To probe alterations in cellular metabolism and structure in the triple transgenic APP/PS1/tauP301L (3xTg) mouse model of Alzheimer’s disease (AD), using MRS and diffusion-weighted MRS in vivo.

Experiments were performed on an 11.7 T Bruker scanner (Gmax=752 mT/m), equipped with a cryoprobe. MRS and DW-MRS were acquired using our new “STE-LASER” sequence¹, which consists in a diffusion-weighted stimulated echo block followed by a LASER localization block (TE/TM/TR=33.4/20/2000 ms). Four 18-months old females 3xTg mice and four control littermates (age and sex-matched) were scanned. A 28 mm³ voxel was positioned around the hippocampus (Fig. 1A). Twelve diffusion-weighted spectra were acquired between b=0 and 20 ms/µm² (which is the maximal value for which SNR allows individual scan phasing in a voxel of this size). Spectra were analyzed with LCModel. An experimentally measured macromolecule (MM) spectrum was included in the basis-set. Signal could be reliably quantified (CRLB<5%) for NAA, total creatine (tCr), choline compounds (tCho), myo-inositol (Ins), taurine (Tau). Due to the excellent spectral resolution, creatine (Cr) and phosphocreatine (PCr) were well resolved at 4.2 ppm and could be separately quantified (CRLB<5% at all b for PCr).

Metabolite concentration were determined from spectra at b=0, relative to tCr (assumed to be at 8 mM). Statistical significance was assessed using unpaired Student’s t-test. Diffusion data were first analyzed using bi-exponential fitting. To get an idea of structural parameters we also performed a fit in randomly oriented infinite cylinders, including cylinder diameter as free parameter². Because diffusion-derived parameters are more sensitive to experimental noise than concentrations, fits were performed on signal attenuations averaged over each group. Significance of the difference was assessed using standard Monte Carlo simulation (500 draws for each group) followed by permutation test on the group-merged Monte Carlo distributions (200,000 permutations of 2x4 parameter values).

Typical spectra are shown for control (Fig. 1B) and 3xTg mice (Fig. 1C), illustrating the striking metabolic remodeling. In the meantime, no atrophy was visible on images. Results are summarized in Fig. 2. Significant variations were observed for different metabolites. Overall, these variations exhibit some similarity with those reported in other mouse models of AD, in particular regarding the decreased NAA and increased Ins in APP/PS1 models (see ref.³ and references within). Results are very similar to those recently observed in a rat model of CNTF-induced astrocytic activation⁴, except for the stable tCho levels reported here (versus increased tCho with CNTF). A striking feature (which was not significant in the rat model of astrocytic activation) is the massive lactate increase, suggesting alteration of energy metabolism. Cr and PCr are rarely resolved in vivo, and to our knowledge this is the first time it is reported in an AD model. Interestingly we measure a close to significance (p=0.06) 15% increased PCr/Cr ratio in 3xTg (as opposed to the decreased ratio reported in the Q111 Huntington’s disease mouse model⁵).

Metabolite signal attenuation for a mouse of each group is shown in Fig. 3A and averaged logarithm signal attenuation is shown in Fig. 3B. Overall, attenuation curves are very similar in both groups. However, some metabolites seem to exhibit some slightly but systematically stronger attenuation in the 3xTg group, in particular for Ins, tCho, tCr and Tau. Diffusion parameters are given in Table 1. Since a relatively short diffusion time was used (t_d=23.2 ms), measured diffusion is mostly sensitive to short distance restriction, in particular fiber diameter². In this context, the increased fast and slow diffusivities (from bi-exponential fit) of Ins, an astrocytic marker (together with tCho, which is also often assumed to be mainly astrocytic, and whose diffusivities also increase close to significance), might reasonably be interpreted as increased astrocytic fiber diameter. This is consistent with the characteristic hypertrophy of activated astrocytes, exhibiting in particular larger fiber diameter, and is further supported by radii extracted from the cylinder fit, which in 3xTg significantly increases by ~10% for Ins (and almost significantly for tCho) (Table 1). tCr, which is supposed to be present both in astrocytes and neurons, and thus should sensitive to astrocytic alterations, also exhibits significantly larger cylinder radius.

The signature of activated, hypertrophic astrocytes (which are known to be present around amyloid plaques), as well as possible alterations of energy metabolism, are detected by MRS and diffusion-weighted MRS in the 3xTg mice hippocampus. Additional experiments are underway to consolidate these results.