Structured Illumination Microscopy for the Investigation of Synaptic Structure and Function

Abstract

The neuronal synapse is a primary building block of the nervous system to which alterations in structure or function can result in numerous pathologies. Studying its formation and elimination is the key to understanding how brains are wired during development, maintained throughout adulthood plasticity, and disrupted during disease. However, due to its diffraction-limited size, investigations of the synaptic junction at the structural level have primarily relied on labor-intensive electron microscopy or ultra-thin section array tomography. Recent advances in the field of super-resolution light microscopy now allow researchers to image synapses and associated molecules with high-spatial resolution, while taking advantage of the key characteristics of light microscopy, such as easy sample preparation and the ability to detect multiple targets with molecular specificity. One such super-resolution technique, Structured Illumination Microscopy (SIM), has emerged as an attractive method to examine synapse structure and function. SIM requires little change in standard light microscopy sample preparation steps, but results in a twofold improvement in both lateral and axial resolutions compared to widefield microscopy. The following protocol outlines a method for imaging synaptic structures at resolutions capable of resolving the intricacies of these neuronal connections.

1 Introduction

Synapses form key communication sites of the nervous system and as such, development and maintenance of healthy synapses are critical for the formation of functional neural networks. Loss of synaptic integrity and proper function has been linked to many developmental and neurodegenerative diseases including autism, schizophrenia, Alzheimer’s disease, and Huntington’s disease [1–4]. Careful analysis of synapse biology, in particular in intact brain tissue, has been limited with conventional light microscopy tools due to synapses’ small size, high density, and complex matrix. Recent advances in the field of super-resolution light microscopy now allow for the investigation of synapse structure and function at a more detailed level, opening a new method to study potential underlying mechanisms of synapse biology in both normal and diseased brains.

Region-specific synapse dysfunction and loss has long been considered a hallmark of Alzheimer’s disease [5–8]; however, during the disease pathogenesis, how or when synapse loss occurs remains elusive. One approach is to examine the integrity of synapse structure in different brain regions of diseased versus control tissue by analyzing the close apposition of various pre- and post-synaptic markers. This has been challenging with conventional microscopy due to the diffraction limiting nature of light and the small size of synaptic structures. Electron microscopy studies show a typical synapse in the hippocampus to have a postsynaptic density ranging in area from 0.07 μm² to 0.42 μm² [9]. However, super- resolution microscopy techniques now provide the ability to visualize synapses in intact brain tissue and to resolve the distribution of proteins on their surface with a light microscope [10, 11]. These techniques also enable quantification of different populations of synapses within brain regions, something previously only possible with labor-intensive techniques such as immuno-electron microscopy or ultra-thin section array tomography.

A number of super-resolution techniques have been developed in recent years. In general, they can be classified under three umbrella terms: Structured Illumination Microscopy (SIM), Single Molecule Localization (SML), and REversible Saturable OpticaL Fluorescence Transitions (RESOLFT) [12]. All three of these techniques have been implemented commercially and are becoming widely accessible in core facilities and individual labs around the world. Although SIM does not attain the same level of resolution improvement that SML or RESOLFT do, SIM is appealing as it does not require large changes to standard light microscopy sample preparation, is relatively fast, and produces limited photo-damage and bleaching. In comparison to ultra-thin section array tomography, SIM provides comparable axial resolution, improved lateral resolution, simpler sample preparation, and eliminates artifacts caused by torn, folded, or incorrectly registered tissue sections.

A Structured Illumination Microscope uses constructive and deconstructive interference of excitation light at the focal plane of an objective to illuminate a sample with a series of sinusoidal stripes of high spatial frequency. These stripes are shifted laterally to illuminate the entire sample, and rotated to effect resolution enhancement in all lateral directions and an image is taken at each position [13, 14]. If the sample possesses high-frequency spatial information, the stripes produce moiré fringes that are observed in Fourier space and can be used to recover super-resolution data that is normally lost as light diffracts through the optics of the microscope. In an ideal sample, SIM produces a twofold improvement in resolution. SIM can also be implemented to form a structured illumination pattern in the axial dimension, thereby producing a twofold resolution improvement in x, y, and z postprocessing [13, 15].

2 Materials

2.1 Solutions

All solutions should be prepared with ultrapure water and analytical- grade reagents.

1. TBS (tris-buffered saline): 7.86 g Tris–HCl, 8.67 g NaCl, Add H₂O to 1 L, Adjust pH to 7.4 with HCl.

2. PFA Solution: 4 % Paraformaldehyde in PBS, freshly made, EM grade.

3. Sucrose Solution: 30 % sucrose in PBS.

4. OCT: Cryosectioning specimen matrix (e.g., Tissue-Tek^® O.C.T. compound; Sakura^® Finetek USA).

5. Blocking Buffer: 20 % goat serum (v/v) in TBS.

6. Antibody Dilution Buffer: 0.3 % Triton X-100 (v/v), 10 % normal goat serum (v/v) in TBS.

7. TDE Mounting Medium: 97 % v/v 2,2′-Thiodiethanol (TDE), 2.4 % v/v 1,4-Diazabicycle[2.2.2]octane (DABCO), 0.6 % v/v PBS, Refractive Index = 1.51.

2.2 Microscope Setup

SIM has now been implemented by a number of commercial manufacturers. This protocol is routinely carried out on a Zeiss ELYRA PS1 microscope; however, it should be widely applicable to all other manufacturers.

Recommended features:

1. Multiple diffraction gratings or a tunable spatial light modulator matched to the wavelength of all laser lines to be used.

2. Optics to collect −1, +1, and 0 diffraction orders (0 order needed for 3D SIM).

3. Axial piezo stage with high accuracy and reproducibility. Step sizes of ~100 nm are needed to ensure Nyquist sampling in the axial dimension.

4. High sensitivity, low read noise, high bit depth camera. Both EMCCD and sCMOS cameras can be used. Pixel spacing should be <50 nm to ensure Nyquist sampling.

5. High numerical aperture (NA) objective lens (≥1.4).

6. Software to perform multichannel alignment (if imaging more than one channel).

7. Software to perform structured illumination calculations.

8. Analysis software for synapse quantification (e.g., Imaris, MatLab).

2.3 Color Alignment

Multicolor imaging requires multiple optical components (diffraction gratings, filter cubes) to be moved in and out of the light path. Each of these components can introduce slight shifts between color channels that can affect colocalization analysis (Fig. 1). Subdiffraction multicolor fluorescent spheres (Tetra-spec beads, 100–200 nm, Life Technologies) can be imaged to develop a transformation matrix that is applied to the final SIM image for color channel alignment.

Fig. 1.

Adjusting for chromatic aberration in SIM. (a–c) 200 nm beads imaged under the following conditions: 488 nm excitation, 495–575 nm emission collection (green); 561 nm excitation, 570–650 nm emission collection (red); 642 nm excitation, 655+ nm emission collection (blue). (a) SIM processed image of 200 nm beads. Enlarged regions in corners identify differing levels of chromatic shift within the same image. (b) Same image as in (a) after color channel alignment. (c) Lateral and axial projections of a single bead before and after color channel alignment

2.4 Mounting Medium

A number of commercial mounting media are available for purchase. However, many super-resolution modalities are very sensitive to index of refraction mismatches. A slight RI mismatch can reduce axial resolution and imaging depth (as the structured illumination pattern quickly degrades) and produce artifacts during SIM processing (Figs. 2 and 3). Therefore, it is prudent to use mounting media matched as closely as possible to the immersion medium of your microscope objective. As most high NA objectives use oil immersion with an RI of approximately 1.51, glycerol-based mounting media are often not ideal (RI = 1.40–1.49; Table 1). If these must be used, it is recommended to allow them to cure for more than 72 h as the curing process will increase the RI. A number of solvent-based mounting media can be used that very closely match the RI of immersion oil and also have tissue clearing properties [16, 17]. As brain tissue is highly scattering, clearing properties are desirable. Caution must be used as some dyes are quenched in solvents, while others will increase in brightness [16]. Aqueous tissue samples will need to be moved through a dilution series of increasing solvent concentration prior to final coverslip mounting [16]. New, high RI aqueous mountants are now becoming available as well [18].

Fig. 2.

Poor refractive index matching results in aberration. 100 nm red fluorescent beads were imaged (561 nm excitation, 570–650 nm emission) and processed using SIM. Beads were mounted in Immersol 518F (RI = 1.518, Carl Zeiss Microscopy) or Immersol W 2010 (RI = 1.33, Carl Zeiss Microscopy) under #1.5 thickness coverslips. Point-spread functions were measured on ten beads, averaged, and presented as lateral (XY) and axial (XZ) projections

Fig. 3.

Processing induced artifacts. 14 μm section of the dLGN of a p5 mouse stained with presynaptic marker VGlut2 and postsynaptic marker homer showing examples of grid line and noise artifacts that can arise during SIM processing. This can occur as a result of a poor signal-to-noise ratio in the sample, incomplete filtering of the grid frequency prior to inverse Fourier transformation, or intensity fluctuations between the raw phase and rotation images at the same Z plane (for example when photobleaching occurs). White arrows indicate regions where SIM processing artifacts were introduced

Table 1.

Refractive indices of common mounting media

Mounting media	Refractive index	Manufacturer
Fluoromount G	1.40	SouthernBiotech
75 % Glycerol	1.44	Sigma
ProLong Gold	Fresh: 1.39 Cured 24 h: 1.40 Cured 4 days: 1.44	Life Technologies
Vectashield	Cured: 1.46	Vector Labs
Mowiol	Cured: 1.49	Sigma
SeeDB2	1.51	–
97 % TDE	1.51	Sigma
100 % DBE	1.56	Sigma
Wintergreen Oil	1.53	Sigma

2.5 Immersion Oil

The RI of immersion oil is measured at or near 23 °C. Differing from this temperature will change the RI of the oil. Depending on environmental conditions, ventilation around the microscope, how long the microscope has been running, and whether or not live-cell environmental stage incubators are present, the temperature near the sample may be far from 23 °C. A number of temperature-corrected immersion oils are commercially available; therefore, it is recommended that multiple oils be compared empirically (see Note 1).

2.6 Coverslips

Most microscope objectives are corrected for #1.5, or 170 μm, thick coverslips (this is indicated on the side of the objective lens; see Note 2).

3 Methods

3.1 Tissue Preparation

1. Deeply anesthetize a mouse with 2.5 % Avertin.

2. Perfuse the mouse with ice-cold 20–30 ml phosphate buffered saline (PBS), and then subsequently with 20–30 ml freshly made PFA Solution. Carefully remove the brain and place it in PFA Solution for 2 h on ice. Alternatively, the PFA perfusion step can be skipped. Perfuse mouse with PBS and drop fix the brain in PFA Solution for 2 h.

3. Rinse the brain 3× with PBS to remove any remaining PFA Solution and immediately cryoprotect by placing it in Sucrose Solution. Leave the brain in sucrose at 4 °C until the tissue sinks to the bottom of the tube (24–48 h).

4. Once the tissue sinks, prepare the brain for sectioning. The following steps are performed in the preparation for cryostat sectioning:

   1. Incubate the tissue for 1 h at RT in 2:1 mixture of Sucrose Solution:OCT in a small tissue mold.

   2. Freeze OCT-embedded brain by placing on dry ice. OCT- embedded brain can be stored at -80 °C for several weeks until ready to be sectioned.

3.2 Staining

1. Section the tissue using a cryostat (14 μm sections) ideally within 2 weeks of preparation and collect on superfrost slides.

2. Warm slides on a 37 °C heat block for 30 min.

3. Wash slides 2× with TBS.

4. Incubate slides with Blocking Buffer for 2 h at room temperature (RT).

5. Incubate slides with 1° antibody (diluted in Antibody Dilution Buffer to a working concentration that was previously empirically determined for the specific antibody) overnight at 4 °C.

6. Wash slides 3× for 10 min with TBS.

7. Incubate slides with 2° antibody (diluted in Antibody Dilution Buffer to a working concentration that was previously empirically determined) for 2 h at RT.

8. Wash slides 3× for 30 min with TBS.

9. Perform dehydration series if using solvent-based mounting medium.

10. Apply a small drop of desired mounting medium to each section and mount a coverslip on the top (22 × 50 mm, No. 1.5).

11. Seal edges of slides with nail polish and store at 4 °C until imaging session.

3.3 Imaging

1. Prepare multicolor bead slide for SIM calibration and channel alignment:

   1. Dilute 100 nm Tetraspec beads (Life Technologies) 1:500 (v/v) in ethanol.

   2. Sonicate for 3 min in a bench top water bath sonicator.

   3. Place two drops of bead solution on a slide, spread across slide with a razor blade, and allow to dry.

   4. Place a drop of mounting media on a coverslip and place over dried beads avoiding the introduction of air bubbles.

2. Image bead slide:

   1. Acquire a full field of view 3D SIM image stack centered on the beads. Z-step = 90 nm, Z-range = 3.5 μm for each color channel present in the sample of interest.

   2. Process the images using the same SIM algorithms to be used for the sample of interest (Fig. 1).

   3. Measure full width half maximum (FWHM) of beads to assess performance of the microscope (if beads are subdif-fraction limit in size).

   4. Calculate a transformation matrix to align each color channel in x, y, and z. You will need an affine transformation; translation alone will not be sufficient (Fig. 1).

3. Image sample:

   1. Individually adjust laser power, camera exposure time, and camera gain (if applicable) to fill a significant portion of the dynamic range of the camera. Avoid saturating any pixels. Avoid amplifying noise, nonspecific staining, autofluoresence, etc. if possible.

   2. Depending on the number of grating rotations and channels imaged, acquisition times can be significant. Ensure all fluorophores used are stable and laser powers are not high enough to induce photobleaching resulting in decreasing levels of fluorescence at later rotations or Z planes as this can introduce artifacts (Fig. 3).

   3. If color channels are imaged sequentially, image higher wavelength channels first to reduce bleaching of yet to be imaged fluorophores.

   4. Acquire a 3D SIM image stack over desired range. Ensure Nyquist sampling in x, y, and z based on the expected resolution after SIM processing.

4. SIM processing:

   1. Process each color channel using SIM algorithms available in commercial or custom software (Fig. 4).

   2. Align each color channel based on the transformation matrix calculated in Subheading 3.3, step 2.

   3. Assess image for artifacts: background noise and/or incomplete removal of grid lines. If artifacts are seen, adjust parameters of SIM processing algorithm and reprocess (Fig. 3).

Fig. 4.

SIM imaging of mouse brain sections. Widefield (a), point scanning confocal (b), and SIM (c) images of a 14 μm section of an adult (3 months) mouse brain stained with presynaptic marker synaptotagmin (green) and postsynaptic homer (red). Images were processed identically using ImageJ: a smoothing filter was applied to all channels, and a 0.5 gamma adjustment was applied to all green channels (to allow visualization of dim synapses, without saturating larger puncta)

3.4 Synapse Quantification Using Imaris and Matlab

1. Open SIM-processed file in Imaris. Make sure that the Volume icon is selected in the top left menu.

2. Confirm the pixel values. Go to the Edit menu and select Image Properties. In this window, the x, y, and z voxel sizes are displayed. These values depend on the microscope, objective, camera, and z-step acquisition and can be found in the software used to acquire the image.

3. In the Display Adjustment window, maximize visualization of all immunoreactive synaptic puncta by increasing brightness for each channel (Fig. 5a).

4. Create spots for each channel:

   1. In the top tool bar, make sure the Surpass mode is selected.

   2. In the Display Adjustment window, select the channel for which to create spots.

   3. Click the Spots icon.

   4. A create window will be displayed in the bottom left. In Algorithm Settings, make sure “Segment ROI” is unchecked and the “Select Different Spot Sizes (Region Growing)” is checked. Click the blue forward button at the bottom.

   5. In the next window, choose one source channel from the pull-down menu. For Spot Detection, select the “Detect Ellipsoids” option and enter expected xy and z diameters. (These values should be determined empirically depending on the particular antibodies used and tissue regions imaged. The average values for the diameters can be determined by measuring the visualized fluorescent puncta in the Slice mode. See Note 3 for some experimentally determined values.) Uncheck background subtraction.

   6. Spot detection: choose “Quality Filter.” Click on histogram and drag until spots created appropriately reflect center (local maxima) of all fluorescent puncta displayed in the source channel. Note the threshold value used. This value can vary between the two channels, but within one specific channel the value should be kept consistent between sets of images that are to be compared.

   7. Choose to threshold using “Absolute Intensity” for region growing.

   8. Define spot growth boundary: confirm whether the threshold displayed in white appropriately reflects the fluorescent image of the source channel. This is determined empirically by rotating the image and moving the view up and down the z-plane to ensure that the white signal is appropriately overlaid on the fluorescent signal. To rotate, select “Navigate” in the Pointer menu on the upper right side of the screen, click and hold image to rotate. To move the viewing plane along the z-axis, click “Select” in the Pointer menu then move the highlighted plane along the z-axis. To return image to original orientation, select “Origin Bottom Left” from the View menu. Click the green “Finish” button.

   9. Confirm that the spots created reflect the fluorescent image in the source channel. One can change color of the spots to a different one from the original fluorescent image to better differentiate the two images.

   10. Click on the “Funnel” (i.e., Filter tab) and select volume from the pull-down menu. Turn on upper threshold and filter out any spots that are deemed too big to be considered synaptic.

   11. Go to the Graph tab and click on “Details” and select any parameters to display. For displaying multiple parameters (volume, diameters, etc.), go to the Wrench icon at the bottom left and select the parameters to record.

   12. Record the number of spots and save the file before continuing.

   13. Create spots for the second channel by following the steps above. Figure 5b displays a maximum intensity projection of spot selection in a two-channel image.

5. Quantify number of colocalized spots:

   1. Go to Image processing, then select Spot Functions, then Colocalize Spots. This will open MatLab.

   2. Select the two spots to colocalize.

   3. Enter the threshold value of the distance between spot centers. This will be dependent on the particular pre- and postsynaptic antibodies used.

   4. Apposed spots will now appear (Fig. 5c, d). Go to the Graphs tab and record number of collocated spots for each spot.

Fig. 5.

Data analysis. 3D-SIM in stratum radiatum of 3 months mouse hippocampus (CA1) in which synapses are immunolabeled with presynaptic marker synaptophysin (green) and postsynaptic marker PSD-95 (red). (a) Maximum intensity projection (MIP) of synaptophysin and PSD-95 immunostaining in 3 months hippocampus. (b) MIP in which spots were created for synaptophysin (green) and PSD-95 (red), respectively. (c, d) Apposed synaptophysin and PSD-95 immunoreactive spot centers (created in b) that are within 0.4 μm (c) or 0.2 μm (d) distance of one another