Synopsis

Keywords: Alzheimer's Disease, fMRI (resting state), Preclinical

MRI is a unique tool to study the complexity of functional and structural communication in the brain. To explore the brain architecture of a mouse model of Alzheimer's Disease and highlight sex-dimorphism in the emergence of AD-like signs, we used a recent mouse model, the APP^NL-F/MAPT double knock-in (dKI). In preclinical imaging, we used resting-state graph theory approaches in a longitudinal study associated with behavioral evaluation. Functional connectivity of perirhinal, dorsal-hippocampus and midbrain nodes were implicated in early memory impairments in dKI female mice. Interestingly, perirhinal-cortex and dorsal-hippocampus are key regions for object-place associative memory and long-term object-recognition.

Introduction

Alzheimer’s disease (AD) is the most common cause of dementia with more than 47M cases worldwide(1). Its early phase is characterized by a slow and progressive decline in episodic and spatial memories. 2/3 of people affected by this disease are women. Preclinical research in mouse models of AD often report age-dependant cognitive deficits but sex dimorphism in early AD is largely unexplored(2). To characterize this dimorphism in the emergence of AD-like signs, we used a recent mouse model: the App^NL-F/MAPT double knock-in (dKI)(3). The KI of these humanized AD-related genes leads to the physiological expression of amyloid precursor protein (APP) bearing two familial AD mutations and the six isoforms of normal human Tau. Hence, the model is characterized by a slow and progressive evolution of the pathology(4). We investigated sex dimorphism (i) in subtle memory deficits overtime using behavioral tests(5) and (ii) through functional and structural connectivity mapping with resting-state functional MRI (rsfMRI) and diffusion tensor imaging (DTI). Our goal was to characterize cognitive performances and their relationship with brain networks configurations at an early phase of the pathology.

Methodology

Animals: The App^NL-F/MAPT dKI mice were bred on a C57BL6/J background. Experiments enrolled four groups of mice (n=17/group): dKI female and male groups and wild type (WT) female and males, imaged at 2 and 4-months each.
Behavioral tests: Spontaneous object exploration tests were used to evaluate subtle impairments in 24h long-term Object Recognition (RO24h) and 5min short-term Object in Place associative memory (OiP).
MRI data acquisition: Brain imaging was performed using a 7T small bore animal scanner and a ¹H four-channel phased array receive-only MRI mouse brain CryoProbe combined with a volume transmission coil 86mm diameter and the ParaVision 6.01 (Bruker, Germany). Anatomical images were acquired with a Turbo-RARE T₂-weighted sequence (TE/TR=20ms/6000ms; resolution=0.07x0.07x0.35mm³; 11min), rsfMRI images with a GE-EPI sequence (TE/TR=15ms/2137ms; resolution=0.14x0.18x0.35mm³; 500 volumes; 17min), and DTI using a DTI-EPI sequence (TE/TR=44ms/6000ms; resolution=0.15³; b=1000s/mm²; 40 diffusion directions; 43min). During fMRI acquisitions, mice were anesthetized by a combination of isoflurane (0.5%) and medetomidine (0.1mg.kg^-1 bolus sc)(6,7,8,9). DTI and anatomical data were collected under isoflurane anesthesia (1,5%).
Data analysis: (i) Behavioral data were analyzed through memory index with a mixed-effect model. (ii) RsfMRI data were corrected for intensity bias, inter-slice-drift and motion. A band-pass filter (0.01-0.1Hz) was applied. Brain images were spatially normalized to the Allen Brain Atlas and 40 ROIs covering major brain areas were selected. From each ROIs, the average BOLD time-course was extracted and functional connectivity (FC) matrices were generated using partial correlation. The matrices were Fisher transformed to obtain the z-scores associated to the connections’ weight between pairs of ROIs and used as inputs for graph theory measures(10). Intergroup statistical comparison of connectivity matrices (2-sample t-test, p<0.01, uncorr.) was performed to identify the most different connections. Machine learning logistic regression (using a sigmoid function) was used to reveal the nodes for which the connectivity features are most discriminative between groups.

Results

Behavioral phenotyping revealed deficits in early AD-sensitive recognition memory paradigms. In both RO24h and OiP tasks, only female dKI mice were significantly impaired at 2-months. At 4-months, both dKI males and females showed a deficit (both p<0.006)(Fig.2A,B). RsfMRI showed genotype and sex related FC modifications. In females, differences between WT and dKI involved connections with the perirhinal-cortex. Indeed, perirhinal-cortex changed its connectivity with frontal pole, orbital, entorhinal, infralimbic, primary somatosensory, piriform areas and septum (Fig.3A) in females dKI. Less differences were identified in males (WT vs. dKI t-test, p<0.01) and involved connections between: midbrain-anterior cingulate cortex, midbrain-retrosplenial, amygdala-dorsal hippocampus, gustatory cortex-ventral hippocampus and midbrain-amygdala (Fig.3B). Interestingly, the retrosplenial cortex was previously identified as a dysfunctional hub in dKI males submitted to the OiP task (11). Finally, females and males dKI mice showed differences in the strength of hippocampus-midbrain, perirhinal-ventral pallidum and pallidum-visceral connections (Fig.3C). Machine learning logistic regression pointed out the connectivity of perirhinal-cortex, dorsal hippocampus and midbrain nodes (score >0.7) as most discriminative between females and males dKI; suggesting that these areas are implicated in the occurrence of early memory impairments in dKI female mice. Particularly, the perirhinal-cortex and dorsal hippocampus are key regions for object-place associative memory and long-term OR (Fig.3D).

Discussion

We combined here brain FC mapping and recognition memory paradigms in the App^NL-F/MAPT dKI mouse model of AD to examine comparatively in male and female mice the early stages of the pathology. Our results indicate a female vulnerability; indeed, only dKI females showed deficits in recognition memory test at 2-months, whereas males only showed these two deficits two months later. RsfMRI data revealed differences between WT and dKI females in perirhinal-entorhinal and perirhinal-orbital connections, known to be recruited in OiP, RO24h and targets for AD pathology. Moreover, these data point out perirhinal, dorsal hippocampus and midbrain areas as potential players in sex specific memory impairments in dKI mice at 2-months; and therefore, possible vulnerable areas in females dKI. Overall, the dKI mouse model showed good promises in the characterization of the female vulnerability at early AD pathology. Further sex specific alterations in memory networks are currently being analyzed using functional (seed-based analysis, ICA(12)), structural(13) (tensor parametric maps, tractography(14), and voxel based-morphometry(15)) MRI at 4 and 12-months.

Acknowledgements

No acknowledgement found.