Supplementary Figure 3: Interpretation of sectioning angle in Allen Institute in situ hybridization tissue sections and HGEA boundary mapping.

(a) Histological processing of brain tissue requires careful sectioning for accurate registration of data to ARA levels. For example, a tissue section of Nov gene expression from the Allen Brain Atlas database does not accurately register to any one ARA atlas level (80, 82, 84), because of oblique cutting angle along the dorsoventral axis during histological sectioning (dashed red lines provide guides for comparison of dorsal, intermediate and ventral parts of the tissue across all images). Rather than register to one atlas level, we used multiple landmarks throughout the tissue section (1-10, 1 = shape of corpus callosum/ external capsule white matter, 2 = position of the CA1/SUB border relative to the tip of the DG granule layer, 3 = shape of the dorsal diencephalon, 4 = position of the posterior commisure, 5 = thalamic nuclei cytoarchitecture, 6 = ventral tip of the DG granule cell layer, 7 = separation of the CA3/CA1 pyramidal layer relative to the rhinal sulcus, 8 = shape of the MM and LM, 9 = position of the CA3/CA1/SUB pyramidal cell layers, 10 = amygdala nuclei cytoarchitecture) to register tissue locally to the appropriate ARA atlas section. Ultimately, the Nov tissue section is better described by a “composite” of ARA levels 80-84, where the ventral part of the hippocampus is closer to ARA 80 and dorsal part of the hippocampus is closer to 84 with a linear progression along the dorsoventral axis. (b) Histological sectioning angle is consistent among tissue sections cut from the same brain, but can be different across many other brains. Here, we show a different tissue section from the same brain showing Nov gene expression and a tissue section cut from a different brain showing Dcn gene expression. The composite ARA sections on the left show that the Nov tissue section contains a similar dorsoventral gradient of 4 atlas levels as shown in a with the dorsal part of the tissue more caudal than the ventral part of the tissue. In contrast, the Dcn tissue section contains a steeper gradient of 7 atlas levels cut where the ventral part of the hippocampus is more caudal than the dorsal part of the hippocampus (opposite of Nov). On the right, are corresponding composite HGEA atlas levels showing that the dorsal limit of positive Dcn gene expression corresponds to the mapping of the CA1i/CA1v boundary at HGEA level 85 (arrow). (c) Oblique tissue sectioning angle can be estimated using the composite ARA atlas levels (each ARA atlas level is 0.1mm apart). For the Nov and Dcn brain tissue sections in b, we can estimate the oblique sectioning angles relative to the ARA atlas (dashed black lines) for each brain as a right triangle where the tangent of the sectioning angle is the displacement of the ARA atlas level between the dorsal and ventral part of the hippocampus over the dorsoventral distance of the hippocampus (~5.0mm). For the Dcn tissue, the sectioning angle was calculated at -8.0 degrees whereas the Nov tissue sectioning angle was calculated at 4.6 degrees. Note, the greater the distance between the two hypotenuses, the greater the inaccuracy for comparison along the dorsoventral axis. (d) Tissue section registration is not only affected by oblique sectioning angle in the dorsoventral axis but also along the mediolateral axis. An example of Calb1 gene expression in a tissue section shows a 2 section ARA atlas difference in the dorsoventral direction and a 2 section ARA atlas difference between the left and right hippocampus (arrows point to corresponding ARA level numbers for each region). Below, a graphical illustration of the plane of sectioning for the Calb1 tissue section (shaded blue) relative to ARA level 91 (clear plane with dashed lines). (e) Because of differences in dorsoventral and mediolateral sectioning angles, it is important to examine the progression of gene expression across multiple rostrocaudal coronal tissue sections. Three adjacent tissue sections containing gene expression for the CA2 marker Amigo2 (with corresponding HGEA levels below) show the rostrocaudal progression of Amigo2 corresponds to the HGEA boundaries of the CA2 with the CA1d, CA1i, and CA1v. At the most caudal level (HGEA 82), a thin lamina of Amigo2 positive cells are located in the deepest part of the pyramidal layer (red and blue boxes correspond to magnified images on right). (f) Examination of the mediolateral progression of gene expression in sagittal-cut tissue can corroborate data from coronal sections (for sagittal-cut tissue sections, differences caused by oblique sectioning angles apply to the dorsoventral and rostrocaudal axis). Amigo2 expression in three adjacent sagittal sections are shown with corresponding HGEA sagittal atlas levels below. Similar to the coronal data, the mediolateral progression shows a similar descent of Amigo2 expression across the dorsoventral axis (dorsal and ventral CA2 are separated at HGEA 4 due to the curve of the hippocampus relative to the straight cut of the tissue). In the most lateral HGEA section 1, a laminar labeling pattern of Amigo2 is shown similar to the coronal section at HGEA 82 (d). Together, examination of both coronal and sagittal data provides corroboration when mapping HGEA boundaries.