Fig 1.
Colonies observed after four days of growth varied by the number of originating progenies.
Of all locations in which colony formation was observed, approximately 35% originated from a single tracked cell, 48% originated from two analyzed cells, and the remainder formed from three or four originating cells that attached near each other at the start of the experiment. Phase contrast images of three representative colonies after four days of growth are displayed, organized by the number of analyzed cells the colonies originated from. Images are montages of nine or sixteen fields of view and were altered post-experiment to artificially color the BMSCs by the progeny they belong to. Cells without color added are BMSCs that migrated into the montaged field of view after initial imaging. Lineage trees of the colonies’ originating cells are also presented, where the width of the lineage lines is representative of the respective cell spread area at each 15 min time point. Lineage trees and images of all colonies at several time points are presented in S1 Fig. Lineage tree outlines indicate the proliferative capacity of the progeny. Dashed line (- -): progeny classified as slow proliferator; dotted line (…): progeny classified as moderate proliferator; solid line (–): progeny classified as fast proliferator. Scale bars = 0.5 mm.
Fig 2.
Colonies varied by size, confluency, number of originating progenies, and degree of isolation after seven days of growth.
Binarized and recolored phase contrast images of all analyzed colonies after seven days of growth. The color of each imaged colony was altered to represent the number of originating cells associated with each colony, as analyzed by time lapse image analysis at the single-cell level up to Day 4. Borders of neighboring colonies that were either touching or overlapping the colonies studied are represented as black arcs and were traced based on the original phase contrast images. The size of the 16 montaged fields of view at Day 7 is also provided for reference. No statistical correlation was found between the number of cells the colony originated from and colony diameter, confluency, or number of neighboring colonies. (See S2 Fig for further details.) Colonies are ordered by approximate diameter. Unaltered phase contrast images of all colonies are provided in S1 Fig.
Fig 3.
Principal component analysis revealed a trend in generation associated with cell behavior.
(A) A total of 128 variables (listed in Methods and S2 Table) related to morphology, location, proliferation, and association with neighboring cells were analyzed using principal component analysis (PCA) for 1384 cells belonging to 28 colonies. k-means clustering revealed no clusters that were especially distinct, indicating that the biophysical properties of these cells behave as a spectrum rather than signifying discrete subpopulations. (B) Color-coding the same PCA plot by generation revealed a trend in PC1-PC2 space. (C) The same PCA plots are color-coded such that only cells from one colony are highlighted, demonstrating that cells from individual colonies did not cluster together. (D) PCA was also conducted on data subsets containing cell properties from only one colony at a time. The trend in generation was observed in individual colonies to varying extents. All plots highlighting individual colonies were chosen at random using the “rand” function in MATLAB.
Fig 4.
Cells within a colony after four days of growth varied by their size, generation, proximity to other cells, and the proliferative capacity of the progeny they belonged to.
Glyphs representative of each colony studied at the single-cell level at Day 4 are presented. Cells are represented by filled circles, which are color-coded by generation and oriented based on the coordinates of the final time point of the single-cell time lapse image analysis. Similarly, the spread area of each cell is represented by the size of the color-filled circles. The outer rings of each colony represent the number and proliferative capacity of the originating progeny; slow dividers had less than three population doublings and fast dividers had more than four population doublings by Day 4. The example provided in the upper-left-hand corner is a representation of a colony after four days of development that originated from two cells, one of which proliferated slowly and one that proliferated moderately. This example colony had 10 cells at Day 4 ranging from first to fifth generation, with the first generation cell being relatively isolated and having a large spread area. The colonies are ordered horizontally by their approximate diameter.
Fig 5.
Most colonies contained senescent cells after four days of development, which is partially detectable by deviations in area dynamics.
The spread area of cells was measured at every fifteen-minute time point for every colony during the first four days of development. Additionally, cells in each colony at Day 4 were monitored via time lapse videos and marked as senescent if they did not divide by Day 7. (A) The number of proliferative and senescent cells at Day 4 is reported. The color of the colony label represents the number of originating progenies, and colonies are rank-ordered by the fraction of cells that were senescent. (B) The Spearman correlation coefficient rs and associated p value for the comparison of average cell area versus generation is presented for all progenies contributing to colony formation. The majority of progenies exhibited a negative trend between area and generation to varying degrees of statistical significance. The dashed horizontal line indicates the Bonferroni-corrected p value threshold for a significant correlation, and the color of the progeny data points represents the number of originating cells their respective colony developed from. (C) Cells belonging to all colonies also showed a negative trend between area and generation when their datasets were pooled, where the area of each cell was calculated as the average area of all time points during the first 0.83 days of their lifetimes. The Spearman correlation coefficient and associated p value is also reported (N = 1384). (D) Line plots of individual cell area in single progenies over time are presented for three progenies. Data lines are color-coded by cell generation, with the exception of senescent cells (labeled black). Many progenies uniformly produced smaller cells with each new generation (left); however, heterogeneity onset occurred when cells deviated from this trend, occasionally due to senescence (middle) or from cells having longer lifetimes (right). Also displayed for each progeny is the Spearman correlation coefficient and associated p- value for the comparison of average cell area versus generation, as presented in (B).
Fig 6.
Proliferation continues to occur within confluent colonies.
The time and location of each cell division–as well as colony confluency–was measured for the first ten days of colony development in one colony. Mitotic events are reported in terms of their location relative to the colony center at each time point, and the shade of the data points is representative of the measured colony confluency (left). Cells continued to proliferate even at the center of the confluent colony after ten days of development. A phase contrast image of the colony at the final time point is provided (right) as a reference point of the measured confluency. Analysis of a second colony is provided in S6 Fig. Scale bar = 0.5 mm.
Fig 7.
Heterogeneity arose in the most uniform single-cell-derived (SCD) colonies within seven days of development.
(A) The lineage tree of one SCD colony studied long-term at the single-cell level demonstrates that even a uniformly-proliferating colony will produce large, senescent cells and asynchronous divisions within its progeny before becoming fully developed. Lineage line widths represent cell spread area at each 15-minute time point. (B) Glyphs similar to those presented in Fig 4 for two SCD colonies (see Fig 4 for key) demonstrate the onset of intracolony heterogeneity. Colony L was classified as a moderate proliferator and consisted of 16 5th generation cells at Day 4, and Colony A consisted of 32 6th-generation cells at Day 4. Both colonies developed a distribution in cell area (represented by the size of the color-filled circles) and generation (represented by color) at the later time points indicated. (C) Area-vs-time curves of individual cells for the same two SCD colonies. Graph lines are again color-coded by cell generation. Both colonies presented a decrease in cell area with each new generation, a trend that ceased at approximately the time point the cell medium was replaced (indicated by vertical dashed lines) as part of the conventional BMSC culture protocols of focus in this study.
Table 1.
All properties measured by CellProfiler at every 15 min time point and considered in further analysis.