Figure 1.
Cerebellar Regions Investigated.
This flatmap diagram displays the four cerebellar regions of interest for our stereological assay. Hemispheric lobules IV–VI (in yellow) are bordered by the preculminate and superior posterior fissures, lateral to the fourth ventricle. Crus I (in blue) is the region bordered by the superior posterior and horizontal fissures. Crus II (in green) is bordered by the horizontal and ansoparamedian fissures. Lobule X, the flocculonodular lobe (in orange), is bordered by the posterolateral fissure. This image is an adaptation of the diagram by Larsell, 1958 [80].
Table 1.
Clinical Characteristics.
Figure 2.
Stereological Assessment of PC Density.
This figure depicts our stereological approach for quantifying PC density in each region of the cerebellum. 2a displays a full sagittal section of the human brain. 2b represents the positioning of contour lines used to select the boundaries around each region of interest, within which sampling occurs. Lobules IV–VI are highlighted in green, crus I is highlighted in pink, crus II is highlighted in blue, and the flocculus of lobule X is highlighted in orange at the bottom of the section. 2c displays a hypothetical placement of counting frames (optical disector probes) within lobules IV–VI. The counting frame size has been increased to aid in visualization for this example. The white dotted line displays the randomly imposed grid over the contour, which regulates the distance between counting frames along the X–Y plane of the section. A counting frame is placed in the top-left of each grid cell if it will include a portion of the region of interest as designated by the contour line. Finally, 2d displays a histological section with an imposed counting frame that includes two PCs to be counted. The use of the optical disector probe dictates that PCs are counted when within the counting frame or at all touching the green line, but never when cells intersect with the red line.
Table 2.
Linear Mixed Models.a
Figure 3.
This graph demonstrates the findings from our linear mixed models (Table 2). Mean PC density was lower in the autism group in each region assayed, but this finding was most prominent in crus I and II (marked with *), where there was 19.8±9% and 21.7±9% lower PC densities, respectively (p = 0.039 and p = 0.032, t-test). (See the results of this test in Table 3). Lobule X PC density was lower in the males with autism (marked with ‡) than in control males (p = 0.05), and lower than female cases in the autism group (p = 0.022). Bars represent mean ± standard deviation. The number of subjects for each test from each diagnostic group and their gender are represented in the results tables (Tables 2 and 3).
Table 3.
Regional Differences in Purkinje Cell Density.
Figure 4.
Overall PC Density Decreases with Age.
Overall PC density obtained from the anterior, posterior, and flocculonodular lobes negatively correlates with age (R2 = −0.39±0.14, p = 0.030). Some cases were missing data from an individual region (as described in Methods) and were not included: one female and one male case from the autism and control groups were not included.
Figure 5.
Lobule X PC Density is Associated with Direct Gaze.
Lobule X PC density in the autism group negatively correlated with ADI-R question 50, which assessed the social and communicative use of eye contact (R2 = −0.75±0.04, p = 0.012). Higher scores on the ADI-R correspond with increased symptom severity. One case from the autism group was lacking sufficient ADI-R data to be included in this analysis.
Figure 6.
These figures represent the importance of estimating PC density in 3-dimensional space rather than along a 1-dimensional line (PC/mm3 rather than PC/mm). 7a and 7b display two views of a 3-dimensional model of a cerebellar folium, in which PCs (black) are arranged in a monolayer apposed to the granule cell layer (blue) that surrounds a central white matter tract (white). 7b displays two slices, one perfectly parallel to a plane of PCs, and the other, more realistically, transverse to this plane. 7c and 7d are cartoons demonstrating the PC arrangement within the parallel and transverse slice, respectively. Notice in 7d the PC layer is thicker, the degree to which depends on the slice position. Also notice that the selection of which PCs lie perfectly along a line is ambiguous (a few examples have been circled in red). This ambiguity introduces human error when performing an estimate of PC/mm that is eliminated in 3-dimensional estimates of PC/mm3 as are utilized in this study.
Figure 7.
Photomicrograph of Transverse Section.
This photomicrograph is an illustrative example of the arrangement of PCs on a transverse section. The PCs (purple) go out of focus to the left as the PC layer curves through the depth of the tissue section. The granule cell layer (blue) is visible below the PC layer, the molecular layer surrounds the PC layer (seen here as sparsely stained space), and the white matter tract is out of view to the right. Nissl-stained section from the celloidin collection used in this study, reference bar = 100 µm.