High-resolution MRI of early-stage mouse embryos

Abstract

Both the availability of methods to manipulate genes and the completion of the mouse genome sequence have led to the generation of thousands of genetically modified mouse lines that provide a new platform for the study of mammalian development and developmental diseases. Phenotyping of mouse embryos has traditionally been performed on fixed embryos by the use of ex vivo histological, optical and high-resolution MRI techniques. Although potentially powerful, longitudinal imaging of individual animals is difficult or impossible with conventional optical methods because of the inaccessibility of mouse embryos inside the maternal uterus. To address this problem, we present a method of imaging the mouse embryo from stages as early as embryonic day (E)10.5, close to the onset of organogenesis in most physiological systems. This method uses a self-gated MRI protocol, combined with image registration, to obtain whole-embryo high-resolution (100 µm isotropic) three-dimensional images. Using this approach, we demonstrate high contrast in the cerebral vasculature, limbs, spine and central nervous system without the use of contrast agents. These results indicate the potential of MRI for the longitudinal imaging of developing mouse embryos in utero and for future applications in analyzing mutant mouse phenotypes.

Fig. 1

Simulation of error propagation due to data exclusion and registration inaccuracies. a) Simulated periodic displacement of the abdomen during maternal respiration at 35 bpm (left). Typical reconstructed image after rigid body transformations and exclusion of 18.5% of k-space lines contaminated by breathing (middle). The final image was recovered with registration (right). b) At 70 bpm the exclusion of 37% of k-space results in artifacts in the phase encoding direction. The registered image still shows artifacts and inaccuracies. c) The applied translations can be recovered accurately at 0 bpm and 35 pbm. At 70 bpm registration errors are superimposed on the aliasing artefacts due to undersampling.

Fig. 2

Comparison of reconstructed 3D images with and without gating and co-registration. A close to midsagittal view of an E14.5 embryo, with anterior (A) and posterior (P) marked. a) Reconstructed image without any respiratory gating with obvious motion artifacts. b) Alleviation of motion artifacts from respiration with rejection of k-space lines contaminated by motion, allowing the visualization of several structures such as spinal cord (SC), third ventricle (3V), choroid plexus (CP), hind limb (HL), and facial features (FF). c) Sharper resolution, especially of vascular features, such as the basilar artery (BA), obtained by co-registration of all the acquired volumes recovered with the use of a rigid-body registration step. d) Respiratory motion characterization with SG signals. Data above the threshold line are excluded from the reconstruction.

Fig. 3

Mid-sagittal views of E10.5 (a), E12.5 (b) and E14.5 (c) embryos that show the difference in contrast between the different types of tissue. For each stage a schematic inset is presented to show the approximate position of the 2D transverse slices taken from the 3D images. Abbreviations: aqueduct (Aq), choroid plexus (CP), fourth ventricle (4V), facial features (FF), hypothalamus (HTh), lateral ventricle (LV), medulla oblongada (Med), mesencephalic vesicle (Mes), mid-hind brain junction (MHB), neopallial cortex (NC), optic recess (OR), thalamus (Th), third ventricle (3V).

Fig. 4

a) Segmentation of the ventricles at E10.5, E12.5 and E14.5. The dashed line represents the lower limit of the head that was used for volumetric calculations. b) A chart showing the volume of the head of the embryos together with the volume of the ventricles. The error bars represent the standard deviation between different embryos. c) The ventricle-to-head volume ratio decreases with increasing developmental stage.

Fig. 5

a) 3D MIPs show the developing vasculature. Individual embryos staged at E10.5, E12.5 and E14.5 are viewed dorsally (top panels) and from the side (bottom panels) showing the increase in vascular pattern complexity at each developmental stage. b) MIPs highlighting the cerebral vasculature of the developing mouse brain that is increasing in complexity from E10.5 (top) to E14.5 (bottom). At E10.5 the jugular vein (JV), and the venous sinus (VS) can be distinguished. At E12.5 and E14.5 additional arteries and veins can also be identified. Abbreviations: basilar artery (BA), dorsal aorta (DA), facial vein (FV), heart (H), jugular vein (JV), liver (L), nasal vein (NV), optical artery (OA), vena cava (VC), and venous sinus (VS).

Fig. 6

a) MIPs of an E12.5 (top) and E14.5 (bottom) embryos highlighting the spinal cord and the hind-limbs. B) Magnified views of the hind-limb at both stages, oriented to show the anterior (A) to posterior (P) arrangement of the developing digits. At E12.5, there is evidence of digits, which are separated by E14.5. c) The intersomitic vessels (ISV) can be distinguished at the two stages from the mid-trunk region to the beginning of the tail.