 752%, and the volume of activation in response to bilateral forepaw stimulation, by 626%, compared to CHOW‐fed TgAD rats, bringing their functional hyperemia to that of the nTg CHOW cohort. Despite extensive literature by us and others showing detrimental effects of HCHF diet on accumulation of AD pathologies, ³⁷ , ⁶² , ⁶³ , ⁶⁴ we herein propose that HCHF may benefit the AD brain function as evidenced by the normalization of functional hyperemia. Notwithstanding, we emphasize that the observed benefits are likely transient, as supported by our earlier work ³⁰ showing a diminution of HCHF diet benefits with extended HCHF feeding period. We thus anticipate the beneficial effects described here to afford only a delay in AD‐related impairments in cerebral hemodynamics. Moreover, we expect the effects of HCHF administration to be AD stage‐dependent, with HCHF feeding prior to amyloidogenesis unlikely to confer any benefit, as we have described previously. ³⁰

The absence of AD transgene and diet effects on resting perfusion suggested that the baseline blood flow in the hippocampus was resilient. We observed that regardless of diet, TgAD rats had hippocampal resting CBF comparable to that of nTg rats. No CBF deficit was seen in APP/PS1 mice fed with a high‐fat diet either, regardless of exacerbated AD pathology. ⁶² HCHF‐fed cohorts (both nTg and TgAD rats) did show a trend of lower resting CBF when compared to CHOW‐fed nTg rats. Attenuated resting CBF is observed in adults with high body mass index (BMI) or obese individuals. ⁶⁵ , ⁶⁶ However, segregating by sex, lower resting CBF of CHOW‐fed TgAD when compared to CHOW‐fed nTg male rats was underlain by constricted capillary diameter, again consistent with data in mouse models of AD (e.g., APP^NL‐G‐F, APP^SWE/PS1ΔE9). ⁵⁶ , ⁵⁷ , ⁵⁸ In patients with AD, Mattsson et al and other groups demonstrated attenuation of hippocampal perfusion ⁶⁷ , ⁶⁸ accompanied by capillary constriction seen post mortem in AD. ⁵⁶ , ⁶⁸ , ⁶⁹ , ⁷⁰ Together, these findings suggest that microvascular reactivity impairments may accompany largely preserved baseline hippocampal perfusion.

4.2. Electrophysiological assessment of the hippocampus

To determine the role of neuronal function change, we assessed hippocampal LFPs and neuronal network synchronization (PAC). The engagement of the hippocampus during forepaw stimulation was seen through the increase in hippocampal gamma band power, while the decreased theta power during stimulation likely resulted from a shift in hippocampal state from memory‐encoding mode to more active processing. ⁷¹ In CHOW‐fed cohorts, TgAD rats exhibited a trend of attenuated resting alpha power and enhanced high‐gamma power, while the change in power during stimulation of TgAD rats was lower when compared to nTg rats. When fed with HCHF, TgAD rats had potentiated resting power as well as increased change in power during stimulation when compared to CHOW‐fed TgAD rats. In addition, HCHF‐fed TgAD rats also showed enhanced hippocampal alpha power, of particular significance given the established role of alpha band‐power in memory processing, ⁷² , ⁷³ with cognitive decline associated with decreasing alpha power. ⁷⁴ The HCHF also induced enhancement of theta power in TgAD rats at baseline and during stimulation, which was most likely beneficial, as hippocampal theta oscillations underlie memory formation, in both animals ⁷⁵ , ⁷⁶ , ⁷⁷ and humans; ⁷⁸ , ⁷⁹ and disruption of hippocampal theta oscillations impairs spatial memory while its potentiation promotes cognitive performance. ⁷⁷

Since the gamma band is involved in short‐term cognitive processes and memory task performance, ⁷⁶ , ⁸⁰ , ⁸¹ , ⁸² higher resting high‐gamma power of TgAD versus nTg rats might indicate a compensatory mechanism, in further support of our prior observations of endogenous cognitive resilience, hippocampal neuronal compensation, and upregulated GABAergic interneuronal markers in 12‐month‐old TgF344‐AD rats. ²⁷ Enhancement of gamma power is also observed in AD patients. ⁸³ In response to forepaw stimulation, TgF344‐AD versus nTg rats showed decreased low‐ and high‐gamma power responses, indicative of disrupted task processing in TgAD animals. HCHF feeding increased low‐ and high‐gamma power responses in both genotypes, but with larger effects in TgAD rats, resulting in a reversal of this contrast.

As PAC has been proposed to be a predictor of cognitive function, ⁵¹ , ⁵⁴ , ⁵⁵ we also examined cross‐frequency coupling between theta and low/high gamma bands. CHOW‐fed TgAD versus nTg rats had higher MI, accompanied by enhanced high‐gamma power. This increased theta‐gamma coupling in CHOW‐fed TgAD here could reflect compensatory mechanisms that oppose synaptic dysfunction and reduce theta power response to stimuli. Alterations in theta‐gamma coupling have been reported in both AD patients ⁸⁴ and AD models, including the TgF344AD, ⁸⁵ , ⁸⁶ , ⁸⁷ as well as APP‐KO mice, ⁸⁸ TgCRND8 mice, ⁸⁹ 5xFAD mice, ⁹⁰ and APP23 mice ⁹¹ ), but the direction of these changes has varied, likely due to differences in experimental paradigms. HCHF feeding increased baseline MI in TgAD rats compared to their CHOW‐fed TgAD littermates. In contrast, HCHF reduced baseline MI in nTg rats relative to CHOW‐fed nTg rats. The opposing effects of HCHF feeding on resting‐state theta‐gamma coupling in TgAD (increased MI) versus nTg (decreased MI) rats suggest that while the diet may support AD‐dysregulated brain metabolism, it has a detrimental impact on hippocampal function in healthy aging. Theta‐gamma coupling has been shown to be associated with hippocampal memory task performance and activity ⁹² in both rodents ⁹³ and humans. ⁵⁴ , ⁹⁴ In response to stimulation, both HCHF‐fed versus CHOW‐fed nTg and TgAD rats showed attenuated MI changes. For TgAD rats, this attenuation occurred despite their elevated baseline MI, suggesting that while HCHF increases resting‐state PAC, it may limit the dynamic range of neural adaptability during stimulation.

This work established an assay of hippocampal neurovascular function for studying the sequelae of AD and its vascular comorbidities that are of high translational potential, given the ease of delivering somatosensory stimuli and limited aging/neurodegeneration‐dependent effects on somatosensation. The attenuation of hippocampal functional hyperemia and local field potentials across different frequency bands (at baseline and during stimulation), along with their normalization with HCHF diet, indicates disrupted neurovascular coupling and neural network signaling in TgAD rats being ameliorated by HCHF feeding. The observed benefits of HCHF diet in symptomatic AD, but not in normal aging, are speculated to result from metabolically dysregulated AD brain suffering an energetic deficit and thus profiting from increased metabolite supply during high‐calorie diet feeding. However, the translation of our findings will require careful AD staging and mitigation of HCHF diet's obesogenic potential via alternative routes, including exercise and administration of metabolic precursors.

CONFLICT OF INTEREST STATEMENT

The authors report no competing interests.

Supporting information

Supporting Information

Supporting Inforamtion

Almanza DLV, Trevisiol A, Koletar MM, et al. High‐caloric intake rescues early symptomatic AD‐induced hippocampal neurovascular coupling deficits. Alzheimer's Dement. 2025;21:e70708. 10.1002/alz.70708