This research was supported by Natural Science Foundation of Zhejiang Province grants (Z23H290001, LY19H280001); National Natural Science Foundation of China grants (82274364, 81673607, and 81774011); as well as the Public Welfare Research Project of Huzhou Science and Technology Grant (2021GY49, 2018GZ24). We appreciate the great help, technical support, and experimental support from the Public Platform of Medical Research Center, Academy of Chinese Medical Science, Zhejiang Chinese Medical University.

1. Material preparation
<ol>
	<li>Culture 2 x 106 Panc02 cells in 12 mL of DMEM medium (10% fetal bovine serum and 100 U/mL of penicillin-streptomycin), and maintain at 37 &#176;C and 95% humidity in an atmosphere of 5% carbon dioxide and 95% air for 48 h.</li>
	<li>Collect Panc02 cells, centrifuge at 28 x g for 2 min, and then discard the supernatant. Ensure that the sample is of an appropriate size (1 x 107 cells), as otherwise, the following fixation and dehydration steps will not work well.</li>
	<li>Add 1 mL of 2.5% glutaraldehyde as a fixative to fix the fresh Panc02 cells in the 1.5 mL micro-centrifuge tubes overnight at 4 &#176;C. 
	NOTE: The samples need to be centrifuged at 1,006 x g for 5 min in each of the following steps (steps 1.3-1.11).</li>
	<li>Aspirate the fixative, and rinse the samples twice with 0.1 M phosphate-buffered saline (PBS), and then rinse twice with double-distilled water (ddH2O) at room temperature for 10 min each.</li>
	<li>Add 50 &#181;L of a solution containing 1% osmium tetroxide (OsO4) and 1.5% potassium ferrocyanide at a ratio of 1:1 at 4 &#176;C for 1 h. Then, rinse twice with 0.1 M PBS for 10 min each time, and rinse with ddH2O for another 10 min.</li>
	<li>Add 1 mL of 1% thiocarbohydrazide (TCH), incubate at room temperature for 30 min to 1 h, and then rinse with ddH2O four times for 10 min each time.</li>
	<li>Add 50 &#181;L of 1% OsO4, fix at room temperature for 1 h, and then rinse with ddH2O four times for 10 min each time.</li>
	<li>Add 1 mL of 2% uranyl acetate at 4 &#176;C overnight, and then rinse with ddH2O four times for 10 min each time.</li>
	<li>Add 1 mL of Walton solution (0.066 g of lead nitrate dissolved in 10 mL of 0.03 mol/L aspartic acid stock solution) at 60 &#176;C for an additional 1 h of incubation.</li>
	<li>Rinse with ddH2O four times for 10 min each time, and then dehydrate in a graded alcohol series for 10 min each time: 50% alcohol, 70% alcohol, 80% alcohol, and 90% alcohol 1x, and 100% alcohol 2x. Then, dehydrate in 100% acetone twice for 10 min each time. Use 1 mL of each solution for the dehydration.</li>
	<li>Mix 200 &#181;L of acetone with Pon 812 epoxy resin at a ratio of 3:1, 1:1, and 1:3. Soak the sample in resin at room temperature for 2 h, 4 h, and 4 h, respectively, and then take the 100% Pon 812 epoxy resin to impregnate overnight. 
	NOTE: When performing the staining and resin embedding steps, it is important to operate in a fume hood as these are toxic substances. Additionally, lab coats and solvent-resistant gloves must be worn during the experiment.</li>
	<li>Put the specimens into an embedding mold, and then place in an oven at 60 &#176;C for 48 h to polymerize. Use flat embedding29 to reduce the appearance of resin around the sample. This simplifies the specimen installation and decreases the impact of charging artifacts on the subsequent image segmentation. 
	NOTE: To subsequently generate a horizontal cutting surface, a small amount of resin can be added into the mold hole, taking care not to overfill it.</li>
	<li>Prepare silicon wafers by first washing them and then hydrophilizing with a concentrated H2SO4/H2O2 solution for 30 min. Collect serially slice strips with a thickness of 70 nm on the hydrophilized silicon wafers by using an ultramicrotome, and dry them in a 60 &#176;C oven for 10 min. 
	NOTE: Slowly capture the slice as the slice floats to the surface of the wafers to prevent slice loss or deformation.</li>
</ol>
2. Image acquisition and three-dimensional reconstruction
<ol>
	<li>Before mounting a silicon wafer on the stage, wash the sections with distilled water three times, and air-dry them. Cut a conductive carbon tape with the size of 1 cm x 0.5 cm, and stick this between the silicon wafer and SEM specimen stage to complete the sample setup (Figure 2E).</li>
	<li>Set the FE-SEM accelerating voltage parameter to 2 kV with a working distance of 4 mm (Figure 3B and Table 1). 
	NOTE: In order to improve the signal-to-noise ratio and reduce the noise in the image, observe at low magnifications whenever possible.As the multiple increases, the scanning range of the SEM decreases, leading to an increase in charge accumulation on the images.</li>
	<li>Click on the H/L icon in the top menu bar. First, at low magnification, orient the first section of the target slice strips, and then click again on the H/L option (Figure 3A) to switch to high-magnification mode; collect an appropriate image for the structure of interest by adjusting the image brightness, contrast, and magnification.</li>
	<li>Select Project View &#62; Open Data, and import the image files to be analyzed into the software.</li>
	<li>Align the pictures by clicking on Align Slices &#62; Edit (Figure 4B), and adjust the Intensity Range value in the lower-left menu bar by modifying the image transparency. Here, use a semi-automatic alignment strategy; specifically, through automatic alignment, make the pictures overlap with a previous picture, and then perform manual fine-tuning. 
	NOTE: In this work, during the process of alignment, the previous image was used as a reference, and the move and rotate operations were utilized to maximally overlap the target structure of the two images. Clicking on Align Current Slice Pair can help to align the images automatically, but misplacement may occur.</li>
	<li>Select the Extract Subvolume module, and clip the aligned data sets to fit the size of the overlapping portion of the whole stack.</li>
	<li>Under the Segmentation subsection (Figure 4C), select Resample &#62; Segmentation &#62; Save. Choose the threshold of the Magic Wand and the size of Brush tool to select the correct range. Here again, adopt a semi-automatic segmentation method by first using the Magic Wand to automatically select a suitable large area, and then use the Brush to accurately describe the details. 
	NOTE: The Magic Wand can be applied to segment images based on the obvious differences in the pixel intensity existing between the structure of interest and the surrounding structures by altering the Masking value and using the Draw Limit Line. It is incapable of distinguishing the charging artifacts generated during image acquisition. Therefore, it may result in incorrect segmentation for the target regions. The Brush tool can be used to segment manually to accurately reconstruct the biological structures.</li>
	<li>Under Segmentation &#62; Selection, click on the + icon to add the selected area (Figure 4C).</li>
	<li>Repeat the above step (2.8) to select the same target structure in different images until the selection is complete for reconstructing the microstructures of interest. 
	NOTE: During the mitochondrial remodeling processes, the selection of the target object must be performed from the picture in the middle of a group of sequential images. If the required area is selected from the first or last picture, the reconstruction result will not be able to present a complete 3D visualization.</li>
	<li>After completing the regional segmentation, generate the image file according to the best results for the size of the object and the picture resolution. Click on Crop Editor, and then enter the number 6 in the Virtual Slider box (Figure 4C).</li>
	<li>In the drop-down menu on the left, right-click on the gray area under the Project subsection, and select Generate Surface &#62; Create &#62; Apply. In the generated file, use the Surface View module to create the surface structure and 3D representation.</li>
</ol>
3. Quantification
<ol>
	<li>Click on Remove Small Spots &#62; Apply (Figure 4D). Under the Project subsection, right-click on the gray area, and select Label Analysis &#62; Apply (Figure 4D).</li>
	<li>Customize the name of the column, and choose a measure group. Select Volume3d and Length3d from the Native Measurements box.</li>
	<li>Click on the Apply button in the lower-left corner of the screen. Copy the data, and graph in the statistical software, such as GraphPad.</li>
</ol>

3D Technique to Visualize Mitochondrial Cristae in Pancreatic Cancer Cells

The authors declare no conflicts of interest.

The method presented here is a useful step-by-step guide for applying the 3D reconstruction technique, which involves applying electron microscopy and image processing technology to the stacking and segmentation of 2D tomographic images generated from serial ultrathin sections. This protocol highlights a limitation of 2D images that can be addressed by 3D visualization of the organelle ultrastructure, which has the advantages of strong reproducibility of the structures at a high-resolution level and higher accuracy. More...

Mitochondria are one of the most important organelles in the cell. They serve as the central hub of cellular bioenergetics and metabolism1,2 and play a critical role in cancer3. Pancreatic cancer (PC) is one of the most difficult cancers4 to treat due to its rapid spread and high mortality rate. Mitochondrial dysfunction, which is mainly caused by changes in the mitochondrial morphology3,5,6,7, has been linked to the disease mechanisms underlying PC8. Mitochondria are also highly dynamic, which is reflected by the frequent and dynamic changes in their network connectivity and cristae structure9. The reshaping of the cristae structure can directly impact the mitochondrial function and cellular state10,11, which are significantly altered during tumor cell growth, metastasis, and tumor microenvironment changes12,13.
In recent years, scientists have studied this organelle using EM observation14,15,16,17; for example, researchers have analyzed the mitochondrial dynamics using 3D reconstruction techniques6,7,18,19. The general concept and method for the 3D reconstruction of electron microscopy images were formally established as early as 196820 and involved combining electron microscopy, electron diffraction, and computer image processing to reconstruct the T4 phage tail. Up until now, electron microscopy 3D imaging technology has made significant advancements in terms of the image resolution21, degree of automation22, and processing volume23 and has been used on an increasingly broad scale in biological research, from the tissue level to the organelle ultrastructure level at the nanometer scale24. In recent years, electron microscopy 3D imaging has also become a promising technology for a wide range of applications25,26,27.
The growing attention on mitochondrial cristae particularly illustrates the essential requirements for ultrastructural volume imaging. Transmission electron microscopy (TEM) has been used to visualize samples collected on a copper grid (400 mesh)28, with the electron beam passing through the section. However, due to the limited range of the copper grid, it is impossible to fully image continuous slices of the same sample29. This complicates the study of target structures during TEM imaging. Additionally, TEM relies on time-consuming and error-prone manual tasks, including cutting and collecting multiple slices and imaging them sequentially21, so it is not adapted for ultrastructural reconstructions of large-volume samples23. At present, the high-resolution reconstruction of large-volume sample imaging is achieved through the use of specialized equipment, such as the TEM camera array (TEMCA)30 or two second-generation TEMCA systems (TEMCA2)31, which enable automated high-throughput imaging in a short time. However, this type of imaging does not have the advantage of being easy to obtain and universal due to the requirement for customized equipment.
Compared with TEM, the method for automatically generating thousands of serial volumetric images for large areas based on SEM32,33 enhances the efficiency and reliability of serial imaging and delivers higher z-resolutions34. For example, serial block-face scanning electron microscopy (SBF-SEM) and focused ion beam scanning electron microscopy (FIB-SEM) have both made it possible to achieve the 3D reconstruction of ultrastructure with high speed, efficiency, and resolution35,36. However, it is inevitable that the block surface is mechanically shaved off by the diamond knife of the SBF-SEM or by milling with the focused ion beam of FIB-SEM33,37. Due to the destructiveness of the two methods to the samples, it is not possible to reconstruct the same target structure again for further analysis38,39,40. In addition, few studies have attempted to reconstruct the 3D organelle ultrastructure of cancer cells using EM to observe pathological changes12. For these reasons, in order to further elucidate the pathological mechanisms of cancer cells, such as pancreatic cancer cells, we propose a novel technology for the 3D reconstruction of serial section images using an ultramicrotome and a field emission scanning electron microscope (FE-SEM) to analyze the mitochondrial ultrastructure at the cristae level; with this technology, high-resolution data can be acquired using an efficient and accessible method. The serial ultrathin sections made using an ultramicrotome can be semi-permanently stored in a grid case and reimaged multiple times, even after several years41. FE-SEM is highly valued as a tool in various research fields due to its ability to provide high-resolution imaging, high magnification, and versatility42. In an attempt to display the fine structure of organelles in 3D, the technique for producing serial 2D image stacks with useful resolution using back-scattered electrons produced by FE-SEM43,44 can also be used to achieve high-throughput and multi-scale imaging of target regions or their associated structures without special equipment45. The generation of charge artifacts directly affects the quality of the acquired images, so it is particularly important to keep the dwell time short.
Thus, the present study elaborates on the experimental procedures employed in this SEM technique to reconstruct the 3D structure of mitochondrial cristae46. Specifically, we show the process developed to achieve the semi-automatic segmentation of mitochondrial regions and digitize the 3D reconstruction using Amira software, which also involves making slice samples using the conventional OTO specimen preparation method44,47, completing the section collection using ultramicrotome slicing, and acquiring sequential 2D data by FE-SEM.

<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always">
	<tbody>
		<tr>
			<td>(1S,3R)-RSL3</td>
			<td>MCE</td>
			<td>HY-100218A</td>
			<td></td>
		</tr>
		<tr>
			<td>Acetone</td>
			<td>SIGMA</td>
			<td>179124</td>
			<td></td>
		</tr>
		<tr>
			<td>Amira</td>
			<td>Visage Imaging</td>
			<td></td>
			<td></td>
		</tr>
		<tr>
			<td>Aspartic acid</td>
			<td>MCE</td>
			<td>HY-42068</td>
			<td></td>
		</tr>
		<tr>
			<td>Dulbecco&#39;s modified Eagle&#8217;s medium</td>
			<td>Gibco</td>
			<td>11995115</td>
			<td></td>
		</tr>
		<tr>
			<td>Ethanol</td>
			<td>Merck</td>
			<td>100983</td>
			<td></td>
		</tr>
		<tr>
			<td>Ferrostatin-1</td>
			<td>MCE</td>
			<td>HY-100579</td>
			<td></td>
		</tr>
		<tr>
			<td>Fetal bovine serum</td>
			<td>Gibco</td>
			<td>10437010</td>
			<td></td>
		</tr>
		<tr>
			<td>Field emission scanning electron microscope</td>
			<td>HITACHI</td>
			<td>SU8010</td>
			<td></td>
		</tr>
		<tr>
			<td>Glutaraldehyde</td>
			<td>Alfa Aesar</td>
			<td>A10500.22</td>
			<td></td>
		</tr>
		<tr>
			<td>Lead nitrate</td>
			<td>SANTA CRUZ</td>
			<td>sc-211724</td>
			<td></td>
		</tr>
		<tr>
			<td>Osmium Tetroxide</td>
			<td>SANTA CRUZ</td>
			<td>sc-206008B</td>
			<td></td>
		</tr>
		<tr>
			<td>Panc02</td>
			<td>European Collection of Authenticated Cell Cultures&#160;</td>
			<td>98102213</td>
			<td></td>
		</tr>
		<tr>
			<td>Penicillin-streptomycin</td>
			<td>Biosharp</td>
			<td>BL505A</td>
			<td></td>
		</tr>
		<tr>
			<td>Phosphate Buffered Saline</td>
			<td>Biosharp</td>
			<td>BL302A</td>
			<td></td>
		</tr>
		<tr>
			<td>Pon 812 Epoxy resin</td>
			<td>SPI CHEM</td>
			<td>GS02660</td>
			<td></td>
		</tr>
		<tr>
			<td>Potassium ferrocyanide</td>
			<td>Macklin</td>
			<td>P816416</td>
			<td></td>
		</tr>
		<tr>
			<td>Thiocarbohydrazide</td>
			<td>Merck</td>
			<td>223220</td>
			<td></td>
		</tr>
		<tr>
			<td>Ultramicrotome</td>
			<td>LEICA</td>
			<td>EMUC7</td>
			<td></td>
		</tr>
		<tr>
			<td>Uranyl Acetate</td>
			<td>RHAWN</td>
			<td>R032929</td>
			<td>2020.2</td>
		</tr>
	</tbody>
</table>

a three-dimensional technique for the visualization of mitochondrial ultrastructural changes in pancreatic cancer cells

Understanding the dynamic features of the cell organelle ultrastructure, which is not only rich in unknown information but also sophisticated from a three-dimensional (3D) perspective, is critical for mechanistic studies. Electron microscopy (EM) offers good imaging depth and allows for the reconstruction of high-resolution image stacks to investigate the ultrastructural morphology of cellular organelles even at the nanometer scale; therefore, 3D reconstruction is gaining importance due to its incomparable advantages. Scanning electron microscopy (SEM) provides a high-throughput image acquisition technology that allows for reconstructing large structures in 3D from the same region of interest in consecutive slices. Therefore, the application of SEM in large-scale 3D reconstruction to restore the true 3D ultrastructure of organelles is becoming increasingly common. In this protocol, we suggest a combination of serial ultrathin section and 3D reconstruction techniques to study mitochondrial cristae in pancreatic cancer cells. The details of how these techniques are performed are described in this protocol in a step-by-step manner, including the osmium-thiocarbohydrazide-osmium (OTO) method, the serial ultrathin section imaging, and the visualization display.

Author Spotlight: A Three-Dimensional Technique for the Visualization of Mitochondrial Ultrastructural Changes in Pancreatic Cancer Cells  

A Three-Dimensional Technique for the Visualization of Mitochondrial Ultrastructural Changes in Pancreatic Cancer Cells

During cell culture (Figure 1A), we first divided the pancreatic cancer cells into a control group cultured with complete culture medium, a (1S,3R)-RSL348 (RSL3, a ferroptosis activator, 100 nM) group, and an RSL3 (100 nM) plus ferrostatin-149 (Fer-1, a ferroptosis inhibitor, 100 nM) group. Through the above experimental steps, the scanning electron microscope acquired 38 (Supplementary&#160;Figure 1), 43 (Supplementary...

Watch this Scientific Journal Video about A Three-Dimensional Technique for the Visualization of Mitochondrial Ultrastructural Changes in Pancreatic Cancer Cells at JoVE.com

This protocol describes how to reconstruct mitochondrial cristae to achieve 3D imaging with high accuracy, high resolution, and high throughput.

Understanding the dynamic features of the cell organelle ultrastructure, which is not only rich in unknown information but also sophisticated from a ...

a-three-dimensional-technique-for-visualization-mitochondrial

Research

JoVE Journal

Cancer Research

1.2K Views.  School of Pharmaceutical Science of Zhejiang Chinese Medical University. This protocol describes how to reconstruct mitochondrial cristae to achieve 3D imaging with high accuracy, high resolution, and high throughput.