Epon Post Embedding Correlative Light and Electron Microscopy

Summary

We present a detailed protocol for Epon post-embedding correlative light and electron microscopy using a fluorescent protein called mScarlet. This method can maintain the fluorescence and the ultrastructure simultaneously. This technique is amenable to a wide variety of biological applications.

Abstract

Correlative light and electron microscopy (CLEM) is a comprehensive microscopy that combines the localization information provided by fluorescence microscopy (FM) and the context of cellular ultrastructure acquired by electron microscopy (EM). CLEM is a trade-off between fluorescence and ultrastructure, and usually, ultrastructure compromises fluorescence. Compared with other hydrophilic embedding resins, such as glycidyl methacrylate, HM20, or K4M, Epon is superior in ultrastructure preservation and sectioning properties. Previously, we had demonstrated that mEosEM can survive osmium tetroxide fixation and Epon embedding. Using mEosEM, we achieved, for the first time, Epon post embedding CLEM, which maintains the fluorescence and the ultrastructure simultaneously. Here, we provide step-by-step details about the EM sample preparation, the FM imaging, the EM imaging, and the image alignment. We also improve the procedures for identifying the same cell imaged by FM imaging during the EM imaging and detail the registration between the FM and EM images. We believe one can easily achieve Epon post embedding correlative light and electron microscopy following this new protocol in traditional EM facilities.

Introduction

Fluorescence microscopy (FM) can be used to obtain the localization and distribution of the target protein. However, the context that surrounds the target protein is lost, which is crucial for investigating the target protein thoroughly. Electron microscopy (EM) has the highest imaging resolution, providing several subcellular details. Nevertheless, EM lacks target labeling. By accurately merging the fluorescence image taken by FM with the gray image acquired by EM, correlative light and electron microscopy (CLEM) can combine the information obtained by these two imaging modes^1,2,3,4.

CLEM is a trade-off between fluorescence and the ultrastructure¹. Because of the limitations of current fluorescent proteins and the traditional EM sample preparation procedures, especially the use of osmic acid (OsO₄) and hydrophobic resins such as Epon, the ultrastructure always compromises fluorescence⁵. OsO₄ is an indispensable reagent in EM sample preparation, which is used to improve the contrast of EM images. Compared with other embedding resins, Epon is superior in ultrastructure preservation and sectioning properties⁵. However, no fluorescent proteins can retain the fluorescence signal after the treatment of OsO₄ and Epon embedding⁶. To overcome the limitations of fluorescent proteins, pre embedding CLEM was developed, in which FM imaging is done before EM sample preparation⁶. However, the drawback of pre embedding CLEM is the imprecise registration between the FM and the EM images⁵.

On the contrary, post embedding CLEM method performs the FM imaging after the EM sample preparation, the registration accuracy of which can reach 6-7 nm^5,6. To retain the fluorescence of fluorescent proteins, very low concentrations of OsO₄ (0.001%)³ or the high-pressure frozen (HPF) and freeze substitution (FS) EM preparation methods^4,7 have been used at the expense of compromised ultrastructure or the contrast of the EM image. The development of mEos4b greatly promotes the progress of post embedding CLEM, although glycidyl methacrylate is used as the embedding resin⁵. With the development of mEosEM, which can survive the OsO₄ staining and Epon embedding, Epon post embedding super-resolution CLEM was achieved for the first time, maintaining the fluorescence and ultrastructure simultaneously⁶. After mEosEM, several fluorescent proteins that can survive the OsO₄ staining and the Epon embedding were developed^8,9,10,11. This greatly promotes the development of CLEM.

There are three key aspects to Epon post-embedding CLEM. The first is the fluorescent protein, which should maintain the fluorescent signal after EM sample preparation. According to our experience, mScarlet is superior to other reported fluorescent proteins. The second is how to find the same cell imaged by FM imaging in EM imaging. To solve this problem, we improve the procedure for this step so that one can readily find the targeted cell. The last is the method to align the FM image with the EM image. Here, we detail the registration between the FM and the EM images. In this protocol, we express mScarlet in VGLUT2 neurons and demonstrate that mScarlet can target secondary lysosomes using Epon post-embedding CLEM. We provide step-by-step details for Epon post-embedding CLEM, without compromising the fluorescence and the ultrastructure.

Protocol

Animal husbandry and experiments were approved by the Institutional Animal Care and Use Committee of Fujian Medical University Medical Center. The step-by-step workflow of the current protocol is shown in Figure 1.

1. Sample preparation

1. Mouse brain
   1. Purchase transgenic mice (see Table of Materials) and oligonucleotide primers (see Table of Materials) to genotype these mice.
   2. Perfuse and remove the intact brain from the cranial vault following previously published protocol^12,13.
   3. Fix the mouse brain using 10 mL of fixative solution (see Table of Materials) at 4 °C overnight.
   4. Rinse off the fixative solution for 3 x 10 min with 0.1 M phosphate buffer (PB).
   5. Slice the mouse brain into sections with 500 µm thickness using an oscillating microtome (see Table of Materials).
   6. Cut the fluorescent regions into small blocks with a scalpel using a stereomicroscope (see Table of Materials).
      NOTE: The volume of the small blocks must not be larger than 1 mm³.
2. Cultured cells
   1. Seed HeLa cells into 6 cm cell culture dishes and maintain them in growth medium (Dulbecco's modified Eagle medium [DMEM] low glucose, supplemented with 10% fetal bovine serum [FBS]).
   2. On the following day, transfect the plasmids containing mScarlet into cells using DNA transfection reagent (see Table of Materials) following the instruction book of the transfection kit.
   3. Forty-eight hours after transfection, trypsinize the transfected cells and centrifuge them to obtain cell pellets at 1,000 × g. The cell pellets are called sample blocks.

2. EM sample preparation

1. Fixation
   1. Fix the sample blocks using a fixative solution (see Table of Materials) at 4 °C overnight.
   2. Remove the fixative solution and rinse off the fixative solution for 3 x 10 min with 0.1 M PB from the sample blocks.
2. Osmium tetroxide (OsO₄) staining
   1. Fix the sample blocks with OsO₄ solution (see Table of Materials) at 4 °C for 1 h.
   2. Remove the OsO₄ solution and rinse off the OsO₄ solution 3 x 10 min with ddH₂O from the sample blocks.
3. Uranyl acetate (UA) staining
   1. Stain the sample blocks using 2% UA solution (see Table of Materials) at 4 °C for 1 h.
   2. Remove the UA solution and rinse the sample blocks with ddH₂O for 3 x 10 min.
4. Gradient dehydration
   1. Dehydrate the sample blocks with gradient ethanol as follows: 50% for 10 min, 70% for 10 min, 80% for 10 min, 90% for 10 min, and 2 x 10 min in 100% ethanol.
   2. Dehydrate the sample blocks with anhydrous acetone for 3 x 10 min.
5. Resin infiltration
   1. Infiltrate the sample blocks with gradient resin (see Table of Materials): acetone to resin 3:1 for 1 h, acetone to resin 1:1 for 2 h, acetone to resin 1:3 for 3 h.
   2. Replace with the pure resin for 3 x 12 h.
6. Resin embedding
   1. Transfer sample blocks into capsules (see Table of Materials) using a toothpick.
   2. Add pure resin to the capsule containing the sample block and polymerize the resin in the oven at 60 °C for 14-16 h.

3. Coating of the coverglass with gold nanoparticles

1. Coverglass cleaning
   1. Wash the coverglasses with cleaning buffer (see Table of Materials) for 2 h at 142 °C.
   2. Rinse off the cleaning buffer 3 x 10 min with ddH₂O.
   3. Transfer the coverglasses carefully into a beaker containing anhydrous ethyl alcohol and incubate overnight.
   4. Air dry the coverglasses on a clean bench for several minutes.
2. Coat the coverglasses with gold nanoparticles.
   1. Fix a glass slide in the center of the rotation of a desktop centrifuge (see Table of Materials) using glass cement.
   2. Mount a coverglass in the center of the glass slide using sticky notes.
   3. Add 75 µL of 1% pioloform (see Table of Materials) onto the coverglass.
      NOTE: We also tested Formvar (see Table of Materials) and it works well.
   4. Spin for 3 min at 1,360 × g (Figure 2A) to produce a thin layer of pioloform on the surface of the coverglass.
   5. Incubate the coverglass surface with 250 µL of 0.1% poly-L-lysine (see Table of Materials) for 1 h at room temperature (Figure 2B).
   6. Rinse with ddH₂O and air dry the coverglass for several minutes.
   7. Dilute 80 nm gold nanoparticles (see Table of Materials) with ddH₂O to 25% (v/v, OD₆₀₀ = 0.07) (see Table of Materials) and sonicate for 15 min.
   8. Incubate the coverglass surface with the diluted gold nanoparticles for 1 h at room temperature (Figure 2C).
   9. Rinse with ddH₂O and air dry the coverglass for several minutes.
      NOTE: It works well 2 weeks after being coated with pioloform and gold nanoparticles. But we recommend using it immediately after being coated.

4. Ultrathin sectioning

1. Trim the surface of the sample block into a right-angled trapezoid. Cut sections at a thickness of 100 nm using a diamond knife (see Table of Materials) with an ultramicrotome (see Table of Materials). Separate a single section from the section band and float the single section in the water bath.
2. Add a drop of ddH₂O onto the center of the gold nanoparticle-coated coverglass. Pick up a section using a sample loop and invert the sample loop onto the surface of the water drop. Remove the sample loop, leaving the section on the surface of the water drop.
3. Air dry the coverglass for several minutes. Draw an annulus around the section on the opposite side of the coverglass using a marker pen.

5. Light microscopy imaging

1. Fluorescence recovery
   NOTE: The workflow is shown in Figure 3A.
   1. Add 10 µL of mounting buffer (see Table of Materials) onto the ultrathin section on the coverglass. Place a new clean coverglass onto the mounting buffer.
      NOTE: The key component of the mounting buffer is DABCO, which is used to recover the fluorescence of fluorescent proteins. This is very important for fluorescence recovery.
   2. Put these two coverglasses together with the ultrathin section into an imaging chamber (see Table of Materials). Make sure the coverglass with the drawn annulus is on the top.
2. Collect the navigation map.
   1. Find the section using the drawn annulus as a marker in brightfield imaging mode of an inverted fluorescence microscope (see Table of Materials). Move to one corner of the section. Turn on the white light and take the brightfield image.
   2. Turn off the white light. Turn on the laser (561 nm) and take fluorescent images of the same field of view (FOV).
   3. Turn off the laser. Move to the adjacent FOV, making sure these two FOVs have a 10% overlap. Take the brightfield image and fluorescent image of the second FOV.
   4. Repeat this operation until the whole section is covered. Record the imaging path that can be used to stitch the FOV.
      NOTE: For higher quality of the image, we recommend using the confocal microscope.
3. Image cells of interest.
   1. Use the navigation map to easily target cells of interest. Turn on the laser (561 nm) and take 300 frames of fluorescent images of the cell of interest. Turn off the laser.
   2. Turn on the white light and take sequential brightfield images (~100 frames).
   3. Move to another cell of interest.

6. Preparation for EM imaging

NOTE: The workflow is shown in Figure 3B.

1. Transfer the ultrathin section to the slot grid.
   1. Fill a glass jar with ddH₂O.
   2. Remove the coverglasses from the chamber and separate the two coverglasses with tweezers. Clamp the coverglass on which the section is located, wash off the mounting buffer, and air dry.
   3. Score a # around the ultrathin section using a single-sided blade and drop 10 µL of 12% hydrofluoric acid (see Table of Materials) at each corner of the #. Put the coverglass slowly into the water.
      NOTE: Hydrofluoric acid reacts with glass, making it easier to remove the pioloform film from the cover glass. Hydrofluoric acid is toxic. Use it in a chemical hood with gloves. Contact with skin requires immediate medical attention.
   4. Float the pioloform film and ultrathin section on the surface of the water together. Put an uncoated slot grid on the section to capture the section in the center of the grid.
   5. Cover a glass slide with parafilm (see Table of Materials) and pick up the grid with the pioloform film. Air dry the grid at room temperature.
2. Section staining
   1. Stain the section with 2% UA and Sato's triple lead. For more detail, refer to the previously published protocol².

7. Image analysis

1. Stitch the navigation map.
   1. Find the location of the raw data. Open the software ImageJ (see Table of Materials) to stitch the navigation map.
   2. Click on Plugins | Stitching | Grid/Collection stitching | (Type) Grid: snake by column | (Order) Up & Left (according to shooting sequence) | Set slice size | Select data path | Set file name | Check Display fusion. Observe the result, which is a navigation map made up of several brightfield images of different FOVs.
   3. Use the same method to stitch a fluorescence navigation map.
2. Merge the brightfield images with the fluorescence images.
   1. In the software ImageJ, click on File | Import | Image Sequence to import the sequential brightfield images of the cell of interest.
   2. Click on Image | Stacks | Z Projection | Sum Slices to get the sum intensity projection (SIP) image of the sequential brightfield images.
   3. Click on Edit | Invert to invert the SIP image of the brightfield so that the signals of the gold nanoparticles become bright white dots.
   4. In ImageJ, click on File | Import | Image Sequence to import the sequential FM images.
   5. Click on Image | Stacks | Z Projection | Sum Slices to get the SIP image of the sequential FM images.
   6. Click on Image | Color | Merge Channels to merge the SIP brightfield image with the SIP FM image as a composite image.
   7. The gold nanoparticles are green, and the fluorescent protein is red. Click on Image | Type | RGB Color to convert the format of the composite image into RGB.

8. EM imaging

1. Put the grid onto the sample hold, keeping the section side of the grid upward.
2. Use the navigation map to roughly identify the position of corresponding cells under an electron microscope through four relative distances to the four different corners of the right trapezoid of the ultrathin section.
3. Confirm the cells using the gold nanoparticles.
4. Take EM images at different magnifications (4,200x, 6,000x, or 8,200x) with a transmission electron microscope (TEM) (see Table of Materials).

9. Registration of the FM image with the EM image

1. Open the registration software (see Table of Materials). Type easyCLEMv0¹⁴ in the search window. Click on easyCLEMv0 to open easyCLEMv0. Click on Image/Sequence | Open to import the EM image and the composite image into the panel of the software.
2. In the EasyCLEMv0 window, click on 2D (x,y,[T]) to choose non rigid (2D or 3D) as the alignment mode.
3. In the EasyCLEMv0 window, click on the dropdown box to the right of Select image that will be transformed and resized (likely FM) to choose Composite (RGB)color.tif. Then, click on the dropdown box to the right of Select image that will not be modified (likely EM) to choose EMimage. tif.
4. Click on Start to begin the registration.
5. In the EM image window named EMimage. tif, click on one gold nanoparticle to put Point 1 onto the gold nanoparticle. In the FM image window named Composite (RGB)color.tif, click on the corresponding gold nanoparticle to put Point 1 onto the gold nanoparticle.
6. Click on Update Transformation to confirm the LM signal with the EM signal of the gold nanoparticles.
7. Repeat the above operation in the EM image window named EMimage. tif and match the LM signal with the EM signal of the same gold nanoparticle one by one.
8. Click on Stop to finish the registration.
9. After image alignment, click on Overlayed image and click on Image/Sequence | Save as to export an image stack containing four channel images (red, green, blue, gray).
10. Open ImageJ and click on File | Open to import the Overlayed image. Click on Image | Stacks | Stack to Images to separate the Overlayed image into four channel images.
11. Click on Image | Color | Merge Channels to merge images from three channels (red, green, and gray) into a composite image.
12. Click on Image | Type | RGB Color to convert the composite image to a CLEM image.

Results

Previous reports demonstrated that mScarlet can target the lysosome¹⁵. In this protocol, AAV expressing mScarlet (rAAV-hSyn-DIO-mScarlet-WPRE-pA) was injected into the M1 (ML: ±1.2 AP: +1.3 DV: -1.5) of Vglut2-ires-cre mouse brain using stereotaxic instruments. Following the protocol described above, the final correlated image is shown in Figure 4A. The FM image can be accurately aligned with the EM image using gold nanoparticles (the green dots) as fiducial markers. As shown in the EM image (Figure 4C,F), mScarlet targeted the lysosome, and the contents inside lysosomes are heterogeneous, which are the typical characteristics of the secondary lysosome.

Figure 1: Schematic overview of correlative light and electron microscopy. (A) Mouse brain preparation. Adeno-associated viruses were injected into the mouse brain. After 30 days, the fluorescent regions of mouse brain slices were cut into small blocks for EM sample preparation. (B) EM sample preparation. (C) Ultrathin section using a diamond knife and the schematic diagram of the coverglass. (D) FM imaging and TEM imaging. Abbreviations: EM = electron microscopy; FM = fluorescence microscopy; TEM = transmission electron microscopy. Please click here to view a larger version of this figure.

Figure 2: Coating coverglasses with gold nanoparticles. (A) Cover the coverglass with pioloform by centrifugation. (B) Incubate the coverglass surface with poly-L-lysine. (C) Incubate the coverglass surface with the diluted gold nanoparticles. (D) Schematic diagram of the coated coverglass. Please click here to view a larger version of this figure.

Figure 3: The schedule of preparations for FM imaging and EM imaging. (A) The workflow of the preparations for FM imaging consists of five steps: (i) scooping the ultrathin section, (ii) Air drying, (iii) placing a second coverglass after adding the buffer, (iv) fixing both coverglasses in the chamber, and FM imaging. (B) The workflow of the preparations for EM imaging also consists of five steps: (i) separating the two coverglasses, (ii) floating the ultrathin section, (iii) scooping up the ultrathin section, (iv) moving the ultrathin section onto the slot grid, and (v) EM imaging. Abbreviations: EM = electron microscopy; FM = fluorescence microscopy; TEM = transmission electron microscopy. Please click here to view a larger version of this figure.

Figure 4: Representative CLEM image of mScarlet. (A) The CLEM image of the whole neuron. (B,E) The CLEM images of the inset. (C,F) The EM images of the inset. (D,G) The fluorescence images of the inset. Green dots indicate gold nanoparticles. Scale bar = 5 µm (A), 1 µm (inset images). Abbreviations: EM = electron microscopy; FM = fluorescence microscopy; CLEM = correlative light and electron microscopy. Please click here to view a larger version of this figure.

Discussion

The protocol presented here is a versatile imaging method, which can combine the localization information of the target protein from light microscopy (LM) and the context surrounding the target protein from electron microscopy (EM)⁶. With the limitations of current fluorescent proteins, the widely used method is pre embedding correlative light and electron microscopy (CLEM), which means the LM imaging is done before the EM sample preparation. Almost all existing fluorescent proteins can be examined in pre embedding CLEM. However, because of the inevitable distortions and shrinkages, accurate alignment of the final image is impossible⁶. Therefore, the information provided by pre embedding CLEM is to check the ultrastructures of the same cell imaged by LM in EM imaging.

The method provided by this study is post-embedding CLEM, in which LM imaging is done after EM sample preparation. The shrinkage caused by the chemical fixation in EM sample preparation does not affect the final image's accurate alignment. Furthermore, because both LM imaging and EM imaging are done on the same section and with the help of gold nanoparticles, the final image alignment can be very accurate. Post-embedding CLEM requires that the fluorescent proteins should retain a fluorescent signal after conventional EM sample preparation. Previously, we had reported the first fluorescent protein called mEosEM, which can retain fluorescent signals using Epon as embedding resin⁸. Compared with other resins as embedding resins, Epon has superior ultrastructure preservation and sectioning properties. Following mEosEM, other resistant Epon-embedding fluorescent proteins had been reported, such as mKate2^11,16, mCherry2^10,17, mWasabi^10,18, CoGFPv0^10,19, mEosEM-E⁸, mScarlet^8,20, mScarlet-I^8,20, mScarlet-H^8,20, and HfYFP⁹.

According to our experience, mScarlet is superior to other fluorescent proteins. Fixative solution can affect the fluorescence of fluorescent proteins. In general, the fixation proceeding speed of paraformaldehyde (PFA) is much slower than that of glutaraldehyde (GA); conversely, PFA penetrates the sample more quickly than the larger GA. A mixture of PFA and GA provides a balance between fixing the sample quickly enough that its quality is maintained but slowly enough that sample damage such as oxidation does not occur. The higher GA concentration will produce higher autofluorescence. From our experience, the fluorescence of mScarlet in resin block can be detected 6 months after polymerization. But we recommend to do CLEM imaging immediately after polymerization.

Another key factor in post embedding CLEM is how to register the LM image with the EM image accurately. Based on previously reported methods^6,21, we made some modifications to the protocol. The first is how to find the same cell imaged by LM in EM imaging. We took different FOVs to stitch the navigation map of the whole ultrathin section using DIC and fluorescence imaging mode. Using the navigation map, the cells of interest could be easily identified. Another improvement is accurate image alignment. Previously, we used the fluorescent signal in fluorescence imaging mode and a high electron contrast signal in the EM imaging mode of the gold nanoparticles as the fiducial alignment markers⁶. However, when using mScarlet, the fluorescent signal of mScarlet was much higher than the fluorescent signal of the gold nanoparticles and it was difficult to detect the fluorescent signal of gold nanoparticles. To solve this problem, we used the brightfield signal of the gold nanoparticles for registration instead of the fluorescent signal. Following this modified protocol, it was easy to perform post-embedding CLEM.

However, there are some limitations of the current protocol, which should be taken into consideration before using it. Although it works well in the overexpression system, the fluorescence is quite faint if the target proteins are under their native promoter²². Another limitation is that although mScarlet is the best fluorescent protein (according to our experience) that can retain enough fluorescence signals after EM sample preparation and works well in mammalian cells, it will be a problem when using mScarlet as a fluorescent tag in neurons, which can lead to the mistaken location of the target proteins¹⁵. In these cases, we suggest using oScarlet¹⁵, a mutation of mScarlet, which works well in neurons.

The potential applications of this modified protocol can be divided into two categories. The first is to zoom into the subcellular level, which means studying the target protein in the subcellular context. In our previously published reports, we used post-embedding CLEM to examine the formation of cytoplasmic virion assembly compartments (cVACs) during infection by a γ-herpesvirus²³. In future applications, post embedding CLEM can be combined with electron tomography to obtain the 3D distribution of the target proteins and solve the 3D structure natively²⁴. The second type of potential application is to zoom out to the cellular level, which can be used to draw and analyze neural circuits. Due to its high-resolution capability, EM has become an effective means of mapping fine brain connections. However, EM cannot accurately provide the identities of neurons, which limits the in-depth analysis of the neuronal circuit. Epon post-embedding CLEM has the potential to bring the identities of neurons into the neuronal circuit without compromising the ultrastructures.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
0.2 M Phosphate Buffer (PB)			NaH2PO4 · 2H2O+Na2HPO4 · 12H2O
0.2 M Tris-Cl (pH 8.5)	Shanghai yuanye Bio-Technology	R26284
25% Glutaraldehyde (GA)	Alfa Aesar	A17876	Hazardous chemical
Abbelight 3D	Nanolnsights
Acetone	SCR	10000418
Ammonium hydroxide	J&K Scientific	335213
BioPhotometer D30	eppendorf
Cleaning buffer of cover glasses			50 mL Ammonium hydroxide, 50 mL Hydrogen peroxide, 250 mL H2O
Coverglass	Warner	64-0715
DABCO	Sigma	290734	Hazardous chemical
DDSA	SPI company	GS02827	Hazardous chemical
Desktop centrifuge	WIGGENS	MINICEN 10E
Diamond knife	DiATOME	MX6353
DMP-30	SPI company	GS02823	Hazardous chemical
DNA transfection reagent	Thermo Fisher	2696953	Lipofectamine 3000 Transfection Kit
Epon 812	SPI company	GS02659	Hazardous chemical
Ethanol	SCR	10009218
Fiji image J	National Institutes of Health
Fixative solution			4% PFA+0.25% GA+0.02 M PB
Formvar	Sigma	9823
Glycerol	SCR	10010618
Gold nanoparticles	Corpuscular	790120-010
Gradient resin			Acetone to resin 3:1, 1:1, 1:3
Hydrofluoric acid	SCR	10011118
Hydrogen peroxide	SCR	10011218
ICY (https://icy.bioimageanalysis.org/about/)			Easy CLEMv0 Plugin
Imaging chamber	Thermo Fisher	A7816
Large gelatin capsules	Electron Microscopy Sciences	70117
Mounting buffer			Mowiol 4-88, Glycerol, 0.2 M Tris-Cl (pH 8.5), DABCO
Mowiol 4-88	Sigma	9002-89-5
Na2HPO4 ž12H2O	SCR	10020318
NaH2PO4 ž2H2O	SCR	20040718
NMA	SPI company	GS02828	Hazardous chemical
Oligonucleotide primers	Takara Biomedical Technology (Beijing)		Three oligonucleotides primers were used to detect Vglut2-ires-Cre and wild-type simultaneously. The primers 5,-ATCGACCGGTAATGCAGGCAA-3, and 5,-CGGTACCACCAAATCTTACGG-3, aimed to detect Vglut2-ires-Cre. The primers  5,-CGGTACCACCAAATCTTACGG-3, and 5,-CATGGTCTGTTTTGAATTCAG-3, aimed to detect wild-type.
Oscillating microtome	Leica	VT1000S
Osmium tetroxide	SCR	L01210302	Hazardous chemical
OsO4 solution			1% Osmium tetroxide+1.5% K4Fe (CN)6·3H2O
Parafilm	Amcor	PM-996
Paraformaldehyde (PFA)	SCR	80096618	Hazardous chemical
Perfusion buffer			4% PFA+0.1 M PB
Pioloform	Sigma	63148-65-2	Hazardous chemical
Poly-L-lysine	Sigma	25986-63-0
Potassium ferrocyanide (K4Fe (CN)6·3H2O)	SCR	10016818
Scalpel blades	Merck	S2771
Scalpel handles	Merck	S2896-1EA
Stereomicroscope	OLYMPUS	MVX10
Transgenic mice	The Jackson Laboratory		Vglut2-ires-Cre mice (strain: 129S6/SvEvTac) were housed in standard conditions (25 °C, a 12 h light/dark cycle, with water and food given ad libitum. Male and Female mice were used at 2–3 months old, weight range 20-30 g.
Transmission electron microscope (TEM)	FEI		TECNAL G2
UA solution (2% UA)			Aqueous solution
Ultramicrotome	Leica	LEICA EM UC6
Uranyl acetate (UA)	TED PELLA	19481	Hazardous chemical