Diffusion MRI Detects Basal Forebrain Cholinergic Abnormalities in the 3xTg-AD Mouse Model of Alzheimer’s Disease

Abstract

Degeneration of the basal forebrain (BF) is detected early in the course of Alzheimer’s disease (AD). Reduction in the number of BF cholinergic (ChAT) neurons associated with age-related hippocampal cholinergic neuritic dystrophy is described in the 3xTg-AD mouse model; however, no prior diffusion MRI (dMRI) study has explored the presence of BF alterations in this model. Here we investigated the ability of diffusion MRI (dMRI) to detect abnormalities in BF microstructure for the 3xTg-AD mouse model, along with related pathology in the hippocampus (HP) and white matter (WM) tracks comprising the septo-hippocampal pathway. 3xTg-AD and normal control (NC) mice were imaged in vivo using the specific dMRI technique known as diffusional kurtosis imaging (DKI) at 2, 8, and 15 months of age, and 8 dMRI parameters were measured at each time point. Our results revealed significant lower dMRI values in the BF of 2 months-old 3xTg-AD mice compared with NC mice, most likely related to the increased number of ChAT neurons seen in this AD mouse model at this age. They also showed significant age-related dMRI changes in the BF of both groups between 2 and 8 months of age, mainly a decrease in fractional anisotropy and axial diffusivity, and an increase in radial kurtosis. These dMRI changes in the BF may be reflecting the complex aging and pathological microstructural changes described in this region. Group differences and age-related changes were also observed in the HP, fimbria (Fi) and fornix (Fx). In the HP, diffusivity values were significantly higher in the 2 months-old 3xTg-AD mice, and the HP of NC mice showed a significant increase in axial kurtosis after 8 months, reflecting a normal pattern of increased fiber density complexity, which was not seen in the 3xTg-AD mice. In the Fi, mean and radial diffusivity values were significantly higher, and fractional anisotropy, radial kurtosis and kurtosis fractional anisotropy were significantly lower in the 2 months-old 3xTg-AD mice. The age trajectories for both NC and TG mice in the Fi and Fx were similar between 2 and 8 months, but after 8 months there was a significant decrease in diffusivity metrics associated with an increase in kurtosis metrics in the 3xTg-AD mice. These later HP, Fi and Fx dMRI changes probably reflect the growing number of dystrophic neurites and AD pathology progression in the HP, accompanied by WM disruption in the septo-hippocampal pathway. Our results demonstrate that dMRI can detect early cytoarchitectural abnormalities in the BF, as well as related aging and neurodegenerative changes in the HP, Fi and Fx of the 3xTg-AD mice. Since DKI is widely available on clinical scanners, these results also support the potential of the considered dMRI parameters as in vivo biomarkers for AD disease progression.

1. Introduction

The basal forebrain (BF) region comprises of the ventral basal ganglia, nucleus basalis, substantia innominate, the diagonal band of Broca (vertical (VDB) and horizontal (HDB)) and the medial septal nucleus (MS) [1]. This highly complex brain region contains a heterogeneous mixture of neuron types that differ in morphology, neurotransmitter content and projection pattern; they are aggregated and non-aggregated cholinergic, GABAergic, glutamatergic and various calcium binding protein containing neurons [1]. The cholinergic basal forebrain (CBF) provides the major cholinergic innervation to the cerebral cortex and hippocampus, and the cholinergic and GABAergic neurons at the MS/VDB complex projects to the hippocampus and receive projections back through the fimbria-fornix bundles, forming the septo-hippocampal pathway [2,3].

The CBF system is involved in the cognitive processing of memory and attention through inputs to the cortex and hippocampus [4–6], and it is known to undergo neurodegenerative alterations during normal aging and in Alzheimer’s disease (AD) [7,8]. Indeed, degeneration of the CBF with loss of cholinergic innervation to the cortex and hippocampus substantially contribute to cognitive decline in AD, as described in the cholinergic hypothesis for the AD pathogenesis [9,10]. Some studies even demonstrate the presence of BF cholinergic neuron abnormalities very early in aging and in the course of AD [11]. Additionally, several human studies have shown that atrophy of the basal forebrain, as measured by a reduced volume in structural MRI, precedes and predicts clinical decline in mild cognitive impairment and even in individuals who have not yet been diagnosed with cognitive impairment [12–16].

The CBF system has been evaluated in biochemical, morphological and imaging studies of mouse models of AD with inconsistent findings [17–21]. Although most mouse models of AD over-expressing mutant APP and/or PS1 genes display an age-dependent Aβ deposition, cortical and hippocampal cholinergic fiber degeneration and memory deficits, some studies report that these changes are not accompanied by a loss of cholinergic BF neurons [18,19]. Other studies, however, have reported BF structural changes in different transgenic mouse models of AD. In one study using structural MRI, Badea et al. [22] reported in the APPswe/ind mice a significant interaction between age and genotype in a variety of brain regions including the MS. In another study, using both structural and diffusion MRI (dMRI), Badea et al. [23] reported in the CVN-AD mice volume decrease for several brain regions including a brain circuit that comprises of the MS, hippocampus, fimbria and fornix. Recently, it has been reported in the 5xFAD mouse model that the septo-hippocampal pathway begins to degenerate before the onset of cognitive impairment (in 4.5-month-old mice) [24]. Furthermore, in this model, significant cholinergic neuronal loss of the MS has been demonstrated in the 6-month-old mice, a time when these mice present with cognitive dysfunction [25].

The triple transgenic mouse model of AD (3xTg-AD) is one of the most studied mouse models of AD. This model possesses the three mutations of human presenilin-1 M146V, human amyloid precursor protein Swedish mutation, and the P301L mutation of human tau, based on human mutations identified in familial AD [26]. It develops both Aβ and neurofibrillary tangles (NFTs) in a temporal and spatial pattern similar to human AD pathology [27–29]. These mice exhibit synaptic plasticity dysfunction, including long-term potentiation (LTP) deficits that occur prior to extracellular Aβ deposition, having cognitive impairment around 3–4 months of age. These observations in 3xTg-AD mice show that physiological and behavioral changes are present at an early age, even before amyloid and tau-related pathologies are observed. In this transgenic model, Perez et al. [30] reported an age-related hippocampal and cortical cholinergic neuritic dystrophy in parallel with the progression of AD pathology. They also reported an age-related reduction in the number of cholinergic neurons in the MS/VDB of these mice. Notably, this age-related reduction in the number of CBF neurons was apparently due to the greater number of MS/VDB cholinergic neurons found in the young 3xTg-AD mice.

In prior work, we showed that diffusional kurtosis imaging (DKI) is able to detect brain morphological abnormalities in the hippocampus, fimbria and fornix of young 2-month-old 3xTg-AD mice [31]. DKI is an extension of diffusion tensor imaging (DTI) that includes dMRI data acquired with stronger diffusion weightings than what is typical for DTI. In this way, DKI provides all the diffusion measures available with DTI plus additional measures related to the kurtosis of the water diffusion displacement probability density. These kurtosis measures quantify the extent to which water diffusion is non-Gaussian and are closely associated with the degree of tissue microstructural complexity [32–34]. Therefore, the goal of this study was to investigate the sensitivity of DTI and DKI in capturing the age and AD pathology induced microstructure alterations in the BF and hippocampus, as well as in the white matter (WM) tracks comprising the septo-hippocampal pathway.

2. Methods

2.1. AD mouse model

All experimental procedures were approved by the Institutional Animal Care and Use Committee at the Medical University of South Carolina (MUSC), and they were conducted in accordance with the National Institutes of Health (NIH) Guide for Care and Use of Laboratory Animals. All animals were purchased from The Jackson Laboratory. The 3xTg-AD (TG) [B6;129-Psen1tm1Mpm Tg(APPSwe, tauP301L)1Lfa/Mmjax; MMRRC Stock No: 34830-JAX|3xTg-AD] and age-matched controls (NC) [101045 B6129SF2/J] were studied at 2 (NC, n=30; TG, n=49), 8 (NC, n=17; TG, n=28) and 15 (NC, n=9; TG, n=12) months of age. The control mice are the recommended controls for genetically engineered strains generated with 129-derived embryonic stem cells and maintained on a mixed B6;129 background. All mice were housed in temperature- and humidity-controlled rooms on a 12-h light/dark cycle (lights on at 6:00 AM) in an accredited animal care facility.

2.2. MRI acquisition

Acquisition was performed as previously reported [31]. Briefly, mice were anesthetized using an isoflurane vaporizer set at the following percentages: 3% for induction, 2% during pilot scanning and data acquisition. After induction, mice were placed on a mouse holder and restrained using a mouse tooth bar (Bruker, T10146) and ear bars (Bruker, T10147) placed in the auditory canal. Compressed air was used as the carrier gas and delivered at a flow rate of 1L/min into a nose cone positioned around the tooth bar, where gases mixed with air and passed over the rodent’s nose. All animals were maintained at 37.0 ± 0.2 °C and respiration ranged between 60 and 80 breaths per minute during scanning.

The in vivo MRI experiments were all performed on a 7T Bruker Biospin 30 cm bore scanner (BioSpec 70/30 USR) running Paravision version 5.1. A 86 mm 1H quadrature volume coil (T128038) was used for signal transmission and an actively decoupled phase array coil (four channel receiver T11765) was used for signal reception. A 2-shot SE-EPI dMRI sequence was employed for the DKI acquisition [32,33]. Imaging parameters were: TR/TE = 3750/34.4 ms, δ/Δ = 5/18ms, slice thickness = 0.7 mm, 15 slices with no gap, data matrix = 128×128, image resolution = 156×156 μm², 2 signal acquisitions, 10 b-value = 0 (b0) images, followed by 30 diffusion encoding gradient directions with 4 b-values for each gradient direction (0.5, 1.0, 1.5, and 2.0 ms/μm²) and fat suppression flip angle = 105°. Total acquisition time = 33 min.

2.3. DKI post-processing and Image analysis

DKI post-processing was performed using DKE [35] (http://nitrc.org/projects/dke). Post-processing included motion correction by aligning all DWIs to the first b=0 image using SPM12 (Wellcome Trust Centre for Neuroimaging, UCL, UK). A principal component analysis approach [36] and the method of moments [37] were used to reduce the effects of signal noise on the dMRI data. Parametric maps were obtained by fitting dMRI signal measurements to the DKI signal model for each voxel using a linearly constrained weighted linear least squares fitting algorithm, which generated the diffusion and kurtosis tensors. These two tensors were then used to calculate parametric maps for all diffusion metrics [33,34]. Figure 1 illustrates (A) the MRI slice selection, (B) a representative fractional anisotropy (FA) maps of a 2-month-old TG mouse exemplifying the selected ROIs, and (C) all diffusion metrics maps from both NC and TG mouse. In our analysis, we considered the four diffusion tensor (DT) parameters of mean diffusivity (MD), axial diffusivity (D_∥), radial diffusivity (D_⊥), and fractional anisotropy (FA). MD corresponds to the diffusivity averaged over all diffusion directions, D_∥ corresponds to the diffusivity in the direction of the principal diffusion tensor eigenvector, D_⊥ corresponds to the diffusivity averaged over all diffusion directions perpendicular to the principal diffusion tensor eigenvector, and FA quantifies the anisotropy of the diffusion tensor. We similarly considered the four diffusional kurtosis (DK) metrics of mean kurtosis (MK), axial kurtosis (K_∥), radial kurtosis(K_⊥), and kurtosis fractional anisotropy (KFA). These are kurtosis analogs of the DT metrics that quantify diffusional non-Gaussianity and provide complementary information about the diffusion dynamics [34,38]. All four b-values (0.5, 1.0, 1.5, 2.0 ms/μm²), for each gradient direction, were used in calculating the diffusion metrics. It is worth noting that, due to the inclusion of non-Gaussian effects, DKI yields more accurate estimates of DT metrics than does conventional DTI [39], as well as enabling a more comprehensive assessment of the diffusion microenvironment in brain tissue.

Figure 1.

(A) Slice positioning for the 15 coronal slices. (B) Representative fractional anisotropy maps of a 2-month-old 3xTg-AD mouse illustrating the selected ROIs: basal forebrain (BF-red), fornix (Fx-blue), fimbria (Fi-orange) and hippocampus (HP-green). (C) Representative dMRI parametric maps of all diffusion metrics, for a single anatomical slice, from a normal control (NC) and a 3xTg-AD mouse. DT measures are mean diffusivity (MD), axial diffusivity (D_∥), radial diffusivity (D_⊥), and fractional anisotropy (FA). DK measures are mean kurtosis (MK), axial kurtosis (K_∥), radial kurtosis (K_⊥), and kurtosis fractional anisotropy (KFA). Scale bars: 0–1 for FA and KFA; 0–2 μm²/ms for MD, D_∥ and D_⊥; 0–3 for MK, K_∥, and K_⊥.

Regions of interest (ROIs) (Figure 1-B) were defined for basal forebrain (BF – one slice) and hippocampus (HP - four slices, including dorsal and ventral areas). Additionally, we assessed fimbria (Fi - three slices) and fornix (Fx - one slice), which are WM tracks relevant to age-related and AD pathology changes, comprising the septo-hippocampal pathway. All ROIs were manually drawn on the averaged b0 image by a neuropathologist (MFF), using ImageJ (http://imagej.nih.gov/ij//) [40]. Anatomical guidelines for outlining these regions were determined by comparing anatomical structures in the MRI slices with a standard mouse atlas [41] and verified with the FA maps to ensure correct anatomical location and to avoid contamination of unintended tissue or cerebrospinal fluid (CSF). To reduce bias when drawing the ROIs, we randomly chose mice ID numbers, which included both NC and TG mice, and sampled the same number of MRI slices for each mouse for each ROI. The regional values for the dMRI measures were obtained by averaging all voxels within an ROI except those with MD > 1.5 μm²/ms, which were excluded to minimize the effect of CSF contamination.

2.4. Statistical analyses

Summary statistics were expressed as group-averaged means ± standard error of the mean (SE). Two-tailed, two-sample t-tests were performed to assess group differences for the dMRI in each ROI. The Benjamini-Hochberg false discovery rate method [42] was used to adjust for multiple testing within each ROI. We also reported the effect size (Cohen’s d) for each dMRI metric when comparing 3×Tg-AD and NC mice.

For the longitudinal data, piecewise linear mixed effects models [43] were used to estimate the slopes for each group between 2–8 and 8–15 months of age. Three covariance structures were considered when accounting for the fact that measurements were made repeatedly over time within subjects: unstructured (empirically-based), compound symmetry (equal correlation over time), and spatial power (decreasing correlation over time), with the final selection for each ROI and dMRI metrics made using Bayesian Information Criterion fit statistics [44]. Model estimate p-values were adjusted for multiple testing using the Benjamini-Hochberg false discovery rate method⁴². An adjusted p-value of ≤0.05 was considered to be statistically significant. Analyses were conducted using SAS v9.4 (SAS Institute, Cary, NC).

3. Results

3.1. Group analysis

Estimates of the diffusion metrics (mean ± standard error of the mean (SE) in each ROI for each time point (2, 8 and 15 months of age) are presented in Tables 1–3 and Figures 2–5. For each brain ROI several dMRI metrics showed significant differences between the two groups at 2 and 8 months of age. In the BF, dMRI metrics values for 2-month-old TG mice were statistically different from the NC mice for D_∥, FA, MK and K_⊥. At 8 months, dMRI metrics values for the TG mice were statistically different from the NC for D_∥, D_⊥, FA and KFA. In the HP, dMRI metrics values for the 2-month-old TG mice were statistically different from the NC mice for MD, D_∥, D_⊥ and FA. At 8 months, the dMRI metrics values for the TG mice were statistically different from the NC mice for MD, D_∥, D_⊥, FA and K_⊥. In the Fi, dMRI metrics values for the 2 and 8 month-old TG mice were statistically different from the NC for MD, D_⊥, FA, K_⊥ and KFA. In the Fx, only K_⊥ in the TG mice was statistically different from the NC mice at 2 months. At 8 months, there was non-significant trends for increased MD, D_⊥ and decreased of KFA in the TG mice. No statistical difference between the two groups were observed in 15-month-old mice in any ROI, for any dMRI metric.

Table 1.

Diffusion metric estimates (mean ± standard error of the mean (SE)) in each brain region of interest for the 2 months old mice. For group comparison, a two-tailed, two-sample t-test with significance at the level of p≤ 0.05 (in bold) after Benjamini-Hochberg false discovery rate was applied to correct for multiple comparisons (8 dMRI metrics compared within each ROI). Mean diffusivity (MD); axial diffusivity (D_∥); radial diffusivity (D_⊥); fraction anisotropy (FA); mean kurtosis (MK); axial kurtosis (K_∥); radial kurtosis (K_⊥); kurtosis fraction anisotropy (KFA); normal control (NC) and 3xTG-AD (TG) groups; basal forebrain (BF), hippocampus (HP), fimbria (Fi) and fornix (Fx).

	MD μm2/ms	D∥ μm2/ms	D⊥ μm2/ms	FA	MK	K∥	K⊥	KFA
BF
NC	0.82 ± 0.004	1.17 ± 0.006	0.65 ± 0.004	0.36 ± 0.004	1.01 ± 0.006	0.97 ± 0.009	1.12 ± 0.012	0.47 ± 0.006
TG	0.81 ± 0.003	1.12 ± 0.005	0.65 ± 0.003	0.34 ± 0.003	0.98 ± 0.008	0.94 ± 0.009	1.05 ± 0.011	0.45 ± 0.006
P-values	0.051	<.0001	0.106	<.0001	0.007	0.073	0.0001	0.071
Cohen’s d	−0.51	−1.30	0.38	−1.33	−0.63	0.07	−1.00	−0.45
HP
NC	0.79 ± 0.004	0.88 ± 0.005	0.74 ± 0.003	0.12 ± 0.001	0.77 ± 0.007	0.78 ± 0.007	0.78 ± 0.007	0.32 ± 0.004
TG	0.80 ± 0.003	0.91 ± 0.005	0.75 ± 0.002	0.13 ± 0.002	0.76 ± 0.004	0.76 ± 0.004	0.78 ± 0.005	0.32 ± 0.002
P-values	<.0001	0.003	0.047	0.003	0.369	0.134	0.999	0.999
Cohen’s d	1.00	0.81	0.53	0.72	−0.25	−0.41	0.03	−0.0001
Fi
NC	0.97 ± 0.007	1.74 ± 0.009	0.63 ± 0.008	0.59 ± 0.005	1.11 ± 0.008	0.78 ± 0.003	1.78 ± 0.025	0.60 ± 0.006
TG	1.01 ± 0.005	1.75 ± 0.007	0.58 ± 0.003	0.56 ± 0.003	1.10 ± 0.005	0.79 ± 0.003	1.70 ± 0.013	0.56 ± 0.003
P-values	<.0001	0.452	<.0001	<.0001	0.452	0.452	0.013	<.0001
Cohen’s d	1.00	0.19	1.29	−1.34	−0.21	0.17	−0.71	−1.40
Fx
NC	0.95 ± 0.011	1.46 ± 0.009	0.70 ± 0.015	0.45 ± 0.009	1.06 ± 0.009	0.83 ± 0.009	1.40 ± 0.011	0.46 ± 0.010
TG	0.95 ± 0.006	1.46 ± 0.007	0.70 ± 0.009	0.44 ± 0.006	1.03 ± 0.008	0.82 ± 0.007	1.34 ± 0.012	0.46 ± 0.008
P-values	0.952	0.952	0.952	0.952	0.097	0.770	0.021	0.998
Cohen’s d	0.05	−0.06	0.07	−0.14	−0.53	−0.25	−0.72	0.0005

Table 3.

Diffusion metric estimates (mean ± standard error of the mean (SD)) in each brain region of interest for the 8 months old mice. For group comparison, a two-tailed, two-sample t-test with significance at the level of p≤ 0.05 (in bold) after Benjamini-Hochberg false discovery rate was applied to correct for multiple comparisons (8 dMRI metrics compared within each ROI). Mean diffusivity (MD); axial diffusivity (D_∥); radial diffusivity (D_⊥); fraction anisotropy (FA); mean kurtosis (MK); axial kurtosis (K_∥); radial kurtosis (K_⊥); kurtosis fraction anisotropy (KFA); normal control (NC) and 3xTG-AD (TG) groups; basal forebrain (BF), hippocampus (HP), fimbria (Fi) and fornix (Fx).

	MD μm2/ms	D∥ μm2/ms	D⊥ μm2/ms	FA	MK	K∥	K⊥	KFA
BF
NC	0.80 ± 0.011	1.09 ± 0.014	0.65 ± 0.011	0.32 ± 0.009	1.09 ± 0.018	1.02 ± 0.012	1.17 ± 0.022	0.43 ± 0.012
TG	0.80 ± 0.010	1.07 ± 0.012	0.66 ± 0.011	0.31 ± 0.005	1.05 ± 0.028	0.99 ± 0.016	1.13 ± 0.037	0.41 ± 0.016
P-values	0.849	0.502	0.502	0.426	0.502	0.426	0.502	0.555
Cohen’s d	0.08	−0.48	0.40	−0.75	−0.52	−0.79	−.040	−0.31
HP
NC	0.79 ± 0.013	0.90 ± 0.013	0.74 ± 0.013	0.13 ± 0.004	0.79 ± 0.011	0.82 ± 0.015	0.80 ± 0.011	0.35 ± 0.023
TG	0.79 ± 0.005	0.90 ± 0.006	0.73 ± 0.005	0.14 ± 0.003	0.78 ± 0.017	0.79 ± 0.014	0.80 ± 0.020	0.34 ± 0.012
P-values	0.959	0.959	0.959	0.959	0.959	0.856	0.959	0.959
Cohen’s d	−0.07	0.14	−0.19	0.43	−0.22	−0.75	0.02	−0.15
Fi
NC	1.01 ± 0.016	1.81 ± 0.029	0.62 ± 0.018	0.59 ± 0.012	1.15 ± 0.023	0.76 ± 0.012	1.88 ± 0.078	0.56 ± 0.011
TG	0.99 ± 0.016	1.78 ± 0.016	0.59 ± 0.017	0.58 ± 0.008	1.19 ± 0.014	0.78 ± 0.009	1.94 ± 0.040	0.57 ± 0.012
P-values	0.529	0.529	0.529	0.529	0.529	0.529	0.529	0.695
Cohen’s d	−0.46	−0.38	−0.40	0.35	0.65	0.50	0.33	0.17
Fx
NC	0.89 ± 0.013	1.46 ± 0.019	0.60 ± 0.016	0.51 ± 0.013	1.11 ± 0.025	0.85 ± 0.023	1.52 ± 0.040	0.53 ± 0.012
TG	0.88 ± 0.014	1.41 ± 0.017	0.61 ± 0.016	0.49 ± 0.011	1.15 ± 0.017	0.89 ± 0.012	1.54 ± 0.036	0.51 ± 0.013
P-values	0.695	0.328	0.695	0.422	0.422	0.328	0.695	0.525
Cohen’s d	−0.23	−0.85	0.17	−0.57	0.61	0.81	0.19	−0.44

Figure 2.

Basal Forebrain diffusion tensor (DT) and diffusional kurtosis (DK) metrics (mean ± standard error) and the slopes for each mouse group between 2–8 and 8–15 months. DT measures are mean diffusivity (MD), axial diffusivity (D_∥), radial diffusivity (D_⊥), and fractional anisotropy (FA). DK measures are mean kurtosis (MK), axial kurtosis (K_∥), radial kurtosis (K_⊥), and kurtosis fractional anisotropy (KFA). Normal control (NC - green) and 3xTG-AD (TG - red). Time points with statistically significant group differences are indicated by a light blue dotted oval shape. Between successive time points, group-specific slopes that are significantly positive (or negative) are represented by a (★) in the color of each mouse group.

Figure 5.

Fornix diffusion tensor (DT) and diffusional kurtosis (DK) metrics (mean ± standard error) and the slopes for each mouse group between months 2–8 and months 8–15. DT measures are mean diffusivity (MD), axial diffusivity (D_∥), radial diffusivity (D_⊥), and fractional anisotropy (FA). DK measures are mean kurtosis (MK), axial kurtosis (K_∥), radial kurtosis (K_⊥), and kurtosis fractional anisotropy (KFA). Normal control (NC - green) and 3xTG-AD (TG - red) groups. Time points with statistical significant group differences are indicated by a light blue dotted oval shape. Between successive time points, group-specific slopes that are significantly positive (or negative) are represented by a (★) in the color of each mouse group.

3.2. Age-related trajectories

Estimates of NC and TG mice slopes for the time period between 2–8 and 8–15 months are presented in Tables 4–5 and in Figures 2–5. In the BF (Tables 4–5 – Figure 2), there were significant age-related changes between 2 and 8 months for several dMRI metrics in both groups, but no changes were present between 8 and 15 months of age. In the HP (Table 4–5 – Figure 3), only FA increased significantly between 2 and 8 months in the TG group, and K_∥ increased significantly between 8 and 15 months in the NC group. In the Fi (Table 4–5 – Figure 4) and Fx (Table 4–5 – Figure 5), there were significant age-related changes for several dMRI metrics in both groups. In the NC group, these changes occurred only between 2 and 8 months of age, but for the TG group these changes were significant both between 2 and 8 and 8 and 15 months, sometimes in opposite directions, e.g. for MD, D_∥, K_∥ and KFA in the Fi and for D_∥ in the Fx.

Table 4.

Diffusion tensor (DT) and diffusional kurtosis (DK) metrics slope estimates in each brain region for the period of 2 to 8 months of age. Normal control (NC) and 3xTG-AD (TG) groups; S: slope estimates; SE: standard error of the slope; mean diffusivity (MD); axial diffusivity (D_∥); radial diffusivity (D_⊥); fraction anisotropy (FA); mean kurtosis (MK); axial kurtosis (K_∥); radial kurtosis (K_⊥); kurtosis fraction anisotropy (KFA); p-values at the level of p≤ 0.05 (in bold) after Benjamini-Hochberg false discovery rate applied to correct for multiple comparisons (8 dMRI metrics compared within each ROI).

BF	S	SE	P-values		S	SE	P-values
MD μm2/ms				MK
Slope NC	−0.00211	0.001258	0.1964	Slope NC	0.003877	0.002703	0.2504
Slope TG	0.000421	0.000982	0.6694	Slope TG	0.01079	0.002109	<.0001
D∥ μm2/ms				K ∥
Slope NC	−0.00704	0.002022	0.0037	Slope NC	0.006126	0.002518	0.0477
Slope TG	−0.00432	0.001578	0.0092	Slope TG	0.008235	0.001965	0.0002
D⊥ μm2/ms				K ⊥
Slope NC	0.000239	0.001117	0.8312	Slope NC	−0.00412	0.0042	0.4306
Slope TG	0.002708	0.000871	0.0038	Slope TG	0.01179	0.003279	0.001
FA				KFA
Slope NC	−0.00307	0.000836	0.0031	Slope NC	−0.00153	0.001679	0.4192
Slope TG	−0.00369	0.000654	<.0001	Slope TG	−0.00524	0.00131	0.0003
HP	S	SE	P-values		S	SE	P-values
MD μm2/ms				MK
Slope NC	−0.00058	0.001055	0.8529	Slope NC	−0.0005	0.001601	0.8661
Slope TG	0.000612	0.000824	0.6138	Slope TG	0.002793	0.001249	0.1013
D∥ μm2/ms				K ∥
Slope NC	0.000206	0.001459	0.8881	Slope NC	−0.00063	0.001651	0.8661
Slope TG	0.001786	0.001141	0.1944	Slope TG	0.00186	0.001289	0.3077
D⊥ μm2/ms				K ⊥
Slope NC	−0.00098	0.000892	0.7329	Slope NC	−0.00069	0.001798	0.8661
Slope TG	−4.83E-6	0.000697	0.9945	Slope TG	0.002975	0.001403	0.1013
FA				KFA
Slope NC	0.000938	0.000521	0.5666	Slope NC	0.002045	0.000978	0.1931
Slope TG	0.001325	0.000407	0.0145	Slope TG	0.000143	0.000763	0.9732
Fi	S	SE	P-values		S	SE	P-values
MD μm2/ms				MK
Slope NC	0.008025	0.001772	<.0001	Slope NC	0.006069	0.001857	0.0028
Slope TG	0.00666	0.001383	<.0001	Slope TG	0.005122	0.00145	0.0013
D∥ μm2/ms				K ∥
Slope NC	0.01672	0.002665	<.0001	Slope NC	−0.0057	0.001044	<.0001
Slope TG	0.01554	0.00208	<.0001	Slope TG	−0.00674	0.000815	<.0001
D⊥ μm2/ms				K ⊥
Slope NC	0.003672	0.001727	0.0428	Slope NC	0.02136	0.005481	0.0005
Slope TG	0.002351	0.00135	0.0867	Slope TG	0.01676	0.00428	0.0005
FA				KFA
Slope NC	0.001382	0.001065	0.1992	Slope NC	−0.00351	0.001271	0.01
Slope TG	0.001638	0.000831	0.0609	Slope TG	−0.00333	0.000993	0.0018
Fx	S	SE	P-values		S	SE	P-values
MD μm2/ms				MK
Slope NC	−0.00943	0.002231	0.0002	Slope NC	0.000249	0.002831	0.9721
Slope TG	−0.00492	0.001741	0.0085	Slope TG	0.009508	0.00221	0.0002
D∥ μm2/ms				K ∥
Slope NC	−0.00019	0.002489	0.9382	Slope NC	0.000375	0.002168	0.9721
Slope TG	0.003209	0.001943	0.1036	Slope TG	0.004499	0.001693	0.0114
D⊥ μm2/ms				K ⊥
Slope NC	−0.01418	0.002801	<.0001	Slope NC	0.009921	0.004312	0.0396
Slope TG	−0.00919	0.002187	0.0002	Slope TG	0.02239	0.003365	<.0001
FA				KFA
Slope NC	0.009179	0.001603	<.0001	Slope NC	0.01148	0.002441	<.0001
Slope TG	0.007258	0.001253	<.0001	Slope TG	0.006112	0.001905	0.0034

Table 5.

Diffusion tensor (DT) and diffusional kurtosis (DK) metrics slope estimates in each brain region for the period of 2 to 8 months of age. Normal control (NC) and 3xTG-AD (TG) groups; S: slope estimates; SE: standard error of the slope; mean diffusivity (MD); axial diffusivity (D_∥); radial diffusivity (D_⊥); fraction anisotropy (FA); mean kurtosis (MK); axial kurtosis (K_∥); radial kurtosis (K_⊥); kurtosis fraction anisotropy (KFA); p-values at the level of p≤ 0.05 (in bold) after Benjamini-Hochberg false discovery rate applied to correct for multiple comparisons (8 dMRI metrics compared within each ROI).

BF	S	SE	P-values		S	SE	P-values
MD μm2/ms				MK
Slope NC	−0.0014	0.001458	0.3881	Slope NC	0.008032	0.003201	0.059
Slope TG	−0.00166	0.001219	0.7148	Slope TG	0.00075	0.002681	0.892
D∥ μm2/ms				K ∥
Slope NC	−0.00513	0.002344	0.0672	Slope NC	0.003224	0.002993	0.3684
Slope TG	−0.00356	0.001961	0.5959	Slope TG	−0.00022	0.002507	0.9295
D⊥ μm2/ms				K ⊥
Slope NC	0.000546	0.001328	0.6825	Slope NC	0.01051	0.004836	0.0672
Slope TG	−0.00063	0.001112	0.892	Slope TG	0.001296	0.004039	0.8561
FA				KFA
Slope NC	−0.00291	0.001032	0.0519	Slope NC	−0.00416	0.002015	0.0672
Slope TG	−0.0007	0.000865	0.892	Slope TG	−0.00062	0.001689	0.892
HP	S	SE	P-values		S	SE	P-values
MD μm2/ms				MK
Slope NC	0.000842	0.001435	0.6386	Slope NC	0.003739	0.001911	0.1465
Slope TG	−0.00267	0.001228	0.1145	Slope TG	0.000606	0.001601	0.7063
D∥ μm2/ms				K ∥
Slope NC	0.001718	0.001494	0.5071	Slope NC	0.006666	0.001965	0.0097
Slope TG	−0.00294	0.001275	0.1145	Slope TG	0.001657	0.001646	0.5086
D⊥ μm2/ms				K ⊥
Slope NC	0.000413	0.001431	0.7734	Slope NC	0.002172	0.002148	0.4357
Slope TG	−0.00252	0.001223	0.1145	Slope TG	−0.00083	0.001799	0.7063
FA				KFA
Slope NC	0.001208	0.000606	0.1395	Slope NC	0.002359	0.002709	0.6184
Slope TG	0.000339	0.000508	0.6747	Slope TG	0.002554	0.002322	0.5334
Fi	S	SE	P-values		S	SE	P-values
MD μm2/ms				MK
Slope NC	−0.0011	0.002073	0.7793	Slope NC	0.001712	0.002173	0.6942
Slope TG	−0.00869	0.001735	<.0001	Slope TG	0.008592	0.001818	<.0001
D∥ μm2/ms				K ∥
Slope NC	−0.00487	0.003117	0.492	Slope NC	0.002071	0.001221	0.492
Slope TG	−0.00979	0.00261	0.0006	Slope TG	0.005322	0.001023	<.0001
D⊥ μm2/ms				K ⊥
Slope NC	0.000724	0.002128	0.735	Slope NC	−0.00258	0.006661	0.735
Slope TG	−0.00769	0.001785	<.0001	Slope TG	0.02007	0.005584	0.001
FA				KFA
Slope NC	−0.00169	0.00129	0.5181	Slope NC	−0.0017	0.001545	0.5484
Slope TG	0.003641	0.001082	0.0015	Slope TG	0.004142	0.001295	0.0022
Fx	S	SE	P-values		S	SE	P-values
MD μm2/ms				MK
Slope NC	−0.00074	0.002606	0.9196	Slope NC	0.005345	0.003279	0.4328
Slope TG	−0.00611	0.002182	0.0181	Slope TG	0.007301	0.002743	0.0198
D∥ μm2/ms				K ∥
Slope NC	0.00047	0.002974	0.8749	Slope NC	0.003187	0.002536	0.5693
Slope TG	−0.0087	0.002491	0.0069	Slope TG	0.00714	0.002122	0.0053
D⊥ μm2/ms				K ⊥
Slope NC	−0.00087	0.003278	0.9196	Slope NC	0.008947	0.005019	0.4328
Slope TG	−0.0041	0.002744	0.1864	Slope TG	0.00951	0.004201	0.0433
FA				KFA
Slope NC	0.000464	0.001869	0.9196	Slope NC	−0.00086	0.002854	0.9196
Slope TG	0.000078	0.001563	0.9606	Slope TG	0.000904	0.002389	0.8073

Figure 3.

Hippocampus diffusion tensor (DT) and diffusional kurtosis (DK) metrics (mean ± standard error) and the slopes for each mouse group between 2–8 and 8–15 months. DT measures are mean diffusivity (MD), axial diffusivity (D_∥), radial diffusivity (D_⊥), and fractional anisotropy (FA). DK measures are mean kurtosis (MK), axial kurtosis (K_∥), radial kurtosis (K_⊥), and kurtosis fractional anisotropy (KFA). Normal control (NC - green) and 3xTG-AD (TG - red). Time points with statistically significant group differences are indicated by a light blue dotted oval shape. Between successive time points, group-specific slopes that are significantly positive (or negative) are represented by a (★) in the color of each mouse group.

Figure 4.

Fimbria diffusion tensor (DT) and diffusional kurtosis (DK) metrics (mean ± standard error) and the slopes for each mouse group between months 2–8 and months 8–15. DT measures are mean diffusivity (MD), axial diffusivity (D_∥), radial diffusivity (D_⊥), and fractional anisotropy (FA). DK measures are mean kurtosis (MK), axial kurtosis (K_∥), radial kurtosis (K_⊥), and kurtosis fractional anisotropy (KFA). Normal control (NC - green) and 3xTG-AD (TG - red) groups. Time points with statistically significant group differences are indicated by a light blue dotted oval shape. Between successive time points, group-specific slopes that are significantly positive (or negative) are represented by a (★) in the color of each mouse group.

4. Discussion

Our results demonstrate that dMRI is able to detect significant group (NC vs.TG) and age-associated differences in the BF, HP, Fi and Fx of both NC and TG mice. The interpretation of changes in dMRI metrics is complex, particularly when taking in account changes that occur during normal brain maturation [45–47], including axonal pruning and myelination, as well as changes that occur during the aging process [48–50] and with pathological neurodegeneration [51].

In the BF, the 2 months-old TG mice exhibited statistically significant lower values compared with NC mice for most of the dMRI metrics. Since DT metrics tends to decrease with diffusion restrictions and DK metrics tends to decrease with diffusional homogeneity, these results may be related to the increased number of cholinergic (ChAT) neurons in the BF of the TG mice at this age [18]. This increase in cellularity creates more diffusion barriers and restrictions, leading to lower DT metrics, and at the same time, a greater proportion of these larger cells (ChAT neurons) occupying this region creates a more homogenous dMRI environment, leading to lower DK metrics. Also, ChAT neurons axons are thinner, thus an increased number of this type of axons can also contribute to restricted diffusion [52]. In 8 month-old mice, changes in the dMRI metrics were generally similar to what we observed at 2 months, but no group differences were detected for the DK metrics, except for a significant lower KFA in TG mice. For the DT metrics, D_∥ and FA were significant lower and D_⊥ was significant higher in the TG mice. These dMRI changes may be reflecting a combination of structural integrity abnormalities detected in young TG mice, and the appearance of pathology related changes in this model at this age. Starting around 5 months, there is a progressive reduction in the number of ChAT neurons in the BF of the TG mice, which will return to similar cell numbers in aged TG mice compared with NC mice [18]. When taking in consideration the age-related changes in the BF, we observed that most of the significant changes were evident only between 2 and 8 months of age. These age-related dMRI changes may be reflecting the complex age-related alterations described in the BF, with diffuse and homogeneous depletion of ChAT neurons [53], more intense and significant in aged TG mice [18]. Although our study lacks a direct morphological correlation, we speculate based on Perez at al. [18] study, which described the BF morphological changes with age in TG mice. The decrease of ChAT neurons overtime, with loss of axons creates a more isotropic distribution of axons which is reflected in the decrease FA and KFA overtime. It also establishes a more heterogeneous environment, with progressive axonal fragmentation and increase of the diffusion barriers, which is reflected in the decrease of D_∥ and in the increase of MK and K_∥ overtime in both groups. There was no significant age-associated change for D_⊥ and K_⊥ in the NC group between 2 and 8 months of age, different from what we observed in the TG group. This may be reflecting the accentuated loss of ChAT unmyelinated axons leading to a higher proportion of GABAergic myelinated axons in this region, which stabilized after 8 months of age in the TG mice. It is noteworthy to point out that there was a trend towards increased MK, K_∥ and K_⊥ between 8 and 15 months for the NC group, while the TG group did not change; although it did not reach significance after adjusting for multiple comparisons, it suggests an increase in the complexity and heterogeneity of the BF in the NC mice, not apparent in the TG mice.

DT and DK metrics group differences were also detected in the HP, Fi and Fx of 2 months-old mice. In the HP, all DT metrics were significantly higher in the TG mice; in the Fi, MD and D_⊥ were significantly higher and K_⊥, FA, and KFA were significantly lower in the TG mice; in the Fx, only K_⊥ was significantly lower in the TG mice compared to NC mice. These results are in agreement with our previous report in 2-month-old TG mice [31] and may be reflecting the previously described significant volumetric and structural integrity differences [54,55] and abnormal myelination patterns in the hippocampal complex and Fi-Fx tracks of young TG mice [56,57]. These dMRI parameter differences continue to be present at 8 months, a time where swollen cholinergic dystrophic neurites are seen in the hippocampal-septal cholinergic axonal projections [18], which increases in an age-dependent manner, in parallel with the progression of Aβ plaque and tau pathology in this model [13]. The dMRI group differences observed in the Fi and Fx may be related to a progressive loss of fibers and decrease in the density of myelinated axons in the WM tracks that forms the septo-hippocampal pathway. No group differences were detected at 15 months, but this may be due to the small sample size at this age which reduced the statistical power. Since BF provides the major cholinergic innervation to the hippocampus and receive projections back through the fimbria-fornix bundles [2], our dMRI results also indicate a relationship between the morphological changes present in the BF and HP, and the WM disruption in the septo-hippocampal pathway, even at a young age. Changes in the age-related trajectories were also observed in the Fi and Fx, but scarcely in the HP. In the HP, there was trends towards increase in MK and KFA and a significant increase in K_∥ after 8 months in NC mice, reflecting a normal pattern of increased fiber density complexity, which was not seen in the TG mice. It is important to point out that aged TG mice have confounding factors related to neuroinflammation and the development of AD pathology in the HP [58,59], which may have masked the detection of age-associated dMRI changes in this region. Only future studies taking all these aspects into consideration and with a morphological correlation can definitively establish the biological interpretation for the HP dMRI changes reported in our study. In the Fi and Fx, the age trajectories for both NC and TG mice had similar progression between 2 and 8 months, except for D_∥ and K_∥, but after 8 months there was a significant decrease in DT metrics associated with an increase in KT metrics in the TG group. The similar dMRI metrics trajectories before 8 months are probably reflecting the progressive decrease in the BF cholinergic axons, resulting in fiber density decrease. This Fi results are in agreement with our previous study in which we studied only 4 TG mice longitudinally, reporting dMRI changes in the Fi with similar trajectory patterns [60]. The HP and the BF have reciprocal pathways; the HP receives both cholinergic and GABAergic projections from cells located in the MS/DB complex through the fimbria-fornix complex. By acting as the major output tract of the hippocampus, the fimbria-fornix fibers play an important role in the formation and consolidation of memories, and they have been shown to be vulnerable to aging as well as AD pathology [23] [61–65]. Thus, the decrease in DT metrics associated with an increase in KT metrics in the Fi and Fx of the TG mice after 8 months of age probably is a secondary effect of the increase in swollen dystrophic neurites and AD pathology progression present in the hippocampus.

This is the first longitudinal dMRI study to report age and AD pathology alterations in the BF, HP and two WM tracks (Fi and Fx) comprising the septo-hippocampal pathway. Little is known about the BF dMRI patterns in normal or transgenic mice. Only two previous studies, both in fixed mouse brain, described changes in the integrity of the BF cholinergic system detected by DTI. Kerbler et al. [66] reported a significant increase in FA, MD, D_∥ and D_⊥ in adult mice with ~25% loss of BF ChAT neurons compared to control animals. Their results are difficult to compare with ours since, unlike in transgenic mouse models of AD, the cholinergic degeneration was caused by the injection of the toxin saporin conjugated to an antibody to a receptor expressed by BF cholinergic neurons (p75 neurotrophin receptor). This approach has no influence from other AD pathological features, and it is a onetime point experiment, with mice being imaged only fourteen days after the injection that causes loss of the BF ChAT neurons. They also reported, using probabilistic tractography analysis of the septo-hippocampal tracts originating from the BF, an increase in streamline MD and D_⊥ in lesioned mice. In the second paper, Hara et al. [21] examined the BF-hippocampal track in a tau mouse model of AD. They found that the streamlines connecting BF and hippocampus, which comprises the Fi-Fx track, were decrease in old (22 months) compared with adult mice (6 months), and these changes were associated with a decrease of cholinergic neurons. They also reported a decrease in FA at the level of the BF in old mice, with no changes in D_∥ and D_⊥. In spite of substantial methodological differences between our study and theirs, we also detected an age-related decrease in FA at the BF.

There is no previous dMRI study assessing changes in the BF in this transgenic AD mouse model. In fact, few dMRI studies [67,68] have investigated any type of brain abnormalities in this model, including our two previous DKI studies where we reported longitudinal WM microstructure changes [60] and brain abnormalities in 2-month-old TG mice [31]. Also, two structural MRI studies have reported significant volumetric differences in the hippocampus and several WM regions, including fimbria, in young TG mice [54,55]. Therefore, this study is the first to use dMRI metrics to investigate early and progressive AD pathology related morphological alterations in the BF, hippocampus, and associated WM tracks.

Our study has two important limitations. First, the relatively small number of mice at later age (15 months old), which reduced the statistical power of the data analysis. Second, our study lacks a direct correlation with morphological data. It should also be noted that some ROIs, particularly Fi and Fx, are adjacent to the ventricles, which makes them susceptible to partial volume effects of CSF. However, DK metrics are known to be less sensitive to CSF contamination [69] and we also mitigated this problem by masking all voxels with MD >1.5 μm²/ms. Thus, these results should be cautiously interpreted. Further studies are needed to confirm the underlying biological mechanisms associated with the group differences and age-associated changes detected by dMRI in the BF, HP, Fi and Fx of both NC and TG mice.

5. Conclusion:

The present in vivo dMRI study of a mouse model of AD (3xTg-AD) demonstrates that dMRI can detect early abnormalities in the BF, HP, Fi and Fx, as well as detect the neurodegenerative process present in this AD mouse model. Specifically, several of the considered dMRI measures are sensitive to changes in brain cytoarchitecture and are promising candidates as in vivo biomarkers for monitoring age-related and disease progression. Since the dMRI methods employed here are easily translatable to clinical MRI, this study also demonstrates their potential for detection of brain abnormalities in human AD patients.

Table 2.

Diffusion metric estimates (mean ± standard error of the mean (SE)) in each brain region of interest for the 8 months old mice. For group comparison, a two-tailed, two-sample t-test with significance at the level of p≤ 0.05 (in bold) after Benjamini-Hochberg false discovery rate was applied to correct for multiple comparisons (8 dMRI metrics compared within each ROI). Mean diffusivity (MD); axial diffusivity (D_∥); radial diffusivity (D_⊥); fraction anisotropy (FA); mean kurtosis (MK); axial kurtosis (K_∥); radial kurtosis (K_⊥); kurtosis fraction anisotropy (KFA); normal control (NC) and 3xTG-AD (TG) groups; basal forebrain (BF), hippocampus (HP), fimbria (Fi) and fornix (Fx).

	MD μm2/ms	D∥ μm2/ms	D⊥ μm2/ms	FA	MK	K∥	K⊥	KFA
BF
NC	0.81 ± 0.005	1.12 ± 0.011	0.65 ± 0.004	0.34 ± 0.005	1.04 ± 0.010	1.00 ± 0.007	1.10 ± 0.020	0.46 ± 0.008
TG	0.81 ± 0.005	1.09 ± 0.008	0.67 ± 0.004	0.31 ± 0.004	1.05 ± 0.011	0.99 ± 0.010	1.12 ± 0.013	0.42 ± 0.006
P-values	0.542	0.042	0.004	0.0002	0.542	0.542	0.542	0.0007
Cohen’s d	0.189	−0.73	1.05	−1.44	0.22	−0.23	0.25	−1.26
HP
NC	0.79 ± 0.004	0.88 ± 0.005	0.74 ± 0.004	0.13 ± 0.002	0.77 ± 0.008	0.77 ± 0.007	0.78 ± 0.009	0.33 ± 0.006
TG	0.81 ± 0.003	0.92 ± 0.004	0.75 ± 0.003	0.14 ± 0.001	0.78 ± 0.005	0.77 ± 0.005	0.80 ± 0.006	0.32 ± 0.004
P-values	0.0002	<.0001	0.006	0.0002	0.182	0.806	0.034	0.117
Cohen’s d	1.00	1.83	0.97	1.41	0.44	0.07	0.73	−0.54
Fi
NC	1.02 ± 0.011	1.84 ± 0.012	0.61 ± 0.012	0.60 ± 0.006	1.15 ± 0.012	0.75 ± 0.004	1.91 ± 0.034	0.58 ± 0.008
TG	1.05 ± 0.007	1.85 ± 0.012	0.65 ± 0.007	0.58 ± 0.004	1.13 ± 0.008	0.75 ± 0.005	1.81 ± 0.021	0.54 ± 0.005
P-values	0.028	0.807	0.003	0.001	0.442	0.615	0.014	0.001
Cohen’s d	0.76	0.07	1.05	−1.19	−0.30	−0.19	−0.87	−1.26
Fx
NC	0.90 ± 0.008	1.46 ± 0.013	0.61 ± 0.013	0.51 ± 0.009	1.07 ± 0.011	0.83 ± 0.014	1.45 ± 0.021	0.53 ± 0.011
TG	0.92 ± 0.007	1.48 ± 0.011	0.65 ± 0.009	0.48 ± 0.007	1.09 ± 0.009	0.85 ± 0.008	1.47 ± 0.015	0.50 ± 0.008
P-values	0.053	0.423	0.053	0.132	0.132	0.423	0.423	0.053
Cohen’s d	0.81	0.27	0.75	−0.56	0.54	0.29	0.25	−0.74