Chapter 17 - Bridging Microscopes: 3D Correlative Light and Scanning Electron Microscopy of Complex Biological Structures

https://doi.org/10.1016/B978-0-12-416026-2.00017-0Get rights and content

Abstract

The rationale of correlative light and electron microscopy (CLEM) is to collect data on different information levels––ideally from an identical area on the same sample––with the aim of combining datasets at different levels of resolution to achieve a more holistic view of the hierarchical structural organization of cells and tissues. Modern three-dimensional (3D) imaging techniques in light and electron microscopy opened up new possibilities to expand morphological studies into the third dimension at the nanometer scale and over various volume dimensions. Here, we present two alternative approaches to correlate 3D light microscopy (LM) data with scanning electron microscopy (SEM) volume data. An adapted sample preparation method based on high-pressure freezing for structure preservation, followed by freeze-substitution for multimodal en-bloc imaging or serial-section imaging is described. The advantages and potential applications are exemplarily shown on various biological samples, such as cells, individual organisms, human tissue, as well as plant tissue. The two CLEM approaches presented here are per se not mutually exclusive, but have their distinct advantages. Confocal laser scanning microscopy (CLSM) and focused ion beam-SEM (FIB-SEM) is most suitable for targeted 3D correlation of small volumes, whereas serial-section LM and SEM imaging has its strength in large-area or -volume screening and correlation. The second method can be combined with immunocytochemical methods. Both methods, however, have the potential to extract statistically relevant data of structural details for systems biology.

Introduction

Correlative light and electron microscopy (CLEM) has become a powerful tool in life science (Jahn et al., 2012). The rationale is to collect data on different information levels from an identical area of one sample and combine these to achieve a holistic understanding of the ultrastructure of living systems. Three-dimensional (3D) imaging techniques in light and electron microscopy (EM) have opened up new possibilities to expand morphological context description into the third dimension at nanometer scale. This chapter focuses on two methods for correlative 3D imaging, combining confocal laser scanning microscopy (CLSM) with focused ion beam-scanning electron microscopy (FIB-SEM), or wide-field light microscopy (LM) and SEM using serial sections of embedded samples. The two approaches are schematically outlined in Fig. 1. A specially adapted sample preparation method based on high-pressure freezing (HPF) for optimum ultrastructural preservation is presented as connecting link between the afore-mentioned imaging techniques.

Since its introduction in the 1960s (Moor, 1973; Riehle, 1968; Riehle & Hoechli, 1973), high-pressure freezing has proven in numerous applications and among various kinds of samples from the kingdoms of life (Dahl & Staehelin, 1989; Hess, 2007; Hohenberg, Mannweiler, & Muller, 1994; Hohenberg, Tobler, & Muller, 1996; McDonald & Morphew, 1993; Müller, 1992; Studer, Hennecke, & Muller, 1992; Studer, Michel, & Muller, 1989; Studer, Michel, Wohlwend, Hunziker, & Buschmann, 1995; Szczesny, Walther, & Muller,, 1996) to preserve biological tissue better and closer to its lifelike state than any other sample preparation method for EM (Murk et al., 2003; Schwarz & Humbel, 2007; Tiedemann, Hohenberg, & Kollmann, 1998). The pinpoint precision of preserving the functional locus of biomolecules has been demonstrated for human skin and other tissues (Kirschning, Rutter, & Hohenberg, 1998; Pfeiffer et al., 2000; Ripper, Schwarz, & Stierhof, 2008; Schwarz, Hohenberg, & Humbel, 1993; Vielhaber, Brade, et al., 2001; Vielhaber, Pfeiffer, et al., 2001) and is summarized in Fig. 2. Human skin biopsies were prepared according to different fixation protocols and immediately immobilized by HPF to preserve the ultrastructural state of the tissue. These frozen samples were then freeze-substituted and embedded in parallel and with identical preparation steps as reported in Pfeiffer et al. (2000) and Biel, Kawaschinski, Wittern, Hintze, and Wepf (2003), and compared to the method of progressive lowering of temperature, or PLT embedding (Robertson, Monaghan, Clarke, & Atherton, 1992).

One sample (Fig. 2(A)) was chemically fixed with Karnovsky’s fixative (1.25% glutaraldehyde and 0.2% formaldehyde in cacodylate buffer) prior to HPF. Paraformaldehyde or formaldehyde is considered a faster but weaker penetrating fixative compared to glutaraldehyde (Elias & Friend, 1975; Karnovsky, 1965). The second sample (Fig. 2(B)) was chemically fixed as mentioned above at 4°C, but was then dehydrated and embedded in HM20 using the PLT method. The third sample (Fig. 2(C)) was high-pressure frozen immediately (i.e., less than 30 sec) after extraction of the punch biopsy and freeze-substituted as reported previously (Biel et al., 2003; Pfeiffer et al., 2000). Ultrathin sections of the embedded tissue were labeled against glucosylated precursor lipids (mouse Glucosylceramide antiserum; Brade, Vielhaber, Heinz, & Brade, 2000) which are very specific to the last viable cell layers of the epidermis and finally are exported into the intercellular space, where they are deglucosylated by beta-glucocerebrosidase (Holleran et al., 1994) and not recognized anymore by the mouse-anti glucosylceramide antibody. Therefore, the localization of this antigen is functionally defined in lamellar bodies, which are transported from the trans-Golgi network in the upper stratum spinosum and stratum granulosum toward the skin surface. The antigens are then secreted into the intercellular space at the “life–death transition interface” between stratum granulosum and the first dead layers of the stratum corneum.

Fig 2 shows clearly that this lipid precursor is delocalized and found in different skin layers even down to the dermis (Fig. 2 (A and B)) after chemical fixation, independent of the follow-up dehydration protocol. The structure–function related localization of such small biomolecules (<1000 MW) is only preserved after HPF and gentle, optimized freeze-substitution (FS) as reported here. Our methods take full advantage of localization techniques, such as immunolocalization for LM and EM or even super-resolution imaging in the near future, and thereby enable bridging between these imaging techniques.

Morphological investigations in life science rely on the description of lipid membranes delineating intracellular compartments, i.e. organelles or vesicles, as well as cell borders and intercellular connections. These membranes are only 4.5–5 nm thick and thus invisible to LM, but resolvable by EM. FIB-SEM has proven to be a valuable application for ultrastructural investigations in life science (Young, Dingle, Robinson, & Pugh, 1993). It offers the combination of precise and site-specific milling of material from a bulk sample, using a focused ion beam, with high-resolution SEM imaging. Using the FIB as a serial-sectioning tool to produce stacks of consecutive image planes allows automated recording of large volumes (Knott, Marchman,Wall, & Lich, 2008; Lucas et al., 2008). Routine overnight FIB-SEM cross-section imaging can extend up to 50 × 40 µm2. With a slice thickness of 5–10 nm, these stacks can expand up to several 10 µm in the x-axis. This makes FIB-SEM a valid alternative to standard serial-section transmission electron microscopy (TEM) for targeted volumes, as the preparative effort is reduced and a better z-resolution compared to serial sections can be achieved.

Modern field emission SEMs allow even at low acceleration voltages of 1 kV to resolve structural details of contrasted or unstained biological samples in the range of 1–2 nm, using eitherback-scattered electron (BSE) or secondary electron (SE) imaging. Therefore, on-section SEM imaging has the lateral resolution power to image lipid membranes in cells and tissues. Necessary adaptations for sample preparation and imaging parameter are reported in this chapter. Hence, SEM imaging of either ultrathin sections or block-faces may be an alternative approach for ultrastructure investigations, which are to date dominated by TEM.

Dyes and specific labels for LM often are incompatible or only partially compatible with heavy metal salts used for EM. Therefore, specimen preparation for 3D CLEM requires a compromise, which ideally results in a sample suitable for all light and electron microscopic approaches described above. En-bloc imaging in CLSM and FIB-SEM poses some challenges: (1) the region of interest (ROI) has to be accessible for imaging as well as for FIB-milling, i.e. near the surface, (2) the material should facilitate stable and artifact-free milling conditions in FIB-SEM, and most important (3) fluorescence imaging should be enabled, while optimum signal-to-noise ratio is provided for BSE imaging. Resin embedding meets these requirements inasmuch as first, the ROI can easily be uncovered by trimming or ultramicrotomy as is routinely done for TEM samples; and second, the material is sufficiently homogenous in terms of hardness and density to enable artifact-free FIB-milling (Knott et al., 2008). Fluorescence labeling during FS followed by embedding in HM20 proved a powerful tool for correlative microscopy (Biel et al., 2003). It enables en-bloc CLSM investigation of the sample prior to preparation of ultrathin sections for TEM.

Obtaining a good signal-to-noise ratio in FIB-SEM, especially with respect to the visibility of lipid membranes, is more challenging, as the samples need to be investigated en-bloc and conventional post-staining of ultrathin sections is not possible. Several protocols have been described, aiming to optimize preembedding membrane staining after chemical fixation, based on potassium ferrocyanide-mediated oxidation of osmium tetroxide, followed by uranyl acetate (De Winter et al., 2009; Knott et al., 2008), or tannic acid (Armer et al., 2009).

For ultrastructural investigations of very fast cellular processes, e.g. virus entry into host cells, HPF is preferable to chemical fixation, as it is faster and avoids fixative-induced artifacts such as shrinkage or membrane perturbation (Griffiths, 1993, Droste et al., 2005; Murk et al., 2003). Enhancing membrane staining during FS after HPF for TEM applications has been discussed intensely. Pfeiffer et al. (2000) and Humbel and Schwarz (1989) showed that FS in acetone with uranyl acetate preserves the membrane details and lipids well. FS in acetone with uranyl acetate and glutaraldehyde was also described as the method of choice by Giddings (2003), when compared to osmium tetroxide and tannic acid in acetone. Hess (2003) on the other hand, concluded that osmium tetroxide in acetone, with optional addition of uranyl acetate, was the best choice for morphological investigations. Tannic acid-mediated osmium impregnation at room temperature following FS in osmium in acetone provided good membrane contrast even in semithin sections used for TEM tomography, and rendered post-staining of the sections unnecessary (Jimenez et al., 2009). Furthermore, addition of water to the FS-medium has been described to enhance membrane visibility (Buser & Walther, 2008; Walther & Ziegler, 2002), which also significantly improved the mordant-mediated osmium staining described by Jimenez et al. (2009). Nevertheless, most of these methods either involve additional post-staining of ultrathin sections or are incompatible with fluorescence labeling.

Array tomography, on the other hand, has different requirements in terms of specimen preparation. As this method employs serial sections of resin-embedded material, labels do not necessarily have to be introduced before embedding. This leaves a margin for adaptations according to the requirements of the respective experiment, and opens up more possibilities for specific labeling either by histological dyes, fluorescent dyes (e.g., DAPI), or by immunocytochemical techniques to identify an ROI. If intense application of heavy metal salts interferes with fluorescence labeling, the sections can be post-stained at a later stage for SEM investigation. The protocol described in this chapter provides a compromise, which allows characterization of the embedded specimen by CLSM prior to either FIB-SEM investigation or serial sectioning for array tomography.

Section snippets

Rationale

The purpose of this chapter is to present and discuss volume imaging techniques for correlative light and scanning electron microscopy, including a protocol for FS adapted for multimodal en-bloc imaging. Fluorescent labeling during FS, prior to embedding, provides histological context information of the ROI and can be employed to relocate the ROI in SEM and FIB-SEM. As an alternative to the en-bloc CLEM approach, serial-section SEM and array tomography are described, and assets and drawbacks of

Methods

The described protocol for FS and the different CLEM approaches are applicable to a wide variety of samples, such as cultured cells, small organisms, tissue samples, or plant specimens. To demonstrate this, we used monolayers of MDCK and HeLa cells, nematodes (C. elegans), biopsies of human skin, and mung bean (Vigna radiata) root nodules. The specimen preparation for these samples is described in the following sections. Pretreatment or culture conditions are omitted. These may vary

High-Pressure Freezing

Instrumentation: We use two models of high-pressure freezers: the Bal-Tec HPM 010 (Bal-Tec AG, Principality of Liechtenstein) and the Leica EM HPM 100 (Leica Microsystems, Austria). Additional instrumentation used for the described experiments include a carbon coater for coating of sapphire discs (BAE 120, Bal-Tec AG, Principality of Liechtenstein), a hand microtome (Windaus Labortechnik GmbH, Germany), and a prevacuum chamber for degassing of samples before freezing, if needed.

Materials:

Discussion

The two 3D-CLEM approaches presented here are not per se mutually exclusive but have their distinct advantages. CLSM combined with FIB-SEM is most suitable for targeted 3D correlation of small volumes, whereas serial-section LM and SEM (i.e., CAT) imaging has its strength in the large field of view and volume screening, and correlation and can be combined with immunocytochemical methods. Both methods have the potential to extract statistically relevant data of structural details, if the

Summary

Fluorescent labeling during FS not only facilitates the task of relocating an ROI, but additionally provides multimodal information of the same ROI in 3D. CLSM investigation prior to EM supplies valuable information on the structural context of the ROI. Such data can be combined with high-resolution imaging of the identical ROI by either FIB-SEM or serial-section SEM. Both approaches, CLSM combined with FIB-SEM, as well as CAT combined with prior in-vivo LM investigation, have the potential to

Acknowledgments

The authors would like to thank all collaborators for providing samples and fruitful discussions: Prof. Hans-Martin Fischer (Institute of Microbiology, ETH Zurich) provided the mung bean root nodules. Dr. Stefanie Krämer (Institute of Pharmaceutical Sciences, ETH Zurich) contributed the MDCK cells and the glucosylceramide antiserum. HeLa cells were obtained from Dr. Jason Mercer (Institute of Biochemistry, ETH Zurich). The nematodes were kindly provided by Dr. Christel Genoud (Friedrich

References (66)

  • G. Vielhaber et al.

    Localization of ceramide and glucosylceramide in human epidermis by immunogold electron microscopy

    Journal of Investigative Dermatology

    (2001)
  • H.E. J. Armer et al.

    Imaging transient blood vessel fusion events in zebrafish by correlative volume electron microscopy

    PLoS ONE

    (2009)
  • S.S. Biel et al.

    From tissue to cellular ultrastructure: closing the gap between micro- and nanostructural imaging

    Journal of Microscopy

    (2003)
  • L. Brade et al.

    In vitro characterization of anti-glucosylceramide rabbit antisera

    Glycobiology

    (2000)
  • C. Buser et al.

    Freeze-substitution: the addition of water to polar solvents enhances the retention of structure and acts at temperatures around −60 degrees C

    Journal of Microscopy

    (2008)
  • A. Cardona et al.

    An integrated micro- and macroarchitectural analysis of the Drosophila brain by computer-assisted serial section electron microscopy

    PLoS Biology

    (2010)
  • R. Dahl et al.

    High-pressure freezing for the preservation of biological structure: theory and practice

    Journal of Electron Microscopy Technique

    (1989)
  • M.S. Droste et al.

    Noninvasive measurement of cell volume changes by negative staining

    Journal of Biomedical Optics.

    (2005)
  • P.M. Elias et al.

    Permeability barrier in mammalian epidermis

    Journal of Cell Biology

    (1975)
  • E.A. Ellis

    Poststaining grids for transmission electron microscopy: conventional and alternative protocols

    Methods in Molecular Biology

    (2007)
  • M. Eppenberger-Eberhardt et al.

    Adult-rat cardiomyocytes cultured in creatine-deficient medium display large mitochondria with paracrystalline inclusions, enriched for creatine-kinase

    Journal of Cell Biology

    (1991)
  • T.H. Giddings

    Freeze-substitution protocols for improved visualization of membranes in high-pressure frozen samples

    Journal of Microscopy

    (2003)
  • B.N. Giepmans et al.

    Correlated light and electron microscopic imaging of multiple endogenous proteins using Quantum dots

    Nature Methods

    (2005)
  • M. Grabenbauer et al.

    Correlative microscopy and electron tomography of GFP through photooxidation

    Nature Methods

    (2005)
  • G. Griffiths

    Fine structure immunocytochemistry

    (1993)
  • P. Hawes et al.

    Rapid freeze-substitution preserves membranes in high-pressure frozen tissue culture cells

    Journal of Microscopy

    (2007)
  • M.W. Hess

    Of plants and other pets: practical aspects of freeze-substitution and resin embedding

    Journal of Microscopy

    (2003)
  • H. Hohenberg et al.

    High-pressure freezing of cell-suspensions in cellulose capillary tubes

    Journal of Microscopy

    (1994)
  • H. Hohenberg et al.

    High-pressure freezing of tissue obtained by fine-needle biopsy

    Journal of Microscopy

    (1996)
  • B.M. Humbel et al.

    Freeze-substitution for immunochemistry

  • M.J. Karnovsky

    A formaldehyde-glutaraldehyde fixative of high osmolality for use in electron microscopy

    Journal of Cell Biology

    (1965)
  • E. Kirschning et al.

    High-pressure freezing and freeze-substitution of native rat brain: suitability for preservation and immunoelectron microscopic localization of myelin glycolipids

    Journal of Neuroscience Research

    (1998)
  • G. Knott et al.

    Serial section scanning electron microscopy of adult brain tissue using focused ion beam milling

    Journal of Neuroscience

    (2008)
  • Cited by (62)

    • On-section correlative light and electron microscopy of large cellular volumes using STEM tomography

      2021, Methods in Cell Biology
      Citation Excerpt :

      To stain nuclei after fixation, various fluorescent dyes can be added to the freeze-substitution cocktail (Lucas, Günthert, Gasser, Lucas, & Wepf, 2012). However, careful establishment is required as fluorescent dyes behave differently depending on the tissue or cell line (Lucas et al., 2012). Nuclei can also be stained in resin sections with Hoechst dyes (Markert et al., 2016).

    • High-throughput screening of mitotic mammalian cells for electron microscopy using classic histological dyes

      2021, Methods in Cell Biology
      Citation Excerpt :

      We aimed to use histological dyes that are known to stain nuclei or DNA (Aescht et al., 2010) to search for mitotic cells. This approach is similar to previous studies, in which specific cellular compartments were stained (Biel, Kawaschinski, Wittern, Hintze, & Wepf, 2003; Lucas, Günthert, Gasser, Lucas, & Wepf, 2012, Lucas, Guenthert, Gasser, Lucas, & Wepf, 2014) for subsequent light microscopy. Here, we identified nuclei/DNA dyes that are soluble in acetone, do not interfere with heavy metals (i.e., uranium, lead or osmium) and keep their fluorescent properties after sample embedding in Epoxy resin.

    • Scanning electron microscopy as a new tool for diagnostic pathology and cell biology

      2020, Micron
      Citation Excerpt :

      However, there is the limitation of surface conductance that requires a thin layer of carbon or heavy metal coating, usually gold. New generation HRSEMs not only have better performance characteristics but many new capabilities and applications that can be usefully employed for diagnostic pathology and cell biology (Lucas et al., 2012). Automated scan generation systems are now available for most high-end systems.

    View all citing articles on Scopus
    View full text