Correlative Super-Resolution Microscopy: New Dimensions and New

---

Review pubs.acs.org/CR

Correlative Super-Resolution Microscopy: New Dimensions and New Opportunities Meghan Hauser,†,∥ Michal Wojcik,†,∥ Doory Kim,†,∥ Morteza Mahmoudi,‡ Wan Li,† and Ke Xu*,†,§ †

Department of Chemistry, University of California, Berkeley, California 94720, United States Center for Nanomedicine and Department of Anesthesiology, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts 02115, United States § Division of Molecular Biophysics and Integrated Bioimaging, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States ‡

ABSTRACT: Correlative microscopy, the integration of two or more microscopy techniques performed on the same sample, produces results that emphasize the strengths of each technique while offsetting their individual weaknesses. Light microscopy has historically been a central method in correlative microscopy due to its widespread availability, compatibility with hydrated and live biological samples, and excellent molecular specificity through fluorescence labeling. However, conventional light microscopy can only achieve a resolution of ∼300 nm, undercutting its advantages in correlations with higher-resolution methods. The rise of superresolution microscopy (SRM) over the past decade has drastically improved the resolution of light microscopy to ∼10 nm, thus creating exciting new opportunities and challenges for correlative microscopy. Here we review how these challenges are addressed to effectively correlate SRM with other microscopy techniques, including light microscopy, electron microscopy, cryomicroscopy, atomic force microscopy, and various forms of spectroscopy. Though we emphasize biological studies, we also discuss the application of correlative SRM to materials characterization and singlemolecule reactions. Finally, we point out current limitations and discuss possible future improvements and advances. We thus demonstrate how a correlative approach adds new dimensions of information and provides new opportunities in the fast-growing field of SRM.

CONTENTS 1. Introduction 2. Correlating Super-Resolution Microscopy with Other Light Microscopy Techniques 3. Correlating Super-Resolution Microscopy with Electron Microscopy 3.1. Advent of Correlative Super-Resolution Light and Electron Microscopy 3.2. Challenges in Correlative Super-Resolution Light and Electron Microscopy Choice of Substrate Electron Microscopy-Related Strong Fixation, Heavy-Metal Staining, and Resin Embedding Sample Dehydration Alignment of Super-Resolution Microscopy and Electron Microscopy Images 3.3. Unsectioned Samples 3.4. Cryosectioned Samples 3.5. Resin-Embedded Samples 3.6. Metal-Replica TEM 4. Correlating Cryo-Super-Resolution Microscopy with Cryo-Electron Microscopy 5. Correlating Super-Resolution Microscopy with Atomic Force Microscopy © XXXX American Chemical Society

6. Correlating Super-Resolution Microscopy with Spectroscopy Fluorescence Spectroscopy Fluorescence Lifetime Infrared Spectroscopy Mass Spectrometry Single-Molecule Force Spectroscopy 7. Correlative Super-Resolution Microscopy for Nonbiological Systems 8. Outlook 8.1. Correlative Super-Resolution Light and Electron Microscopy Further Development of Fixation- and Embedding-Resistant Fluorophores and Epitopes Dual-Contrast Agents as Correlative Probes for Super-Resolution Correlative Light and Electron Microscopy Correlating Electron Microscopy with Live-Cell Super-Resolution Microscopy

B C E E G G G H H I K K M

R R S S S T T U U U

V V

Special Issue: Super-Resolution and Single-Molecule Imaging

N

Received: September 2, 2016

P A

DOI: 10.1021/acs.chemrev.6b00604 Chem. Rev. XXXX, XXX, XXX−XXX

Chemical Reviews Integrated Correlative Super-Resolution Light and Electron Microscopy Correlating Super-Resolution Microscopy with Other Electron Microscopy-Related Methods 8.2. Correlative Cryo-Super-Resolution Microscopy Synthetic Dyes for Cryo-Super-Resolution Microscopy High Numerical Aperture Objective Lenses for Cryo-Super-Resolution Microscopy Correlating Cryo-Super-Resolution Microscopy with Cryo-Soft X-ray Tomography 8.3. Correlative Super-Resolution Microscopy and Atomic Force Microscopy Concurrent Super-Resolution Microscopy− Atomic Force Microscopy Correlating Super-Resolution Microscopy with High-Speed Atomic Force Microscopy 8.4. Concluding Remarks Author Information Corresponding Author ORCID Author Contributions Notes Biographies Acknowledgments References

Review

corroborated information about a system regarding morphology, functionality, dynamics, cellular context, and chemical composition.6−8 Light microscopy is a pivotal method in correlative microscopy due to its ease of access and the above-mentioned benefits. However, the resolution of light microscopy was historically limited to about half the wavelength of light, or ∼300 nm, due to diffraction. The significant difference in resolution between light microscopy and high-resolution microscopy methods like EM and atomic force microscopy (AFM) produces difficulty in fully utilizing the strengths of light microscopy in correlative applications. The past decade has witnessed the rapid rise of superresolution microscopy (SRM), the collective name given to a host of emerging fluorescence microscopy methods that break the conventional resolution limit by reinventing how fluorescence signal is generated, detected, and processed.9−15 By achieving resolution down to ∼10−20 nm, SRM overcomes the standing resolution disparity between light microscopy and EM/AFM and can provide more powerful and meaningful correlation results. From the standpoint of SRM itself as an emerging class of techniques, correlation with other advanced microscopy and spectroscopy techniques also helps validate new results as well as adding new dimensions of information. The new opportunities afforded by correlative SRM have hence motivated extensive new research. Challenges abound. It is naturally difficult to combine two distinct microscopy techniques: sample preparation protocols optimized for one technique may be entirely incompatible with the other, and results from the two modalities can be difficult to correlate due to differences in contrast mechanism and sample damage during handling steps between the two experiments. While these issues are relevant to all correlative approaches, they are especially challenging for correlative SRM: as SRM pushes the resolution of light microscopy from its traditional limit of ∼300 nm to ∼10 nm, it also raises the bar for how well structures need to be preserved and correlated. In terms of sample preparation, this means new strategies that are compatible with both SRM and its correlated methods need to be developed. Fluorescent tags must attain a labeling density much higher than that is necessary for conventional fluorescence microscopy, so that the true structure is not obscured by voids in fluorescence due to incomplete labeling at the nanoscale. Moreover, SRM often imposes stringent requirements on the photophysics of the fluorescent probes,14,16,17 including photoswitchablilty and photostability, and these special properties of the probes need to be preserved during sample preparation. For correlating the results, alignment between images obtained by SRM and the other methods needs to be achieved to a precision down to ∼10 nm; distortions beyond this length scale, due to either degradation of sample or imaging artifacts, would defeat the very purpose of high-resolution correlative SRM. Other constraints of SRM, including the typical requirement of oil-immersion objectives to achieve the highest possible numerical aperture (which, in turn, necessitates glass coverslips), low sample background fluorescence, and aqueous imaging buffers, further complicate the equation. Overcoming these challenges has thus become a major endeavor, for which researchers have devised vastly diverse strategies and experimental designs: these form the basis of this review. We note that by SRM, we refer to the recently developed farfield methods that are readily applied to biological samples and

V V W W W W W W W W W W X X X X X X

1. INTRODUCTION Imagine, for a moment, a child outside playing in the grass. She can see that the blades are short, thin, and verdant. She can feel with her hands that they are short, thin, and cool to her touch. She can smell that fresh-mown-lawn scent when she puts her face close enough to the earth. Altogether, she forms a clear image in her mind of the structure and even composition of these living objects by correlating, integrating, and synthesizing the information she receives from multiple senses. Scientists, meanwhile, have expanded the array of human “senses” to probe objects at previously unimaginable scales, notably through microscopy. However, just as with the child in the grass, no single type of technique can provide all information, or even all structural information, about a (biological) system. Put another way, every microscopy technique has its intrinsic limitations. For example, while electron microscopy (EM) boasts subnanometer resolution, it is generally incompatible with live or wet samples, achieves limited molecular specificity, and often suffers from a restricted viewing window. In contrast, light microscopy works well with live and wet samples across a large range of sample dimensions and, with fluorescence microscopy, achieves excellent molecular specificity for multiple targets via the proper tagging of fluorescent labels, for example, encoding of fluorescent proteins (FPs) or immunolabeling of synthetic dyes. Consequently, although traditional light microscopy offers limited (∼300 nm) resolution, its combination with EM has been highly successful in overcoming the aforementioned limits of EM.1−5 More generally, when any two or more microscopy/ spectroscopy modalities are correlatively combined to probe the same targeted area, either simultaneously or in tandem, the strengths of each can often be reinforced, while the drawbacks are mitigated. Correlative microscopy thus arrives on the scientific scene to attain multidimensional, multiscale, and B

DOI: 10.1021/acs.chemrev.6b00604 Chem. Rev. XXXX, XXX, XXX−XXX

Chemical Reviews

Review

Figure 1. An example of correlating diffraction-limited fluorescence microscopy to identify targets for SRM. (A) Diffraction-limited two-color epifluorescence image of Alexa Fluor 647−phalloidin-labeled actin filaments (green) and Alexa Fluor 555-immunolabeled MAP2 (magenta), a marker for dendrites, for a cultured neuron sample. (B) Three-dimensional (3D) STORM image of actin corresponding to the box in panel A, revealing contrasting actin ultrastructure in the MAP2-positive dendrite and MAP2-negative axons. (C) Epifluorescence image of Alexa Fluor 647− phalloidin-labeled actin filaments (green) and Alexa Fluor 555-immunolabeled NrCAM (magenta), a marker for axon initial segments. (D) 3D STORM image of actin corresponding to the box in panel C. Color is used to present axial (z) positions in panels B and D according to the color bar in panel D. Adapted with permission from ref 35. Copyright 2013 AAAS.

hence were popularized over the past decade.9−15 Earlier nearfield approaches, in particular near-field scanning optical microscopy,18,19 also break the diffraction limit of resolution, but their application is essentially limited to sample surfaces. Far-field SRM methods are generally divided into two categories: one overcomes the diffraction limit by engineering the illumination patterns to effectively reduce the size of the point-spread function. Methods in this category include stimulated emission depletion (STED) microscopy,20,21 ground-state depletion (GSD) microscopy,22,23 and [saturated] structured-illumination microscopy ([S]SIM)/saturated patterned excitation microscopy (SPEM).24−26 The other category achieves super-resolution through locating single molecules that undergo stochastic “on−off” fluorescence photoswitching. This approach was introduced as stochastic optical reconstruction microscopy (STORM),27 photoactivated localization microscopy (PALM),28 and fluorescence photoactivation localization microscopy (FPALM).29 Variants are given many different names, including PALMIRA (photoactivated localization microscopy with independently running acquisition),30,31 dSTORM (direct stochastic optical reconstruction microscopy),32 and GSDIM (ground-state depletion followed by single-molecule return).33 In a related approach called PAINT (points accumulation for imaging in nanoscale topography),34 the reversible binding of fluorescent molecules to targets is used to achieve single-molecule localization and SRM. In this review we use single-molecule localization microscopy (SMLM) to generally refer to the class of SRM methods based on singlemolecule localization, but when discussing specific work, we adopt the name used in each study without modification.

be used to carry out both experiments. Although SRM provides unmatched optical resolution, at this time it still faces a few key limitations that may be complemented by correlating with conventional, diffraction-limited light microscopy methods. To begin with, SRM is relatively low in throughput. Acquisition of a single, high-resolution SRM image may take minutes; the use of high-magnification objective lenses further limits the imaging window to