Growth dynamics of bacterial protocells in densely packed aggregates and in isolation.

(A) Image sequences showing the development of bacterial protocells derived from wall-deficient B. subtilis in an aggregated protocolony (upper) and in isolation (lower). The data was from representative cases. The first panel in each row is a phase-contrast image. The green false color displays the fluorescence from nucleic acids (stained by SYTO16; Methods), which served as a proxy of the biomass of cells. For cells in protocolonies, the area occupied by nucleic acids kept increasing, while scattered hollow areas appeared within the protocolony due to cell death; also see Movie S1. For isolated protocells, the area occupied by nucleic acids was decreasing and cells gradually lysed; lysis events were identified by sudden loss of membrane integrity (Movie S2; Methods). Note that the area with weak fluorescence enclosed by dashed lines are nucleic acids remnants of recently lysed cells; the fluorescence disappeared shortly as nucleic acids diffused away. Scale bars: 50 μm (upper panels) and 20 μm (lower panels). (B) Temporal dynamics of nucleic acid area during the growth of cells in protocolonies (lines and data points in red color) and of isolation protocells (green line). The nucleic acid area was computed from SYTO16 fluorescence images (Methods) and it served as a proxy of biomass. For protocolonies (N=27), each light-colored line represents data from one protocolony; the data points connected by straight lines are the averaged biomass of all protocolonies, with the error bars representing the standard error of the mean. For isolated protocells, the sum of nucleic acid area in a population (N=291) was plotted. The schematic diagrams illustrate the spatial distribution of aggregated and isolated cells. In both panels T=0 min corresponds to the time of inoculation.

Analysis of surface-volume balance for actively growing bacterial protocells.

(A) The surface-volume balance ratio (η) plotted against growth time for actively growing cells (see Fig. S3, panels C,H). Red and green points correspond to cells in protocolonies and isolated protocells, respectively. The time is normalized by the maximal observation period (>4 hr), or by the lifespan of a cell if it lysed within the observation period. Lines connecting the data points serve as guides to the eyes. Error bars represent standard error of the mean (N= 866 and 187 cells in protocolonies and in isolation, respectively). (B,C) Dependence of η on geometrical parameters (colormap) and the evolutionary trajectory of growing bacterial protocells in the geometrical space (white triangles). Panel B and C correspond to cells in protocolonies and in isolation, respectively. Colormap in each panel was computed based on our mathematical analysis and the experimentally measured growth parameters (averaged growth rates of surface area and volume) (Methods); it shows the dependence of η on effective radius and circularity, with the color bar to the right of each panel indicating the value of η and the black dotted line indicating the contour of η=1. The white triangles in each panel represents the average effective radius and circularity of cells at different times of growth as shown in panel A (time increasing from left to right).

Correlation between shape deformation and division of bacterial protocells.

(A,B) Effective radius (panel A) and circularity (panel B) of cells plotted against the growth time of bacterial protocells. Red and green data correspond to cells in protocolonies and isolated cells, respectively. T=0 min corresponds to the time of inoculation. Error bars represent standard deviation (N>50 cells). Lines connecting the data points serve as guides to the eyes. For data in panel A, two-tailed Mann-Whitney-Wilcoxon test was performed to examine whether each pair of red and green data sets at a specific time point were drawn from the same distribution, and the test showed that every pair of data sets were significantly different (P < 0.05). For data in panel B, Levene’s test was performed to examine the equality of variances of each pair of red and green data sets at a specific time point, and the test showed that the variances of every pair of data sets were significantly different (P < 0.05). Also see Fig. S5. (C) Protocell division rate (number of divisions per cell per hour) plotted against the extent of cell deformation defined as (1 – C), where C represents the circularity of cells. Data were acquired in 9 protocolonies (gray lines); red line shows the average of data from all gray lines, with the error bars indicating standard deviation. For each protocolony, all cells and all the division events throughout the entire observation period were included in the analysis.

Membrane tension measurement based on the fluorescent probe FliptR.

(A) Boxplot showing the fluorescence lifetime distributions of FliptR on bacterial protocell membranes. Due to the quasi-2D interstitial confinement (Fig. 1A), most of a cell’s membrane is in contact with the double-layered agar pad and the multi-well culture plate; this part of cell membrane is denoted as “bulk part”. The membrane at the interface between neighboring cells in a protocolony (i.e., brighter lines in the fluorescence intensity plot in panel B) is denoted as “edge”. Fluorescence lifetime of FliptR on the membrane was analyzed pixel-wise for 5 protocolonies and 162 isolated cells (Methods), and data from all the pixels were grouped and plotted here. The line inside each box represents the median, the top and bottom edges of each box represent the quartiles, and the ends of whiskers represent the largest (or lowest) data point within 1.5-fold of the interquartile range (i.e., the box height) from the upper (or lower) edge of the box. Significance test was performed by two-tailed Mann-Whitney-Wilcoxon non-parametric test for the equality of means of data sets that do not follow Gaussian distributions (***P < 1.0 × 10−10). (B) Fluorescence intensity image of FliptR staining cell membrane in protocolonies. The higher fluorescence intensity at the interface between neighboring cells than the bulk part of cells is due to the higher density of lipid membrane at cell-cell contact areas. Scale bar, 20 μm. (C) Spatial distribution of FliptR fluorescence lifetime obtained by FLIM imaging for the protocolonies shown in panel B. False color scale represents fluorescence lifetime in units of nanoseconds.

Cellular Potts Model simulations of protocell growth dynamics.

(A,B) Comparison of protocolony morphology between experiment (panel A) and Cellular Potts Model simulation (panel B). The experimental image sequence shown in panel A was segmented according to the fluorescence images of CellROX-stained cells (Methods). To initiate the simulation 120 circular cells were seeded at hexagonal close-packing density (Methods). Scale bar, 10 μm. (C) Membrane tensile strain of protocells plotted against growth rate in simulations. Red and green data correspond to cells in protocolonies and in isolation, respectively. For cells in protocolonies, the membrane tensile strain is computed for the periphery area (upper panel) and the bulk part (lower panel); for isolated protocells, the overall membrane tensile strain is computed (see Methods). Scattered points represent the time-averaged membrane tensile strain and the prescribed growth rate of individual cells. Data points in the line plots represent the mean membrane tensile strain of cells with the same prescribed growth rate, with error bars representing standard deviation of membrane tensile strain (N > 45 simulated cells). Lines connecting the data points serve as guides to the eyes. (D) Simulated protocell division rate (number of divisions per cell per hour) plotted against the extent of cell deformation defined as (1 – C), where C represents the circularity of cells. Data were acquired in 11 protocell populations (gray lines); red line shows the average of data from all gray lines, with the error bars indicating standard deviation. The analysis for each protocell population was performed by taking into account all cells and division events across all simulation steps, following the same calculation method as in Fig. 3C. (E) Rate of circularity change plotted against membrane tensile strain in simulations. Red and green data correspond to cells in protocolonies and in isolation, respectively. Each scattered point represents the instantaneous membrane tensile strain (calculated in the same manner as in the upper panel of Fig. 5C) and the instantaneous rate of circularity change for a cell. The Spearman correlation coefficient of membrane tensile strain and circularity change rate is -0.69 (p ≈ 0) and -0.02 (p=0.04) for cells in protocolonies and in isolation, respectively. (F) Time evolution of protocell proliferating probability depends on seeding density. The seeding density was varied by tuning the average cell-cell distance in simulated protocell populations, with a density of 1.0 corresponding to hexagonal close packing (Methods).

Experimental setup for long-term observation of bacterial protocell growth.

(A) Schematic diagram of the quasi-2D culture method of bacterial protocells derived from wall-deficient B. subtilis L-forms. The double-layered agar pad consists of a thick layer of 4.0% agar (yellow) and a thin layer of 0.6% agar (red), both of which are infused with an osmoprotective nutrient medium (Methods). The thick layer of 4% agar (yellow) was solidified between two coverslips (blue). The thin layer of 0.6% agar (red) was solidified between the 4% agar and a coverslip. The prepared double-layered agar was then placed on top of the protocell suspension (green) deposited on a multi-well culture plate. The thin 0.6% agar layer, with a thickness of ∼5 µm, is in direct contact with cells and the multi-well culture plate’s bottom surface; it reduces sliding of the 4% agar pad and allows long-term observation of cells. (B) Lifespan distribution of isolated bacterial protocells. The lifespan of an isolated protocell is defined as the duration from inoculation to lysis.

Growth rate of bacterial protocells in densely packed aggregates and in isolation.

(A) Boxplot showing the volume growth rate of cells in aggregated protocolonies (red) and of isolated protocells (green). The cytoplasm-occupied area (A) of each individual cell is proportional to cell volume because of the pancake shape of cells under quasi-2D confinement. The cytoplasm-occupied area was measured by cell profiling based on fluorescence images of the cytoplasm stain CellROX Deep Red (Methods). It was fitted exponentially as A = A0exp(βt), where β is defined as the volume growth rate and A0 is the cell’s cytoplasm-occupied area at the beginning of observation (Methods; Fig. S4). The average volume growth rate of cells in protocolonies and in isolation was 0.50 ± 0.67 h−1 (mean±S.D., N=1762) and 0.38 ± 0.32 h−1 (mean±S.D., N=291), respectively. Statistical features in the boxplot are interpreted in the same way as in Fig. 4A. Significance test was performed by two-tailed Mann-Whitney-Wilcoxon test for data sets following non-Gaussian distribution. n.s., not significant. (B,C) Distribution of volume growth rate β of cells in protocolonies (panel B) and in isolation (panel C). A negative growth rate here indicates a reduction in cell size, which was likely due to cell mass loss caused by shearing of neighboring cells. The statistical features of the distributions are shown in panel A. Also see Fig. S3.

Analysis of growth dynamics of bacterial protocells in aggregates and in isolation.

(A,B) Pie chart showing the proportions of five categories of protocell volume growth dynamics in aggregated protocolonies (panel A; N=1762) and in isolation (panel B; N=291). The classification criteria were based on the temporal dynamics of cell volume growth shown in panels C-K (Methods). The color coding in the legends applies to pie charts in both panels. (C-K) Temporal dynamics of cytoplasm-occupied area during the growth of cells in protocolonies (panels C-G) and in isolation (panels H-K). The color coding of the growth curves in panels C-K is the same as that in panels A,B. Each line corresponds to data from an individual cell, with T=0 min corresponding to the time of inoculation. The cytoplasm-occupied area, which is proportional to cell volume given the quasi-2D geometry, was measured by cell profiling based on fluorescence images of the cytoplasm stain CellROX Deep Red (Methods). Characteristics of each growth curve were evaluated, including standard deviation of data in the time sequence, growth rate obtained by exponential fit of the growth curve, and goodness of growth rate fitting (Methods). Cells were then classified into five categories according to criteria set based on these characteristics (Methods), which form the basis of color coding in all panels of this figure.

Division of cells in a protocolony.

(A) Fluorescence image sequence showing the morphological change of CellROX-stained bacterial protocells in a protocolony (Methods). As shown in the images, CellROX fluorescence intensity is higher at the interface between cells, which allows for efficient cell segmentation. A representative protocell undergoing division is highlighted in red. T=0 min corresponds to the start time of the recording. Scale bar, 5 μm. (B) Cell number in protocolonies plotted against time. T=0 min corresponds to the time of inoculating the protocolonies. For each protocolony, the cell number was normalized by the cell number at the beginning of the observation. The data in red color (associated with left vertical axis) plots the average normalized cell number of protocolonies over time; error bars represent standard deviation (N>50 protocolonies), and lines connecting the data points serve as guides to the eyes. Data were acquired in 9 protocolonies of variable duration of observation, and the gray line (associated with the right vertical axis) plots the number of protocolonies from which the data point at a specific time was acquired. The result in panel B shows that the cell number in protocolonies started to increase at ∼2.5 hr after inoculation, suggesting that cell division events took place primarily after ∼2.5 hr post inoculation.

Impact of cell division on cell size in protocolonies.

In both panels T=0 min corresponds to the time of inoculation, error bars represent standard deviation (N>50 cells), and lines connecting the data points serve as guides to the eyes. (A) Effective radius of cells in protocolonies plotted against growth time. The size or effective radius of a bacterial protocell is defined as , where A(t) is 2D projected cell area (Methods). Data in red color plot the actual average cell size, and data in violet color plot the average effective radius of virtual mother cells calculated based on the total area of daughter cells belonging to the same lineage (i.e., the area of a virtual mother cell at a specific time is the sum of all its surviving off-springs). Significance test for each pair of red and violet data sets at a specific time point was performed by two-tailed Mann-Whitney-Wilcoxon test to examine whether the two data sets were drawn from the same distribution (∗ P < 0.05). As shown in this panel, the red and violet data sets started to deviate from each other at T= ∼140 min, which corresponds to the onset of division events in the densely packed population; this result is consistent with data shown in Fig. S4B. (B) Comparison of the protocell size in aggregates and in isolation. Data in violet and green color correspond to virtual mother cells in protocolonies and isolated cells, respectively; the violet-color data are the same as those shown in panel A. Note that data of virtual mother cells are used for the comparison here in order to eliminate the impact of cell division on the cell size distribution. Significance test for each pair of green and violet data sets at a specific time point was performed in the same manner as in panel A. This panel shows that the average size of virtual mother cells in protocolonies is significantly smaller than that of isolated protocells during the initial growth stage (from T= 0 min to ∼150 min), but then it became comparable or even greater after T=∼150 min. Combining both panels of this figure and the main text Fig. 3A, we conclude that cell division is the primary cause of the significant size difference between cells in protocolonies and isolated protocells at the later stage of growth (> ∼150 min); the initial cell size difference (before ∼150 min) is presumably due to the fact that a larger proportion of cells in protocolonies were non-growing (37.3% in protocolonies vs. 22.0% for isolated cells; see Fig. S3A,B).

Parameters for Cellular Potts Model simulations.