Elsevier

NeuroImage

Volume 142, 15 November 2016, Pages 656-662
NeuroImage

Variability of brain anatomy for three common mouse strains

https://doi.org/10.1016/j.neuroimage.2016.03.069Get rights and content

Highlights

  • Outbred mice (CD-1) are in general no more variable than inbred mice (C57, 129).

  • PCA finds distinct anatomically-specific modes of shape variability.

  • Three of the most important modes are related to strain differences.

  • PCA-based dimensionality reduction can capture strain differences using only 2 variables.

Abstract

The way in which brain structures express different morphologies is not fully understood. Here we investigate variability in brain anatomy using ex vivo MRI of three common laboratory mouse strains: in two inbred strains (C57BL/6 and 129S6) and one outbred strain (CD-1). We use Generalised Procrustes Analysis (GPA) to estimate modes of anatomical variability. We find three distinct bilateral modes of anatomical surface variability associated with the motor cortex, the anterior somatosensory, the retrosplenial and the entorhinal cortex. The modes of variability that are associated with the motor cortex and anterior somatosensory cortex are predominantly due to genetic, i.e. strain differences. Next, we specifically test if a particular strain is more variable. We find that only the mode associated with motor cortex size has a slightly larger variance in the outbred CD-1 mice compared to the two inbred strains. This suggests that the hypothesis that outbred strains are more variable in general is not true for brain anatomy and the use of outbred CD-1 mice does probably not come at the price of increased variability. Further, we show that the first two principal components distinguish between the three strains with 91% accuracy. This indicates that neuroanatomical strain differences are captured by considerably fewer dimensions than necessary for atlas-based or voxel-wise testing. Statistical comparisons based on shape models could thus be a powerful complement to traditional atlas and voxel-based methods at detecting gene-related brain differences in mice. Finally, we find that the principal components of individual brain structures are correlated, suggesting a tightly coupled network of interdependent developmental trajectories. These results raise the question to what degree neuroanatomical variability is directly genetically determined or the result of experience and epigenetic mechanisms.

Introduction

Unlike outbred strains, inbred strains are strains with little or no genetic variation between individuals due to inbreeding over multiple generations. Inbred strains are popular in research because they exist in many genetically modified versions that address specific cellular and molecular processes. Neuroanatomical studies have shown that within-strain variance is smaller than between strain variance in inbred strains. More specifically, MRI-base segmentation of the brain of eleven inbred mouse strains showed that volume variation between strains was significantly larger than within strains for 23 out of 33 regions (Badea et al., 2009). This suggests that different strains form relatively well-separated tight clusters of anatomical variation.

Inbred strains have been described as more stable, uniform, and genetically better defined than outbred strains. It could thus be argued that the use of genetically identical mice should lead to more consistent results, thus requiring fewer individuals to demonstrate the significance of a specific effect However, in the literature there is still no clear consensus whether outbred strains are indeed more variable. Some studies have argued that inbred strains are more variable or just as variable as outbred strains in terms of body size and physiological characteristics (Biggers, J.D., et al., 1958, McLaren, A. and Michie, D., 1956). Some studies suggest that outbred strains are more variable (Jay, G.E., 1955, Stehelin, D., et al., 1976). The argument for larger variance in outbred strains usually involves the hypothesis that phenotypic variance is composed of genetic and environmental components (Chia et al., 2005). Following this logic, inbred strains should vary less due to the absence of genetic variation. Due to their genetic heterogeneity, however, outbred strains do allow one to draw conclusions that are less likely to reflect strain-specific idiosyncrasies (Miller et al., 1999). Therefore, outbred strains might constitute a better model of humans and their natural genetic variability. In addition, outbred mice are more efficient breeders and have large litter sizes (Aldinger et al., 2009). In light of these advantages of outbred mice, it is curious that so few studies have specifically investigated whether they do indeed come at the price of increased phenotypic variability (Vaickus et al., 2010). Prior studies suggest that certain anatomical features, such as mandible shape, are more variable (Stehelin et al., 1976). For neuroscientific studies it is of particular interest whether strain differences in anatomical variability are related to variability in brain function and behaviour.

Chen et al. (2006) report a measure of morphological variability averaged across the whole brain using ex-vivo MRI of two inbred (CD-1, 129S1) and one outbred strain (C57BL/6). Thus it is not possible to attribute increased variance to specific brain regions. Badea et al. (2009) report MRI-based volumes of individual anatomical regions, but only within inbred strains. Quantitative trait locus (QTL) mapping studies probe the link between phenotype and genotypic data (Kearsey, 1998). Sizes of the neocortex (Gaglani et al., 2009) and the barrel cortex (Jan et al., 2008) measured on histological sections have been found to be associated with QTLs on specific chromosomes across large numbers of inbred mouse strains. These studies suggest that particular genes might determine the size and shape of specific brain regions (Hager et al., 2012). This hypothesis raises the question whether the increased genetic heterogeneity of outbred mice is associated with increased variance in the size and shape of neuroanatomy.

Here we use ex vivo magnetic resonance imaging (MRI) to test specifically for variability in intact brain anatomy using two inbred strains (C57BL/6, 129S6) and one outbred strain (CD-1). Mouse models in conjunction with advanced imaging methods have proven to be a valuable tool to investigate how genetic and environmental factors affect brain structure (Lerch et al., 2011a). Anatomical MR imaging followed by automatic registration-based volume estimation is sensitive enough to even detect the subtle morphological changes associated with skill training (Lerch, J.P., et al., 2011b, Scholz, J., et al., 2014) and environmental enrichment (Scholz et al., 2015). However, the effect sizes are small and close to minimal detection thresholds. Future studies could thus benefit from a better characterization of the variability associated with a specific genetic background. A procedural question is how to best characterize this variability. To answer these questions we demonstrate that neuroanatomical shape variability can be quantified using surface shape models in conjunction with principal component analysis (PCA). Specifically, we test whether variability is localized to functionally relevant anatomical regions and whether this variability is strain-dependent. Second, we test whether variability is increased in the outbred strain. Third, we test whether anatomical variability is a distinguishing feature between the three strains.

Briefly, we base our estimates of anatomical variability on principal components derived from applying PCA to surface shape models of the whole brain and specific anatomical structures, including the hippocampus and striatum. We use the Generalised Procrustes Analysis (GPA) framework which brings the surface point clouds into alignment and optionally adjusts for differences in scale. Variants of PCA have been successfully used to characterize phenotypic variation in the mouse cranium (Hallgrímsson, B., et al., 2009, Hallgrímsson, B., et al., 2007) and morphological asymmetry in the primate and human brain (Gómez-Robles et al., 2013). Here we show that GPA can reveal meaningful components of neuroanatomical variation in the mouse brain, some of which are associated with strain differences.

Section snippets

Mice

Three mouse strains were used: 129S6/SvEvTac/J (N = 25), C57BL6J/Tcp (N = 25), and Crl:CD1(ICR)/Tcp (N = 27). Strains are referred to as 129S6, C57BL/6, and CD-1 for brevity throughout the manuscript. All mice were male and 8 (± 0.17) weeks old. Brains were perfusion-fixed and subsequently immersion fixed for 24 h (Cahill et al., 2012).

MR image acquisition

Brains were scanned within their skulls to minimize perturbations (Badea et al., 2007). A custom-built 16-coil solenoid array was used to image 16 samples concurrently (

Whole brain surface

The GPA of the whole brain surface resulted in a number of PCs of quickly decreasing importance (N = 77, Fig. 1). While the first PC explains about 64% of total variance, the following PCs explain only 13, 4, 4, and 2%. Seven components are sufficient to explain 90% of total brain surface shape variance.

We used Levene's test on the first seven PC scores to test for differences in variance between strains. The strains differed only in the variance associated with motor cortex (M1) size, i.e. PC2

Discussion

Here we present the first systematic evaluation of neuroanatomical shape variability across three common laboratory mouse strains including both inbred and outbred strains. Among the components that explained 90% of total variance we found at least three that seemed to have some degree of anatomical or even functional relevance as suggested by their spatial specificity and symmetry. Further, we were able to quantify inter-strain variance as well as strain-independent variance, with the former

Limitations

Our aim was the investigation of the anatomical variability across several common laboratory mouse strains. We chose PCA-based shape analysis as a method that is well established in related anatomical studies of faces (Claes et al., 2014) and skeletal features (Hallgrímsson, B., et al., 2007, Hallgrímsson, B., et al., 2009, Cates, J., et al., 2008). We could have varied our methods in several ways to conduct a comparable analysis. First, PCA can be applied to voxel-wise data as well to reduce

Conclusions

In summary, using two inbred strains and one outbred strain we were able to quantify strain-dependent neuroanatomical shape variance. Our analysis suggests that outbred strains might not come at the price of larger variability. Neuroanatomical studies could thus benefit from their increased disease resistance, high fecundity, and low neonatal mortality without sacrificing sensitivity and power of the analysis. Future studies will have to investigate how the variability of molecular, cellular

References (47)

  • A. Sotiras et al.

    Finding imaging patterns of structural covariance via non-negative matrix factorization

    NeuroImage

    (2015)
  • J.F.P. Ullmann et al.

    A segmentation protocol and MRI atlas of the C57BL/6J mouse neocortex

    NeuroImage

    (2013)
  • K.a. Aldinger et al.

    Genetic variation and population substructure in outbred CD-1 mice: implications for genome-wide association studies

    PLoS One

    (2009)
  • J.D. Biggers et al.

    Variance control in the animal house

    Nature

    (1958)
  • J. Cates et al.

    A shape analysis framework for small animal phenotyping with application to mice with a targeted disruption of HOXD11

  • R. Chia et al.

    The origins and uses of mouse outbred stocks

    Nat. Genet.

    (2005)
  • S. Cui et al.

    Genetic variation within and between strains of outbred Swiss mice

    Lab. Anim.

    (1993)
  • P. Claes et al.

    Modeling 3D facial shape from DNA

    PLoS Genet.

    (2014)
  • I.L. Dryden

    Shapes Package

    (2015)
  • I.L. Dryden et al.

    Statistical Shape Analysis

    (1998)
  • M.F.W. Festing

    Inbred strains should replace outbred stocks in toxicology, safety testing, and drug development

    Toxicol. Pathol.

    (2010)
  • B.L. Finlay et al.

    Linked regularities in the development and evolution of mammalian brains

    Science (New York, N.Y.)

    (1995)
  • S.M. Gaglani et al.

    The genetic control of neocortex volume and covariation with neocortical gene expression in mice

    BMC Neurosci.

    (2009)
  • Cited by (0)

    View full text