Monitoring membrane viscosity in differentiating stem cells using BODIPY-based molecular rotors and FLIM

Membrane fluidity plays an important role in many cell functions such as cell adhesion, and migration. In stem cell lines membrane fluidity may play a role in differentiation. Here we report the use of viscosity-sensitive fluorophores based on a BODIPY core, termed “molecular rotors”, in combination with Fluorescence Lifetime Imaging Microscopy, for monitoring of plasma membrane viscosity changes in mesenchymal stem cells (MSCs) during osteogenic and chondrogenic differentiation. In order to correlate the viscosity values with membrane lipid composition, the detailed analysis of the corresponding membrane lipid composition of differentiated cells was performed by time-of-flight secondary ion mass spectrometry. Our results directly demonstrate for the first time that differentiation of MSCs results in distinct membrane viscosities, that reflect the change in lipidome of the cells following differentiation.


General Materials and Methods
The manipulation of all air and/or water sensitive compounds was carried out using standard inert atmosphere techniques. All chemicals were used as received from commercial sources without further purification. Anhydrous solvents were used as received from commercial sources. Analytical thin layer chromatography (TLC) was carried out on Merck aluminium backed silica gel 60 GF254 plates and visualisation when required was achieved using UV light or I 2 . Flash column chromatography was performed on silica gel 60 GF254 using a positive pressure of nitrogen with the indicated solvent system. Where mixtures of solvents were used, ratios are reported by volume. Nuclear magnetic resonance spectra were recorded on 400 MHz spectrometers at ambient probe temperature. Chemical shifts for 1 H NMR spectra are recorded in parts per million from tetramethylsilane with the solvent resonance as the internal standard (methanol: δ = 3.31 ppm). 13 C NMR spectra were recorded with complete proton decoupling. Chemical shifts are reported in parts per million from tetramethylsilane with the solvent resonance as the internal standard ( 13 CD 3 OD: 49.00 ppm). 19 F NMR spectra were recorded with complete proton decoupling. Chemical shifts are reported in parts per million referenced to the standard hexafluorobenzene: −164.9 ppm. Mass spectra were carried out using ElectroSpray Ionisation (ESI), and only molecular ions are reported. BODIPY 1. To a solution of BODIPY 4 (220 mg, 0.48 mmol) in 16 mL of dry THF was added N,N,N',N'-tetramethyl-1,3-propanediamine (3 mL, 17.9 mmol) under N 2 atmosphere. The reaction mixture was stirred at 18 0 C for 24 h. The resulting dark-red precipitate was filtered, washed with THF (50 mL) and diethyl ether (100 mL) and dried under vacuum to give the corresponding mono-charged intermediate which was used without further purification. Then, the mono-charged intermediate was dissolved in DMF (3 mL) and iodomethane (1 mL, 6.5 mmol) was added to the solution under N 2 atmosphere. The reaction mixture was stirred overnight at room temperature, and then the solvent was evaporated under reduced pressure to give a dark red crude product which was purified by column chromatography on silica gel (methanol, and then a mixture of 3:1 methanol:0.5M NH 4 Cl), R f 0.2. Fractions were evaporated at 30°C to give a mixture of BODIPY 1 and NH 4 Cl which was further dissolved in methanol and successively filtered to remove most of NH 4 Cl. The red-orange crude, which was still contaminated with NH 4 Cl according to 1 H-NMR, was dissolved in methanol and a saturated solution of NH 4 PF 6 in H 2 O was added in order to exchange counter-ions. The precipitate was isolated by filtration, washed thoroughly with H 2 O, methanol and diethyl ether. Finally, the red-orange solid was dissolved in acetone and passed through a Dowex 1x8 200 mesh ion-exchange column (H 2 O). Fractions were evaporated to dryness (at 30°C) to give BODIPY 1 as a red-orange wax. Yield: 108 mg (41%). 1 Figure S4. Fluorescence time resolved traces of BODIPY 1 recorded as a function of the decreasing viscosity of the environment (M0→M10), obtained in glycerol at 1.2-100 0 C. The decays were fitted using a monoexponential decay function and the obtained lifetimes were plotted against viscosity in Figure S5.

BODIPY 2 staining
Initially we stained control and differentiated MSCs with molecular rotor BODIPY 2, that was previously successfully used in 2D and 3D cell culture and in vivo (2,3). We observed the poor cell staining for this rotor. Higher concentrations of BODIPY 2 in the staining medium resulted in brighter images, however, it was clear from the decay analysis that the lifetimes obtained from cellular membranes change significantly depending on the incubation concentration, which is a signature of dye aggregation.

BODIPY 1 staining
We used two channel detection FLIM: green channel predominantly recording the emission of the monomers and the red channel recording the emission of the aggregates (if present) to ascertain whether BODIPY 1 biexponential decay kinetics is caused by aggregation. It is well known that in the presence of aggregates, the red channel shows a very different decay kinetics, with a significantly larger contribution of the long decay component due to the aggregated species (4). In this present case, both channels show a very similar decay kinetics, which allows us to exclude aggregation of BODIPY 1.   In all examined locations at the plasma membrane BODIPY 1 had a biexponential decay with a short component (τ1) in the region of 500 ps (corresponding to ca 20-30 cP) and a long component (τ2) in the region of 4-5 ns (corresponding to several hundred cP up to 1500 cP).
The short component of these decays relates to a BODIPY orientation close to the head region of the bilayer, while the long component corresponds to the viscosity in the inner, highly hydrophobic tail region of the bilayer. The longer component (τ2) was converted to viscosity, as previously discussed (5,6). α 1 and α 2 parameters were used to assess the correctness of the obtained data on fluorescence lifetimes and characterized the contribution of short component (τ1) and long component (τ2), which may reflect the distribution of BODIPY between two locations of the membrane, as previously discussed (5,6).
Therefore, it should be noted that α1 and α2 parameters do not relate to the viscosity of the lipid tail region of the membrane, as 1 and averaged are never converted to viscosities.

Undifferentiated cells: FLIM and mass spectrometry studies
We first performed FLIM measurements with BODIPY1 to compare undifferentiated cells on days 0-21 and differentiated MSCs on days 7-21. We observed a marked viscosity increase in all cell types with cultivation time: 1) undifferentiated day 21 cells were significantly more viscous than undifferentiated day 7 and 14 cells, and these were more viscous than undifferentiated day 0 cells. 2) osteogenically and chondrogenically differentiated cells on day 21 were more viscous than respected day 7 and 14 cells, and all were more viscous than undifferentiated day 0 cells. The fact that undifferentiated MSCs showed a significant increase in membrane viscosity on day 21 may be related to a significantly denser layer of MSCs. However, the membrane viscosity rise in osteogenically and chondrogenically differentiated cell when compared to the undifferentiated day 0 cells may be associated with both an increase in confluence and a change in the membranes composition. Figure S13. Viscosity analysis of MSCs lipid tail region via FLIM of molecular rotor BODIPY 1 during differentiation. Dynamics of the viscosity change during differentiation.* statistically significant differences compared with undifferentiated MSCs, p≤0, 05.
In order to gain insight on whether the lipid composition could cause the above viscsoity changes, we performed the time-of-flight secondary ion mass spectrometry on all cell samples. This is, however, a surface-sensitive technique, which requires dehydrated sample in vacuum conditions. It may be considered as a drawback in the context of our work since it is not possible to track changes in lipid composition over time for exactly the same cells. Nevertheless TOF-SIMS could provide valuable information comparing different cells or the same cell culture that is taken at different stages of the process (days 0, 7, 14, 21 in our case). Cell density on the substrate is the main challenge in the analysis of this data, since lipid ion yield is proportional to the area occupied by cells.
ToF-SIMS analysis of cell culture required chemical fixation, rinsing and air drying as described in materials and methods section. The main aim of rinsing is removing of salts from the cell surfaces, since the excess of salts leads to organic ions signal suppression and dividing of a single species signal into multiple channels. However, we found that rinsing procedure, even when carefully controlled, leads to loss of adhesion of some MSCs. The amount of cells removed from the substrate surface varies hence the protocol could not be considered as reproducible. The application of a layer of polylysine on the substrate solves this problem. Polylysine layer does not affect data quantification for undifferentiated MSCs, osteogenic and hondrogenic differentiated MSCs on day 21 since the cells density on the substrates reaches its maximum (confluent cells). However, it severely complicates the analysis for growing undifferentiated MSCs with low density, due to the interference of lysine peaks from the substrate with peaks originating from cells. It is unclear what exact percentage of the examined area is occupied by cells and, hence, how to compare ion yields of differentiated MSCs at day 0 with others cells. Nevetheless specific lipid ions could not originate from the substrate covered by polylysine. Hence, the ratio of these ions does not depend on cell density and can be compared. We used phosphatidylcholine peak (m/z 224) as a reference, which should be proportional to the surface cell coverage and is not affected by polylysine signal. Fig. S14 shows that SM/PC and cholesterol/PC ratios significantly differs for undifferential cells on day 0 and day 21. Interestingly, the ratios change in the opposite direction. While SM/PC ratio decreases for undifferentiated MSCs day 21 compare to day 0, cholesterol/PC ratio increases. A high level of sphingomyelin in undifferentiated MSCs at day 0 may be associated with their high metabolic activity. These cells actively proliferate, that is associated with the activation of signaling pathways. Sphingomyelin is known to play an important role in cell signaling (7). Figure S14. Lipid analysis by ToF-SIMS. Relative ion yield of sphingomyelin and cholesterol relative to phosphatidylcholine ion yield.
Fatty acids analysis (Fig. S15) does not reveal significant differences in saturated and overall unsaturated fatty acids yields. A significant decrease in amount of polyunsaturated fatty acids is clearly observed for undifferentiated MSCs on day 21 compare to day 0. Moreover, chondrogenically and osteogenically differentiated cells also show lower relative PUFAs ion yield. The decrease in PUFAs relative ion yield between undifferentiated cells day 0 and day 21 is in agreement with viscosity measurements provided by FLIM. Figure S15. Lipid analysis by ToF-SIMS. Fatty acids analysis. Ion yields are normalized to ion yield of saturated fatty acids.
Thus, these data showed that changes in the viscosity of the membrane in cells on day 21 can be caused by changes in membrane lipids.