We describe a method for the qualitative and quantitative analysis of stress granule formation in mammalian cells after the cells are challenged with bacteria and a number of different stresses. This protocol can be applied to investigate the cellular stress granule response in a wide range of host-bacterial interactions.
Fluorescent imaging of cellular components is an effective tool to investigate host-pathogen interactions. Pathogens can affect many different features of infected cells, including organelle ultrastructure, cytoskeletal network organization, as well as cellular processes such as Stress Granule (SG) formation. The characterization of how pathogens subvert host processes is an important and integral part of the field of pathogenesis. While variable phenotypes may be readily visible, the precise analysis of the qualitative and quantitative differences in the cellular structures induced by pathogen challenge is essential for defining statistically significant differences between experimental and control samples. SG formation is an evolutionarily conserved stress response that leads to antiviral responses and has long been investigated using viral infections1. SG formation also affects signaling cascades and may have other still unknown consequences2. The characterization of this stress response to pathogens other than viruses, such as bacterial pathogens, is currently an emerging area of research3. For now, quantitative and qualitative analysis of SG formation is not yet routinely used, even in the viral systems. Here we describe a simple method for inducing and characterizing SG formation in uninfected cells and in cells infected with a cytosolic bacterial pathogen, which affects the formation of SGs in response to various exogenous stresses. Analysis of SG formation and composition is achieved by using a number of different SG markers and the spot detector plug-in of ICY, an open source image analysis tool.
Visualizing host-pathogen interactions on a cellular level is a powerful method for gaining insights into pathogenic strategies and for identifying key cellular pathways. Indeed, pathogens can be used as tools to pinpoint important cellular targets or structures, as pathogens have evolved to subvert central cellular processes as a strategy for their own survival or propagation. Visualization of cellular components can be achieved by recombinantly expressing fluorescently-tagged host proteins. While this allows for real-time analysis, the generation of cell lines with specifically-tagged host proteins is highly laborious and may result in undesirable side effects. More convenient is the detection of cellular factors using specific antibodies, because multiple host factors can be analyzed simultaneously and one is not limited to a particular cell type. A drawback is that only a static view can be captured as immunofluorescence analysis necessitates host cell fixation. However, an important advantage of immunofluorescence imaging is that it readily lends itself to both qualitative and quantitative analysis. This in turn can be used to obtain statistically significant differences to provide new insights into host-pathogen interactions.
Fluorescent image analysis programs are powerful analytical tools for performing 3D and 4D analysis. However, the high cost of software and its maintenance make methods based on free open source software more widely attractive. Careful image analysis using bio-analysis software is valuable as it substantiates visual analysis and, when assigning statistical significances, increases confidence in the correctness of a given phenotype. Previously, SGs have been analyzed using the free ImageJ software, which necessitates the manual identification of individual SGs4. Here we provide a protocol for the induction and analysis of cellular SG formation in the context of bacterial infections using the free open source bio-image analysis software ICY (http://icy.bioimageanalysis.org). The bio-image analysis software has a built-in spot detector program that is highly suitable for SG analysis. It allows the fine-tuning of the automated detection process in specified Regions Of Interest (ROIs). This overcomes the need for manual analysis of individual SGs and removes sampling bias.
Many environmental stresses induce the formation of SGs, which are phase dense cytosolic, non-membranous structures of 0.2 – 5 µm in diameter5,6. This cellular response is evolutionarily conserved in yeast, plants and mammals and occurs when global protein translation is inhibited. It involves aggregation of stalled translation initiation complexes into SGs, which are considered holding places for translationally-inactive mRNAs, allowing selective translation of a subset of cellular mRNAs. Upon removal of the stress, SGs dissolve and global rates of protein synthesis resume. SGs are composed of translation elongation initiation factors, proteins involved in RNA metabolism, RNA-binding proteins, as well as scaffolding proteins and factors involved in host cell signaling2, although the exact composition can vary depending on the stress applied. Environmental factors that induce SG formation include amino acid starvation, UV irradiation, heat shock, osmotic shock, endoplasmic reticulum stress, hypoxia and viral infection2,7,8. Much progress has been made in understanding how viruses induce and also subvert SG formation, while little is still known about how other pathogens, such as bacterial, fungal or protozoan pathogens, affect this cellular stress response1,7.
Shigella flexneri is a gram-negative facultative cytosolic pathogen of humans and the causative agent of severe diarrhea or shigellosis. Shigellosis is a major public health burden and leads to 28,000 deaths annually in children under 5 years of age9,10. S. flexneri infects the colonic epithelium and spreads cell-to-cell by hijacking the host's cytoskeletal components11,12. Infection of the epithelium supports the replication of S. flexneri within the cytosol but infected macrophages die through an inflammatory cell death process called pyroptosis. Infection leads to a massive recruitment of neutrophils and severe inflammation that is accompanied by heat, oxidative stress and tissue destruction. Thus, while infected cells are subject to internal stresses induced by infection, such as Golgi disruption, genotoxic stress and cytoskeletal rearrangements, infected cells are also subjected to environmental stresses due to the inflammatory process.
Characterization of the effect of S. flexneri infection on the ability of cells to respond to environmental stresses using a number of SG markers has demonstrated that infection leads to qualitative and quantitative differences in SG composition3. However, little is known about other bacterial pathogens. Here we describe a methodology for the infection of host cells with the cytosolic pathogen S. flexneri, the stressing of cells with different environmental stresses, the labeling of SG components, and the qualitative and quantitative analysis of SG formation and composition in the context of infected and non-infected cells. This method is widely applicable to other bacterial pathogens. In addition, the image analysis of the SG formation may be used for infections by viruses or other pathogens. It can be used to analyze SG formation upon infection or the effect of infection on SG formation in response to exogenous stresses.
1. Preparation of Bacteria and Host Cells
2. Bacterial Challenge of Host Cells
3. Inducing Stress Granule Formation by the Addition of Exogenous Stressors
4. Fixation and Immunofluorescence Analysis of Stress Granule Formation
NOTE: Process the control and experimental coverslips at the same time to avoid staining differences that may impact image analysis in subsequent steps. The control samples include no infection with and without SG inducing treatment, and infected samples with and without SG inducing treatment.
5. Fluorescence Imaging
NOTE: Refer to the user manual of the microscope for optimization of the set up.
6. Image Analysis
NOTE: Here, SG analysis on collapsed stacks using freeware ICY is described. Image analysis of 3D reconstructions may also be done using other specialized software. SG detection may be performed via a fully automated workflow established for this protocol (available at http://icy.bioimageanalysis.org/protocol/Stress_granule_detection_in_fluorescence_imaging). To use this protocol, the nuclei need to be stained with a DNA stain for the software to find the center of each cell, and the cell edges need to be marked by either a cytoplasmic marker (such as eIF3b) or actin stain for the software to identify the cell boundaries. The automatic workflow can be used to validate manually-derived results or directly for analysis when cell boundaries are detected with high confidence, which will mostly depend on the density of cells and the marker used to detect the cell boundaries.
To explain and demonstrate the protocol described in this manuscript, we characterized the image of clotrimazole-induced SGs in HeLa cells infected or not with the cytosolic pathogen S. flexneri. An outline of the procedure is presented in Figure 1, and includes virulent and avirulent S. flexneri streaked on Congo Red plates, bacteria preparation, infection, addition of environmental stress, sample fixation and staining, sample imaging and quantitation, as well as image analysis. A number of different stresses may be used to induce SG formation and a variety of SG markers are available for interrogation. eIF3b is a canonical SG marker that also clearly stains the cytoplasmic compartment of the cell and can be used to identify cell edges. G3BP1 is a widely-used SG marker that aggregates into SGs without any background staining (Figure 2A, B & D). Cells are best imaged with a confocal microscope, and the stacks covering the whole cell depth are loaded onto the imaging analysis freeware ICY (Figure 2A – C). To better identify infected cells and to put cells into either the infected or non-infected group for later analysis, it is useful to increase the intensity of the nucleic acid stain (Figure 2D). Within the imaging analysis software, the spot detector is then used to identify SGs within each of the designated ROIs (Figure 2E) and a binary image is generated that displays all the identified SGs (Figure 2F).
SG analysis needs to be tailored to each SG marker analyzed and the appropriate setting for analysis must be carefully chosen. Focusing on the SG marker G3BP1, changing the size requirement of the spot detected or changing the sensitivity of the detection parameters will lead to varying results, as demonstrated in Figure 3A – C. The scale selection (1 – 3, although scales can be added) is based on pixel size and thus at higher scales, smaller SGs will not be counted and the numbers of SGs will decrease (Figure 3C). Sensitivity is measured from 1 – 100, with 100 being the most sensitive. By increasing the sensitivity, the number of SGs detected increases (Figure 3C). Thus, the correct settings must be found to minimize false positives and false negatives. This is best done by carefully analyzing the visuals obtained for each setting. For G3BP1, a scale of 2 with a sensitivity of 100 (2 – 100, highlighted in red) gave the best result. A scale of 2 with a lower sensitivity (50 or 25) left small SGs uncounted (see orange arrows). Similarly, a scale of 3 left small SG unaccounted while a scale of 1 oversampled. Importantly, additional scales can be added to focus only on large SGs, if this is warranted. For eIF3b, a scale of 2 with a sensitivity of 55 (2 – 55) gave the best result for eIF3b (highlighted in red) although red arrows highlight either under or oversampling in a scale 2 – 50 & 2 – 55, respectively.
Once the parameters for SG analysis are properly set, the data can be analyzed in many ways (Figure 4). The spot detector gives the number of SGs detected within each ROI such that the uninfected and infected cells can be compared (Figure 4A). Similarly, the size, in this case number of pixels, is given from which the surface area can be calculated (Figure 4B). Frequency distribution plots are also useful to highlight shifts in the size of SGs found in different cell populations. Here, a complete absence of large SGs in S. flexneri infected cells, as well as a definite shift in the distribution becomes clearer (Figure 4C). In addition, the intensity (shown here is minimum and maximum intensity) provides information on the quality of the SGs analyzed. For S. flexneri infected cells, SGs are significantly less intense. These analyses provide statistically relevant information on both the qualitative and quantitative nature of SGs formed in response to exogenous stress when cells are infected with or without S. flexneri.
Figure 1: Outline of the Experimental Procedure to Infect Cells with S. Flexneri and to Analyze the Effect of Infection SG Formation. A Congo-red colony, and a nonvirulent Congo-red-negative colony, are picked and grown O/N in tryptic soy broth. The overnight culture is subcultured to the late exponential phase before bacteria are coated with PLL to promote adherence. Host cells grown on glass coverslips in 12-well plates are infected with bacteria for 30 min. Control cells are not infected. Cells are washed and treated with stress inducers to induce SG formation either by replacing the medium with stress-containing medium or subjecting the cells to stress conditions, such as heat. Cells are then fixed, permeabilized, stained with immunofluorescent SG-specific markers, and imaged using confocal microscopy. Z-projections are made using ImageJ and analyzed using the spot detector of the image analysis software. The data is then analyzed. Please click here to view a larger version of this figure.
Figure 2: Spot Detector Analysis of HeLa Cells Infected with S. Flexneri for 1.5 h before the Addition of Clotrimazole for 1 h. A. Immunofluorescent image of a z-projection showing cells stained with the SG markers eIF3b and G3BP1, DAPI and actin. B. Same image as in (A) but without the actin stain to highlight the use of eIF3b to delineate cell boundaries. C. Screen shot of the image analysis software with the spot detector module. D. Gating of the infected and non-infected cells as seen using different channels. To see all bacteria, the intensity is increased. Cells with more than 1 bacterium present in the cell are labeled with a red star. E. Output of the spot detector analysis of individual ROIs, indicating the ROI name, the number of spots and their location. F. Image of spots detected in the ROIs. Scale bar = 30 µm. Please click here to view a larger version of this figure.
Figure 3: Comparisons of Different Scale and Sensitivity Settings of Spot Detector for the Detection of SGs through Different SG Markers. A. Zoom-in to HeLa cells infected or not with S. flexneri and stained with DAPI and the SG marker G3BP1, and analyzed at different scales (pixel size cut off, 1 – 3) and sensitivity (1 – 100). Graphical representation of SG numbers in uninfected and infected cells analyzed at different scales (B) and sensitivities (C). Visuals (D) and graphical representation (E) of SG number detection of the SG marker eIF3b for the same image when changing the sensitivity limit without changing the spot size cut-off. Red writing and red symbols indicate the best parameters. Red arrows point towards false positives and orange arrows to a lack of SG detection. Values include mean with standard deviation. Best results are highlighted in red. Selected statistical analysis included for clarity using the Wilcoxon rank sum test p-values on the left and the variance F-test on the right. * = <0.05, ** = <0.01, *** = <0.001, ns = nonsignificant. Scale bar = 10 µm. Please click here to view a larger version of this figure.
Figure 4: Analysis of Spot Detector Generated SG Data Obtained from Clotrimazole-treated HeLa Cells Infected with S. Flexneri. A. Number of G3BP1 SGs per cell in infected and uninfected HeLa cells. B. Surface area of G3BP1-SGs in infected and uninfected cells, and their percent distribution frequency (C). D. Minimum and maximum intensity values (arbitrary) of G3BP1-SGs. Values include mean with standard error. Selected statistical analysis included for clarity using the Wilcoxon rank sum test p-values on the left and the variance F-test on the right. * = <0.05, ** = <0.01, *** = <0.001, ns = nonsignificant. Please click here to view a larger version of this figure.
The protocol outlined here describes the induction, localization, and analysis of SGs in non-infected cells and cells infected with the cytosolic pathogen S. flexneri in the presence or absence of exogenous stress. Using free imaging software, the protocols allows for the precise qualitative and quantitative analysis of SG formation to identify and statistically address differences in given phenotypes.
There are several critical steps within the protocol for the infection, SG-induction and imaging parts. For the infection, it is important that the bacterial interaction with host cells, such as invasion as outlined protocol here, is synchronized as much as possible. This is particularly important when investigating the effects of bacterial challenge early during the infection process. We previously showed that some phenotypes, such as eIF3b localization after S. flexneri infection, are perturbed early during wild-type S. flexneri challenge while other phenotypes, such as aggregation of G3BP1-containing SGs, are more affected at later time points3. Thus, to precisely describe the temporal effect of bacterial challenge, synchronization of infection is advisable. Notably, to analyze SG formation, cells should be in exponential phase of growth, and thus cell density at the time of stress addition is also an important parameter. It also limits SG analysis to non-confluent cell infection models. Another critical aspect is the co-processing of control and experimental samples during both the processing of the samples for imaging and during image acquisition. For immunofluorescent processing, all samples should be stained with the same antibody working dilutions for the same amount of time and under the same conditions (i.e. RT Vs. 4 °C), while also taking care to treat each coverslip similarly during the washing stages and mounting of the coverslips. This will ensure comparable immunofluorescent stainings that can be analyzed both qualitatively and quantitatively. Differences in mounting medium can have significant effects on bleaching of the samples during image acquisition and therefore should be kept constant. Similarly, image acquisition for all samples to be compared in analysis should be performed within one sitting and with the same settings to minimize differences arising from laser strength or sample set-up.
When using exogenous stresses, it is important to properly store the reagent and to frequently make new working stocks. Extended periods (>4 months for clotrimazole, for example) lead to decreased potency of the drug and will affect SG formation. Reagents should be added to culture medium immediately before addition to the cells; if a reduction in SG formation in control cells is observed or the aggregation of SGs appears aberrant, reagents should be tested for their activity and new stocks should be made.
One limitation is that SG identification using spot detector becomes less reliable when too much background fluorescence is present. While this is not a problem for many SG markers, some, such as eIF3b, which is a clear cytosolic marker under both SG-inducing and non-inducing conditions, are more difficult to analyze as demonstrated in Figure 3. In addition, scales and sensitivities need to be empirically determined for each SG marker. Another limitation is that the analysis through the image analysis software is best with 2D images, and therefore works well on optical slices or collapsed stacks, which however results in a loss of 3D information. The analysis on z-projections therefore tends to overstate the size of SGs and understate the number of SGs. 3D analysis of SGs may be performed with Imaris to give an even more precise spatial characterization, if deemed necessary for a particular phenotype.
SG characterization can be performed quickly and in a standardized manner using the automated spot detector of the image analysis software. In contrast, manual methods to identify and delineate SGs as previously performed4, leave the analysis open to more bias and is considerably more time consuming. SG analysis using spot detector can also be tailored to include or exclude small SG aggregates by selecting different sensitivity and size thresholds, thereby allowing flexibility in evaluating different aspects of SG formation. SG analysis can also be performed in a completely automated manner using an established automated workflow3. Within this workflow, the center of the cell is identified through a DAPI stain and the program then lets a malleable sphere grow outwards to delineate the cell boundaries based on either a cytoplasmic stain (such as eIF3b) or actin. At the same time, using a semi-automated analysis by manually delineating individual cells for analysis, can be advantageous when cells grow in clusters and cell boundaries are difficult for the automated program to delineate clearly. In these circumstances, defining cell boundaries manually can eliminate false positive or negative values.
The protocol may be adapted to other SG-inducing conditions and to other pathogens, including viruses, bacteria, yeast and protozoans. Of particular interest may be other cytosolic pathogens such as Rickettsia spp., Francisella tularensis, Burkholdieria pseudomallei and Listeria spp.14 For each infectious agent, and possibly also different cell lines, the infection protocol will have to be adapted and the sample times of exogenous stresses empirically determined. In addition, SG characterization using image analysis software may also be extended to live-cell imaging in the future. Real-time SG analysis would necessitate the expression of one or more fluorescently-tagged SG markers in the host cells but would thereby allow questions regarding the temporal and special dynamics of SG formation to be addressed under different experimental conditions. Fluorescently-labeled bacteria would then be needed to complement the real-time SG analysis in infected and uninfected cells.
The authors have nothing to disclose.
PS is a recipient of the Bill and Melinda Gates Grand Challenge Grant OPP1141322. PV was supported by a Swiss National Science Foundation Early Postdoc Mobility fellowship and a Roux-Cantarini postdoctoral fellowship. PJS is supported by an HHMI grant and ERC-2013-ADG 339579-Decrypt.
Primary Antibodies | |||
eIF3b (N20), origin goat | Santa Cruz | sc-16377 | Robust and widely used SG marker. Cytosolic staining allows cell delineation. Dilution 1 in 300 |
eIF3b (A20), origin goat | Santa Cruz | sc-16378 | Same target as eIF3b (N20) and in our hands was identical to eIF3b (N20). Dilution 1 in 300 |
eIF3A (D51F4), origin rabbit (MC: monoclonal) | Cell Signaling | 3411 | Part of multiprotein eIF3 complex with eIF3b . Dilution 1 in 800 |
eIF4AI, origin goat | Santa Cruz | sc-14211 | Recommended by (Ref # 13). Dilution 1 in 200 |
eIF4B, origin rabbit | Abcam | ab186856 | Good stress granule marker in our hands. Dilution 1 in 300 |
eIF4B, origin rabbit | Cell Signaling | 3592 | Recommended by Ref # 13. Dilution 1 in 100 |
eIF4G, origin rabbit | Santa Cruz | sc-11373 | Widely used SG marker. (Ref # 13): may not work well in mouse cell lines. Dilution 1 in 300 |
G3BP1, origin rabbit (MC: monoclonal) | BD Biosciences | 611126 | Widely used SG marker. Dilution 1 in 300 |
Tia-1, origin goat | Santa Cruz | sc-1751 | Widely used SG marker. Can also be found in P bodies when SG are present (Ref # 13). Dilution 1 in 300 |
Alexa-conjugated Secondary Antibodies | |||
A488 anti-goat , origin donkey | Thermo Fisher | A-11055 | Cross absorbed. Dilution 1 in 500 |
A568 anti-goat, origin donkey | Thermo Fisher | A-11057 | Cross absorbed. Dilution 1 in 500 |
A488 anti-mouse, origin donkey | Thermo Fisher | A-21202 | Dilution 1 in 500 |
A568 anti-mouse, origin donkey | Thermo Fisher | A10037 | Dilution 1 in 500 |
A647 anti-mouse, origin donkey | Thermo Fisher | A31571 | Dilution 1 in 500 |
A488 anti-rabbit, origin donkey | Thermo Fisher | A-21206 | Dilution 1 in 500 |
A568 anti-rabbit, origin donkey | Thermo Fisher | A10042 | Dilution 1 in 500 |
Other Reagents | |||
Shigella flexneri | Available from various laboratories by request | ||
Tryptone Casein Soya (TCS) broth | BD Biosciences | 211825 | Standard growth medium for Shigella, application – bacterial growth |
TCS agar | BD Biosciences | 236950 | Standard growth agar for Shigella, application – bacterial growth |
Congo red | SERVA Electrophoresis GmbH | 27215.01 | Distrimination tool for Shigell that have lost the virulence plasmid, application – bacterial growth |
Poly L lysine | Sigma-Aldrich | P1274 | Useful to coating bacteria to increase infection, application – infection |
Gentamicin | Sigma-Aldrich | G1397 | Selective killing of extracellular but not cytosolic bacteria, application – infection |
HEPES | Life Technologies | 15630-056 | PH buffer useful when cells are incubated at room-temperature, application – cell culture |
DMEM | Life Technologies | 31885 | Standard culture medium for HeLa cells, application – cell culture |
Fetal calf serum | Biowest | S1810-100 | 5% supplementation used for HeLa cell culture medium, application – cell culture |
Non-essential amino acids | Life Technologies | 11140 | 1/100 dilution used for HeLa cell culture medium, application – cell culture |
DMSO | Sigma-Aldrich | D2650 | Reagent diluent, application – cell culture |
Sodium arsenite | Sigma-Aldrich | S7400 | Potent stress granule inducer (Note: highly toxic, special handling and disposal required), application – stress inducer |
Clotrimazole | Sigma-Aldrich | C6019 | Potent stress granule inducer (Note:health hazard, special handling and disposal required), application – stress inducer |
PFA | Electron Microscopy Scences | 15714 | 4% PFA is used for standard fixation of cells, application – fixation |
Triton X-100 | Sigma-Aldrich | T8787 | Used at 0.03% for permeabilizationof host cells before immunofluorescent staining, application – permeabilization |
A647-phalloidin | Thermo Fisher | A22287 | Dilution is at 1/40, best added during 2ary antibody staining, application – staining |
DAPI | Sigma-Aldrich | D9542 | Nucleid acid stain used to visualize both the host nucleus and bacteria, application – staining |
Parafilm | Sigma-Aldrich | BR701501 | Paraffin film useful for immunofluorescent staining of coverslips, application – staining |
Prolong Gold | Thermo Fisher | 36930 | Robust mounting medium that works well for most fluorophores , application – mounting |
Mowiol | Sigma-Aldrich | 81381 | Cheap and robust mounting medium that works well for most fluorophores, application – mounting |
24-well cell culture plate | Sigma-Aldrich | CLS3527 | Standard tissue culture plates, application – cell culture |
12-mm glass coverslips | NeuVitro | 1001/12 | Cell culture support for immunofluorescent applications, application – cell support |
forceps | Sigma-Aldrich | 81381 | Cheap and obust mounting medium that works well for most fluorophores, application – mounting |
Programs and Equipment | |||
Prism | GraphPad Software | Data analysisand graphing program with robust statistical test options, application – data analysis | |
Leica SP5 | Leica Microsystems | Confocal microsope, application – image acquisition | |
Imaris | Bitplane | Professional image analysis program, application – data analysis | |
Excel | Microsoft | Data analysis and graphing program, application – data analysis |