Two-Photon Excitation STED Microscopy for Imaging Synapses and Glia in Acute Brain Slices

Abstract

STED (stimulated emission depletion) microscopy is a novel fluorescence microscopy technique that provides unprecedented access to dynamic processes of living cells with a spatial resolution well below 100 nm, even down to a few nanometers. By breaking the classical diffraction barrier of optical microscopy, it offers the chance to close the gap in spatial resolution between electron microscopy and conventional light microscopy and is attracting attention as a potential game-changing imaging technology.

The first-generation STED microscopes typically suffered from significant disadvantages over confocal or two-photon microscopy, regarding labeling, live-cell compatibility, color contrast, and depth penetration. However, due to several recent technical developments, these teething problems have been largely overcome. Similar to confocal microscopy, STED microscopy can be readily combined with a powerful array of complementary techniques, such as photo-uncaging, fluorescence recovery after photobleaching (FRAP), and patch-clamp electrophysiology, giving experimental access to a rich set of molecular, structural, and functional data.

Here, we present a primer on two-photon excitation STED microscopy, its basic principles and instrumentation requirements, providing a detailed how-to guide for the construction and operation of a STED microscope for two-color nanoscale imaging of neural morphology deep inside living brain slices.

1 Introduction

1.1 Development of STED Microscopy

Fluorescence microscopy has become one of the dominant experimental techniques in the biosciences, because it can capture dynamic events inside living cells with exquisite sensitivity and specificity. However, the spatial resolution of conventional optical microscopy is limited by the diffraction of light to around half the wavelength of the light used in the microscope [1], which makes it difficult to investigate many subcellular compartments such as synapses or glial processes, because they can be considerably smaller than the diffraction limit (<200 nm).

Fortunately, recent advances in optical microscopy techniques have broken the diffraction limit, starting the new era of superresolution fluorescence microscopy. Stimulated emission depletion (STED) microscopy was the first viable concept in this respect [2, 3], but other powerful techniques based on single-molecule imaging (PALM/STORM) were developed soon afterward [4–6].

As a result, it is now possible to perform sub-diffraction cellular imaging with a spatial resolution well below 100 nm, opening up exciting research avenues for neuroscience and synapse research in particular.

Initially, STED microscopy was used for studying the distribution of immunolabeled synaptic proteins in fixed neuronal cultures [7–9]. Subsequently, it became possible to also use it for live-cell imaging applications, e.g., revealing synaptic vesicle dynamics in nerve terminals [10] or imaging the morphology of dendritic spines in organotypic brain slices [11].

Since then, many aspects of STED microscopy have been improved, regarding setup complexity and robustness, labeling and multicolor imaging, and sample preparation and handling, substantially expanding the scope and usefulness of this novel technique for neurobiologists.

Like most other superresolution techniques, STED microscopy was initially limited to monolayers of dissociated cells or cells right on the surface of the sample. But it has now been successfully extended to thick tissue preparations, including brain slices [12] and intact brain in vivo [13], through the use of glycerol objectives equipped with a correction collar, which reduces spherical aberrations as compared with oil objectives. In addition, STED microscopy has recently been combined with two-photon excitation using long-working distance water objectives and an upright microscope design [14, 15].

These were important steps because many biological preparations, such as the popular acute brain slice preparation, require deep optical access, i.e., beyond the layer of dead cells and debris on the surface, and there needs to be enough space for inserting pipettes for electrophysiological recordings or local perfusion.

In this chapter, we will describe how STED microscopy can be used for live-cell imaging in two colors with a lateral resolution of around 50 nm. We start out by going over the basic principles of STED microscopy, before describing the technical challenges and practical solutions concerning the construction and operation of a homebuilt upright STED microscope, which is based on two-photon excitation and pulsed fluorescence quenching. We close by presenting a couple of application examples of STED imaging of the morphology of dendrites and microglia cells deep inside (~50 μm) acute brain slices.

1.2 Basic Concept of STED

The spatial resolution of a scanning fluorescence microscope is defined by the size of the focal excitation volume, termed point spread function (PSF) of the microscope. The spatial extent of the PSF therefore defines the minimal distance at which two closely spaced objects can still be resolved. Because of the diffraction of light, its size is on the order of a few 100 nm in the lateral dimension, depending on the wavelength of light (λ) and the numerical aperture (NA) of the objective, according to the Abbe formula [1]:

$$ d=\frac{\lambda }{2* NA} $$

(1)

The basic idea of STED microscopy is to reduce the size of the PSF and hence improve the spatial resolution of the microscope, by spatially constricting the fluorescence signal. This is accomplished by a second, red-shifted laser (the STED laser), which quenches the fluorescence by the process of stimulated emission. Because the STED laser features a doughnut-shaped intensity distribution with a deep zero in its center, the fluorescence inhibition is restricted to the edge of the PSF allowing signal light to be collected from the center. By increasing the intensity of the STED laser, the size of the central region, from where fluorescence is permitted to occur, can be made in principle arbitrarily small. As a result, the Abbe formula takes on this new form:

$$ d=\frac{\lambda }{2* NA*\sqrt{1+{I}_{STED}/{I}_S}\operatorname{}} $$

(2)

where I _STED denotes the incident power of the STED laser and I _s the intensity at which half of the molecules are quenched [16].

According to Eq. 2, d tends to zero as I _STED increases. In this way, a resolution of 5.8 nm has been achieved for samples of diamond crystals, which is about two orders of magnitude smaller than the wavelength of the excitation light [17].

The STED laser is usually tuned to the long wavelength tail end of the emission spectrum, where it does not excite the molecules, and where it can be separated well from the fluorescence signal by standard dichroic mirrors and emission filters (Fig. 1a–c).

Fig. 1

Basic principle of STED microscopy. (a) Principle of STED microscopy. A diffraction-limited two-photon excitation spot is scanned across the sample, and its emission signal is detected. In STED microscopy, a doughnut-like quenching beam is superimposed on the excitation beam. The quenched molecules in the periphery of the STED beam do not fluoresce anymore, and, hence, the detected emission signal originates only from the center of the STED doughnut. The resolution of the STED microscope is then, in addition to Abbe’s law (λ wavelength, NA numerical aperture), also defined by the power of the STED beam (I) and a property of the dye (I _s ). (b) Spectral schema and reflection from gold beads for spatial alignment. As the two-photon beam is in the near-IR spectrum (910 nm), the emission signal can be detected over a broad range (<600 nm). The STED beam at 600 nm allows simultaneous quenching of GFP and YFP. The reflections form gold beads are detect with a PMT and used for spatial alignment of the excitation and the STED beam. Excitation action cross section from Warren Zipfel, Cornell University, USA (Scale bars, 500 nm). (c) Synchronization of two-photon and STED pulses. A line scan of a calcein solution demonstrates the quenching effect of the STED laser if excitation, and STED pulses are synchronized and optimally delayed in time. A quenching efficiency of 80 % can be achieved. Varying the relative delay between the pulses allows for direct measurement of the STED pulse width and the lifetime of the molecule. (d) A resolution of 62 nm can be observed on 40-nm-large fluorescent beads. This is strongly contrasted to the 368 nm measured in two-photon mode alone and illustrates a sixfold improvement in lateral resolution

Because the time that the molecules spent in the excited state is short (the fluorescence lifetime of fluorescent proteins and green organic dyes like Alexa 488 is a few nanoseconds), it is better to use pulsed lasers, which achieve much more efficient fluorescence quenching for a given amount of STED laser intensity.

2 Materials

2.1 Optical Components for Beam Steering

To begin with, great care must be taken to create a diffraction-limited PSF of the excitation and the STED light. To achieve this, the choice of optical components is critical, and only high-quality optical components with an optical flatness of λ/10 should be used. Because of the dependence of two-photon excitation on pulse width, care has to be taken not to stretch the pulses in time excessively while propagating through the optical components. For beam steering, dedicated two-photon mirrors (for excitation) and dielectric visible mirror (for STED light and fluorescence) are available. High-grade silver mirrors can be used when all wavelengths (two-photon, STED, and emission signal) are combined as they reflect (>98 %) over a broad range of wavelengths (375–1,000 nm) while keeping pulse dispersion to a minimum. It is helpful to use a motorized 45° flip mirror as the final mirror before the objective so that it can be flipped out of the way to allow for wide-field imaging.

In addition, various components such as lens telescopes, half-wave plates (λ/2), quarter-wave plates (λ/4), and polarizing beam splitters as well as instruments and tools for beam diagnostics and alignment such as beam profiler, power meter, spectrometer, shear plate, and oscilloscopes are essential.

2.2 Microscope Base and Objective

To minimize temperature drift and mechanical vibrations, it is important to mount all components as close as possible to the table (1" posts). For an upright microscope, it is therefore recommended to raise the essential components via a breadboard to the height of the scan-tube lens axis (Fig. 2). The use of a modular research-grade upright microscope has the advantage that essential optical components are included, such as condenser, light illuminator, and phase contrast, facilitating the combination with an electrophysiology setup. Our system is developed around an Olympus BX-51WI in conjunction with a motorized sample stage (380FM, Luigs & Neumann, Reutlingen, Germany).

Fig. 2

Design of homebuilt two-photon excitation STED microscope. (a) Overview of the custom build two-photon STED microscope. (b) Its light sources. (c) The optical beam path. An upright Olympus BX51-WI is modified so that dedicated scan-tube lenses can be incorporated. The beam is deflected via a motorized flip mirror toward the piezo-actuator-controlled objective. Electrophysiological recording equipment is mounted on the motorized stage (1). The excitation laser (910 nm) originates from a Ti–Sapphire laser (Millenia/Tsunami) (2). This laser is synchronized to the STED laser, consisting of the Ti–Sapphire laser Mai Tai (3), pumping an OPO (4). The STED light is then coupled into a 20-m-long single-mode fiber (5). All laser attenuation is realized via electro-optical modulators (6). The STED laser passes a polarizing beam splitter (7), the vortex phase mask (8), and a λ/2 wave plate (9). STED and 2P lasers are combined using a dichroic mirror (10). Signal light is separated from the excitation and quenching lasers via a dichroic mirror (11) before passing suitable blocking and emission filters (12), focused on multimode fibers and detected via avalanche photo diodes (13). Two-photon and STED lasers are routed through a λ/4 (14) and into a beam scanner (15). 3D illustration performed with free version of Trimble SketchUp

As opposed to conventional light microscopy, where the numerical aperture of the objective and the excitation wavelength define the spatial resolution, in STED microscopy the optical resolution also depends on the power of the STED light. Nevertheless, when combining STED microscopy with two-photon excitation, high-NA objectives have the advantage of minimizing the excitation PSF, allowing for efficient two-photon excitation. A water-dipping objective of NA 1.1, long-working distance (1.5 mm) equipped with a correction collar and high visible and near-IR transmission (Olympus LUMFI 60x, NA 1.1) can be used to this end [15].

2.3 Two-Photon Excitation Laser Source, Synchronization with the STED Laser

In principle, the light source used for STED quenching can be either continuous wave (CW) or pulsed lasers. CW lasers are easier to implement while pulsed lasers have the advantage of restricting the optical power to the time the fluorophores is excited. This greatly reduces the photon load and hence photodamage of the sample. To ensure that the STED pulses have the right timing in respect to the excitation pulses, the excitation two-photon laser needs to have the option of electronically varying the cavity length so that it can be synchronized to the STED laser. This can be achieved by phase-locked loop electronics, which are commercially available from the two-photon laser companies Spectra-Physics and Coherent (Lok-to-Clock or Synchrolock). This solution can be implemented on a Spectra-Physics Tsunami HP fs 15 W with Model 3930 Lok-to-Clock for excitation but is not compatible with the modern generation of Ti–Sapphire lasers (i.e., Mai Tai or Chameleon). A Pockels cell (Conoptics, Danbury, CT) can be used to control the laser beam power for the imaging beam path.

2.4 STED Laser and Optical Parametric Oscillator

The STED laser is an essential component of the system and should be chosen according to the emission spectrum of the desired fluorophores. For green/yellow fluorophores such as GFP and YFP, a combination of a Ti–Sapphire laser (e.g., Spectra-Physics Mai Tai HP, wavelength: 797 nm, repetition rate ~80 MHz, pulse duration ~200 fs) pumping an optical parametric oscillator (such as OPO BASIC Ring fs RTP, APE, Berlin, Germany) is a powerful solution. This STED-OPO system provides the trigger signal for the synchronization with the two-photon laser. The incident STED laser power at the back focal plane is controlled via a dedicated Pockels cell.

2.5 STED Pulse Dispersion in Glass Rod and Polarization-Preserving Fiber

The OPO is tuned to 598 nm and delivers pulses of 200 fs duration with a repetition rate that is defined by the pump laser (~80 MHz). These pulses need to be stretched in time to allow for efficient fluorescence quenching and to avoid unintended multiphoton excitation. A 25-cm glass rod (SF6) stretches the pulses to around ~2 ps before being coupled into a 20-m-long polarization-maintaining single-mode fiber (Schäfter + Kirchhoff, Hamburg, Germany). The pre-broadening of the pulses in the glass rod protects the fiber from being damaged by the high-energy pulses and reduces spectral broadening within the fiber. A band-pass filter (593/40, AHF Analysentechnik, Tübingen, Germany) is used to spectrally clean up the light after the fiber.

2.6 Pulse Delay Generator

The synchronization of the laser pulses is realized via dedicated phase-locked loop electronics from the manufacturer of the excitation two-photon laser (Spectra-Physics, Model 3930 Lok-to-Clock). The exact pulse frequency of the lasers can be read out from fast photodiodes via a reflection of the laser beam or from the internal photodiodes. The ability to adjust the relative temporal delay between the pulses is essential for the optimization of STED quenching. The system in use allows for a temporal delay of 2 ns. Alternatively, the relative temporal delay can be optimized by changing the optical path length. As this involves the movement of optical components, beam misalignment during the delay adjustments can occur. A fast photodiode (such as 3932-LX, Spectra-Physics) under the objective can serve as a readout for the initial temporal coarse alignment.

2.7 Helical Vortex Phase Mask

The shape and the quality of the central null of the STED doughnut are essential to the performance of the microscope. A helical vortex phase plate (RPC Photonics, Rochester, NY), mounted on a xy-translation stage, is used for the generation of the STED doughnut. A polarizing beam splitter before the phase mask “cleans” the polarization state of the incident light so that the phase plate can create a destructive interference pattern in the center of the beam resulting in the doughnut-like intensity distribution in the focal plane. To shape the doughnut, λ/2 and λ/4 wave plates before the scanner are used to optimize both the symmetry and the central null intensity of the doughnut.

2.8 Dichroic Filters for Merging the Two Beams and Filtering Out the Emission Signal

The dichroic mirrors in use have to fulfill the λ/10 criterion and are typically dielectrically coated and 5 mm thick. We use a short-pass dichroic (F73-700UV, AHF) to combine both the STED and the two-photon beam before they pass a long-pass dichroic (580 DCXRUV, AHF) to separate the emission signal and direct it to the photo detectors (see Sect. 2.10).

2.9 Beam Scanner and Fast Nano-positioner of Microscope Objective

For live-cell applications and the combination of imaging and electrophysiology, beam scanning is preferred over stage scanning, even though the implementation is more challenging. Creating a stationary beam at the back focal plane is critical and realized via a scan and tube lens combination (FV300/U-TLUIR, Olympus, Hamburg, Germany) in conjunction with a telecentric galvo scanner (Yanus IV, TILL Photonics, Gräfelfing, Germany). On upright systems, it is helpful to define an objective height position and adjusting the stage so that a stable beam position can be achieved. Fast scanning in the z-dimension can be realized via a piezo-driven objective positioner (P-725, Physik Instrumente, Germany). Note that electronic noise and ground loops can interfere with the beam scanning, creating unwanted artifacts in the image.

2.10 Photodetectors

Due to the low signal count of STED images, high-quality detectors are very important. Both avalanche photo diodes (APDs) and photomultiplier tubes (PMTs) have been used on STED microscopes. APDs have the clear advantage of higher quantum efficiency and very low dark counts while the PMTs offer a higher dynamic range and larger detection area. New-generation hybrid detectors might fill this gap in the foreseeable future. Our system is equipped with two APDs (SPCM-AQR-13; PerkinElmer, Fremont, CA) to allow for two-channel spectral detection of the emission signal. A dichroic mirror (514RS, Semrock) separates the emission light in two channels. Both channels are focused via long focal length ( f = 300 mm) lenses onto multimode fibers with a 100-μm core diameter, corresponding to about 120 % of the back-projected Airy disk.

Because of the use of high-power STED and two-photon lasers, care has to be taken to efficiently shield the detectors from the excitation/STED light by suitable optical filters, such as a STED blocking filter (594S-25), a two-photon blocking filter (680SP-25), and an emission band-pass filter (520-70, all Semrock, Rochester, NY, USA). Due to the small detection area of APDs, detection has to occur in descanned mode. For imaging deeper inside the tissue, i.e., beyond the scattering length of photons (~100 μm), proximity PMT detectors (in a non-descanned arrangement) can be used, which collect scattered photons more efficiently [14].

2.11 Perfusion Recording Chamber

The recording chamber is equipped with temperature control (via the stage) and a perfusion system, designed for coverslips with a diameter of 18 mm. Due to the upright design of the system, standard electrophysiological chambers can be used.

3 Methods

3.1 General Alignment of Beams

As STED microscopy relies on the modulation of the excitation PSF by the STED beam, a precise alignment of the two beams is critical. Coarse beam path alignment is realized by sending either a green test diode laser or by using the STED laser backward, starting at the objective position. The scan mirrors need to be parked in their central position during this process. Because of the upright design, it can be difficult getting the test laser diode at the correct position. One solution is to couple the diode laser into a single-mode optical fiber and mounting the collimated output instead of the objective. Additional problems arise from the poor performance of Ti–Sapphire laser mirrors in the visible spectrum and the dichroic mirrors that transmit/reflect the laser diode light to only one beam path. Either a red diode laser can be used or the backward alignment is terminated at two pinholes that serve as reference to the forward beam alignment. The STED beam path is well aligned when the test laser can be coupled through the single-mode fiber that usually carries the STED light. The detectors are coarsely aligned via the test diode laser (taking care not to send the test laser directly onto the detectors) or by exciting a dye solution (i.e., calcein) and aligning the resulting fluorescence that can be seen on a white card in darkness. Two-photon lasers are class 4 lasers and require the use of eye safety equipment.

3.2 Making Doughnuts

The shape and quality of the STED doughnut is essential to the performance of the microscope. Using gold nanospheres (diameter = 150 nm; BB International, Cardiff, UK) and a pellicle beam splitter (Thorlabs, Maisons-Laffitte, France), the reflection of the gold beads can be directed toward a detector, such as a PMT. Here, one should consider the detection range of the PMT since both the ~600-nm STED light and the near-infrared two-photon light have to be detected. The reflection from the gold bead is a good indicator for the alignment of the beams, verifying that both beams are superimposed and the reflections do not show a tilt in one or the other direction. When the helical phase plate is correctly placed, it produces a clearly visible dark spot in the profile of the laser beam at some distance from the phase plate. This dark spot has to be centered in the beam but can be slightly adjusted with the xy-translation stage. By adjusting both the λ/2 and λ/4 wave plates before the scanner, the symmetry of the doughnut and the quality of the central null can be optimized while observing the gold bead reflections. Rotating the λ/2 will allow the user to optimize the symmetry while rotation of the λ/4 has a strong effect on the central minimum. A slight tilt of the λ/4 might yield a slightly better central minimum. The STED beam is optimally modulated when the doughnut is symmetric and the central null approaches 1 % of the intensity of the rim. Alignment of the beams in the z-direction can be achieved by slightly de-collimating one or both beams.

3.3 Evaluating Microscope Performance

As STED microcopy crucially relies on the ability of the STED laser to quench the fluorescence of the excited molecules, testing the quenching efficiency is one of the first steps in determining the performance of the microscope. Only when the beams are aligned in space and time, can a quenching efficiency of ~80 % be expected. The resolution of the STED system is most commonly assessed by imaging sub-diffraction-sized fluorescent beads that emit in the desired wavelength. Note that 40-nm-sized beads are hard to image with a two-photon microscope because they bleach very rapidly so that for the daily routine, 170-nm beads will provide enough information to evaluate the correct alignment.

3.4 Acute Brain Slices

All acute brain preparations can be used on this system, among them the widely used acute hippocampal slice preparation. Protocols on the preparation of acute brain slices can be found elsewhere [18]. We use a standard NaCl-based artificial cerebrospinal fluid (ACSF) solution consisting of (in mM) 124 NaCl, 3 KCl, 26 NaHCO₃, 1.25 NaH₂PO₄, 10 glucose, 2 CaCl₂, 1 MgCl₂, and 0.6 Trolox (pH 7.4, osmolarity ~305 mOsm/L). Trolox is an antioxidant that protects cells from free radicals produced by fluorescence excitation. Because of the relatively low signal of STED images, we selected strongly fluorescent mice lines, such as the Thy1-YFP and the CX3CR1-GFP.

3.5 Imaging Living Brain Slices

The quality of the slices is essential for obtaining high standard images. Lower-quality slices have a thick layer of damaged cells on the surface, which typically introduces high background fluorescence. To obtain images from healthy cells, the objective height should be adjusted to at least ~20 μm below the surface. Strong background fluorescence induced by the STED beam alone can be observed in non-healthy slices, indicating the need to terminate the experiment. Laser intensities, image size, pixel size, and dwell time have to be adapted for the desired resolution, fluorophore, and depth. Two-photon laser intensities are typically in the range of 10–25 mW and slightly higher for STED (20–40 mW), measured at the back focal plane of the objective. Pixel size should adhere to the Nyquist sampling theory and is usually between 20 and 40 nm for STED microscopes.

3.6 Two-Color STED Imaging by Linear Unmixing of GFP and YFP

Our dual-color schema relies on the fact that the emission of two highly overlapping spectra can be separated by their relative contributions on the two installed detectors [19]. Post-hoc analysis of a known cell population allows the calculation of the so-called mixing matrix, which in turn can be used for the identification of the fluorescence contributions for each pixel. The same STED wavelength can be used for quenching GFP and YFP, with the advantage that both colors are acquired simultaneously and are inherently co-aligned [19].

3.7 Image Analysis

As STED imaging employs raster scanning, no post-hoc analysis is necessary to produce a superresolved image as in other superresolution techniques (such as SIM, PALM, STORM). Measurement of fine structures, such as spine necks, should be performed on raw data of single frames instead of z-projections. This decreases the risk of including pixel from structures above and below the structure of interest. Using averaged intensity profiles from 2 to 3 pixel wide lines reduces the contribution of outlier values. Quantification of the line profiles should report the full width at half maximum (FWHM) done either on plotted observed values or by using nonlinear fit functions (such as Lorentzian or Gaussian functions). Two-photon STED image profiles seem to be fitted better with Lorentzian than with Gaussian functions. For presentation purposes only, images can be de-noised by using median or Gaussian filters.

4 Notes

• 1. Using a “hands-off” Ti–Sapphire laser (such as the Mai Tai or the Chameleon series) as the pump laser for the OPO has the advantage of minimizing alignment procedures during start-up of the system.

• 2. Since the resolution of STED microscopy depends on the intensity of the STED light, lower NA objectives can be used as well. We have acquired superresolved images using a water-dipping objective of NA 1.0 (data not shown).

• 3. It is critical not to introduce any translation of the beam while operating a laser power attenuation device such as a motorized, rotating λ/2 wave plate, placed in front of a polarizing beam splitter as this results in small, nevertheless detrimental horizontal translation of the STED beam. An electro-optical modulator, placed in front of the STED fiber, works very well. In addition, it is recommended to send only the necessary STED power through the fiber to minimize damage on the fiber core. Therefore, the power should be adjusted before the fiber input coupling.

5 Application Examples

5.1 Alignment and Performance

Figure 1b shows an image created by the reflection of gold beads on a PMT. The correct alignment of the two beams in xy- and z-direction and the quality of the STED doughnut can be verified with this test. In Fig. 1c, the fluorescence intensity of a calcein solution is strongly reduced by an optimally aligned and synchronized STED beam. If synchronization is disabled, despite the same optical power reaching the sample, quenching of the STED beam is strongly diminished. By varying the relative delay of the synchronized pulses, an estimate of both the width of the STED as well as the lifetime of the molecule can be obtained as proposed previously [20]. Using 40-nm-sized fluorescent beads, the optical resolution of the microscope can be assessed (Fig. 1d).

5.2 Life Cell Imaging

All presented images originate from acute brain slices. Figure 3 shows several examples of two-photon STED images recorded in either CA1 of the hippocampus or in the cortex. In Fig. 1a, the line profile across a spine neck illustrates that with conventional two-photon microscopy, this spine would be described as having a very large spine neck. Using STED microscopy, it becomes apparent that the structure consists of two, closely spaced, spines necks. The individual spine heads are located along above the plane presented and hardly visible because of the diffraction-limited z-resolution. In Fig. 3b, we have acquired several z-stacks along the primary dendrite of a cortical neuron, starting at the soma and spanning a total length of 540 μm. The higher magnification images B1 and B2 illustrate the sub-diffraction sized spine necks along the dendrite. In Fig. 3c, two examples of microglial processes (green) in close proximity to superresolved spines are presented. The left image (maximum intensity projection) was created by assigning each channel a color (green and red) and simply merging the two colors. The right image (single frame) was linearly unmixed, using an ImageJ Plugin (“Spectral Unmixing” by Walter J). For Fig. 4, we acquired the same z-stack every 2 min and present maximum intensity projections of the same structure. A modest change over time is detectable on most structures while obvious photodamage is not observed. Again, spine neck diameters, well below the diffraction limit of two-photon microscopy can be measured.

Fig. 3

Nanoscale imaging of neural morphology in acute brain slices. (a) Comparison of a hippocampal CA1 dendritic segment recorded from an acute brain slice in two-photon versus two-photon STED mode. The line profile of the spine neck (indicated by white line) clearly resolves two structures when imaged in STED mode. The FWHM of the individual spine neck indicates diameters of 115 nm and 132 nm, respectively. The images were recorded 49 μm below the surface of the slice. Presented is a crop of a median filtered image from a z-stack, each 512 × 512 pixel, 20-nm pixel size, 30-μs pixel dwell time. (b) Reconstruction of a cortical Layer 4/5 neuron from an acute brain slice. The main dendrite of a single neuron was reconstructed by recording 17 stacks, each 1,024 × 1,024 pixel, 40-nm pixel size, 15 frames per stack (∆z = 500 nm), 30-μs pixel dwell time. The dendrite was located 45–90 μm below the surface of the slice. Maximum intensity projections were arranged using an ImageJ Plugin (“2D stitching” by Preibisch, S). White boxes B1 and B2 indicate position of the higher magnification images below. Spine neck widths of 192 and 186 nm can be observed. Using this approach, neuronal morphology and fine structures can be studied in a high-throughput manner. (c) Microglial processes in close proximity to neuronal spines of cortical neurons. Both merging of channels (left ) and linear unmixing (right) allow sufficient contrast to distinguish cell types. Spine neck widths of 135 and 145 nm can be measured in single planes

Fig. 4

Time-lapse two-photon excitation STED imaging of neuronal morphology. Crop of median filtered maximum intensity projections of 6 stacks, recorded every 2 min at a depth of – 23 μm in cortex (6 frames per stack, ∆z = 500 nm, every frame is 20 μm × 20 μm, 512 × 512 pixel, pixel size 40 nm, 30 μs pixel dwell time). Several spine necks with diameters well below the diffraction limit of classic two-photon microscopy are visible. The arrows point at structures, which are morphologically dynamic under physiological conditions, in the absence of signs of photodamage (such as blebbing)