Three-dimensional histology: new visual approaches to morphological changes during neural regeneration.

Three-dimensional (3D) histology utilizes tissue clearing techniques to turn intact tissues transparent, allowing rapid interrogation of tissue architecture in three dimensions. In this article, we summarized the available tissue clearing methods and classified them according to their physicochemical principles of operation, which provided a framework for one to choose the best techniques for various research settings. Recent attempts in addressing various questions regarding the degenerating and regenerating nervous system have been promising with the use of 3D histological techniques.

Introduction

The study of neural tissue architecture and cellular morphology is important in neuroscientific research. Visualizing and understanding the three-dimensional (3D) spatial relationships of cells and molecules have been difficult using conventional histology. Serial sections of tissues are often required and this method is time- and labor-intensive, and leads to physical damage and depletion of tissue samples. Advances in optical microscopy and fluorescent probes have contributed to the emergence of tissue-clearing methods that allow intact, transparent tissues to be optically sectioned, imaged, and reconstructed in 3D in silico. Here, we briefly introduce the techniques that facilitate the rigorous histological evaluation of intact transparent tissues in 3D and better visualization of neuronal tracts and fibers, which could allow researchers to gain extraordinary insights into the processes of neurodegeneration and neuroregeneration.

The differences in refractive indices between different parts of the cell scatter visible light, lead to the perceived tissue opacity. Therefore, the key to transforming an opaque tissue into a transparent one is to homogenize these differences in refractive indices by immersing the tissue in an appropriate agent. The transparency of the tissue can be enhanced by prior delipidation (e.g., clear lipid-exchanged acrylamide-hybridized rigid imaging/immunostaining/in situ hybridization-compatible tissue hydrogel (CLARITY) by Chung et al., 2013) or dehydration (e.g., 3DISCO (3D imaging of solvent-cleared organs) by Ertürk et al., 2012), which also alters the nature of the tissue and thus its compatibility with various fluorescent labeling methods. Currently, there are no more than 20 tissue-clearing methods with multiple variants, each having its own strengths, difficulties, and issues of compatibility with fluorescent labeling methods and tissue types. Detailed description of each of these methods is beyond the scope of this brief review article; interested readers can refer to the excellent reviews published recently (Susaki et al., 2016; Treweek et al., 2016). To facilitate discussion, here we divide these tissue-clearing methods into three broad categories in terms of their operations and working principles: (1) aqueous-based refractive index homogenization (aqueous-based n-homogenization), (2) delipidation-facilitated n-homogenization, and (3) organic solvent-based n-homogenization. The delipidation step can be optional or partial (Lai et al., 2016), while all organic solvent-based n-homogenization methods require prior tissue dehydration to permit the homogeneous infiltration of hydrophobic organic solvents (). Such categorization allows the investigator to deduce the physicochemical changes made to the tissues and thus the method of choice for a particular research setting ().

An introduction of 3D histology and some image examples.

(A) General workflow of 3D histology. Fixed tissue is cleared by refractive index homogenization, during which the chemical reagent used can be aqueous or organic solvent-based. The tissue transparency achieved can be enhanced by prior delipidation. Fluorescent labeling (not shown) can be performed prior to tissue harvesting, or at any time before the final refractive index homogenization step. Finally, the fluorescently-labelled, optically-cleared tissue is then imaged with optical sectioning microscopy. (B) 3D histology images generated using CLARITY and FASTClear; all are color-coded projections of Z-stacks (scale bars as labeled), where a more red-shifted color indicates the further the signal from the objective. i) A layer II neuron within the motor cortex of a Thy1-GFP mouse, tissue processed using the CLARITY method. Inset: Enlarged view showing dendritic spines on the dendritic tree of the featured neuron. ii) A tiled Z-stack of a Thy1-YFP mouse brain slice consisting of 858 images stitched together, tissue processed using the FASTClear method (Z-depth 1,100 μm). iii) GFAP staining of wild-type mouse cortical tissue with tyramide signal amplification, demonstrating astrocytic processes around a blood vessel forming the blood-brain barrier; tissue processed using the FASTClear method (Z-depth 24 μm). The shallow Z-depth was due to the limited working distance of the high-magnification objective. iv) A full layer tiled Z-stack of the motor cortex of a Thy1-GFP mouse, demonstrating sparsely-labelled layer V neurons and a layer III neuron, tissue processed using the CLARITY method (Z-depth 370 μm). CLARITY: Clear Lipid-exchanged Acrylamide-hybridized rigid imaging/immunostaining/in situ hybridization-compatible tissue; FASTClear: free-of-acrylamide, SDS-based tissue clearing; GFAP: glial fibrillary acidic protein; 3D: three-dimension.

As a general rule, aqueous-based n-homogenization (Susaki et al., 2016) best preserves tissue structures and fluorescence, but the lack of adequate permeabilization makes it incompatible or poorly compatible with immunohistochemistry. Moreover, the transparency of the tissues is not as good as the other two categories. In delipidation-facilitated tissue-clearing methods, the tissues are washed free of lipids in detergents with subsequent n-homogenization using a suitable aqueous medium (Chung et al., 2013; Susaki et al., 2016). The aqueous media and permeabilization result in good preservation of endogenous fluorescence and compatibility with immunohistochemistry, respectively, but they are incompatible with most lipophilic tracers and limited by the long time required for adequate tissue delipidation. In organic solvent-based n-homogenization methods, the tissues are dehydrated and immersed in a hydrophobic organic solvent for homogenization (Ertürk et al., 2012). They are compatible with immunohistochemistry (Renier et al., 2014), but commonly destroy the fluorescent proteins expressed using genetic and viral tools if the pH is not well-controlled (Schwarz et al., 2015). With the differing characteristics and advantages of different methods of tissue-clearing, one size does not fit all (). Thus, when choosing a method of tissue-clearing for specific research questions and goals, a thorough understanding of the physicochemical basis of individual methodologies and their impact on tissues is important, as it allows the researcher to select the methodology appropriate for the desired application.

The novel perspectives brought by this added dimension in histology are obvious, especially since these techniques can be applied to previously insoluble questions. Some examples related to neural regeneration include the convenient delineation of corticospinal projection pathways (Soderbolm et al., 2015), visualization of axonal branching patterns after optic nerve crush injury (Luo et al., 2014) and peripheral nerve injury (Jung et al., 2014), and the assessment of axonal regeneration and glial responses after spinal cord injury (Ertürk et al., 2011). Our laboratory has compiled a series of 3D images of tissues processed with different optical clearing methods () and rendered them in the form of a video that clearly demonstrates the power of this technology ().

Despite the significant advances in tissue-clearing, several methodological challenges remain unresolved. First, the limited diffusion and penetration of antibodies: even with maximal tissue permeabilization by complete delipidation of tissues, the dense location of antigens at shallower depths can rapidly consume antibodies, limiting deep immunostaining. This also contributes to difficulty in the translation of these techniques to human neural tissues, which has been inherently difficult due to the substantially larger tissue volumes and variable sample quality (Lai et al., unpublished observations). Finally, with the use of ultrafast light-sheet microscopic imaging, the volume of data generated commonly lies in the gigabyte to terabyte range, requiring rigorous and robust computational infrastructures for storage and analyses (Susaki et al., 2016). Since these techniques are still in their infancy, they need to be individually optimized by users in order to obtain the best results for their own projects, emphasizing once again the importance of understanding the working principles underlying these methodologies.

The perfect combination of tissue-clearing, fluorescent labeling, and optical sectioning microscopy has led to the birth of 3D histology, providing new, exciting, and powerful visualization approaches to the study of the nervous system. Reminiscent of the invention of microtome sectioning, tissue stains, and microscopy that gave birth to histology centuries ago, we envision that the continued methodological innovations in interrogating the brain will achieve new heights in neuroscience research.

Table 1

Selected tissue clearing methods