These scripts are designed for processing and analysis of fluorescence images of live HEK293 cells, heterologously expressing soluble mCherry and YFP(H148Q/I152L)-CFTR channels.
Here we provide information on how to use the MATLAB code for the image analysis described in “High-content assay for precision medicine discovery in cystic fibrosis” (Prins et al., 2019). The instructions will allow you to download a set of images, get a copy of the script and run the analysis to obtain information on CFTR membrane density, and CFTR ion channel function.
Install MATLAB R2017b (https://uk.mathworks.com/products/matlab.html)
the Image Processing Toolbox https://www.mathworks.com/products/image.html
the Optimization Toolbox https://uk.mathworks.com/products/optimization.html
Go to https://github.com/stellaprins/CFTRimg, and download the ZIP folder called CFTRimg-master. Unpack the folder, place it on the Desktop and rename it CFTRimg.
The folder should contain three folders: CFTRimg, Data and Input Files
- The CFTRimg folder contains the scripts and functions that are used to analyse the data
- The Data folder contains sets of example images, suitable for the localisation analysis (local folder), the I- first quenching analysis (quench_Ifirst folder) and the I- last quenching analysis (quench_Ilast folder)
- The Input Files folder contains the scripts that provide image specifications
- Open MATLAB and add the CFTRimg folder and its subfolders to the MATLAB path.
You can find the example data in CFTRimg>Data>example_data_local. The CFTRimg_Local.m script will use the functions in CFTRimg>CFTRimg>functions Local to segment and analyse images as described in Prins et al., (2019). In short, images of cytosolic mCherry are used for identification of single cells using thresholding and a watershed-based segmentation. After background correction, the YFP and mCherry fluorescence intensity of each cell is normalized to the median YFP and mCherry fluorescence intensities in WT-CFTR expressing cells. A band within the border of each cell is defined as the membrane zone, and CFTR membrane density is estimated by dividing the average normalized YFP fluorescence intensity within the membrane zone by the average normalized mCherry fluorescence over the entire cell.
The example_input_local.m file (in Input Files), is customized by the experimenter and contains image specifications necessary for the analysis (for example pixel binning, number of images taken per well, microscope objective and normalisation conditions). Furthermore, it allows the filenames of the images to be matched to the experimental conditions used.
- Open CFTRimg_Local.m located in the CFTRimg folder and run the script (choose Add to path if you have not already added this folder to the path).
- A window will pop up asking you whether you want to save cropped images of segmented cells after the image analysis. Choose either Yes or No.
- Folders to save the results in will be created in the main folder. The workspace and excel files with the results will be saved in CFTRimg>Results>ResultsLocal.
The analysis results can be found on the Desktop in CFTRimg>Results>ResultsLocal
The workspace resulting from running CFTRimg_Local.m contains the structures plate and resultsLocal, containing the analysis results. The structure called plate contains the results per plate, per image. To access the fields with information about plate n, type plate(n) in the command window.
>> plate(1)
> plateStr: 'Plate_15771'
> experimentStr: 'Experiment 1'
> normConditionStr: 'WT 28°C'
> image: [48×1 struct]
Plate(n) contains a structure called image that can be accessed to view the fields with information about each image. To access the information about image m on plate n, type plate(n).image(m) in the command window.
>> plate(1).image(1)
> conditionStr: 'WT 28°C'
> redPath: 'C:\Users\USERNAME\Desktop\CFTRimg\Data\example_data_local\TimePoint_1\local_C02_s1_w2.TIF'
> yelPath: 'C:\Users\ USERNAME \Desktop\CFTRimg\Data\example_data_local\TimePoint_1\local_C02_s1_w1.TIF'
> binning: 0.5000
> boundingBox: [36×4 double]
> cellLocation: {36×4 cell}
> redEntire: [36×1 double]
> redOutside: [36×1 double]
> yelEntire: [36×1 double]
> yelOutside: [36×1 double]
> yelMembrane: [36×1 double]
> yelMembraneAbsolute: [36×1 double]
> yelEntireAbsolute: [36×1 double]
> redEntireAbsolute: [36×1 double]
> redBackground: 0.0147
> yelBackground: 0.0187
> cellN: [37 36]
> yelMembraneNeg: 1
> yelEntireNeg: 1
> redEntireNeg: 0
> deleteNeg: 1
> objective: 60
To access the fields with information about condition number n, type resultsLocal(n) in the command window.
>> resultsLocal(1)
> condition: 'F508del + 10 µM VX-809 28°C'
> normCondition: 'WT 28°C'
> cellLocation: {528×4 cell}
> yelMembrane: [528×1 double]
> yelEntire: [528×1 double]
> redEntire: [528×1 double]
> yelMembraneNorm: [528×1 double]
> yelEntireNorm: [528×1 double]
> redEntireNorm: [528×1 double]
> yelMembraneAbsolute: [528×1 double]
> yelEntireAbsolute: [528×1 double]
> redEntireAbsolute: [528×1 double]
> memDens: [528×1 double]
> logMemDens: [528×1 double]
> localCellN: 528
> yelMembraneNeg: 8
> yelEntireNeg: 8
> redEntireNeg: 0
> deleteNeg: 8
In addition to saving the workspace, running CFTRimg_Local.m results in the generation of three excel files that are saved in Results>ResultsLocal. The tables with field/header descriptions below, can be used to describe the headers in the excel files.
-
fullLocalResult_yyyy_mm_dd_HHMM.xlsx For each cell the rows in the excel sheet contain the following information: experimentStr, plateIdx, condition, normCondition, yelMembrane, yelEntire, redEntire, yelMembraneNorm, yelEntireNorm, redEntireNorm, memDens and logMemDens
-
LocalTable_yyyy_mm_dd_HHMM.xlsx Per plate, per condition, this excel sheet lists: experimentStr, condition, meanLogMemDens, meanRedEntireNorm and meanRedEntire
-
LocalSummary_yyyy_mm_dd_HHMM.xlsx The mean log10ρ on each plate is determined per condition and forms the subsample mean. For every condition, this excel sheet lists: condition, N, mean meanLogMemDens, lower CI, upper CI, cellN, deleteNeg, redEntireNeg, yelEntireNeg and yelMembraneNeg
If you have chosen to save cropped images of segmented cells after the image analysis, images will appear in Results>ResultsLocal>CellImages. The white lines in the images denote the cell segmentation boundary and the ρ is stamped on each image. Descriptions of the fields in structures of the workspace and in the headers of the excel files are listed in the table below.
| field / header | Description |
|---|---|
| plateStr | Plate label as specified in the example_input_local.m file |
| experimentStr | Experiment label as specified in the example_input_local.m file |
| condition | Label of the experimental conditions as specified in the example_input_local.m file |
| normCondition | Label of the condition used for normalisation (in this case “WT 28°C”), as specified in the example_input_local.m file |
| redPath | Filepaths for mCherry images |
| yelPath | Filepaths for YFP images |
| binning | With n×n pixel binning, this field is set to 1/n (in the example_input_local.m file 2×2 binning is used so this field is set to 1/2) |
| objective | 60 for 60× objective power |
| redEntire | The average background corrected mCherry fluorescence intensity inside the cell selection |
| redOutside | The average background corrected mCherry fluorescence intensity outside the cell selection and within the boundingBox |
| yelEntire | The average background corrected YFP fluorescence intensity inside the cell selection |
| yelOutside | The average background corrected YFP fluorescence intensity outside the cell selection and within the boundingBox |
| yelMembrane | The average background corrected YFP fluorescence intensity in the membrane zone |
| yelMembraneAbsolute | The average YFP fluorescence intensity in the membrane zone |
| yelEntireAbsolute | The average YFP fluorescence intensity inside the cell selection |
| redEntireAbsolute | The average mCherry fluorescence intensity inside the cell selection |
| redBackground | The average mCherry background fluorescence intensity |
| yelBackground | The average YFP background fluorescence intensity |
| cellN | Group size of segmented cells ([before selection] [after selection]) |
| yelMembraneNeg | Number of cells with a negative average YFP fluorescence intensity in the membrane zone after background correction |
| yelEntireNeg | Number of cells with a negative average YFP fluorescence intensity inside the cell selection after background correction |
| redEntireNeg | Number of cells with a negative average mCherry fluorescence intensity inside the cell selection after background correction |
| deleteNeg | Number of cells deleted because of negative average YFP and / or mCherry fluorescence intensity after correction for background |
| boundingBox | Pixel indices of boxes around selected cells |
| cellLocation | Column 1: plateStr Column 2: plate number (n) Column 3: image number (m) Column 4: cell number |
| yelMembraneNorm | The average background corrected YFP fluorescence intensity in the membrane zone, normalized to the median YFP fluorescence intensities in WT-CFTR expressing cells on the same plate |
| yelEntireNorm | The average background corrected YFP fluorescence intensity in the entire cell, normalized to the median YFP fluorescence intensities in WT-CFTR expressing cells on the same plate |
| redEntireNorm | The average background corrected mCherry fluorescence intensity in the entire cell, normalized to the median mCherry fluorescence intensities in WT-CFTR expressing cells on the same plate |
| memDens | CFTR membrane density (ρ) defined by yelMembraneNorm/redEntireNorm |
| logMemDens | The common logarithm of CFTR membrane density (log10ρ) defined by log10(memDens) |
| header | Description |
|---|---|
| plateIdx | plate number (n) |
| meanLogMemDens | mean logMemDens |
| meanRedEntireNorm | mean redEntireNorm |
| meanRedEntire | mean redEntire |
| N | number of subsample log10ρ means |
| mean meanLogMemDens | Mean of the subsample log10ρ means |
| lower CI | lower limit of the 95% confidence interval of log10ρ (can only be calculated if N>1) |
| upper CI | upper limit of the 95% confidence interval of log10ρ (can only be calculated if N>1) |
For assessment of CFTR function, two different protocols were used; the I- first protocol (Langron et al., 2017) and the I- last protocol (Langron et al., 2018). You can find example data collected with the I- first protocol in CFTRimg>Data>quench_Ifirst and example data collected with the I- last protocol in CFTRimg>Data>quench_Ilast.
For both protocols, cells were selected based on the mCherry fluorescence by means of thresholding. The YFP fluorescence in the cell selection is corrected for background and the average YFP fluorescence intensity is normalised to the time point before I- addition. For the images collected with the I- first protocol, the maximal influx of I- after the addition of Forskolin (test) or DMSO (control) is determined together with the time point at which I- influx is highest. For the I- last protocol, quenching traces are fit to a mathematical model (as described in Langron et al., 2018), to estimate CFTR conductance (GCFTR).
As for the localisation analysis, the input files example_input_quench_Ifirst.m and example_input_quench_Ilast.m, are customized by the experimenter and provide specifications necessary for image analysis (for example sampling frequency, timepoints of fluid addition and normalisation conditions). These input files also allow the filenames of the images to be matched to the experimental conditions used.
- Open CFTRimg_Quench_Ifirst.m in CFTRimg and run the script
- The resulting workspace (quench_Ifirst_yyyy_mm_dd_HHMM.mat) and an excel file with the results (fullQuenchTimeline_yyyy_mm_dd_HHMM.xlsx) will be saved in Results>ResultsQuench>Ifirst.
- Open CFTRimg_Quench_Ilast.m in CFTRimg and run the script
- The resulting workspace (quench_Ilast_yyyy_mm_dd_HHMM.mat) and excel files with the results (fullQuenchTimeline_yyyy_mm_dd_HHMM.xlsx and fullQuenchFitting_yyyy_mm_dd_HHMM.xlsx) will be saved in Results>ResultsQuench>Ilast.
The analysis results can be found on the Desktop in CFTRimg>Results>ResultsQuench
Both types of quenching analysis, the I- first as well as the I- last image analysis, result in a workspace that contains the structure called plate.
The structure plate contains the analysis results per plate, per well. To access the fields with information about plate n, type plate(n) in the command window.
>> plate(1)
> plateStr: '04-12-17 Plate_15771'
> experimentStr: '04-12-17'
> normCondition: 'WT + VX-770 28°C'
> well: [23×1 struct]
To access the fields with information about well m on plate n, type plate(n).well(m) in the command window.
>> plate(1).well(1)
> condition: 'WT + VX-770 28°C'
> test_control: 'test'
> redPath: {1×2 cell}
> yelPath: {1×70 cell}
> timeline: [5 26 70]
> timeStep: 2
> yelInsideOverT: [70×1 double]
> redInsideNorm: 0.8079
> redInside: 0.0381
> redMaskChange: 0.2063
> maxGrad: 0.1332
> maxGradLoc: 38
Running the quenching analysis generates and saves an excel files in CFTRimg>Results>ResultsQuench>Ifirst for the I- first quenching analysis, and two excel files in Results>ResultsQuench>Ilast for the I- last quenching analysis.
-
fullQuenchTimeline_yyyy_mm_dd_HHMM.xlsx This sheet contains the normalised fluorescence intensity over time (TimePoint_1, TimePoint_2, TimePoint_3, … TimePoint_70) for each well. For each well it also contains the experimentStr, plateStr, condition, test_control, redInside, redInsideNorm.
-
fullQuenchFitting_yyyy_mm_dd_HHMM.xlsx This file is only saved for the I- last quenching analysis. The file contains the fitting results (TimePoint_1, TimePoint_2, TimePoint_3, … TimePoint_70) for each well. For each well it also contains the experimentStr, plateStr, condition, test_control, redInside, redInsideNorm.
| Field | Description |
|---|---|
| plateStr | Experiment label as specified in the example_input_local.m file |
| experimentStr | Label of the experimental conditions as specified in the example_input_local.m file |
| normCondition | Label of the condition used for normalisation as specified in the example_input_local.m file |
| test_control | Addition of forskolin (test conditions) or DMSO (control conditions) |
| timeline | [6 26 70] [timepoint after first fluid addition; timepoint after second fluid addition; last timepoint] |
| timeStep | Time interval (in seconds) between successive images (2 seconds in this study) |
| yelInsideOverT | Normalised fluorescence intensity over time |
| redInside | The average background corrected mCherry fluorescence intensity inside the cell selection |
| redMaskChange | The proportion of area in the cell selection that has no overlap between the first and last mCherry image. |
| maxGrad | The maximal I- influx rate after after the addition of forskolin or DMSO. |
| maxGradLoc | The timepoint at which the I- influx rate is maximal |
| Header | Description |
|---|---|
| label | condition_FSK/DMSO_Platestr |
| FSK/DMSO | Specifies whether there was addition of forskolin (FSK, test conditions) or DMSO (control conditions) |
| y_0 y_3 y_5 ... y_39 |
y_0: normalised fluorescence intensity 0 seconds after the addition of I- y_3: normalised fluorescence intensity 3 seconds after the addition of I- y_5: normalised fluorescence intensity 5 seconds after the addition of I- y_5: normalised fluorescence intensity 5 seconds after the addition of I- ... y_39: normalised fluorescence intensity 39 seconds after the addition of I- |
| error ratio (free/fixed) | Error_free / Error_fixed |
| G_free | Estimated steady-state CFTR conductance (model with four free parameters) |
| G_free_norm | G_free / redInsideNorm |
| Vm_free | Estimated membrane potential at time of I- addition (model with four free parameters) |
| G_trans_free | Estimated transient endogenous anion conductance (model with four free parameters) |
| TAU_trans_free | Estimated exponential decay time constant for transient endogenous anion conductance (model with four free parameters) |
| Error_free | The sum of squared residuals between the measured and predicted normalised fluorescence (model with four free parameters) |
pred_y_free_0 pred_y_free_3 pred_y_free_5 ... pred_y_free_39 |
Using the model with four free parameters: y_0: predicted normalised fluorescence intensity 0 seconds after the addition of I- y_3: predicted normalised fluorescence intensity 3 seconds after the addition of I- y_5: predicted normalised fluorescence intensity 5 seconds after the addition of I- ... y_39: predicted normalised fluorescence intensity 39 seconds after the addition of I- |
| G_fixed | Estimated steady-state CFTR conductance (model with two free parameters) |
| G_fixed_norm | G_fixed / redInsideNorm |
| Vm_fixed | Estimated membrane potential at time of I- addition (model with two free parameters) |
| G_trans_fixed | Transient endogenous anion conductance fixed to average value measured for DMSO controls |
| TAU_trans_fixed | Exponential decay time constant for transient endogenous anion conductance fixed to average value measured for DMSO controls |
| Error_fixed | The sum of squared residuals between the measured and predicted normalised fluorescence (model with two free parameters) |
pred_y_fixed_0 pred_y_fixed_3 pred_y_fixed_5 ... pred_y_fixed_39 |
Using the model with two free parameters: y_0: predicted normalised fluorescence intensity 0 seconds after the addition of I- y_3: predicted normalised fluorescence intensity 3 seconds after the addition of I- y_5: predicted normalised fluorescence intensity 5 seconds after the addition of I- ... y_39: predicted normalised fluorescence intensity 39 seconds after the addition of I- |
Prins, S., Langron, E., Hastings, C., Hill, E., Stefan, A.C., Griffin, L.D., Vergani, P., (2019). High-content assay for precision medicine discovery in cystic fibrosis. bioRxiv 631614; doi: https://doi.org/10.1101/631614
Langron, E, Simone, MI, Delalande, CMS, Reymond, JL, Selwood, DL, Vergani, P (2017). Improved fluorescence assays to measure the defects associated with F508del-CFTR allow identification of new active compounds. Br J Pharmacol 174: 525– 539.
Langron E, Prins S, Vergani P (2018). Potentiation of the cystic fibrosis transmembrane conductance regulator by VX-770 involves stabilization of the pre-hydrolytic O1 state. Br J Pharmacol 175: 3990–4002