Cell division and polarity in E. coli. (A) Assignment of old (red) and new (blue) poles and daughters. The starting cell is white because it was randomly picked to start the experiment and its polarity was unknown. Because the division plane (dashed line) cuts the white cell at the midpoint of the long axis, the poles formed at the division point are new and the distal poles are old. At the next division, the new daughter acquires the new pole, while the old daughter receives the old pole. Note that two divisions must be tracked to determine the old and new daughters from the starting white cell. The outlines of the bottom four daughters in the figure are colored red and blue to identify them as old and new daughters, while the intracellular red and blue colors identify the old and new poles, also designated as O and N. Although old and new poles and daughters are here tracked for only 2 generations, the notation and tracking methods can be extended into generations 3, 4, 5, 6, 7, and further if needed. For example, if the old and new poles of any of the four daughters after 2 divisions are known as in Fig. 1A, and these daughters elongate, become a mother, and divide to produce two grand-daugthers, the polarity of the grand-daughters can be determined by the same tracking methods.(B) Time-lapse images of an E. coli bacterium dividing into two and four cells. Top row: Phase contrast. Second row: Assignment of old (red) and new (blue) poles from the top row cells. Third row: Fluorescence images matching phase images in time and position. Bottom row: Heat-map of fluorescence images reporting ribosome density in the top row cells, showing lesser intensity by the old poles (blue color spots) than the new poles and inside the cells (purple color). Scale on the right shows intensity gradient. (C) Schematic showing division of a cell into two halves containing either the new pole or the old pole. The ratio of the fluorescence in the new half divided by that in the old half (or pole ratio) was used to normalize and pool different old and new daughter pairs, when each pair descended from a different mother. (D) Schematic showing division of a cell into four length quartiles denoted NP, L2, L3 and OP in relation to the old and new poles. These quartiles were used to quantify the distribution of fluorescence along the length of a cell.

Ratios of ribosome density and elongation rates within and between cells. Because ratios are a form of normalization, their values were derived from not normalized data. Asterisks in figures indicate levels of significance from comparison between ratios of new (blue bar) versus old (red bar) mothers (* ** ***, significance at p < 0.05, 0.01, and 0.001). Error bars are SEM. Values of p in text below are for the significance probability that a ratio is greater than 1.0, unless indicated otherwise. Sample size for ratios is for pairs of data: n = 89 pairs correspond to 89 old/new daughters of the same mother. Each pair is used to obtain one ratio to yield 89 ratios. (A) Ribosome ratio of daughters (new/old) at birth from old (red) and new (blue) mother cells. The ratio was 1.11 ± 0.018 from old mothers (p = 6.89×10-6, one-tailed paired t-test, n = 89 pairs) and 1.04 ± 0.015 from new mothers (p = 0.04, one-tailed paired t-test, n = 91 pairs). The two ratios were also significantly different from each other (p= 0.007, two-tailed non-paired t-test, ** in figure). (B) Ribosome ratio (new/old) of the two polar halves (Fig. 1C) from old (red) and new (blue) mothers at division. The ratio in old mothers was 1.10 ± 0.015 (p = 6.23×10-5, one-tailed paired t-test, n = 89 pairs) and from new mothers 1.03 ± 0.010 (p = 0.08, one-tailed paired t-test, n = 91 pairs). The difference between the two ratios was significant (p= 1.3×10-4, two-tailed non-paired t-test, *** in figure). A comparison of the daughter and polar half ratios from old mothers (Figs. 2A, 2B; red bars) found no significant difference (p = 0.62; two-tailed paired t-test). A likewise comparison for new mothers (Figs. 2A, 2B; blue bars) also found no significance (p = 0.27). (C) Elongation rate ratio of new over old daughters from new (blue) and old (red) mothers. The ratio from old mothers was 1.11 ± 0 .017 (p = 1.3×10- 10, one-tailed paired t-test, n = 216 pairs) and from new mothers 1.07 ± 0.011 (p = 2.76×10-7, one-tailed paired t-test, n = 198 pairs). The two ratios were significantly different from each other (p= 0.02, two-tailed non-paired t-test; * in figure). Note that these elongation rate ratios parallel the ribosomal pattern of a higher asymmetry in daughters from old mothers (Fig. 2A).

Variation in ribosome density and elongation rate in old and new daughters from old and new mothers. The measurements were made from the same cells in Fig. 2C. All density and elongation values were normalized for this analysis. Means, variances, and other parameters extracted from these data are presented in Fig. 4. (A) Top panel: Normalized density distribution of ribosome density of old (red line) and new (blue line) daughters from old mothers. Red and blue dots on the x-axis indicate the mean density for each distribution. Dashed lines represent the density distribution of the old and new daughters combined into one pooled total population. D (black arrow) shows the distance between peaks of old and new daughter curves (*** p=2.2×10-16 for significance of D>0; 1-tailed paired t-test; n=216 daughter pairs; see Fig. 4A). Bottom panel: Normalized ribosome density of each old (red) and new (blue) daughter in pairs from old mothers. The zero point is set as the average ribosome density for each pair. As shown, the old daughter in each pair more often ends up on the minus side of the pair’s zero point, i.e. having lower ribosome density. (B) Same as (A), but from new mothers (* p=0.014 for D>0; 1-tailed paired t-test; n=198 daughter pairs; see Fig. 4A). (C) Normalized elongation rate distributions (*** p=2.0×10-11 for D>0; 1-tailed paired t-test; n=216 daughter pairs; see Fig. 4B), but otherwise as (A). (D) Same as (C), but from new mothers (*** p=2.1×10-7 for D>0; 1-tailed paired t-test; n=198 daughter pairs; see Fig. 4B).

Variance components of ribosome density and elongation rates. Total (VT), stochastic (VE) and deterministic (D2/4) components of variances estimated from Fig. 3 (see Materials and Methods). The deterministic proportion of total variance VT was h2 = (D2/4) / VT. The stochastic component of VT is therefore VE = VT (1 – h2). Significance testing for D>0 as described in Fig. 3 (*, **, and *** denote p<0.05, 0.01, and 0.001). (A) Estimates for ribosomal density. (B) Estimates for elongation rates. (C) Comparison of VE estimates. Testing of significance difference between VE was by a Bartlett’s test of homogeneity of variances. p-values1 are for comparisons across rows (old versus new mothers). p-values2 are for significance of VE down columns (elongation rate vs. ribosome density).

Correlation between normalized elongation rate and ribosome density of old and new daughters from old and new mothers. Because the old and new daughter values were jointly normalized, they were not independent and their combination into a single plot could not be assessed by statistical models assuming independence. Thus, all analyses for the figure were conducted by randomizing the data and obtaining a null distribution to estimate the significance values (see Materials and Methods). (A) Ribosome density versus elongation rate between new and old daughters from old mothers. Slope of linear regression = 0.498, correlation r = 0.387, p < 1 x 10-5, n = 216 daughter pairs. (B) Ribosome density versus elongation rate between new and old daughters from new mothers. Slope of linear regression = 0.301, correlation r = 0.233, p = 4 x 10-4, n = 198 daughter pairs.

Ratio of ribosome density in old and new mothers over quartile time from birth to division. Ratios correspond to density in the new pole half divided by density in the old pole half (Fig. 1C) of the mothers. Regression slopes were significantly greater than zero for old (slope=0.0177, * p=0.026, df=278) but not for new mothers (slope=0.0039, n.s. p=0.59, df=278). Regression was based on ratios that were binned by the four time quartiles (Old mothers: n=71, 74, 46, 89; df=278. New mothers: n=68, 72, 49, 91; df=278), although the plots only show the ribosome means of the four time bins for illustration. Old mothers began with a higher asymmetry ratio and elevated to even higher levels at division. On the other hand, new mothers had a lower and more symmetrical ratio that was held more constant from birth to division.

Spatial and temporal distribution of ribosome density in old and new mothers. (A) Density in length quartiles over time quartiles. Lines presented are from regressions of ribosome densities for one length quartile that were binned by the four time quartiles. All length quartiles for a mother type had time bins of the same sample size (old mothers: n=71, 74, 46, 89; new mothers: n=68, 72, 49, 91). The single dots presented are ribosome density means of the four time bins and are presented for illustration. (B) Values of regression slopes of the 8 lines in Fig. 7A. All regression slopes were significantly greater or less than, i.e. not equal to, zero (Old mothers: p=0.015, 1.46×10- 11, 0.00029, 7.8×10-6; df=278. New mothers: p=0.0012, 1.7×10-10, 3.5×10-10, 4.2×10-6; df=278). To test whether old mothers were more asymmetrical than new ones, the regression slopes of the pairs NP vs. OP and L2 vs. L3 for each mother type were compared (see Materials and Methods for details). The comparisons test whether length quartiles of the mothers on the left new pole side mirrors the right old pole side. The horizontal brackets in the figure denote the comparisons and whether they were significant. Old mothers were asymmetrical because their comparisons were all significantly different (NP vs. OP and L2 vs. L3; p=0.04 and 0.008; df=158 and 112). New mothers were symmetrical as the differences were not significant (NP vs. OP and L2 vs. L3; p=0.11 and 0.44; df=156 and 115).

Fluorescence profile along new and old daughter pairs’ cell lengths at birth. The fluorescence and length values were normalized (n = 89 pairs). Color designation is as in Fig. 1A for old (red) and new (blue) poles and daughters. Error bars show standard error of the mean (s.e.m.). These transects are shown only for visualization and were not used in the final analyses, which were based on halves and quartiles (Figs. 1C and 1D).