Figure 1.
Our compound screening concept is based on a simple cell culture platform optimised for 3D spheroid cultures complemented with an easy to use proprietary image analysis program and data analysis tools.
(A) ibidi Angiogenesis μ-slides and μ-plates have a unique well-in-a-well design that facilitates 3D cell culturing between two layers of extracellular matrix on a very narrow focal plane. (B) Time and operation schedule for a typical compound screen including all the major steps from cell seeding to image analysis and visualisation of morphometric responses.
Figure 2.
Flow diagram illustrating key functionalities of the AMIDA image analysis program.
(A) Flowchart presenting four main steps in image analysis. After parameterisation all numerical data is written into an Excel file. (B) An overview of AMIDA's simple user-friendly interface and its basic operations.
Table 1.
List of mathematical preliminaries utilized in AMIDA.
Table 2.
List of morphological parameters implemented in AMIDA.
Figure 3.
Evaluation of key parameters analysed by AMIDA.
(A: left panel) Six representative PC-3 spheroids, all treated with different compounds in order to manipulate the morphology, stained for viable cells (Calcein AM) and imaged with spinning disk confocal microscope (5x objective). (A: right panel) The same spheroids segmented with AMIDA. The table in A shows numerical values appointed by AMIDA for selected morphological features. (B) Representative panel of PC-3 spheroids with red dots added by image manipulation in certain number and distribution to exemplify the power and preciseness of RGB functions (B: table).
Figure 4.
Exemplary screen based on the PC-3 spontaneous invasive transformation model.
PC-3 spheroids were treated with 19 compounds mainly targeting integrity, function and organization of the actin cytoskeleton. 172–424 multicellular structures for each treatment were analysed with AMIDA program. (A) A morphometric heatmap showing standardized differences in medians between the treatments and the control for 15 morphological parameters and all 19 compound treatments. Morphological responses clustered into three functional groups. Increasing cytotoxicity, measured by the AreaRatioR parameter – based on presence of dead cells stained with ethidium homodimer - is indicated by the red gradient arrow. (B) Correlation map (nonparametric Spearman) indicating the similarity (positive correlation, red) or dissimilarity (negative correlation, blue) for 21 of AMIDAs morphometric parameters. (C) Bonferroni-corrected and Mann-Whitney U-test filtered morphometric heatmap (threshold p>0.05) focusing on four selected, most informative parameters (AppIndex, AreaRatioR, Roundness, Area). The graph highlights compounds causing mainly growth-inhibition and cytotoxicity (group II), and those that enhance spheroid symmetry and reduce number of invasive protrusions (group I). (D) The image panel shows representative, segmented PC-3 spheroids for groups I and II, compared to DMSO and paclitaxel controls, after six days of drug treatment.
Table 3.
List of compounds used in exemplary screens.
Figure 5.
Histograms showing the distribution of morphological response data in the exemplary screen.
The data is shown for three key parameters, Area (representing spheroid size), Roundness and AppIndex (representing symmetry), and for three experimental compounds each one representing one of the response groups (DMSO: control, NF023: no response, EHT-1864: growth inhibition, gallein: anti-invasive).
Figure 6.
Validation of morphological responses with 9 additional prostate and 3 breast cancer cell lines.
The heatmaps illustrate changes in spheroid growth (A: Area) and general symmetry (B: Roundness) in response to the 19 compound treatments (5031–16415 multicellular structures analysed for each cell line). (A) CCG-1423, KH7, latrunculin A and narciclasine are preferentially cytotoxic and/or antiproliferative compounds across all cell lines, as highlighted by red boxes. Paclitaxel, at a concentration of 5 nM, shows partial cytotoxic/antiproliferative effects only in some of the cell lines. (B) Effects of mainly anti-invasive compounds IPA3 or NSC23766 were reproducible in many of the spontaneously invasive (or branching) cell lines PC-3, PC-3M Pro4, and RWPE-1. (C) Images segmented and analysed with AMIDA. Effective anti-invasive functions of the compounds IPA3, BPIPP and NSC23766 against the most aggressive, motile and invasive cell lines PC-3M, ALVA31, MDA-MB-231 (both SA and parental ATCC), and RWPE1. The extremely invasive PC3-derivative ALVA31 was not affected, however. (D) Blebbistatin and Y-27632 show invasion-inducing function in spheroids formed by the LNCaP and DU145 lines which typically form round spheroids and lack invasive properties.
Table 4.
Summary of drug treatments of PC-3 cells in monolayer and organotypic culture.