1.

Introduction

Neurons rely on the complex vascular network to supply them with metabolites and oxygen. Dysfunction of the microvasculature is associated with several cerebral disorders, including schizophrenia, traumatic brain injury, dementia, and stroke.^1–3 Fundamental to exploring these relationships is a consistent and reliable methodology to visualize the three-dimensional cerebral microvasculature under normal and pathological conditions. Researchers would benefit in particular from whole-brain models of the cerebrovascular structure. The crucial and challenging task of creating useful whole-organ visualizations requires the pairing of compatible labeling, clearing, and imaging techniques using repeatable and robust methodologies and equipment.

Vascular labeling techniques include secondary antibody staining via diffusion, transgenic endothelial markers, and vessel labeling perfusion-based methods. While secondary antibody staining requires long labeling periods (up to seven days for mouse hemibrains),^4,5 perfusion-based vascular labeling is quick and easily deliverable.^6–8 However, the quality of vascular labeling can vary with perfusion-based labels due to aggregate formation, leakage, rapid photobleaching, and incomplete perfusion.^6,9–11 For example, for immunolabeling, antibody diffusion from the exposed surfaces inwards can lead to heterogeneous vessel labeling, especially for deeper vessel structures.¹² For cardiac perfusion methods, capillary labeling can be heterogeneous, presumably due to insufficient pressure to maintain consistent perfusion of all capillaries.¹³

Tissue clearing procedures have progressed rapidly, enabling whole-organ clearing using aqueous solvents, organic solvents, or hydrogel infusion with electrophoretic tissue clearing.^4,14,15 Recent advances in light-sheet microscopy (LSM) enable detailed three-dimensional imaging of intact cleared organs with excellent optical sectioning capabilities.^4,12,16 LSM can achieve short whole-organ imaging time ($\sim 1$ to 2 h for a mouse brain) while maintaining $\mu \mathrm{m}$-level resolution and mitigating photobleaching. Recent reports demonstrate whole-organ vascular visualization in the brain and teeth.^14,17 However, the use of complicated clearing procedures necessitating lengthy labeling techniques and non-standard equipment has precluded the routine use of these methodologies for scientific research.^6,18 Here, we describe a simple protocol combining perfusion-based labeling with a two-day clearing step that facilitates whole-brain, three-dimensional microvascular imaging and characterization.

2.

Materials and Methods

3.

Results

3.1.

Cerebral Vasculature Visualization with Lectin-Dylight-649

To validate the efficacy of Lectin-Dylight-649 as a vascular label in cleared tissue, we imaged 1-mm-thick coronal brain sections from C57BL/6J mice. Across different samples from separate animals, the cerebral vasculature is visualized with high contrast [Fig. 1(a)]. Representative MIP [Fig. 1(b)], three-dimensional [Fig. 1(c)], and fly-through views (Video 1) provide complementary visualizations of the dense microvascular network in the mouse brain. Qualitatively, perfusion of Lectin-Dylight-649 provides a robust method for vascular labeling.

Fig. 1

Visualization of Lectin-Dylight-649 in cleared 1-mm-thick coronal brain sections. (a) Single z-stack view of a representative section from different brains ($n=4$). Scale bar is $100\mu \mathrm{m}$. (b) Representative MIP of 10 fluorescence emission images ($60\mu \mathrm{m}$ thickness), showing the dense microvascular network. Scale bar is $100\mu \mathrm{m}$. (c) Three-dimensional view of 80 fluorescence emission images taken from a single z-stack. Scale bar is $100\mu \mathrm{m}$. Fly-through rendering of the dataset rendered in three dimensions in (c). A total of 80 fluorescence emission images are contained in this video, spanning a thickness of $\sim 470\mu \mathrm{m}$. (Video 1, MPG, 1948 KB [URL: https://doi.org/10.1117/1.NPh.8.2.025004.1]).

3.2.

Lectin-Dylight-649 Labels Vasculature with High Contrast-to-Background Ratio

To evaluate the labeling capacity of perfused Lectin-Dylight-649, we compared its CBR to the Tie2-GFP transgenic vascular model. We measured fluorescence intensities of vascular and background perivascular ROIs for each fluorophore in the same region of dual-labeled brain slices using confocal microscopy ($n=4$ slices) [Figs. 2(a)–2(d)]. Vascular intensity values were shown to be higher with Lectin-Dylight-649, while background intensity values were higher for Tie2-GFP fluorescence ($n=100$ ROIs each) [Fig. 2(e)]. Median GFP and Lectin-Dylight-649 fluorescence emission from the vascular regions were similar ($p=0.71$), but the median GFP and Lectin-Dylight-649 fluorescence emission from the tissue regions differed (mean rank fluorescence emission difference = 820, $p<0.0001$) [Fig. 2(e)]. The measured fluorescence CBR of Tie2-GFP is $\sim \mathrm{three}$ times lower than that of Lectin-Dylight-649 [Fig. 2(e)]. Collectively, these data demonstrate improved vascular visibility with the use of Lectin-Dylight-649 over endogenous GFP, in combination with iDISCO+ optical clearing.

Fig. 2

With iDISCO+ optical clearing, Lectin-Dylight-649 labeling exhibits an $\sim \text{threefold}$ higher CBR than Tie2-GFP. (a), (b) Lectin and Tie2-GFP fluorescence confocal images, respectively, shown as MIPs taken from the same region of the brain. Both images are on the same colorscale. (c), (d) Representative ROIs taken from (a), (b) in perivascular (green) and vascular (red) regions. (e) Comparison of Lectin and Tie2-GFP intensity profiles in vascular and perivascular regions ($n=400$ ROIs each).

3.3.

Whole-Brain Cerebral Microvascular Imaging in the Intact Brain

To visualize whole brain microvasculature, Lectin-Dylight-649 was administered via retro-orbital injection followed by cardiac perfusion and the modified iDISCO+ protocol on intact excised mouse brains ($n=3$). The modified iDISCO+ protocol considerably increased the transparency of the intact brain, compared with its appearance before treatment [Fig. 3(a)]. LSM microscopy yielded several tiles, which were stitched and rendered in three dimensions while maintaining $\mu \mathrm{m}$ resolution over the cm-scale organ. Three-dimensional visualization depicts the detailed cerebrovascular structure [Figs. 3(b), 3(c), Video 2 and Video 3]. Inspection of $200\text{-}\mu \mathrm{m}$ MIPs revealed high-resolution visibility of the vasculature throughout the brain, including internal sections [Figs. 3(d)–3(k)].

Fig. 3

Three-dimensional whole-brain images allow dynamic structure analysis from micrometer to centimeter scales. (a) The entire brain cleared with modified iDISCO+. Renderings of the cerebrovasculature: (b) top view and (c) sagittal view. (d)–(g) $200\text{-}\mu \mathrm{m}$-thick MIPs of transverse virtual sections taken in steps of 1 mm through the brain. (h)–(k) Magnified view outlined in (d) (scale bar $500\mu \mathrm{m}$). Scale bar in all $\text{images}=500\mu \mathrm{m}$. Fly-through rendering of 2.2 mm of intact, cleared brain with vasculature labeled with Lectin-Dylight-649. Images collected with the Zeiss Z.1 light-sheet microscope. Imaris software was used to render the dataset (Video 2, MPG, 24,810 KB [URL: https://doi.org/10.1117/1.NPh.8.2.025004.2]). Stitched fly-through rendering of 3.9 mm of an intact, cleared brain labeled with Lectin-Dylight-649. Light-sheet microscope was used with dual-side fusion and pivot scanning to equalize emissions across the whole brain and reduce shadowing effects, respectively. Imaris software was used to stitch and render a $4\times 3$ array of tiles. Total volume size is $31.6\mathrm{mm}\times 23.6\mathrm{mm}\times 3.9\mathrm{mm}$ (Video 3, MPG, 24,132 KB [URL: https://doi.org/10.1117/1.NPh.8.2.025004.3]).

4.

Discussion

Here, we created a simple method to enable visualization of the cerebral microvasculature in an intact mouse brain with micrometer resolution. We found that Lectin-Dylight-649, which is reported to be more stable than the FITC derivatives,^10,11 to be an effective method for labeling blood vessels. Our data demonstrate that Lectin-Dylight-649, coupled with a modified iDISCO+ protocol for clearing, achieves superior CBR as compared with endogenous GFP in transgenic Tie2-GFP mice. Preparation of Lectin-Dylight-649 and administration into the bloodstream via retro-orbital injection both are techniques that are simple to master. Thus, our method for vessel labeling should be easy to integrate into studies with other mouse models.

We demonstrated whole-brain imaging and rendering of the cerebral microvasculature is possible with a two-day protocol. We found that using a lectin-based labeling technique and adaptation of the iDISCO+ methodology allowed for a chemically compatible protocol that was achievable in two days. Antibody-based labeling techniques can take $\sim \text{one week}$ to complete due to the need to wait for the diffusion of the antibodies and labels from the solution into the tissue sample. The required time can be reduced with the use of tissue sectioning, but with an associated loss of connectivity of features within the individual sections.

Whole-brain vascular visualization was recently demonstrated, as well as the ability to see vasculature in other organs.^14,15,17 To achieve such results, groups used complicated clearing procedures necessitating lengthy labeling techniques ($\sim \text{weeks}$) and non-standard equipment (e.g., electrophoretic tissue-clearing chambers and hydrogel embedding equipment), which may impede the routine use of these methods for scientific research.^6,18 Our group and others have used cardiac perfusion to administer DiI as a vessel labeling agent.^9,23,24 However, recent work by Salehi et al.¹³ suggests that adjuvant methods (i.e., sodium nitroprusside) to dilate the vasculature is required for DiI to label the vasculature robustly. Our previous data²⁵ and data presented here demonstrate that Lectin-Dylight-649 enters the circulatory system via retro-orbital injection and can label vasculature while the animal is alive. We postulate that this in vivo vessel labeling approach is a much more straightforward and more effective approach for vessel labeling.

Although the present study describes a translatable protocol for vascular labeling, we did not test the compatibility and visibility of other labels, including fluorophore-conjugated antibodies, with our optical clearing protocol. For example, our previous work centered on FocusClear-based optical clearing has included the use of nuclear stains (i.e., DAPI), fibrillar beta-amyloid (Thioflavin S), and hemosiderin labels (Prussian Blue).^24,26 Also, previous work demonstrates that organic solvent-based approaches may quench endogenous fluorescence; thus, the Tie2GFP fluorescence may have been quenched and resulted in lower CBR that we report here.²⁷ Future work is required to demonstrate the degree to which fluorescence of Lectin-Dylight-649 is preserved with time.

Future work will involve systematic refinement of our clearing protocol with alternate versions of the “*DISCO” protocol (e.g., FDISCO, SDISCO, and UDISCO). These newer protocols are designed to increase photostability, preserve endogenous fluorescent signals, and clear the whole body.^28–30 Additionally, our data suggest that the vessel labeling and clearing approach can be applied to investigate both normal and abnormal vascular architecture in other organs.

5.

Conclusion

We describe an experimental protocol to visualize the three-dimensional microvascular architecture of the intact mouse brain. The combination of retro-orbital injection of Lectin-Dylight-649 to label the vasculature, the clearing process of a modified iDISCO+ protocol, and light-sheet imaging collectively enables a comprehensive view of the cerebrovasculature. Eliminating additional dehydration/rehydration steps and immersion labeling reduced clearing time to two days. Our protocol is expected to facilitate rapid three-dimensional visualization of the microvascular network for a wide variety of biological and biomedical applications.