Literature DB >> 31527839

The mesoSPIM initiative: open-source light-sheet microscopes for imaging cleared tissue.

Fabian F Voigt1,2, Daniel Kirschenbaum3, Evgenia Platonova4, Stéphane Pagès5,6, Robert A A Campbell7, Rahel Kastli8,9, Martina Schaettin9,10, Ladan Egolf8,9, Alexander van der Bourg8,9, Philipp Bethge8,9, Karen Haenraets9,11, Noémie Frézel9,11, Thomas Topilko12, Paola Perin13, Daniel Hillier14,15, Sven Hildebrand16, Anna Schueth16, Alard Roebroeck16, Botond Roska17,18, Esther T Stoeckli9,10, Roberto Pizzala19, Nicolas Renier11, Hanns Ulrich Zeilhofer9,11, Theofanis Karayannis8,9, Urs Ziegler4, Laura Batti5, Anthony Holtmaat6, Christian Lüscher6,20, Adriano Aguzzi3, Fritjof Helmchen8,9.   

Abstract

Light-sheet microscopy is an ideal technique for imaging large cleared samples; however, the community is still lacking instruments capable of producing volumetric images of centimeter-sized cleared samples with near-isotropic resolution within minutes. Here, we introduce the mesoscale selective plane-illumination microscopy initiative, an open-hardware project for building and operating a light-sheet microscope that addresses these challenges and is compatible with any type of cleared or expanded sample ( www.mesospim.org ).

Entities:  

Mesh:

Year:  2019        PMID: 31527839      PMCID: PMC6824906          DOI: 10.1038/s41592-019-0554-0

Source DB:  PubMed          Journal:  Nat Methods        ISSN: 1548-7091            Impact factor:   28.547


Over the course of the past decade, tissue clearing methods have reached a high level of sophistication with a wide variety of approaches now available[1]. Common clearing techniques fall into two categories: Approaches using organic solvents, for example the DISCO family of protocols[2-5], and methods using aqueous solutions such as CLARITY[6] and CUBIC[7, 8]. To image samples processed with these methods, a wide range of commercial light-sheet microscopes can be utilized (Supplementary Note 1). Nonetheless, many users still experience significant shortcomings when using existing instruments to image cleared samples: For example, the imaging chamber, sample holders, and sample stages of microscopes designed for time-lapse imaging in developing embryos are usually undersized for cm-sized cleared samples. Even if the setup was specifically designed with clearing in mind, accommodating large samples can be a significant challenge: Modern clearing techniques can render a whole mouse central nervous system (CNS) or even entire mice[4, 5] transparent; yet, there are no instruments capable of imaging such samples without cutting. In addition, many instruments achieve optimal image quality only for a limited selection of immersion media, often restricted by the specifications of existing microscope objectives. As typical refractive indices in light-sheet microscopy range from 1.33 (for water) to 1.56 (for a mixture of benzyl alcohol and benzyl benzoate, BABB[9]), this limits several commercial microscopes to a narrow subset of clearing techniques (Supplementary Table 1). To overcome these limitations, we set out to design a modular light-sheet microscope that combines a large imaging volume, excellent image quality over large fields-of-view (FOV) with simple and versatile sample handling (Fig. 1a, Supplementary Note 2, Supplementary Video 1). To allow multi-view acquisitions, we adopted an instrument layout similar to the original selective plane illumination microscope (SPIM)[10] with a horizontal detection path and a vertical sample rotation axis (Supplementary Fig. 1 and 2). The instrument is equipped with a zoom macroscope in the detection path which allows FOVs of 2-21 mm in combination with a sCMOS camera. This enables users to view large samples and then zoom in to reveal minute details such as individual axons (Fig. 1c-e). Therefore, we have termed the instrument the mesoscale selective plane-illumination microscope (mesoSPIM).
Figure 1

Example demonstrations of the mesoSPIM light-sheet mesoscope.

a) Overview of the mesoSPIM instrument (Version 4). Left inset: 3D-printed sample holder with magnetic quick-exchange mount. Alternatively, samples can be mounted in a cuvette. Right inset: Photograph of a Thy1-YFP mouse brain during image acquisition. b) Comparison of axial image quality achieved in a CLARITY-cleared VIP-tdTomato mouse brain for standard light-sheet illumination (left) and for the axially swept light-sheet mode, ASLM (right). Images are maximum intensity projections over 250-μm range. c) Whole-CNS imaging with the mesoSPIM. A whole central nervous system was dissected from a Glycine Transporter-2 EGFP (GlyT2-EGFP) mouse and cleared using the X-CLARITY protocol. The inset shows glycinergic neurons in a subregion of the spinal cord. d) Overview image of a CLARITY-cleared TPH2Cre-tdTomato mouse brain. e) Volume rendering of sparsely labeled Purkinje cells and their axonal projections (arrow) from the dashed box in (d). Inset: individual Purkinje cell (arrow head). Panels (d) and (e) use the same colorbar as (b). Scale bars of all insets: 200 μm. The imaging experiments were conducted once using animals aged 6 weeks (c) and 2 months (d).

To streamline sample handling, we use magnetic quick-exchange mounts for the immersion cuvettes and sample holders. These mounts allow rapid switching between different immersion media and samples within less than a minute (Supplementary Video 2). Samples are usually mounted in a cuvette[11] or clamped in a 3D-printed holder (Fig. 1a, Supplementary Fig. 3). The mesoSPIM light-sheet is generated by rapidly scanning a Gaussian beam in the vertical plane using a galvo scanner similar to a digital scanned light-sheet microscope (DSLM)[12]. This approach has several advantages when imaging large cleared samples: Firstly, it results in uniform image brightness as each part of the FOV is illuminated with the same intensity. Secondly, when changing the detection FOV using the detection zoom, the height of the light-sheet can be easily adapted by regulating the amplitude of the galvo waveform. Finally, DSLM illumination reduces shadow artefacts[13]: In light-sheet microscopy, any sample feature that absorbs or refracts the excitation light casts a shadow across the FOV. The resulting images are full of stripes along the illumination direction which complicates image analysis and visualization. Because in DSLM each part of the sample is illuminated with a scanned cone of excitation light, the shadow zone behind absorbing objects can be shortened by increasing the opening angle of the cone (equivalent to increasing the excitation NA). In the mesoSPIM, we use an NA of 0.15 to achieve homogenous illumination with minimal shadow artifacts. A Gaussian beam with increased excitation NA does, however, have a reduced Rayleigh range which leads to axially blurred images outside of the narrow waist region of the light-sheet (Fig. 1b). As we deemed uniform axial resolution to be absolutely necessary to achieve the highest possible data quality, we integrated axially scanned light-sheet microscopy (ASLM)[14] in the mesoSPIM (Fig. 1b, see Supplementary Note 3, and Supplementary Videos 3-5). In our ASLM implementation, we shift the excitation beam waist through the sample using an electrically tuneable lens (ETL) and synchronize this motion with the rolling shutter of the sCMOS camera. Therefore, only the axially most confined region of the light-sheet contributes to image formation comparable to earlier approaches using mechanical translation of the sample[15]. In ASLM mode the mesoSPIM achieves an axial resolution of 5.57 ± 0.03 μm (FWHM, n = 2170 beads, nD=1.45) across a 3.29-mm FOV and 6.52 ± 0.07 μm (FWHM, n = 322 beads, nD=1.45) across a FOV of 13.29 mm (Supplementary Note 4, Supplementary Figure 4). These features enable us to image a whole mouse brain (≈ 1 cm3) with isotropic sampling (6.5 μm) within 7-8 minutes resulting in a relatively small dataset (12-16 GB). The microscope software (mesoSPIM-control, Supplementary Software) is written in Python and allows users to specify sequences of z-stacks using a table-based acquisition manager (Supplementary Note 5, Supplementary Fig. 5, Supplementary Video 6). The software can also be used to acquire large-scale tiling acquisitions, for example to visualize fine neurites in the developing nervous system of a 7-day old chicken embryo resulting in a 880 GB dataset (Figure 2, Supplementary Video 7). To achieve optimum optical sectioning in ASLM, the amplitude and offset of the ETL waveform need to be adapted when changing the excitation wavelength, zoom, or the immersion medium and can even depend on the local refractive properties of the sample. Therefore, mesoSPIM-control allows users to select configuration files with default ASLM settings for different immersion media and to manually optimize ASLM parameters (Supplementary Video 8).
Figure 2

Large scale-dataset acquired with the mesoSPIM:

Depth-coded original XY (left) and resliced YZ (right) projections of a dataset taken from a 7-day old chicken embryo (neurofilament labeling) cleared using BABB. Throughout the dataset (acquired at 1.6×1.6×2 μm3 sampling), neurites are visible in great detail. Because of the ASLM mode the same holds true for the original (transverse) and the resliced (axial) direction. The assignment of color to Z-position is similar for both the XY and YZ view. Scale bars of all insets: 200 μm. The imaging experiment was conducted once.

With a travel range of 44 × 44 × 100 mm, large samples such as a whole mouse CNS can be imaged in their entirety (Fig. 1c, Supplementary Video 9). After acquiring overview datasets, users can zoom in and record multidimensional data at higher resolution by mosaic acquisitions, for example revealing cellular distribution and long-range axonal projections of Purkinje cells in the mouse cerebellum (Fig. 1d-e, Supplementary Videos 10-12). We tested the instrument in combination with all major clearing techniques (Supplementary Note 6) including active[7] and passive[11] CLARITY (Fig 1b-e, Supplementary Fig. 6 and 7, Supplementary Video 13) and CUBIC-X[9] (Supplementary Fig. 8). Among organic solvent methods, we tested iDISCO[4] (Supplementary Fig. 9, Supplementary Video 14) and BABB[2] (Fig. 2, Supplementary Fig. 10). To demonstrate multi-view acquisitions with the mesoSPIM, we imaged a BABB-cleared chicken embryo from multiple directions (Supplementary Fig. 11) and fused the resulting datasets using BigStitcher[16] (Supplementary Fig. 12). Given its flexible sample holders, the mesoSPIM is compatible with a wide range of sample types ranging from Drosophila melanogaster (Supplementary Fig. 13, Supplementary Video 15) to cleared human cortex processed using MASH[17] (Supplementary Fig. 14, Supplementary Videos 16 and 17). Inspired by the openSPIM[18] and openSPIN[19] projects, the mesoSPIM hardware documentation and software are freely available. Depending on the configuration, a mesoSPIM requires a budget of $170000-$240000 (Supplementary Table 2) and can be installed in a day if all parts are available (Supplementary Video 18). Currently, five mesoSPIM setups are in operation across Europe and several more instruments are under construction. The mesoSPIM is the ideal instrument to quickly bridge scales from the μm- to the cm-level which enables it to serve as an excellent tool for detailed three-dimensional anatomical investigations in neuroscience and developmental biology. We have designed the mesoSPIM as a versatile and modular imaging platform and expect that it will be extended towards even larger samples, combined with novel clearing methods, and integrated with other imaging modalities such as optical projection tomography[20].
  46 in total

1.  DEVILS: a tool for the visualization of large datasets with a high dynamic range.

Authors:  Romain Guiet; Olivier Burri; Nicolas Chiaruttini; Olivier Hagens; Arne Seitz
Journal:  F1000Res       Date:  2020-11-27

Review 2.  Tutorial: practical considerations for tissue clearing and imaging.

Authors:  Kurt R Weiss; Fabian F Voigt; Douglas P Shepherd; Jan Huisken
Journal:  Nat Protoc       Date:  2021-05-21       Impact factor: 13.491

3.  Performance tradeoffs for single- and dual-objective open-top light-sheet microscope designs: a simulation-based analysis.

Authors:  Kevin W Bishop; Adam K Glaser; Jonathan T C Liu
Journal:  Biomed Opt Express       Date:  2020-07-24       Impact factor: 3.732

4.  Value-guided remapping of sensory cortex by lateral orbitofrontal cortex.

Authors:  Giuseppe Parente; Jasper Teutsch; Abhishek Banerjee; Christopher Lewis; Fabian F Voigt; Fritjof Helmchen
Journal:  Nature       Date:  2020-09-03       Impact factor: 49.962

Review 5.  Whole-Brain Profiling of Cells and Circuits in Mammals by Tissue Clearing and Light-Sheet Microscopy.

Authors:  Hiroki R Ueda; Hans-Ulrich Dodt; Pavel Osten; Michael N Economo; Jayaram Chandrashekar; Philipp J Keller
Journal:  Neuron       Date:  2020-05-06       Impact factor: 17.173

6.  Immune regulation of islet homeostasis and adaptation.

Authors:  Jinglong Guo; Wenxian Fu
Journal:  J Mol Cell Biol       Date:  2020-10-01       Impact factor: 6.216

Review 7.  Harnessing non-destructive 3D pathology.

Authors:  Jonathan T C Liu; Adam K Glaser; Kaustav Bera; Lawrence D True; Nicholas P Reder; Kevin W Eliceiri; Anant Madabhushi
Journal:  Nat Biomed Eng       Date:  2021-02-15       Impact factor: 25.671

8.  A versatile oblique plane microscope for large-scale and high-resolution imaging of subcellular dynamics.

Authors:  Etai Sapoznik; Bo-Jui Chang; Jaewon Huh; Robert J Ju; Evgenia V Azarova; Theresa Pohlkamp; Erik S Welf; David Broadbent; Alexandre F Carisey; Samantha J Stehbens; Kyung-Min Lee; Arnaldo Marín; Ariella B Hanker; Jens C Schmidt; Carlos L Arteaga; Bin Yang; Yoshihiko Kobayashi; Purushothama Rao Tata; Rory Kruithoff; Konstantin Doubrovinski; Douglas P Shepherd; Alfred Millett-Sikking; Andrew G York; Kevin M Dean; Reto P Fiolka
Journal:  Elife       Date:  2020-11-12       Impact factor: 8.140

9.  Rapid methods for the evaluation of fluorescent reporters in tissue clearing and the segmentation of large vascular structures.

Authors:  Nils Kirschnick; Dominik Drees; Esther Redder; Raghu Erapaneedi; Abel Pereira da Graca; Michael Schäfers; Xiaoyi Jiang; Friedemann Kiefer
Journal:  iScience       Date:  2021-05-26

10.  A deep learning algorithm for 3D cell detection in whole mouse brain image datasets.

Authors:  Adam L Tyson; Charly V Rousseau; Christian J Niedworok; Sepiedeh Keshavarzi; Chryssanthi Tsitoura; Lee Cossell; Molly Strom; Troy W Margrie
Journal:  PLoS Comput Biol       Date:  2021-05-28       Impact factor: 4.475

View more

北京卡尤迪生物科技股份有限公司 © 2022-2023.