Production and Multi-Parameter Live Cell Fluorescence Lifetime Imaging Microscopy (FLIM) of Multicellular Spheroids (2024)

Choosing the appropriate spheroid formation method
The selected spheroid formation method can greatly influence spheroids' size, shape, cell density, viability, and drug sensitivity (Figure 2). Previously, the effects of multiple high-throughput (SphericalPlate 5D, lab-made micromolds, and MicroTissue molds) and the 'medium throughput' low attachment (Biofloat and Lipidure-coated 96-well plates) methods were compared on spheroids viability and oxygenation²⁹.

Here, different formation methods result in spheroids of different sizes, even with the same initial concentration of 500 cells/ spheroid. HCT116 spheroids formed in low attachment 96 well plates were significantly larger compared to high-throughput methods after 5 days of formation (Figure 2 and Table 2). Furthermore, low attachment methods led to the evident development of a necrotic core detected by the propidium iodide staining (red), while spheroids generated with other methods (MicroTissue, lab-made molds, and Sphericalplate 5D) demonstrated the diffused distribution of dead cells across the spheroid body (Figure 2A).This difference in cell viability across the volume of the spheroid can also affect oxygen diffusion through the media, which in turn can be affected by O₂ partial pressure, consumption rate, temperature, and O₂ solubility. Additionally, in high throughput methods, nutrients get depleted, and waste products accumulate faster as a larger number of spheroids are within the same volume andtherefore require more frequent media exchange (Figure 2A).

The overall spheroid oxygenation was measured using the recently characterized and validated MMIR1 probe²⁹, with a detailed protocol previously reported²⁵. Oxygenation could only be compared between the high-throughput methods and low-attachment method separately (Figure 2B),as the signal-to-noise ratio is different in all compared spheroid formation methods. The Spherical plate 5D spheroids showed lower overall oxygenation compared to the MicroTissue (P-value = 0.0697) and the lab-made micromolds (P-value = 0.0005) (Figure 2C). Spheroids produced with the Spherical plate 5D are grown at a bigger distance to the media air interface surface compared to the agarose mold methods, causing a slower O₂ delivery in the static culture. This hypothesis is valid as oxygenation values between both low attachment plates and, additionally, between both agarose mold methods are statistically similar (Figure 2D).

Figure 2: Spheroid formation methods affect morphology, size, viability, and oxygenation. (A) HCT116 spheroids (initial seeding density 500 cells) were grown for 5 days prior to 1 h-long staining with Propidium Iodide (cell necrosis, red, 1 µg/mL) and Calcein Green-AM (viable cells, green, 1 µg/mL). Low-attachment formation methods produce bigger spheroids containing a necrotic core (red). Scale bar is 100 µm. (B) Oxygenation in HCT116 spheroids (initial seeding of 500 cells with addition of 20 µg/mL MMIR1 during spheroid formation, 6 days) produced by different methods showed more oxygenated spheroids with the MicroTissue and lab-made micromolds than the SphericalPlate 5D. (C) Overall spheroid oxygenation calculated by the ratio measurements. (D) Spheroids size in area square after 5 days. Results show the average ± SEM of 6-16 spheroids. 5D = SphericalPlate 5D, MM = Micromold method, MT = MicroTissue method. ***P-value < 0.001 and ****P-value <0.0001 Please click here to view a larger version of this figure.

Table 2: Different spheroid formation methods lead to differences in spheroid size, measured by area square.

When selecting a specific spheroid formation method for the experiments, it must be kept in mind that the addition of nanoparticles can influence the spheroid's shape. When using the lipid-based coating with round bottom microwells, probe precipitation can occur, leading to the formation of "satellite spheroids" or non-circular spheroids (Figure 3). The addition of the O₂ probe to the commercial Biofloat plate results in a non-compact spheroid periphery, suggesting the potential interference of the nanoparticles with the microplate coating. The morphology of the spheroids can also be affected by the dust attached to the pipet tips. Cells in suspension can stick to the dust, hereby affecting the shape. Therefore, it is recommended to use refillable pipette tips already supplied in a rack.

Figure 3: Transmission light microscopy images with examples of non-ideal spheroid formation. Incorrect storage of nanoparticles can cause probe precipitation, interfering with spheroid formation. Examples of interference of probe and dust. Scale bar is 100 µm. Please click here to view a larger version of this figure.

Clearly, spheroid formation methods affect the spheroid's size, morphology, and viability, and lead to differences in their oxygenation. The high throughput method with agarose molds (e.g., commercially available, such as MicroTissue mold) is compatible with most experiments. It results in a more reproducible shape, does not induce nanoparticle precipitation, is reusable, and can be compatible with direct follow-up microscopy. However, for large spheroids (>800 µm) and long-term culturing, the low attachment method is recommended as high throughput methods would require frequent media exchange. In such a case, the nanoparticle precipitation problem and its influence on spheroid formation can be avoided by the overnight pre-staining of the 2D monolayer culture before spheroid formation (Figure 4).

Figure 4: Suggested algorithm for selecting the spheroid formation method for live fluorescence microscopy. Please click here to view a larger version of this figure.

Multi-parameter FLIM of spheroidsNAD(P)H imaging via two-photon FLIM microscopy reveals metabolic heterogeneity in HCT116 spheroids
A prominent autofluorescence marker in label-free metabolic imaging, NAD(P)H shows a shorter lifetime for glycolysis and a longer lifetime for OxPhos, enabling deciphering the spatial distribution of metabolic activity in such 3D models as spheroids and organoids³¹. To conveniently analyze the NAD(P)H distribution in HCT116 spheroids, a phasor-based quantitative method for autofluorescence NAD(P)H analysis was used, revealing glycolysis and oxidative phosphorylation (OxPhos)-linked states. While fluorescence intensity images of HCT116 spheroids did not reveal significant differences between spheroids, fast FLIM imaging of NAD(P)H revealed a clear difference within the spheroid optical sections. In fast FLIM images, spheroids with an internal area displaying a shorter fluorescence lifetime were marked with a white circle, identifying this as a glycolytic core (Figure 5A). Consequently, the spheroids were divided into two groups, the glycolytic core group and the non-glycolytic core group, based on the presence or lack of a glycolytic core. A centroid-based phasor analysis was employed to statistically analyze the differences between two distinct groups and measure the distance from the centroid of the pixel cluster to the free NAD(P)H point. As illustrated in Figure 5B, there was an obvious pixel cluster difference between the glycolytic group and the non-glycolytic group on the phasor plot. To accurately measure the distance, the free NAD(P)H point on the phasor plot was precisely marked using the FLIM module in LAS X software, and then exported into FIJI software for the exact coordinates' determination. Subsequently, following protocol 2.3, centroid coordinates were measured in each group's phasor plot using FIJI software. The distance (D) from the centroid to free NAD(P)H point, using the Pythagorean theorem, was calculated hereby facilitating analysis of NAD(P)H profiles to distinguish OxPhos spheroids and spheroids with a glycolytic core. The results, as displayed in the boxplot, show the significant distance differences between the glycolytic core spheroids and the non-glycolytic core group (***P-value < 0.001) (Figure 5C). This result was consistent with fast FLIM imaging, demonstrating that this phasor analysis method is suitable for quantitative analysis of the NAD(P)H divergence of spheroid populations. Moreover, to validate if this approachwas applicable to compare the distance between the two spheroid groups, a linear regression analysis for centroids from both groups and free NAD(P)H point was performed (Figure 5D). The result showed that the centroids are aligned with free NAD(P)H, with a high coefficient of correlation (R² = 0.997), validating the feasibility of comparing the distance to infer the NAD(P)H difference between spheroid groups.

Figure 5: Phasor analysis of HCT116 spheroids made using low-attachment plates: glycolytic and non-glycolytic cores comparison of the NAD(P)H distribution via two-photon FLIM. Acquisition parameters: laser intensity: 15%, resolution: 512 x 512 pixels, ex. 741 nm/em. 411-491 nm.(A) The fluorescence intensity and fast FLIM images display NAD(P)H distribution in HCT116 spheroids, distinguishing spheroids with glycolytic core (left) from those without (right). White circle in the fast FLIM image displayed a shorter lifetime area, defining a glycolytic core. Initial seeding density: 50 cells (left) and 500 (right), incubation time: 6 days, scale bar: 50 µm. (B) The methodology for measuring the distance from the centroid of HCT116 spheroids to free NAD(P)H point on phasor plots, using LAS X FLIM module and FIJI software across two distinct spheroid groups (D: distance). For HCT116 spheroids, a wavelet filter and threshold value 10 were employed for the phasor plot. (C) Boxplot showing the difference of distance from the centroid to free NAD(P)H point, comparing glycolytic core and non-glycolytic core groups (***P < 0.001). (D) Linear fitting of phasor plot centroids from glycolytic core, non-glycolytic core groups, and free NAD(P)H theoretical coordinates demonstrated accurate linear alignment with R² = 0.997, allowing for direct comparison of distance difference between glycolytic core and non-glycolytic core spheroids. Please click here to view a larger version of this figure.

Multiplexed analysis of FAD autofluorescence and hypoxia in human dental pulp stem cell (DPSC) spheroids
Dental pulp stem cell spheroids are an attractive experimental tool for the biofabrication of different tissues, including osteoblasts^25,66,67. However, their viability and metabolism are rarely studied. These stem cell-derived spheroids display rather small sizes (<200 µm), bright green autofluorescence from FAD/Flavins (later referred to as FAD) (ex. 460 nm, em. 550 nm) and the presence of 'direct' and 'inverted' hypoxic gradients²⁹. Figure 6 demonstrates the results of combined confocal ratiometric imaging of spheroids oxygenation (with the help of O₂-sensing nanoparticles MMIR) and autofluorescence FLIM of FAD. The 'classical' directoxygenation gradient was observed in spheroids of 69 mm diameter, while larger (141 µm) spheroid showed an 'inverted' gradient. Adding FAD-FLIM to these measurements helps validate differences in oxidative metabolism, as the fast FLIM and phasor plots demonstrate: small spheroids displayed a more prominent fraction of longer FAD lifetimes, potentially indicating higher glycolytic activity in the smaller-size spheroids.

Figure 6:Example of multiparametric imaging of flavin autofluorescence (ex. 460 nm/em. 510-590 nm) and oxygenation (intensity ratio analysis with MMIR O₂ sensitive nanoparticles: ex. 614 nm/reference em. 631-690 nm / sensitive em. 724-800 nm) in hDPSC spheroids of different size (small: 188 cells/spheroid, big: 820 cells/spheroid), produced by high-throughput self-produced agarose micromolds method (confocal FLIM). (A,B) Representative example of oxygenation intensity ratio and flavin-autofluorescence lifetime imaging in big (A, "B Sphs") and small (B, "S Sphs") hDPSC spheroids. (C,D) Comparative analysis of flavin-autofluorescence lifetime using phasor plot cloud analysis, based on the comparison of phasor cloud geometrical centers. (E) The pixel centroids (geometrical centers) coordinates of flavin autofluorescence from big hDPSC spheroid (B Sphs) and small hDPSC spheroid (S Sphs) on phasor plot. Two different color codes applied for different FLIM analysis approaches, where color coding with τ shows average photon arrival time (FAST-FLIM images) distribution, while color coding with τ φ – tau phase corresponds to phasor-FLIM analysis of fluorescence lifetime in the same spheroids. Please click here to view a larger version of this figure.

Multiplexed analysis of FAD autofluorescence and cell death in human iPSC spheroids
Human induced pluripotent stem cell (iPSC) spheroids are often used as a first step during the induction of tissue-specific differentiation and the production of organoids, e.g., in the case of neural organoid culture. Depending on their handling and the specific generation protocol, viability and subsequent reproducibility of organoids can be severely affected. Non-destructive investigation of organoids and neural progenitor cell-containing spheroids is crucial for the structural assessment, monitoring, and predicting the quality of growing and assembled neural tissues during their development^68,69. Figure 7A shows that the non-destructive FLIM of autofluorescent Flavin/FAD molecules can give us information about the viability of iPSC spheroids. Imaging was performed 4 days after seeding 9,000 iPSCs in commercial ultra-low attachment plate wells (Corning). Before imaging, spheroids were stained for 1 h with 0.5 µg/mL Propidium Iodide (PI) to visualize dead cells. No necrotic cores were observed in these iPSC spheroids with a seeding density of 9,000 cells after 4 days. Three main patterns can be distinguished on the corresponding phasor plot of flavin/FAD autofluorescence (Figure 7B left panel): pattern of media autofluorescence (magenta color), pattern (with average τ φ – tau phase, ~2.6 ns) corresponding to no PI regions (ROI1) and pattern (with average τ φ ~3.1 ns) of regions correlating with the cell death marker – staining (ROI2). An additional comparison of PI-treated and non-treated 2D cultures of iPSCs demonstrated no impact of PI on the appearance of long-lifetime flavin / FAD phasor patterns (Figure 7C),indicating that this increase of flavins / FAD lifetime can be an independent marker of compromised viability of iPSCs and their derivative cells. While this observation is in line with therecent report on intensity-based imaging⁷⁰, we cannot completely rule out the presence of propidium iodide intensity with FAD emission channel (506-582 nm). A further investigation is needed to prove the link between flavin / FAD lifetime changes and potential events of cell death.The simple multiparametric phasor-based analysis of ROI regions demonstrates an elegant way for quick screening and data analysis.

Figure 7: Flavin autofluorescence confocal FLIM and propidium iodide staining in iPSC spheroids to assess cell viability. (A) Flavin autofluorescence (left) and Propidium Iodide – PI (right) FLIM of an iPSC spheroid 4 days after seeding 9,000 cells in an Ultra-low attachment plate (Corning) and PI staining (0.5 µg/mL, 1 h) prior to imaging. (B) Phasor-plot analysis of flavin autofluorescence (left) and propidium iodide (right) overall images and corresponding ROIs (ROI1 – green and ROI2 – red, media autofluorescence – magenta). ROIs were selected based on the presence of propidium iodide labeling. (C) fast-FLIM and phasor plot analysis of flavins autofluorescence in 2D culture of iPSC treated and non-treated with PI (0.5 µg/mL, 1 h). No impact of PI staining on the appearance of the long fluorescence lifetime of flavins (corresponds to a phasor pattern within a pink circle) was detected when collecting theFlavin excitation within a range from 469-542 nm. This detector range does not overlap with the excitation spectrum of PI (550-720 nm). When a broader excitation range for FAD is collected (469-590 nm) an impact of PI can be noticed. Image color-coding corresponds to the average photon arrival time values (fast-FLIM). Imaging parameters were T = 35 °C, 25X/0.95 water-immersion objective (A-B), 40X/1.25 Glycerol objective (C). Flavins: ex. 460 nm, em. 506-582nm (A-B), pinhole 4 AU. PI: ex. 535 nm, em. 584-667 nm, pinhole 1 AU. Channels: Intensity and Tau. The repetition rate of 40 MHz to collect full decay of Flavin autofluorescence. Phasor analysis settings; Harmonic 1, Threshold 1 (A-B), 47 (C), Median Filter 17 (A-B), 11 (C). Please click here to view a larger version of this figure.

Supplementary File 1: Additional protocols on spheroid formation. Please click here to download this File.

Supplementary File 2: FLIM data of glycolytic core and non-glycolytic core.lif. Please click here to download this File.

Supplementary File 3: FLIM data of glycolytic core.ptu. Please click here to download this File.

Supplementary File 4: FLIM data of non-glycolytic core.ptu. Please click here to download this File.

Supplementary File 5: Glycolytic core fast FLIM.tif Please click here to download this File.

Supplementary File 6: Non-glycolytic core fast FLIM.tif Please click here to download this File.

Supplementary File 7: Glycolytic core free NAD(P)H point.tif Please click here to download this File.

Supplementary File 8: Glycolytic core phasor plot.tif Please click here to download this File.

Supplementary File 9: Non-glycolytic core free NAD(P)H point.tif Please click here to download this File.

Supplementary File 10: Non.glycolytic core phasor plot.tif Please click here to download this File.