Rapid phenotypic individualization of bacterial sister cells

A growing bacterium typically divides into two genetically identical and morphologically similar sister cells and eventually gives rise to a clonal population. Nevertheless, significant phenotypic differentiation among isogenic cells frequently occurs, with the resulting heterogeneity in cellular behavior often ensuring population level growth and survival in complex and unpredictable environments. Although several mechanisms underlying the generation of phenotypic heterogeneity have been elucidated, the speed with which identical sister cells tend to phenotypically diverge from each other has so far remained unaddressed. Using Escherichia coli as a model organism, we therefore examined the timing and dynamics of phenotypic individualization among sister cells by scrutinizing and modeling microscopically tracked clonally growing populations before and after a semi-lethal heat challenge. This analysis revealed that both survival probability and post-stress physiology of sister cells shift from highly similar to uncorrelated within the first decile of their cell cycles. This nearly-immediate post-fission randomization of sister cell fates highlights the potential of stochastic fluctuations during clonal growth to rapidly generate phenotypically independent individuals.

Supplemental text: survival is not governed by epigenetic 21 determinants nor by predispositions 22 23 As sister cells did not display a tendency to share the same fate in our assay, we set 24 forward to further investigate the potential absence of epigenetic factors and 25 predisposition factors governing survival in our assay.

27
In a first step, we investigated whether the observed intermicrocolony variation could be 28 caused by differentiation events occurring before cells were placed on the agarose pad 29 (instead of being the consequence of random stochasticity). On average, microcolonies Given the limited number of monitored microcolonies (n = 29), we set forward to examine 45 these findings further using a set of complementary approaches. We first employed 46 bootstrapping to assess the potential variability in average microcolony level survival. To 47 this end, we generated 10 000 bootstrap samples by sampling with replacement from our Wilk and Anderson-Darling tests on the empirical distribution indicated that the fraction 54 of surviving cells within a microcolony likely is normally distributed (p-value = 2.52 x 10 -55 1 and 1.17 x 10 -1 , respectively), a finding strengthened by the quantile-quantile plot 56 ( Figure S1C). Together, these findings again suggest that average microcolony level 57 survival is not significantly affected by epigenetically inherited factors predating the 58 beginning of TLFM recording, as these would give rise to highly variable, non-normal 59 microcolony survival frequencies.

61
We also examined whether any of the variability in survival frequency could be attributed 62 to a measurable property of the microcolonies. To this end, we looked for potential 63 correlations as these would allow us to identify possible determinants of increased or 64 decreased survival frequencies. However, only microcolony growth rate, expressed in 65 terms of area increase, appeared to marginally correlate with average survival within 66 microcolonies (Pearson's r = 0.3932, p-value = 3.77 x 10 -2 ), with slower microcolony 67 growth rates leading to slightly increased cellular survival ( Figure S1D). Other attributes, 68 such as the total number of cells or the area of the microcolony, did not correlate with 69 4 cellular survival frequency, altogether indicating that no major survival determinant was 70 active on the microcolony level ( Figure S1D).

72
Besides variation between microcolonies, we also investigated whether epigenetically 73 inherited determinants of cell fate (leading to increased or decreased chance of survival) 74 were active within individual microcolonies. If such factors were to exist and be decisive 75 for the outcome of our assay, surviving (and non-surviving) cells should be more closely 76 related to each other than expected by chance (random distribution). Therefore, we first 77 looked at the correlation between the survival rate of the progeny of second and third direct cell fate predisposition factors. Only instantaneous growth rate, expressed in terms 120 of length increase during the last 7 min of growth, appeared able to slightly (but not 121 significantly, Fisher's exact test, α = 0.01) influence a cell's fate, with, similar as to what 122 was observed on the microcolony level (Fig. S1D), slower growth rates leading to 123 increased chances of survival (Fig. S2B). This effect, however, appeared rather small and 124 would in no way be able to explain all the observed variability in survival.     (> 10 µm) were excluded from the analysis due to their tendency to give aberrant results.

325
The bisector is shown as a dashed orange line.