Neuroanatomical phenotypes in the Reeler mouse

www.elsevier.com/locate/ynimg
NeuroImage 34 (2007) 1363–1374
Neuroanatomical phenotypes in the Reeler mouse

Alexandra Badea,a Peter J. Nicholls,b G. Allan Johnson,a,⁎ and William C. Wetselb,c

aCenter for In Vivo Microscopy, Box 3302, Duke University Medical Center, Durham, NC 27710, USAbDepartment of Psychiatry and Behavioral Sciences, Duke University Medical Center, Durham, NC 27710, USAcCell Biology, Mouse Behavioral and Neuroendocrine Analysis Core Facility, Duke University Medical Center, Durham, NC 27710, USA

Received 6 June 2006; revised 13 September 2006; accepted 18 September 2006Available online 20 December 2006

The reeler mouse (Reln) has been proposed as a neurodevelopmentalmodel for certain neurological and psychiatric conditions and has beenstudied by qualitative histochemistry and electron microscopy. Usingmagnetic resonance microscopy (MRM), we have quantitated for thefirst time the neuromorphology of Reln mice at a resolution of21.5 μm. The neuroanatomical phenotypes of heterozygous andhomozygous mutant Reln mice were compared to those of wild type(WT) littermates using morphometry and texture analysis. Thecortical, hippocampal, and cerebellar phenotypes of the heterozygousand homozygous mutant Reln mice were confirmed, and new featureswere revealed. The Relnrl/rl mice possessed a smaller brain, and bothRelnrl/+ and Relnrl/rl mice had increased ventricles compared to WTcontrols. Shape differences were found between WT and Relnrl/rl

brains, specifically in cerebellum, olfactory bulbs, dorsomedial frontaland parietal cortex, certain regions of temporal and occipital lobes, aswell as in the lateral ventricles and ventral hippocampus. Thesefindings suggest that certain brain regions may be more severelyimpacted by the Reln mutation than others. Gadolinium-based activestaining demonstrated that layers of the hippocampus were disorga-nized in Relnrl/rl mice and differences in thickness of these layers wereidentified between WT and Relnrl/rl mice. The intensity distributionscharacteristic to the dorsal, middle, and ventral hippocampus werealtered in the Relnrl/rl, especially in the ventral hippocampus. Thesedifferences were quantified using skewness and modeling the intensitydistributions with a Gaussian mixture. Our results suggest thatstructural features of Relnrl/rl brain most closely phenocopy those ofpatients with Norman–Roberts lissencephaly.© 2006 Elsevier Inc. All rights reserved.

Keywords: Magnetic resonance microscopy; Reelin; Brain; Hippocampus;Ventricles; Cerebellum

⁎ Corresponding author. Fax: +1 919 684 7158.E-mail address: [email protected] (G.A. Johnson).Available online on ScienceDirect (www.sciencedirect.com).

1053-8119/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.neuroimage.2006.09.053

Introduction

Diagnosis of neurological and psychiatric conditions is oftenmade according to phenomenological or behavioral criteria (Amer-ican Psychiatric Association, 2000; Rowland, 2005). Postmortemstudies have identified structural features in the central nervoussystem (CNS) that are characteristic for many neurologicalsyndromes and the approach is somewhat successful for certainpsychiatric conditions, such as schizophrenia (Roberts et al., 2005;Rowland, 2005). Unfortunately, this analysis is applicable only todeceased individuals and it cannot be used in the initial diagnosis.For this reason, investigators are combining clinical assessment withmagnetic resonance imaging (MRI) and/or functional MRI to rendera more concrete diagnosis of the disorder (McCarley, 2002).Although this approach has the potential to advance identificationand improve the accuracy of diagnoses, in most cases it has notrevealed basic mechanisms that underlie the condition so thattherapies can be developed to more efficiently ameliorate theseverity or expression of the syndrome. Despite this limitation, someresearchers are beginning to identify genes and gene products thatmay underlie structural and/or behavioral features that accompanycertain neurological and psychiatric conditions (Katsel et al., 2005).

Alterations in cytoarchitecture of brain may be attributed tosequelae that occur during development and/or maturation of theCNS (Rowland, 2005). Although neurodevelopmental events areoften attributed to epilepsy and deficits in learning andmemory, theymay also underlie autism, schizophrenia, and certain neurologicalconditions that include hydrocephaly (Rowland, 2005). Whileinvestigators are identifying candidate or collections of genes thatmay predispose a given patient to present a certain syndrome, otherscientists are using this information to study pre-existing mutants orto develop genetically modified animals to model neurodevelop-mental morphological features of the condition.

Reln is a mutation that arose spontaneously in mice thatdisplayed ataxia, hypotonia, and fine tremor (Falconer, 1951).Subsequently, reelin was found to be an extracellular matrixglycoprotein that is secreted by certain neurons (D’Arcangelo etal., 1995; Ogawa et al., 1995), such as Cajal–Retzius cells in thecortical marginal zones and granule cells in cerebellum (Tissir andGoffinet, 2003). Reelin appears to act as a stop signal for migrating

mailto:[email protected]

http://dx.doi.org/10.1016/j.neuroimage.2006.09.053

1364 A. Badea et al. / NeuroImage 34 (2007) 1363–1374

neurons where it seems to control the positioning and orientation ofneurons in their respective target layers. Hence, it participates inlaminar development and development of architectonic structureswithin the CNS (Landrieu and Goffinet, 1981; Goffinet et al.,1984; Del Rio et al., 1997; Förster et al., 2006a,b). As a result, Relnmice display abnormalities in neural migration that lead todistortions in the dendritic trees and axonal projections, especiallyin cerebral cortex, hippocampus, and cerebellum. Since theseneurodevelopmental alterations bear similarities to certain aspectsof those in schizophrenia, autism, and lissencephaly, Relnrl/+ andRelnrl/rl mice have been proposed to represent preclinical structuralmodels for these conditions (Fatemi, 2001).

To date, Reln brain has been examined by various neuroanato-mical procedures (Caviness and Sidman, 1973; Mikoshiba et al.,1980). These traditional methods are prone to distortion, can causetissue damage, are typically qualitative, and do not preserve thethree-dimensional (3D) spatial relationships between anatomicalstructures. We have analyzed Reln brain using magnetic resonancemicroscopy (MRM). This procedure obviates many concerns oftraditional methods and the sensitivity of the technique is sufficientto detect changes within 0.03% in mouse brain volume (Cyr et al.,2005). In the present study, we have used MRMwith a resolution ashigh as 21.5 μm to assess morphometric alterations in Reln brain.

Materials and methods

Animal preparation and imaging protocol

The Reln mutation originally arose spontaneously in mice(Falconer, 1951). Male and female adult WT, heterozygous, andhomozygous mutant Reln mice (Jackson Laboratory, Bar Harbor,Maine, USA) were used for this study. The animals wereapproximately 5 months of age at the time of investigation. Micewere housed in groups of 3–5 animals/cage in a temperature- andhumidity-controlled room with a 14:10 light–dark cycle (lights on at0700 h). Food and water were provided ad libitum. All experimentswere conducted in accordance with NIH guidelines for the care anduse of animals and with approved animal protocols from the DukeUniversity Institutional Animal Care and Use Committee.

In preparation for morphology studies, mice were anesthetizedwith 100 mg/kg pentobarbital (i.p.) and transcardially perfusedwith a solution of 10% formalin and 10% ProHance (BraccoDiagnostics, Princeton, NJ) as described by Johnson andcolleagues (2002). Heads were stored overnight in formalin andscanned the next day. The brains were placed in fomblin-filledtubes and were scanned in the skull to avoid distortions of brain ordamage to tissue due to handling.

All imaging was performed on a 9.4-T Oxford vertical boremagnet with a GE EXCITE Console (EPIC 11.0) that has beenspecially adapted for MRM. Images were acquired with aradiofrequency refocused 3D spin warp sequence with TE/TR of5.1/50 ms, a field of view of 22×11×11 mm, and zero filling wasused, resulting in a 1024×512×512 matrix and an isotropicresolution of 21.5 μm. The sequence used non-uniform radial gainand asymmetric data collection resulting in a scan time of 2 h and7 min per brain (Johnson, in preparation).

Manual segmentation

The brains were manually segmented for ventricles, hippo-campus, and cerebellum using Voxstation (MRPath, Durham,

NC). Only one plane of orientation was used for segmenting aspecific structure; however, a 3D viewer was used to assess theaccuracy of segmentation in all three cardinal planes (MacKenzie-Graham et al., 2004). The ventricles were segmented as onesystem comprising the lateral, third, and fourth ventricles insagittal slices. Manual segmentation of the ventricles wasfacilitated by the high contrast that cerebrospinal fluid has inthe MRM images, but it was made difficult, especially in WT andheterozygous mice by their thin structure. In several places, anarbitrary cut was required to separate the ventricles from clefts inthe brain filled with fluid.

The hippocampus was segmented in coronal slices startinganteriorly at the level of the hippocampal commissure using themouse brain atlas of Paxinos and Franklin (2001) as a guide. Theposterior limit was established with the help of the sagittal andhorizontal planes, at the level of the ventral nucleus of the laterallemniscus. In general, the in-plane borders were established usinghigh contrasting white matter tracts like the corpus callosum(external capsule) or alveus for the lateral side, the brachium of thesuperior colliculus, optic tract, or cerebral peduncles on the medialside. The ventricles provided high contrast for determining thelateral borders of the hippocampus. Aside from hippocampus, thecerebellum was segmented in two mid-sagittal slices in all brains toshow the dramatic change in size of the Relnrl/rl brain compared tothat in the WT and Relnrl/+ mice.

Layer thickness measurements

Cortical layer measurements were performed in coronal slicesat approximately −2.1 mm from bregma and in the region of theprimary somatosensory cortex (Hof et al., 2000). Additionally,hippocampal length, from its upper medial to the most distal lateralaspect, and the thickness of hippocampal layers were measured inthe same slice.

Image preprocessing

Brain volumes were obtained after stripping the skull using analgorithm based on mathematical morphology and connectedcomponent analysis similar to the one described by Badea et al.(2003). To increase the time efficiency of the procedure, the imagedata were down-sampled by a factor of two along each dimensionat this step. The algorithm started with smoothing the data, andapplying a number of erosions. A seed was selected automaticallyin the center of the volume, and region growing was started fromthat seed. Pixels with values between fixed thresholds andconnected to the initial region were added to the brain mask. Atthe end, dilation was applied for the same number of times as weapplied erosion. Empirically, we established that the optimalnumber of erosions/dilations to be used was 6 for our data, whichcome as a 512×256×256 matrix.

Image intensity normalization was performed in the followingway. The lowest 5% and the highest 2% of the intensity valueswere discarded and assigned as the minimum (IMin) and maximum(IMax) image values. The original values (I) were mapped to I’ afterscaling to the same global range of image values (IGlobalMax andIGlobalMin):

I V¼ I � IMinð Þ* IGlobalMax � IGlobalMin

IMax � IMinþ IMin:

1 Supplementary material and figures can be found at Duke Center for InVivo Microscopy website; http://www.civm.duhs.duke.edu/neuro/mouse.html.

1365A. Badea et al. / NeuroImage 34 (2007) 1363–1374

The segmented structures were registered with an affine transformusing Image Registration Toolkit (Rueckert et al., 2003) to removeglobal sources of variability, which may arise from changes in theposition of a structure within the brain, its size, or its orientation. Afterthresholding and smoothing the mask volumes, BRAINSUITE(Shattuck and Leahy, 2002) and MATLAB (The MathWorks, Natick,MA) were used to generate surface meshes from the iso-surfaces thatseparate the binary segmentations from the background and to reducetheir dimension to 3000 faces per structure. The meshes were thenanalyzed using software written in MATLAB.

Morphometry

The total volume of whole brain, ventricles, hippocampus, and inparticular the dentate gyrus, as well as the mid-sagittal area of thecerebellum, weremeasured. Brain size was not normalized since thisparameter was dramatically affected by cerebellar atrophy in thehomozygous mutant mice. Absolute volumetric measurements weremade for segmented structures using voxel counting andmultiplyingthe number of voxels by the voxel volume (9.94 μl). Areas werecalculated by computing the sum of areas for elementary patches thatmake up the surfacemeshes of the segmented structures. Themesheswere used for shape analysis in the following way. Correspondencebetween meshes was established based on the minimum distancebetween vertices of two meshes. The mean shapes for each of thestructures were generated at each node of the mesh using the linearaverage of the closest vertices for each of the brains. Distance mapswere calculated to give estimates of local shape changes. The rootmean square distance between each of the original meshes and theirmeanwas used to pseudocolor themeanmesh and estimate the shapevariability within groups. Global parameters like Hausdorff distance(HD), the mean distance between shapes, and the fractal dimension(FD) were calculated for each of the three genotypes of mice. Thesame morphometric parameters were used for the between groupanalysis.

The HD measures the degree of mismatch between two shapes,represented by the vertices of the triangulated meshes. The directHD (h) identifies the point from the first shape (S1) that is farthestaway from any point in the second shape (S2) and measures thedistance between the point in S1 and its closest neighbor in S2. HDwas calculated as the maximum of minimum distances (ri to rj)between one shape and another, computed forward and backward.

HDðS1;S2Þ ¼ maxðhðS1;S2Þ; hðS1;S2ÞÞ; where hðS1;S2Þ¼ max

rjas1minrjas2

tri� rjt:

This distance metric is, however, sensitive to “outliers.” Hence,the mean distance between shapes was computed as well.

The FD is a measure of the shape complexity of an object. Itreflects the rate at which the area increases as the scale ofmeasurement is reduced. The FD was calculated using the boxcounting method (Barabási and Stanley, 1995). In this method theobject is covered with boxes of various sizes. The FD was estimatedfrom the slope of the log–log plot that relates the number of boxesnecessary to cover the volume of interest and their size as follows:

FD ¼ �limIbY0

log10NbðIbÞlog10Ib

;

where Nb(lb) was the number of boxes with size lb needed tocompletely cover the structure.

Grayscale analysis

A grayscale analysis was performed for the hippocampus. Theimage intensity of each hippocampus was normalized to the sameglobal range of values, according to the gray level distribution of theindividual head scans. The minimum and maximum intensities wereobtained after discarding the lowest 5% and the highest 2% values ofthe distribution characteristic of the head scans. The mean, standarddeviation, skewness, and kurtosis for the hippocampus were mea-sured. The intensity distribution for the hippocampus was modeledwith a mixture of two Gaussian distributions using expectationmaximization. In this model, we assumed that one of the dis-tributions would model the intensities characteristic of the cell layersof the hippocampus which appear dark in the images, whereas thesecond distribution would model the remainder of the hippocampalbody. The distance between the means of the two Gaussian distri-butions was measured. This analysis was performed on whole hippo-campus, and separately for the dorsal,middle, and ventral hippocampus.

Statistics

The data are presented as means and standard deviations withsome analyses by MATLAB. Most of the data were analyzed usingSPSS (SPSS Inc., Chicago, IL) where univariate ANOVA andrepeated measures ANOVA were used; the latter was used toanalyze differences in skewness of the intensity distributions fordifferent regions of the hippocampus. Post hoc tests wereperformed by Bonferroni corrected pair-wise comparisons. Ap<0.05 was considered significant.

Results

As a number of morphological features have been identified to beabnormal in neurological and psychiatric patients (Lawrie andAbukmeil, 1998; Rowland, 2005), MRM was used to study thebrains of Relnmice. A qualitative examination of the images revealedthat the appearance of WT and Relnrl/+ brains differed dramaticallyfrom those ofRelnrl/rlmutants (Fig. 1, see also Supplementary Figs. 1–3).1 Gross inspection of the images showed the ventricles to beenlarged, the cerebellum atrophied, and the layered organization of thecortex and hippocampus to be greatly disturbed in brains of thehomozygous mutants. By comparison, these structures appeared to besimilar between the WT and Relnrl/+ brains.

To perform a quantitative morphometric analysis, brains weresemi-automatically segmented and extraneous tissue was removed.The ventricles, hippocampus, and mid-sagittal slices through thecerebellum were manually segmented for all three genotypes. Slicesin the three cardinal planes through representative brains are shown(Fig. 1; see also Supplementary Figs. 1–3). The results ofsegmentations were used to reconstruct the 3D structures (Fig. 2)and measure the volumes, areas, and shape parameters of the brainsas detailed below.

Volume measurements

The absolute volumes of the manually segmented structureswere measured for brain, as well as for ventricles and hippocampus

Fig. 1. Mouse brain structures visible by MRM with resolution to 21 μm in Reln mice. Horizontal (Left), coronal (Middle), and sagittal (Right) sections throughthe brains of a WT control (Top), Relnrl/+ (Middle), and Relnrl/rl mouse (Bottom). The stain emphasizes cellular layers of the hippocampus where the pyramidal(Py) and granule cells (GrDG) can be visualized. These cell layers have high contrast and a characteristic shape in WT and Relnrl/+ mice but appear disorganizedin Relnrl/rl animals. The ventricles appear enlarged in the Relnrl/+ and Relnrl/rl mice; the cerebellum is atrophied in the homozygous mutant animals.


(Table 1). As expected, the variability of volumes for the total brainand ventricles increased from WT to Relnrl/+ to Relnrl/rl mice.Compared to WT animals, the total volume of brain andhippocampus in the Relnrl/rl animals was reduced by approximately19% and 17%, respectively; whereas ventricular volume wasincreased by 205% (Table 1). Alterations were also observed incomparisons between WT and Relnrl/+ mice where total brain,ventricular, and hippocampal volumes in the latter group wereenhanced by approximately 6%, 82%, and 7%, respectively.

A univariate ANOVA for brain volumes revealed significantgenotype effects [F(2,9)=22.58, p=0.0001] and Bonferroni testsdemonstrated that volumes of the WT and Relnrl/+ brains weresimilar but greater than those from Relnrl/rl mice (p<0.002).Similarly, univariate ANOVA detected a significant main effect ofgenotype for ventricular volumes [F(2,9)=5.38, p<0.029] and

Fig. 2. Segmentation of the ventricles and hippocampus in Relnmice. The lateral ve(Middle), and Relnrl/rl (Right) animals in a background of the brain. Enlarged ventrsmaller for the homozygous mutants. The olfactory bulbs appear disorganized in thecontrols. Layers of the cerebral cortex can be visualized in the WT animal but are

Bonferroni comparisons found that ventricular volumes for WTand Relnrl/+ mice were similar, with the WT ventricular volumesignificantly smaller than that from the homozygous mutants(p<0.030). It should be noted, however, ventricular volumes in theRelnrl/+ animals were highly variable and were approximately 1.8-fold larger than those in WT littermates. ANOVA for hippocampusrevealed statistically significant differences in the volumes for thethree groups [F(2,9)=18.12, p<0.001]. A post hoc analysisshowed that the hippocampal volumes were not different betweenWT and the Relnrl/+ mice but were greater than those for Relnrl/rl

animals (ps<0.007). Together, these findings indicate that volumesof total brain, ventricles, and hippocampus from the homozygousmutants were notably different from those of WT mice, whereasonly total brain and hippocampal volumes in the Relnrl/rl animalswere significantly changed from those in heterozygous animals.

ntricles (cyan) and hippocampus (yellow) are depicted in WT (Left), Relnrl/+

icles can be seen for the Relnrl/+ and Relnrl/rl mice; the hippocampus is alsoRelnrl/rl animals and they seem deformed in Relnrl/+ mice relative to the WTnot as easily seen in Relnrl/rl mice.

Table 2Fractal dimension (FD) for the hippocampus, ventricles, and total brain

Structure a WT Relnrl/+ Relnrl/rl

Brain 2.52±0.01 2.52±0.01 2.49±0.01 b, c

Ventricle 2.08±0.07 1.93±0.05 b 1.83±0.05 b, c

Hippocampus 2.34±0.04 2.48±0.24 2.35±0.01Dorsal hippocampus 2.44±0.01 2.56±0.20 2.47±0.01Ventral hippocampus 2.38±0.02 2.49±0.23 2.37±0.03

a The data are presented as means and standard deviations.b p<0.05 from the WT control.c p<0.05 from the heterozygous mice.

Table 1Volume and area measurements for the hippocampus, ventricles, and total brain

Structure a Volume (mm3) Area (mm2) Area/volume (mm−1)

WT Relnrl/+ Relnrl/rl WT Relnrl/+ Relnrl/rl WT Relnrl/+ Relnrl/rl

Brain 552.3±14.2 582.8±33.5 448.5 b, c±36.2 407.88±21.16 443.45±38.06 397.07±9.54 0.74±0.05 0.77±0.09 0.89±0.08Ventricle 3.8±0.7 6.9±2.2 11.6 b±5.4 40.63±3.57 62.41 b±8.49 77.53 b±12.48 10.96±2.05 9.50±1.91 6.29 b±1.42Hippocampus 29.4±2.8 31.5±0.8 24.3 b, c±0.9 86.56±1.09 87.89±2.41 94.28 b, c±4.23 2.97±0.31 2.80±0.11 3.88 b, c±0.17a The data are presented as means and standard deviations.b p<0.05 from the WT control.c p<0.05 from the heterozygous mice.


Since there did not seem to be a significant difference betweenWT and Relnrl/+ volume measurements for hippocampus, wecompared volume measurements for the dentate gyrus. The volumefor the dentate gyrus for the Relnrl/rl mice (5.71±1.16 mm3)represented about 76% of the WT volume (7.47±0.35 mm3); thesedata were significant [t(1,6)=2.908, p<0.027].

Area measurements

Large variability in areas was noticeable for brain, hippocam-pus, and ventricular regions, especially for the homozygous mutant(Table 1). For instance, the standard deviation for ventricular areawas ∼12 mm2 for Relnrl/rl animals, compared to ∼4 mm2 for theWT controls. Despite variability in brain, ventricular, andhippocampal areas in the Relnrl/rl mice, Levin's tests forheterogeneity of variance were not significant in any of theseincidences. ANOVA for brain areas was not significant. AnANOVA for ventricular areas found a main effect of genotype[F(2,9)=10.23, p<0.005]. Bonferroni tests showed ventricularareas from both the Relnrl/+ and Relnrl/rl mice were significantlydifferent from those of the WT littermates (ps<0.02). ANOVA forhippocampal areas found a main effect of genotype [F(2,9)=8.22, p<0.0093]. Bonferroni corrections indicated that hippocam-pal areas for WT and Relnrl/+ mice were smaller than for Relnrl/rl

animals (ps<0.036) but were not different from each other.Collectively, the analyses of areas revealed high variability withinRelnrl/rl mice, where ventricular and hippocampal areas wereincreased relative to the other genotypes.

Since the cerebellum was distorted in shape in the Relnrl/rl mice,mid-sagittal slices from this brain region were compared as areas.The mean areas were 10.3±0.62 mm2 for WT, 10.7±0.52 mm2 forRelnrl/+, and 2.6±0.19 mm2 for Relnrl/rl animals. Hence, cerebellarareas were similar between WT and heterozygous mice, whereasareas were reduced by ∼76% in the homozygous mutants. Aunivariate ANOVA found a main effect of genotype [F(2.9)=358.56, p<0.001] and Bonferroni corrections revealed thecerebellar area was significantly reduced in Relnrl/rl mice comparedto that in the WT and Relnrl/+ animals (ps<0.001). Hence, the mid-sagittal cerebellar area was greatly reduced in the homozygousmutants.

Global shape characteristics

Apart from examining differences in volume and areas, ratios ofareas to volumes are characteristics of shapes of structures (Table 1).These ratios were found to differ for ventricles for Relnrl/rl incontrast to WT and Relnrl/+ animals, and for hippocampus for WTand Relnrl/+ compared to Relnrl/rl mice. For brain, the differences inthe area/volume ratio approached significance [F(2,9)=4.07,

p<0.055] and genotype differences could not be discerned withBonferroni corrections. In contrast, ratios were significant for theventricles [F(2,9)=6.89, p<0.015] and Bonferroni tests showedthat the area/volume ratios for ventricles for Relnrl/rl animals weresmaller than for WT mice (p<0.017), whereas the ratios for theRelnrl/+ and Relnrl/rl animals approached significance (p<0.1).

Aside from the ventricles, significant differences were alsodetected for the hippocampal area to volume ratio [F(2,9)=29.55,p<0.001]; post hoc analyses revealed that the ratio wassignificantly smaller for the homozygous mutants compared tothat of the other two genotypes (ps<0.001), which did not differfrom each other. These ratio results show that the shapes of theventricles and hippocampus are clearly different between WT andRelnrl/rl mice.

Besides examining the area to volume ratio, the fractaldimension (FD) can also be used to quantitatively characterizethe global shape and its complexity for segmented structures inbrain (Table 2). Analyses of FD for the total brain and ventriclesshowed these indices to be significantly different [F(2,9)=13.23,p<0.002 for brain; F(2.9)=21.39, p<0.001 for ventricles]. Posthoc analyses indicated differences between WT and Relnrl/rl andbetween Relnrll+ and Relnrl/rl mice (ps<0.005 for brain; ps<0.02for ventricles). Parenthetically, differences in FD for ventricles forWT and Relnrl/+ mice were also significant (p<0.050). Genotypedifferences in FD were not detected in hippocampus. Thus, forbrain and ventricles the FDs were larger for WT than homozygousmutants, and the ventricular FDs were greater for the WT than theheterozygous animals.

Another way to examine shape characteristics is to analyzeboundaries of the segmented structures. The surfaces extractedfrom the segmentations were used to generate mean meshes foreach of the three genotypes (Fig. 3). The HD between the meanmeshes for the WT, Relnrl/+, and Relnrl/rl mice, together with themean distances are summarized in Table 3. The results indicatedlarger HD distances between WT and the Relnrl/rl mice than

Fig. 3. Three-dimensional rendering of the whole brain, ventricles, and hippocampus of Reln mice. The mean surfaces are pseudo-colored according to the rootmean square of the distance within the WT group (Left). The mean surfaces for the Relnrl/+ (Middle) and Relnrl/rl mice (Right) are pseudo-colored according tothe distances between the respective mutant group of mice and the WT animals. Histograms of the distance distributions are depicted for whole brain (Top),ventricles (Middle), and hippocampus (Bottom). Most of the values for whole brain in WT and Relnrl/rl mice are concentrated in the lower range of distances,whereas those of the Relnrl/rl animals are more evenly distributed across the whole range of values. With regards to the ventricles, distance values for the WTcontrols are restricted to 0.01–0.8 mm, while those for Relnrl/+ and Relnrl/rl mice span a much larger range (0.01–2.65 and 0.01–2.77 mm, respectively). Asimilar pattern is seen for hippocampus for the three genotypes.


between the WT and Relnrl/+ animals for brain (3.17 mm versus2.10 mm), ventricles (2.71 mm versus 2.58 mm), and hippocampus(0.75 mm versus 0.36 mm). These same relationships can be seenin comparisons of mean distances, which are less sensitive to noise.To better understand these differences, we evaluated the HD withingroups and compared that to the values between groups. The HDbetween the WT and Relnrl/rl mice was greater than the distanceswithin groups for whole brain, ventricles, and hippocampus (Table3). In summary, when the HD was used as a landmark featurefor each genotype, distances were greater between the WT and

Table 3Global distance measures between the segmented shapes

Structure Between groups Within groups

WT vs. Relnrl/+ WT vs. Relnrl/rl WT

Distance a MDb HD MD HD MD HBrain 0.24 2.10 0.34 3.17 0.23±0.05 2Ventricles 0.11 2.58 0.29 2.71 0.15±0.06 2Hippocampus 0.12 0.36 0.16 0.75 0.10±0.03 0a Distances are given in millimeters.b Abbreviations: MD=mean distance, HD=Hausdorff distance.

homozygous mutants than between the heterozygous animals andeither of the other two genotypes.

Local shape changes

The analysis of global changes suggested possible local shapechanges within total brain, ventricles, and, perhaps, hippocampusfor Reln mice. Local shape changes were observed after thresh-olding the distance maps used to pseudocolor the shapes of thesegmented structures (Fig. 3). When examining the signed distance

Relnrl/+ Relnrl/rl

D MD HD MD HD.71±0.82 0.27±0.16 1.95±0.63 0.22±0.07 1.66±0.80.13±0.94 0.10±0.03 0.87±0.43 0.16±0.07 1.23±0.80.45±0.19 0.11±0.04 0.57±0.31 0.13±0.03 0.68±0.10


maps of total brain, inward deformations of the cerebellum andolfactory bulbs were clearly evident in the Relnrl/rl mice comparedto the WT controls (Fig. 4). Shape differences were found betweenWT and Relnrl/rl animals in medial frontal and parietal cortex, andtemporal and occipital lobes. Parenthetically, the brain stem areasof WT and Relnrl/rl mice appear to be highly variable. These latterresults may be due both to removal of the brain from the skull forscanning, as well as true variability in brain structures.

After analyzing total brain and selected regions, we evaluatedlocal shape changes in the ventricles and hippocampus. Theventricles were found to vary mostly in the region of the thirdventricle, which was very thin, and along the lateral ventricles—especially on the anterior side and midway along the dorsal toventral axis at the curvature of the ventricles (Fig. 3). Besides theventricles, the dorsal and ventral aspects of the hippocampusappeared changed (Figs. 3 and 5). The signed distance map showeda contraction of the Relnrl/rl shape along the dorsolateral aspect ofhippocampus (Fig. 5). It should be emphasized that brain,ventricle, and hippocampal distances of the Relnrl/rl animalsshowed much greater variability and dispersion relative to those forthe WT littermate controls.

Grayscale analysis

It is evident that the organization of the laminated structures inthe Relnrl/rl mouse brain is greatly disturbed (Caviness and Sidman,1973). For example, the cell layers of the cerebral cortex andhippocampus appear dramatically different in the homozygousmutant relative to those in the WT animals (Fig. 1; see alsoSupplementary Figs. 1 and 3). The layers appear disorganized, thecontrast between them is reduced, and the cortex and hippocampusappear more homogenous than in the WT controls. In order toquantify differences in texture between genotypes, we performedgrayscale analysis and restricted the investigation to the hippo-campus because it was the most easy to quantify. Binary masks ofhippocampi were intersected with the grayscale images and therange of intensities was normalized to the same global values forall brains included in the analysis. Despite normalization and theseprecautions, the shapes of the histograms were different. Althoughmean values and kurtosis of the intensity distributions did notclearly distinguish among the genotypes, analysis of skewness ofthe distributions was sufficient in this regard. A slice-by-sliceanalysis along the rostrocaudal axis was performed to examinedifferences in skewness—which reflected the presence of darklystained gray matter layers in the hippocampus. A repeatedmeasures ANOVA showed that the skewness of the histogram

Fig. 4. Shape analysis for the surface of the whole brain for Relnrl/rl mouse. Signed dinward deformations in the regions of the olfactory bulbs and cerebellum. Tempo“warm” color scale, whereas the negative distances are depicted in the “cold” coloupon Hotteling tests, which show that shapes are significantly reduced in the Relnr

lobes, cerebellum, and brainstem.

curves was statistically different for the most dorsal, the middle,and ventral regions of the hippocampus (Fig. 6; see alsoSupplementary Fig. 4). For rostral hippocampus, the repeatedmeasures ANOVA for multiple levels of grayscale was significant[F(39,351)=2.514, p<0.001], as was the levels by genotypeinteraction [F(78,351)=1.674, p<0.001]. Bonferroni correctedpair-wise comparisons of the interaction revealed the WT controlshad higher levels of skewness than the Relnrl/+ mice (ps<0.052)and it approached significance for the Relnrl/rl animals (ps<0.10).With regards to middle hippocampus, grayscale levels approachedsignificance [F(69,621)=1.286, p<0.067], whereas the levels bygenotype interaction was significant [F(138,621) = 2.048,p<0.001]. Bonferroni tests of the interaction showed that theWT and Relnrl/rl mice were different (ps<0.045). Analyses ofventral hippocampus demonstrated levels [F(69,621)=6.093,p<0.001] and the levels by genotype interaction to be significant[F(138,621)=2.653, p<0.001]. Post hoc comparisons of theinteraction showed skewness to be significantly higher in theWT than Relnrl/rl mice (ps<0.042), with no differences discernedbetween WT and Relnrl/+ animals.

Aside from slice-by-slice analyses, the hippocampus was dividedinto the dorsal, middle, and ventral parts and analyzed by modelingthe intensity distributions with two Gaussian distributions, whereone Gaussian distribution (Gauss1) represented the stained celllayers and the other distribution (Gauss2) was the unstained areas(cells and fibers). Since the WT and Relnrl/+ mice were notstatistically different on the slice-by-slice analyses, only WT andRelnrl/rl animals were compared. Gauss1 was significant in thedorsal [t(1,6)=2.777, p<0.032], middle [t(1,6)=2.499, p<0.047],and ventral hippocampus [t(1,6)=3.003, p<0.024]. Similarly,Gauss2 was also significant in these same areas of the hippocampus[dorsal: t(1,6)=2.650, p<0.038; middle: t(1,6)=2.738, p<0.034;ventral: t(1,6)=2.861, p<0.029]. Collectively, these results indicatethat the organization of the cell layers in the dorsal, middle, andventral hippocampus are abnormal in the homozygous mutant Relnmouse.

A quantitative analysis of the thickness of layered structures ofthe hippocampus was performed in a coronal plane approximately−2.10 mm from bregma (Supplementary Fig. 5; see atlas of Hof etal., 2000, as guide). These linear measurements are summarized inTable 4. Overall cortical thickness was diminished [t(1,6)=6.66,p<0.002] and hippocampal length was reduced [t(1,6)=3.89,p<0.008] in Relnrl/rl mice compared to the WT controls. Layers ofthe cortex in the homozygous mutants were diffused and could notbe clearly distinguished as layers I to VI in WT animals. In CA1hippocampus, the pyramidal cell layer was bifurcated and it

istance map between the mean meshes for the WTand Relnrl/rl mice indicateral cortex (blue) is also deformed. The positive distances are shown in ther scale. To the right of each brain are the statistical maps of distances basedl/rl olfactory bulbs, medial frontal and parietal cortex, temporal and occipital

Fig. 5. Caudal view of whole hippocampus of WTand Relnrl/rlmice. (A) The mean shapes for the WT (green) and Relnrl/rl (red) hippocampus. Differences can bediscerned between genotypes in the dorsal, middle, and ventral hippocampus where this brain area is more compressed in the mutant mouse. (B) Unsigneddistance maps quantify differences in shape between WT and Relnrl/rl hippocampus. The greatest distance between groups is present in the ventral or tail regionof the hippocampus. (C) Signed distance map shows a compression in shape of dorsolateral hippocampus in Relnrl/rl mice.


extended over a larger portion of this region in Relnrl/rl than in WTmice [t(1,6)=12.6, p<0.001]. By comparison, the thickness of thestratum radiatum [t(1,6)=4.35, p<0.005] was decreased in Relnrl/rl

hippocampus. Similar to the CA1 pyramidal cell layer, thethickness of this layer in CA3 hippocampus was also greatercompared to WT littermates [t(1,6)=3.73, p<0.010]. In themutants, the dentate gyrus appeared diffuse and the granular andpolymorphic layers were difficult to discriminate. Nevertheless, theupper blade of the molecular cell layer was decreased [t(1,6)=4.38,p<0.005], while the upper blade of the granule cell layer wasthicker in the homozygous mutants than that in the WT controls[t(1,6)=6.34, p<0.001]. Collectively, these findings show thatmany features of the cerebral cortex and hippocampus are alteredin the Relnrl/rl brain.

Table 4Cortical and hippocampal thickness measurements

Structure a WT Relnrl/rl p

Somatosensory cortex 1055.70±29.70 b 835.92±58.95 0.002

Discussion

The Reln mouse was first described as having an abnormalneuroanatomical phenotype over 50 years ago (Falconer, 1951).Since that time, abnormalities have been observed in the cerebralcortex, hippocampus, and cerebellum (Caviness, 1976; Stanfield

Fig. 6. Grayscale characterization of hippocampus in Reln mice. A slice-by-slice analysis of skewness values is used to evaluate genotype differencesalong the rostral to caudal axis of the hippocampus. Inspection of thehistograms reveals that the skewness values for hippocampus are similarfor WT and Relnrl/+ mice and these values are different from those of theRelnrl/rl animals. Although differences are observed in the dorsal, middle,and ventral hippocampus, the most robust changes occur in the dorsal (inset,right) and ventral (inset, left) hippocampus.

and Cowan, 1979; Mikoshiba et al., 1980). In cortex andhippocampus, lamination of cells is perturbed (Ogawa et al.,1995; Gebhardt et al., 2002) and the cerebellum often lacksfoliation and appears atrophied due to reductions in Purkinje andgranule cells (Mikoshiba et al., 1983; Heckroth et al., 1989;Soriano et al., 1997). Aside from these regions, neurons in theolfactory bulb, cochlear nuclei, inferior olive, and variousmesencephalic nuclei are aberrant (Wyss et al., 1980; Martin,1981; Blatt and Eisenman, 1985; Yip et al., 2000). Since reelinappears to serve as a stop signal for the migration of cells from thefloorplate to the cortical plate (Landrieu and Goffinet, 1981;Goffinet et al., 1984; Del Rio et al., 1997; Hack et al., 2002;Förster et al., 2006a,b), the Reln mutant mice have served as auseful neurodevelopmental model to study laminar developmentand development of architectonic structures within the mammalian

Layer 1 of cortex/polymorphic layer c

106.69±15.50 102.61±7.99 0.795

Hippocampus (length) 3061.06±28.75 2558.50±256.20 0.008Stratum oriens (CA1) 146.42±27.98 136.79±24.97 0.626Stratum pyramidale (CA1) 69.50±6.90 167.27±13.90 0.000Stratum radiatum (CA1) 214.52±8.16 117.01±44.14 0.005Stratum lacunosum

moleculare (CA1)61.22±7.23 68.16±6.99 0.106

Pyramidal cell layer (CA3) 84.61±5.12 132.49±25.19 0.010Molecular cell layer-upper

blade (DGd)243.76±14.06 156.04±37.62 0.005

Granule cell layer-upperblade (DG)

72.46±5.71 101.53±7.19 0.001

Polymorphic cell layer e

(DG)190.81±13.41 185.98±109.96 0.933

Granule cell layer—lowerblade (DG)

61.22±7.23 68.16±6.99 0.217

Molecular cell layer (DG)(lower blade)

130.61±23.47 99.12±33.40 0.174

a The structures were named according to the nomenclature of Hof et al.(2000).b The thicknesses are given in μm as means and standard errors of the

mean.c Layer 1 of WT mouse and polymorphic layer of cerebral cortex of

Relnrl/rl mouse (Caviness, 1976).d DG=dentate gyrus.e Corresponds to hilus area.


CNS. In our own studies, we observed changes within many brainregions and concentrated our experiments on whole brain,ventricles, and hippocampus to provide the most robust quantita-tion of differences among the Reln mice. A unique aspect of ourstudies was to use high-resolution (21.5 μm) MRM to analyzebrain structures in Reln mice and to provide quantitative analysesin non-distorted brain. Briefly, we compared total brain, ventricles,and hippocampus according to volume, area, and global and localshapes. Grayscale analyses were also applied to hippocampus andthickness measurements were applied to cerebral cortex andhippocampus. Due to dramatic deformation of the cerebellum inRelnrl/rl mice, area measurements were applied to mid-sagittalsections in these mutants. Although the cortical, hippocampal, andcerebellar phenotypes of the Relnrl/+ and Relnrl/rl mice wereconfirmed, some novel features were observed. Brains from bothmutants had increased ventricular sizes compared to that of WTcontrols. Relnrl/rl mice also possessed a smaller brain. Shapedifferences were found between WT and Relnrl/rl brains,specifically in cerebellum, olfactory bulbs, dorsomedial frontaland parietal cortex, certain regions of temporal and occipital lobes,as well as in the lateral ventricles and ventral hippocampus.Additionally, the thicknesses of the cerebral cortex and hippo-campus were also altered. These findings suggest the Relnmutation exerts heterogeneous effects on brain – perhaps throughinteractions with other signal transduction systems (Beffert et al.,2004) – where some locations are more affected than others.

Various investigators have emphasized qualitative differencesin hippocampus, cerebral cortex, and olfactory bulbs of the Relnrl/rl

mice (Caviness, 1976; Stanfield and Cowan, 1979; Wyss et al.,1980). Although we saw similar changes, dramatic differences involumes of the cerebellum and ventricles were observed. Ourmorphometric analysis also showed Relnrl/rl whole brain, hippo-campus, and its dentate gyrus to be smaller in volume than that inWT controls and Relnrl/+ animals. Global shape analysis indicatedsignificant differences between WT and Relnrl/rl mice, such that theventricles were enlarged, but hippocampus was reduced in themutants. Local shape analysis demonstrated significant atrophy ofthe olfactory bulbs and cerebellum in Relnrl/rl animals. In addition,regions of the dorsolateral cortex appeared to be significantlychanged. When shape analysis was limited to hippocampus,alterations were localized to its ventral aspect. Although the thirdand fourth ventricles appeared abnormal, the shape of the lateralventricles was clearly aberrant in Relnrl/rl mice.

Besides regional changes, there is considerable literaturesupporting disorganization of the laminar structures in the cerebralcortex, cerebellum, and hippocampus of Relnrl/rl mice (Mikoshibaet al., 1983; Ogawa et al., 1995; Gebhardt et al., 2002). Althoughwe observed alterations in the cortex and cerebellum, changeswithin hippocampus were most amenable to quantification. Usingactive staining, disorganization of cellular layers of the dentategyrus and hippocampus proper were easy to visualize. By usinggrayscale analysis, we were able to discern differences betweenWT and Relnrl/rl animals in the dorsal, middle, and ventralhippocampus. The former and latter regions were particularlydeformed. In dorsal hippocampus, the pyramidal cell layer in CA1and CA3 hippocampus were thicker in Relnrl/rl than in WT mice. Incontrast, in CA1 the thickness of the stratum radiatum was thinner.Although in mutants the dentate gyrus was diffuse and the granularand polymorphic layers were smaller and difficult to discriminate,the upper blade of the molecular cell was thinner, while the upperblade of the granule cell layer was thicker in the Relnrl/rl mice than

that in the WT controls. In summary, morphometric, grayscale, andthickness analyses demonstrated that the hippocampus of thehomozygous mutants was very different from the WT mice;many of these same indices were indistinguishable for WT andheterozygous animals.

Structural aberrations in Reln mouse brain have beencompared to abnormalities in brains of patients with schizo-phrenia (Ballmaier et al., 2002). These changes typically includeincreased ventricular volume, especially in the lateral ventricles(Lawrie and Abukmeil, 1998); a smaller cerebellar vermis(Okugawa et al., 2003); and reduced volumes of hippocampusand amygdala (Vita et al., 2006). Changes in shape in thehippocampal tail (Styner et al., 2004) and increased volumes ofbasal ganglia have been reported (Hokama et al., 1995). At amore detailed level, schizophrenic brains show significantcellular disorganization compared to normal controls (Conradet al., 1991). These changes are observed in the hippocampus inrelation to abnormal pyramidal cell density and orientation(Conrad et al., 1991); smaller neuron size in the subiculum, CA1hippocampus, and layer II of entorhinal cortex (Arnold et al.,1995); reduced dendritic spine density in CA1 hippocampus andcerebral cortex; and decreased Purkinje cells in cerebellar vermis(Tran et al., 1998). Aside from these changes, analysis of theRELN gene in postmortem brain reveals its promoter to behypermethylated in some schizophrenic patients (Grayson et al.,2005). This result may underlie the reduced expression of RELNin layer I neurons in schizophrenic cerebral cortex compared tothat of healthy controls (Eastwood and Harrison, 2006).Expression in deeper layers of cortex is unchanged (Eastwoodand Harrison, 2003). Due to reduction, but not complete absenceof reelin protein, Relnrl/+ mice have been proposed to representa structural model of schizophrenia (Costa et al., 2002). In thisregard, structural changes have been reported for this mutant incerebral cortex, hippocampus, and mesolimbic pathways (Liu etal., 2001; Ballmaier et al., 2002). We have observed increasedventricular volume and area in this mouse. Although behavioralstudies show prepulse inhibition to be abnormal in Relnrl/+ mice(Costa et al., 2002), other investigators have failed to replicatethis result and find no evidence of a schizophrenia-likephenotype using a large battery of behavioral tests (Podhornaand Didriksen, 2004). Thus, if the Reln mouse is considered astructural model of schizophrenia, it must be used with cautionand with knowledge of the many limitations of the model.

Besides schizophrenia, the Reln mouse has also been proposedas a model for autism (Fatemi, 2001). Structural imaging studiesshow the ventricles to be enlarged and the hippocampus smaller inautistic patients than in normal controls (Brambilla et al., 2003).Some investigators have also reported the overall brain is largerand the vermis of cerebellum is smaller in autistic subjects;however, these findings are controversial (Palmen and vanEngeland, 2004). Hence, relationships of the Reln mouse to autismare highly speculative. With regard to genetics, there is littleevidence that autistic patients harbor a mutation in RELN (Bonoraet al., 2003). However, many individuals with Norman–Robertslissencephaly have mutations at two different locations in RELNand some of these patients may present with autistic-spectrumdisorders (Hong et al., 2000). Interestingly, the Relnrl/rl mousephenocopies many of the structural attributes of lissencephalybecause these patients have enlarged ventricles, small hippocam-pus, hypoplastic cerebellum, and abnormal cerebral and hippo-campal architecture (Kato and Dobyns, 2003). Taken together,


these data suggest that the Relnrl/rl mouse most closely represents astructural animal model for lissencephaly.

It should be emphasized that the neuroanatomical and behavioralphenotype of the Reln mouse is shared by several other mutants.Reelin protein is known to bind the ApoE2 and VLDL receptors andmice that have disruptions in either gene do not display the samephenotype as Reln mutants. However, mutants lacking both genespossess a phenotype that is virtually identical to that of Relnrl/rl mice(Trommsdorff et al., 1999). Another mutant that possesses similarstructural deformations is the disabled 1 mutant (Howell et al.,1997). This gene encodes an adapter protein for tyrosine kinases thatappear in the reelin signaling pathway (Rice et al., 1998). Inaddition, fyn and src also seem to participate in this same pathwaybecause disruption of Fyn produces an intermediate phenotype(Yuasa et al., 2004), whereas mice lacking both the Fyn and Srcgenes phenocopy the Relnmutation (Kuo et al., 2005). Furthermore,compound mutant analysis has identified a distal member of thereelin signaling pathway. Lis1 appears to be important formicrotubule stabilization and neuronal migration (Sasaki et al.,2000; Brandon et al., 2004). When heterozygous Lis1 mice arecrossed with animals that lack other genes in the reelin pathway, theReln phenotype is recapitulated (Assadi et al., 2003). Moreover,LIS1 is a well-known causative gene for human lissencephaly, acondition which may predispose the patient to autistic-spectrumdisorders (Kato and Dobyns, 2003). Interestingly, Lis1 binds to thedisrupted in schizophrenia 1 protein and it has been implicated inschizophrenia (Lipska et al., 2006; Tabares-Seisdedos et al., 2006).Hence, interactions of genes involved in reelin signaling mayprovide insights into etiologies not only for lissencephaly, but alsofor various neurodevelopmental disorders that include certainaspects of autism and schizophrenia.

Acknowledgments

We wish to thank Ms. Jiechun Zhou for breeding andmaintaining the Reln mice, Ms. Boma Fubara and Mr. JamesPollaro for help with the animal preparation and imaging, and Dr.Martin Styner for helpful discussions. This work was madepossible through collaborations with the Mouse BioinformaticsResearch Network (MBIRN), through NIH/NCRR grant U24RR021760. All experiments were conducted at the Duke Center forIn Vivo Microscopy, an NCRR/NCI National Biomedical Technol-ogy Resource Center, and were partially supported by unrestrictedfunds to W.C.W. and by grants P41-RR005959 and R24-CA092656 to G.A.J.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.neuroimage.2006.09.053.

References

American Psychiatric Association, 2000. Diagnostic and Statistical Manualof Mental Disorders, 4th ed. American Psychiatric Association,Arlington, VA. Text Revision (DSM-IV-TR).

Arnold, S.E., Franz, B.R., Gur, R.C., Gur, R.E., Shapiro, R.M., Moberg, P.J.,Trojanowski, J.Q., 1995. Smaller neuron size in schizophrenia inhippocampal subfields that mediate cortical–hippocampal interactions.Am. J. Psychiatry 152, 738–748.

Assadi, A.H., Zhang, G., Beffert, U., McNeil, R.S., Renfro, A.L., Niu, S.,Quattrocchi, C.C., Antalffy, B.A., Sheldon, M., Armstrong, D.D.,Wynshaw-Boris, A., Herz, J., D'Arcangelo, G., Clark, G.D., 2003.Interaction of reelin signaling and Lis1 in brain development. Nat.Genet. 35, 270–276.

Badea, A., Kostopoulos, G.K., Ioannides, A.A., 2003. Surface visualizationof electromagnetic brain activity. J. Neurosci. Methods 127, 137–147.

Ballmaier, M., Zoli, M., Leo, G., Agnati, L.F., Spano, P., 2002. Preferentialalterations in the mesolimbic dopamine pathway of heterozygous reelermice: an emerging animal-based model of schizophrenia. Eur. J.Neurosci. 15, 1197–1205.

Barabási, A.-L., Stanley, H.E., 1995. Fractal Concepts in Surface Growth.Cambridge Univ. Press, New York.

Beffert, U., Weeber, E.J., Morfini, G., Ko, J., Brady, S.T., Tsai, L.H., Sweatt,J.D., Herz, J., 2004. Reelin and cyclin-dependent kinase 5-dependentsignals cooperate in regulating neuronal migration and synaptictransmission. J. Neurosci. 24, 1897–1906.

Blatt, G.J., Eisenman, L.M., 1985. A qualitative and quantitative lightmicroscopic study of the inferior olivary complex of normal, reeler, andweaver mutant mice. J. Comp. Neurol. 232, 117–128.

Bonora, E., Beyer, K.S., Lamb, J.A., Parr, J.R., Klauck, S.M., Benner, A.,Paolucci, M., Abbott, A., Ragoussis, I., Poustka, A., Bailey, A.J.,Monaco, A.P., 2003. International molecular genetic study of autism(IMGSAC). Analysis of reelin as a candidate gene for autism. Mol.Psychiatry 8, 885–892.

Brambilla, P., Hardan, A., di Nemi, S.U., Perez, J., Soares, J.C., Barale, F.,2003. Brain anatomy and development in autism: review of structuralMRI studies. Brain Res. Bull. 61, 557–569.

Brandon, N.J., Handford, E.J., Schurov, I., Rain, J.C., Pelling, M., Duran-Jimeniz, B., Camargo, L.M., Oliver, K.R., Beher, D., Shearman, M.S.,Whiting, P.J., 2004. Disrupted in Schizophrenia 1 and Nudel form aneurodevelopmentally regulated protein complex: implications forschizophrenia and other major neurological disorders. Mol. Cell.Neurosci. 25, 42–55.

Caviness Jr., V.S., 1976. Patterns of cell and fiber distribution in theneocortex of the reeler mutant mouse. J. Comp. Neurol. 170, 435–447.

Caviness Jr., V.S., Sidman, R.L., 1973. Retrohippocampal, hippocampal andrelated structures of the forebrain in the reeler mutant mouse. J. Comp.Neurol. 147, 235–254.

Conrad, A.J., Abebe, T., Austin, R., Forsythe, S., Scheibel, A.B., 1991.Hippocampal pyramidal cell disarray in schizophrenia as a bilateralphenomenon. Arch. Gen. Psychiatry 48, 413–417.

Costa, E., Davis, J., Pesold, C., Tueting, P., Guidotti, A., 2002. Theheterozygote reeler mouse as a model for the development of a newgeneration of antipsychotics. Curr. Opin. Pharmacol. 2, 56–62.

Cyr, M., Caron, M.G., Johnson, G.A., Laakso, A., 2005. Magneticresonance imaging at microscopic resolution reveals subtle morpholo-gical changes in a mouse model of dopaminergic hyperfunction.NeuroImage 26, 83–90.

D'Arcangelo, G., Miao, G.G., Chen, S.C., Soares, H.D., Morgan, J.I.,Curran, T., 1995. A protein related to extracellular matrix proteinsdeleted in the mouse mutant reeler. Nature 374, 719–723.

Del Rio, J.A., Heimrich, B., Borrell, V., Forster, E., Drakew, A., Alcantara,S., Nakajima, K., Miyata, T., Ogawa, M., Mikoshiba, K., Derer, P.,Frotscher, M., Soriano, E., 1997. A role for Cajal–Retzius cells andreelin in the development of hippocampal connections. Nature 385,70–74.

Eastwood, S.L., Harrison, P.J., 2003. Interstitial white matter neuronsexpress less reelin and are abnormally distributed in schizophrenia:towards an integration of molecular and morphologic aspects of theneurodevelopmental hypothesis. Mol. Psychiatry 8, 821–831.

Eastwood, S.L., Harrison, P.J., 2006. Cellular basis of reduced corticalreelin expression in schizophrenia. Am. J. Psychiatry 163,540–542.

Falconer, D.S., 1951. Two new mutants, Trembler and Reeler, withneurological actions in the house mouse. J. Genet. 50, 192–201.

Fatemi, S.H., 2001. Reelin mutations in mouse and man: from reeler mouse

http://dx.doi.org/doi:10.1016/j.neuroimage.2006.09.053


to schizophrenia, mood disorders, autism and lissencephaly. Mol.Psychiatry 6, 129–133.

Förster, E., Jossen, Y., Zhao, S., Chai, X., Frotscher, M., Goffinet, A.M.,2006a. Recent progress in understanding the role of Reelin in radialneuronal migration, with special emphasis on the dentate gyrus. Eur. J.Neurosci. 23, 901–909.

Förster, E., Zhao, S., Frotscher, M., 2006b. Laminating the hippocampus.Nat. Rev., Neurosci. 7, 259–268.

Gebhardt, C., Del Turco, D., Drakew, A., Tielsch, A., Herz, J., Frotscher, M.,Deller, T., 2002. Abnormal positioning of granule cells alters afferentfiber distribution in the mouse fascia dentata: morphologic evidencefrom reeler, apolipoprotein E receptor 2-, and very low densitylipoprotein receptor knockout mice. J. Comp. Neurol. 445, 278–292.

Goffinet, A.M., So, K.F., Yamamoto, M., Edwards, M., Caviness Jr., V.S.,1984. Architectonic and hodological organization of the cerebellum inreeler mutant mice. Brain Res. 318, 263–276.

Grayson, D.R., Jia, X., Chen, Y., Sharma, R.P., Mitchell, C.P., Guidotti, A.,Costa, E., 2005. Reelin promoter hypermethylation in schizophrenia.Proc. Natl. Acad. Sci. U. S. A. 102, 9341–9346.

Hack, I., Bancila, M., Loulier, K., Carroll, P., Cremer, H., 2002. Reelin is adetachment signal in tangential chain-migration during postnatalneurogenesis. Nat. Neurosci. 5, 939–945.

Heckroth, J.A., Goldowitz, D., Eisenman, L.M., 1989. Purkinje cellreduction in the reeler mutant mouse: a quantitative immunohistochem-ical study. J. Comp. Neurol. 279, 546–555.

Hof, P., Young, W.G., Bloom, F.E., Belichenko, P.V., Celio, Z.M.R., 2000.Comparative Cytoarchitectonic Atlas of the C57BL/6 and 129/SvMouseBrains. Elsevier, Amsterdam, Netherlands.

Hokama, H., Shenton, M.E., Nestor, P.G., Kikinis, R., Levitt, J.J., Metcalf,D., Wible, C.G., O'Donnell, B.F., Jolesz, F.A., McCarley, R.W., 1995.Caudate, putamen, and globus pallidus volume in schizophrenia: aquantitative MRI study. Psychiatry Res. 61, 209–229.

Hong, S.E., Shugart, Y.Y., Huang, D.T., Shahwan, S.A., Grant, P.E.,Hourihane, J.O., Martin, N.D., Walsh, C.A., 2000. Autosomal recessivelissencephaly with cerebellar hypoplasia is associated with humanRELN mutations. Nat. Genet. 26, 93–96.

Howell, B.W., Hawkes, R., Soriano, P., Cooper, J.A., 1997. Neuronalposition in the developing brain is regulated by mouse disabled-1.Nature 389, 733–737.

Johnson, G.A., Cofer, G.P., Gewalt, S.L., Hedlund, L.W., 2002. Morpho-logic phenotyping with MR microscopy: the visible mouse. Radiology222, 789–793.

Kato, M., Dobyns, W.B., 2003. Lissencephaly and the molecular basis ofneuronal migration. Hum. Mol. Genet. 12, R89–R96.

Katsel, P.L., Davis, K.L., Haroutunian, V., 2005. Large-scale microarraystudies of gene expression in multiple regions of the brain inschizophrenia and Alzheimer's disease. Int. Rev. Neurobiol. 63,41–82.

Kuo, G., Arnaud, L., Kronstad-O'Brien, P., Cooper, J.A., 2005. Absence ofFyn and Src causes a reeler-like phenotype. J. Neurosci. 25, 8578–8586.

Landrieu, P., Goffinet, A., 1981. Inverted pyramidal neurons and their axonsin the neocortex of reeler mutant mice. Cell Tissue Res. 281, 291–301.

Lawrie, S.M., Abukmeil, S.S., 1998. Brain abnormality in schizophrenia. Asystematic and quantitative review of volumetric magnetic resonanceimaging studies. Br. J. Psychiatry 172, 110–120.

Lipska, B.K., Peters, T., Hyde, T.M., Halim, N., Horowitz, C., Mitkus, S.,Weickert, C.S., Matsumoto, M., Sawa, A., Straub, R.E., Vakkalanka, R.,Herman, M.M., Weinberger, D.R., Kleinman, J.E., 2006. Expression ofDISC1 binding partners is reduced in schizophrenia and associated withDISC1 SNPs. Hum. Mol. Genet. 15, 1245–1258.

Liu, W.S., Pesold, C., Rodriguez, M.A., Carboni, G., Auta, J., Lacor, P.,Larson, J., Condie, B.G., Guidotti, A., Costa, E., 2001. Down-regulationof dendritic spine and glutamic acid decarboxylase 67 expressions in thereelin haploinsufficient heterozygous reeler mouse. Proc. Natl. Acad.Sci. U. S. A. 98, 3477–3482.

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 atlasof the C57BL/6J mouse brain. J. Anat. 204, 93–102.

Martin, M.R., 1981. Morphology of the cochlear nucleus of the normal andreeler mutant mouse. J. Comp. Neurol. 197, 141–152.

McCarley, R.W., 2002. Structural magnetic resonance imaging studies inschizophrenia. In: Davis, K.L., Charney, D., Coyle, J.T., Nemeroff, C.(Eds.), Neuropsychopharmacology: The Fifth Generation of Progress.Lippincott Williams and Wilkins, Philadelphia, pp. 757–773.

Mikoshiba, K., Nagaike, K., Kohsaka, S., Takamatsu, K., Aoki, E., Tsukada,Y., 1980. Developmental studies on the cerebellum from reeler mutantmouse in vivo and in vitro. Dev. Biol. 79, 64–80.

Mikoshiba, K., Terada, S., Takamatsu, K., Shimai, K., Tsukada, Y., 1983.Histochemical and immunohistochemical studies of the cerebellum fromthe reeler mutant mouse. Dev. Neurosci. 6, 101–110.

Ogawa, M., Miyata, T., Nakajima, K., Yagyu, K., Seike, M., Ikenaka, K.,Yamamoto, H., Mikoshiba, K., 1995. The reeler gene-associated antigenon Cajal–Retzius neurons is a crucial molecule for laminar organizationof cortical neurons. Neuron 14, 899–912.

Okugawa, G., Sedvall, G.C., Agartz, I., 2003. Smaller cerebellar vermis butnot hemisphere volumes in patients with chronic schizophrenia. Am. J.Psychiatry 160, 1614–1617.

Palmen, S.J., van Engeland, H., 2004. Review on structural neuroimagingfindings in autism. J. Neural Transm. 111, 903–929.

Paxinos, G., Franklin, K.B.J., 2001. The Mouse Brain in StereotaxicCoordinates, 2nd ed. Academic Press, New York.

Podhorna, J., Didriksen, M., 2004. The heterozygous reeler mouse:behavioural phenotype. Behav. Brain Res. 153, 43–54.

Rice, D.S., Sheldon, M., D'Arcangelo, G., Nakajima, K., Goldowitz, D.,Curran, T., 1998. Disabled-1 acts downstream of Reelin in a signalingpathway that controls laminar organization in the mammalian brain.Development 125, 3719–3729.

Roberts, R.C., Roche, J.K., Conley, R.R., 2005. Synaptic differences in thepostmortem striatum of subjects with schizophrenia: a stereologicalultrastructural analysis. Synapse 56, 185–197.

Rowland, L.P., 2005. Merritt's Neurology, 11th ed. Lippincott Williams andWilkins, Philadelphia.

Rueckert, D., Frangi, A.F., Schnabel, J.A., 2003. Automatic construction of3-D statistical deformation models of the brain using nonrigidregistration. IEEE Trans. Med. Imaging 22, 1014–1025.

Sasaki, S., Shionoya, A., Ishida, M., Gambello, M.J., Yingling, J.,Wynshaw-Boris, A., Hirotsune, S., 2000. A LIS1/NUDEL/cytoplasmicdynein heavy chain complex in the developing and adult nervoussystem. Neuron 28, 681–696.

Shattuck, D.W., Leahy, R.M., 2002. BrainSuite: an automated corticalsurface identification tool. Med. Image Anal. 6, 129–142.

Soriano, E., Alvarado-Mallart, R.M., Dumesnil, N., Del Rio, J.A., Sotelo,C., 1997. Cajal–Retzius cells regulate the radial glia phenotype in theadult and developing cerebellum and alter granule cell migration.Neuron 18, 563–577.

Stanfield, B.B., Cowan, W.M., 1979. The morphology of the hippocampusand dentate gyrus in normal and reeler mice. J. Comp. Neurol. 185,393–422.

Styner, M., Lieberman, J.A., Pantazis, D., Gerig, G., 2004. Boundary andmedial shape analysis of the hippocampus in schizophrenia. Med. ImageAnal. 8, 197–203.

Tabares-Seisdedos, R., Escamez, T., Martinez-Gimenez, J.A., Balanza, V.,Salazar, J., Selva, G., Rubio, C., Vieta, E., Geijo-Barrientos, E.,Martinez-Aran, A., Reiner, O., Martinez, S., 2006. Variations in genesregulating neuronal migration predict reduced prefrontal cognition inschizophrenia and bipolar subjects from Mediterranean Spain: apreliminary study. Neuroscience 139, 1289–1300.

Tissir, F., Goffinet, A.M., 2003. Reelin and brain development. Nat. Rev.,Neurosci. 4, 496–505.

Tran, K.D., Smutzer, G.S., Doty, R.L., Arnold, S.E., 1998. Reduced Purkinjecell size in the cerebellar vermis of elderly patients with schizophrenia.Am. J. Psychiatry 155, 1288–1290.

Trommsdorff, M., Gotthardt, M., Hiesberger, T., Shelton, J., Stockinger,


W., Nimpf, J., Hammer, R.E., Richardson, J.A., Herz, J., 1999.Reeler/Disabled-like disruption of neuronal migration in knockoutmice lacking the VLDL receptor and ApoE receptor 2. Cell 97,689–701.

Vita, A., De Peri, L., Silenzi, C., Dieci, M., 2006. Brain morphology in first-episode schizophrenia: a meta-analysis of quantitative magneticresonance imaging studies. Schizophr. Res. 82, 75–88.

Wyss, J.M., Stanfield, B.B., Cowan, W.M., 1980. Structural abnormalities inthe olfactory bulb of the Reeler mouse. Brain Res. 188, 566–571.

Yip, J.W., Yip, Y.P., Nakajima, K., Capriotti, C., 2000. Reelin controlsposition of autonomic neurons in the spinal cord. Proc. Natl. Acad. Sci.U. S. A. 97, 8612–8616.

Yuasa, S., Hattori, K., Yagi, T., 2004. Defective neocortical development inFyn-tyrosine-kinase-deficient mice. NeuroReport 15, 819–822.

Documents

Neuroanatomical phenotypes in the Reeler mouse