Princeton RAtlas: A Common Coordinate Framework for Fully cleared, Whole Rattus norvegicus Brains

Background

Recent developments in tissue clearing and imaging allow us to gather whole-brain connectivity in intact brains, together with classic neuronal tracing techniques.Though tissue fixation and preparation predated neuroscience, its use was refined by Golgi and further modified and popularized by Ramón y Cajal (1911). To this day, after experiments, neuroscientists often remove the brain and slice it into sections that are then mounted to microscope slides for imaging. This allows for both post-hoc verification of experimental targeting and, if paired with staining or immunohistochemistry, learning about the tissue. However, this process is hard to automate and time-consuming: each step of fixation, removal, slicing, and imaging for each slice can take hours. Additional immunohistochemistry or staining can take weeks. To decrease the amount of labor required to process each brain, researchers will frequently only keep a subset of the brain slices near the relevant regions of study. Often this is more than sufficient, but this approach can easily miss potentially relevant information in the discarded sections, and sectioning itself can introduce information and aberrations to data that are difficult to identify post-hoc. Address this issue, several groups have developed methods to clear the lipids and pigments from whole brains, allowing for imaging through the entire tissue without the need for sectioning and often dramatically decreasing imaging time.

Building on advances from the early 1900s, tissue clearing has experienced a resurgence in neuroscience over the last 20 years. Methods like CUBIC (Susaki et al., 2015), CLARITY (Chung and Deisseroth, 2013), FluoClearBABB (Schwarz et al., 2015), and the DISCO methods (Renier et al., 2014; Pan et al., 2016) have become popular with researchers studying mice (see Molbay et al., 2021 for a review) yet those of us studying rats have not yet fully embraced this new methodology. One major impediment for broad acceptance of these methods in larger-brained animals is that existing clearing techniques do not consistently produce fully cleared, intact brains, prohibiting the study of subcortical structures in whole brains and decreasing their utility over slice histology. However, several advances have been made: Woo and colleagues nearly doubled the length of the mouse-centric CUBIC method to 20 days to improve clearing rats (Woo et al., 2018); Branch and colleagues successfully cleared entire rat brains by splitting samples into two hemispheres before clearing, together with increasing the timing of stages of the mouse iDISCO protocol (Branch et al., 2019); and a third method, FluoClearBABB, cleared the majority of tissue from an intact rat brain in 8 days while controlling the pH during dehydration, which, similar to uDISCO, preserved fluorescence, though the clearing at the center of the tissue is unclear from their manuscript (Stefaniuk et al., 2016).

Similar advances in imaging techniques allow for faster image acquisition. Typically, slices are imaged on confocal or 2-photon microscopes, which use point scanning methods that scan across and through tissues to excite and observe fluorescence. In contrast, light-sheet microscopy images planes instead of points, decreasing the time needed to go through the same volume of tissue. Like other methods key to modern developments, light-sheet microscopy also was first proposed in 1902 (Siedentopf and Zsigmondy, 1902), but recent developments in CMOS sensors that can acquire large images quickly, as well as advances in light sculpting (Keller et al., 2008), moving illumination and detection (Ahrens et al., 2013), and others [see review Hillman et al. (2019)] were key in making whole-brain imaging available to neuroscientists.

Inspired by these advances, we set out to produce two reliable, worker-friendly protocols for efficient clearing of intact rat brains, based on the uDISCO and iDISCO protocols. Coupled with light-sheet imaging, we can now image intact rat brains, including subcortical structures. We use these clear brains to form a light-sheet-based rat brain template for the rat neuroscience community.

Whole-brain clearing of Rattus norvegicus brains

We first set out to improve existing clearing techniques to work consistently on whole adult Rattus norvegicus brains of any size or sex. Inspired by many, we primarily focused on the DISCO techniques and the timing of stages, as this was the most effective in our preliminary explorations. We successfully made a worker-friendly protocol (Figure1A, maximum 8 h days, compared with 10–16 h days in other protocols) that also resulted in consistently clear brains, visible in the field-standard qualitative evaluations (Figure1B and 1C) and a newer quantitative evaluation method, Fourier ringcorrelation–based quality estimation (FRC-QE) (Figure 1E) (Preusser et al., 2021). When applied appropriately, FRC-QE allows for comparable measurements across experiments and provides quantitative data in addition to the field-standard qualitative data.

Figure 1. Modifications to iDISCO and uDISCO protocols allow for fully cleared ratbrains. A. Schematics of iDISCO protocol and the modified rat iDISCO and uDISCOprotocols (abbreviations DCM, BABB, and DBE all refer to solutions inRecipes). B. Example images of brains before clearing and after clearing with each protocol from the schematic. The uncleared brain is approximately 2.5 cm long and 1.7 cm wide. All images were taken on the same piece of paper for easy comparison. C. Computational slices through each clearedbrain in B. Labels (a253, k320, z268) reflect the animal’s name. D. Drawing showing the orientation of the brain (left) and the FRC-QE scores (right) for each brain in B along the coronal axes in slice units. E. Quantification of the values in (D), with a Bonferroni-corrected t-test. Horizontal white bars are median values, and the violin limits are the 10% (bottom) and 90% (top) of the data. Asterisks indicate significance p < 0.05 compared with iDISCO a253.

Princeton RAtlas

Once we could reliably obtain fully cleared brains, we set out to make a common coordinate framework for rats using whole-brain, light-sheet imaging. We imaged 18 brains and chose the best eight for an averaged template brain. We chose these eight based on a manual inspection of quality and based on external genitalia: we chose four male and four female brains without obvious damage or imaging aberrations, like streaking. The four female brains were all of sufficient quality that we did not need to procure additional brains, and we chose four of the male brains to match our number of females. We used an iterated, paired averaging scheme to create the seed brain (Figure 2A) and then aligned each brain to this seed, averaged the result to create the new seed, and repeated this process two additional times. The final averaged result of all brains aligned to this final seed brain became the Princeton RAtlas (PRA, Figure 2B). This coordinate space is agnostic to the sex of the animal. We also provide mPRA, a common coordinate space using only animals with visible penis and testes, and fPRA, a common coordinate space using only animals without visible penis or testes.

Figure 2. Creation of the Princeton RAtlas (PRA). A. Schematics showing the paired alignments to create the seed brain (left, middle) and the multilevel refinement resulting in the PRA (right). B.Summed axial projection of the PRA. mBrains and fBrains are male and femalebrains respectively, defined by external genitalia.