A Myocardial Tissue Clearing Protocol for High-Resolution 3D Microvascular Imaging

Introduction: Cardiac diseases such as fibrosis or coronary microvascular disease cause structural changes to microvascular networks within the myocardium. Although histology may be used to identify tissue as healthy or diseased, characterizing the geometry and structure of capillary networks requires a three-dimensional (3D) method. Fluorescent proteins that target endothelial cells can be used to identify the vascular wall under confocal or fluorescence microscopy. However, many tissues, especially the myocardium, have high levels of autofluorescence and heterogeneous refractive indexes (RI), both of which are fundamentally detrimental to microscopic resolution. Different tissue clearing strategies have thus been identified to address these issues and maximize image clarity. Broadly, high-resolution 3D imaging of microvascular networks can be accomplished through the following steps: fluorescent vessel staining, autofluorescence quenching, tissue clearing, and confocal microscopy visualization. Protocols for myocardial vasculature imaging often prepare samples by utilizing a Langendorff perfusion apparatus to circulate reagents through an entire heart. However, whole-heart perfusion is not always feasible when working with large animals or donor samples.

In this abstract, we present a protocol for confocal microscopy of microvascular networks in thin sections of myocardial tissue, without relying on whole-heart perfusion. The thin myocardial tissue sections are incubated in reagents designed for endothelial cell staining, autofluorescence quenching, and tissue clearing (delipidation and RI matching). This protocol enables successful 3D imaging and computational reconstruction of myocardial microvasculature. This method has been demonstrated using murine tissue, and results suggest that it may also be applicable to tissues from larger animals such as swine.

Materials and

Methods: Heart Preparation: PS19 mice (strain C57BL/6) were anesthetized with isoflurane then sacrificed. Hearts were fixed in 4% PFA overnight at 4ºC, and subsequently washed in PBS and mounted in 4% agarose. The gel-enclosed hearts were excised in a cubicle section and mounted onto a vibratome, then transversely sliced at 300 μm thickness.

Reagents: Two home-made CUBIC reagents (Reagent I and Reagent II) [1-2], 1X wash buffer, and TrueBlack solution were prepared according to Table 1.

Clearing, Staining, and Quenching Procedure: Myocardium sections were incubated in 50% Reagent I (diluted in dH2O) and 100% Reagent I at 37ºC on a shaking incubator for 18 hours and 24 hours, respectively. Sections underwent washes in PBS prior to permeabilization with 0.3% Triton X-100 in 1X wash buffer at 37ºC. After PBS rinsing, sections were covered with TrueBlack solution or PBS and incubated for 5 minutes. Sections then were placed in a 37ºC shaking incubator for 48 hours in 1% tomato lectin-DyLight 594 (Vector Lab, DL-1177-1) diluted in PBS. Finally, the sections were incubated in 50% Reagent II (diluted in PBS) for one hour and 100% Reagent II overnight. The sections were mounted in slides with Reagent II.

Imaging: Samples were visualized using a TIRF/FRAP/CSU-X1 Spinning Disk Confocal System. Nikon NIS-Elements software was used for image processing. Image resolution was 0.65x0.65 μm, with 1 μm z-stack steps. Frangi filters designed to detect vessel-like structures were used to create masks. 3D structures were visualized in paraview.

Results, Conclusions, and Discussions: The clearing, staining, and quenching protocol produced highly resolvable networks of blood vessels under a confocal microscope (Figures 1 and 2).

Figure 1 shows 3x3 field of views for tissue samples treated with PBS and TrueBlack (panels A and B, respectively). Tissue quenching with TrueBlack is a key component of our protocol. Quenching appears to lead to reduced tissue autofluorescence, but could lead to issues such as increased laser absorbance. Overall, in our experience, the benefits of tissue quenching outweigh its shortcomings. This is apparent in Figure 1B (tissue quenched with TrueBlack solution), which offers a higher signal-to-noise ratio compared with results in Figure 1A (tissue treated with PBS).

Figure 2 illustrates some aspects of our image processing protocol using tissue treated with TrueBlack. These consist mainly of denoising (gray-scale image on top panel), and application of a Frangi filter designed to detect vessel-like structures, followed by a threshold-based segmentation (bottom panel).

Our results demonstrate that the primary hindrances of coronary microvascular imaging were overcome. Specifically, autofluorescence was minimized, and heterogeneous RI’s were corrected. Total tissue preparation time took about 5 days; though, this exact timing can span from 3 to 14 days. Given that autofluorescence is so prevalent in myocardial samples, the lengthy steps aimed at autofluorescent management, namely incubations in CUBIC reagent I, are the greatest determinants of overall experimental efficiency. Developing a balance between the bottle-necking step of delipidation and maximum autofluorescence elimination marks one of the most challenging facets of high-resolution 3D myocardium imaging.

The proposed protocol enabled high-resolution imaging of capillary beds in murine myocardial tissue. Given these promising results, future research will expand this protocol to tissue samples from swine or donor hearts. Unlike prior studies, this work would require only small tissue sections. Ultimately, microvascular networks derived through this protocol will be used to develop computational models of coronary flow and tissue perfusion. This framework will make it possible to understand differences in coronary function on healthy vs. pathological states.

[1] Matsumoto et al. Nature Protocols 2019.
[2] Susaki et al. Nature Protocols 2015.

Acknowledgements (Optional):