Phenotypic integration of neurocranium and brain

Abstract

Evolutionary history of Mammalia provides strong evidence that the morphology of skull and brain change jointly in evolution. Formation and development of brain and skull co-occur and are dependent upon a series of morphogenetic and patterning processes driven by genes and their regulatory programs. Our current concept of skull and brain as separate tissues results in distinct analyses of these tissues by most researchers. In this study, we use 3D computed tomography and magnetic resonance images of pediatric individuals diagnosed with premature closure of cranial sutures (craniosynostosis) to investigate phenotypic relationships between the brain and skull. It has been demonstrated previously that the skull and brain acquire characteristic dysmorphologies in isolated craniosynostosis, but relatively little is known of the developmental interactions that produce these anomalies. Our comparative analysis of phenotypic integration of brain and skull in premature closure of the sagittal and the right coronal sutures demonstrates that brain and skull are strongly integrated and that the significant differences in patterns of association do not occur local to the prematurely closed suture. We posit that the current focus on the suture as the basis for this condition may identify a proximate, but not the ultimate cause for these conditions. Given that premature suture closure reduces the number of cranial bones, and that a persistent loss of skull bones is demonstrated over the approximately 150 million years of synapsid evolution, craniosynostosis may serve as an informative model for evolution of the mammalian skull.

Fig. 1

3D reconstruction of CT images of the craniofacial skeleton and MR images of the central nervous system of a 21-week-old child with right unilateral synostosis of the coronal suture (RUCS). Features consistent with the diagnosis of RUCS include a flattened frontal bone on the side of the fused suture, and a “twisting” of the facial skeleton and cranial base (not shown). The four panels include: (a) anterior view of 3D reconstruction of skull; (b) anterior view of 3D reconstruction of brain; (c,d) anterior, lateral views of 3D reconstruction of skull superimposed (and ghosted for transparency) over the 3D brain reconstruction to show anatomical relationships of brain and skull. CT and MR images were acquired separately. Consequently, the superimposition used in this figure is based on anatomical knowledge rather than any superimposition or registration algorithm.

Fig. 2

Superior view of 3D reconstructions of brain (left) and skull (right) of a morphologically normal child (a), a child with SS (b), and a child with RUCS (c). In all views, the anterior aspect of the head is at the top while the posterior aspect is below. The brain and skull of the SS individual (b) is elongated along the anteroposterior axis and reduced medio-laterally. The sagittal suture is obliterated. The brain of the RUCS individual (c) is obviously deformed on the right anterior aspect and appears somewhat wider than normal. The RUCS skull shows dysmorphology of the right frontal bone and distortion of the sagittal and lambdoid suture patterns. The right coronal suture is closed while the left remains patent. Images are not to scale.

Fig. 3

Brain and skull landmarks used in this analysis labeled with abbreviations described in Table 3. The top row represents 3D CT reconstructions of the skull of a child diagnosed with RUCS. Views from left to right are: lateral view of the skull, superior view of the endocranial base, and inferior view of the ectocranial base. Starting at left, the bottom row shows 3D MRI reconstructions of: right lateral surface of the brain (posterior at left, anterior to right), and superior surface of the cerebrum (anterior at top, posterior at bottom). The figure at right represents a lateral view of landmarks located on a model of subcortical structures within the 3D MR images (posterior at left, anterior to right). The approximate location of landmarks AC, PC and 4VP are shown on this view though their true anatomical location lies on the sagittal plane.

Fig. 4

A plot of the elements of the correlation matrices for our two samples. The magnitudes of the raw correlation coefficients are plotted on the Y-axis, while the LDPs are on the X-axis. Each open square represents the correlation coefficient for a pair of linear distances measured on the RUCS sample, one measured on the brain and another measured on the skull. Correlation coefficients for pairs of linear distances measured on SS are designated by closed triangles. The measurement pairs are sorted from left to right in terms of the magnitude of the differences in correlation between the samples, thus the correlations for the two synostosis groups are plotted in the same order. An interactive version of this figure is available on our laboratory website (http://getahead.psu.edu) which allows the user to point to any given symbol and the value for any specific LDP is provided.

Fig. 5

Differences in raw correlation values and associated confidence interval limits for comparison of correlations between measures of skull and brain in RUCS and SS. Correlation difference values are shown as filled diamonds with the associated 95% confidence interval limits (CI upper and CI lower) estimated by bootstrapping shown as open squares. Correlation differences are sorted from minimum (indicating the correlation value for RUCS is greater than the associated value in the SS sample) to maximum. The black horizontal line intersecting the Y-axis at 0 represents the null hypothesis of no difference between RUCS and SS correlations. The shaded gray square includes those linear distances in which the correlation coefficients calculated for RUCS are shown to be significantly different from SS by confidence interval. An interactive version of this figure is available on our laboratory website (http://getahead.psu.edu) which allows the user to point to any given symbol and the value for any specific LDP is provided.

Fig. 6

Correlation coefficients for the 99 LDPs that show significantly stronger correlation in RUCS as compared to SS by confidence interval. The correlation differences shown in the shaded gray square on the far left of Fig. 5 were computed from the correlation coefficients shown on this graph. The osseous linear distance of the skull-brain LDPs are labeled along the X-axis. An interactive version of this figure is available on our laboratory website (http://getahead.psu.edu), which allows the user to point to any given symbol and the value for any specific LDP is provided.

Fig. 7

CT reconstruction of the skull (right half of skull and mandible shown) to show distances on the skull (in red) and the brain (in blue and green) that are associated differently in RUCS and SS. Orientation of the views are given with reference to the orientation of the face. When the nasal bones point directly at the reader, the skull is at 0°. As the skull rotates to its right, we provide views (from top to bottom in the panel) at approximately 30° (top), 90°, 120°, and 170°. This figure depicts those measures of the brain (in blue) that are significantly more strongly associated in RUCS with two measures on the cranial base shown in red: VSJ-to-RFOV and RACP-to-RFOV. Linear distances among neural landmarks shown in green are more strongly associated with the distance between cranial base landmarks RACP-to-RFOV in RUCS as compared to SS. Refer to text for further discussion.

Fig. 8

CT reconstruction of the skull (right half of skull and mandible shown) to show distances on the skull (in red) and the brain (in blue) that are associated differently in RUCS and SS. Orientations of the views are the same as given in Figure 7. This figure depicts those measures of the brain that are significantly more strongly associated with two measures on the cranial base shown in red: RAST-to-BAS and RAST-to-OPI. Linear distances among neural landmarks shown in blue are more strongly associated with both skull measures in RUCS. See text for further discussion.

Fig. 9

CT reconstruction of the skull (right half of skull and mandible shown) to show distances on the skull (in red) and the brain (in blue) that are associated differently in RUCS and SS. Orientations of the views are the same as those in Figure 7. This figure depicts those measures of the brain that are significantly more strongly associated in RUCS with a single measure of the cranial base shown in red: RFOV-to-BAS.

Fig. 10

CT reconstruction of the skull (right half of skull and mandible shown) to show distances on the skull (in red) and the brain (in blue and green) that are associated differently in RUCS and SS. Orientations of the views are the same as those in Figure 7. This figure depicts those measures of the brain that are significantly more strongly associated with two measures of the anterior cranial vault shown in red: RPTNP-to-BRG and RFZJ-to-BRG. Linear distances among neural landmarks shown in blue are more strongly associated with both skull measures in RUCS. The single measure of the brain shown in green (RFSS-to-ROP) is more strongly associated with RFZJ-to-BRG. Refer to text for further discussion.

Fig. 11

Variances of linear distance data (ln) for brain, skull, and brain skull combined (total) for the RUCS and SS samples. Variance calculations follow VanValen (2005).