Simultaneously defining cell phenotypes, cell cycle, and chromatin modifications at single-cell resolution

Abstract

Heterogeneity of cellular phenotypes in asynchronous cell populations placed in the same biochemical and biophysical environment may depend on cell cycle and chromatin modifications; however, no current method can measure these properties at single-cell resolution simultaneously and in situ. Here, we develop, test, and validate a new microscopy assay that rapidly quantifies global acetylation on histone H3 and measures a wide range of cell and nuclear properties, including cell and nuclear morphology descriptors, cell-cycle phase, and F-actin content of thousands of cells simultaneously, without cell detachment from their substrate, at single-cell resolution. These measurements show that isogenic, isotypic cells of identical DNA content and the same cell-cycle phase can still display large variations in H3 acetylation and that these variations predict specific phenotypic variations, in particular, nuclear size and actin cytoskeleton content, but not cell shape. The dependence of cell and nuclear properties on cell-cycle phase is assessed without artifact-prone cell synchronization. To further demonstrate its versatility, this assay is used to quantify the complex interplay among cell cycle, epigenetic modifications, and phenotypic variations following pharmacological treatments affecting DNA integrity, cell cycle, and inhibiting chromatin-modifying enzymes.

Keywords: epigenetics; high-throughput; microscopy.

Figure 1.

Simultaneous measurements of relative histone acetylation, DNA content, and cell/nucleus morphology at single-cell resolution. A–D) Typical micrographs showing nuclear DNA (A) and acetylated histone (B) stains and corresponding heat maps of DNA intensity (C) and histone acetylation intensity (D) in each cell. Scale bar = 100 μm. E–G) Distributions of overall histone H3 content per cell (E), global histone H3 acetylation per cell (F), and H3 acetylation normalized by overall histone H3 content per cell (G). H, I) Distribution of DNA content per cell normalized by the fluorescence intensity at the G₀/G₁ peak (H) and the corresponding fraction of cells in the G₀/G₁, S, and G₂/M cell cycle phases (I). Panel I shows the mean ± se. J–N) Distributions of nucleus size (J), nucleus circularity (K), cell size (L), cell circularity (M), and F-actin content per cell (N). Two biological repeats of 2 duplicate samples were conducted for a total of 2309 cells for all panels.

Figure 2.

Variations in histone acetylation predict variations in cell and nucleus morphology. A–D) Global acetylation on histone H3 as a function of normalized DNA content per cell (A), nucleus size (B), cell size (C), and cell circularity (D) at single-cell resolution. E–H) Acetylation on histone H3 normalized by overall histone H3 content per cell as a function of normalized DNA content per cell (E), nucleus size (F), cell size (G), and cell circularity (H) at single-cell resolution. Insets show the same data binned by the x-axis parameter (mean±se AcH3) and statistically compare each bin to the preceding bin using 1-way ANOVA tests. Two biological repeats of 2 duplicate samples were conducted for a total of 1832 cells for all panels. ns, not significant (P>0.05). *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 3.

Heterogeneities in histone acetylation and nucleus size as a function of DNA content. A) Narrow slices were computationally isolated from the DNA content distribution at each of the 3 cell cycle phases, G₀/G₁ (red, 221 cells), S (blue, 105 cells), and G₂/M (green, 101 cells). B–D) Distributions of nucleus sizes (B), histone H3 acetylation values (C), and histone H3 acetylation normalized by overall histone H3 content (D) were plotted for the whole cell population (gray bars) and for the cells in each of the 3 DNA content slices. E) Narrow slices were isolated from the nucleus size distribution at small nucleus size (pink, 158 cells) and large nucleus size (yellow, 97 cells). F–H) Distributions of global AcH3 values (F), normalized AcH3 values (G), and DNA contents (H) were plotted for the whole cell population (gray bars) and for the cells in each of the 2 nucleus size slices. I) Narrow slices were isolated from the global AcH3 distribution at low acetylation (orange, 114 cells) and high acetylation (purple, 103 cells). J–L) Distributions of normalized AcH3 values (J), DNA contents (K), and nucleus sizes (L) were plotted for the whole cell population (gray bars) and for the cells in each of the two AcH3 slices. M) Narrow slices were isolated from the normalized AcH3 distribution at low (teal, 135 cells) and high values of normalized acetylation (brown, 118 cells). N–P) Distributions of DNA contents (N), nucleus sizes (O), and AcH3 values (P) were plotted for the whole cell population (gray bars) and for the cells in each of the 2 normalized AcH3 slices. For panels A–P, insets show the same data with the y axis magnified to better show distributions. Q) Global histone H3 acetylation plotted with binned values of increasing nucleus size for each narrow slice of DNA content. R) DNA content plotted with binned values of increasing nucleus size for each narrow slice of global histone H3 acetylation. S) Global histone H3 acetylation plotted with binned values of increasing DNA content for each narrow slice of nucleus size. For panels Q–S, bars show means ± se, and statistics compare bars to the preceding bar using 1-way ANOVA tests. For all panels, two biological repeats of two duplicate samples were conducted for a total of 2309 cells in the overall population. ns, not significant (P>0.05). *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 4.

Histone acetylation as a function of cell cycle. A) Cell-cycle distribution of untreated control cells (black bars), compared to that of serum-starved cells (red curve), TSA-treated cells (purple curve), and DMSO-treated cells (the drug-vehicle control for TSA, green curve). B) Corresponding fractions of cells in the G₀/G₁, S, and G₂/M cell cycle phases for each of the 4 conditions. C–F) Average nucleus size (C), global histone H3 acetylation (D), histone H3 acetylation normalized by overall histone H3 content (E), and F-actin content (F) for the whole cell population, as well as within each of the cell cycle phases for the same 4 conditions. For panels B–F, bars show means ± se, and statistics compare each condition to the untreated control (black bars) using 2-way ANOVA tests. For all panels, 2 biological repeats of 2 duplicate samples were conducted for a total of 2337 untreated cells, 1341 serum-starved cells, 849 DMSO-treated cells, and 801 TSA-treated cells. ns, not significant (P>0.05). *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 5.

Cellular variations in histone acetylation predict phenotypic variations. A, B) Coefficient of variation (CV) of acetylated histone H3 values as a function of nucleus size (A) and as a function of CV of nucleus size (B) for 5 subpopulations of each of the 4 listed conditions. CV of acetylated histone H4 values as a function of nucleus size (C) and as a function of CV of nucleus size (D) for 5 subpopulations of each of the 4 listed conditions. Black lines show the best-fit line through all 20 points for each plot. For panels A and B, 2 biological repeats of 2 duplicate samples were conducted for a total of 1301 untreated cells, 810 serum-starved cells, 838 DMSO-treated cells, and 417 TSA-treated cells. Each population was divided into 5 subpopulations by binned nucleus size. For panels C and D, 2 biological repeats of 2 duplicate samples were conducted for a total of 1100 untreated cells, 676 serum-starved cells, 1212 DMSO-treated cells, and 214 TSA-treated cells.