Cells were grown from one cell into a monolayer microcolony that contained up to 500 cells, and time-lapse images (see Video S1
for an example film) were analyzed with custom automated software designed for this purpose. We followed 94 such colonies, resulting in the complete record from division to division, of 35,049 cells. As the history and physical parameters of each cell in the microcolony are known, and the identity of each pole is tracked, the complete lineage can be determined. The resulting lineages from each film were averaged by each unique cell position within the lineage. This can be represented as a single, bifurcating tree, where each branch point is an average cell for that position in the lineage, and the length of the lines connecting cells to their progeny are proportional to the growth rate of the cell (). At each division in the tree, the cell inheriting the old pole of the progenitor cell is represented on the right branch of the sibling pair (in red), while the cell inheriting the new pole is on the left branch (in blue).
Average Lineage Showing Old Pole Effect on Growth Rate
The pattern of fast and slow growth rates in this average lineage gives striking evidence for reproductive asymmetry between the progeny cells, as the cells that show a cumulatively slowed rate of growth (shorter lines) are those cells that have more often inherited an old pole during their ancestry. To verify that this pattern in the average lineage is actually due to a difference in growth rate between new and old pole cells, we performed a pairwise comparison of every set of sister cells that was produced at the eighth generation in each of the films. As sister cells share temporal and spatial surroundings, this comparison controls for potential environmental variation within the microcolony. The comparison (two-tailed t-test) includes cells of all division ages and shows that the average growth rate of the old pole progeny cell is 2.2% (+/−0.1%) slower than that of the new pole cell. This analysis, performed on 7,953 cell pairs, conclusively demonstrates (p < 0.00001, t = 14.40, df = 7952) that the cell that inherits the old pole grows slower than the new pole cell produced in the same division. Two factors from this same dataset demonstrate the lack of a juvenile phase in E. coli. First, comparison of the progeny cells shows that the new pole cell is slightly larger on average (0.9+/−0.1%; p < 0.00001, t = 5.62, df = 7952) than the old pole cell (the contrary would be expected in the presence of a juvenile phase). Second, the young pole cell is marginally more likely to divide sooner than the old pole cell (in about 15% of the cases cells divide within the same 2-min time point; of those where the two cells divide in different time points, 54% of the time the new pole cell divides first [significant; p < 0.00001, t = 5.02, df = 4812]), which is also not consistent with a phase where the young cell must grow or differentiate before reproduction. These differences are consistent for all generations during steady-state growth (data not shown). Therefore, while a juvenile phase is absent, there is a consistent functional asymmetry between the two progeny cells that is disadvantageous to the old pole.
Each cell is defined not just by its preceding division, however, but also by all previously recorded divisions, back to the initial cell in the analysis. Therefore, each old pole cell can be categorized by the number of consecutive, final old pole divisions that occurred to produce the current cell (thus giving the age in divisions of its old pole). Equivalently, each new pole cell can be assigned a number of divisions that it sequentially divided as the new pole cell. Comparing these values with the growth rates of the cells, we find that the older the old pole of a cell is, the slower the growth rate of that cell, while cells with more consecutive new pole divisions exhibit increasing growth rates (A). Furthermore, a pairwise comparison shows that the difference in the growth rate between the old pole sibling (the mother cell) and the new pole sibling (daughter cell) increases with the increasing age of the mother (B). Therefore, the difference between pairs of progeny cells, as well as the pattern seen in the average lineage, is not only due to a decrease in growth rate of cells that have inherited the old pole, but also to an increase in the growth rate (for at least three divisions; subsequent divisions do not detectably improve) of cells that have repeatedly inherited the new pole.
The Effects of Consecutive Divisions as an Old or New Pole on Growth Rate
To determine the longer-term effect of inheriting the old pole, we performed a second pairwise analysis, comparing the total length of offspring produced by sister cells from the fifth generation until the end of tracking (this generation was chosen as each cell has the opportunity to progress through about three divisions). As the bacteria are rod shaped, the total length of cells produced is proportional to the biomass of the offspring. The results show that old pole cells produce less offspring biomass compared to their new pole sisters (3.1+/−0.3% less, p
< 0.00001, t
= 9.29, df
= 1565). Therefore, it appears that the slower growth rate of the old pole cells also results in a longer-term decreased ability to produce offspring biomass. Another long-term effect of aging is the probability of survival of the organism over time. During the growth of the microcolonies, sixteen cells were observed to cease growing; these cells never resumed growth during the course of the experiment. We have defined these cells as potentially dead cells and have analyzed their locations in the lineages. While these apparent deaths may ultimately be due to stochastic events, they show a statistically significant bias (p
= 0.01; see Materials and Methods
) toward increased divisions spent as an old pole (over the total observation history). This observation is consistent with the hypothesis that aged cells are more susceptible to harmful events and/or less likely to survive them. It is unlikely that these cells represent a growth arrested “persister” state, as it has recently been demonstrated that persister cells that arise during exponential growth occur at a frequency of approximately 1.2 × 10−6
]; the appearance of apparently dead cells in our study (about 4.6 × 10−4
) is almost 400 times more frequent.