Single cell organization and cell cycle characterization of DNA stained multicellular tumor spheroids

Abstract

Multicellular tumor spheroids (MCTSs) can serve as in vitro models for solid tumors and have become widely used in basic cancer research and drug screening applications. The major challenges when studying MCTSs by optical microscopy are imaging and analysis due to light scattering within the 3-dimensional structure. Herein, we used an ultrasound-based MCTS culture platform, where A498 renal carcinoma MCTSs were cultured, DAPI stained, optically cleared and imaged, to connect nuclear segmentation to biological information at the single cell level. We show that DNA-content analysis can be used to classify the cell cycle state as a function of position within the MCTSs. We also used nuclear volumetric characterization to show that cells were more densely organized and perpendicularly aligned to the MCTS radius in MCTSs cultured for 96 h compared to 24 h. The method presented herein can in principle be used with any stochiometric DNA staining protocol and nuclear segmentation strategy. Since it is based on a single counter stain a large part of the fluorescence spectrum is free for other probes, allowing measurements that correlate cell cycle state and nuclear organization with e.g., protein expression or drug distribution within MCTSs.

Figure 1

On-chip formation, culture, staining and imaging of MCTSs. The microwell plate (a) consists of a 22 × 22 × 0.3 mm³ silicon wafer with 100 etched microwells (350 × 350 µm²) bonded to a 170 µm thick coverglass. Ultrasonic radiation forces, used in microwell cell aggregation, were induced in a device (b) where the chip was placed on a transducer (1) with a piezo ceramic plate (2) connected to a SMB connector (3). The microwell chip was secured against the piezo by plastic (4) and aluminum (5) frames mounted with springs (6) and nuts (7). The workflow for MCTS culture and processing (c) starts with seeding cells (I) which sedimented into the microwells (II). Transducer actuation for 24 h induces ultrasonic radiation forces acting on the cells in each microwell causing aggregation (III). After 24 h of active ultrasound culture, the chip is further kept in passive culture for 0–72 h (IV) before on-chip staining, clearing (V) and imaging (VI).

Figure 2

Nuclear segmentation of confocal z-stacks. The full volume of DAPI stained and Histodenz cleared A498 MCTSs was acquired by confocal microscopy (3D rendering and optical section at z = 84 µm) (a). Outline of an in-house developed MATLAB script (b) used to segment all individual nuclei (c) for further volumetric and fluorescence intensity analysis. Scalebars are 20 µm.

Figure 3

Correlation between cell cycle and integrated DAPI fluorescence in segmented nuclei. The cell cycle phases G1 (red), transition between G1/S (yellow) and G2/S (a) can be studied with the FUCCI construct which was transfected in A498 MCTSs (N = 40) (b). Based on integrated red and green intensity within each segmented nucleus, cells were classed as G1, G1/S or S/G2 positive (c). The FUCCI positive distribution of normalized integrated DAPI intensity per nucleus (d) could then be divided based on cell cycle position (e–f). Normalized DAPI intensity threshold against percentage of false positive rate (FPR) (red line), true positive S/G2 rate (TPR) (green line) and the difference between TPR and FPR (blue line) (g). The optimal threshold (1.25) is indicated at the TPR-FPR maximum (dashed orange line).

Figure 4

A machine learning-based support vector machine (SVM) trained on volumetric data and integrated DAPI intensity improves S/G2 classification. Density-scatter plots of normalized DAPI intensity against volume for each nucleus in G1 (red), G1/S (yellow) and S/G2 (green) (a). Volume boxplots for G1, G1/S and S/G2 showing 25th and 75th percentiles with median marked with red lines (b). Whiskers shows the furthest observation within 1.5 times the interquartile length away from the box edge and outliers are indicated by blue dots. Pairwise significance determined by Mann–Whitneys U-test and overall significance by Kruskal–Wallis (***p < 0.0001). A SVM was trained on the integrated DAPI intensity in combination with volumetric data (c) from FUCCI positive cells (N = 1270). Each volumetric parameter for a nucleus (blue) was normalized against the mean properties of the 5 closest nuclei (r_1-5, orange) to limit spatial SMV dependencies (d). Scatterplots of integrated DAPI intensity per nucleus against volume shows correctly classified G1 (red) and S/G2 (green) cells as dots and misclassified cells as orange crosses for unseen test data (e). Confusion matrix describing the SVM classification performance shows the correctly (green) and incorrectly (red) classified cells, recall (true positive and false negative rate), precision (positive predictive value and false discovery rate) and overall accuracy (blue) for unseen test data (f).

Figure 5

Spatial distribution of S/G2 positive cells in A498 MCTSs cultured for 24- and 96-h. Normalized integrated DAPI intensity per nucleus data of segmented nuclei from A498 MCTSs cultured for 24 h (N = 10, orange) and 96 h (N = 10, blue) (a). The integrated intensity and volumetric parameters were used to classify G1 and S/G2 cells (percentages in each histogram). The surface of each MCTS was approximated and the distance between each nucleus and the MCTS perimeter was measured and placed in concentric layers automatically using MATLAB (b). The number of S/G2 positive cells (c) and fraction of S/G2 positive cells (d) was calculated for each concentric 10 µm wide layer, from 0 to 60 µm, for each MCTS. Bar plot distributions show mean and standard deviation for 24 (orange) and 96 (blue) hours old MCTSs. Scalebar is 20 µm.

Figure 6

Nucleus volume and local cell density with nucleus-MCTS perimeter distance as a continuous variable. Scatter plots shows the nucleus-MCTS perimeter distance, as a continuous variable (a), against nucleus volume (b) for A498 MCTSs cultured for 24 h (orange dots) and 96 h (blue dots). The alignment angle α, calculated as the angle between the MCTS center-nucleus center vector (r₁) and the plane spanned by the major (e₁) and intermediate (e₂) eigen vectors of the ellipsoid with the same second central momentum as the nucleus (d), is shown in the scatter plot against the nucleus-MCTS perimeter distance (e). The nuclear aspect ratio was studied by measuring the ratio between the major axis (p_major) and minor axis (p_minor) lengths (g) plotted against nucleus-MCTS perimeter distance (h). The nucleus-MCTS perimeter distance dependent trend in the scatter plots (red line with crosses) is calculated as the mean and median within 5 µm wide concentric layers for the nucleus volume, alignment angle and major/minor axes ratio respectively (b,d,h). Dotted line indicates the standard deviation and median absolute deviation. Box plot charts summarizes the data side by side for easier comparison (c,f,i). Significance was tested with two-way ANOVA followed by Tukey’s post-hoc multiple comparison (**p < 0.001). Scalebar is 20 µm.