Elsevier

NeuroImage

Volume 172, 15 May 2018, Pages 357-368
NeuroImage

Deciphering the microstructure of hippocampal subfields with in vivo DTI and NODDI: Applications to experimental multiple sclerosis

https://doi.org/10.1016/j.neuroimage.2018.01.061Get rights and content

Highlights

  • NODDI can delineate the internal anatomy of the mouse hippocampus in vivo.

  • Quantitative NODDI and DTI data can be collected in vivo in a single hippocampal layer.

  • AD and MD correlate with dendritic damage in the molecular layer of EAE mice.

  • NODDI data fail to capture dendritic damages in the molecular layer of EAE mice.

  • DTI may be more sensitive than NODDI in detecting early changes in the hippocampal layers.

Abstract

The hippocampus contains distinct populations of neurons organized into separate anatomical subfields and layers with differential vulnerability to pathological mechanisms. The ability of in vivo neuroimaging to pinpoint regional vulnerability is especially important for better understanding of hippocampal pathology at the early stage of neurodegenerative disorders and for monitoring future therapeutic strategies. This is the case for instance in multiple sclerosis whose neurodegenerative component can affect the hippocampus from the early stage. We challenged the capacity of two models, i.e. the classical diffusion tensor imaging (DTI) model and the neurite orientation dispersion and density imaging (NODDI) model, to compute quantitative diffusion MRI that could capture microstructural alterations in the individual hippocampal layers of experimental-autoimmune encephalomyelitis (EAE) mice, the animal model of multiple sclerosis. To achieve this, the hippocampal anatomy of a healthy mouse brain was first explored ex vivo with high resolution DTI and NODDI. Then, 18 EAE mice and 18 control mice were explored 20 days after immunization with in vivo diffusion MRI prior to sacrifice for the histological quantification of neurites and glial markers in each hippocampal layer. Fractional anisotropy (FA), axial diffusivity (AD), radial diffusivity (RD) and mean diffusivity (MD) maps were computed from the DTI model while the orientation dispersion index (ODI), the neurite density index (NDI) and the volume fraction of isotropic diffusivity (isoVF) maps were computed from the NODDI model. We first showed in control mice that color-coded FA and ODI maps can delineate three main hippocampal layers. The quantification of FA, AD, RD, MD, ODI, NDI and isoVF presented differences within these 3 layers, especially within the molecular layer of the dentate gyrus which displayed a specific signature based on a combination of AD (or MD), ODI and NDI. Then, the comparison between EAE and control mice showed a decrease of AD (p = 0.036) and of MD (p = 0.033) selectively within the molecular layer of EAE mice while NODDI indices did not present any difference between EAE and control mice in any layer. Histological analyses confirmed the differential vulnerability of the molecular layer of EAE mice that exhibited decreased dendritic length and decreased dendritic complexity together with activated microglia. Dendritic length and intersections within the molecular layer were independent contributors to the observed decrease of AD (R2 = 0.37 and R2 = 0.40, p < 0.0001) and MD (R2 = 0.41 and R2 = 0.42, p < 0.0001). We therefore identified that NODDI maps can help to highlight the internal microanatomy of the hippocampus but NODDI still presents limitations in grey matter as it failed to capture selective dendritic alterations occurring at early stages of a neurodegenerative disease such as multiple sclerosis, whereas DTI maps were significantly altered.

Introduction

The hippocampus is involved in declarative memory and hippocampal damage has been correlated to memory decline in several neurological disorders (Kandel et al., 2014). The hippocampus is not a unitary and homogeneous entity but a complex and highly connected archeocortical structure. Indeed, it is composed of distinct hippocampal subfields and layers which are very different in their morphological, molecular, electrophysiological and functional profiles (Kandel et al., 2014). Disorders affecting the hippocampus, such as Alzheimer, schizophrenia, post-traumatic stress disorder, hypoxia and multiple sclerosis among others, can start by differentially targeting one specific hippocampal subfield, depending on the underlying pathological process: a principle known as differential vulnerability (Small, 2014). Mapping such vulnerability and capturing subtle histological abnormalities in a particular subfield at the early stage of the disease is highly desirable in order to develop and monitor future neuroprotective strategies.

Diffusion tensor imaging (DTI), the most commonly used model of water diffusion in tissue (Le Bihan, 2013), could be able to pinpoint the hippocampal subfield most affected early in the course of the disease. DTI is based on one Gaussian model of water diffusion for both intra and extra-cellular compartments in order to reflect brain microstructure through numeric indices called fractional anisotropy (FA), axial diffusivity (AD, water diffusion along tracts), radial diffusivity (RD, water diffusion perpendicular to tracts) and mean diffusivity (MD) (Le Bihan, 2013). In rodents, DTI can individualize the layered hippocampal anatomy (Laitinen et al., 2010, Shepherd et al., 2006, Zhang et al., 2002). Furthermore, it has been recently demonstrated that DTI is able to capture early dendritic alterations selectively present within the molecular layer of the dentate gyrus and responsible for memory deficit prior to measurable hippocampal atrophy at the early stage of experimental multiple sclerosis (Planche et al., 2017).

Although DTI could be a promising early in vivo marker of hippocampal differential vulnerability in multiple sclerosis (and probably other conditions), DTI is the simplest model of water molecule diffusion that does not take into account restricted and hindered diffusion. The link between tissue architecture and DTI parameters is therefore indirect and DTI parameters lack specificity as they can be altered similarly by several diseases. Furthermore, DTI measurements are hardly analyzable in areas of partial-volume, crossing, kissing or fanning axons and dendrites as they estimate FA and diffusivity indexes poorly.

Consequently, an important field of research in diffusion magnetic resonance imaging (MRI) consists in developing alternative model-based strategies that describe the cerebral microstructure more accurately. One of these strategies is NODDI (for Neurite Orientation Dispersion and Density Imaging), which has proved promising because the acquisition protocol for this type of model is typically applicable in clinical settings (Zhang et al., 2012). The model is built with three diffusion-compartments: (i) the cerebrospinal fluid (CSF) compartment with Gaussian isotropic diffusion; (ii) the intra neurite compartment, which is modeled as a set of cylinders of zero radius to capture highly restricted diffusion perpendicular to the neurites and unhindered diffusion along them, and which takes into account numerous possibilities of orientation dispersion; and (iii) the extra neurite compartment, which refers to the space occupied by somas, glial cells and the space around neurites and which is modeled with Gaussian anisotropic diffusion.

NODDI provides volume fraction of isotropic diffusivity (isoVF), neurite density index (NDI) and orientation dispersion index (ODI) maps which could reflect the morphology of axons and dendrites and their branching complexity. NDI and ODI might disentangle factors contributing to DTI parameters (Zhang et al., 2012) and become more specific surrogate biomarkers of neurological diseases. NODDI has already been applied to patients with multiple sclerosis presenting altered ODI and NDI in normal appearing white matter (Schneider et al., 2017), and to the whole hippocampus of an Alzheimer's mouse model (Colgan et al., 2016). However, NODDI has never been used to precisely analyze the hippocampus micro-anatomy at the level of the subfields and layers in which diseases can start. Furthermore, NODDI has been built to be clinically feasible and thus relies on assumptions (fixed diffusivities for each compartment and Watson distribution of the orientations of fiber segments) that might not hold true in pathological conditions.

Thus, we wanted to test whether or not NODDI could be a step forward compared to DTI and provide additional and more specific information on early hippocampal microstructural changes responsible for cognitive impairment. To test this hypothesis, we developed a high-resolution diffusion MRI acquisition procedure capable of revealing hippocampal architecture, first ex vivo, and then in live mice in an acceptable timeframe, in order to evaluate the capacity of DTI and NODDI to identify the hippocampal subfield microstructure. Then, we assessed the capacity of these two techniques to pinpoint early regional hippocampal vulnerability in mice with experimental multiple sclerosis (experimental autoimmune encephalomyelitis, EAE) by performing MRI-histological correlations with a large panel of immunofluorescence staining for neurites, glial cells and myelin.

Section snippets

Animals and Experimental Autoimmune Encephalomyelitis (EAE)

We used EAE which is the most widely accepted animal model of multiple sclerosis (t Hart et al., 2011). Briefly, 7-to-9 week-old female C57BL6/J mice (Janvier Labs) were injected subcutaneously at the base of the tail with 200 μg of Myelin Oligodendrocyte Glycoprotein peptide 35-55 (MOG35-55, Anaspec) emulsified in 200 μL of Complete Freund's Adjuvant (CFA, Difco) containing 6 mg/mL of desiccated Mycobacterium Tuberculosis (H37Ra, Difco). Animals received intraperitoneal injections of Pertussis

Immunostaining

To allow MRI-to-histology correlations, animals were sacrificed immediately following MRI. Mice were deeply anesthetized with pentobarbital and perfused transcardially with PBS containing 2% PFA and 0.2% picric acid, for 20 min (7–8 mL/min). The brain was then removed and transferred into a Tris-buffered saline (TBS) solution containing 30% sucrose for cryoprotection and 0.05% sodium azide for conservation and left at 4 °C until use. A 0.8 mm block containing the dorsal hippocampus (i.e; from

Statistical analyses

Data are presented as mean ± SEM. The Gaussian distribution of the data was tested with the Shapiro-Wilk normality test. First, we looked at the hippocampal internal anatomy and we compared MRI data (DTI and NODDI parameters) between the 3 main hippocampal layers in control mice (CFA) using ANOVA (and Tukey's multiple comparisons test for post-hoc multiple comparisons) or the nonparametric Friedman test (and Dunn's multiple comparison test) as appropriate. We also tested whether each layer

Results

Our optimized protocol was successfully implemented on the 7 T and 4.7 T MR-systems. For in vivo acquisitions, 39 diffusion-weighted images distributed among 5 mice (1.59% of the total number of diffusion-weighted images collected from the 36 mice) were scored as artifacted data during the quality control procedure, due to movement, blurring and ghosting effects. Consequently, these 5 mice were excluded from the analysis conducted on the 31 remaining mice (15 control CFA and 16 EAE mice) with

Discussion

In this study, we challenged the capacity of the NODDI model to pinpoint the differential vulnerability of a specific hippocampal layer at the early stage of a neurodegenerative disease such as multiple sclerosis. We showed that high-resolution ODI mapping could highlight each of the three main hippocampal layers and that the quantitative signature of an individual layer can be identified by combining DTI and NODDI indices. However, at the early stage of experimental multiple sclerosis, NODDI

Funding

This work was supported by public grants from the French Agence Nationale de la Recherche within the context of the Investments for the Future program referenced ANR-10-LABX-57 named TRAIL (project GM-COG), ANR-10-LABX-43 named BRAIN (Project MEMO-MS) and ARSEP foundation “Fondation pour l'aide à la recherche sur la sclérose en plaques”. The 4.7 T animal scanner was supported by FLI (ANR-11-INBS-0006). AC received a grant in the framework of the Master 2 Bioscience of the Ecole Normale

Acknowledgements

The microscopy experiments were performed in the Bordeaux Imaging Center, a unit of the CNRS-INSERM and Bordeaux University, and member of the national France BioImaging infrastructure. We gratefully thank Nicolas Renaud (INCIA, CNRS UMR5287) for his important technical support with MRI post processing.

References (45)

  • B. Lampinen et al.

    Neurite density imaging versus imaging of microscopic anisotropy in diffusion MRI: a model comparison using spherical tensor encoding

    Neuroimage

    (2017)
  • N. Mikuni et al.

    Postnatal expressions of non-phosphorylated and phosphorylated neurofilament proteins in the rat hippocampus and the Timm-stained mossy fiber pathway

    Brain Res.

    (1998)
  • G. Nair et al.

    Myelination and long diffusion times alter diffusion-tensor-imaging contrast in myelin-deficient shiverer mice

    Neuroimage

    (2005)
  • V. Planche et al.

    Selective dentate gyrus disruption causes memory impairment at the early stage of experimental multiple sclerosis

    Brain Behav. Immun.

    (2017)
  • T.M. Shepherd et al.

    Structural insights from high-resolution diffusion tensor imaging and tractography of the isolated rat hippocampus

    Neuroimage

    (2006)
  • S.A. Small

    Isolating pathogenic mechanisms embedded within the hippocampal circuit through regional vulnerability

    Neuron

    (2014)
  • S.K. Song et al.

    Demyelination increases radial diffusivity in corpus callosum of mouse brain

    Neuroimage

    (2005)
  • S.W. Sun et al.

    Selective vulnerability of cerebral white matter in a murine model of multiple sclerosis detected using diffusion tensor imaging

    Neurobiol. Dis.

    (2007)
  • B.A. t Hart et al.

    EAE: imperfect but useful models of multiple sclerosis

    Trends Mol. Med.

    (2011)
  • M. Tariq et al.

    Bingham-NODDI: mapping anisotropic orientation dispersion of neurites using diffusion MRI

    Neuroimage

    (2016)
  • C.F. Westin et al.

    Q-space trajectory imaging for multidimensional diffusion MRI of the human brain

    Neuroimage

    (2016)
  • G.P. Winston et al.

    Advanced diffusion imaging sequences could aid assessing patients with focal cortical dysplasia and epilepsy

    Epilepsy Res.

    (2014)
  • Cited by (0)

    1

    Dr Hiba and Prof. Tourdias jointly directed this work and share co-senior authorship.

    View full text