Name: workflow for high-content, individual cell quantification of fluorescent markers from universal microscope data, supported by open source software
Uploaded: 2014-12-16T13:00:00
Duration: 9 min 57 s
Description: Watch this Scientific Journal Video about Workflow for High-content, Individual Cell Quantification of Fluorescent Markers from Universal Microscope Data, Supported by Open Source Software at JoVE.com

This work was supported by grants CRUK 15043 and CRUK 14251.
We thank Daniel Wetterskog and Ka Kei Ho for technical assistance and critical reading of the manuscript.

1. Experimental Perturbation and Cell Labeling for Response Markers (Reverse Transfection siRNA Screen)
<ol>
	<li>In a sterile tissue culture hood pipette 70 &#181;l of 200 nM siRNA in 1x siRNA buffer in wells of a sterile, plain 96 well plate. Dilute transfection lipid into 40 volumes of serum-free DMEM media and dispense 105 &#181;l into each well containing siRNA. 
	NOTE: Diluting 262.5 &#181;l lipid into 10.5 ml of serum-free DMEM yields a master mix suitable for a whole 96 well plate of siRNA, delivering 2.6 &#181;l of lipid per well. Use of 200 nM siRNA starting concentration at this step will deliver a working concentration of 20 nM in step 1.3, but procedures will work for working concentrations down to 5 nM, with the starting concentration adjusted accordingly (i.e. 50 nM). Lower working concentration may reduce off-target false positive scores, although they can reduce the magnitude of on-target response, leading to increase of on-target false negative rates.</li>
	<li>Mix the plate by gentle vibration for ten minutes at room temperature. Sub-divide the resulting 175 &#181;l into three 50 &#181;l replicates per target onto an opaque, tissue culture treated, 96-well plate with a transparent base.</li>
	<li>Reverse transfect by dispensing 8,000 cells per well in 150 &#181;l DMEM containing 10% serum directly onto the 50 &#181;l lipid-siRNA complexes. Use HCT116 human colorectal cells stably expressing a GFP-tagged marker reporting CDK2 activity5,8. No further mixing is necessary. Seal the plate with a sterile, adhesive breathable membrane to control humidity and prevent plate &#8216;edge-effects&#8217; and place the plate into a humidified incubator at 37 &#176;C, 5% CO2 for 48 hr.</li>
	<li>Aspirate the media such that a small residual amount of media remains in the wells. Fix the cells by adding 100 &#181;l of 4% buffered formaldehyde to each well and incubate in a fume hood for 10 min at room temperature.</li>
	<li>Remove the fixing solution by aspirating the plate. At this point either stop the experiment by washing the plate three times with 100 &#181;l phosphate-buffered saline (PBS) and then store sealed, under 100 &#181;l of PBS in the dark at 4 &#176;C for up to a week, or proceed with the permeabilization of the cells. 
	NOTE: We recommend processing plates as soon as possible after fixation, and generally prefer storage of fully processed plates. Biocidal preservatives such as thimerosal, sodium azide, or commercial alternatives may be added to prevent micoroganismal growth. Addition of phosphatase inhibitors helps to preserve phospho-epitopes, and other means to preserve protein modification states may be useful in relevant assay contexts</li>
	<li>Remove PBS from the plate and permeabilize the cells by adding 100 &#181;l of permeabilization solution. Incubate for 10 min at room temperature without shaking. Aspirate the permeabilization solution using a multichannel pipette. Repeat this step three times.</li>
	<li>Block the cells by adding 100 &#181;l block solution per well for 30 min at room temperature. Remove the block solution by aspirating the plate, then probe with 50 &#181;l of anti P-S780 RB1 antibody diluted 500-fold in the block solution for 2 hr in the dark at room temperature.</li>
	<li>Wash the plate three times with 100 &#181;l plate wash solution, leaving the solution on the plate for 5 min each time. Probe the plate overnight in the dark at 4 &#176;C with 50 &#181;l fluorescently-tagged secondary antibody diluted 1,000-fold in block solution supplemented with 2 &#181;M of the chromatin-specific DNA dye Bisbenzimide. Wash the plate three times as before and store sealed, under 100 &#181;l PBS in the dark at 4 &#176;C. Image the plate within two weeks.</li>
</ol>
2. Imaging and Image Segmentation
<ol>
	<li>Use a confocal or spinning-disk fluorescence microscope with a 20X objective to take separate 16-bit, greyscale TIFF images in three channels corresponding to the DNA dye, GFP and immuno-staining fluorophores. Capture many non-overlapping image sets, referred to here as frames, to image approximately 1,000-2,000 cells per well.</li>
	<li>Name the image files systematically so that each file name is a unique combination of&#160; &#8216;experiment name&#8217;, &#8216;well address&#8217;, &#8216;frame number&#8217; and &#8216;channel identifier&#8217;, in this order (Figure 3). The example data set uses &#8220;Blue&#8221; (chromatin DNA staining) or &#8220;Green&#8221; (GFP) or &#8220;Red&#8221; (the immuno-stained fluorophore) as channel identifiers. The well address, frame number and channel identifier are further on referred to as the image metadata. Use the underscore symbol to avoid confusing well and frame metadata.</li>
	<li>Name the files with these metadata elements in the specified order. This is necessary to ensure that the subsequent software steps correctly group sets of images for analysis.</li>
	<li>Download and install the freeware Cell Profiler, Active Perl Community Edition, R statistical programming environment and RStudio. Accept all default options during installation; PC users installing Active Perl should enable all options relating to PATH, file extension association and script mapping where prompted. Active Perl is optional for Mac users, but they will otherwise need to run the Perl script in step 3.2 from the Terminal command line rather than using icon clicking.</li>
	<li>Open the Cell Profiler software, click &#8216;File&#8217;, &#8216;Import Pipeline&#8217; then &#8216;From file&#8217; and select the file 3_channels_pipeline.cppipe (Figures S1A &#38; S1B). The file contains instructions necessary for the software to interpret image file metadata from the file name convention described. Cell Profiler now relates the images, extracts nuclear DNA and antibody intensities from these and uses the GFP channel to calculate the ratio of nuclear versus cytoplasm intensity for each cell detected (Figures 4 &#38; 5).</li>
	<li>Click the button &#8216;View output settings&#8217; in the lower left corner of the Cell Profiler window. At the top of the new screen are text boxes labeled &#8216;Default Input Folder&#8217; and &#8216;Default Output Folder&#8217;. One at a time, click the folder-icons to the right of these boxes and select the location of the image files for analysis and the destination for the extracted data, respectively (Figure S1C).</li>
	<li>Begin the image analysis by pressing the &#8216;Analyze Images&#8217; button in the lower left corner of Cell Profiler. At the bottom of the screen observe the remaining time for the data extraction, the &#8216;Stop Analysis&#8217; and &#8216;Pause&#8217; buttons. If required pause the analysis by selecting the &#8216;Pause&#8217; button at any time, which is useful when watching the images being analyzed (described in step 2.8).</li>
	<li>Optionally, open the windows for any of the image analysis steps by clicking the eye-icons in the panel on the extreme left of the program window (Figure S1D). Observe the &#8216;IdentifyPrimaryObjects&#8217; window and those for &#8216;Secondary&#8217; and &#8216;Tertiary Objects&#8217; to check that the current settings in Cell Profiler to perform image segmentation are suitable (see Figure 1 and Discussion for advice on modifying these settings).</li>
	<li>Click &#8216;Ok&#8217; in the message box that appears when the analysis is complete. Go to the location &#8216;Default Output Folder&#8217; where all the data files with the results are saved as comma-separated-value (.csv) files (Figure S2A).</li>
</ol>
3. Data Extraction
<ol>
	<li>Find the new file &#8216;Nuclei.csv &#8217;, which is included among the output from Cell Profiler. This file contains individual cell data for fluorescent nuclear antibody intensity, nuclear DNA intensity and GFP-CDK2 reporter ratio values (Figures 6A &#38; S2A). 
	NOTE: Different laboratories will want to process this type of data to suit the nature of their own assays. Suggested for the current data is the gating of the cells from each treatment condition according to the antibody data and the GFP-CDK2 reporter values using the provided Perl script &#8216;2_gate_classifier.pl&#8217;.</li>
	<li>Copy the provided Perl script file &#8216;2_gate_classifier.pl&#8217; into the same folder as the &#8216;Nuclei.csv&#8217; data file (Figure S2A). Double-click the icon for the Perl script and, when prompted, type the full name of the data file followed by a &#8216;.csv&#8217; filename name for the file in which the cells are to be gated and finally the gate values for the antibody fluorescence and GFP-CDK2 reporter data. 
	NOTE: How to principally determine gate settings and apply these for analysis of data are discussed below in the Representative Data section and Figure 6 (to analyze the data provided use &#8216;0.004&#8217; and &#8216;1.5&#8217;, respectively). Mac users should run the Perl script from the Terminal command line by typing: &#8216;perl 2_gate_classifier.pl&#8217;.</li>
	<li>Observe the newly created file which combines the raw individual cell assay values from the original Cell Profiler data with sub-population labels that show how each cell from each well performs against both gates (Figure 6C).</li>
	<li>Plot the data for each experimental condition using the individual cell subpopulation labels by opening the RStudio software. Click &#8216;File&#8217; and &#8216;Open File&#8217;, then select the provided &#8216;analysis.r&#8217; file. Observe the commands to plot Figures 6B, 7 and 8 in the upper left window of RStudio (Figure S2B). In the upper left window, between the double quote symbols on lines 5 and 6, type the computer address of the folder containing the gated data. Include the drive letter and the name of the file itself, respectively (e.g., &#8220;C:/analysis folder/analysis output&#8221; and &#8220;nuclei_gated.csv&#8221;). 
	NOTE: If RStudio is used for the first time on a given computer, the R graphics package &#8216;ggplot2&#8217; will need to be installed first. This is a once only step for a new installation of RStudio, after which this step becomes redundant. To install &#8216;ggplot2&#8217;, click the tab called &#8216;Packages&#8217; above the window in the lower right corner of RStudio, click the &#8216;Install Packages&#8217; button that appears beneath this. A new window will appear. Type &#8216;ggplot2&#8217; (omitting quotations) into the &#8216;Packages&#8217; space in this new window and finally click the &#8216;Install&#8217; button to close the window, install the necessary ggplot2 functions and return to the main RStudio window to continue from step 3.6.</li>
	<li>Highlight lines 1 through 17 in the upper left window of RStudio, then click the &#8216;Run&#8217; button. This will enter the experimental data, threshold values and well location details into R (Figure S2C). R will now temporarily hold the relevant data for plotting.</li>
	<li>Highlight individual blocks of the remaining code beneath line 17 and create the corresponding plots by clicking the &#8216;Run&#8217; button as before. Observe the plots in the window in the lower right corner of RStudio and save number of formats by clicking the &#8216;Export&#8217; button (Figure S2D).</li>
	<li>While closing RStudio, click &#8216;Don&#8217;t Save&#8217; when prompted. This prevents confusion on the next use of RStudio, which will otherwise hold data from the previous session.</li>
</ol>

<ol>
	<li>Carpenter, A. E., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=CellProfiler:+image+analysis+software+for+identifying+and+quantifying+cell+phenotypes.">CellProfiler: image analysis software for identifying and quantifying cell phenotypes.</a> Genome Biol. 7 (10), 100 (2006).</li><li>Khan, A., Eldaly, H., Rajpoot, N. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+gamma-gaussian+mixture+model+for+detection+of+mitotic+cells+in+breast+cancer+histopathology+images.">A gamma-gaussian mixture model for detection of mitotic cells in breast cancer histopathology images.</a> J Pathol Inform. 4, 11 (2013).</li><li>Selzer, P., Beibel, M., Gubler, H., Parker, C. N., Gabriel, D. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Comparison+of+multivariate+data+analysis+strategies+for+high-content+screening.">Comparison of multivariate data analysis strategies for high-content screening.</a> J Biomol Screen. 16 (3), 338-347 (2011).</li><li>Lyman, S. K., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=High+content,+high-throughput+analysis+of+cell+cycle+perturbations+induced+by+the+HSP90+inhibitor+XL888.">High content, high-throughput analysis of cell cycle perturbations induced by the HSP90 inhibitor XL888.</a> PLoS One. 6 (3), e17692 (2011).</li><li>Richardson, E., Stockwell, S. R., Li, H., Aherne, W., Cuomo, M. E., Mittnacht, S. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Mechanism-based+screen+establishes+signalling+framework+for+DNA+damage-associated+G1+checkpoint+response.">Mechanism-based screen establishes signalling framework for DNA damage-associated G1 checkpoint response.</a> PLoS One. 7 (2), e17692 (2012).</li><li>Heynen-Genel, S., Pache, L., Chanda, S. K., Rosen, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Functional+genomic+and+high-content+screening+for+target+discovery+and+deconvolution.">Functional genomic and high-content screening for target discovery and deconvolution.</a> Expert Opin Drug Discov. 7 (10), 955-968 (2012).</li><li>Krausz, E. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=High-content+siRNA+screening.">High-content siRNA screening.</a> Mol Biosyst. 3 (4), 232-240 (2007).</li><li>Gu, J., Xia, X., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Cell+Cycle-dependent+Regulation+of+a+Human+DNA+Helicase+That+Localizes+in.">Cell Cycle-dependent Regulation of a Human DNA Helicase That Localizes in.</a> DNA Damage Foci. Mol Biol Cell. 15 (7), 3320-3332 (2004).</li><li>Mittnacht, S. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Control+of+pRB+phosphorylation.">Control of pRB phosphorylation.</a> Curr Opin Genet Dev. 8 (1), 21-27 (1998).</li><li>Mittnacht, S. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+retinoblastoma+protein--from+bench+to+bedside.">The retinoblastoma protein--from bench to bedside.</a> Eur J Cell Biol. 84 (2-3), 97-107 (2005).</li><li>Nybo, K. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=GFP+imaging+in+fixed+cells.">GFP imaging in fixed cells.</a> BioTechniques. 52 (6), 359-360 (2012).</li><li>Schwartz, R. L., Foy, B. D., Phoenix, T. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=.">.</a> Learning Perl - Making Easy Things Easy and Hard Things Possible. , 1-363 (2011).</li><li>Wickham, H. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=.">.</a> ggplot2: Elegant Graphics for Data Analysis. , 978-970 (2009).</li></ol>

The authors have nothing to disclose

The workflow described constitutes a procedure for multiwell perturbance of cells using siRNA, subsequent marker detection and finally use of a series of software-supported steps to facilitate extraction of quantitative data from the resulting fluorescent microscopy images. The approach is focused on the delivery of nuclear and cytoplasmic intensity values for individual cells, which has broad practical application in many cell-based applications. The example data used here was generated in an siRNA screen setting in which two fluorescent assays for G1 phase cell cycle transit are tested and correlated back to a more direct biophysical measure of nuclear DNA content.
The use of a fluorescent DNA stain to image nuclear DNA is an indispensable step in the image segmentation process as it allows identification of individual cells and the resulting &#8216;Nuclei mask&#8217; serves as the starting point to identify corresponding cytoplasmic regions. The GFP-tagged CDK2 reporter, which is stably expressed in the cells, gives a variable yet consistently higher than background signal in the cytoplasm by which this compartment can be delineated. The same analysis pipeline should be applicable to the analysis of protein translocation events using other suitable fluorescence-linked reporters and their response to perturbance. Also, substituting the GFP-CDK2 reporter with cytoplasm-specific fluorescent dyes would allow the alternative use of this algorithm to measure the dimensions of the cytoplasm and the relative sizes of the cells in the images.
Another design consideration in the image segmentation strategy described here is the use of Cell Profiler to deliver integrated intensity values for the DNA quantification. Integration of the intensity values for the nuclear DNA staining data allow for possible variations in nucleus size, and represents a close match for the quantification profiles seen for propidium iodide stained FACS data. However, integrated intensity may not provide an appropriate means to assess protein function where average concentration, exemplified by mean intensity of antigen fluorescence, is more biologically relevant than the integrated total amount of protein (and associated fluorescence) within a cell compartment. Therefore mean intensity values were used for the P-S780 RB1 and GFP data. The option to alter between the two modes (mean or integrated) of data assessment is found on the &#8216;ExportToSpreadsheet&#8217; panel of the Cell Profiler software.
The analysis settings in the 3_channels_pipeline.cppipe file are optimized for the images in the example data set. Analysis of new images sets with this protocol will require that the file names adopt the naming convention described above (Figure 3). Also, sensitivity values to suit the brightness of nuclear DNA staining and thresholds for background intensities in the new image sets may need to be adjusted within the Cell Profiler settings. Given the key role the DNA staining holds for building the various image segmentation masks, the application of correct sensitivity settings for this channel is key to the successful analysis of new image data with the Cell Profiler software. The provided Cell Profiler settings file (3_channels_pipeline.cppipe) contains notes on the most frequently useful parameters for adapting the analysis to new data. These notes are in the text box at the top of the screen in the Cell Profiler main window and include guidance on changing the sensitivity settings and adjusting the number of channels to be analyzed. As indicted in Protocol section 2.8, to test settings for new image data it may be necessary to observe the image segmentation during image analysis by clicking open the &#8216;eye&#8217; icons for each of the &#8216;Identify&#8230;Objects&#8217; protocol steps (Figure S1D). In particular, visualization of the image data through &#8216;IdentifyPrimaryOjects&#8217; will show if the Nuclei mask is correctly identified from images of the DNA staining. On the Cell Profiler software page for the &#8216;IdentifyPrimaryOjects&#8217; module is the threshold correction factor. Trial and error adjustment of this value will fix most nuclear recognition errors. The values balance the DNA channel against the background intensity for each image. Threshold correction factor values hinge around 1, where greater than this is more stringent (good for clear images) and less than 1 is lenient (suited to images with less contrast between staining and background).
The raw output of individual cell data from Cell Profiler can be analyzed in varying ways to suit the needs of other studies. Shown here is the use of a Perl script to apply gates to two of the parameters measured per cell in order to assist extracting biological trends from the data and permit cross-referencing of the identified subpopulations with additional measurements. Although it is equally possible to include elements of gating within the framework of Cell Profiler, the alternative route used here provides greater flexibility and speed, specifically if large data sets need to be assessed. The slowest stage in the post-image acquisition phases of the current protocol is the running of the Cell Profiler software. Cell profiler here is run without imposing gates to produce an un-gated raw data set which can be reanalyzed with the subsequent Perl script more quickly and, if needed, iteratively with different gate values. Not all studies will know in advance the suitable gate values as this may vary with reagents on any given set of data, and potentially over time. It is, therefore, recommended to generate histograms depicting the raw data distribution obtained from Cell Profiler for positive controls and mock-perturbed cells in order to identify suitable gate values for the parameters of interest.
The Perl scripts are written to accept a rigidly defined column structure of data from Cell Profiler and may stop working if a user modifies the number of parameters output by Cell Profiler using the &#8216;ExportToSpreadsheet&#8217; settings. To help implement modification of the settings notes are included within the Perl script files. To see these view the script in a text editor, preferably a programmer&#8217;s text editor set to color code Perl elements (e.g., http://www.activestate.com/komodo-edit). These notes indicate where to adjust the script to adapt to changes in data format. Similar to the Perl scripts, the R-code file provided (analysis.r), containing the instructions for plotting the figures from the image analysis data, can be read in a text editor or RStudio software to see additional notes on use and adaptation. These notes can be supplemented with details on regular expressions and Perl12 and the ggplot213 package for R, both of which form the basis for how the data is read, annotated and plotted, respectively.
New studies using fluorescence microscopy as well as raw data deposited with open source publications are amenable to methods of analysis such as those described here. The very nature of high-content data lends itself to recursive analysis with different analytic emphases depending on the research interests of any given observer. Although the questions that can be asked of the data are limited by the probes originally used, image data can often be meaningfully reanalyzed beyond the scope of the studies which generated them.

The work presented here describes the use of the freely available software Cell Profiler to perform algorithm-guided breakdown of fluorescent microscopy images of adherent cells to identify individual cells and defined subcellular regions. This approach, referred to as image segmentation, allows the subsequent multi-parametric analysis of the imaged cells by quantifying fluorescently labeled markers localized to each cell or subcellular region (referred to as segmented objects). This workflow constitutes a basis for enabling high-content analysis and is intended to serve as a tool that can be further developed and modified to suit multi-parametric, individual cell analyses in laboratories without access to specialized high-content instruments or proprietary software. The files supplied with this manuscript include a test set of relevant raw image data, algorithm settings and supporting scripts to generate the analysis described. The provided algorithm settings for Cell Profiler are optimized for the example data set and the Discussion section details what adjustments may be necessary to enable use of image data from other studies.
Once quantitative data has been extracted using Cell Profiler, different laboratories may have different requirements for how to use the information presented by the individual cell values in the raw data; shown here is one approach by which gates are applied to the raw data for each assay. Using these gates, the data are transformed into binary terms of response, allowing visualization of trends linking different treatments with the subpopulations of cells undergoing response defined by the gates. The gates are set based on observations of the data distributions obtained for appropriate negative and positive controls for each relevant measurement. The use of gates is just one example of how to manage the raw, cell-based measurements. Also shown here is the use of nuclear DNA intensity measurements in their raw form as a continuous range of values in combination with the gated data. Other approaches to managing image analysis data should be considered, depending on the nature of the study; statistical alternatives to using gates for assigning cells to subpopulations have been reported2 and systematic comparisons of strategies to summarize high-content data across large numbers of parameters have been reported3.
High-content analyses of image data have found use in cellular studies of drug-response, reverse genetics and environmental stress signalling4&#8211;6. The merit of high-content analysis stems from the fact that algorithmic analysis of fluorescence microscopy data allows quantitative and spatial parameters to be considered simultaneously across individual cells7. In this way, cellular outcomes for multiple assays can be cross-referenced, differential behavior of assay-defined cell subpopulations can be tracked within experimental conditions and assays can include consideration of morphological variables. The strategies and analysis workflow discussed here, as for other high-content approaches, are capable of delivering multiplexed data that are cross-referenced to individual cells. High-content methods suit studies that generate fluorescent microscope images and are applicable to analysis of data ranging from tens of images produced in low throughput conventional fluorescence-based microscopy through to the thousands of images produced using automated high-content screening platforms.
The workflow is illustrated here with example data from which separate assays are measured in terms of either nuclear fluorescent marker intensities or nuclear/cytoplasmic translocation of a fluorescent reporter protein, respectively. The workflow is flexible in that these assays can be considered separately or in combination depending on each given research question by different investigators. The example data are produced as part of a RNA interference (RNAi) experiment (Figure 1). Small interfering RNA oligonucleotides (siRNA) are used to knockdown specific proteins in HCT116 human colorectal carcinoma cells which result in changes for two fluorescent reporters of cyclin-dependent kinase (CDK) activity. The CDK6-dependent phosphorylation of the nuclear retinoblastoma protein at serine 780 (P-S780 RB1) is assessed by antibody staining. In the same cells, a green fluorescent protein-tagged reporter of CDK2 activity (GFP-CDK2 reporter) is assessed by its nuclear to cytoplasmic ratio where in the absence of CDK2 activity the reporter resides in the nucleus and upon CDK2 activation shuttles into the cytoplasm8. Additionally, the nuclear DNA of each cell is stained using a DNA-intercalating dye, Bisbenzimide, which serves as a means to identify cells and define nuclei borders in the images as well as a measure of DNA abundance providing information on cell cycle position of the cell (Figure 2).
The activities of CDK6 and CDK2 are detectable as cells transit from G1 to S phase of the cell cycle5 and succeed each other9,10 and, as such, close concordance between the two reporters in individual cells is expected. The demonstration dataset used here analyzes as an example the effect of siRNA targets CDK6, retinoblastoma protein (RB1) and a non-targeting negative control (Table 1). Knockdown of CDK6 should elicit both a decrease of the P-S780 RB1 epitope and an accumulation of cells in G1 phase of the cell cycle. The RB1 knockdown serves as a reagent control for the specificity of the phospho-S780 antibody. Fluorescence microscope images from formalin fixed11, fluorescently stained HCT116 tissue culture cells are used for algorithmic image analysis. The resulting numeric data is then used to cross-reference the reporters and gauge the impact of the different knockdown states.
The potential size of the data produced by this type of analysis can present a challenge to normal analysis tools. For example, the individual cell data can be larger than some spreadsheet software will accommodate. Included are Perl scripts which perform simple, highly-repetitive, supervised processing of the data to aid analysis of large datasets. The Perl scripts are written specifically for the output files produced by Cell Profiler, when processing image files with a specific file naming convention (Figure 3), and allow for variable numbers of fields per well to be used in the analysis. It is frequently important to gate individual cell assay data to track trends in cell subpopulations5 and shown here is the use of a Perl script to flag each cell based on a set gate predetermined for each assay type. Also included are optional Perl scripts which summarize the data outcome for individual wells (or conditions), delivering: percentage of cells within the set gate and the mean values of the raw assay scores. The latter, more homogeneous way of viewing the data, is valid where responses affect all or the majority of cells within a well. As discussed above, such assessment is less useful than that afforded by the individual cell data gating where response is confined to a subset of cells within a population.
The utility of the described workflow is not limited to perturbation by siRNA or the marker assays described. Studies have used this approach to assay responses in tissue culture experiments using combinations of siRNA, chemical inhibitors and radiation treatment and for assessment of markers other than CDK6 and CDK2 activity5.
Conceptually, the experimental strategy allows a variety of biologically useful subcellular regions to be automatically registered in individual cells present in fluorescent microscope images. As such, this approach can yield quantitative, multiplexed data revealing biological information that may be missed through techniques that focus on populations rather than individual cells. With minor modifications, the approach and analysis workflow described can yield quantitative, individual cell data for any fluorescence-based assay outputs and cell-biological responses, where quantitative assessment of DNA content, quantification of nuclear or cytoplasmic fluorescence or the shuttling of markers between these two compartments either individually or in a multiplexed manner is of interest. As publishing requirements increasingly tend towards submission of openly accessible raw data, access to and familiarity with free tools for microscopy image analysis such as those described here will also be of direct interest to labs looking to reanalyze published data.

<table border="0" cellpadding="0" cellspacing="0" width="604">
	<tbody>
		<tr height="42">
			<td height="42" style="height:42px;width:216px;">Name of Material/ Equipment</td>
			<td style="width:103px;">Company</td>
			<td style="width:119px;">Catalog Number</td>
			<td style="width:167px;">Comments/Description</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;">AllStars negative control siRNA</td>
			<td style="width:103px;">Qiagen</td>
			<td align="right" style="width:119px;">1027280</td>
			<td style="width:167px;">Negative control siRNA.</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">CDK6 siRNA</td>
			<td style="width:103px;">Dharmacon/ Custom synthesis</td>
			<td style="width:119px;">NA</td>
			<td style="width:167px;">Antisense sequence: 5&#39; CUCUAGGCCAGUCUUCUUCUU</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">RB1 siRNA</td>
			<td style="width:103px;">Dharmacon/ Custom synthesis</td>
			<td style="width:119px;">NA</td>
			<td style="width:167px;">Antisense sequence: 5&#39; GGUUCAACUACGCGUGUAATT</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">5x siRNA buffer</td>
			<td style="width:103px;">Thermo Scientific</td>
			<td>B-002000-UB-100</td>
			<td style="width:167px;">To be diluted with nuclease-free, non-DEPC treated water.</td>
		</tr>
		<tr height="42">
			<td height="42" style="height:42px;width:216px;">Nulcease-free water (non-DEPC)</td>
			<td style="width:103px;">Applied Biosystems</td>
			<td>AM9937</td>
			<td style="width:167px;">For dilution of siRNA buffer.</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:216px;">Hiperfect</td>
			<td style="width:103px;">Qiagen</td>
			<td align="right">301705</td>
			<td style="width:167px;">Transfection lipid.</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">DMEM</td>
			<td style="width:103px;">Life Technologies</td>
			<td align="right">41966052</td>
			<td style="width:167px;">Pyruvate and high-glucose supplemented tissue culture media.</td>
		</tr>
		<tr height="126">
			<td height="126" style="height:126px;width:216px;">96 well tissue culture plate</td>
			<td style="width:103px;">Falcon</td>
			<td align="right">3072</td>
			<td style="width:167px;">Plain, 96-well tissue culture plate for the parallel, compartmentalized mixing of transfection complexes.</td>
		</tr>
		<tr height="189">
			<td height="189" style="height:189px;width:216px;">Packard Viewplate</td>
			<td style="width:103px;">Perkin Elmer LAS</td>
			<td align="right">6005182</td>
			<td style="width:167px;">96 well TC plate with optical base on which cells are transfected, grown, fixed and eventually imaged. Supplied with opaque, adhesive plate seals, which are used during storage of used plates.</td>
		</tr>
		<tr height="42">
			<td height="42" style="height:42px;width:216px;">Breathable membrane</td>
			<td style="width:103px;">Alpha Labs</td>
			<td style="width:119px;">LW2783</td>
			<td style="width:167px;">Sterile, gas-permeable, adhesive membrane.</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">Neutral buffered formalin, 10%</td>
			<td style="width:103px;">Sigma-Aldrich</td>
			<td>HT5012-1CS</td>
			<td style="width:167px;">10% Formalin (4% formaldehyde) fixative used neat in protocol.</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:216px;">Triton X-100</td>
			<td style="width:103px;">Sigma-Aldrich</td>
			<td>X100PC</td>
			<td>Non-ionic detergent</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">Tris</td>
			<td style="width:103px;">Sigma-Aldrich</td>
			<td style="width:119px;">T1503</td>
			<td style="width:167px;">Trizma base (tris(hydroxymethyl)aminomethane)</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:216px;">Tween 20</td>
			<td style="width:103px;">Sigma-Aldrich</td>
			<td style="width:119px;">P2287</td>
			<td style="width:167px;">For plate wash solution.</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:216px;">Anti-P-S780 RB1 antibody</td>
			<td style="width:103px;">Abcam</td>
			<td>ab32513</td>
			<td>Rabbit monoclonal</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">AlexaFluor647 Anti-rabbit</td>
			<td style="width:103px;">Invitrogen</td>
			<td>A21245</td>
			<td style="width:167px;">Highly cross-adsorbed, fluorescently labelled secondary antibody.</td>
		</tr>
		<tr height="42">
			<td height="42" style="height:42px;width:216px;">Hoechst 33342 (Bisbenzimide)</td>
			<td style="width:103px;">Sigma-Aldrich</td>
			<td>B2261</td>
			<td style="width:167px;">Fluorescent, chromatin-intercalating, DNA dye.</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">Permeabilization solution</td>
			<td style="width:103px;">NA</td>
			<td style="width:119px;">NA</td>
			<td style="width:167px;">0.1% Triton X-100 in 50 mM Tris-buffered saline, pH 8.0.</td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:216px;">Plate wash solution</td>
			<td style="width:103px;">NA</td>
			<td style="width:119px;">NA</td>
			<td style="width:167px;">Tris-buffered saline containing 0.1 % Tween-20.</td>
		</tr>
		<tr height="126">
			<td height="126" style="height:126px;width:216px;">Block solution</td>
			<td style="width:103px;">NA</td>
			<td style="width:119px;">NA</td>
			<td style="width:167px;">5% powdered milk in Tris-buffered saline and 0.1% Tween-20. For blocking plate prior to immuno-staining and dilution of antibodies.</td>
		</tr>
		<tr height="42">
			<td height="42" style="height:42px;width:216px;">Cell Profiler 2.1</td>
			<td style="width:103px;">Broad Institute</td>
			<td style="width:119px;"></td>
			<td>http://www.cellprofiler.org/download.shtml</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:216px;">Active Perl Community Edition</td>
			<td style="width:103px;">ActiveState</td>
			<td style="width:119px;"></td>
			<td>http://www.activestate.com/activeperl/downloads</td>
		</tr>
		<tr height="42">
			<td height="42" style="height:42px;width:216px;">R programming environment</td>
			<td style="width:103px;">The R Foundation</td>
			<td style="width:119px;"></td>
			<td>http://www.r-project.org</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:216px;">Rstudio</td>
			<td style="width:103px;">Rstudio</td>
			<td style="width:119px;"></td>
			<td>http://www.rstudio.com/</td>
		</tr>
	</tbody>
</table>

workflow for high-content, individual cell quantification of fluorescent markers from universal microscope data, supported by open source software

Advances in understanding the control mechanisms governing the behavior of cells in adherent mammalian tissue culture models are becoming increasingly dependent on modes of single-cell analysis. Methods which deliver composite data reflecting the mean values of biomarkers from cell populations risk losing subpopulation dynamics that reflect the heterogeneity of the studied biological system. In keeping with this, traditional approaches are being replaced by, or supported with, more sophisticated forms of cellular assay developed to allow assessment by high-content microscopy. These assays potentially generate large numbers of images of fluorescent biomarkers, which enabled by accompanying proprietary software packages, allows for multi-parametric measurements per cell. However, the relatively high capital costs and overspecialization of many of these devices have prevented their accessibility to many investigators.
Described here is a universally applicable workflow for the quantification of multiple fluorescent marker intensities from specific subcellular regions of individual cells suitable for use with images from most fluorescent microscopes. Key to this workflow is the implementation of the freely available Cell Profiler software1&#160;to distinguish individual cells in these images, segment them into defined subcellular regions and deliver fluorescence marker intensity values specific to these regions. The extraction of individual cell intensity values from image data is the central purpose of this workflow and will be illustrated with the analysis of control data from a siRNA screen for G1 checkpoint regulators in adherent human cells. However, the workflow presented here can be applied to analysis of data from other means of cell perturbation (e.g., compound screens) and other forms of fluorescence based cellular markers and thus should be useful for a wide range of laboratories.

Watch this Scientific Journal Video about Workflow for High-content, Individual Cell Quantification of Fluorescent Markers from Universal Microscope Data, Supported by Open Source Software at JoVE.com

Workflow for High-content, Individual Cell Quantification of Fluorescent Markers from Universal Microscope Data, Supported by Open Source Software

Presented is a flexible informatics workflow enabling multiplexed image-based analysis of fluorescently labeled cells. The workflow quantifies nuclear and cytoplasmic markers and computes marker translocation between these compartments. Procedures are provided for perturbation of cells using siRNA and reliable methodology for marker detection by indirect immunofluorescence in 96-well formats.

Advances in understanding the control mechanisms governing the behavior of cells in adherent mammalian tissue culture models are becoming increasingly ...

Research

JoVE Journal

Biology

A Neuronal and Astrocyte Co-Culture Assay for High Content Analysis of Neurotoxicity

High Content Analysis (HCA) assays combine cells and detection reagents with automated imaging and powerful image analysis algorithms, allowing ...

Imaging C. elegans Embryos using an Epifluorescent Microscope and Open Source Software

Imaging C. elegans Embryos using an Epifluorescent Microscope and Open Source Software

Cellular processes, such as chromosome assembly, segregation and cytokinesis,are inherently dynamic. Time-lapse imaging of living cells, using ...

A Quantitative Fitness Analysis Workflow

Quantitative Fitness Analysis (QFA) is an experimental and computational workflow for comparing fitnesses of microbial cultures grown in parallel1,2,3,4. ...

A High-content Imaging Workflow to Study Grb2 Signaling Complexes by Expression Cloning

Signal  transduction by growth factor receptors is essential for cells to maintain  proliferation and differentiation and requires tight control. Signal  ...

Profiling Individual Human Embryonic Stem Cells by Quantitative RT-PCR

Heterogeneity of stem cell population hampers detailed understanding of stem cell biology, such as their differentiation propensity toward different ...

Tracking and Quantifying Developmental Processes in C. elegans Using Open-source Tools

Tracking and Quantifying Developmental Processes in C. elegans Using Open-source Tools

Quantitatively capturing developmental processes is crucial to derive mechanistic models and key to identify and describe mutant phenotypes. Here ...

Open Source High Content Analysis Utilizing Automated Fluorescence Lifetime Imaging Microscopy

We present an open source high content analysis instrument utilizing automated fluorescence lifetime imaging (FLIM) for assaying protein interactions ...

Cellular Redox Profiling Using High-content Microscopy

Reactive oxygen species (ROS) regulate essential cellular processes including gene expression, migration, differentiation and proliferation. However, ...

Open-source Single-particle Analysis for Super-resolution Microscopy with VirusMapper

Super-resolution fluorescence microscopy is currently revolutionizing cell biology research. Its capacity to break the resolution limit of around 300 nm ...

IR-TEx: An Open Source Data Integration Tool for Big Data Transcriptomics Designed for the Malaria Vector Anopheles gambiae

IR-TEx: An Open Source Data Integration Tool for Big Data Transcriptomics Designed for the Malaria Vector Anopheles gambiae

IR-TEx is an application written in Shiny (an R package) that allows exploration of the expression of (as well as assigning functions to) transcripts ...