Lighting Up the Pathways to Caspase Activation Using Bimolecular Fluorescence Complementation

This protocol discusses the use of split fluorescent proteins to measure caspase induced proximity. Split Venus was chosen for this technique because it is very bright, highly photostable and the refolding is fast ¹³. Thus, the analysis of Venus refolding upon caspase induced proximity can provide close to real-time estimations of caspase protein interaction dynamics. Venus is split into two slightly overlapping fragments, the N terminus of Venus (Venus-N or VN) comprising amino acids 1-173 and the C terminal fragment (Venus-C or VC) comprising residues 155-239. Other split fluorescent proteins exist but some, such as split mCherry, require incubation at 4 °C for up to 2 h for refolding ²¹. Additional split fluorescent proteins including split Cerulean (a version of cyan fluorescent protein) and the far red split mLumin have yet to be tested in this context ^13,22. Ultimately, if these fluorescent proteins prove suitable for measuring caspase induced proximity, this could allow for the simultaneous evaluation of multiple caspases.

Generally, the prodomain (Pro) region of the caspase is fused to the split fluorescent fragments. This region contains the minimal portion of the caspase required to bind to the activation platform and encompasses the CARD or DED interaction motif. The advantage of using the prodomain is that it does not introduce any enzymatic activity to the cells. This allows for the simultaneous monitoring of downstream events due to the endogenous caspase. Research to date has shown little difference in the kinetics and localization of caspase BiFC between the prodomain and the full length caspase ¹⁴. However, the full-length constructs tend to express with lower efficiency, necessitating the transfection of increased amounts of plasmid DNA. In addition, it is recommended to mutate the catalytic cysteine to a serine to prevent caspase-induced apoptosis upon expression. Despite these drawbacks, certain investigations may require the concurrent analysis of the full length BiFC pairs to determine if there are context-dependent differences in caspase induced proximity in the presence or absence of the catalytic domains.

This protocol is specifically adapted for HeLa cells. The same protocol works well for additional adherent cell lines including the mammary carcinoma cell line, MCF-7, the glioblastoma cell line, LN18, and mouse embryonic fibroblasts (MEF). It is recommended that the user optimizes both plating and transfection conditions when adapting this protocol to a new cell type. The protocol describes two main microscopy approaches that can be used to evaluate the cells: epifluorescence and confocal microscopy. When using confocal microscopy, cells must be plated on glass coverslips because the majority of high numerical aperture objectives that are used in conjunction with confocal microscopy will not penetrate the thickness of plastic dishes or glass slides. Thus, it is recommended to use dishes with embedded no. 1.5 glass coverslips that have the optimal coverslip thickness of 0.17 mm. A number of different types of these dishes are available that range from individual 3 cm dishes, to chamber slides, to standard size multi-well plates. Tables 1 and 2 outline how to adapt the plating and transfection instructions for the different well sizes. For the epifluorescence imaging that is described here (4.1), standard plastic tissue culture plates are sufficient as long as the microscope is equipped with long pass objectives.

Because this approach is effectively a "lights on" method to measuring caspase activation, the inclusion of a fluorescent reporter is essential both to allow visualization of the cells prior to or in the absence of activation of the caspase reporter and to identify the transfected cells, when transient transfection is used. A number of different reporters can be used and the only criteria when choosing one are: 1) the fluorophore is sufficiently bright so that it can be easily detected by the imaging protocol; and 2) its fluorescent spectrum does not overlap with that of YFP. For example, green fluorescent protein (GFP) should not be used as a reporter because it will be detected by the YFP filter or laser used to detect Venus. This is due to the spectral overlap of the two fluorophores. Choosing a fluorophore that localizes to a specific organelle can provide additional qualitative information. For example, use of a protein that targets the mitochondria, such as the red fluorescent protein DsRed-mito, can provide additional information on the overall viability of the cell. Mitochondria tend to undergo fragmentation (fission) during the early stages of cell death ²³ and this can be visualized with a reporter like DsRed-mito. The use of an organelle marker as a reporter can also allow conclusions about subcellular localization of caspase BiFC to be made.

There are a number of ways that caspase BiFC data can be acquired and analyzed. Deciding which method to use will be determined by the parameters to be explored (endpoint population analysis versus single cell positional or kinetic information) and the availability of instrumentation and imaging software. This protocol outlines a few microscope-based approaches to collecting the results of a caspase BiFC experiment for acquiring both single time points and time-lapse data. However, it should be noted that a number of variations and combinations of these approaches to be used to creatively interrogate caspase activation pathways.

Single time point data acquisition (4.1) is used to determine the percentage of cells with caspase induced proximity at a single time point. For this protocol, one simply counts the number of transfected cells that are Venus-positive. Thus, highly specialized equipment is not required and only the most basic epifluorescence microscope with a GFP (or YFP) and RFP filter and a hand tally counter is needed. Some imaging systems can be automated to automatically take a set number of images per well of a plate. If such instrumentation is available, the cells in the acquired images can be similarly counted by visual inspection, or quantitated using imaging software. Three-dimensional imaging (4.2) provides high resolution images of caspase BiFC throughout the focal planes of the cell. A confocal microscope with lasers that excite Venus and RFP (or the chosen transfection reporter) and Z-stack capabilities is required. A number of 3D rendering software modules are available that provide different ways to visualize the data. For both these types of analysis, it is possible to fix the cells and image or assess them at a later point. However, the process of fixation does diminish the Venus signal, which could add artifacts to the analysis and therefore this approach is not recommended. Time-lapse imaging (4.3) can be used to provide simultaneous kinetic and positional analysis of caspase BiFC. A microscope equipped with a motorized stage can be set up to take images of multiple positions and even compare different treatment groups in the same experiment.

For time-lapse, a number of additional considerations should be taken into account. The laser light can cause artifacts due to phototoxicity or photobleaching and both need to be controlled for. In general, both can be avoided by using the lowest possible amount of light (input low percentage laser power and short exposure times in Steps 4.3.6-4.3.7). Phototoxicity results when the laser light induces death in the cells and can be controlled for by imaging untreated cells under the same conditions and confirming that cell division proceeds as normal. Photobleaching occurs when multiple images taken of the same cell decreases the fluorescence. The Venus fluorescent protein is quite photostable, so this is rarely a problem in practice. The effects of photobleaching can be determined by taking numerous sequential images of a Venus-positive cell and ensuring that the fluorescence remains stable. This only needs to be done once, as long as the same laser power and exposure settings are used for every experiment. Effects due to phototoxicity or photobleaching can also be mitigated by reducing the total number of images captured. This can be achieved by lengthening or shortening the time interval between captures (Step 4.3.9). Using 5-10 min intervals over a 16 h time-lapse is a good place to start. If no photoxicity is observed or experiments of shorter duration are required, the interval between frames can be reduced. If caspase activation is rapid, shorter experiments with shorter time intervals can be used. For example, for a 16 h experiment with 10 min intervals, a total of 192 images will be taken (96 for each fluorophore). A 2 h experiment with 1.5 m intervals will yield 160 images. Therefore, the long and short experiments would each expose the cells to similar total amounts of laser light. For short duration experiments the caspase activating stimulus can be added directly to the media immediately prior to initiation of the time-lapse capture (after Step 4.3.11 instead of at Step 3). To do this, remove the lid from the plate and gently drop in a solution containing the stimulus using a pipette. Take care not to jostle the plate and quickly check that the cells are still in focus before proceeding.

When carrying out time-lapse experiments, it is also critical to make sure the immediate environment of the cells closely resembles that of a tissue culture incubator. A number of environmental controls are available on different microscope set ups. These include smaller stage top incubators as well as environmental chambers that encapsulate the entire microscope. Both of these devices usually deliver heat, CO₂, and humidity to the chamber in a manner that can be precisely controlled. The precise control of CO₂ and temperature is essential to the success of the experiment because unexpected temperature fluctuations can lead to focal drift and absence of CO₂ causes the pH of the media to rise that can be toxic to the cells. If a CO₂ source is not available, the media can be buffered by the addition of Hepes (see Step 3.1.1). Even if a CO₂ source is available, it is recommended to include Hepes in the imaging medium as an extra measure to keep the cells healthy. However, even in the presence of Hepes, the absence of CO₂ for prolonged periods of time can cause cell death. Therefore, it is important to inspect untreated cells, imaged under the same conditions, and the color of the media at the conclusion of the experiment to ensure that the pH has remained stable. Note that phenol red does not need to be omitted from any of the media used in this protocol since the contribution of autofluorescence to the Venus signal is minimal.

There are a number of available methods to analyze the imaging data resulting from a time lapse experiment. Measuring the mean intensity of Venus pixels per transfected cell is the most straightforward way to analyze the timing of caspase induced proximity. As shown in Figure 3, Figure 4, Figure 5, the caspase-2 BiFC often appears as one or more puncta located in different subcellular compartments. Imaging analysis approaches that measure numbers of foci, object size, or additional similar factors can yield informative quantitative data about these structures. Not all caspase BiFC appears as puncta. For example, the inflammatory caspases have shown different BiFC patterning including diffuse fluorescence, filamentous structures, and single or multiple puncta per cell ¹⁸. These patterns were dependent on the caspase activated as well as the upstream activating signals. Therefore, the type of analysis that is most suitable for the time-lapse results will largely be dictated by the type of fluorescent patterning that is observed.

Many of the treatment stimuli that are assessed when using this protocol will induce apoptosis either directly through the caspase analyzed or indirectly by engaging a different cell death pathway. This causes the cells to shrink and detach from the plate, which makes it very difficult to image the cells and almost impossible to determine any specific localization of the caspase-BiFC. Inclusion of a pan-caspase inhibitor will prevent this cell death. The recruitment of caspases to their activation platforms and the associated caspase BiFC is not dependent on the catalytic activity of the caspase. Therefore, caspase inhibition will have no effect on this step, but will inhibit the downstream physical characteristics of apoptosis including blebbing, detachment, and eventual loss in plasma membrane integrity. Thus, if carrying out endpoint analysis (Step 4.1) or Z stacks (Step 4.2), the inclusion of a caspase inhibitor is essential. For time-lapse experiments in contrast, if apoptosis-related events such as MOMP, annexin V binding (to detect phosphatidyl serine exposure) or propidium iodide uptake (to detect plasma membrane permeabilization) are to be analyzed alongside caspase BiFC, the caspase inhibitor should be omitted to prevent inhibition of these events. A caspase inhibitor such as qVD-OPh (10-20 µM) should be added to in the media that contains the apoptosis-inducing stimulus (Step 3.1.1). If co-expressing a pro-apoptotic protein with the caspase BiFC components, the caspase inhibitor should be added at Step 2.1.12.

While caspase BiFC is one of the few available methods that can be used to visualize events in caspase activation pathways in live cells, a number of considerations must be taken into account when designing these experiments. Because this approach is overexpression-based, controls for specificity are important. This can be accomplished in one of two ways. First, introduction of single point mutations into the CARD or DED of the caspase prodomain such that they disrupt the binding between the caspase and its activation platform should decrease both the intensity of caspase BiFC and the percentage of Venus-positive cells. For example, mutation of the residue D59 to E in the prodomain of caspase-1 disrupts binding to the adaptor protein ASC (apoptosis-associated speck-like protein containing a CARD) and equally disrupts ASC-induced caspase-1 BiFC ^18,24. Second, use of cells that are deficient in the adaptor proteins that recruit the caspase to the activation platform will decrease specific BiFC signals. This can be achieved using knockout cells, if available, or using siRNA or CRISPR/Cas9-based approaches to deplete the cells of the adaptor. Accordingly, depletion of the caspase-2 adaptor protein RAIDD completely blocked caspase-2 BiFC induced by heat shock or DNA damage ^14,17. For these experiments, the knockout or knockdown cells should be compared with matched control cells (i.e. litter-matched wild type MEF or control knockdown cells) and, if a new cell line is used, optimal transfection conditions should be assessed to ensure equal expression of the BiFC reporter across cell lines.

A second limitation of this approach is that, when using transient transfection, it is difficult to ensure that each cell is expressing equal amounts of each protein. If expression is unequal, one Venus fragment could be recruited to the platform at higher frequencies. This could mask the full fluorescent signal and lead to false negative results. To overcome this problem, a new version of the caspase BiFC reporter for stable cell line generation was recently reported ¹⁷. This new version consisted of a construct that expressed the C2 Pro-BiFC components on a single open reading frame separated by the viral 2A self-cleaving peptide ²⁵. Therefore, the BiFC components are transcribed as a single mRNA from the same promoter but translated to produce stoichiometrically equal amounts of the VC and VN fusion proteins. The cell lines generated to stably express this construct demonstrated similar trends of activation to the transiently expressed reporters but were both more sensitive and displayed lower background fluorescence. Thus, the above protocol can be greatly simplified if used in conjunction with reporter cell lines that stably express the caspase BiFC components.

Data to date suggest minimal interference between the caspase BiFC components and the endogenous proteins. For example, caspase-2 is known to be activated upstream of MOMP but caspase-2 BiFC did not block the progression of MOMP ¹⁴. Thus, these reporters do not appear to act in a dominant negative fashion. However, if simultaneously evaluating downstream events, this should be investigated within each set of experiments. Including a control non-transfected well in the experiment is the best way to approach this. If the kinetics of cell death are unchanged with or without expression of the caspase sensor, this will indicate that the caspase-BiFC components are not interfering with endogenous function. Conversely, the endogenous caspase could possibly affect the BiFC readout, either in efficiency or in the size or shape of the structures. Expressing the reporter in cells deficient for the caspase will control for this.

Despite these limitations, the caspase BiFC technique is a useful method to interrogate caspase activation pathways that can complement and even improve on existing approaches to measure initiator caspase activation. By measuring induced proximity, the first step in activation, this technique overcomes problems associated with measuring later steps such as caspase auto-processing associated with activation. For example, while using western blot to monitor the cleavage of the executioner caspases, caspase-3 and caspase-7, provides an accurate measure of activation ², using the same approach for initiator caspases can be flawed. This is because cleavage of initiator caspases, is not sufficient for activation ^4,26,27. In fact, many initiator caspases are cleaved during apoptosis by caspase-3 without producing active enzymes ²⁸. Similarly, techniques based on using peptide substrates designed as targets for specific caspases need to take into account the overlapping specificities of these sequences for the caspases ²⁹. This does not mean that such approaches cannot be used to measure caspase activation, but the drawbacks should be acknowledged when interpreting the results. Caspase BiFC provides an alternative approach that can help confirm results using more conventional methods. In addition, the ability to track the assembly of the caspase activation complexes in real time and to distinctly determine the size, shape and localization of the complexes is a significant advantage of this technique that is not possible to assess by other means.