Super-resolution microscopy

Super-resolution microscopy is a form of light microscopy. Due to the diffraction of light, the resolution of conventional light microscopy is limited, as stated by Ernst Abbe in 1873. A good approximation of the resolution attainable is the full width at half maximum (FWHM) of the point spread function, and a precise widefield microscope with high numerical aperture and visible light usually reaches a resolution of ~250 nm.

Super-resolution techniques allow images to be taken with a higher resolution than the diffraction limit. They fall into two broad categories: "true" super-resolution techniques, which capture information contained in evanescent waves, and "functional" super-resolution techniques, which use clever experimental techniques and known limitations on the matter being imaged to reconstruct a super-resolution image.

True subwavelength imaging techniques include those that utilize the Pendry Superlens and near field scanning optical microscopy, the 4Pi Microscope and structured illumination microscopy technologies like SIM and SMI. However, the majority of techniques of importance in biological imaging fall into the functional category.

There are two major groups of methods for functional super-resolution microscopy:

1. Deterministic super-resolution: The most commonly used emitters in biological microscopy, fluorophores, show a nonlinear response to excitation, and this nonlinear response can be exploited to enhance resolution. These methods include STED, GSD, RESOLFT and SSIM.
2. Stochastic super-resolution: The chemical complexity of many molecular light sources gives them a complex temporal behavior, which can be used to make several close-by fluorophores emit light at separate times and thereby become resolvable in time. These methods include Super-resolution optical fluctuation imaging (SOFI) and all single-molecule localization methods (SMLM) such as SPDM, SPDMphymod, PALM, FPALM, STORM and dSTORM.

On October 8, 2014, the Nobel Prize in Chemistry was awarded to Eric Betzig, W.E. Moerner and Stefan Hell for "the development of super-resolved fluorescence microscopy," which brings "optical microscopy into the nanodimension".

History

In 1978, the first theoretical ideas had been developed to break the Abbe limit using a 4Pi Microscope as a confocal laser scanning fluorescence microscope where the light is focused ideally from all sides to a common focus that is used to scan the object by 'point-by-point' excitation combined with 'point-by-point' detection.

Some of the following information was gathered (with permission) from a chemistry blog's review of sub-diffraction microscopy techniques Part I and Part II. For a review, see also reference.

In 1986, the super-resolution optical microscope based on stimulated emission was patented by Okhonin.

True super-resolution techniques

Near-field scanning optical microscope (NSOM)

Main article: Near-field scanning optical microscope

Near-field scanning is also called NSOM. Probably the most conceptual way to break the diffraction barrier is to use a light source and/or a detector that is itself nanometer in scale. Diffraction as we know it is truly a far-field effect: The light from an aperture is the Fourier transform of the aperture in the far-field, but, in the near-field, all of this is not necessarily the case. Near-field scanning optical microscopy (NSOM) forces light through the tiny tip of a pulled fiber — and the aperture can be on the order of tens of nanometers. When the tip is brought to nanometers away from a molecule, the resolution is limited not by diffraction but by the size of the tip aperture (because only that one molecule will see the light coming out of the tip). An image can be built by a raster scan of the tip over the surface to create an image.

The main down-side to NSOM is the limited number of photons you can force out a tiny tip, and the minuscule collection efficiency (if one is trying to collect fluorescence in the near-field). Other techniques such as ANSOM (see below) try to avoid this drawback.

Local enhancement / ANSOM / optical nano-antennas

Instead of forcing photons down a tiny tip, some techniques create a local bright spot in an otherwise diffraction-limited spot. ANSOM is apertureless NSOM: it uses a tip very close to a fluorophore to enhance the local electric field the fluorophore sees. Basically, the ANSOM tip is like a lightning rod, which creates a hot spot of light.

Bowtie nanoantennas have been used to greatly and reproducibly enhance the electric field in the nanometer gap between the tip's two gold triangles. Again, the point is to enhance a very small region of a diffraction-limited spot, thus improving the mismatch between light and nanoscale objects—and breaking the diffraction barrier.

Near-field optical random mapping (NORM) microscopy

NORM (near-field optical random mapping) microscopy is a method of optical near-field acquisition by a far-field microscope through the observation of nanoparticles' Brownian motion in an immersion liquid.

NORM utilizes object surface scanning by stochastically moving nanoparticles. Through the microscope, nanoparticles look like symmetric round spots. The spot width is equivalent to the point spread function (~ 250 nm) and is defined by the microscope resolution. Lateral coordinates of the given particle can be evaluated with a precision much higher than the resolution of the microscope. By collecting the information from many frames one can map out the near field intensity distribution across the whole field of view of the microscope. In comparison with NSOM and ANSOM this method does not require any special equipment for tip positioning and has a large field of view and a depth of focus. Due to the large number of scanning "sensors" one can get shorter time of image acquisition.

4Pi

A 4Pi microscope is a laser scanning fluorescence microscope with an improved axial resolution. The typical value of 500–700 nm can be improved to 100–150 nm, which corresponds to an almost spherical focal spot with 5–7 times less volume than that of standard confocal microscopy.

The improvement in resolution is achieved by using two opposing objective lenses, both of which are focused to the same geometrical location. Also, the difference in optical path length through each of the two objective lenses is carefully aligned to be minimal. By this, molecules residing in the common focal area of both objectives can be illuminated coherently from both sides and also the reflected or emitted light can be collected coherently, i.e. coherent superposition of emitted light on the detector is possible. The solid angle $\Omega$ that is used for illumination and detection is increased and approaches the ideal case. In this case the sample is illuminated and detected from all sides simultaneously.

Up to now, the best quality in a 4Pi microscope was reached in conjunction with the STED principle.

Structured illumination microscopy (SIM)

Comparison of the resolution obtained by confocal laser scanning microscopy (top) and 3D structured illumination microscopy (3D-SIM-Microscopy, bottom). Shown are details of a nuclear envelope. Nuclear pores (anti-NPC) red, nuclear envelope (anti-Lamin) green, chromatin (DAPI-staining) blue. Scale bars: 1µm.

There is also the wide-field structured-illumination (SI) approach to breaking the diffraction limit of light. SI—or patterned illumination—relies on both specific microscopy protocols and extensive software analysis post-exposure. The main concept of SI is to illuminate a sample with patterned light and increase the resolution by measuring the fringes in the Moiré pattern (from the interference of the illumination pattern and the sample). "Otherwise-unobservable sample information can be deduced from the fringes and computationally restored."

SI enhances spatial resolution by collecting information from frequency space outside the observable region. This process is done in reciprocal space: The Fourier transform (FT) of an SI image contains superimposed additional information from different areas of reciprocal space; with several frames with the illumination shifted by some phase, it is possible to computationally separate and reconstruct the FT image, which has much more resolution information. The reverse FT returns the reconstructed image to a super-resolution image.

• Images of cell nuclei and mitotic stages recorded with 3D-SIM Microscopy.
• 

  Comparison confocal microscopy – 3D-SIM

• 

  Cell nucleus in prophase from various angles

• 

  Two mouse cell nuclei in prophase.

• 

  mouse cell in telophase

Spatially modulated illumination (SMI)

SMI + TIRF of human eye tissue affected by macular degeneration

SMI microscopy is a light optical process of the so-called point spread function-Engineering. These are processes that modify the Point Spread Function (PSF) of a microscope in a suitable manner to either increase the optical resolution, to maximize the precision of distance measurements of fluorescent objects that are small relative to the wavelength of the illuminating light, or to extract other structural parameters in the nanometer range.

The Vertico SMI microscope achieves this in the following manner: the illumination intensity within the object range is not uniform, unlike conventional wide field fluorescence microscopes, but is spatially modulated in a precise manner by the use of one or two opposing interfering laser beams along the axis. The object is moved in high-precision steps through the wave field, or the wave field itself is moved relative to the object by phase shift. This results in an improved axial size and distance resolution.

SMI can be combined with other super resolution technologies, for instance with 3D LIMON or LSI-TIRF as a total internal reflection interferometer with laterally structured illumination. This SMI technique allows one to acquire light-optical images of autofluorophore distributions in the sections from human eye tissue with previously unmatched optical resolution. Use of three different excitation wavelengths (488, 568 and 647 nm), enables one to gather spectral information about the autofluorescence signal. This has been used for human eye tissue affected by macular degeneration AMD.

Deterministic functional techniques

RESOLFT microscopy is an optical microscopy with very high resolution that can image details in samples that cannot be imaged with conventional or confocal microscopy. Within RESOLFT the principle of STED microscopy and GSD microscopy are generalized. Also, there are techniques with other concepts than RESOLFT or SSIM, for example fluorescence microscopy using optical AND gate property of nitrogen-vacancy center.

Stimulated emission depletion (STED)

A figure showing the resolution improvement between traditional confocal microscopy and STED microscopy.

Main article: STED microscopy

STED (stimulated emission depletion microscopy) uses two laser pulses, the excitation pulse for excitation of the fluorophores to their fluorescent state and the STED pulse for the de-excitation of fluorophores by means of stimulated emission. In practice, the excitation laser pulse is first applied whereby a STED pulse soon follows (But STED without pulses using continuous wave lasers is also used). Furthermore, the STED pulse is modified in such a way that it features a zero-intensity spot, which coincides with the excitation focal spot. Due to the non-linear dependence of the stimulated emission rate on the intensity of the STED beam, all the fluorophores around the focal excitation spot will be in their off state (the ground state of the fluorophores). By scanning this focal spot one retrieves the image. The FWHM of the PSF of the excitation focal spot can theoretically be compressed to an arbitrary width by raising the intensity of the STED pulse, according to equation (1).

$\Delta r \approx \frac{\Delta}{\sqrt{1+I_\max/I_s}}$   (1)

∆r is the lateral resolution, ∆ is the FWHM of the diffraction limited PSF, I_max is the peak intensity of the STED laser, and $I_{s}$ is the threshold intensity needed in order to achieve saturated emission depletion.

The main disadvantage of STED, which has prevented its widespread use, is that the machinery is complicated. On the one hand, the image acquisition speed is relatively slow for large fields of view because of the need to scan the sample in order to retrieve an image. On the other hand, it can be very fast for smaller fields of view: Recordings with up to 80 frames per second have been shown. Due to a large I_s value associated with STED, there is a need for a high-intensity excitation pulse, which may cause damage to the sample.

Ground state depletion (GSD)

GSD microscopy or ground state depletion microscopy, uses the triplet state of a fluorophore as the off-state and the singlet state as the on-state, whereby an excitation laser is used to drive the fluorophores at the periphery of the singlet state molecule to the triplet state. This is much like STED, where the off-state is the ground state of fluorophores, which is why equation (1) also applies in this case. The $I_{s}$ value is smaller than in STED making super-resolution imaging possible at a much smaller laser intensity. Compared to STED though, the fluorophores used in GSD are generally less photostable and the saturation of the triplet state may be harder to realize.

Saturated structured illumination microscopy (SSIM)

SSIM (saturated structured-illumination microscopy) exploits the nonlinear dependence of the emission rate of fluorophores on the intensity of the excitation laser. By applying a sinusoidal illumination pattern with a peak intensity close to that needed in order to saturate the fluorophores in their fluorescent state one retrieves Moiré fringes. The fringes contain high order spatial information that may be extracted by computational techniques. Once the information is extracted a super-resolution image is retrieved.

SSIM requires shifting the illumination pattern multiple times, effectively limiting the temporal resolution of the technique. In addition there is a need of very photostable fluorophores due to the saturating conditions. These conditions also induce radiation damage to the sample that restricts the possible applications in which SSIM may be used.

Structured illumination microscopy enhances the resolution only by a factor of 2 (because the SI pattern cannot be focused to anything smaller than half the wavelength of the excitation light). To further increase the resolution, one can introduce nonlinearities, which show up as higher-order harmonics in the FT. In reference, Gustafsson uses saturation of the fluorescent sample as the nonlinear effect. A sinusoidal saturating excitation beam produces the distorted fluorescence intensity pattern in the emission. This nonpolynomial nonlinearity yields a series of higher-order harmonics in the FT.

Each higher-order harmonic in the FT allows another set of images that can be used to reconstruct a larger area in reciprocal space, and thus a higher resolution. In this case, Gustafsson achieves less than 50-nm resolving power, more than five times that of the microscope in its normal configuration.

The main problems with SI are that, in this incarnation, saturating excitation powers cause more photodamage and lower fluorophore photostability, and sample drift must be kept to below the resolving distance. The former limitation might be solved by using a different nonlinearity (such as stimulated emission depletion or reversible photoactivation, both of which are used in other sub-diffraction imaging schemes); the latter limits live-cell imaging and may require faster frame rates or the use of some fiduciary markers for drift subtraction. Nevertheless, SI is certainly a strong contender for further application in the field of super-resolution microscopy.

Examples for this microscopy are shown under section 2.4 Structured illumination microscopy (SIM), Images of cell nuclei and mitotic stages recorded with 3D-SIM Microscopy.

Stochastic functional techniques

Localization microscopy

Single-molecule localization microscopy (SMLM) summarizes all microscopical techniques that achieve super-resolution by isolating emitters and fitting their images with the point spread function (PSF). Normally, the width of the point spread function (~ 250 nm) limits resolution. However, given an isolated emitter, one is able to determine its location with a precision only limited by its intensity according to equation (2).

$\Delta \mathrm{loc} \approx \frac{\Delta}{\sqrt{N}}$   (2)

Here, Δloc is the localization precision, Δ is the FWHM of the PSF and N is the number of collected photons. This fitting process can only be performed reliably for isolated emitters (see Deconvolution), and interesting biological samples are so densely labeled with emitters that fitting is impossible when all emitters are active at the same time. SMLM techniques solve this dilemma by activating only a sparse subset of emitters at the same time, localizing these few emitters very precisely, deactivating them and activating another subset.

Generally, localization microscopy is performed with fluorophores. Suitable fluorophores reside in a non-fluorescent dark state for most of the time and are activated stochastically, typically with an excitation laser of low intensity. A readout laser stimulates fluorescence and bleaches or photoswitches the fluorophores back to a dark state, typically within 10–100 ms. The photons emitted during the fluorescent phase are collected with a camera and the resulting image of the fluorophore (which is distorted by the PSF) can be fitted with very high precision even on the order of a few Angstroms. Repeating the process several thousand times ensures that all fluorophores can go through the bright state and are recorded. A computer then reconstructs a super-resolved image.

The desirable traits of fluorophores used for these methods, in order to maximize the resolution, are that they should be bright. That is, they should have a high extinction coefficient and a high quantum yield. They should also possess a high contrast ratio (ratio between the number of photons emitted in the light state and the number of photons emitted in the dark state). Also, a densely labeled sample is desirable according to the Nyquist criteria.

The multitude of localization microscopy methods differ mostly in the type of fluorophores used.

Localization microscopy SPDM

Single YFP molecule super-resolution microscopy using SPDMphymod

SPDM (Spectral Precision Distance Microscopy), the first described localization microscopy technology 1997 is a light optical process of fluorescence microscopy, which allows position, distance and angle measurements on "optically isolated" particles (e.g. molecules) well below the theoretical limit of resolution for light microscopy.

"Optically isolated" means that at a given point in time, only a single particle/molecule within a region of a size determined by conventional optical resolution (typically approx. 200–250 nm diameter) is being registered. This is possible when molecules within such a region all carry different spectral markers (e.g. different colors or other usable differences in the light emission of different particles). The structural resolution achievable using SPDM can be expressed in terms of the smallest measurable distance between two in their spatial position determined punctiform particle of different spectral characteristics ("topological resolution"). Modeling has shown that under suitable conditions regarding the precision of localization, particle density etc., the "topological resolution" corresponds to a "space frequency," which in terms of the classical definition is equivalent to a much-improved optical resolution.

SPDM is a localization microscopy that achieves an effective optical resolution several times better than the conventional optical resolution, represented by the half-width of the main maximum of the effective point image function. By applying suitable laser optical precision processes, position and distances can be measured with nanometer accuracy between targets with different spectral signatures.

TMV Virus Super Resolution Light Microscopy

SPDMphymod

Dual color localization microscopy SPDMphymod/super-resolution microscopy with GFP & RFP fusion proteins

Localization microscopy for many standard fluorescent dyes like GFP, Alexa dyes and fluorescein molecules is possible if certain photo-physical conditions are present. With this so-called SPDMphymod (physically modifiable fluorophores) technology a single laser wavelength of suitable intensity is sufficient for nanoimaging in contrast to other localization microscopy technologies that need two laser wavelengths when special photo-switchable/photo-activatable fluorescence molecules are used. A further example for the use of SPDMphymod is an analysis of Tobacco mosaic virus (TMV) particles or the study of virus–cell interaction.

Based on singlet triplet state transitions it is crucial for SPDMphymod that this process is ongoing and leading to the effect that a single molecule comes first into a very long-living reversible dark state (with half-life of several seconds even) from which it returns to a fluorescent state emitting many photons for several milliseconds before it returns into a very long-living so-called irreversible dark state. SPDMphymod microscopy uses fluorescent molecules that are emitting the same spectral light frequency but with different spectral signatures based on the flashing characteristics. By combining two thousands images of the same cell, it is possible using laser optical precision measurements to record localization images with significantly improved optical resolution.

Standard fluorescent dyes already successfully used with the SPDMphymod technology are GFP, RFP, YFP, Alexa 488, Alexa 568, Alexa 647, Cy2, Cy3, Atto 488 and fluorescein.

STORM, PALM and FPALM

Main article: Photoactivated localization microscopy

Stochastic optical reconstruction microscopy (STORM), photo activated localization microscopy (PALM) and fluorescence photo-activation localization microscopy (FPALM) are super-resolution imaging techniques that utilize sequential activation and time-resolved localization of photoswitchable fluorophores to create high resolution images. During imaging, only an optically resolvable subset of fluorophores is activated to a fluorescent state at any given moment, such that the position of each fluorophore can be determined with high precision by finding the centroid position of the single-molecule images of particular fluorophore. The fluorophore is subsequently deactivated, and another subset is activated and imaged. Iteration of this process allows numerous fluorophores to be localized and a super-resolution image to be constructed from the image data. These three methods were published independently during a short period of time and their principle is identical. STORM was originally described using Cy5 and Cy3 dyes attached to nucleic acids or proteins, while PALM and FPALM was described using photoswitchable fluorescent proteins. In principle any photoswitchable fluorophore can be used, and STORM has been demonstrated with a variety of different probes and labeling strategies. Using stochastic photoswitching of single fluorophores, such as Cy5, STORM can be performed with a single red laser excitation source. The red laser both switches the Cy5 fluorophore to a dark state by formation of an adduct and subsequently returns the molecule to the fluorescent state. Many other dyes have been also used for STORM. In addition to single fluorophores, dye-pairs consisting of an activator fluorophore (such as Alexa 405, Cy2, and Cy3) and a photoswitchable reporter dye (such as Cy5, Alexa 647, Cy5.5, and Cy7) can be used for STORM. In this scheme, the activator fluorophore, when excited near its absorption maximum, serves to reactivate the photoswitchable dye to the fluorescent state. Multicolor imaging has been performed by using different activation wavelengths to distinguish dye-pairs based on the activator fluorophore used or using spectrally distinct photoswitchable fluorophores either with or without activator fluorophores. Photoswitchable fluorescent proteins can be used as well. Highly specific labeling of biological structures with photoswitchable probes has been achieved with antibody staining, direct conjugation of proteins, and genetic encoding. STORM has also been extended to three-dimensional imaging using optical astigmatism, in which the elliptical shape of the point spread function encodes the x, y, and z positions for samples up to several micrometers thick, and has been demonstrated in living cells. To date, the spatial resolution achieved by this technique is ~20 nm in the lateral dimensions and ~50 nm in the axial dimension and the temporal resolution is as fast as 0.1–0.33s.

Label-free localization microscopy

Optical resolution of cellular structures in the range of about 50 nm can be achieved even in label-free cells using the localization microscopy SPDM.

Beside a substantial resolution improvement of autofluorescent structures, using two different laser wavelengths SPDM reveals cellular objects which are not detectable under conventional fluorescence wide-field imaging conditions.

As a control the positions of the detected objects in the localization image match these in the bright-field image.

Label-free superresolution microscopy has also been demonstrated using the fluctuations of surface enhanced Raman scattering signal on a highly uniform plasmonic metasurface.

Direct stochastical optical reconstruction microscopy (dSTORM)

dSTORM utilizes the photoswitching of a single fluorophore. In dSTORM, fluorophores are embedded in an oxidizing and reducing buffer system (ROXS) and fluorescence is excited. Sometimes, stochastically, the fluorophore will enter a triplet or some other dark state which is sensitive to the oxidation state of the buffer. As the molecules return stochastically they can be excited to fluoresce so that single molecule positions can be recorded. Development of the dSTORM method occurred in 3 independent laboratories at about the same time and was called 'Reversible photobleaching microscopy -RPM", "Ground state depletion microscopy followed by individual molecule return" -GSDIM as well as the now generally accepted moniker dSTORM.

Software for localization microscopy

Localization microscopy depends heavily on software that can precisely fit the point spread function (PSF) to millions of images of active fluorophores within a few minutes. Since the classical analysis methods and software suites used in natural sciences are too slow to computationally solve these problems, often taking hours of computation for processing data measured in minutes, specialised software programs have been developed. Once molecule positions have been determined, the locations need to be displayed and several algorithms for display have been developed e.g.

Super-resolution optical fluctuation imaging (SOFI)

Main article: Super-resolution optical fluctuation imaging

It is possible to circumvent the need for PSF fitting inherent in single molecule localization microscopy (SMLM) by directly computing the temporal autocorrelation of pixels. This technique is called super-resolution optical fluctuation imaging (SOFI) and has been shown to be more precise than SMLM when the density of concurrently active fluorophores is very high.

Combination of techniques

3D light microscopical nanosizing (LIMON) microscopy

3D Dual Colour Super Resolution Microscopy with Her2 and Her3 in breast cells, standard dyes: Alexa 488, Alexa 568 LIMON

3 D LIMON (Light MicrOscopical nanosizing microscopy) images using the Vertico SMI microscope are made possible by the combination of SMI and SPDM, whereby first the SMI and then the SPDM process is applied.

The SMI process determines the center of particles and their spread in the direction of the microscope axis. While the center of particles/molecules can be determined with a 1–2 nm precision, the spread around this point can be determined down to an axial diameter of approx. 30–40 nm.

Subsequently, the lateral position of the individual particle/molecule is determined using SPDM, achieving a precision of a few nanometers.

As a biological application in the 3D dual color mode the spatial arrangements of Her2/neu and Her3 clusters was achieved. The positions in all three directions of the protein clusters could be determined with an accuracy of about 25 nm.

See also

• Multifocal plane microscopy (MUM)
• Stimulated emission depletion microscope (STED)
• Photoactivated localization microscopy (PALM)
• Stochastic optical reconstruction microscopy (STORM)
• Deconvolution
• Photoactivatable probes

External links

• List of Superresolution Microscope Instruments and Manufacturers (organized by filter options)
• Light Microscopy: An ongoing contemporary revolution (Introductory Review)
• Everyday Scientist: Breaking the Diffraction Limit of Light, part 1
• Everyday Scientist: Breaking the Diffraction Limit of Light, part 2
• MBInfo: Super Resolution Microscopy
• History of Super Resolution Microscopy / Optical Nanoscopy