Localization of fluorescently labeled structures in frozen-hydrated samples using integrated light electron microscopy

Abstract

Correlative light and electron microscopy is an increasingly popular technique to study complex biological systems at various levels of resolution. Fluorescence microscopy can be employed to scan large areas to localize regions of interest which are then analyzed by electron microscopy to obtain morphological and structural information from a selected field of view at nm-scale resolution. Previously, an integrated approach to room temperature correlative microscopy was described. Combined use of light and electron microscopy within one instrument greatly simplifies sample handling, avoids cumbersome experimental overheads, simplifies navigation between the two modalities, and improves the success rate of image correlation. Here, an integrated approach for correlative microscopy under cryogenic conditions is presented. Its advantages over the room temperature approach include safeguarding the native hydrated state of the biological specimen, preservation of the fluorescence signal without risk of quenching due to heavy atom stains, and reduced photo bleaching. The potential of cryo integrated light and electron microscopy is demonstrated for the detection of viable bacteria, the study of in vitro polymerized microtubules, the localization of mitochondria in mouse embryonic fibroblasts, and for a search into virus-induced intracellular membrane modifications within mammalian cells.

Introduction

The fusion of a gene of a protein of interest with the gene of a fluorescent protein has revolutionized the use of fluorescence microscopy (FM) for cell biology (Heim et al., 1995). FM allows for multi-dimensional live-cell imaging of biological systems and tissues with multiple colors at a resolution of a few hundred nanometers, or better when super-resolution methods are applied (Carlton et al., 2010, Hell, 2007, Huang et al., 2010). A large variety of markers are available to cell biologists (Giepmans et al., 2006), e.g. for the study of protein expression levels, localization, molecular dynamics and activity state. Despite the current push in further improving super-resolution techniques, fluorescence microscopy cannot unravel the structure of the protein of interest, nor its unlabeled cellular context.

Electron microscopy (EM) can reveal biological ultrastructure at nm-scale resolution, electron tomography (ET) can provide three dimensional data. Unfortunately, the superior spatial resolution can only be obtained if individual images with a limited field of view (e.g. 2 × 2 μm²) are acquired. Recently, tools for the automated acquisition of EM images over significant larger areas, such as complete cross-sections of cells, tissues or even larger regions have been described (Anderson et al., 2011, Cardona et al., 2012, Faas et al., 2012). The collection of large EM panoramas would greatly facilitate the correlation between light and electron microscopy (CLEM) and provide confidence in the relation between the location of fluorophores and their associated cellular ultrastructure. However, ideally, one would be able to seamlessly zoom in on a fluorescent area and study the underlying structures without having to collect large amounts of EM data outside the region of interest.

Over the last decades, a wealth of biological information has been obtained by imaging specimens that were (chemically) fixed, dehydrated and stained for transmission electron microscopy (TEM). A general limitation of TEM is that the specimens have to be sufficiently thin for imaging, i.e. depending on the acceleration voltage used, less than ∼0.5 μm. To access thicker biological specimens it becomes necessary to physically trim the sample. This is traditionally done by embedding the sample in a resin that, once polymerized, can be cut into sufficiently thin slices using an ultramicrotome. The employed protocols have been shown to yield excellent morphologies for a wide range of samples.

Unfortunately, these conventional protocols quench the fluorescence signal of most fluorophores. The combination of dehydration and metal staining can be detrimental for fluorescence imaging. Commonly used fixatives such as aldehydes show some fluorescence themselves, which further masks the signal of interest. Therefore, the most popular approach to CLEM is to perform fluorescence imaging prior to EM sample preparation, which abates the success rate for finding the correlations. Newer protocols have been developed to mitigate such problems (Karreman et al., 2012, Kukulski et al., 2011, Nixon et al., 2009, Watanabe et al., 2011), and are based on cryo-immobilization of the specimen, embedding in a hydrophilic resin, and the use of minimal amounts of heavy atom stains. These pioneering protocols show that it is possible to preserve significant fluorescence with only mild compromises on sample morphology and contrast in the electron microscope. Nevertheless, though extremely powerful to study the architecture of tissue and cells, resin embedding and metal staining jeopardizes the high resolution imaging of macromolecular structures.

Since the early 1980’s, alternative and complementing methods have evolved in parallel to the resin embedding and metal staining protocols. These alternative techniques are based on cryo immobilization of the samples (vitrification). All subsequent steps, including imaging, are done at cryogenic temperatures (Chang et al., 1981, Dubochet, 2012). A key characteristic of cryo electron microscopy (cryo-EM) is that macromolecular structures remain hydrated and relatively unperturbed, thus preserving the high-resolution information. Concomitantly, the fluorescence remains preserved as well. Depending on the specimen thickness, specimens are cryo immobilized by plunge-freezing or high-pressure freezing (Dubochet et al., 1988, Cardona et al., 2012, Studer et al., 2001, Studer et al., 1995). Thick biological specimens can be physically trimmed by cryo sectioning (Al-Amoudi et al., 2004, McDowall et al., 1983) or focused ion beam milling (Hayles et al., 2010, Marko et al., 2006, Rigort et al., 2012, Strunk et al., 2012).

In recent years, several correlative methodologies emerged aimed at combining cryo fluorescence light microscopy with cryo electron microscopy (Briegel et al., 2010, Jun et al., 2011, Le Gros et al., 2009, Plitzko et al., 2009, Sartori et al., 2007, Schwartz et al., 2007). All these implementations rely on a two-step approach. First, regions of interest are selected by imaging the sample on a fluorescence microscope equipped with a cryo specimen stage. Next, the sample is transferred to a cryo electron microscope and the areas of interest are re-located based on the recorded fluorescence data. The advantage of such a two-step approach is that one can choose the best suited microscopy technique for both modalities. The drawbacks are the increased chance for sample contamination and damage during the transfer between the microscopy modalities and the time, effort and skills needed for the transfer and relocation.

Here, we describe the use of an integrated light and electron microscope (ILEM) for cryo applications. Previously, we reported the design of the ILEM and some room temperature applications (Agronskaia et al., 2008). We present a second-generation ILEM (named ILEM2) dedicated to work at cryogenic temperatures and describe the hardware and software modifications needed to optimize the cryo CLEM workflow. Examples are presented of cryo correlative microscopy in biological systems that are representative for a variety of potential applications and fluorescent markers. The use of an integrated light and electron microscope increases the correlation success rate in cryo correlative microscopy and opens up new avenues for structure–function studies.