- Email: [email protected]
MRI of cellular layers in mouse brain in vivo
NeuroImage 47 (2009) 1252–1260
Contents lists available at ScienceDirect
NeuroImage j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / y n i m g
MRI of cellular layers in mouse brain in vivo Susann Boretius a,⁎, Lars Kasper a, Roland Tammer a,b, Thomas Michaelis a, Jens Frahm a,b a b
Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut für biophysikalische Chemie, 37070 Göttingen, Germany DFG Center of Molecular Physiology of the Brain, 37073 Göttingen, Germany
a r t i c l e
i n f o
Article history: Received 6 April 2009 Revised 6 May 2009 Accepted 29 May 2009 Available online 8 June 2009 Keywords: High-resolution MRI Mouse Cortical layer Brain
a b s t r a c t Noninvasive imaging of the brain of animal models demands the detection of increasingly smaller structures by in vivo MRI. The purpose of this work was to elucidate the spatial resolution and structural contrast that can be obtained for studying the brain of C57BL/6J mice by optimized T2-weighted fast spin-echo MRI at 9.4 T. As a prerequisite for high-resolution imaging in vivo, motion artifacts were abolished by combining volatile anesthetics and positive pressure ventilation with a specially designed animal bed for fixation. Multiple substructures in the cortex, olfactory bulb, hippocampus, and cerebellum were resolved at 30 to 40 μm in-plane resolution and 200 to 300 μm section thickness as well as for relatively long echo times of 65 to 82 ms. In particular, the approach resulted in the differentiation of up to five cortical layers. In the olfactory bulb the images unraveled the mitral cell layer which has a thickness of mostly single cells. In the hippocampus at least five substructures could be separated. The molecular layer, Purkinje layer, and granular layer of the cerebellum could be clearly differentiated from the white matter. In conclusion, even without the use of a contrast agent, suitable adjustments of a widely available T2-weighted MRI sequence at high field allow for structural MRI of living mice at near single-cell layer resolution. © 2009 Elsevier Inc. All rights reserved.
Introduction In recent years MRI has developed into an indispensable research tool for noninvasive imaging of the brain. Complementing histological studies, in vivo MRI offers the detection of more global changes in intact animals but does not suffer from shrinking or other preparation artifacts and allows for a sequential follow-up of the same animal. In addition, because most MRI techniques can be performed in both humans and experimental animals, the method provides a missing link in translational research. In this context, the successful use of (genetically modified) mouse models enhances the demand for detecting increasingly smaller brain structures by in vivo MRI. In terms of absolute spatial resolution this corresponds to isotropic image voxel dimensions of less than 100 μm or voxel sizes of less than 1 nl. In addition, the discrimination of specific structures requires sufficient contrast, or more precisely a certain contrast-to-noise ratio (CNR) that also takes into account the signal-to-noise ratio (SNR) and available measurement time. In fact, because a reduction of the voxel volume in MRI results in a linear decrease of the SNR, the reduced partial volume effect and better contrast between adjacent structures at high resolution might be compromised by a lower SNR. Similarly, a better tissue contrast based on a more pronounced relaxation time weighting is also at the expense of SNR due to more pronounced T1 saturation or T2 attenuation.
⁎ Corresponding author. Fax: +49 511 201 1307. E-mail address: [email protected] (S. Boretius). 1053-8119/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2009.05.095
Recently, several attempts have been made for high-resolution imaging of the cortical anatomy in humans (Barbier et al., 2002, Walters et al., 2003) advancing the spatial resolution to 240 × 240 × 1000 μm3 (Duyn et al., 2007). In mouse brain, however, high-resolution MRI at 20 to 60 μm isotropic resolution (corresponding to voxel sizes of 8 pl to 0.2 nl) has so far only been achieved ex vivo using sacrificed animals (Badea et al., 2007; Benveniste and Blackband, 2002; Kovacevic et al., 2005). Although of basic value in closely linking MRI to histological assessments, such studies do not fully exploit the specific advantages of in vivo MRI over conventional histology. Nevertheless, only few MRI studies attempted to visualize cortical substructures in living mice (Angenstein et al., 2007; Chahboune et al., 2007; Lee et al., 2005; MacKenzie-Graham et al., 2004). A first challenge stems from movements due to spontaneous breathing (Ma et al., 2008) which bears the risk of impairing or even precluding access to microscopic resolution. Continuous endotracheal intubation and positive pressure ventilation may help to overcome this problem (Brown et al., 1999) and has been shown to allow for repeated measurements of mice, each with a total duration of several hours (Merkler et al., 2005; Boretius et al., 2009). A second difficulty stems from the relatively small amount of white matter (Zhang and Sejnowski, 2000). In humans, the many strongly myelinated structures are particularly useful for mapping the brain into anatomically and functionally distinct areas. For example, in the striate cortex T1weighted MRI at high spatial resolution reveals the stria of Gennari (Barbier et al., 2002; Clark et al., 1992). The aim of this study was to explore the practical limits of in vivo MRI of mouse brain with respect to resolution and (native) contrast
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
that can be achieved by suitable MRI sequences on commercially available high-field MRI systems and within reasonable measuring times. These techniques were exploited for the detection of cerebral layers in normal mice, where we were able to achieve near single-cell layer resolution in several brain structures. Methods
1253
Table 1 In vivo protocols for the identification of mouse brain structures using fast spin-echo MRI at 9.4 T. Brain structure
Resolution/μm
Echo time/ms
Measurement time/min
Cerebral cortex Olfactory bulb Hippocampal formation Cerebellum
30 × 30 × 300 40 × 40 × 300 53 × 53 × 234 53 × 53 × 234
65–82 65 65–82 51–65
90 60 60 60
Animals All experiments were performed in compliance with relevant laws and institutional guidelines. Animal experiments were approved by local authorities. For MRI adult C57BL/6J mice (n = 4) were initially anesthetized using a chamber pervaded with 5% isoflurane in oxygen. Subsequently, the mice were intubated and kept under anesthesia with 1 to 1.5% isoflurane in a 1:1.5 mixture of oxygen and ambient air. Respiration was monitored by a pressure transducer under the ventral thoraco-abdominal region. The animals were placed in a prone position with the head fixed to a teeth and palate holder as demonstrated in Fig. 1. This specially designed Göttingen animal bed allowed for a reproducible and reliable fixation of the mouse head and mounted receive coil in the isocenter of the magnet. The rectal temperature was held constant at 36 ± 0.5 °C by temperaturecontrolled water flowing through pipes that form a chamber. MRI MRI studies were performed at 9.4 T (Bruker Biospin, Ettlingen, Germany) with the use of a birdcage resonator (inner diameter 70 mm) for excitation and a saddle-shaped phased-array surface coil (4 elements) for signal detection (Bruker Biospin, Ettlingen, Germany). T2-weighted images were acquired with a multislice fast spinecho MRI sequence (repetition time TR = 4200 ms, 8 differently phase-encoded echoes, number of slices 14–20, matrix size 512 × 512). Experiments involved four different (mean) echo times TE and acquisitions at different spatial resolutions. Typically, the in-plane resolution varied between 30 and 80 μm, while the section thickness was about 200 to 300 μm (see below and Table 1). The smallest field of view was 15.4 × 15.4 mm2 covering a full brain section. The receiver bandwidth was adapted to the respective echo time yielding pairs of
36 ms/100 kHz, 51 ms/50 kHz, 65 ms/35 kHz, and 82 ms/25 kHz. To minimize chemical shift artifacts and in particular to avoid any contamination of the cortex by overlapping signals from subcutaneous fat, the orientation of the readout gradient was chosen such as to shift lipid signals into an anterior direction. Finally, to assess the potential of high-resolution T1-weighted MRI, manganese-enhanced 3D gradient-echo MRI was acquired 24 h after subcutaneous injection of 40 mg/kg MnCl2 using a 3D FLASH sequence (TR/TE = 17.0/3.8 ms, flip angle 25°) at 30 × 30 × 300 μm3 resolution. In all cases, images were calculated without zero filling, so that voxel dimensions refer to truly acquired resolutions. Cortical profiles To directly compare the cortical lamination pattern seen on T2weighted images with the underling cyto- and myeloarchitecture, MRI intensity profiles in the parietal cortex were extracted from an axial T2-weighed image as defined in Fig. 2. Similarly, profiles from corresponding histological sections were obtained for myelin, cresyl violet, and Nissl stains using the open source software ImageJ (version 1.34s, http://rsb.info.nih.gov/ij/). For this purpose the histological sections were mildly smoothed by a Gaussian filter to improve the visualization of the lamination pattern. For each technique 5 intensity profiles were normalized according to their length and averaged. To account for intermodal differences in the intensity and gray value ranges all profiles were z-normalized (Eickhoff et al., 2005). Results Experimental adjustments of a fast spin-echo MRI sequence substantially improved the detection of several small brain structures
Fig. 1. The Göttingen animal bed for in vivo MRI of rodent brain. (A) Entire setup as well as magnified views of (B) the animal bed and (C) the mouse head without receive coil. 1 = magnet bore, 2 = gliding channel, 3 = animal bed, 4 = supply lines of anesthetic gas, 5 = supply lines of warm water, 6 = clip for fixation of the surface coil, 7 = teeth and palate holder, 8 = surface coil, 9 = temperature-controlled chamber, 10 = endotracheal tube.
1254
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
in living mice. The resulting main parameters for resolving different cellular layers in the cerebral cortex, olfactory bulb, hippocampus, and cerebellum are summarized in Table 1. These general choices with respect to resolution and T2 contrast are supported by Fig. 3 comparing horizontal T2-weighted images of the mouse brain at different in-plane resolutions (30 vs 80 μm) and echo times (36 vs 82 ms). In general, the in-plane resolution turns out to be the most important parameter for a delineation of small brain structures such as the myelinated striatal fibers, while a strengthening of the T2 contrast by a prolongation of the echo time improves the visibility of substructures in the cortex. Cerebral cortex Fig. 2. Definition of intensity profiles as a function of cortical depth. (Left) Axial T2weighted image of the cerebral cortex of a mouse brain in vivo and (right) magnified view of the analyzed region (white rectangle). Profiles cover the cerebral cortex from the surface (cortical depth 0%) to the white matter (wm, cortical depth 100%).
Fig. 3. Horizontal T2-weighted images of a mouse brain in vivo (section thickness 300 μm) as a function of (top) in-plane resolution (80 vs 30 μm) at TE = 65 ms and (bottom) echo time (36 vs 82 ms) at 30 μm in-plane resolution. For details see text.
Best results for a visualization of cortical layers were achieved at 30 to 40 μm in-plane resolution and 65 to 82 ms echo time. As shown in Fig. 4 (top right), a resolution of 40 × 40 × 300 μm3 (TE = 65 ms) unravels up to 5 different layers in the cerebral cortex of living mice. They resemble – but not exactly correspond to – the cortical layers seen in histological sections by light microscopy (Fig. 4, bottom left). In technical terms, the most important aspect was a high in-plane resolution. This observation was supported by the fact that the layer differentiation was impaired when reducing the slice thickness at the expense of the in-plane resolution while keeping the absolute voxel size constant (data not shown). The influence of the echo time or T2 contrast on the layer detection is shown in Fig. 5 for two horizontal sections at 30 × 30 × 300 μm3 resolution. Whereas no layer-like structures are visible at an echo time of 36 ms, echo times of 65 or 82 ms allow for the identification of up to 5 different layers. However, the optimal echo time depends on the cortical area of interest. For instance, in a more dorsal section (Fig. 5, top) covering the primary auditory cortex and the dorsal area of the secondary auditory cortex, echo times of higher than 65 ms do not improve the situation. In contrast, in a more ventral horizontal section (Fig. 5, bottom) depicting the secondary auditory cortex, an echo time of 82 ms leads to a better differentiation.
Fig. 4. Magnified horizontal T2-weighted images of the cerebral cortex of a mouse brain in vivo as a function of in-plane resolution at TE = 65 ms and 300 μm section thickness. Layer-like structures in the cortex (yellow lines) are compared to histologically defined layers (black lines) in a corresponding Nissl stain indicating areas of different neuronal density (bottom left, http://www.brainmaps.org/).
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
1255
Fig. 5. Magnified views of two horizontal T2-weighted sections (top, bottom) of the cerebral cortex of a mouse brain in vivo as a function of echo time at 30 × 30 × 300 μm3 resolution. Layer-like structures in the cortex (yellow lines) are compared to histologically defined layers (black lines) in corresponding cresyl violet stains indicating areas of different neuronal density (right, http://www.mbl.org/).
Fig. 6. Magnified axial T2-weighted images of the cortex of a mouse brain in vivo at (A) 40 × 40 × 300 μm3 resolution (TE = 65 ms) and (B) 30 × 30 × 300 μm3 resolution (TE = 82 ms) in comparison to (C) a Nissl stain indicating neuronal density (http://mouse.brain-map.org/atlas/ARA/Coronal/) and (D) a myelin stain (http://www.hms.harvard.edu/research/ brain/). (E) Magnified view of image (A) indicating MRI-detectable T2 layers a to e in the cortex (yellow lines). (F) Overlay of part of image (E) onto the Nissl stain indicating histologically defined layers I to VI. (G) Overlay of part of image (E) on the myelin stain. 1,2,3 = layers in piriform cortex; bfd = barrel field; fl = forelimb area; hl = hindlimb area; p1, p2 = areas 1 and 2 in parietal cortex.
1256
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
layer). These cortical laminae are best seen in areas 1 and 2 of the parietal cortex, while a layer separation was not possible in sensorimotor areas representing the forelimb and hindlimb. When directly comparing the T2-weighted images to the corresponding histological sections by partial overlay (Figs. 6F, G), the brighter outer lamina (T2 layer a) seems to correspond to cortical layers I and II, the next darker lamina (T2 layer b) to layers III and IV, the inner brighter lamina (T2 layer c) to layer V, and the inner darker and brighter laminae (T2 layers d and e) to layer VI. The MRI identification of a most inner bright lamina (T2 layer e) is supported by an overlay with the Nissl stain (Fig. 6F) and probably indicates a part of cortical layer VI with lowered cell density (see also the cresyl violet staining in Fig. 5). Finally, at higher spatial resolution and longer echo time (Fig. 6B) the observation of an additional dark substructure in only a part of the cortex most likely refers to the barrel field which is known to exhibit a high neuronal density. Furthermore, the chosen experimental conditions allow for the detection of 3 different layers (1 to 3) in the piriform cortex (Figs. 6B, C). The relationship between MRI and underlying histology is further illustrated in Fig. 7 depicting cortical profiles for myelin, T2-weighted MRI signal intensity, and cellular density as reflected by cresyl violet and Nissl stains. This comparison is in general agreement with the findings of Fig. 6 and suggests that the T2 contrast is mainly determined by the degree of myelination and the cell density. In particular, myelination is low at the cortical surface (high z-score) and increases with cortical depth (lower z-scores). This observation parallels the general trend seen for T2-weighted MRI signal intensities. However, their z-scores reveal an additional modulation which resembles the pattern found for cell density. For example, the
Fig. 7. Mean intensity profiles of the parietal cortex (compare Fig. 2) from (top) a myelin stain, (middle) a T2-weighted (T2w) image of a mouse brain in vivo, and (bottom) a cresyl violet as well as Nissl stain reflecting cell density. To allow for a comparison of profiles from different techniques respective intensities are z-normalized and the distance from the cortical surface is given in percent of cortical depth.
Fig. 6 demonstrates the appearance of MRI-detectable cortical layers in axial sections at 40 × 40 × 300 μm 3 resolution with TE = 65 ms (Fig. 6A) and at 30 × 30 × 300 μm3 resolution with TE = 82 ms (Fig. 6B). The images are compared to a Nissl stain (Fig. 6C) reflecting neuronal density and a myelin stain (Fig. 6D). In a magnified view of the cortex (Fig. 6E) the above mentioned 5 MRI layers are denoted as T2 layers a (most outer layer) to e (most inner
Fig. 8. Magnified horizontal T2-weighted images of the olfactory bulb of a mouse brain in vivo as a function of echo time at 30 × 30 × 300 μm3 resolution indicating access to the mitral layer (yellow arrows). Bottom row: Magnified view and corresponding Nissl stain (http://www.brainmaps.org/). aob = accessory olfactory bulb; epl = external plexiform layer; gl = glomerular layer; gr = granule layer; mi = mitral layer; ipl = inner plexiform layer; wm = white matter.
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
initial sharper decline than that of the myelin z-scores may be caused by the steep decrease of the z-scores for the cell density. Moreover, the relatively strong increase of the z-score for T2-weighted MRI in the region of layer V may result from both the reduced myelination and the much lower cell density than in the neighboring regions IV and VI. Olfactory bulb To distinguish the different layers of the olfactory bulb a minimum in-plane resolution of 40 μm was required. In addition, Fig. 8 demonstrates that an echo time of at least 51 ms is necessary in order to achieve sufficient contrast. This especially applies to a discrimination of the mitral cell layer and the external and internal plexiform layers. With these MRI parameters the granular layer appears as a gray–black layer that may be well distinguished from the brighter external plexiform layer. The mitral cell layer, which in histology mainly presents as a cellular monolayer, exhibits low signal intensity in T2-weighted images because of its dense packing which opposes the bright signal from the internal plexiform layer (Fig. 8, bottom). Hippocampal formation To differentiate the molecular layer, granule cell layer, and polymorphic cell layer of the dentate gyrus, T2-weighted MRI required a resolution of 53 × 53 × 243 μm3 or higher and a strong T2 contrast with echo times of 65 ms or longer. As shown in Fig. 9 an echo time of 36 ms did not allow for a proper separation of the molecular layer and granule layer which becomes possible at 65 ms. In addition, at an echo time of 82 ms (Fig. 9, bottom left), the polymorphic layer and granule cell layer may be identified. These MRI parameters also delineate the pyramidal cell layer (in CA1 to CA3), the stratum oriens, and the white matter fibers of the alveus and external capsule which mark the outer border of the hippocampus. Based on pronounced T2 contrast, the
1257
densely packed pyramidal cell layer appears dark, the relatively cellfree stratum oriens bright, and the myelinated fibers dark. Noteworthy, manganese-enhanced MRI reveals complementary information (Fig. 9, bottom right). The approach exploits the fact that Mn2+ ions behave like Ca2+ ions because of their similar valence and size. It has been shown that Mn2+ ions can enter neurons via voltagegated calcium channels (Yokel and Crossgrove, 2004, Gadjanski et al., in press). With manganese uptake and subsequent axonal transport, the polymorphic and granule cell layers of the dentate gyrus as well as the pyramidal cells of the CA3 region experience a strong signal enhancement, whereas the pyramidal cells of the CA1 region remain undetectable. In contrast, the T2-weighted images allow for a visualization of the entire pyramidal cell layer due to their dependence on cellular density. On the other hand, the stratum radiatum and stratum lacunosum-moleculare may be distinguished using manganese-enhanced MRI, but not with native T2 contrast at high spatial resolution. Cerebellum Horizontal T2-weighted images of the cerebellum at different echo times and spatial resolutions are shown in Figs. 10 and 11 respectively. Echo times of 36 ms and slightly better of 51 ms provide good contrast between short-T2 structures such as white matter and Purkinje cells (dark) and those characterized by longer T2 relaxation times like the granular layer (bright). The molecular layer with a low cellular density (and long T2 relaxation time) exhibits an even higher MRI signal. The longest echo time of 82 ms results in very similar signal intensities (low contrast) for white matter, granular cell layer, and Purkinje cell layer. Fig. 11 demonstrates a slightly different behavior for resolving substructures of the cerebellum as found for most other brain structures. For example, even at 80 × 80 × 300 μm3 resolution the light-gray molecular layer may be clearly detected (data not shown).
Fig. 9. Magnified horizontal T2-weighted images of the hippocampal formation of a mouse brain in vivo as a function of echo time at 30 × 30 × 300 μm3 resolution and in comparison to a corresponding Giemsa stain (http://www.brainmaps.org/) and a manganese-enhanced T1-weighted image (Mn2+). gldg = granule cell layer; hf = hippocampal fimbria; mlgd = molecular layer; podg = polymorphic layer; s = subiculum; so = stratum oriens; sr = stratum radiatum; slm = stratum lacunosum-moleculare; wm = white matter.
1258
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
Fig. 10. Magnified horizontal T2-weighted images of a mouse cerebellum in vivo as a function of echo time at 30 × 30 × 300 μm3 resolution.
Increasing the spatial resolution to 53 μm or 40 μm allows for the additional delineation of the thin and dark Purkinje layer, the slightly brighter granular layer, and the dark white matter. A reduction in slice thickness from 300 to 234 μm further improved the delineation. In contrast, acquisitions at even higher spatial resolution were partially comprised by reduced SNR and enhanced motion sensitivity due to its close spatial relationship to the medulla oblongata eventually affected by respiratory movements. Discussion
cortical area, layers I and II and in part also layer III are poorly myelinated (Paxinos, 2004). Considering cell density, degree of myelination, and proportional size of the cortical layer, the bright appearance of the outer T2 layer a (Fig. 6E) most likely corresponds to cortical layers I and II. The low cellular density and the sparse myelination both support a relatively long T2 relaxation time that leads to bright intensities on T2-weighted images. This interpretation is further supported by a comparison of intensity profiles that have been obtained across the parietal cortex for T2-weighted MRI and histology (Fig. 7). In the area of layers I and II and partly of layer III the slope of the decreasing z-score for T2-weighted MRI is larger than that of the slowly increasing myelination, but smaller than that of the rapidly decreasing cell density. Thus, the area of T2 layer a corresponds to the cortical layer I and partly also to layers II–III (Fig. 7). The cell-rich internal granular layer IV is most prominent in sensory areas and poorly developed or even absent in areas with motor function. The high cell density and the strong myelination of layer IV (and partly also of layer III) cause a relatively short T2 relaxation time. This behavior leads to low intensities in T2-weighted images, so that the first dark band in areas 1 and 2 of the parietal cortex denoted T2 layer b (Fig. 6E) best matches with cortical layers III and IV (Figs. 6F and 7). Another cell-rich region and correspondingly dark zone in area 1 most likely belongs to the barrel field of layer IV (Fig. 6B). The internal pyramidal layer V contains medium to large pyramidal neurons. Especially in area 1 of parietal cortex, layer V consists of fewer cells compared to neighboring layers IV and VI and therefore presents with a brighter intensity on T2-weigted images. Its main part may be identified with T2 layer c (Figs. 6E and 7). In the forelimb and hindlimb areas of the parietal cortex, however, layer V exhibits more densely packed pyramidal cells and can hardly be separated from layers IV and VI, neither in the Nissl stain (Fig. 6C) nor on T2-weighted MRI (Fig. 6B).
High-resolution T2-weighted MRI at 9.4 T allows for the identification of multiple cellular layers in the cerebral cortex, olfactory bulb, hippocampus, and cerebellum of intact living mice. In some areas this capability includes the detection of layers with predominantly singlecell thickness — without any application of a contrast agent and within reasonable measurement times of 60 to 90 min. However, depending on the underlying contrast properties of the various tissue components, the structures differentiated by MRI do not necessarily correlate with the structures seen in histology. Pertinent relationships will be discussed below to better understand the MRI correlates of conventional histological stains. Isocortex The mammalian isocortex shows a lamination into 6 layers (I to VI) that are histologically well described, for example for humans (Yamamori and Rockland, 2006; Zilles and Palomero-Gallagher, 2001) and mice (Caviness, 1975; Polleux et al., 1997). While the separation of different layers is mainly based on their cyto- and myeloarchitecture, the layers may be further characterized by a – functionally relevant – receptor expression and chemical composition (Paxinos, 2004). Moreover, size and cell density as well as the degree of myelination of individual layers differ between different cortical areas. The outermost cortical layer, that is molecular layer I, contains only few scattered neurons and consists mainly of extensions of apical dendrites and horizontally-oriented axons. In contrast, the external granular layer II comprises small pyramidal neurons and numerous stellate neurons (Paxinos, 2004; Peters and Jones, 1984). Layer III, the external pyramidal layer, reveals a nearly similar cell density and is predominantly built of small and medium-sized pyramidal neurons as well as of non-pyramidal neurons. Moreover, independent of the
Fig. 11. Magnified horizontal T2-weighted images of a mouse cerebellum in vivo as a function of spatial resolution at TE = 65 ms and in comparison to a corresponding Giemsa stain (http://www.brainmaps.org/). gl = granular layer; ml = molecular layer; pl = Purkinje cells; wm = white matter.
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
The inner multiform layer VI contains few large pyramidal neurons and many small spindle-like pyramidal and multiform neurons. On T2-weighted MRI layer VI mainly appears dark and especially in areas 1 and 2 may be well separated from the brighter layer V. However, a comparison of cortical intensity profiles reveals an underestimation of the thickness of layer V by the corresponding T2 layer c (Fig. 7). In fact, T2 layer d seems to reflect a transition zone including parts of layers V and VI. The T2 layer d can be further distinguished from an inner brighter lamina denoted T2 layer e (Fig. 6E) that seems to reflect the lower cell density in the innermost part of layer VI as indicated by a cresyl violet and a Nissl stain (compare Figs. 5 and 7) and the slightly reduced myelin content (higher z-score in Fig. 7). Previous MRI studies at ultrahigh 43 μm isotropic resolution described the layers in the cortex of formalin-fixed mouse brains also using T2-weighted MRI at 9.4 T (Sharief and Johnson, 2006; Badea et al., 2007). These in situ MRI reports distinguished 5 cortical T2 layers of alternating brightness which seem to be in close agreement with the findings presented here. Though studies of fixed brains pose no limitations with respect to residual motion or measuring time and certainly offer highest resolution, it should be kept in mind that the MRI contrasts in living and dead tissue are sensitive to a variety of mechanisms including temperature, intra- and extracellular compartmentalization, oxygenation, and water content. In fact, all these parameters are known to change during formalin fixation. Alternatively, the availability of native MRI contrasts may be enhanced by the use of contrast agents. In small animal research, an attractive tool is the application of Mn2+ as an analogue to Ca2+. In mice systemic application of high doses of MnCl2 revealed an unequal distribution of Mn2+ ions – as evidenced by a corresponding Mn2+induced signal enhancement – over the cortical laminae (Angenstein et al., 2007; Lee et al., 2005), while rats presented with a signal increase in layers II and V but not in layers I and III (Silva et al., 2008). Apart from the principal possibility to address functional aspects of cortical organization, disadvantages of the Mn2+ approach are its neurotoxicity (Michalke et al., 2007) and the long retention time in the brain with a biological half-life of 51 to 74 days (Takeda et al., 1995), which hampers follow-up studies in the same animal. Piriform cortex In contrast to the 6-layered isocortex the piriform cortex consists of only three layers. Layer 1 mainly contains fibers and apical dendrites from the underlying layer 2 which consists of densely packed pyramidal cells. In comparison to layer 2, layer 3 is a moderately dense cell layer comprising intrinsic fibers though to a lesser extent than layer I (Paxinos, 2004). According to its low cell density the outermost layer 1 causes a long T2 relaxation time and appears bright on T2-weighted images (Fig. 6B). In contrast, the cell-rich layer 2 is characterized by a shorter T2 relaxation time and a dark appearance which clearly delimits the less cell-dense layer 3 with its brighter intensity. The best separation of these layers is achieved with an inplane resolution of 30 μm and an echo time of 82 ms (Fig. 6B). Previous MRI studies of the piriform cortex of rodents have been performed in animal models of epilepsy using T2-weighted imaging (Bhagat et al., 2005) and diffusion-weighted imaging (Eidt et al., 2004; Engelhorn et al., 2007). To the best of our knowledge, however, this is the first demonstration of the piriform layers in vivo and without the application of a contrast agent. So far, a similar 3-layer structure could only be achieved after Mn2+ administration which enhanced the pyramidal layer 2 and therefore allowed for its separation from the darker layers 1 and 3 (Boretius et al., 2009). Olfactory bulb The olfactory bulb is divided into two distinct structures, the main olfactory bulb and the accessory olfactory bulb. The main
1259
olfactory bulb has a multi-layered cellular architecture: the outer glomerular layer contains neuropil-rich spheroid structures surrounded by a distinctive shell of small neurons and glia cells. The adjacent external plexiform layer is a layer of very dense dendrites and low cell density, whereas the nearby mitral cell layer is densely packed with neurons. The internal plexiform layer is a thin layer with many axons and dendrites but a low density of cells. The inner granule cell layer contains many small granule cell neurons and surrounds the inner white matter fibers of the main olfactory bulb (Paxinos, 2004). Using a spatial resolution of 40 × 40 × 300 μm3 and an echo time of 65 ms these 5 layers could be directly distinguished and identified. Layers with a relatively low cell density such as the external and internal plexiform layer appear bright, whereas more cell-dense layers like the glomerular layer and the granule cell layer present with somewhat brighter intensities. The mitral cell layer, which is mainly arranged as a monolayer of mitral cells with a diameter of 25 to 30 μm in rats (Paxinos, 2004), was visible as a thin dark line. Several MRI studies have addressed the anatomical details of the mouse olfactory bulb using Mn2+ as a contrast agent (Pautler and Koretsky, 2002; Bock et al., 2006; Chuang et al., 2009; Boretius et al., 2009). In contrast to the cerebral cortex, where staining of specific layers requires high doses, the layers of the olfactory bulb become easily visible with low Mn2+ levels (Lee et al., 2005). Noteworthy, this T1 contrast is inverse to that seen on T2-weighted images presented here. For example, layers with high cell density lead to a high uptake of Mn2+ and therefore bright intensities on T1-weighted images, whereas on T2-weighted images such layers appear dark due to their short T2 relaxation times. Hippocampal formation The hippocampal formation comprises three distinct structures: the dentate gyrus, the hippocampus with its three fields CA1 to CA3, and the subiculum. The dentate gyrus has a 3-layer structure comprising the outer relatively cell-free molecular layer, the principal or granule cell layer which is densely packed with columnar stacks of granule cells, and the polymorphic cell layer. All layers could be clearly separated by T2-weighted MRI at 53 × 53 × 234 μm3 resolution and with an echo time of 65 ms or longer (Fig. 9). These parameters also allow for a clear delineation of the dark-appearing, densely packed pyramidal cell layer throughout CA1 to CA3, the brighter relatively cell-free stratum oriens, and the darker myelinated white matter of the alveus and external capsule at the outer border of the hippocampus. Substructures of the hippocampus have mainly been studied with the use of T1-weighted MRI sequences. Especially at low field strengths and even for relatively low spatial resolution (e.g., 117 × 156 × 156 μm3) the shorter T1 relaxation time of the dentate gyrus and parts of the pyramidal cell layer can be exploited by T1weighted MRI to distinguish these structures from surrounding tissue (Natt et al., 2002). The native T1 differences of hippocampal structures can be enforced by administration of Mn2+ (Lee et al., 2005; Watanabe et al., 2002). High-resolution T2-weighted MRI, however, yields additional information and, for example, allows for the complete delineation of the pyramidal cell layer, whereas Mn2+enhanced MRI fails to identify the pyramidal cells within CA1. Cerebellum The cerebellar cortex is organized in 3 cortical layers. Next to white matter, the innermost cell-rich granular layer is followed by the thin Purkinje layer which contains only one type of cell bodies, the large Purkinje cells. The outermost molecular layer consists of interneurons, dendritic arbors of Purkinje neurons, and parallel fiber tracts from the granule cells (Palay and Chan-Palay, 1974). The choice of T2-weighted
1260
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
images at 53 × 53 × 234 μm3 resolution and with echo times of 51 to 65 ms ensured a clear delineation of all 3 layers and a direct correspondence with histology: the molecular layer at intermediate intensities (light-gray), the Purkinje layer as a dark central rim, and the slightly brighter granular layer, which attaches to the darkappearing white matter (Figs. 10 and 11). So far, an in vivo differentiation of cerebellar layers has only been described after systemic Mn2+ administration (Lee et al., 2005; Wadghiri et al., 2004; Watanabe et al., 2002). A Mn2+-induced signal enhancement was observed in the granular cell layer and correlated with granular cell density (Wadghiri et al., 2004). Concluding remarks Optimization of T2-weighted MRI sequences and suitable adjustments for different brain structures resulted in significant improvements for imaging the mouse brain in vivo. Within measuring times of 60 to 90 min cross-sectional images with an in-plane resolution of 30 to 40 μm and a section thickness of 200 to 300 μm unraveled the cellular layers of all major brain structures and even visualized layers known to represent single cells. The MRI approach was strongly supported by a serious attempt to prevent any head motion by adequate fixation and positive pressure ventilation. The brain tissue T2 contrast used here primarily reflects the degree of myelination and the density of cell bodies (Fig. 7). Both pronounced myelination and high cellular density restrict the microscopic and macroscopic mobility of the water molecules detected by MRI. Dipolar relaxation of the water protons then causes a rapid T2 relaxation process, so that respective tissues appear dark on T2-weighted images. Depending on the tissue composition, however, it is obvious that the MRI appearance does not in all cases exhibit a direct correspondence to the structures indicated on conventional histological maps. For example, the 5 layers detected by T2-weighted MRI in the mouse isocortex partly overlap with the 6-layer structure unraveled by Nissl staining. On the other hand, the MRI structures in the olfactory bulb, hippocampal formation, and cerebellum of living mice seem to directly match the structural differences seen in histology. Further improvements of the in vivo MRI resolution and contrast are desirable to allow for an even more detailed access to the brain anatomy of the intact living mouse. References Angenstein, F., Niessen, H.G., Goldschmidt, J., Lison, H., Altrock, W.D., Gundelfinger, E.D., Scheich, H., 2007. Manganese-enhanced MRI reveals structural and functional changes in the cortex of Bassoon mutant mice. Cereb. Cortex 17, 28–36. Badea, A., Ali-Sharief, A.A., Johnson, G.A., 2007. Morphometric analysis of the C57BL/6J mouse brain. NeuroImage 37, 683–693. Barbier, E.L., Marrett, S., Danek, A., Vortmeyer, A., van Gelderen, P., Duyn, J., Bandettini, P., Grafman, J., Koretsky, A.P., 2002. Imaging cortical anatomy by high-resolution MR at 3.0T: detection of the stripe of Gennari in visual area 17. Magn. Reson. Med. 48, 735–738. Benveniste, H., Blackband, S., 2002. MR microscopy and high resolution small animal MRI: applications in neuroscience research. Prog. Neurobiol. 67, 393–420. Bhagat, Y.A., Obenaus, A., Hamilton, M.G., Mikler, J., Kendall, E.J., 2005. Neuroprotection from soman-induced seizures in the rodent: evaluation with diffusion- and T2weighted magnetic resonance imaging. Neurotoxicology 26, 1001–1013. Bock, N.A., Kovacevic, N., Lipina, T.V., Roder, J.C., Ackerman, S.L., Henkelman, R.M., 2006. In vivo magnetic resonance imaging and semiautomated image analysis extend the brain phenotype for cdf/cdf mice. J. Neurosci. 26, 4455–4459. Boretius, S., Michaelis, T., Tammer, R., Tonchev, A., Ashery-Padan, R., Frahm, J., Stoykova, A., 2009. In vivo MRI of altered brain anatomy and fiber connectivity in adult Pax6 deficient mice. Cerebr. Cortex a (electronic publication ahead of print, March 27, 2009) doi:10.1093/cercor/bhp057. Brown, R.H., Walters, D.M., Greenberg, R.S., Mitzner, W., 1999. A method of endotracheal intubation and pulmonary functional assessment for repeated studies in mice. J. Appl. Physiol. 87, 2362–2365. Caviness Jr., V.S., 1975. Architectonic map of neocortex of the normal mouse. J. Comp. Neurol. 164, 247–263.
Chahboune, H., Ment, L.R., Stewart, W.B., Ma, X., Rothman, D.L., Hyder, F., 2007. Neurodevelopment of C57B/L6 mouse brain assessed by in vivo diffusion tensor imaging. NMR Biomed. 20, 375–382. Chuang, K.H., Lee, J.H., Silva, A.C., Belluscio, L., Koretsky, A.P., 2009. Manganese enhanced MRI reveals functional circuitry in response to odorant stimuli. NeuroImage 44, 363–372. Clark, V.P., Courchesne, E., Grafe, M., 1992. In vivo myeloarchitectonic analysis of human striate and extrastriate cortex using magnetic resonance imaging. Cereb. Cortex 2, 417–424. Duyn, J.H., van Gelderen, P., Li, T.Q., de Zwart, J.A., Koretsky, A.P., Fukunaga, M., 2007. High-field MRI of brain cortical substructure based on signal phase. Proc. Natl. Acad. Sci. U. S. A. 104, 11796–11801. Eickhoff, S., Walters, N.B., Schleicher, A., Kril, J., Egan, G.F., Zilles, K., Watson, J.D.G., Amunts, K., 2005. High-resolution MRI reflects myeloarchitecture and cytoarchitecture of the human cerebral cortex. Hum. Brain Map. 24, 206–215. Eidt, S., Kendall, E.J., Obenaus, A., 2004. Neuronal and glial cell populations in the piriform cortex distinguished by using an approximation of q-space imaging after status epilepticus. AJNR 25, 1225–1233. Engelhorn, T., Weise, J., Hammen, T., Bluemcke, I., Hufnagel, A., Doerfler, A., 2007. Early diffusion-weighted MRI predicts regional neuronal damage in generalized status epilepticus in rats treated with diazepam. Neurosci. Lett. 417, 275–280. Gadjanski, I., Boretius, S., Williams, S.K., Lingor, P., Knöferle, J., Sättler, M.B., Fairless, R., Hochmeister, S., Sühs, K.W., Michaelis, T., Frahm, J., Storch, M.K., Bähr, M., Diem, R., in press. Role of N-type voltage dependent calcium channels in autoimmune optic neuritis, Ann. Neurol. 66 (Electronic publication ahead of print, March 23, 2009). doi:10.1002/ana.21668. Kovacevic, N., Henderson, J.T., Chan, E., Lifshitz, N., Bishop, J., Evans, A.C., Henkelman, R.M., Chen, X.J., 2005. A three-dimensional MRI atlas of the mouse brain with estimates of the average and variability. Cereb. Cortex 15, 639–645. Lee, J.H., Silva, A.C., Merkle, H., Koretsky, A.P., 2005. Manganese-enhanced magnetic resonance imaging of mouse brain after systemic administration of MnCl2: dosedependent and temporal evolution of T1 contrast. Magn. Reson. Med. 53, 640–648. Ma, Y., Smith, D., Hof, P.R., Foerster, B., Hamilton, S., Blackband, S.J., Yu, M., Benveniste, H., 2008. In vivo 3D digital atlas database of the adult C57BL/6J mouse brain by magnetic resonance microscopy. Front. Neuroanat. 2, 1. MacKenzie-Graham, A., Lee, E.F., Dinov, I.D., Bota, M., Shattuck, D.W., Ruffins, S., Yuan, H., Konstantinidis, F., Pitiot, A., Ding, Y., Hu, G., Jacobs, R.E., Toga, A.W., 2004. A multimodal, multidimensional atlas of the C57BL/6J mouse brain. J. Anat. 204, 93–102. Merkler, D., Boretius, S., Stadelmann, C., Ernsting, T., Michaelis, T., Frahm, J., Brück, W., 2005. Multicontrast MRI of remyelination in the central nervous system. NMR Biomed. 18, 395–403. Michalke, B., Halbach, S., Nischwitz, V., 2007. Speciation and toxicological relevance of manganese in humans. J. Environ. Monit. 9, 650–656. Natt, O., Watanabe, T., Boretius, S., Radulovic, J., Frahm, J., Michaelis, T., 2002. High-resolution 3D MRI of mouse brain reveals small cerebral structures in vivo. J. Neurosci. Methods 120, 203–209. Palay, S.L., Chan-Palay, V., 1974. Cerebellar Cortex: Cytology and Organization. Springer, Berlin. Pautler, R.G., Koretsky, A.P., 2002. Tracing odor-induced activation in the olfactory bulbs of mice using manganese-enhanced magnetic resonance imaging. NeuroImage 16, 441–448. Paxinos, G., 2004. The Rat Nervous System, 3rd ed. Elsevier, Sydney. Peters, A., Jones, E.G., 1984. Cerebral Cortex. Plenum, New York. Polleux, F., Dehay, C., Kennedy, H., 1997. The timetable of laminar neurogenesis contributes to the specification of cortical areas in mouse isocortex. J. Comp. Neurol. 385, 95–116. Sharief, A.A., Johnson, G.A., 2006. Enhanced T2 contrast for MR histology of the mouse brain. Magn. Reson. Med. 56, 717–725. Silva, A.C., Lee, J.H., Wu, C.W., Tucciarone, J., Pelled, G., Aoki, I., Koretsky, A.P., 2008. Detection of cortical laminar architecture using manganese-enhanced MRI. J. Neurosci. Methods 167, 246–257. Takeda, A., Sawashita, J., Okada, S., 1995. Biological half-lives of zinc and manganese in rat brain. Brain Res. 695, 53–58. Wadghiri, Y.Z., Blind, J.A., Duan, X., Moreno, C., Yu, X., Joyner, A.L., Turnbull, D.H., 2004. Manganese-enhanced magnetic resonance imaging (MEMRI) of mouse brain development. NMR Biomed. 17, 613–619. Walters, N.B., Egan, G.F., Kril, J.J., Kean, M., Waley, P., Jenkinson, M., Watson, J.D., 2003. In vivo identification of human cortical areas using high-resolution MRI: an approach to cerebral structure–function correlation. Proc. Natl. Acad. Sci. U. S. A. 100, 2981–2986. Watanabe, T., Natt, O., Boretius, S., Frahm, J., Michaelis, T., 2002. In vivo 3D MRI staining of mouse brain after subcutaneous application of MnCl2. Magn. Reson. Med. 48, 852–859. Yamamori, T., Rockland, K.S., 2006. Neocortical areas, layers, connections, and gene expression. Neurosci. Res. 55, 11–27. Yokel, R.A., Crossgrove, J.S., 2004. Manganese toxicokinetics at the blood-brain barrier. Res. Rep. Health Eff. Inst. 7–58 discussion 59–73. Zhang, K., Sejnowski, T.J., 2000. A universal scaling law between gray matter and white matter of cerebral cortex. Proc. Natl. Acad. Sci. U. S. A. 97, 5621–5626. Zilles, K., Palomero-Gallagher, N., 2001. Cyto-, myelo-, and receptor architectonics of the human parietal cortex. NeuroImage 14, S8–20.
Contents lists available at ScienceDirect
NeuroImage j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / y n i m g
MRI of cellular layers in mouse brain in vivo Susann Boretius a,⁎, Lars Kasper a, Roland Tammer a,b, Thomas Michaelis a, Jens Frahm a,b a b
Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut für biophysikalische Chemie, 37070 Göttingen, Germany DFG Center of Molecular Physiology of the Brain, 37073 Göttingen, Germany
a r t i c l e
i n f o
Article history: Received 6 April 2009 Revised 6 May 2009 Accepted 29 May 2009 Available online 8 June 2009 Keywords: High-resolution MRI Mouse Cortical layer Brain
a b s t r a c t Noninvasive imaging of the brain of animal models demands the detection of increasingly smaller structures by in vivo MRI. The purpose of this work was to elucidate the spatial resolution and structural contrast that can be obtained for studying the brain of C57BL/6J mice by optimized T2-weighted fast spin-echo MRI at 9.4 T. As a prerequisite for high-resolution imaging in vivo, motion artifacts were abolished by combining volatile anesthetics and positive pressure ventilation with a specially designed animal bed for fixation. Multiple substructures in the cortex, olfactory bulb, hippocampus, and cerebellum were resolved at 30 to 40 μm in-plane resolution and 200 to 300 μm section thickness as well as for relatively long echo times of 65 to 82 ms. In particular, the approach resulted in the differentiation of up to five cortical layers. In the olfactory bulb the images unraveled the mitral cell layer which has a thickness of mostly single cells. In the hippocampus at least five substructures could be separated. The molecular layer, Purkinje layer, and granular layer of the cerebellum could be clearly differentiated from the white matter. In conclusion, even without the use of a contrast agent, suitable adjustments of a widely available T2-weighted MRI sequence at high field allow for structural MRI of living mice at near single-cell layer resolution. © 2009 Elsevier Inc. All rights reserved.
Introduction In recent years MRI has developed into an indispensable research tool for noninvasive imaging of the brain. Complementing histological studies, in vivo MRI offers the detection of more global changes in intact animals but does not suffer from shrinking or other preparation artifacts and allows for a sequential follow-up of the same animal. In addition, because most MRI techniques can be performed in both humans and experimental animals, the method provides a missing link in translational research. In this context, the successful use of (genetically modified) mouse models enhances the demand for detecting increasingly smaller brain structures by in vivo MRI. In terms of absolute spatial resolution this corresponds to isotropic image voxel dimensions of less than 100 μm or voxel sizes of less than 1 nl. In addition, the discrimination of specific structures requires sufficient contrast, or more precisely a certain contrast-to-noise ratio (CNR) that also takes into account the signal-to-noise ratio (SNR) and available measurement time. In fact, because a reduction of the voxel volume in MRI results in a linear decrease of the SNR, the reduced partial volume effect and better contrast between adjacent structures at high resolution might be compromised by a lower SNR. Similarly, a better tissue contrast based on a more pronounced relaxation time weighting is also at the expense of SNR due to more pronounced T1 saturation or T2 attenuation.
⁎ Corresponding author. Fax: +49 511 201 1307. E-mail address: [email protected] (S. Boretius). 1053-8119/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2009.05.095
Recently, several attempts have been made for high-resolution imaging of the cortical anatomy in humans (Barbier et al., 2002, Walters et al., 2003) advancing the spatial resolution to 240 × 240 × 1000 μm3 (Duyn et al., 2007). In mouse brain, however, high-resolution MRI at 20 to 60 μm isotropic resolution (corresponding to voxel sizes of 8 pl to 0.2 nl) has so far only been achieved ex vivo using sacrificed animals (Badea et al., 2007; Benveniste and Blackband, 2002; Kovacevic et al., 2005). Although of basic value in closely linking MRI to histological assessments, such studies do not fully exploit the specific advantages of in vivo MRI over conventional histology. Nevertheless, only few MRI studies attempted to visualize cortical substructures in living mice (Angenstein et al., 2007; Chahboune et al., 2007; Lee et al., 2005; MacKenzie-Graham et al., 2004). A first challenge stems from movements due to spontaneous breathing (Ma et al., 2008) which bears the risk of impairing or even precluding access to microscopic resolution. Continuous endotracheal intubation and positive pressure ventilation may help to overcome this problem (Brown et al., 1999) and has been shown to allow for repeated measurements of mice, each with a total duration of several hours (Merkler et al., 2005; Boretius et al., 2009). A second difficulty stems from the relatively small amount of white matter (Zhang and Sejnowski, 2000). In humans, the many strongly myelinated structures are particularly useful for mapping the brain into anatomically and functionally distinct areas. For example, in the striate cortex T1weighted MRI at high spatial resolution reveals the stria of Gennari (Barbier et al., 2002; Clark et al., 1992). The aim of this study was to explore the practical limits of in vivo MRI of mouse brain with respect to resolution and (native) contrast
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
that can be achieved by suitable MRI sequences on commercially available high-field MRI systems and within reasonable measuring times. These techniques were exploited for the detection of cerebral layers in normal mice, where we were able to achieve near single-cell layer resolution in several brain structures. Methods
1253
Table 1 In vivo protocols for the identification of mouse brain structures using fast spin-echo MRI at 9.4 T. Brain structure
Resolution/μm
Echo time/ms
Measurement time/min
Cerebral cortex Olfactory bulb Hippocampal formation Cerebellum
30 × 30 × 300 40 × 40 × 300 53 × 53 × 234 53 × 53 × 234
65–82 65 65–82 51–65
90 60 60 60
Animals All experiments were performed in compliance with relevant laws and institutional guidelines. Animal experiments were approved by local authorities. For MRI adult C57BL/6J mice (n = 4) were initially anesthetized using a chamber pervaded with 5% isoflurane in oxygen. Subsequently, the mice were intubated and kept under anesthesia with 1 to 1.5% isoflurane in a 1:1.5 mixture of oxygen and ambient air. Respiration was monitored by a pressure transducer under the ventral thoraco-abdominal region. The animals were placed in a prone position with the head fixed to a teeth and palate holder as demonstrated in Fig. 1. This specially designed Göttingen animal bed allowed for a reproducible and reliable fixation of the mouse head and mounted receive coil in the isocenter of the magnet. The rectal temperature was held constant at 36 ± 0.5 °C by temperaturecontrolled water flowing through pipes that form a chamber. MRI MRI studies were performed at 9.4 T (Bruker Biospin, Ettlingen, Germany) with the use of a birdcage resonator (inner diameter 70 mm) for excitation and a saddle-shaped phased-array surface coil (4 elements) for signal detection (Bruker Biospin, Ettlingen, Germany). T2-weighted images were acquired with a multislice fast spinecho MRI sequence (repetition time TR = 4200 ms, 8 differently phase-encoded echoes, number of slices 14–20, matrix size 512 × 512). Experiments involved four different (mean) echo times TE and acquisitions at different spatial resolutions. Typically, the in-plane resolution varied between 30 and 80 μm, while the section thickness was about 200 to 300 μm (see below and Table 1). The smallest field of view was 15.4 × 15.4 mm2 covering a full brain section. The receiver bandwidth was adapted to the respective echo time yielding pairs of
36 ms/100 kHz, 51 ms/50 kHz, 65 ms/35 kHz, and 82 ms/25 kHz. To minimize chemical shift artifacts and in particular to avoid any contamination of the cortex by overlapping signals from subcutaneous fat, the orientation of the readout gradient was chosen such as to shift lipid signals into an anterior direction. Finally, to assess the potential of high-resolution T1-weighted MRI, manganese-enhanced 3D gradient-echo MRI was acquired 24 h after subcutaneous injection of 40 mg/kg MnCl2 using a 3D FLASH sequence (TR/TE = 17.0/3.8 ms, flip angle 25°) at 30 × 30 × 300 μm3 resolution. In all cases, images were calculated without zero filling, so that voxel dimensions refer to truly acquired resolutions. Cortical profiles To directly compare the cortical lamination pattern seen on T2weighted images with the underling cyto- and myeloarchitecture, MRI intensity profiles in the parietal cortex were extracted from an axial T2-weighed image as defined in Fig. 2. Similarly, profiles from corresponding histological sections were obtained for myelin, cresyl violet, and Nissl stains using the open source software ImageJ (version 1.34s, http://rsb.info.nih.gov/ij/). For this purpose the histological sections were mildly smoothed by a Gaussian filter to improve the visualization of the lamination pattern. For each technique 5 intensity profiles were normalized according to their length and averaged. To account for intermodal differences in the intensity and gray value ranges all profiles were z-normalized (Eickhoff et al., 2005). Results Experimental adjustments of a fast spin-echo MRI sequence substantially improved the detection of several small brain structures
Fig. 1. The Göttingen animal bed for in vivo MRI of rodent brain. (A) Entire setup as well as magnified views of (B) the animal bed and (C) the mouse head without receive coil. 1 = magnet bore, 2 = gliding channel, 3 = animal bed, 4 = supply lines of anesthetic gas, 5 = supply lines of warm water, 6 = clip for fixation of the surface coil, 7 = teeth and palate holder, 8 = surface coil, 9 = temperature-controlled chamber, 10 = endotracheal tube.
1254
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
in living mice. The resulting main parameters for resolving different cellular layers in the cerebral cortex, olfactory bulb, hippocampus, and cerebellum are summarized in Table 1. These general choices with respect to resolution and T2 contrast are supported by Fig. 3 comparing horizontal T2-weighted images of the mouse brain at different in-plane resolutions (30 vs 80 μm) and echo times (36 vs 82 ms). In general, the in-plane resolution turns out to be the most important parameter for a delineation of small brain structures such as the myelinated striatal fibers, while a strengthening of the T2 contrast by a prolongation of the echo time improves the visibility of substructures in the cortex. Cerebral cortex Fig. 2. Definition of intensity profiles as a function of cortical depth. (Left) Axial T2weighted image of the cerebral cortex of a mouse brain in vivo and (right) magnified view of the analyzed region (white rectangle). Profiles cover the cerebral cortex from the surface (cortical depth 0%) to the white matter (wm, cortical depth 100%).
Fig. 3. Horizontal T2-weighted images of a mouse brain in vivo (section thickness 300 μm) as a function of (top) in-plane resolution (80 vs 30 μm) at TE = 65 ms and (bottom) echo time (36 vs 82 ms) at 30 μm in-plane resolution. For details see text.
Best results for a visualization of cortical layers were achieved at 30 to 40 μm in-plane resolution and 65 to 82 ms echo time. As shown in Fig. 4 (top right), a resolution of 40 × 40 × 300 μm3 (TE = 65 ms) unravels up to 5 different layers in the cerebral cortex of living mice. They resemble – but not exactly correspond to – the cortical layers seen in histological sections by light microscopy (Fig. 4, bottom left). In technical terms, the most important aspect was a high in-plane resolution. This observation was supported by the fact that the layer differentiation was impaired when reducing the slice thickness at the expense of the in-plane resolution while keeping the absolute voxel size constant (data not shown). The influence of the echo time or T2 contrast on the layer detection is shown in Fig. 5 for two horizontal sections at 30 × 30 × 300 μm3 resolution. Whereas no layer-like structures are visible at an echo time of 36 ms, echo times of 65 or 82 ms allow for the identification of up to 5 different layers. However, the optimal echo time depends on the cortical area of interest. For instance, in a more dorsal section (Fig. 5, top) covering the primary auditory cortex and the dorsal area of the secondary auditory cortex, echo times of higher than 65 ms do not improve the situation. In contrast, in a more ventral horizontal section (Fig. 5, bottom) depicting the secondary auditory cortex, an echo time of 82 ms leads to a better differentiation.
Fig. 4. Magnified horizontal T2-weighted images of the cerebral cortex of a mouse brain in vivo as a function of in-plane resolution at TE = 65 ms and 300 μm section thickness. Layer-like structures in the cortex (yellow lines) are compared to histologically defined layers (black lines) in a corresponding Nissl stain indicating areas of different neuronal density (bottom left, http://www.brainmaps.org/).
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
1255
Fig. 5. Magnified views of two horizontal T2-weighted sections (top, bottom) of the cerebral cortex of a mouse brain in vivo as a function of echo time at 30 × 30 × 300 μm3 resolution. Layer-like structures in the cortex (yellow lines) are compared to histologically defined layers (black lines) in corresponding cresyl violet stains indicating areas of different neuronal density (right, http://www.mbl.org/).
Fig. 6. Magnified axial T2-weighted images of the cortex of a mouse brain in vivo at (A) 40 × 40 × 300 μm3 resolution (TE = 65 ms) and (B) 30 × 30 × 300 μm3 resolution (TE = 82 ms) in comparison to (C) a Nissl stain indicating neuronal density (http://mouse.brain-map.org/atlas/ARA/Coronal/) and (D) a myelin stain (http://www.hms.harvard.edu/research/ brain/). (E) Magnified view of image (A) indicating MRI-detectable T2 layers a to e in the cortex (yellow lines). (F) Overlay of part of image (E) onto the Nissl stain indicating histologically defined layers I to VI. (G) Overlay of part of image (E) on the myelin stain. 1,2,3 = layers in piriform cortex; bfd = barrel field; fl = forelimb area; hl = hindlimb area; p1, p2 = areas 1 and 2 in parietal cortex.
1256
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
layer). These cortical laminae are best seen in areas 1 and 2 of the parietal cortex, while a layer separation was not possible in sensorimotor areas representing the forelimb and hindlimb. When directly comparing the T2-weighted images to the corresponding histological sections by partial overlay (Figs. 6F, G), the brighter outer lamina (T2 layer a) seems to correspond to cortical layers I and II, the next darker lamina (T2 layer b) to layers III and IV, the inner brighter lamina (T2 layer c) to layer V, and the inner darker and brighter laminae (T2 layers d and e) to layer VI. The MRI identification of a most inner bright lamina (T2 layer e) is supported by an overlay with the Nissl stain (Fig. 6F) and probably indicates a part of cortical layer VI with lowered cell density (see also the cresyl violet staining in Fig. 5). Finally, at higher spatial resolution and longer echo time (Fig. 6B) the observation of an additional dark substructure in only a part of the cortex most likely refers to the barrel field which is known to exhibit a high neuronal density. Furthermore, the chosen experimental conditions allow for the detection of 3 different layers (1 to 3) in the piriform cortex (Figs. 6B, C). The relationship between MRI and underlying histology is further illustrated in Fig. 7 depicting cortical profiles for myelin, T2-weighted MRI signal intensity, and cellular density as reflected by cresyl violet and Nissl stains. This comparison is in general agreement with the findings of Fig. 6 and suggests that the T2 contrast is mainly determined by the degree of myelination and the cell density. In particular, myelination is low at the cortical surface (high z-score) and increases with cortical depth (lower z-scores). This observation parallels the general trend seen for T2-weighted MRI signal intensities. However, their z-scores reveal an additional modulation which resembles the pattern found for cell density. For example, the
Fig. 7. Mean intensity profiles of the parietal cortex (compare Fig. 2) from (top) a myelin stain, (middle) a T2-weighted (T2w) image of a mouse brain in vivo, and (bottom) a cresyl violet as well as Nissl stain reflecting cell density. To allow for a comparison of profiles from different techniques respective intensities are z-normalized and the distance from the cortical surface is given in percent of cortical depth.
Fig. 6 demonstrates the appearance of MRI-detectable cortical layers in axial sections at 40 × 40 × 300 μm 3 resolution with TE = 65 ms (Fig. 6A) and at 30 × 30 × 300 μm3 resolution with TE = 82 ms (Fig. 6B). The images are compared to a Nissl stain (Fig. 6C) reflecting neuronal density and a myelin stain (Fig. 6D). In a magnified view of the cortex (Fig. 6E) the above mentioned 5 MRI layers are denoted as T2 layers a (most outer layer) to e (most inner
Fig. 8. Magnified horizontal T2-weighted images of the olfactory bulb of a mouse brain in vivo as a function of echo time at 30 × 30 × 300 μm3 resolution indicating access to the mitral layer (yellow arrows). Bottom row: Magnified view and corresponding Nissl stain (http://www.brainmaps.org/). aob = accessory olfactory bulb; epl = external plexiform layer; gl = glomerular layer; gr = granule layer; mi = mitral layer; ipl = inner plexiform layer; wm = white matter.
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
initial sharper decline than that of the myelin z-scores may be caused by the steep decrease of the z-scores for the cell density. Moreover, the relatively strong increase of the z-score for T2-weighted MRI in the region of layer V may result from both the reduced myelination and the much lower cell density than in the neighboring regions IV and VI. Olfactory bulb To distinguish the different layers of the olfactory bulb a minimum in-plane resolution of 40 μm was required. In addition, Fig. 8 demonstrates that an echo time of at least 51 ms is necessary in order to achieve sufficient contrast. This especially applies to a discrimination of the mitral cell layer and the external and internal plexiform layers. With these MRI parameters the granular layer appears as a gray–black layer that may be well distinguished from the brighter external plexiform layer. The mitral cell layer, which in histology mainly presents as a cellular monolayer, exhibits low signal intensity in T2-weighted images because of its dense packing which opposes the bright signal from the internal plexiform layer (Fig. 8, bottom). Hippocampal formation To differentiate the molecular layer, granule cell layer, and polymorphic cell layer of the dentate gyrus, T2-weighted MRI required a resolution of 53 × 53 × 243 μm3 or higher and a strong T2 contrast with echo times of 65 ms or longer. As shown in Fig. 9 an echo time of 36 ms did not allow for a proper separation of the molecular layer and granule layer which becomes possible at 65 ms. In addition, at an echo time of 82 ms (Fig. 9, bottom left), the polymorphic layer and granule cell layer may be identified. These MRI parameters also delineate the pyramidal cell layer (in CA1 to CA3), the stratum oriens, and the white matter fibers of the alveus and external capsule which mark the outer border of the hippocampus. Based on pronounced T2 contrast, the
1257
densely packed pyramidal cell layer appears dark, the relatively cellfree stratum oriens bright, and the myelinated fibers dark. Noteworthy, manganese-enhanced MRI reveals complementary information (Fig. 9, bottom right). The approach exploits the fact that Mn2+ ions behave like Ca2+ ions because of their similar valence and size. It has been shown that Mn2+ ions can enter neurons via voltagegated calcium channels (Yokel and Crossgrove, 2004, Gadjanski et al., in press). With manganese uptake and subsequent axonal transport, the polymorphic and granule cell layers of the dentate gyrus as well as the pyramidal cells of the CA3 region experience a strong signal enhancement, whereas the pyramidal cells of the CA1 region remain undetectable. In contrast, the T2-weighted images allow for a visualization of the entire pyramidal cell layer due to their dependence on cellular density. On the other hand, the stratum radiatum and stratum lacunosum-moleculare may be distinguished using manganese-enhanced MRI, but not with native T2 contrast at high spatial resolution. Cerebellum Horizontal T2-weighted images of the cerebellum at different echo times and spatial resolutions are shown in Figs. 10 and 11 respectively. Echo times of 36 ms and slightly better of 51 ms provide good contrast between short-T2 structures such as white matter and Purkinje cells (dark) and those characterized by longer T2 relaxation times like the granular layer (bright). The molecular layer with a low cellular density (and long T2 relaxation time) exhibits an even higher MRI signal. The longest echo time of 82 ms results in very similar signal intensities (low contrast) for white matter, granular cell layer, and Purkinje cell layer. Fig. 11 demonstrates a slightly different behavior for resolving substructures of the cerebellum as found for most other brain structures. For example, even at 80 × 80 × 300 μm3 resolution the light-gray molecular layer may be clearly detected (data not shown).
Fig. 9. Magnified horizontal T2-weighted images of the hippocampal formation of a mouse brain in vivo as a function of echo time at 30 × 30 × 300 μm3 resolution and in comparison to a corresponding Giemsa stain (http://www.brainmaps.org/) and a manganese-enhanced T1-weighted image (Mn2+). gldg = granule cell layer; hf = hippocampal fimbria; mlgd = molecular layer; podg = polymorphic layer; s = subiculum; so = stratum oriens; sr = stratum radiatum; slm = stratum lacunosum-moleculare; wm = white matter.
1258
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
Fig. 10. Magnified horizontal T2-weighted images of a mouse cerebellum in vivo as a function of echo time at 30 × 30 × 300 μm3 resolution.
Increasing the spatial resolution to 53 μm or 40 μm allows for the additional delineation of the thin and dark Purkinje layer, the slightly brighter granular layer, and the dark white matter. A reduction in slice thickness from 300 to 234 μm further improved the delineation. In contrast, acquisitions at even higher spatial resolution were partially comprised by reduced SNR and enhanced motion sensitivity due to its close spatial relationship to the medulla oblongata eventually affected by respiratory movements. Discussion
cortical area, layers I and II and in part also layer III are poorly myelinated (Paxinos, 2004). Considering cell density, degree of myelination, and proportional size of the cortical layer, the bright appearance of the outer T2 layer a (Fig. 6E) most likely corresponds to cortical layers I and II. The low cellular density and the sparse myelination both support a relatively long T2 relaxation time that leads to bright intensities on T2-weighted images. This interpretation is further supported by a comparison of intensity profiles that have been obtained across the parietal cortex for T2-weighted MRI and histology (Fig. 7). In the area of layers I and II and partly of layer III the slope of the decreasing z-score for T2-weighted MRI is larger than that of the slowly increasing myelination, but smaller than that of the rapidly decreasing cell density. Thus, the area of T2 layer a corresponds to the cortical layer I and partly also to layers II–III (Fig. 7). The cell-rich internal granular layer IV is most prominent in sensory areas and poorly developed or even absent in areas with motor function. The high cell density and the strong myelination of layer IV (and partly also of layer III) cause a relatively short T2 relaxation time. This behavior leads to low intensities in T2-weighted images, so that the first dark band in areas 1 and 2 of the parietal cortex denoted T2 layer b (Fig. 6E) best matches with cortical layers III and IV (Figs. 6F and 7). Another cell-rich region and correspondingly dark zone in area 1 most likely belongs to the barrel field of layer IV (Fig. 6B). The internal pyramidal layer V contains medium to large pyramidal neurons. Especially in area 1 of parietal cortex, layer V consists of fewer cells compared to neighboring layers IV and VI and therefore presents with a brighter intensity on T2-weigted images. Its main part may be identified with T2 layer c (Figs. 6E and 7). In the forelimb and hindlimb areas of the parietal cortex, however, layer V exhibits more densely packed pyramidal cells and can hardly be separated from layers IV and VI, neither in the Nissl stain (Fig. 6C) nor on T2-weighted MRI (Fig. 6B).
High-resolution T2-weighted MRI at 9.4 T allows for the identification of multiple cellular layers in the cerebral cortex, olfactory bulb, hippocampus, and cerebellum of intact living mice. In some areas this capability includes the detection of layers with predominantly singlecell thickness — without any application of a contrast agent and within reasonable measurement times of 60 to 90 min. However, depending on the underlying contrast properties of the various tissue components, the structures differentiated by MRI do not necessarily correlate with the structures seen in histology. Pertinent relationships will be discussed below to better understand the MRI correlates of conventional histological stains. Isocortex The mammalian isocortex shows a lamination into 6 layers (I to VI) that are histologically well described, for example for humans (Yamamori and Rockland, 2006; Zilles and Palomero-Gallagher, 2001) and mice (Caviness, 1975; Polleux et al., 1997). While the separation of different layers is mainly based on their cyto- and myeloarchitecture, the layers may be further characterized by a – functionally relevant – receptor expression and chemical composition (Paxinos, 2004). Moreover, size and cell density as well as the degree of myelination of individual layers differ between different cortical areas. The outermost cortical layer, that is molecular layer I, contains only few scattered neurons and consists mainly of extensions of apical dendrites and horizontally-oriented axons. In contrast, the external granular layer II comprises small pyramidal neurons and numerous stellate neurons (Paxinos, 2004; Peters and Jones, 1984). Layer III, the external pyramidal layer, reveals a nearly similar cell density and is predominantly built of small and medium-sized pyramidal neurons as well as of non-pyramidal neurons. Moreover, independent of the
Fig. 11. Magnified horizontal T2-weighted images of a mouse cerebellum in vivo as a function of spatial resolution at TE = 65 ms and in comparison to a corresponding Giemsa stain (http://www.brainmaps.org/). gl = granular layer; ml = molecular layer; pl = Purkinje cells; wm = white matter.
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
The inner multiform layer VI contains few large pyramidal neurons and many small spindle-like pyramidal and multiform neurons. On T2-weighted MRI layer VI mainly appears dark and especially in areas 1 and 2 may be well separated from the brighter layer V. However, a comparison of cortical intensity profiles reveals an underestimation of the thickness of layer V by the corresponding T2 layer c (Fig. 7). In fact, T2 layer d seems to reflect a transition zone including parts of layers V and VI. The T2 layer d can be further distinguished from an inner brighter lamina denoted T2 layer e (Fig. 6E) that seems to reflect the lower cell density in the innermost part of layer VI as indicated by a cresyl violet and a Nissl stain (compare Figs. 5 and 7) and the slightly reduced myelin content (higher z-score in Fig. 7). Previous MRI studies at ultrahigh 43 μm isotropic resolution described the layers in the cortex of formalin-fixed mouse brains also using T2-weighted MRI at 9.4 T (Sharief and Johnson, 2006; Badea et al., 2007). These in situ MRI reports distinguished 5 cortical T2 layers of alternating brightness which seem to be in close agreement with the findings presented here. Though studies of fixed brains pose no limitations with respect to residual motion or measuring time and certainly offer highest resolution, it should be kept in mind that the MRI contrasts in living and dead tissue are sensitive to a variety of mechanisms including temperature, intra- and extracellular compartmentalization, oxygenation, and water content. In fact, all these parameters are known to change during formalin fixation. Alternatively, the availability of native MRI contrasts may be enhanced by the use of contrast agents. In small animal research, an attractive tool is the application of Mn2+ as an analogue to Ca2+. In mice systemic application of high doses of MnCl2 revealed an unequal distribution of Mn2+ ions – as evidenced by a corresponding Mn2+induced signal enhancement – over the cortical laminae (Angenstein et al., 2007; Lee et al., 2005), while rats presented with a signal increase in layers II and V but not in layers I and III (Silva et al., 2008). Apart from the principal possibility to address functional aspects of cortical organization, disadvantages of the Mn2+ approach are its neurotoxicity (Michalke et al., 2007) and the long retention time in the brain with a biological half-life of 51 to 74 days (Takeda et al., 1995), which hampers follow-up studies in the same animal. Piriform cortex In contrast to the 6-layered isocortex the piriform cortex consists of only three layers. Layer 1 mainly contains fibers and apical dendrites from the underlying layer 2 which consists of densely packed pyramidal cells. In comparison to layer 2, layer 3 is a moderately dense cell layer comprising intrinsic fibers though to a lesser extent than layer I (Paxinos, 2004). According to its low cell density the outermost layer 1 causes a long T2 relaxation time and appears bright on T2-weighted images (Fig. 6B). In contrast, the cell-rich layer 2 is characterized by a shorter T2 relaxation time and a dark appearance which clearly delimits the less cell-dense layer 3 with its brighter intensity. The best separation of these layers is achieved with an inplane resolution of 30 μm and an echo time of 82 ms (Fig. 6B). Previous MRI studies of the piriform cortex of rodents have been performed in animal models of epilepsy using T2-weighted imaging (Bhagat et al., 2005) and diffusion-weighted imaging (Eidt et al., 2004; Engelhorn et al., 2007). To the best of our knowledge, however, this is the first demonstration of the piriform layers in vivo and without the application of a contrast agent. So far, a similar 3-layer structure could only be achieved after Mn2+ administration which enhanced the pyramidal layer 2 and therefore allowed for its separation from the darker layers 1 and 3 (Boretius et al., 2009). Olfactory bulb The olfactory bulb is divided into two distinct structures, the main olfactory bulb and the accessory olfactory bulb. The main
1259
olfactory bulb has a multi-layered cellular architecture: the outer glomerular layer contains neuropil-rich spheroid structures surrounded by a distinctive shell of small neurons and glia cells. The adjacent external plexiform layer is a layer of very dense dendrites and low cell density, whereas the nearby mitral cell layer is densely packed with neurons. The internal plexiform layer is a thin layer with many axons and dendrites but a low density of cells. The inner granule cell layer contains many small granule cell neurons and surrounds the inner white matter fibers of the main olfactory bulb (Paxinos, 2004). Using a spatial resolution of 40 × 40 × 300 μm3 and an echo time of 65 ms these 5 layers could be directly distinguished and identified. Layers with a relatively low cell density such as the external and internal plexiform layer appear bright, whereas more cell-dense layers like the glomerular layer and the granule cell layer present with somewhat brighter intensities. The mitral cell layer, which is mainly arranged as a monolayer of mitral cells with a diameter of 25 to 30 μm in rats (Paxinos, 2004), was visible as a thin dark line. Several MRI studies have addressed the anatomical details of the mouse olfactory bulb using Mn2+ as a contrast agent (Pautler and Koretsky, 2002; Bock et al., 2006; Chuang et al., 2009; Boretius et al., 2009). In contrast to the cerebral cortex, where staining of specific layers requires high doses, the layers of the olfactory bulb become easily visible with low Mn2+ levels (Lee et al., 2005). Noteworthy, this T1 contrast is inverse to that seen on T2-weighted images presented here. For example, layers with high cell density lead to a high uptake of Mn2+ and therefore bright intensities on T1-weighted images, whereas on T2-weighted images such layers appear dark due to their short T2 relaxation times. Hippocampal formation The hippocampal formation comprises three distinct structures: the dentate gyrus, the hippocampus with its three fields CA1 to CA3, and the subiculum. The dentate gyrus has a 3-layer structure comprising the outer relatively cell-free molecular layer, the principal or granule cell layer which is densely packed with columnar stacks of granule cells, and the polymorphic cell layer. All layers could be clearly separated by T2-weighted MRI at 53 × 53 × 234 μm3 resolution and with an echo time of 65 ms or longer (Fig. 9). These parameters also allow for a clear delineation of the dark-appearing, densely packed pyramidal cell layer throughout CA1 to CA3, the brighter relatively cell-free stratum oriens, and the darker myelinated white matter of the alveus and external capsule at the outer border of the hippocampus. Substructures of the hippocampus have mainly been studied with the use of T1-weighted MRI sequences. Especially at low field strengths and even for relatively low spatial resolution (e.g., 117 × 156 × 156 μm3) the shorter T1 relaxation time of the dentate gyrus and parts of the pyramidal cell layer can be exploited by T1weighted MRI to distinguish these structures from surrounding tissue (Natt et al., 2002). The native T1 differences of hippocampal structures can be enforced by administration of Mn2+ (Lee et al., 2005; Watanabe et al., 2002). High-resolution T2-weighted MRI, however, yields additional information and, for example, allows for the complete delineation of the pyramidal cell layer, whereas Mn2+enhanced MRI fails to identify the pyramidal cells within CA1. Cerebellum The cerebellar cortex is organized in 3 cortical layers. Next to white matter, the innermost cell-rich granular layer is followed by the thin Purkinje layer which contains only one type of cell bodies, the large Purkinje cells. The outermost molecular layer consists of interneurons, dendritic arbors of Purkinje neurons, and parallel fiber tracts from the granule cells (Palay and Chan-Palay, 1974). The choice of T2-weighted
1260
S. Boretius et al. / NeuroImage 47 (2009) 1252–1260
images at 53 × 53 × 234 μm3 resolution and with echo times of 51 to 65 ms ensured a clear delineation of all 3 layers and a direct correspondence with histology: the molecular layer at intermediate intensities (light-gray), the Purkinje layer as a dark central rim, and the slightly brighter granular layer, which attaches to the darkappearing white matter (Figs. 10 and 11). So far, an in vivo differentiation of cerebellar layers has only been described after systemic Mn2+ administration (Lee et al., 2005; Wadghiri et al., 2004; Watanabe et al., 2002). A Mn2+-induced signal enhancement was observed in the granular cell layer and correlated with granular cell density (Wadghiri et al., 2004). Concluding remarks Optimization of T2-weighted MRI sequences and suitable adjustments for different brain structures resulted in significant improvements for imaging the mouse brain in vivo. Within measuring times of 60 to 90 min cross-sectional images with an in-plane resolution of 30 to 40 μm and a section thickness of 200 to 300 μm unraveled the cellular layers of all major brain structures and even visualized layers known to represent single cells. The MRI approach was strongly supported by a serious attempt to prevent any head motion by adequate fixation and positive pressure ventilation. The brain tissue T2 contrast used here primarily reflects the degree of myelination and the density of cell bodies (Fig. 7). Both pronounced myelination and high cellular density restrict the microscopic and macroscopic mobility of the water molecules detected by MRI. Dipolar relaxation of the water protons then causes a rapid T2 relaxation process, so that respective tissues appear dark on T2-weighted images. Depending on the tissue composition, however, it is obvious that the MRI appearance does not in all cases exhibit a direct correspondence to the structures indicated on conventional histological maps. For example, the 5 layers detected by T2-weighted MRI in the mouse isocortex partly overlap with the 6-layer structure unraveled by Nissl staining. On the other hand, the MRI structures in the olfactory bulb, hippocampal formation, and cerebellum of living mice seem to directly match the structural differences seen in histology. Further improvements of the in vivo MRI resolution and contrast are desirable to allow for an even more detailed access to the brain anatomy of the intact living mouse. References Angenstein, F., Niessen, H.G., Goldschmidt, J., Lison, H., Altrock, W.D., Gundelfinger, E.D., Scheich, H., 2007. Manganese-enhanced MRI reveals structural and functional changes in the cortex of Bassoon mutant mice. Cereb. Cortex 17, 28–36. Badea, A., Ali-Sharief, A.A., Johnson, G.A., 2007. Morphometric analysis of the C57BL/6J mouse brain. NeuroImage 37, 683–693. Barbier, E.L., Marrett, S., Danek, A., Vortmeyer, A., van Gelderen, P., Duyn, J., Bandettini, P., Grafman, J., Koretsky, A.P., 2002. Imaging cortical anatomy by high-resolution MR at 3.0T: detection of the stripe of Gennari in visual area 17. Magn. Reson. Med. 48, 735–738. Benveniste, H., Blackband, S., 2002. MR microscopy and high resolution small animal MRI: applications in neuroscience research. Prog. Neurobiol. 67, 393–420. Bhagat, Y.A., Obenaus, A., Hamilton, M.G., Mikler, J., Kendall, E.J., 2005. Neuroprotection from soman-induced seizures in the rodent: evaluation with diffusion- and T2weighted magnetic resonance imaging. Neurotoxicology 26, 1001–1013. Bock, N.A., Kovacevic, N., Lipina, T.V., Roder, J.C., Ackerman, S.L., Henkelman, R.M., 2006. In vivo magnetic resonance imaging and semiautomated image analysis extend the brain phenotype for cdf/cdf mice. J. Neurosci. 26, 4455–4459. Boretius, S., Michaelis, T., Tammer, R., Tonchev, A., Ashery-Padan, R., Frahm, J., Stoykova, A., 2009. In vivo MRI of altered brain anatomy and fiber connectivity in adult Pax6 deficient mice. Cerebr. Cortex a (electronic publication ahead of print, March 27, 2009) doi:10.1093/cercor/bhp057. Brown, R.H., Walters, D.M., Greenberg, R.S., Mitzner, W., 1999. A method of endotracheal intubation and pulmonary functional assessment for repeated studies in mice. J. Appl. Physiol. 87, 2362–2365. Caviness Jr., V.S., 1975. Architectonic map of neocortex of the normal mouse. J. Comp. Neurol. 164, 247–263.
Chahboune, H., Ment, L.R., Stewart, W.B., Ma, X., Rothman, D.L., Hyder, F., 2007. Neurodevelopment of C57B/L6 mouse brain assessed by in vivo diffusion tensor imaging. NMR Biomed. 20, 375–382. Chuang, K.H., Lee, J.H., Silva, A.C., Belluscio, L., Koretsky, A.P., 2009. Manganese enhanced MRI reveals functional circuitry in response to odorant stimuli. NeuroImage 44, 363–372. Clark, V.P., Courchesne, E., Grafe, M., 1992. In vivo myeloarchitectonic analysis of human striate and extrastriate cortex using magnetic resonance imaging. Cereb. Cortex 2, 417–424. Duyn, J.H., van Gelderen, P., Li, T.Q., de Zwart, J.A., Koretsky, A.P., Fukunaga, M., 2007. High-field MRI of brain cortical substructure based on signal phase. Proc. Natl. Acad. Sci. U. S. A. 104, 11796–11801. Eickhoff, S., Walters, N.B., Schleicher, A., Kril, J., Egan, G.F., Zilles, K., Watson, J.D.G., Amunts, K., 2005. High-resolution MRI reflects myeloarchitecture and cytoarchitecture of the human cerebral cortex. Hum. Brain Map. 24, 206–215. Eidt, S., Kendall, E.J., Obenaus, A., 2004. Neuronal and glial cell populations in the piriform cortex distinguished by using an approximation of q-space imaging after status epilepticus. AJNR 25, 1225–1233. Engelhorn, T., Weise, J., Hammen, T., Bluemcke, I., Hufnagel, A., Doerfler, A., 2007. Early diffusion-weighted MRI predicts regional neuronal damage in generalized status epilepticus in rats treated with diazepam. Neurosci. Lett. 417, 275–280. Gadjanski, I., Boretius, S., Williams, S.K., Lingor, P., Knöferle, J., Sättler, M.B., Fairless, R., Hochmeister, S., Sühs, K.W., Michaelis, T., Frahm, J., Storch, M.K., Bähr, M., Diem, R., in press. Role of N-type voltage dependent calcium channels in autoimmune optic neuritis, Ann. Neurol. 66 (Electronic publication ahead of print, March 23, 2009). doi:10.1002/ana.21668. Kovacevic, N., Henderson, J.T., Chan, E., Lifshitz, N., Bishop, J., Evans, A.C., Henkelman, R.M., Chen, X.J., 2005. A three-dimensional MRI atlas of the mouse brain with estimates of the average and variability. Cereb. Cortex 15, 639–645. Lee, J.H., Silva, A.C., Merkle, H., Koretsky, A.P., 2005. Manganese-enhanced magnetic resonance imaging of mouse brain after systemic administration of MnCl2: dosedependent and temporal evolution of T1 contrast. Magn. Reson. Med. 53, 640–648. Ma, Y., Smith, D., Hof, P.R., Foerster, B., Hamilton, S., Blackband, S.J., Yu, M., Benveniste, H., 2008. In vivo 3D digital atlas database of the adult C57BL/6J mouse brain by magnetic resonance microscopy. Front. Neuroanat. 2, 1. MacKenzie-Graham, A., Lee, E.F., Dinov, I.D., Bota, M., Shattuck, D.W., Ruffins, S., Yuan, H., Konstantinidis, F., Pitiot, A., Ding, Y., Hu, G., Jacobs, R.E., Toga, A.W., 2004. A multimodal, multidimensional atlas of the C57BL/6J mouse brain. J. Anat. 204, 93–102. Merkler, D., Boretius, S., Stadelmann, C., Ernsting, T., Michaelis, T., Frahm, J., Brück, W., 2005. Multicontrast MRI of remyelination in the central nervous system. NMR Biomed. 18, 395–403. Michalke, B., Halbach, S., Nischwitz, V., 2007. Speciation and toxicological relevance of manganese in humans. J. Environ. Monit. 9, 650–656. Natt, O., Watanabe, T., Boretius, S., Radulovic, J., Frahm, J., Michaelis, T., 2002. High-resolution 3D MRI of mouse brain reveals small cerebral structures in vivo. J. Neurosci. Methods 120, 203–209. Palay, S.L., Chan-Palay, V., 1974. Cerebellar Cortex: Cytology and Organization. Springer, Berlin. Pautler, R.G., Koretsky, A.P., 2002. Tracing odor-induced activation in the olfactory bulbs of mice using manganese-enhanced magnetic resonance imaging. NeuroImage 16, 441–448. Paxinos, G., 2004. The Rat Nervous System, 3rd ed. Elsevier, Sydney. Peters, A., Jones, E.G., 1984. Cerebral Cortex. Plenum, New York. Polleux, F., Dehay, C., Kennedy, H., 1997. The timetable of laminar neurogenesis contributes to the specification of cortical areas in mouse isocortex. J. Comp. Neurol. 385, 95–116. Sharief, A.A., Johnson, G.A., 2006. Enhanced T2 contrast for MR histology of the mouse brain. Magn. Reson. Med. 56, 717–725. Silva, A.C., Lee, J.H., Wu, C.W., Tucciarone, J., Pelled, G., Aoki, I., Koretsky, A.P., 2008. Detection of cortical laminar architecture using manganese-enhanced MRI. J. Neurosci. Methods 167, 246–257. Takeda, A., Sawashita, J., Okada, S., 1995. Biological half-lives of zinc and manganese in rat brain. Brain Res. 695, 53–58. Wadghiri, Y.Z., Blind, J.A., Duan, X., Moreno, C., Yu, X., Joyner, A.L., Turnbull, D.H., 2004. Manganese-enhanced magnetic resonance imaging (MEMRI) of mouse brain development. NMR Biomed. 17, 613–619. Walters, N.B., Egan, G.F., Kril, J.J., Kean, M., Waley, P., Jenkinson, M., Watson, J.D., 2003. In vivo identification of human cortical areas using high-resolution MRI: an approach to cerebral structure–function correlation. Proc. Natl. Acad. Sci. U. S. A. 100, 2981–2986. Watanabe, T., Natt, O., Boretius, S., Frahm, J., Michaelis, T., 2002. In vivo 3D MRI staining of mouse brain after subcutaneous application of MnCl2. Magn. Reson. Med. 48, 852–859. Yamamori, T., Rockland, K.S., 2006. Neocortical areas, layers, connections, and gene expression. Neurosci. Res. 55, 11–27. Yokel, R.A., Crossgrove, J.S., 2004. Manganese toxicokinetics at the blood-brain barrier. Res. Rep. Health Eff. Inst. 7–58 discussion 59–73. Zhang, K., Sejnowski, T.J., 2000. A universal scaling law between gray matter and white matter of cerebral cortex. Proc. Natl. Acad. Sci. U. S. A. 97, 5621–5626. Zilles, K., Palomero-Gallagher, N., 2001. Cyto-, myelo-, and receptor architectonics of the human parietal cortex. NeuroImage 14, S8–20.