CalQuo: automated, simultaneous single-cell and population-level quantification of global intracellular Ca2+ responses

Detecting intracellular calcium signaling with fluorescent calcium indicator dyes is often coupled with microscopy techniques to follow the activation state of non-excitable cells, including lymphocytes. However, the analysis of global intracellular calcium responses both at the single-cell level and in large ensembles simultaneously has yet to be automated. Here, we present a new software package, CalQuo (Calcium Quantification), which allows the automated analysis and simultaneous monitoring of global fluorescent calcium reporter-based signaling responses in up to 1000 single cells per experiment, at temporal resolutions of sub-seconds to seconds. CalQuo quantifies the number and fraction of responding cells, the temporal dependence of calcium signaling and provides global and individual calcium-reporter fluorescence intensity profiles. We demonstrate the utility of the new method by comparing the calcium-based signaling responses of genetically manipulated human lymphocytic cell lines.

Calcium triggering protocol. All cells were labeled with 4μM Fluo-4 AM (F-14201; Invitrogen, Paisley UK) for 30min at room temperature with 2.5mM probenecid (P-36400; Invitrogen, Paisley UK) in RPMI (Sigma-Aldrich, UK) without supplements. Cells were then washed in HBS (51558; Sigma, UK) and medium changed to HBS containing 2.5mM probenecid before adding to antibody-coated slides or to SLBs. Cells were imaged using 10x objective and a NA=0.45 on a spinning disk confocal microscope (Carl Zeiss AG, Overkochen, Germany) under 488 nm laser excitation, with an exposure time of 350 ms and a time between frames of 500 ms for 840 frames.

Protein expression. Lck and TCRβ expression was quantified by FACS using a Beckman Coulter CyAn
Analyser. For surface labelling of TCR complex an anti-CD3ε-alexa647 was used (purified from hybridoma supernatant, clone UCHT1 and labeled with antibody labeling kit from Molecular Probes, Invitrogen). For the intracellular staining of Lck an anti-Lck antibody (clone 73A5, Cell Signalling) was used. Lentiviral infected JCaM 1.6 T cells were fixed in 1% formaldehyde for 15 minutes at room temperature. Cells were then incubated with appropriate dilutions of the antibody. Primary antibody was detected using a donkey anti-rabbit-APC (Molecular Probes, Invitrogen) antibody. Jcam 1.6 cells expressing Lck-Halo were also stained with HaloTag® TMR ligand (Promega) as per manufacturer recommendation.
Calcium triggering analysis. CalQuo Software was used to detect cell landing events on protein-coated glass surfaces and record fluorescence intensities for each recorded frame. Sharp changes in fluorescence intensity above background levels (>3 fold) were indicative of calcium releases associated with TCR triggering in T-cells. Time lapse between "landing" and "triggering" events were also obtained directly from the CalQuo output.

CALQUO SOFTWARE
The CalQuo Software is written in the latest release of MATLAB version R2015a (Mathworks, UK) and was licensed with Isis Innovation (isis-innovation.com) permitting reuse within proprietary software provided all copies of the licensed software include a copy of the Isis License terms and the copyright notice.

The
CalQuo Software package consists of two programs 'CalQuo_masterfile.m' and 'CalQuoAnalysis_masterfile.m'. 'CalQuo_masterfile.m' reads multiple 8bit stack-files in tif-format from a userdefined folder. The image stacks can comprise different pixel sizes and cell densities but need to have the same number of time frames. Subsequently, 'CalQuo_Analysis.m' further analyses the output data in the MATLAB workspace of 'CalQuo_masterfile.m'. Relevant parameters for software control and selection of appropriate signaling response functions are edited in 'CalQuo_parameters.m' present in the main folder. This parameter file is independently read by both programs 'CalQuo_masterfile.m' and 'CalQuoAnalysis_masterfile.m'. This way, the user can dynamically adjust and edit relevant parameters after the raw-data were read in and workspaces were calculated. It is therefore recommended to save the original workspaces after raw-data reading to then optimize analysis parameters without the need to pre-process the raw images again. The output of CalQuo is organized in file-structures and can be found in the MATLAB workspace saved as the structure QUANTDATA including the sub-structure STATISTICS and files. STATISTICS comprises the relevant parameters: number of triggering cells including the averaged response function, the time-points of signaling, and the decay times of the calcium signal. The parameters are given for the individual cells as well as an average over all cells (or multiple files) along with standard statistical parameters such as standard deviations. The corresponding values for the individual files can be found in the structure files. 'CalQuo_masterfile.m' uses feature recognition (FR, MATLAB function 'findfeature2d', see http://people.umass.edu/kilfoil/downloads.html) and the novel distance regularized level set evolution (DRLS) algorithms to segment the rawdata image stacks and detect the cell features such as the signaling response functions as detailed in [16][17][18]. The user can run the cell feature detection in two different modes: the first mode uses the feature edge for determination of the cell's location and the second mode uses a rough estimation of the cell periphery, a more robust approach for cells drifting within the time-coarse of the experiment. In practice, the FR algorithm splits each raw-data image in the image stack into multiple sub-regions where it recognizes the cell features. Subsequently, the DRLS algorithm finds the precise location and edges of the cell features. The users can decide to exclude very bright and fluorescence saturated cells with the feature threshold parameter because the curves can alter the average response behavior. For example, one can set the 'feature_thresh=0.98' to deselect only the 2% brightest cells in the image stack. The feature recognition function requires the user to set three parameters that can be directly determined from the raw data. The first parameter is the typical 'feature_size' in pixels (i.e. the cell size), the second parameter is the approximate 'subregion_size' in arbitrary units, and the third parameter is the 'feature_thresh' in percentage intensity units. CalQuo's DRLS algorithm uses a dynamically shrinking polygon to find the feature edges and then calculate their intensity profiles in time. For this, the user is asked to define an approximate minimal and maximal size of the inner and outer edge of the feature cells with the two parameters 'iter_inner 'and 'iter_outer'.

SOFTWARE CONTROL
Optimal DRLS and FR parameters allow sufficient image segmentation and feature recognition but are limited in the selection of representative calcium response curves. The software does not include dynamic feature tracking and therefore relies on minimal movements of the cells once they have been detected. In the section 'Parameters software control' in the function 'CalQuo_parameters.m', the user is asked to define a name of the experiment and the frame-rate as well as the first and last frame number for the data analysis. The parameter 'firstframenumber' defines the frame at which the cell features are recognized and further analyzed by the FR and DRLS algorithms. Note, the parameter 'firstframenumber' is defined as the number of frames minus one while the second parameter 'lastframenumber' equals the total number of frames. If the parameter 'lastframenumber' is chosen to be smaller than the total number of frames, the remaining frames are excluded from the analysis.

CALIBRATION AND SELECTION OF CALCIUM RESPONSE FUNCTIONS
In the section 'Parameters Profile control' in the function 'CalQuo_parameters.m', the user is asked to set the five parameters alpha, beta, gamma, delta, and epsilon to sufficiently calibrate the selection automation of the software (Fig. S1). All parameters can be read off and dynamically adjusted response curves by running the 'CalQuoAnalysis_masterfile.m' multiple times until readout data such as the signaling fractions and response curves no longer improves. The parameter alpha defines the ratio of fluorescence intensities between the landing plateau and background signal in arbitrary units at early time-points prior landing. The parameter beta equals the approximate intensity value at the frame of calcium triggering in the units counts where the fluorescence peak originates. The parameter gamma defines the minimum ratio of fluorescence intensity of the maximum peak and the average landing intensity at long times in arbitrary units. This parameter can be deactivated by setting it to zero. The parameter delta defines the minimum fluorescence intensity ratio of the last and first ten frames in arbitrary units. The last parameter epsilon defines the minimum distance between landing and triggering time in the units' seconds.

Running the CALQUO SOFTWARE
'CalQuo_masterfile.m' can be started by typing its name into the MATLAB command line and pressing enter. Subsequently, 'CalQuoAnalysis_masterfile.m' is run by typing its name into the MATLAB command line and pressing enter as well. All relevant output parameters are saved to the workspace as described above. The distribution of triggering times pops up in form of a histogram in counts as a function of time in seconds and the averaged response curves are presented in graph showing normalized intensities as a function of time in seconds. Fig. S1: Selection parameters for response curves R(t). The five parameters α, β, γ, δ, and ε allow the user to select the response functions. . we determined for each cell absolute (i.e. un-normalized) fluorescence intensity levels of the basal and maximum signal, I0 and dI = Imax -I0, respectively. Both the peak (a) and the basal levels (b) of the response curves dI(t)/I0 did not change for the different experimental conditions. Error bars = s.d.m. from averaging over all investigated cells (for total numbers of cells see Table 1).

Fig. S3
: Average over all response curves I(t) in Jurkat T-cells generated from the image data with increasing pixel size, resulting in less pixels per cell, as labeled. I(t) was hardly affected by the number of pixels per cell at our experimental setup (300 pixels per cell in our case with the 10x microscope objective). This result indicates that a magnification of down to 5x would have been sufficient to accurately detect the calcium responses. Error bars = s.d.m. resulting from averaging over N = 366 cells.

SUPPORTING MOVIES
Supplementary Movie S1. Time-lapse images of Fluo4 fluorescence (maximum projection over 780 frames) detected for Jurkat T-cells when landing on microscope cover glass coated with αCD3ε and αCD28 antibodies. The cells were pipetted onto the culture medium of the imaging dish and then imaged. The majority of cells stopped movement and flashed brightly when touching the glass surface indicating calcium triggering. Scale bar: 100 μm. Total duration: ~6 min.
Supplementary Movie S2. Time-lapse images of Fluo4 fluorescence (maximum projection over 780 frames) detected for J.Cam1.6 T-cells when landing on microscope glass coated with αCD3ε and αCD28 antibodies. The cells were pipetted onto the culture medium of the imaging dish and then imaged. The majority of cells stopped movement but did not flash when touching the glass surface, indicating the absence of calcium release. Scale bar: 100 μm. Total duration: ~6 min.
Supplementary Movie S3. Time-lapse images of Fluo4 fluorescence (maximum projection over 780 frames) detected for J.Cam1.6-wthLCK T-cells when landing on microscope glass coated with αCD3ε and αCD28 antibodies. The cells were pipetted onto the culture medium of the imaging dish and then imaged. The majority of cells stopped movement and flashed when touching the glass surface, indicating calcium triggering. Scale bar: 100 μm. Total duration: ~6 min.