|Home | About | Journals | Submit | Contact Us | Français|
We report the development of a modular and optimized thin-sheet laser imaging microscope (TSLIM) for nondestructive optical sectioning of organisms and thick tissues such as the mouse cochlea, zebrafish brain/inner ear, and rat brain at a resolution that is comparable to wide-field fluorescence microscopy. TSLIM optically sections tissue using a thin sheet of light by inducing a plane of fluorescence in transparent or fixed and cleared tissues. Moving the specimen through the thinnest portion of the light sheet and stitching these image columns together results in optimal resolution and focus across the width of a large specimen. Dual light sheets and aberration-corrected objectives provide uniform section illumination and reduce absorption artifacts that are common in light-sheet microscopy. Construction details are provided for duplication of a TSLIM device by other investigators in order to encourage further use and development of this important technology.
The ability to selectively visualize threedimensional (3-D) biological structures is useful for understanding structure, function, and dysfunction relationships in tissues and organs, and is one of the initiatives of the National Institutes of Health (NIH) Roadmap for Medical Research. Three-dimensional reconstruction of anatomical structures requires that tissues be sectioned (either mechanically or nondestructively) and structures segmented for 3-D rendering. Nondestructive optical sectioning produces well-aligned sections and can be performed, at high resolution using confocal and multiphoton microscopy, but imaging depth is limited to a few hundred microns. For thick specimens, micro MRI or micro CT can be used, but at lower resolution and without the advantages offered by immunofluorescence. Light sheet-based microscopy is an important, innovative tool that offers nondestructive optical sectioning of selectively stained thick tissues at a spatial resolution between that of micro MRI and confocal microscopy. Development of this technology has been hindered by the lack of a commercial device, although several investigators have constructed light sheet-based microscopes.
Optical sectioning using a plane of light was described as early as 1903 by Siedentopf and Zsigmondy (1), but given the unfortunate name of ultramicroscopy, which now refers to electron microscopy. The basis for light-sheet optical sectioning and the materials for the construction of a device are quite simple. A thin light sheet is produced using a cylindrical lens and is projected through either a transparent or fixed and cleared specimen to illuminate a thin plane (i.e., optical section) within the tissue. The optical section is observed orthogonal to the light sheet, and by moving the specimen through the thin plane of light, a z stack of serial sections is produced. Pioneering work on a light sheet microscope was done by Voie et al. (2-4) and called orthogonal-plane fluorescence optical sectioning (OPFOS). Their device collected real-time, 2-D optical sections in cleared tissues using a light sheet that was produced using a single laser, beam expander, and cylindrical lens. Fuchs et al. (5) constructed a similar device, called thin-light sheet microscopy (TLSM) to optically section specimens in seawater. Another device was developed by Huisken et al. (6) called selective plane illumination microscopy (SPIM). This method rotates an agaroseembedded specimen through a single laser light sheet, but requires complex algorithms to produce a z stack of optical sections. Dodt et al. (7) called their method ultramicroscopy and added dual light sheet illumination for the optical sectioning of larger specimens such as the mouse brain.
The quality of the optical sections produced by all of these devices is dependent upon the optical geometry of the light sheet. A light sheet produced by a cylindrical lens has a Gaussian intensity profile and reaches a minimal thickness (called the beam waist) in close proximity to the focal plane of the lens (see Supplementary Materials and Supplementary Figure 1 for details). The beam waist has a relatively constant thickness over a region called the confocal parameter, which is also twice the Raleigh range. For small specimens whose dimensions do not exceed the confocal parameter, the optical section appears to have nearly uniform resolution across the width of the specimen. However, for large specimens, optimized resolution and focus is present only in the tissue region illuminated within the confocal parameter of the light sheet. For large specimens and full-frame imaging, one can choose lenses that result in a larger confocal parameter, but this configuration also produces thicker light sheets and lower image resolution. In order to achieve high-resolution across the full width of large specimens, Buytaert and Dirckx (8) moved a specimen through the beam waist and stitched image columns together to produce a high-resolution, well-focused composite image. However, a light sheet-based device has not yet been developed which incorporates the best features of the previous devices (2-8). This study reports on the development of such a device, which is modular and optimized, and called a thin-sheet laser imaging microscope (TSLIM). In addition, we provide details for the construction of a TSLIM device by other investigators to encourage its further use and development (see Supplementary Materials).
Adult CBA/BL6 mice and 5-day-old rat pups were euthanized by CO2 inhalation, decapitated, and mouse cochleas and rat brains were removed and fixed by immersion in 4% paraformaldehyde/1% glutaraldehyde for 24 h. Paraformaldehyde-fixed Casper mutant zebrafish (Danio rerio) 6-8 weeks old were shipped to the University of Minnesota for TSLIM imaging. All specimens (except rat brains) were decalcified in a 10% solution of disodium ethylene diaminetetraacetic acid for 3 d and bleached in a 5% solution of H2O2 for 24 h. Mouse cochleas were separated from surrounding tissues and rat brains were hemisected and cut into thirds. The anterior 5 mm of the zebrafish head containing the brain and inner ears were used for imaging. All tissues were dehydrated in ascending concentrations of ethanol, immersed in hexane, and then cleared to transparency using Spalteholz fluid (9) which consists of 5:3 methyl salicylate:benzyl benzoate. Rat brains were cleared in 2:1 benzyl benzoate:benzyl alcohol (8) as this solution appeared to clear brain tissue better than Spalteholz fluid. The refractive index of the cleared specimens and the fluid-filled specimen chamber was ~1.56. Tissue fluorescence, which is necessary for TSLIM imaging, was induced either by chemical fixation autofluorescence (paraformaldehyde/glutaraldehyde), or by immersion in Rhodamine B isothiocyanate (1 mg/200 mL in Spalteholz for 24 h). A 590-nm bandpass filter was used to block scattered laser light from entering the CCD camera which was used to capture images.
Design and specifications of the TSLIM device are shown in computer-aided design (CAD) diagrams (Figure 1, Supplementary Figure 2). TSLIM consists of five primary components: two thin-sheet illuminators with aberration corrected objectives, a specimen chamber, a microscope with digital camera, motorized micropositioners and rotating stage, and control and imaging software. A complete list of materials used for the TSLIM device can be found in Supplementary Table 1 and a parts diagram is outlined in Supplementary Figure 2. Assembly and alignment procedures are also available in the Supplementary Materials. TSLIM contains two opposing laser illuminators that are mounted on a horizontal optical bench rail, which project their light sheets into the specimen chamber. Each illuminator consists of a 15 mW, green (λ = 532 nm) frequency-doubled Nd:YAG laser, a 10× or 20× Galilean beam expander, a cylindrical lens, and a 5× microscope objective. A 532-nm solid-state laser was selected because it excites and causes emission of a wide variety of fluorescent markers that are used for biological research. The laser beam is expanded and collimated using a Galilean beam expander, and then travels through a cylindrical lens, which focuses the beam in the y direction. The cylindrical lens and microscope objective assemble a Keplerian beam expander, which means that the beam leaving the microscope objective is collimated in the y direction. As the cylindrical lens does not affect the z component of the beam, the microscope objective has a focusing effect on the beam that results in a diffraction-limited light-sheet thickness in the z direction. The improvement in image quality by the addition of an objective lens is shown in Supplementary Figure 3; this was first used by Greger et al. (10). The light sheet then passes through the specimen chamber, which is positioned orthogonal to the optical axis of a horizontally mounted, Olympus MVX10 microscope (Olympus America, Inc., Center Valley, PA, USA). A glass cuvette or a custom-designed specimen chamber with an open top is filled with clearing fluid and the specimen is attached to a black Delrin rod (Small Parts Inc., Miramar, FL, USA) that extends into the middle of the chamber. The specimen attaching rod is connected to an optional, motorized rotating stage for convenient rotation/orientation of the specimen. The light sheet enters and leaves the chamber through the side windows and the fluorescent image plane in the tissue is viewed through the back window of the chamber nearest the MVX10 objective. Micropositioners (Newport Corp., Irvine, CA, USA) move the specimen (not the chamber) in the x,y,z directions (QImaging, Surrey, BC, Canada) through the illumination plane and at the focal point of the microscope objective. A custom LabVIEW program (version 8.6; National Instruments, Austin, TX, USA) was used to control the micropositioners and collect images using a Retiga 2000 (1600 × 1200 px) digital camera attached to the MVX10 microscope. Micropositioner control, image stitching and stack collection were automated and run on a Windows XP-based PC (See program flowchart in Supplementary Figure 4). The program controlled x-axis micropositioner movement while building a composite image from columns collected at each x-axis step. Column width was chosen to coincide with the confocal parameter of the light sheet and supplied to the CCD camera as a region of interest (ROI). After saving each composite image, the z axis was incremented and the next optical section was generated. Images were processed in Adobe Photoshop (Adobe CS3; Adobe Systems Incorporated, San Jose, CA, USA) and ImageJ (version 1.41; National Institutes of Health, Bethesda, MD, USA). After processing, stacks were loaded into Amira software (Visage Imaging Inc., Carlsbad, CA, USA) for reconstruction of individual tissue structures. See Supplementary Materials for information regarding obtaining a copy of our custom LabVIEW program or TSLIM community resources.
Tissues from the mouse, zebrafish, and rat were used to illustrate some of the capabilities of TSLIM. However, TSLIM can be used for optical sectioning of many different types of tissues and organisms. Since TSLIM is modular, it can be configured with different lasers, beam expanders, lenses, and specimen chambers for different types of specimens. For high resolution of small structures, a thin beam waist is required and TSLIM could be configured for single-beam, full-frame imaging. Exposure times varied depending upon tissue f luorescence, specimen thickness, the number of optical sections desired, and whether single-or dual-beam illumination was used. A typical length of time to obtain a single, full-frame optical section using dual-beam illumination was ~1 s. However, for high-resolution structures in a large specimen, a full-frame image would appear focused only within the confocal region of the light sheet. This is shown in Figure 2 using the mouse cochlea and zebrafish head. In Figure 2A and 2C the arrow indicates the approximate position of the beam waist and in that region the specimen appears well focused. To the left and right of the beam waist, the optical section appears out of focus due to increasing thickness of the light sheet away from the focal point of the lens. To provide optimal resolution and focus across the full width of a specimen, the specimen was moved across the beam waist of the light sheet and image columns (the size of the confocal region) were obtained and stitched together to form a well focused, composite image (Figure 2, B and D). It took 46 s and 74 s to produce the composite image in Figure 2, B and D, respectively. In addition, horizontal lines are noticeable in the left portion of Figure 2A: these are produced by the uneven absorption of light by certain tissue structures as the light passes through the specimen. These absorption artifacts, which are common in light-sheet microscopy, are minimized by dual-beam illumination in TSLIM (7,11). Huisken et al. (11) also used beam oscillation to reduce absorption lines, but their method required complex equipment and software.
In some specimens that were repeatedly imaged over a period of several months, or exposed to the beam for a long period of time, we noticed a loss in fluorochrome emission (i.e., photobleaching). This loss was much less than observed in wide-field fluorescence microscopy. Photobleaching in light-sheet microscopy is minimized since only a thin portion of the tissue (i.e., the plane illuminated by the light sheet) is exposed during optical sectioning. In addition, tissue clearing, tissue staining, and the use of a sensitive digital camera requires relatively low-watt light sources (15 mW each) and short exposure times. A preliminary experiment was performed to quantify photobleaching by TSLIM for full-frame and column stitching imaging. In full-frame imaging, tissue was exposed to the light sheet for ~1 s or ~48 s to produce a composite image by stitching. Pixel intensity decreases were measured in optical sections that were obtained by full-frame versus column stitching. As expected, photobleaching was less after a 1-s full-frame image (pixel intensity decrease = 0.85%) compared with a 48-s stitched image (pixel intensity decrease = 8.85%). However, in dye-stained tissue, fluorochrome emission could be restored by restaining the tissue by immersion in Rhodamine B isothiocyanate.
High-magnification TSLIM imaging resolves many cellular details within the cochlea. Figure 3A is a 3-D perspective image (using Amira) from a stack of 134 serial TSLIM sections through the mouse cochlea. The last 2-D section of the stack and virtual, oblique sections through the organ of Corti are shown in this figure. Three rows of the outer hair cells and the smaller, single row of pillar cells are clearly visible, as well as a number of other important cochlear structures (e.g., the stria vascularis, tectorial and Reissner's membranes, and spiral ganglion neurons). A movie (Supplemental Movie 1) of this stack is available in the Supplementary Materials. Figure 3B shows an image from a stack of optical sections through the spiral ganglion of another mouse cochlea. The axons, cell bodies, and nuclei are clearly visible in this figure and in Supplemental Movie 2, showing 82 1-μm sections of the spiral ganglion (see Supplementary Materials).
TSLIM was also used to image the zebrafish head. Figure 3C shows imaging of the head of the Casper mutant (12) and 3-D reconstruction of the inner ear using Amira. The eyes, which contain pigment, were opaque, but other structures such as the brain and membranous labyrinth of the inner ear were clearly resolved in 3-D renderings by Amira, including the semicircular canals, saccule, utricle, lagena, and branches of the VIII nerve. High magnification of the brain resolved many structures such as cell bodies, blood vessels, ventricles and nerve fiber pathways that are traceable in a stack of well-aligned serial sections. Supplemental Movie 3 shows 100 20-μm serial sections through a zebrafish head. TSLIM imaging was also used to optically section the rat brain. Figure 3D is a cross-section through the rat brain which clearly shows the hippocampus.
Like other light-sheet based devices (2-8,10-11) TSLIM is a powerful imaging technology that permits nondestructive, high-resolution, rapid, and efficient optical sectioning of thick tissues, whole organs, and even small organisms. TSLIM's primary advantage over previous systems is that it is modular, optimized, and incorporates the best features from other designs. It uses off-the-shelf components and can be constructed for a reasonable cost of ~$22,000 for a dual-beam system (not including the Olympus microscope and Retiga camera). TSLIM imaging is well-suited for optically sectioning large, fixed, and cleared specimens, and produces a stack of well-aligned images that can be volume-rendered or segmented for isosurface 3-D reconstruction of individual structures. It may also be used for imaging live tissue and organisms in an aqueous solution, provided that they are transparent and contain a fluoro-chrome. The construction of TSLIM by other investigators will serve to establish the capabilities and limitations of light-sheet based optical sectioning in a variety of different specimens.
This work was supported by the Capita Foundation, the Digital Technology Center, and the Supercomputing Institute at the University of Minnesota. It was also supported in part by the National Institute on Deafness and Other Communication Disorders (grant no. RO1 DC007588, to P.S.). We thank Dave Hultman for the construction of custom parts, Richard M. White for generously sending us paraformaldehyde-fixed zebrafish, and Marilyn Carroll and Justin Anker for providing the rat brains. All authors made significant contributions to this research and the writing of the manuscript. S.J., P.S., and T.G. processed the tissues and prepared the images. S.J. and P.G. wrote the LabVIEW program for specimen movement and image collection. M.H. prepared the CAD design of TSLIM and developed a method to test and align the light sheets. P.S. wrote the manuscript, and J.L. provided expertise on optics selection and measurement. This paper is subject to the NIH Public Access Policy.
Supplementary material for this article is available at www.BioTechniques.com/article/113087.
The authors declare no competing interests.