We developed a novel technique in electron microscopy, "flash-and-freeze," that enables the visualization of membrane dynamics with ms temporal resolution. This technique combines the optogenetic stimulation of neurons with high-pressure freezing. Here, we demonstrate the procedures and describe the protocols in detail.
Cells constantly change their membrane architecture and protein distribution, but it is extremely difficult to visualize these events at a temporal and spatial resolution on the order of ms and nm, respectively. We have developed a time-resolved electron microscopy technique, "flash-and-freeze," that induces cellular events with optogenetics and visualizes the resulting membrane dynamics by freezing cells at defined time points after stimulation. To demonstrate this technique, we expressed channelrhodopsin, a light-sensitive cation channel, in mouse hippocampal neurons. A flash of light stimulates neuronal activity and induces neurotransmitter release from synaptic terminals through the fusion of synaptic vesicles. The optogenetic stimulation of neurons is coupled with high-pressure freezing to follow morphological changes during synaptic transmission. Using a commercial instrument, we captured the fusion of synaptic vesicles and the recovery of the synaptic vesicle membrane. To visualize the sequence of events, large datasets were generated and analyzed blindly, since morphological changes were followed in different cells over time. Nevertheless, flash-and-freeze allows the visualization of membrane dynamics in electron micrographs with ms temporal resolution.
Visualizing membrane and protein dynamics within a cell is a key step towards understanding the cell biology of particular processes. Dynamic trafficking events can be captured using light or fluorescence microscopy. However, the subcellular context is largely missing in such images because subcellular structures cannot be completely "painted" by dyes or fluorescent probes and resolved spatially and spectrally1,2. On the other hand, while electron microscopy can delineate subcellular architecture in exquisite detail, it cannot capture cellular dynamics, because specimens must be fixed prior to imaging. Thus, it is typically not sufficient to completely understand cellular dynamics using only one imaging modality.
To overcome the limitations of light and electron microscopy, correlative microscopy techniques have been developed. Correlative Light and Electron Microscopy (CLEM) visualizes intracellular dynamics using light microscopy and underlying subcellular structures with electron microscopy. In CLEM, cells engaged in various processes, such as cytokinesis and endocytosis3,4,5,6, are live-imaged and then processed for electron microscopy. Although CLEM captures certain aspects of intracellular dynamics, there are four factors that limit the utility of this approach. First, the temporal resolution is limited by how quickly the cells can be immobilized, which typically takes s – min due to the slow diffusion and reaction of fixatives7. Second, the subcellular architecture is observed post facto8; thus, the dynamic morphological changes cannot be captured using this approach. Third, the fluorescence and electron micrographs cannot be precisely aligned due to tissue shrinkage caused by dehydration during the sample preparation for electron microscopy9,10. Fourth, events like cytokinesis and endocytosis do not take place at the same time in every cell5,11, and thus, a particular cell that is engaged in the event must be identified from a large population of cells. This process is often laborious. Thus, a new method is necessary to induce particular events in every cell and to capture the resulting cellular dynamics by the rapid immobilization of cells at defined time points.
Recently, several tools have been developed to induce particular cellular dynamics using light (optogenetics). Channelrhodopsin is a light-sensitive, non-selective cation channel isolated from Chlamydomonas reinhardtii12,13. When channelrhodopsin is expressed in neuronal membranes, a brief flash of light induces an influx of sodium ions into neurons and triggers an action potential14,15. The action potential then propagates into the synaptic terminals, where synaptic vesicles fuse within milliseconds16,17,18 Therefore, channelrhodopsin induces neuronal activity. To follow membrane dynamics at synaptic terminals, neurons must be immobilized at defined time points after stimulation with ms precision.
To capture membrane dynamics after inducing neuronal activity, we coupled light stimulation with high-pressure freezing17,18,19. High-pressure freezing allows for the near-instantaneous immobilization of cells with reduced ice crystal formation20. Ice crystals can rupture membranes and disrupt the subcellular architecture21. By varying the time intervals between stimulation and freezing, membrane trafficking within synaptic terminals was captured following the induction of an action potential.
Here, we demonstrate experimental procedures using a commercialized high-pressure freezer that couples a ms temporal control of light stimulation with high-pressure freezing. Unlike other instruments that require an external device to control light stimulation and freezing, light stimulation is fully integrated in this system and can be applied with ms precision19. This process involves multiple steps. 1) Mouse hippocampal neurons are cultured on sapphire disks and infected with lentivirus carrying an expression vector for channelrhodopsin18. 2) Neurons are stimulated and frozen at defined time points after stimulation. 3) The vitrified water is substituted with an organic solvent, while lipids and proteins are cross-linked by fixatives to preserve the intracellular architecture. 4) The samples are infiltrated and embedded in epoxy resin. 5) Ultrathin sections are collected using an ultramicrotome. 6) Thin sections are imaged on a transmission electron microscope. 7) Image acquisition and analysis are performed blindly with respect to time points or genotypes. Cellular dynamics can be determined through the reconstruction of time-resolved images17,18. Sample preparation (steps 2 – 5 above) requires a week, but the subsequent image analysis requires months to a year.
All of the experiments were performed according to the rules and regulations of animal use by the National Institutes of Health. The protocol was approved by the Animal Care and Use Committee (IACUC) at Johns Hopkins School of Medicine.
1. Isolation and Culture of Mouse Hippocampal Neurons
2. Flash-and-Freeze
3. Freeze Substitution in the Automated Freeze-substitution Unit
4. Infiltration and Plastic Embedding with Epoxy Resin
5. Mounting Samples
6. Sectioning
7. Imaging Using a Transmission Electron Microscope (TEM)
8. Image Analysis
Using the protocol described above, we performed "flash-and-freeze" experiments in mouse hippocampal neurons expressing channelrhodopsin. These neurons were frozen either 15 ms or 100 ms after light onset. We have previously shown that the exocytosis and endocytosis of synaptic vesicles occur in the synaptic terminals at the 15 ms and 100 ms time points, respectively18. These events were successfully captured at the appropriate times (Figure 2), suggesting that flash-and-freeze experiments can be successfully performed on the chosen specialized high-pressure freezer (see the Table of Materials).
Figure 1. Sample Loading and Programming in the High-pressure Freezer. A) Sample loading table of a high-pressure freezer. The middle plate, shown in the inset for structural detail, is placed in a CLEM holder for sample loading. One of the half-cylinders is placed at the bottom part of the sample loading table, and the other is attached with a clip to the top cover. Once the sample is loaded, the middle plate is pushed forward to the bottom half-cylinder and the cover is closed to initiate the freezing. B) Specimen assembly. The sapphire disk containing neurons is placed in the well of the middle plate, with the cell-side facing up. A 100 µm ring is placed directly above the sapphire disk inside the well. Then, an empty sapphire disk dipped in physiological saline is placed with the solution-side down. Air bubbles must be avoided. Finally, a 100 µm ring and a 400 µm ring are snugly placed above. Any extra liquid is removed with filter paper. C) A cross-section of an embedding capsule with the sapphire disk submerged in epoxy resin. The sapphire disk is placed at the bottom of the capsule, with the cell-side facing up and covered with epoxy resin for infiltration and embedding. D) Programming the high-pressure freezer for a single, 10 ms stimulus. The specimens are frozen 90 ms after the light pulse. E) Programming the high-pressure freezer for 10 stimuli at 20 Hz. The specimens are frozen 5 ms after the last light pulse. Please click here to view a larger version of this figure.
Figure 2. Visualization of Exocytosis and Endocytosis in Mouse Hippocampal Neurons. Hippocampal neurons are stimulated once and frozen at the indicated times. Electron micrographs show the exocytosis of a synaptic vesicle A) and ultrafast endocytosis B). PSD, post-synaptic Density. Please click here to view a larger version of this figure.
The "flash-and-freeze" approach visualizes membrane dynamics by inducing a particular cellular event with optogenetics and by freezing cells at defined time points after stimulation19. In this demonstration, we used channelrhodopsin, a light-sensitive cation channel, to stimulate neurons and captured the fusion and recovery of synaptic vesicles at the synaptic terminals. In recent years, many optogenetic tools have been developed22,23, all of which are compatible with flash-and-freeze. For example, organelle trafficking can be induced using light-induced heterodimerization of cryptochrome and CIB124. Similarly, the lipid composition of the plasma membrane can be altered by the light-induced translocation of phosphoinositide phosphatases to the plasma membrane25. Furthermore, small, light-sensitive compounds like azobenzene change conformation depending on the illumination wavelengths. This conformational change can be used to activate ligand-gated channels or to change lipid composition in the membrane26,27. Caged compounds can also be used to induce cellular activity. However, the LED used in the current setup may not produce sufficient energy for uncaging; thus, further optimizations of the system are likely necessary. Nevertheless, the applications of these light-activatable tools are flexible-many cellular events can be induced by a flash of light. "Flash-and-freeze" can capture the resulting membrane dynamics.
There are two main limitations to the "flash-and-freeze" method. First, it captures "snapshots" of a particular event from different cells. In other words, it is not possible to follow membrane dynamics in one cell over a period of time. Thus, for the reconstruction of any cellular event, one must acquire and analyze a large number of images from each sample and at each time point. Furthermore, in neurons, an even larger number of images is necessary, since the fusion of synaptic vesicles only takes places in 20 – 30% of the synapses in mouse hippocampal neurons18,28. The analysis of such a large dataset requires tremendous amounts of time. In the future, image acquisition and analysis need to be automated to make the approach more efficient29,30.
The second limitation is imposed by the nature of the high-pressure freezing technique. When cells freeze, cellular water rearranges to form ice crystals if the freezing rate is below 100 K/s21. These ice crystals can penetrate membranes or concentrate solutes to alter local osmotic pressure, resulting in the rupture of membranes. To avoid ice crystals, high pressure (~2,000 atm) is applied to the specimens. Due to the super-cooling effect, a freezing rate of 100 K/s is sufficient to prevent water from forming ice crystals at this pressure21. In theory, specimens as thick as 500 µm can be frozen without ice crystals, but approximately 200 µm is likely the practical limit, as cuboidal forms of ice tend to form in thick tissue, compromising morphology. When processing specimens thicker than 5 µm, the use of a proper cryo-protectant, such as BSA, is necessary. However, BSA will change the osmolarity of the solution and may affect the physiological response of cells. Therefore, extensive control experiments are required to validate the use of BSA in particular systems. Ice crystals can also form after high-pressure freezing if the specimens are accidently removed from the liquid nitrogen bath. Thus, it is critical to keep the specimens in the liquid nitrogen at all times and to use pre-cooled forceps to manipulate them.
When planning experiments, the following three points should be considered. First, the maximal intensity of light (the 460 nm line) is 5.5 – 8.0 mW/mm2. Whether this intensity is sufficient to induce activity must be verified with live-cell imaging on a fluorescence microscope prior to flash-and-freeze experiments. Second, experiments must be performed at physiological temperature. The stage of the high-pressure freezer is warmed up to 37 °C for the experiments with mouse hippocampal neurons31. Finally, the time points must be carefully chosen to capture the membrane dynamics. Initial studies indicated that endocytosis is complete after 100 ms of stimulation. Thus, three additional time points (15 , 30, & 50 ms) were also examined to follow the membrane dynamics17,18. These time points were necessary to visualize membrane trafficking events during synaptic transmission. However, the requirement for the number of time points are different in each cellular event. Therefore, a few time points should be sampled before initiating large dataset collection.
The authors have nothing to disclose.
This work was supported by funding from the Johns Hopkins University (SW). We thank the Johns Hopkins School of Medicine Microscope Facility for their technical support. We thank Erik Jorgensen and Christian Rosenmund and members of their laboratories for the development of the technique. We thank M. Wayne Davis for the design of the initial device. We thank Paul Wurzinger, Cveta Tomova, and Delgermaa Luvsanjav for their technical assistance. We also thank Natalie R. Hamilton and Grant F. Kusick for their critical reading of the manuscript.
Freeze Substitution and Low Temperature Embedding System for Light and Electron Microscopy -AFS II | Leica | ||
High Pressure Freezer – EM-ICE | Leica | EM ICE is a specialized high pressure freezer that allows precise control of light stimulation and freezing. | |
Osmometer | Gonotec | ||
Ultramicrotome UC7 | Leica | ||
Oven | Blue M | ||
Razor blade | Personna | ||
Glutaraldehyde | EMS | 16530 | |
Osmium tetroxide | EMS | RT19132 | Toxic, open only under certified chemical hood |
Acetone | EMS | RT10016 | |
HEPES | Emdmillipore | 391340-250GM | |
Glucose | Sigma | 49159-1KG | |
KCl | Sigma | 746436-1KG | |
NaCl | Sigma | S7653-1KG | |
CaCl2 | Sigma | 21115-250ML | |
MgCl2 | J.T.Baker | 2444-01 | |
Liquid epoxy resin Eponate 12 | Ted Pella | 18028 | |
Bisphenol A epoxy resin Araldite 502 | Ted Pella | 18028 | |
Dodecenyl succinic anhydride (DDSA) | Ted Pella | 18028 | |
Benzyl dimethyl amine (BDMA) | Ted Pella | 18241 | |
Special embedding (BEEM) capsule | EMS | 70021 | |
Copper Grid | Ted Pella | 1GC12H | |
Polyvinyl acetate (Pioloform F) | Ted Pella | 19244 | |
Uranyl acetate | Polysciences | 21447-25 | |
Ethyl cyanoacrylate (Super glue) | Scotch | 170497 | |
Trypsin-EDTA | Themo scientific | 25300-120 | |
DMEM | Thermo scientific | 10569-044 | Should be warmed at 37°C before use |
FBS | Thermo scientific | 26140-079 | |
Pen-Strep | Thermo scientific | 15140-122 | |
Fluoro-deoxyuridine (FUDR) | Sigma | F0503 | |
Glass cover slip | Fisher | S175223 | Should be acid-washed |
Sapphire disc | Technotrade | 616-100 | |
Acetic acid | Emdmillipore | 1000631011 | |
Poly-D-Lysine | Sigma | P6407 | |
Rat tail collagen | Thermo scientific | A10483-01 | |
Neurobasal A | Fisher | 10888022 | Should be warmed at 37°C before use |
L-alanyl-L-glutamine (Glutamax) | Fisher | 35050-079 | |
Seraum free supplement (B-27) | Fisher | 17504044 | |
Hanks Balanced Salt Solution (HBSS) | Thermo scientific | 24020117 | |
Papain | Worthington | LS003126 | Active Unit should be calculated for each batch |
Thermomixer | Eppendorf | Thermomixer C | |
Trypsin inhibitor | Sigma | T9253 | |
NBQX | Tocris | 03-731-0 | |
Bicculine | Tocris | 01-091-0 | |
Whatman I filter paper | GE | ||
Transmission Electron Microscope | Philips CM20 |