Abstract Magnetic resonance microscopy, a non-inva- sive imaging technique was used for a longitudinal fol- low-up of mouse embryonic development in utero and for the assessment of embryonic kidney function using 50 nm magnetite dextran particles. Even though the mor- phologic proton images obtained were still far from clas- sical histological slices quality, an in-plan resolution of 195 µm was achieved for a slice thickness of 800 µm.
Mouse embryos sub-structures such as the fourth ventri- cle, the mesencephalic vesicle, the aorta or the liver can be revealed as early as E11/12. Heart, diaphragm, spinal cord, third, fourth and lateral ventricles were unambigu- ously seen at E13/14; whereas skeleton, tail, kidney and digit can only be seen from E15/16. Kidney and bladder were certainly identified from E16 on. MR microscopy offers a possibility for in utero phenotyping of mice and can therefore be a powerful tool for post-genomic appli- cations.
Keywords Magnetic Resonance microscopy · In utero anatomy · Contrast agent · Magnetite
Introduction
In the fast-developing era of transgenic mouse models, any tool providing information on the effect of gene ma-
nipulation is of great interest for the biologist. The mouse (genetically engineered or wild) is the major model system for studying the genetic basis of mammali- an development. A systematic descriptive reference of the phenotype of the living mouse embryo that would help analyzing the vast amount of data needed to under- stand gene function is still under construction. Conven- tional embryo atlases (Rugh 1968; Theiler 1989; Kauf- man 1992) provide a limited number of pictures and sec- tion planes, but can now be usefully completed by 3-D digital voxel models of mouse embryos calculated from serial histological sections such as those of the E9 mouse embryo (Brune et al 1999).
Magnetic resonance imaging is widely used clinically, but would be of equally great interest in pre-clinical studies, since it allows, for example, non-invasive evalu- ation of anatomy and function associated with normal or pathological modifications, such as those encountered in Alzheimer’s disease (Lemaire et al 2002), or brain matu- ration (Miot-Noirault et al 1997) or, furthermore, in comportmental studies such as in depression (Wrynn et al 2000). Hardware improvements of the MRI tool have lead to MR micro-images (Callaghan 1991; Blackband et al. 1999) of near-histological resolution. On a fixed chicken embryo, 3D-images with an isotropic resolution of 31 µm were acquired in less than 1 h (Hogers et al.
2001), whereas on fixed mouse embryos, isotropic 20–80 µm images were obtained for a 3D, MRI-based at- las (Dhenain et al 2001). Recently, 3D-in utero imaging of rat embryos (Smith et al 1998) and 2D images of E13 and E17 mice were presented, (Hogers et al 2000) and have shown the feasibility of imaging those embryos with an in-plan resolution under 200 µm in less than 30 min in the first case, and in less than 5 min in the sec- ond. The work presented in this study provides (i) a day- to-day follow-up of mouse-embryo development in ute- ro, using high resolution MRM and (ii) an attempt to run functional imaging of mouse embryos in utero using par- ticular contrast agents.
C. Chapon · J.J. Le Jeune · L. Lemaire (
✉
)INSERM ERIT-M 0104, Ingénierie de la Vectorisation, Université d’Angers, 10 rue Boquel, 49100 Angers, France e-mail: [email protected]
Tel.: +33-0241-735006, Fax: +33-0241-735007 F. Franconi
Service Commun d’Analyses Spectroscopiques, Université d’Angers, France
J. Roux
Service Commun Animalerie Hospitalo-Universitaire, Université d’Angers, France
L. Marescaux
Ecole Nationale Vétérinaire, Unité de Pathologie Chirurgicale, Nantes, France
DOI 10.1007/s00429-002-0281-6
O R I G I N A L A R T I C L E
C. Chapon · F. Franconi · J. Roux · L. Marescaux J. J. Le Jeune · L. Lemaire
In utero time-course assessment of mouse embryo development using high resolution magnetic resonance imaging
Accepted: 23 September 2002 / Published online: 13 November 2002
© Springer-Verlag 2002
Materials and methods
Animals and anesthesia
Nine to eleven week-old virgin female OF1 mice were mated for 48 h (n=5), 24 h (n=12) and 12 h (n=6), start- ing at 8:00 a.m. with males of the same genetic constitu- tion. We could have considered a theoretical fecundation time in the middle of the mating period, however, this would have led to inaccurate, staging of the embryo.
Therefore, embryo stages presented in this paper take this confidence interval into account and are labelled ac- cordingly. For functional experiments, the mother tail vein was catheterised in order to allow injection of the contrast agent solution (10 mg Fe/Kg). Two non-preg- nant mice received injections of the contrast agent solu- tion in order to characterise the glomerular filtration of the nanoparticles in adults.
During MRI examination, mice were anaesthetised by spontaneous inhalation of a mixture of isoflurane/oxygen (4%–1.5%)/(1–1.5 l/min), and body temperature was controlled at 36–37 C. Animal care was administered in strict accordance to the French Ministry of Agriculture regulations.
Contrast agent and imaging
Magnetite dextran nanoparticles were prepared and puri- fied as previously described (Pouliquen et al 1989).
Briefly, a ferrous and ferric salt mixture is co-precipitat- ed in the presence of dextran 40000 (Sigma; Saint Quen- tin, France) with dropwise addition of NaOH at 65 C un- der continuous mechanical stirring. The average size of the magnetite crystal is 12 nm, and the hydrodynamic di- ameter of the particles is 50 nm. Magnetic relaxivities at 7T are: r1=1.2 mM-1×s-1and r2=247 mM–1×s-1(Messag- er et al 1999).
Experiments were performed on a Bruker Avance DRX 300 equipped with a vertical superwide-bore mag- net operating at 7 T and shielded gradient system (maxi- mum gradient strength 144 mT/m, rising time <300 µs – Bruker, Wissembourg, France). The resonant circuit of the NMR probe were either a 38 mm or a 64 mm diame- ter birdcage (Bruker, NMR Microscopy Application, Rheinstetten, Germany) depending on the mouse size.
Mice were carefully fixed in the animal holder, then po- sitioned vertically along the magnetic field direction.
Sagittal, axial and coronal scout images were performed to localize embryos, and, if required, re-axing scans for MR imaging were undertaken in order to obtain pure sagittal or coronal virtual slicing of the targeted embryo.
Images were then acquired using a field of view (FOV) of 25 mm×25 mm or 30 mm×30 mm (depending on the size of the mouse) and a matrix 128×128 leading to an in-plan resolution of 198 or 234 µm. if necessary, digital filtering was used in the frequency encoding direction to avoid aliasing. Five to nine 0.8 –1 mm contiguous slices were acquired to cover the entire embryo.
T2 weighted images were acquired using an imaging method known as the multi-slice RARE method (Henning et al. 1986). This method employs a single excitation step followed by the collection of multiple phase encod- ed echoes. The TR was 2000 ms whereas the TE was 7.5 ms. A train of 8 echoes was used to fill the k-space, the effective TE was 31.7 ms, 8 or 12 averages for each phase encoding were performed, resulting in a total ac- quisition time of 4 min and 22 s in the first case and 6 min 32 s in the second.
Bruker ParaVision software was used for data acquisi- tion and processing on a Silicon Graphics O2 worksta- tion with the Irix 6.5 operating system.
Results
A mating period of up to 12 h led to a fecundation rate of 0% (n=6), whereas increasing it to 48 h led to a 80% fe- cundation rate (n=5) but to an inappropriate staging of the embryo. It appeared that the 24 h mating period led to a reasonable fecundation rate of 42% (n=12) and al- lowed reasonably accurate staging of the embryo.
The litter was examined for the first time 9–10 days after fecundation. Embryos (4 mm×3 mm) appeared ho- mogenous on the T2-weighted set of images surrounded by a fine appearance of hypointense structure (Fig. 1A).
This structure may correspond to the embryonic vesicle with rich vascularisation and the consequently large amount of paramagnetic iron contained in the red cells, leading to highly efficient T2* relaxation, which is re- sponsible for the drop in signal intensity.
Imaging pregnant mice 11–12 days after fecundation showed signal heterogeneity within the sagittally sliced embryos (Fig. 1B). Those heterogeneities correspond to macroscopic organisation of some embryonic sub-struc- tures, especially in the developing brain, identified ac- cording to the classical Kaufman histological atlas (Kaufman 1992) as the fourth ventricle and the mesence- phalic vesicle. The cardiac cavity, the liver, and the aorta are also visible. Embryos measured 11 mm×7 mm. The same mouse was then imaged 2 days later (E13–14) (Fig. 2). On the sagittal slices, the embryo measured 12.5×8 mm. Anatomically, the heart, aorta, pleural cavi- ty, diaphragm, liver, spinal cord, umbilical cord, third, fourth and lateral ventricles were unambiguously depict- ed. Forelimb bud were also visible on peripheral slices.
Figure 3 shows the coronal view of E14–15 embryos.
Their size was estimated at 16 mm×10 mm. At this stage, fingers and kidneys can be depicted. Their presence was unambiguously demonstrated at E15–16 (Fig. 4), and cal- cification was undoubtedly present, as shown by the hy- pointense signal corresponding to the vertebra. In this im- aging plan, the common carotid can also be pointed out.
Embryos then measured 18 mm×10 mm.
At later stages (E17–18 or E18–19) post mating, all mice images showed great anatomical detail as displayed in Fig. 5. The size of the embryo was assessed at 20–22 mm×11–13 mm.
In term of functional imaging, experiments on preg- nant or non-pregnant mice showed that the mother kid- neys (mature kidney) were impermeable to the magnetite dextran nanoparticles, since no significant changes of the
bladder signal intensity was observed on the T2-weight- ed set of images (Fig. 6). However and even though a large impregnation of the placenta with the contrast agent was noticed, no significant changes in kidney sig- Fig. 1 In utero 1 mm thick im-
ages of E9/10 (left) and E11/12 (right) mouse embryos. The E11/12 embryo has its head pointing to the right side of the image, the back pointing up- wards. A 5 mm bar is added as a scale
Fig. 2 In utero imaging of E13/14 mouse embryo. Three consecutive 0.8 mm thick, sag- ittal slices of an embryo posi- tioned with the head pointing downwards, its back on the right side of the image. Both images at the bottom are identi- cal, but the image on the right is without label to allow better visualisation. A 5 mm bar is added as a scale
Fig. 3 In utero imaging of an E14/15 mouse embryo. Two, 1 mm thick coronal slices of an embryo positioned with the head pointing upwards. A 5 mm bar is added as a scale
Fig. 4 In utero imaging of an E15/16 mouse embryo. Coronal and sagittal 1 mm thick slices of the mature kidneys. A 5 mm bar is added as a scale
Fig. 5 In utero imaging of an E18/19 mouse embryo. A 5 mm bar is added as a scale
nal intensity were measured in E15 and later pups (Fig. 7).
Discussion
In utero imaging of developing embryos is a challenging feature, especially with the rapid development of geneti- cally engineered mice, for which early depiction of the anatomical modification associated with gene mutation is of prime importance. Non-invasive magnetic reso- nance imaging (MRI) has proven to be an suitable tool for addressing sub-millimetric structures, especially those of fixed animals (Hogers et al. 2001; Dhenain et al.
2001). However, to counterbalance the low intrinsic sen- sibility of the technique, long acquisition times are re- quired. This time constraint is seldom adapted to the im- aging of living animals, that have to be anaesthetized.
Nevertheless, when a relatively static structure such as the brain is addressed, acquisition time can reach half an hour without compromising the welfare of the animal,
nor the quality of the images. However, when the struc- tures targeted are within the abdominal cavity, where un- controllable random motion can occur, imaging time has to be reduced to a minimum. This is the case in in utero mice imaging. Some of these constraints can be partially overcome, as previously shown (Hogers et al 2000), by using high field MR microscopy. Indeed, increasing the magnetic field provides a better signal-to-noise ratio and therefore allows either increasing the image resolution or reducing the experimental time (Hoult and Richards 1976). However, as magnetic field intensity increases, relaxation times are significantly affected, as shown, for example, on MRM of rat spinal cord (Franconi et al 2000; Narayana et al1999) and can therefore notably re- duce the quality of the image by altering the contrast.
Therefore, a compromise has to be reached. A second technical limitation is due to the maximal strength of the magnetic field gradient coils. These gradients are used to select the position of the imaging plane, its thickness and the in-plan resolution. When in utero imaging is ad- dressed, positioning of the embryo cannot be modified by the investigator, the choice of the imaging plane has therefore to be made according to the gradients which explains why the images presented in this paper have a thickness ranging from 0.8 to 1 mm dependent on the hardware limitations of our system. One way to over- come this slice thickness limitation will be to perform 3D-spin echo acquisitions, but this will lead to longer ac- quisitions, unless 3D-gradient echo imaging sequences are used. However, such gradient echo sequences are not sensitive to the particular contrast agent used to assess in utero renal function, and moreover, the contrast between tissues is rather poor in the field used in this study.
Nevertheless, a longitudinal follow-up of mouse em- bryo development can be achieved using high field MR microscopy. A particularity of the T2-weighted MRM sequence used in this study is its sensitivity to motion.
Even though this characteristic can be a potential draw- back in terms of image quality, (heart conspicuity for ex- ample) it appeared to be of major advantage when ad- dressing flux. Indeed, as spins moved along magnetic field gradients used to encode the image, they shifted in phase, inducing a signal loss due to intravoxel phase dis- persion. Thus, vessels such as the aorta and veins were unambiguously represented by fine hypointense lines.
In terms of time course development follow-up, we attempted to address the kidney maturation regarding anatomy and functionality. Both have been, and still are, extensively studied (Bard et al.2001; Phillips et al 2001).
Compiling figures 3 & 4 allows such an anatomical fol- low-up. Even though invasive histological studies (Kauf- mann, 1992) have shown that metanephros appears to be divided into an outer cortical zone and an inner medullar region as early as E12–13, such a distinction was not possible, since metanephros were not observed on our T2-weighted image set, probably because of their small size. From E14–15, the metanephros can be pointed out.
It appeared as an ovoid structure, 1–1.5 mm in length with a medullar and a cortical component. However, the Fig. 6 Assessment of signal change in the bladder (filled circles)
and kidneys (open squares) of a non-pregnant mouse induced by an i.v. injection of 50 nm superparamagnetic iron oxide nanoparti- cles
Fig. 7 Change in signal of mother kidney (open squares), fetal kidney (open circles) and placenta (closed triangles) after an i.v.
injection 50 nm superparamagnetic iron oxide nanoparticles
resolution on the image did not allow reaching better de- lineation between the kidney substructures.
Nevertheless, the challenging feature of this work was mainly on the in utero functional imaging perspective.
Previous experiments conducted on adult rats (Pouliquen et al 1994), as well as the present work on adult mice, have shown that 50 nm magnetite dextran particles do not pass through the mature kidney, whereas 30 nm par- ticles do so, and can therefore be used to assess renal perfusion (Yang et al 2001). Extravasation of particles across the glomerular capillaries is dependent on their size and charge (Brenner et al. 1978; Takakura et al 1998). The peculiarity of those 50 nm particles, in term of glomerular exclusion, was the basis of our hypothesis, that even though those nanoparticles could not pass through the mature kidney, they could potentially pass through the immature kidney and can therefore be used as a marker for kidney functionality establishment fol- low-up. Despite the tight placental microcirculation which limits the crossing of the fetoplacental barrier (Firth et al. 1996), diffusive Gadolinium-based contrast agents can cross the placenta due to the existence of con- tinuous extracellular pathways and be recovered in the embryonic bladder (Tanaka et al 2001). However, to our knowledge, no such data are available for magnetite dex- tran nanoparticles. After intra-venous injection, we no- ticed a large impregnation of the placenta by the particu- lar contrast agent (Fig. 7), but no crossing of this barrier, since no significant changes in signal intensity were measured in E15 or later embryos. This fact has to be compared to the observation in women with microbubble contrast agent (average diameter of 2.7 µm), in which air bubbles are attached to the surface of galactose micro- particles designed for ultrasonic studies that were trapped by the fetoplacental barrier (Orden et al. 1999).
Conclusions
High-field nuclear magnetic resonance microscopy is a unique tool that allows the longitudinal follow-up of mouse development at sub-millimetric resolution in a non invasive manner and without injection of any chemi- cals. The attempt to challenge the kidney function in ute- ro has not been successful, due to the placenta imperme- ability to the tested magnetite dextran nanoparticles, even though no technical limitation has been pointed out.
In the new era of genetically engineered models, having in hand a tool which allows non-invasive phenotyping of rodents is of major interest.
Acknowledgement The authors thank Dr. J.B.L. Bard from the anatomy department of Edinburgh University for his advice, and P. Legras and D. Gilbert for the housing and care of the animals used in this study.
References
Bard JB, Gordon A, Sharp L, Sellers WI (2001) Early nephron formation in the developing mouse kidney. J Anat 199:
385–392
Brenner BM, Hostetter TH, Humes HD (1978) Glomerular perm- selectivity; barrier function based on distribution of molecular size and charge. Am J Physiol 234:455-460
Brune RM, Bard JBL, Dubreuil C, Guest E, Hill W, Kaufman M, Stark M, Davidson D and Baldock RA (1999) A three-dimen- sional model of the mouse at embryonic day 9. Devel Biology 216:457–468
Blackband SJ, Buckley DL, Bui JD, Phillips MI (1999) NMR mi- croscopy-beginnings and new directions. MAGMA 9:112–116 Callaghan PT (1991) Principles of nuclear magnetic resonance mi-
croscopy. Oxford University Press, New York
Dhenain M, Ruffins SW, Jacobs RE (2001) Three-dimensional digital mouse atlas using high-resolution MRI. Dev Biol 232:458–470
Firth JA, Leach L (1996) Not trophoblast alone; a review of the contribution of the fetal microvasculature to transplacental ex- change. Placenta 17:89–96
Franconi F, Lemaire L, Marescaux L, Jallet P, Le Jeune JJ (2000) In vivo quantitative microimaging of rat spinal cord at 7 T.
Magn Reson Med 44:893–898
Henning J, Nauerth A and Friedburg H (1986) RARE imaging; a fast imaging method for clinical MR. Magn Reson Med 3:823–833
Hogers B, Gross D, Lehmann V, Zick K, De Groot HJ, Gittenberger- De Groot AC, Poelmann RE (2000) Magnetic resonance microscopy of mouse embryos in utero. Anat Rec 260:
373–377
Hogers B, Gross D, Lehmann V, De Groot JM, De Roos A, Gittenberger-de Groot AC, Poelmann RE (2001) Magnetic resonance microscopy at 17.6 Tesla on chicken embryos in vitro. J. Magn Reson Imaging 14:83–86
Hoult DI, Richards RE (1976) The SNR of the NMR experiment.
J Magn Reson 24:71–85
Kaufman MH (1992) The atlas of mouse development. Academic Press, London
Lemaire L, Fournier J, Ponthus C, Le Fur Y, Confort-Gouny S, Vion-Dury J, Keane P and Cozzone PJ (2002) MR imaging of the neuroprotective effect of Xaliproden in rats. Invest Radiol 37:321–327
Messager T, Franconi F, De Bray JM, Denizot B, Lemaire L, Jallet P, Le Jeune JJ (1999) Contrast agents relaxivities evolu- tion with magnetic field strenth 0.47–7 Teslas. Proc. ES- MRMB Abst.500
Miot-Noirault E, Barantin L, Akoka S, Le Pape A. (1997) T2 relaxation time as a marker of brain myelination: experimental MR study in two neonatal animal models. J Neurosci Methods 72:5–14
Narayana P, Fenyes D, Zacharopoulos N (1999) In vivo relaxation times of gray matter and white matter in spinal cord. Magn Reson Imaging 17:623–626
Pouliquen D, Okon P, Kovaleva ZV, Stepanova TP, Lavit SG, Kudryavtsev BN, Jallet P (1994) Biodegradation of magnetite dextran nanoparticles in the rat. A histologic and biophysical study. Lab Invest 71:895–903
Orden MR, Gudmundsson S, Kirkinen P (1999) Intravascular ultrasound contrast agent; an aid in imaging intervillous blood flow? Placenta 20:235–240
Phillips CL, Arend LJ, Filson AJ, Kojetin DJ, Clendenon JL, Fang S, Dunn KW (2001) Three-dimensional imaging of embryonic mouse kidney by two-photon microscopy. Am J Pathol 158:49–55
Pouliquen D, Perdrisot R, Ermias A, Akoka S, Jallet P, Le Jeune JJ (1989) Superparamagnetic iron oxide nanoparticles as a liver MRI contrast agent: a contribution of microencapsulation to improved biodistribution. Magn Reson Imaging 7:619–27
Rugh R (1968) The mouse: its reproduction and development.
Burgess, Minneapolis. [reprinted 1990, Oxford Univ Press, Oxford]
Smith BR, Shattuck MD, Hedlund LW, Johnson GA (1998) Time- course imaging of rat embryos in utero with magnetic reso- nance microspcopy. Magn Reson Med 39:673–677
Takakura Y, Mahato RI, Hashida M (1998) Extravasation of mac- romolecules. Adv Drug Delivery 34:93–108
Tanaka YO, Sohda S, Shigemitsu S, Niitsu M, Itai Y (2001) High temporal resolution dynamic contrast MRI in a high risk group for placenta accreta. Magn Reson Imaging, 19:635–642
Theiler K (1989) The house mouse: atlas of embryonic develop- ment. Springer, Berlin Heidelberg New York
Wrynn AS, Mac Sweeney CP, Franconi F, Lemaire L, Pouliquen D, Herlidou S, Leonard BE, Gandon J, de Certaines JD (2000) An in-vivo magnetic resonance imaging study of the olfactory bulbectomized rat model of depression. Brain Res 879:
193–199
Yang D, Ye Q, Williams M, Sun Y, Hu TCC, Williams DS, Moura JMF and Ho C (2001) USPIO-enhanced dynamic MRI:
evaluation of normal and transplanted rat kidneys. Magn Re- son Medicine 46:1152–1163
