The Chemical Potential of Plasma Membrane Cholesterol: Implications for Cell Biology  Artem G. Ayuyan, Fredric S. Cohen  Biophysical Journal  Volume 114, Issue 4, Pages 904-918 (February 2018) DOI: 10.1016/j.bpj.2017.12.042 Copyright © 2018 Terms and Conditions

Figure 1 Calibration for percentage of saturation of cholesterol bound to 3 mg/mL MBCD (defined as S in text) versus cholesterol concentration in hexadecane (solid circles) or squalane (open triangles). (Inset) Raw data were fit with Langmuir binding curves scaled to account for the percentage of saturation not reaching 100%. The subsaturation in plotting raw data occurred because some MBCD binding sites were occupied by the solvent and some MBCD partitioned into the organic phase. Cholesterol concentration was normalized so that when saturated in the organic phase, C = 1. Cholesterol saturation in MBCD was also normalized to 1. A scaling factor was applied to all data of the inset so that when cholesterol was saturated in the organic solvent, cholesterol binding to MBCD would also be saturated. The same scaling factor was applied to all data points. Biophysical Journal 2018 114, 904-918DOI: (10.1016/j.bpj.2017.12.042) Copyright © 2018 Terms and Conditions

Figure 2 The procedure used to measure the natural value of cholesterol’s activity and chemical potential in RBCs also yields their relationship to cholesterol concentration. (A) The intersection of initial versus final chemical potential with the diagonal line yields μCH of unperturbed RBCs. Data shown for RBCs obtained from three individuals. (B) In performing the measurement, the cholesterol content of the RBCs is diminished if the initial chemical potential of the bathing solution is less than μCH, and content is increased if the initial chemical potential is the greater. This allows μCH (B) and activity (C) versus cholesterol concentration to be obtained. To derive activities from measured chemical potentials, the reference value of μ0 in the RBC membrane is required. For this figure, we assumed μ0 = 0. The activity versus concentration curves would be contracted or expanded vertically if μ0 was > or <0, respectively. Cholesterol concentration is 0.45 for unperturbed RBCs. Biophysical Journal 2018 114, 904-918DOI: (10.1016/j.bpj.2017.12.042) Copyright © 2018 Terms and Conditions

Figure 3 Plots of initial versus final chemical potential of cholesterol in aqueous MBCD solutions (A) show that plasma membrane cholesterol chemical potential increases with increasing cell densities (B). (A) The intersection of initial versus final chemical potential curves with the diagonal line representing equal initial and final chemical potential is μCH for MCF7 (open squares) and MDA-MB-231 (open circles) cells. The curves are third-order polynomial fits; straight lines are least-square linear fits. A curve and a line fit to the same data set intersect the diagonal equilibrium line at virtually the same point. Each displayed μCH point is the mean ± SE obtained by averaging the intersection points of linear fits of data from four sequential cell densities. That is, cell densities were rank-ordered and a moving average of four intersection points was calculated. (B) μCH of the highly metastatic MDA cells (solid circles) is higher than that of MCF7 cells (solid squares), for all cell densities. DNA content within a well of confluent cells was used as the measure of cell density. The p values of an ANOVA comparing each point against that of the lowest density point (i.e., the first point) were calculated; for each of the two cell lines, asterisks mark where statistically significant differences occurred. Whether the dips in μCH (e.g., ∼0.8 μg/cm2 of DNA for MCF7 cells) are chance occurrences or due to consequences of density remains to be determined. ∗ denotes p < 0.05, ∗∗p < 0.03, ∗∗∗p < 0.005, ∗∗∗∗p < 0.003. Biophysical Journal 2018 114, 904-918DOI: (10.1016/j.bpj.2017.12.042) Copyright © 2018 Terms and Conditions

Figure 4 Cell protein (A) and cholesterol (B) content normalized to total DNA decreases with cell density; cholesterol content normalized to protein (C) is independent of cell density. For each parameter, the same patterns were observed for MDA (solid circles) and MCF7 cells (open squares). Biophysical Journal 2018 114, 904-918DOI: (10.1016/j.bpj.2017.12.042) Copyright © 2018 Terms and Conditions

Figure 5 μCH (A) and activity (B) versus cellular cholesterol content in plasma membranes of MDA-MB-231 cells. (A) μCH as a function of the normalized amount of cholesterol in the cells after cholesterol transfer brought cholesterol between the plasma membrane and the aqueous solution to equilibrium. This amount was calculated from experimentally derived curves illustrated in Fig. 3 A. Equilibrium was quickly achieved in these experiments (Fig. S1), allowing the final chemical potential to be calculated. The shapes of the curves are complex and depend on cell density; circles are for the lowest density—0.11 ± 0.024 μg DNA per cm2 of growth area, squares for intermediate density—0.24 ± 0.013, and triangles for highest density—0.98 ± 0.03. (B) Chemical activities were calculated from chemical potentials for a reference value of μ0 = 0. Biophysical Journal 2018 114, 904-918DOI: (10.1016/j.bpj.2017.12.042) Copyright © 2018 Terms and Conditions

Figure 6 Varying the clamped cholesterol chemical potential does not greatly affect cellular cholesterol content. For MDA cells (solid bars), the cholesterol content of clamped cells in defined media was statistically lower (∗p < 0.03) for μCH clamped to either −2.1 or −1.35 kBT, but not to −0.6 kBT, compared to control cells. The small differences in values for cholesterol content in clamped MDA cells were statistically significant at a level of p < 0.03. For MCF7 cells (open bars), cholesterol content was not statistically affected by clamping μCH to any value. Biophysical Journal 2018 114, 904-918DOI: (10.1016/j.bpj.2017.12.042) Copyright © 2018 Terms and Conditions

Figure 7 Cells that have been clamped for a long time quickly assume a new cholesterol chemical potential value when reclamped at that value. MCF7 cells were clamped for 24 h to values of μCH (−2.1 kBT, circles; −1.35 kBT, squares; −0.6 kBT, triangles). The actual chemical potential was then measured for each batch of cells that had been clamped for 24 h. The bathing solutions were changed to new initial chemical potentials, and the solution’s final chemical potentials were measured 15 min later. If the initial cholesterol chemical potential in solution used for the measurement was less than the 24-h clamped μCH, the experimental points were slightly to the left of the 45° diagonal. But for an initial chemical potential in solution that was equal to the clamped μCH, the experimental point was (within experimental error) on the line. That is, the clamping procedure set μCH to the intended value, demonstrating the reliability of the clamping procedure. Once μCH of cells had been clamped for long times, they quickly (within the 15 min necessary to measure μCH) adjusted their clamped value of μCH to the chemical potential of the bathing solution—upon increasing the chemical potential in solution, the points laid, within experimental scatter, along the 45° line. Biophysical Journal 2018 114, 904-918DOI: (10.1016/j.bpj.2017.12.042) Copyright © 2018 Terms and Conditions

Figure 8 STAT3 activation is dependent on cholesterol chemical potential. Densitometry for expression of total STAT3 (solid bars), p-S-STAT3 (open bars), and STAT3 phosphorylated at Tyr-705 (p-Y-STAT3, shaded bars). With increasing cholesterol chemical potential, p-Y-STAT3 levels clearly increase (ANOVA, p < 0.01), and p-S- STAT3 modestly increase as well (ANOVA, p < 0.01). The total expression of STAT3 protein is not affected. Biophysical Journal 2018 114, 904-918DOI: (10.1016/j.bpj.2017.12.042) Copyright © 2018 Terms and Conditions