A mother Escherichia coli cell divides asymmetrically into two daughter cells that have different elongation rates (Stewart et al., 2005; Chao, 2010; Shi et al., 2020; Proenca et al., 2018; Proenca et al., 2019; Łapińska et al., 2019). The new daughter elongates faster while the old daughter is slower. The asymmetry and old-new designation of the daughters result from the fact that the rod-shaped mother divides by forming new poles at its midplane. As a result, all E. coli cells are polarized at birth with a new and an old pole (Figure 1A). When a polarized cell in turn becomes a mother and divides, its daughters will also be polarized but they can also be identified as new and old daughters depending on whether they acquired the mother’s new or old pole. Note that mothers can also be denoted as new or old, depending on whether they were new or old daughters at birth. The difference in elongation rate between old and new daughters was found by Lindner et al., 2008 to correlate negatively with the amount of damaged proteins (inclusion bodies) in individual cells. They also found that protein aggregates were more often associated with the old pole of the mother, in which case damaged proteins are partitioned asymmetrically by the mother to its daughters. However, a study by Govers et al., 2018 has reported a weaker and nonsignificant negative correlation. Because the two studies used different external stressors (to induce damage) and fluorescent reporters, they are not fully comparable and the relationship between elongation rate and the damaged proteins in E. coli remains unresolved (see ‘Discussion’ for additional details). Despite the differences between Lindner et al. and Govers et al., protein aggregates often have a negative qualitative effect on cell health and viability (Maisonneuve et al., 2008; Gatti-Lafranconi et al., 2011; Schramm et al., 2020; Rang et al., 2018), and more so in old daughters (Vedel et al., 2016; Winkler et al., 2010).

Figure 1 with 1 supplement see all

Download asset Open asset

The asymmetry of both the elongation rate difference between old and new daughters and the damage partitioning by the mother in E. coli has evolutionary consequences. The asymmetry creates fitness variance between daughters that increases the effectiveness of natural selection (Chao, 2010; Chao et al., 2016), and an asymmetrical lineage has more progeny over time relative to a symmetrical one. An analogy comes from economic models. Two asymmetrical $500 bank accounts at 10 and 6% yield more money over time than two symmetrical $500 accounts at 8%. However, the asymmetry comes with costs. While the higher elongation rate of the new daughter rises in the lineage over generations, the rate of the old daughter declines (Stewart et al., 2005; Chao, 2010). Under more benign culture conditions, the rise and the decline can achieve stable values and the lineages are immortal (Chao, 2010; Proenca et al., 2018; Łapińska et al., 2019; Rang et al., 2011; Rang et al., 2012). Under more stressful and damage-inducing conditions, the old daughter lineage can die because its elongation rate declines to zero and the death probability is dosage dependent (Proenca et al., 2018; Proenca et al., 2019). The bacterial population survives despite the death of the old daughter lineage because the new daughter lineage is still able to achieve a stable rate that is greater than zero (Chao, 2010; Proenca et al., 2018; Proenca et al., 2019). Because the elongation rate declines and death of the old daughter lineage is a form of physiological aging, bacteria can be used to investigate the evolutionary origins of biological aging. Despite a quantitative correlation between the amount of damaged proteins and elongation rates remaining unresolved (see above), the asymmetries here described for elongation rates and death strongly suggest that damage stress in E. coli may play a key, if not causal, role in cell aging. We note that the asymmetrical partitioning of damage is not an alternative to protein repair, or vice versa. Protein repair (including reprocessing) is occurring (Ezraty et al., 2017; Merdanovic et al., 2011) and its effect is to lower the damage rate experienced by the cell. However, if damage aggregates are present in cells, the damage rate must necessarily exceed the repair rate. There is likely a cost to repair, in which case the repair rate may become limited by diminishing returns. Thus, the asymmetry could be an evolutionary adaptation for dealing with a low repair rate. If repair rates were sufficiently high to negate damage, E. coli should not have evolved asymmetry.

To develop further our understanding bacterial aging, our previous study examined the distribution of expressed native (not fused to another protein) GFP molecules in E. coli (Shi et al., 2020). We found that the GFP density was higher in new poles than in old ones and in cells with a faster elongation rate. To take the analysis one step further, we present here results from our examination of the distribution of ribosomes in the poles and daughters of single E. coli cells. We used the yfp (yellow fluorescent protein) and S2 subunit (rpsB gene) fusion by Bakshi et al., 2012 as a reporter for ribosomal density. The S2-YFP construct is an effective fluorescent reporter. The S2 ribosomal subunit makes a desirable reporting partner because free subunits are negligible and nearly all cellular S2-YFP is incorporated into ribosomes (Bakshi et al., 2012; Sanamrad et al., 2014). Almost all of the detected fluorescent S2-YFP comes from copies are already incorporated in ribosomes (Bakshi et al., 2012). Bakshi et al., 2012 used S2-YFP to successfully make a more detailed map of the nucleoid/ribosome segregation. Most importantly, because fluorescent reporters can produce artifacts (Govers et al., 2018; Swulius and Jensen, 2012; Landgraf et al., 2012), YFP and S2-YFP were extensively tested to validate that the reported fluorescence was not artifactual. These tests are detailed next.

The tests validating YFP and S2-YFP followed the methods of Landgraf et al., 2012 by which validation requires that either fluorescent aggregation, clusters, or foci are reported when non-uniformity is expected by an assay or alternatively not reported when uniformity is expected. An artifact occurs when uniformity is reported when non-uniformity is expected (and vice versa). Because reporting artifacts depend on the protein to which a fluorescent reporter is fused (Landgraf et al., 2012), a reporter that makes artifacts with one protein does not necessarily make it with a second. Thus, many of the tests were targeted specifically to the S2-YFP construct. (i) A first test focused on native YFP (not fused to any protein) because an earlier study suggested that YFP produced artifactual non-uniformity due to self-aggregation (Govers et al., 2018). This test was needed because the self-aggregation results were based on a fusion construct of the chaperone IbpA and YFP. However, was the self-aggregation a property of YFP or an interaction with IbpA? Bakshi et al., 2012 found that native YFP reported a uniform distribution, in which case YFP is not inherently prone to self-aggregation artifacts (see also ‘IbpA-YFP self-aggregation’ in ‘Discussion’). (ii) If YFP is fused to OmpR, which is a regulator that binds to the OmpC and OmpF loci, and the three loci (OmpR-YFP, OmpC, and OmpF) are expressed from the bacterial chromosome (hereafter strain A), the fluorescence distribution is uniform (Batchelor and Goulian, 2006). Because the chromosome has a low copy number, most OmpR-YFP is distributed over the entire cell. Some OmpR-YFP is bound to OmpC and OmpF but they are too few to produce a fluorescence focus. If OmpC and OmpF are cloned into the multi-copy plasmid pBR322, which forms clusters in cells (Pogliano et al., 2001), and the plasmid is inserted into strain A (making strain B), a fluorescence focus reporting the plasmid cluster emerges (Batchelor and Goulian, 2006). Additionally, if native OmpR (not fused to YFP or any fluorescent reporter) is over-expressed from a second plasmid that is inserted into strain B (making strain C), the fluorescence focus disappears because the more numerous native OmpR outcompetes OmpR-YFP in binding to the pBR322/OmpC-OmpF cluster. Thus, OmpR-YFP is responding correctly when either uniformity or non-uniformity is expected. (iii) To confirm that the nucleoid/ribosome segregation they observed was not due to YFP reporting artifacts, Bakshi et al. repeated their experiments by adding rifampicin and chloramphenicol, two antibiotics that affect the distribution of the nucleoid DNA (Cabrera et al., 2009). Whereas the nucleoid is expanded and rendered more uniform by rifampicin, it is compacted and clumped by chloramphenicol. Bakshi et al., 2012 found that the fluorescence reported by S2-YFP became more uniform with rifampicin but more clumped and non-uniform with chloramphenicol. As the nucleoid expanded with rifampicin, the ribosome distribution also expanded as it filled the spaces created by the nucleoid expansion. On the other hand, as the nucleoid contracted with chloramphenicol, the ribosomes were excluded from the space occupied by the more clumped nucleoid and they in turn became less uniform. In summary, these YFP, OmpR-, and S2- controls give us confidence that S2-YFP is a reliable reporter for ribosomes and that reported fluorescence foci are better explained by the non-uniform properties of the cell rather than any inherent property or artifact of YFP. We find these tests that functionally validate S2-YFP to be more convincing than examining S2 fused with more fluorescent reporters. If two reporters disagree, which one is the artifact? If they agree, are they both truthful or artifactual?

Our examination of the distribution of ribosomes with S2-YFP could be particularly instructive. First, while Bakshi et al. have shown that ribosomes congregate in the poles, it is not known if ribosomal densities differ between old and new daughters or their poles. Second, our previous report of more expressed gene products in new poles and daughters (Shi et al., 2020) could have resulted from regulatory factors acting on a specific gene or at a higher upstream level such as ribosomal abundance. In the latter case, the consequences would be much more fundamental as they could affect the entire proteome. The fact that ribosomal density and elongation rates are highly correlated in bacteria (Schaechter et al., 1958; Neidhardt and Magasanik, 1960; Kjeldgaard and Kurland, 1963; Poulsen et al., 1995; Greulich et al., 2015) additionally reinforces the importance of knowing the distribution of ribosomes in the poles and daughters of E. coli. Lastly, given that ribosomes occupy nearly 40% of the cytoplasmic volume of an E. coli cell (Woldringh and Nanninga, 1985) and that both ribosomes and damage aggregates tend to reside in the poles, the possibility that the two could be excluding each other at the poles becomes a probability. However, a space-limiting model as a hypothesis for explaining bacterial aging requires first that the distribution of ribosomes in old and new poles or daughters be characterized. Here, we present the results of our study quantifying the temporal and spatial distribution of ribosomes in single E. coli cells by time-lapse microscopy.

Following ‘Materials and methods’, the distribution of ribosomes was measured in the E. coli strain AFS55 with a fluorescent yfp reporter fused to the ribosomal subunit S25 (Bakshi et al., 2012). Fluorescence levels were quantified from time-lapse images of single cells cultured on agarose pads. Because the level recorded for a cell on an image is affected by the light that the cell loses by diffraction and gains by diffraction from neighboring cells, the diffractional scatter of all images was corrected by deconvolution. Red and blue colors in the figures represent old and new daughters or mothers, respectively. Ratio data (Figure 1C) were derived from estimates that were not normalized. All other data for cell length, ribosomal density, and time were normalized when pooling and comparing cells. All error bars are ± SEM and statistical details are presented in figure legends.

We first examined the density of ribosomes in colonies starting from a single cell. A quick visual inspection of the fluorescent images and their heatmaps for a few representative cells showed clearly that the ribosomal densities tended to be highest near the poles (Figure 1B). However, high densities also built up in the mid-region of a mother cell right before division. As the mid-region leads to the formation of the new poles of the daughters, the buildup explains how those poles acquired their higher levels. These polar concentrations of ribosomal densities matched the previous ones reported by Bakshi et al., 2012. To obtain a more quantitative analysis of the changes in ribosomal densities during cell growth, we collected a larger sample size and compared measurements between old and new mothers, old and new daughters, between their poles, along the length of their cells, and during their elongation from birth to division.

New daughters at birth from old mothers have more ribosomes

A first comparison revealed that if a pair of old and new daughters descending from the same mother were compared, ribosome density measured by the fluorescence was higher for new daughters. The ratio of reported fluorescence between new/old daughters (the daughter ratio) was 1.08 ± 0.012, which was significantly greater than 1.00 (p=4.38 × 10^–6). However, if the mothers were split into pools of old and new mothers (Figure 2A), the daughter ratio from old mothers was 1.11 ± 0.018 and significantly greater than 1.00 (p=6.89 × 10–6) and the ratio from new mothers was 1.04 ± 0.015 and barely significant (p=0.04). Thus, the higher ribosome density in new daughters resulted primarily from old mothers. The difference between old and new mothers was further demonstrated by a strong significant difference between their daughter ratios (1.11 vs. 1.04; p=0.007). Thus, new daughters have more ribosomes relative to old daughters produced by the same mother, but asymmetry is higher in pairs from old mothers. This higher density of ribosomes in new daughters correlates qualitatively with our previously reported higher levels of expressed gene products and elongation rates in new daughters (Chao, 2010; Shi et al., 2020; Proenca et al., 2018; Proenca et al., 2019).

Figure 2

Download asset Open asset

Stochastic and deterministic components of ribosomal variance in daughter cells

Although new and old daughters can differ significantly in ribosome density, there was also substantial variation within each daughter population. A plot of the distributions shows clearly the within-population variation, although the difference between old and new daughters is also apparent (Figure 3A and B). Old and new daughters form their own distributions, but their mean values are displaced by a difference D, which is larger when the daughters are derived from old mothers. The variation within the old and new daughter distributions is presumably due to stochastic noise, although the stochasticity could also result from an unknown deterministic source. The latter is well illustrated by the discovery that E. coli mothers produce asymmetrical daughters. For example, without knowledge about the asymmetry, old and new daughters would have been pooled into one population with a total variance V_T (Figure 3A and B; dashed lines), which would have been interpreted to result from stochasticity. In reality, V_T would actually have a deterministic component due to the difference between old and new daughters. As we have shown for expressed gene products (Shi et al., 2020), V_T = (V_Old +V_New)/2 + D²/4, when there is deterministic asymmetry (D > 0) and V_Old and V_New are the variances of the unpooled old and new daughter populations (Figure 3A and B; red and blue lines). If D = 0, V_T is the average of V_Old and V_New. Thus, D²/4 represents the deterministic component of V_T, and the contribution of deterministic asymmetry expressed as a percentage is h² = (D²/4)/V_T. Our estimates of D showed that h² was 39.9% from old mothers and 2.4% from new mothers (Table 1A). The higher h² for daughters from old mothers is consistent with our results in Figure 3A and B that show old mothers having more different daughters than new mothers. These results were also consistent with our previous report that h² is higher for expressed gene products in daughters of old than of new mothers (h² of 40.1 and 10.1%) (Shi et al., 2020). Thus, a claim of stochasticity to explain the variance of ribosomal density of E. coli cells could have introduced an error of nearly 40% without knowledge of the asymmetry between old and new daughters. Without knowing more, we presume that the variances V_Old and V_New result from stochasticity, but it may also be that a deterministic process still lies hidden.

Figure 3

Download asset Open asset

Table 1

(A) Ribosomal density
	Vr		D		D2/4		h2
Old mothers	0.00634		0.101 ***		0.00253		0.399
New mothers	0.00347		0.0184 *		0.00008		0.024
(B) Elongation rates
	Vr		D		D2/4		h2
Old mothers	0.00913		0.0821 ***		0.00168		0.185
New mothers	0.00581		0.0533 ***		0.00071		0.122
(C) VE estimates
		Old mothers		New mothers		p-value1
Elongation rates		0.00744		0.0051		0.0073**
Ribosome density		0.00381		0.00338		0.40 n.s.
p-value2		0.000001***		0.0041**

Asymmetry and variance components for elongation rates in daughter cells

To determine whether the observed ribosome patterns reported in Figure 3A and B had a downstream effect, we measured elongation rates of the same daughters and mothers. When new and old daughters were pooled irrespective of whether they originated from old or new mothers, new daughters were found to elongate faster than old daughters (new/old ratio of 1.09 ± 0.011; p=3.9 × 10^–16). When the old and new daughters were separated as descending from old or new mothers, the ratio was respectively 1.11 ± 0.017 (p=1.3 × 10^–10) and 1.07 ± 0.011 (p=2.8 × 10^–7) (Figure 2C). Although both ratios were significantly greater than 1.00, the asymmetry was greater between daughters from old mothers and the two ratios were significantly different (p=0.02). As we did for the ribosome data, the variance of elongation rate measurements was also portioned into a deterministic and stochastic component to estimate the percentage h² (Figures 3C, D, 4B). The deterministic percentage of the variance was found to be h² = 18.5% and 12.2% for daughters originating from old and new mothers, respectively. As we observed for ribosome densities, the h² for old mothers was still higher, but not to the same magnitude (see ‘Discussion’). The overall similarity between the ribosome and elongation rate data reflects the fact that the two phenotypes are well known to be positively correlated in E. coli (Schaechter et al., 1958; Neidhardt and Magasanik, 1960; Kjeldgaard and Kurland, 1963; Poulsen et al., 1995; Greulich et al., 2015). An examination of the relationship between our measurements for ribosome and elongation yielded similar outcomes (Figure 4). Daughters from both old and new mothers had ribosome densities that correlated positively with elongation rates (r = 0.387, p<1 × 10^–5 and r = 0.233, p=4 × 10^–4, respectively; see Figure 4 for more details). The correlation from new mothers was weaker because its deterministic component h² was smaller (Table 1B), in which case the distributions of new and old daughters were less separated (Figure 4A and B).

Figure 4

Download asset Open asset

Spatial distribution of ribosomes in mother cells from birth to division

Our observation that the asymmetry of the pole halves of mothers determines the distribution of ribosomes of daughters at birth (Figure 2A and B) raises the question of how the spatial distribution of ribosomes changes temporally in the mother cell, and whether the changes provide insights or hypotheses for the origin of the asymmetry. Thus, we increased the resolution of the process by sampling over a finer spatial (lengthwise) and temporal (birth to division) grain of the ribosomal density in a mother cell. The length of the mother cell was divided into four quartiles, which were termed NP (new pole), L2, L3, and OP (old pole) (Figure 1D). The time from birth to division of the mother cell was also divided into four quartiles termed birth (B), T2, T3, and division (D). Ribosome density and the quartiles of length and time were all normalized for pooling the cells. Note that because ribosome densities are normalized, they indicate relative rather than absolute amounts and the former can change even when the latter is constant.

We first examined whether the new length quartile results were consistent with our previous ribosome pole ratios (Figure 2B) for the new and old pole halves in the mothers. These new ribosome pole ratio data were obtained by taking the ratio of the length quartiles (NP +L2) / (L3 + OP) and plotting the values over the time quartiles (Figure 5). The ratios increased from birth to division in both old and new mothers, but increase was significant only for the old mothers (p=0.026 vs. 0.59, respectively). The rapid rise in the old mothers also began from a higher value at birth. More importantly, the final ratio values at quartile D (1.11 and 1.02 for old and new mothers, respectively) were comparable to Figure 2C (1.11 and 1.04), which was also timed to division. Thus, old mothers magnified the asymmetry of their ribosome distribution over time to achieve the higher ratio that we used above to explain the larger asymmetry of their daughters. On the other hand, the new mothers began with a pole ratio close to 1.02 and did not change over time, which explained the higher symmetry of their daughters.

Figure 5

Download asset Open asset

By using the length into quartiles NP, L2, L3, and OP as individual measurements, we were able to examine in more detail the trajectories for the changes of ribosome densities of mothers from birth to division (Figure 6A). Several key patterns emerged. First, the trajectories decreased in NP and OP and increased in L2 and L3 in both old and new mothers. Moreover, the density at the final time quartile D was L2 > NP and L3 > OP. Because L2 and L3 are the new poles of the daughters after division (Figure 1A and Figure 6A), the daughters were already being provisioned by their mothers to be biased to have more ribosomes at their future new poles at the expense of their future old poles. Second, the trajectories of new mothers in NP-L2 were mirrored by L3-OP (Figure 6A; blue lines), that is, NP was more similar to OP, and L2 more similar to L3. The trajectory slopes of NP vs. OP (–0.00520 ± 0.00313 to –0.00809 ± 0.00340; p>0.05) and L2 vs. L3 (0.00675 ± 0.00200 vs. 0.00655 ± 0.00198; p>0.05) were not significantly different (Figure 6B). Thus, ribosomes were being redistributed from NP and OP to L2 and L3 in an equal manner and the lower asymmetry of new mothers was being maintained (Figure 2B). Third, the trajectories for the old mothers had a higher asymmetrical signature (Figure 6B; red lines). They resided above the new mother trajectories in NP and declined until they were below in OP. Moreover, the trajectory slopes of NP vs. OP (–0.00390 ± 0.00313 to –0.00803 ± 0.00347; p<0.05) and L2 vs. L3 (0.00788 ± 0.00220 vs. 0.00405 ± 0.00217; p<0.05) were now significantly different (Figure 6B). Thus, relative to L3 and OP, NP and L2 not only began at birth with more ribosomes, but also NP lost fewer and L2 acquired more ribosomes over time. As a result, the old mothers became even more asymmetric than new mothers at division.

Figure 6

Download asset Open asset

The discovery that the division of mother E. coli cells produces two asymmetric daughters that have different elongation rates was groundbreaking (Stewart et al., 2005). The old daughter receiving the maternal old pole (Figure 1A) elongates more slowly than the new daughter receiving the maternal new pole. The difference is not due to genetic mutations because the daughter with the slower rate can in turn produce a grand-daughter with the elevated rate (Stewart et al., 2005; Chao, 2010; Proenca et al., 2018; Proenca et al., 2019; Rang et al., 2011). These results have been replicated on agar surfaces (Stewart et al., 2005; Shi et al., 2020; Rang et al., 2018; Rang et al., 2012) and in liquid media (Proenca et al., 2018; Proenca et al., 2019; Łapińska et al., 2019; Rang et al., 2011), although the tracking of cells by microscopy requires the cells to be a monolayer in both methods. Population growth rates during microscopy are comparable to those in shaking liquid cultures (Proenca et al., 2018; Proenca et al., 2019; Łapińska et al., 2019; Lindner et al., 2008; Wang et al., 2010). The causal trigger for the asymmetry between the daughters remains unknown, although damaged proteins (Lindner et al., 2008), expressed gene products (GFP), and now ribosomes (Figure 5; see below) have been shown to correlate with the differences in elongation rate. Damage may have the strongest causal link because the addition of external damage agents such as heat, phototoxicity, and antibiotics increases differentially the death probability of old daughters (Proenca et al., 2018; Proenca et al., 2019). Because new daughters are born with less damage, they are not killed by the damage agents and allow for the survival of the population (Chao, 2010; Proenca et al., 2019). The elongation rate differences have an epigenetic effect because slow mothers produce slower daughter pairs, although the new daughter is still quicker within the pair (Chao, 2010; Proenca et al., 2018; Proenca et al., 2019; Rang et al., 2011). If the causal factor were damaged proteins, the epigenetic interpretation is that the slower mother has more damage than average, and therefore the old and new daughters receive also more than their respective averages but the asymmetry is preserved because the former still gets more than the latter.

Because old daughters are more vulnerable to cell death than new daughters (see above), the elongation rate asymmetry can be considered a form of biological aging (Stewart et al., 2005). Thus, E. coli is a model for the evolutionary origin of biological aging from single-celled organisms. We would extend the argument to suggest that the old daughters are more properly considered to be the continuation of mother, and the new daughters are the true daughters. Given an epigenetic factor that slows the elongation rate of a cell, the mother retains a larger pool during division and it ages. The daughter is allocated less and it is rejuvenated. The larger pool that the mother does not transmit to the daughter is the original soma. Aging is the price that a mother pays to produce healthy daughters. Because the elongation rate asymmetry between mothers and daughters maximizes the fitness of the lineage (see ‘Introduction’), aging is indirectly favored by natural selection (Chao, 2010; Chao et al., 2016). Because the origin of life required organization of genetic material into protocells (Chao, 1991) in hazardous damage-inducing conditions (Miller and Urey, 1959; Michaelian, 2011), biological aging may have evolved in the first life forms.

The higher vulnerability of old daughters (or old mothers) in the presence of added damage agents was quantified by estimating the mean a and the variance s² of the regression (or mapping) coefficient of the doubling time (DT) of daughters onto mothers (Proenca et al., 2018). The vulnerability is high if (a² + s²) < 1 because the daughter could then acquire a very long DT and fail to divide. For example, if a = 1.1, a mother with DT = 40 min would have a daughter with a longer DT = 1.1 * 40 min, and the DT of each subsequent daughter in the lineage is even longer. If s² is large, then by chance the DT of a daughter could also be vulnerably long. When damage agents were added, the old daughters were the first ones to die because their values of a and s² increased above the threshold (Proenca et al., 2019). Thus, the asymmetry between old and new daughters is extended to include the values of a and s².

A question that is often raised is why evolution by natural selection is not able to minimize a and s² to diminish the window of vulnerability. An answer is that it does, but the damage agents are pathological conditions beyond the physiological limits. We suggest an additional answer for s² based on our variance measurements for ribosome density and elongation rate (Table 1). From those measurements, we estimated the stochastic and deterministic components of the variances. The deterministic component, which resulted from the old and new daughter asymmetry, was D²/4. The stochastic component is therefore V_E = V_T – D²/4 = V_T (1 – h²). V_E is comparable to s² because both describe phenotypic variation in a clonal population and are generated by stochastic noise. From our measured values (Table 1C), V_E for ribosome density was 0.0038 and 0.0034 in old and new mothers and not significantly different (p=0.40). For elongation rate, V_E was 0.074 and 0.051, significantly higher in old mothers (p=.007), and comparable with the trend reported for s² (Proenca et al., 2019). A cross-comparison of V_E for ribosome and elongation showed that the latter was significantly higher both in old mothers (0.00381 vs. 0.00744; p=1 × 10^–6) and in new mothers (0.00338 vs. 0.00510; p=0.0041). This inflation of the elongation V_E is consistent with the hypothesis that downstream traits are inherently noisier because they are at the end of many converging pathways or networks, each one with its own inherent noise (Price and Schluter, 1991; Houle, 1992). Because ribosomes are farther upstream, they have fewer inputs and a lower final V_E. Thus, another answer is that although complex traits such as elongation are important for a cell and under strong natural selection, they are most likely downstream and their V_E are noisier and higher in value.

Previous studies have shown that old and new daughters in E. coli are asymmetrical for elongation rates (Stewart et al., 2005; Proenca et al., 2018; Proenca et al., 2019; Łapińska et al., 2019; Lindner et al., 2008; Chao et al., 2016; Rang et al., 2011), gene expression (Shi et al., 2020), and vulnerability to dying in the presence of damage agents (Proenca et al., 2019). We show here that the variances (V_E) of ribosome density and elongation rates are also asymmetric. If asymmetry is a central characteristic of bacterial aging, identifying its cause becomes an important goal. We began our studies by examining asymmetry for elongation rates, and then to gene products and now to ribosomes. Because ribosomes are positioned highly upstream in the pathway of these cells and their abundance correlates strongly with elongation rates in E. coli (Schaechter et al., 1958; Neidhardt and Magasanik, 1960; Kjeldgaard and Kurland, 1963; Poulsen et al., 1995), their physiological role is both fundamental and general. As we have argued, our previous report on the asymmetry of gene products was instructive but left uncertain whether the outcome was unique to the gene that was studied or a more general phenomenon. Our present results on ribosome asymmetry move the process to a more fundamental level because ribosomes can affect more genes and more downstream processes. More importantly, existing measurements of ribosome cytology and cell growth in E. coli, in conjunction with our new results, allow us to quantitatively evaluate, understandingly not to prove, our previous suggestion (Shi et al., 2020) that space competition between damage aggregates and ribosomes could be responsible for the asymmetry.

The nucleoid and ribosomes in a rod-shaped E. coli occupy non-overlapping space along the length of the cell (Bakshi et al., 2012). At birth, the cells are smaller and the nucleiod resides at the mid-region of the rod and the ribosomes are displaced to the poles. As the cell elongates, the nucleoid splits to create the two nucleoids for the daughters. The splitting opens space in mid-region and the void is then filled by the ribosomes. Cell division ensues with a cleavage at the mid-region, and the ribosomes at the mid-region are distributed to the new poles of the daughters. However, the data used to generate this description did not distinguish between old and new poles or daughters. As a result, the cells were oriented randomly, and the resulting profiles are generic averages and necessarily symmetrical. When we distinguished between the old and new poles and daughters, the ribosome distribution resembles the generic profile only in new mothers and is asymmetrical in old mothers (Figure 6). Because damage aggregates also tend to reside at the poles (Lindner et al., 2008), we need to evaluate whether the volume of added aggregates displaced a volume of ribosomes that could quantitatively account for the elongation rate difference between old and new daughters from old mothers (Figure 2C). Such an analysis is possible using published information.

Single-cell measurements estimate elongation rates for E. coli to be 2.148 and 1.964 hr^–1 for new and old daughters in rich media (Proenca et al., 2018). The well-known linear relationship between elongation rate and ribosomal content (Schaechter et al., 1958; Neidhardt and Magasanik, 1960; Kjeldgaard and Kurland, 1963; Greulich et al., 2015) can be quantified from the data (Poulsen et al., 1995) as Y = 337.046 ∙ X+87.167, where Y is ribosomal content (A.U.) and X is elongation rate (hr^–1). The relationship between ribosome density and elongation rate in Figure 4 could not be used because the values were normalized. Letting X_New = 2.148 and X_Old = 1.964 hr^–1 from above, Y_New = 811.02 and Y_Old = 749.08 A.U. The proportional reduction of ribosomes in the old daughter relative to the new daughter is therefore p = (811.02–749.08)/811.02 = 0.0764. The total volume occupied by ribosomes in rich media is estimated to be V_ribo = 0.79 μm³ (Woldringh and Nanninga, 1985). The volume needed to reduce the elongate rate of the old daughter to 1.964 hr^–1 becomes ∆ = p∙V_ribo = 0.0764 ∙ 0.79 = 0.0603 μm³. Aggregate radii have been estimated from diffusion constants for E. coli cells in rich media (Coquel et al., 2013). By converting the radii to a sphere as (4π/3) ∙ radius³, the largest aggregates had mean values of 0.0852 μm³, which is well in the range of our estimate of ∆ = 0.0603 μm³. Thus, space competition between aggregates and ribosomes could quantitatively account for the lower elongation rate of old daughters of old mothers.

Our above space competition model implies that elongation rates and the size of damage aggregates should be negatively correlated. However, while the initial study by Lindner et al., 2008 found a strong and significant negative correlation (p<0.0001), a later study by Govers et al., 2018 found one that was negative but nonsignificant (p=0.08). Lindner et al. and Govers et al., respectively, used different stressors to induce damage (streptomycin and heat shock) and fluorescent reporters (YFP and msfGFP) to quantify aggregate size. The YFP and msfGFP used in both studies were fused to the chaperone IbpA, which associates to aggregates of damaged proteins. Because Govers et al. believed that IbpA-YFP formed self-aggregates they suggested that the correlation of Lindner et al. was an artifact of YFP self-aggregation. However, the rejection of Lindner et al. may be premature. First, the suggestion of YFP self-aggregation by Govers et al., 2018 was based on their demonstration that IbpA-YFP increased the expression of IbpA (at another locus), which is induced by perturbations to protein homeostasis. The expression was not increased in the IbpA-msfGFP control. Because the assay was conducted in the absence of heat shock or another stressor, they surmised that it was YFP self-aggregates that had perturbed the homeostasis. However, a recent report that IbpA self-represses its own translation because over-expression is harmful in the absence of stress (Miwa et al., 2021) suggests an alternative explanation for why IbpA-YFP increased expression. It could be that IbpA-YFP was unable to self-repress because of its larger size. Second, msfGFP-YFP may have its own issues because super-folding GFPs are more prone to photobleaching (Valbuena et al., 2020; Pédelacq et al., 2006). Could photobleaching have weakened the correlation of Govers et al. and rendered it nonsignificant? Third, most importantly, the correlations of Lindner et al. and Govers et al. must be compared with caution. While Lindner et al. induced cell damage with 10 µg ml^–1 of streptomycin, where the minimal inhibitory concentration is 16–32 µg ml^–1 (Tashiro et al., 2017; Regoes et al., 2004), Govers et al. used 47°C heat shock for 15 min, where 50°C at 15 min is lethal (Govers et al., 2018). Although both Lindner et al. and Govers et al. grew their cells in LB broth, the elongation rate in the two studies, respectively, centered about 0.35 hr^–1 and 0.25 hr^–1 (19 and 27 min doubling times). To have a doubling time of 27 min in LB, the cells in Govers et al. must have experienced severe stress and damage. In relative terms, the two treatments, respectively, were 10/16 to 10/32 = 63% to 31% vs. 47/50 = 94% of stopping growth. It is likely that at 94% lethality the stress level in Govers et al. is collapsing most cellular processes, including protein aggregation. This possibility is supported by Govers et al.’s measurement of the mean IbpA-msfGFP fluorescence per cell as a function of increasing heat shock temperature. The mean response increased linearly from 42°C up to 48°C, and then from 48°C to 50°C it crashed abruptly back to the control level (shock-free) (Govers et al., 2018). If the accuracy of laboratory incubators is ± 1°C, Govers et al. were using IbpA-msfGFP dangerously near the failure point of the reporter, which could have underestimated (or estimated with more error) the true amount of protein damage in the cells. If a 47°C heat shock induced damage to other cellular components besides proteins (e.g., DNA), the IbpA-msfGFP would not have reported them. High heat can damage DNA and elevate mutation rates in E. coli (Zamenhof and Greer, 1958; Chu et al., 2018). The choice of 47°C was justified because Govers et al. wanted to test whether protein aggregates could protect cells from new damage. However, if the underestimates or statistical analyses weakened by measurement errors compromised the correlation of Govers et al., the two correlations are not equivalent and cannot be used to reject each other.

Until new studies prove otherwise, we believe that the elongation rate and aggregate correlations of Lindner et al. and Govers et al. should be viewed as alternative models to be further tested. Although the correlations do not bear directly on our ribosome results, they are relevant to our discussion of the relationship of ribosomes, damage aggregates, and elongation rates. Our space competition model (see above) was evaluated to test whether our data was (or was not) quantitatively consistent with the correlation reported by Lindner et al. We look forward to other tests or challenges as connecting the two correlations will be critical. The protection uncovered by Govers et al. is an exciting new phenomenon. However, we are most excited by the possibility that both correlations are correct and protein aggregates are harmful and protective under low and high stress, respectively. At low stress levels, proteins could be the main target of damage and the amount of protein aggregates correlates with elongation rate. The damaged proteins are aggregated and then partitioned asymmetrically between the daughters. Protein repair is occurring (Ezraty et al., 2017; Merdanovic et al., 2011) and its effect is to lower the damage rate experienced by the cell. If aggregates are observed, damage rates must be exceeding repair rates. However, a full stress response (e.g., SOS, efflux transporters, and RpoS) (Roemhild et al., 2022) is too costly and unnecessary. At high stress levels and near lethality, proteins, DNA, and other cellular processes (including protein aggregation and IbpA-msfGFP reporting) are all failing due to damage and the correlation between elongation rate and protein aggregates is weakened. In order to survive, the cell needs a full stress response and aggregates become beneficial because they are present to act as a trigger (or the ‘epigenetic memory’ of Govers et al.). In effect, protein aggregates at high levels are inducing persister states (Leszczynska et al., 2013; Dewachter et al., 2021; Goode et al., 2021; Balaban et al., 2019).

In summary, previous reports have documented in E. coli the asymmetrical associations between bacterial aging, elongation rates, cell poles, damage aggregates, and expressed gene products, and we have added one more phenotype, the density of ribosomes. Because ribosomes have a fundamental role in determining cell function, our results indicate that bacterial aging and the asymmetry are resulting from processes that are central to cell function and not just a subset of genes or pathways. Moreover, ribosomal function has been well characterized and the information allowed us to show that space competition between ribosomes and damage aggregates, both of which tend to reside in the poles, could quantitatively account for the elongation rate asymmetry between old and new daughters. The latter calculation does not prove a causal role for damage aggregates, but we hope it will motivate and guide new research to evaluate hypotheses emanating from the work by us and others.

All data analyzed and presented in Figures are available in Dryad in https://doi.org/10.5061/dryad.59zw3r2jg.

1. 1. Chao L
   2. Chan CK
   3. Shi C
   4. Rang CU

   (2024) Dryad Digital Repository

   Spatial and temporal distribution of ribosomes in single cells reveals aging differences between old and new daughters of Escherichia coli.

   https://doi.org/10.5061/dryad.59zw3r2jg

1. 1. Bakshi S
   2. Siryaporn A
   3. Goulian M
   4. Weisshaar JC

   (2012) Superresolution imaging of ribosomes and RNA polymerase in live Escherichia coli cells

   Molecular Microbiology 85:21–38.

   https://doi.org/10.1111/j.1365-2958.2012.08081.x

   • PubMed
   • Google Scholar
2. 1. Balaban NQ
   2. Helaine S
   3. Lewis K
   4. Ackermann M
   5. Aldridge B
   6. Andersson DI
   7. Brynildsen MP
   8. Bumann D
   9. Camilli A
   10. Collins JJ
   11. Dehio C
   12. Fortune S
   13. Ghigo J-M
   14. Hardt W-D
   15. Harms A
   16. Heinemann M
   17. Hung DT
   18. Jenal U
   19. Levin BR
   20. Michiels J
   21. Storz G
   22. Tan M-W
   23. Tenson T
   24. Van Melderen L
   25. Zinkernagel A

   (2019) Definitions and guidelines for research on antibiotic persistence

   Nature Reviews. Microbiology 17:441–448.

   https://doi.org/10.1038/s41579-019-0196-3

   • PubMed
   • Google Scholar
3. 1. Batchelor E
   2. Goulian M

   (2006) Imaging OmpR localization in Escherichia coli

   Molecular Microbiology 59:1767–1778.

   https://doi.org/10.1111/j.1365-2958.2006.05048.x

   • PubMed
   • Google Scholar
4. 1. Biggs DSC
   2. Andrews M

   (1997) Acceleration of iterative image restoration algorithms

   Applied Optics 36:1766–1775.

   https://doi.org/10.1364/ao.36.001766

   • PubMed
   • Google Scholar
5. 1. Cabrera JE
   2. Cagliero C
   3. Quan S
   4. Squires CL
   5. Jin DJ

   (2009) Active transcription of rRNA operons condenses the nucleoid in Escherichia coli: examining the effect of transcription on nucleoid structure in the absence of transertion

   Journal of Bacteriology 191:4180–4185.

   https://doi.org/10.1128/JB.01707-08

   • PubMed
   • Google Scholar
6. 1. Chao L

   (1991) Levels of selection, evolution of sex in RNA viruses, and the origin of life

   Journal of Theoretical Biology 153:229–246.

   https://doi.org/10.1016/s0022-5193(05)80424-2

   • PubMed
   • Google Scholar
7. 1. Chao L

   (2010) A model for damage load and its implications for the evolution of bacterial aging

   PLOS Genetics 6:1001076.

   https://doi.org/10.1371/journal.pgen.1001076

   • PubMed
   • Google Scholar
8. 1. Chao L
   2. Rang CU
   3. Proenca AM
   4. Chao JU

   (2016) Asymmetrical damage partitioning in bacteria: a model for the evolution of stochasticity, determinism, and genetic assimilation

   PLOS Computational Biology 12:e1004700.

   https://doi.org/10.1371/journal.pcbi.1004700

   • PubMed
   • Google Scholar
9. 1. Chu XL
   2. Zhang BW
   3. Zhang QG
   4. Zhu BR
   5. Lin K
   6. Zhang DY

   (2018) Temperature responses of mutation rate and mutational spectrum in an Escherichia coli strain and the correlation with metabolic rate

   BMC Evolutionary Biology 18:126.

   https://doi.org/10.1186/s12862-018-1252-8

   • PubMed
   • Google Scholar
10. 1. Coquel ASS
    2. Jacob JPP
    3. Primet M
    4. Demarez A
    5. Dimiccoli M
    6. Julou T
    7. Moisan L
    8. Lindner AB
    9. Berry H

    (2013) Localization of protein aggregation in Escherichia coli is governed by diffusion and nucleoid macromolecular crowding effect

    PLOS Computational Biology 9:e1003038.

    https://doi.org/10.1371/journal.pcbi.1003038

    • Google Scholar
11. 1. Dewachter L
    2. Bollen C
    3. Wilmaerts D
    4. Louwagie E
    5. Herpels P
    6. Matthay P
    7. Khodaparast L
    8. Khodaparast L
    9. Rousseau F
    10. Schymkowitz J
    11. Michiels J

    (2021) The dynamic transition of persistence toward the viable but nonculturable state during stationary phase is driven by protein aggregation

    mBio 12:e0070321.

    https://doi.org/10.1128/mBio.00703-21

    • PubMed
    • Google Scholar
12. 1. Ezraty B
    2. Gennaris A
    3. Barras F
    4. Collet JF

    (2017) Oxidative stress, protein damage and repair in bacteria

    Nature Reviews. Microbiology 15:385–396.

    https://doi.org/10.1038/nrmicro.2017.26

    • PubMed
    • Google Scholar
13. 1. Gatti-Lafranconi P
    2. Natalello A
    3. Ami D
    4. Doglia SM
    5. Lotti M

    (2011) Concepts and tools to exploit the potential of bacterial inclusion bodies in protein science and biotechnology

    The FEBS Journal 278:2408–2418.

    https://doi.org/10.1111/j.1742-4658.2011.08163.x

    • PubMed
    • Google Scholar
14. 1. Goode O
    2. Smith A
    3. Zarkan A
    4. Cama J
    5. Invergo BM
    6. Belgami D
    7. Caño-Muñiz S
    8. Metz J
    9. O’Neill P
    10. Jeffries A
    11. Norville IH
    12. David J
    13. Summers D
    14. Pagliara S

    (2021) Persister Escherichia coli cells have a lower intracellular pH than susceptible cells but maintain their pH in response to antibiotic treatment

    mBio 12:e0090921.

    https://doi.org/10.1128/mBio.00909-21

    • PubMed
    • Google Scholar
15. 1. Govers SK
    2. Mortier J
    3. Adam A
    4. Aertsen A

    (2018) Protein aggregates encode epigenetic memory of stressful encounters in individual Escherichia coli cells

    PLOS Biology 16:e2003853.

    https://doi.org/10.1371/journal.pbio.2003853

    • PubMed
    • Google Scholar
16. 1. Greulich P
    2. Scott M
    3. Evans MR
    4. Allen RJ

    (2015) Growth-dependent bacterial susceptibility to ribosome-targeting antibiotics

    Molecular Systems Biology 11:796.

    https://doi.org/10.15252/msb.20145949

    • PubMed
    • Google Scholar
17. 1. Houle D

    (1992) Comparing evolvability and variability of quantitative traits

    Genetics 130:195–204.

    https://doi.org/10.1093/genetics/130.1.195

    • PubMed
    • Google Scholar
18. 1. Kjeldgaard NO
    2. Kurland CG

    (1963) The distribution of soluble and ribosomal RNA as a function of growth rate

    Journal of Molecular Biology 6:341–348.

    https://doi.org/10.1016/S0022-2836(63)80093-5

    • Google Scholar
19. 1. Landgraf D
    2. Okumus B
    3. Chien P
    4. Baker TA
    5. Paulsson J

    (2012) Segregation of molecules at cell division reveals native protein localization

    Nature Methods 9:480–482.

    https://doi.org/10.1038/nmeth.1955

    • PubMed
    • Google Scholar
20. 1. Łapińska U
    2. Glover G
    3. Capilla-Lasheras P
    4. Young AJ
    5. Pagliara S

    (2019) Bacterial ageing in the absence of external stressors

    Philosophical Transactions of the Royal Society B 374:20180442.

    https://doi.org/10.1098/rstb.2018.0442

    • Google Scholar
21. 1. Leszczynska D
    2. Matuszewska E
    3. Kuczynska-Wisnik D
    4. Furmanek-Blaszk B
    5. Laskowska E

    (2013) The formation of persister cells in stationary-phase cultures of Escherichia coli is associated with the aggregation of endogenous proteins

    PLOS ONE 8:e54737.

    https://doi.org/10.1371/journal.pone.0054737

    • PubMed
    • Google Scholar
22. 1. Lindner AB
    2. Madden R
    3. Demarez A
    4. Stewart EJ
    5. Taddei F

    (2008) Asymmetric segregation of protein aggregates is associated with cellular aging and rejuvenation

    PNAS 105:3076–3081.

    https://doi.org/10.1073/pnas.0708931105

    • Google Scholar
23. 1. Lucy LB

    (1974) An iterative technique for the rectification of observed distributions

    The Astronomical Journal 79:745.

    https://doi.org/10.1086/111605

    • Google Scholar
24. 1. Maisonneuve E
    2. Ezraty B
    3. Dukan S

    (2008) Protein aggregates: an aging factor involved in cell death

    Journal of Bacteriology 190:6070–6075.

    https://doi.org/10.1128/JB.00736-08

    • PubMed
    • Google Scholar
25. 1. Merdanovic M
    2. Clausen T
    3. Kaiser M
    4. Huber R
    5. Ehrmann M

    (2011) Protein quality control in the bacterial periplasm

    Annual Review of Microbiology 65:149–168.

    https://doi.org/10.1146/annurev-micro-090110-102925

    • PubMed
    • Google Scholar
26. 1. Michaelian K

    (2011) Thermodynamic dissipation theory for the origin of life

    Earth System Dynamics 2:37–51.

    https://doi.org/10.5194/esd-2-37-2011

    • Google Scholar
27. 1. Miller SL
    2. Urey HC

    (1959) Organic compound synthesis on the primitive earth

    Science 130:245–251.

    https://doi.org/10.1126/science.130.3370.245

    • PubMed
    • Google Scholar
28. 1. Miwa T
    2. Chadani Y
    3. Taguchi H

    (2021) Escherichia coli small heat shock protein IbpA is an aggregation-sensor that self-regulates its own expression at posttranscriptional levels

    Molecular Microbiology 115:142–156.

    https://doi.org/10.1111/mmi.14606

    • PubMed
    • Google Scholar
29. 1. Mosleh A
    2. Langlois JMP
    3. Green P

    (2014) Image deconvolution ringing artifact detection and removal via PSF frequency analysisLecture

    Notes in Computer Science 01:10593-2_17.

    https://doi.org/10.1007/978-3-319-10593-2_17

    • Google Scholar
30. 1. Neidhardt FC
    2. Magasanik B

    (1960) Studies on the role of ribonucleic acid in the growth of bacteria

    Biochimica et Biophysica Acta 42:99–116.

    https://doi.org/10.1016/0006-3002(60)90757-5

    • PubMed
    • Google Scholar
31. 1. Pédelacq JD
    2. Cabantous S
    3. Tran T
    4. Terwilliger TC
    5. Waldo GS

    (2006) Engineering and characterization of a superfolder green fluorescent protein

    Nature Biotechnology 24:79–88.

    https://doi.org/10.1038/nbt1172

    • PubMed
    • Google Scholar
32. 1. Pogliano J
    2. Ho TQ
    3. Zhong Z
    4. Helinski DR

    (2001) Multicopy plasmids are clustered and localized in Escherichia coli

    PNAS 98:4486–4491.

    https://doi.org/10.1073/pnas.081075798

    • Google Scholar
33. 1. Poulsen LK
    2. Licht TR
    3. Rang C
    4. Krogfelt KA
    5. Molin S

    (1995) Physiological state of Escherichia coli BJ4 growing in the large intestines of streptomycin-treated mice

    Journal of Bacteriology 177:5840–5845.

    https://doi.org/10.1128/jb.177.20.5840-5845.1995

    • PubMed
    • Google Scholar
34. 1. Price T
    2. Schluter D

    (1991) On the low heritability of life-history traits

    Evolution 45:853–861.

    https://doi.org/10.1111/j.1558-5646.1991.tb04354.x

    • Google Scholar
35. 1. Proenca AM
    2. Rang CU
    3. Buetz C
    4. Shi C
    5. Chao L

    (2018) Age structure landscapes emerge from the equilibrium between aging and rejuvenation in bacterial populations

    Nature Communications 9:3722.

    https://doi.org/10.1038/s41467-018-06154-9

    • PubMed
    • Google Scholar
36. 1. Proenca AM
    2. Rang CU
    3. Qiu A
    4. Shi C
    5. Chao L

    (2019) Cell aging preserves cellular immortality in the presence of lethal levels of damage

    PLOS Biology 17:e3000266.

    https://doi.org/10.1371/journal.pbio.3000266

    • Google Scholar
37. 1. Rang CU
    2. Peng AY
    3. Chao L

    (2011) Temporal dynamics of bacterial aging and rejuvenation

    Current Biology 21:1813–1816.

    https://doi.org/10.1016/j.cub.2011.09.018

    • PubMed
    • Google Scholar
38. 1. Rang CU
    2. Peng AY
    3. Poon AF
    4. Chao L

    (2012) Ageing in Escherichia coli requires damage by an extrinsic agent

    Microbiology 158:1553–1559.

    https://doi.org/10.1099/mic.0.057240-0

    • PubMed
    • Google Scholar
39. 1. Rang CU
    2. Proenca A
    3. Buetz C
    4. Shi C
    5. Chao L

    (2018) Minicells as a damage disposal mechanism in Escherichia coli

    mSphere 3:e00428-18.

    https://doi.org/10.1128/mSphere.00428-18

    • PubMed
    • Google Scholar
40. 1. Regoes RR
    2. Wiuff C
    3. Zappala RM
    4. Garner KN
    5. Baquero F
    6. Levin BR

    (2004) Pharmacodynamic functions: a multiparameter approach to the design of antibiotic treatment regimens

    Antimicrobial Agents and Chemotherapy 48:3670–3676.

    https://doi.org/10.1128/AAC.48.10.3670-3676.2004

    • PubMed
    • Google Scholar
41. 1. Roemhild R
    2. Bollenbach T
    3. Andersson DI

    (2022) The physiology and genetics of bacterial responses to antibiotic combinations

    Nature Reviews. Microbiology 20:478–490.

    https://doi.org/10.1038/s41579-022-00700-5

    • PubMed
    • Google Scholar
42. Book

    1. Sambrook J
    2. Fritsch EF
    3. Maniatis T

    (1989)

    Molecular cloning: a laboratory manual

    CSH Press.

    • Google Scholar
43. 1. Sanamrad A
    2. Persson F
    3. Lundius EG
    4. Fange D
    5. Gynnå AH
    6. Elf J

    (2014) Single-particle tracking reveals that free ribosomal subunits are not excluded from the Escherichia coli nucleoid

    PNAS 111:11413–11418.

    https://doi.org/10.1073/pnas.1411558111

    • PubMed
    • Google Scholar
44. 1. Schaechter M
    2. Maaloe O
    3. Kjeldgaard NO

    (1958) Dependency on medium and temperature of cell size and chemical composition during balanced grown of Salmonella typhimurium

    Journal of General Microbiology 19:592–606.

    https://doi.org/10.1099/00221287-19-3-592

    • PubMed
    • Google Scholar
45. 1. Schramm FD
    2. Schroeder K
    3. Jonas K

    (2020) Protein aggregation in bacteria

    FEMS Microbiology Reviews 44:54–72.

    https://doi.org/10.1093/femsre/fuz026

    • Google Scholar
46. 1. Shi C
    2. Chao L
    3. Proenca AM
    4. Qiu A
    5. Chao J
    6. Rang CU

    (2020) Allocation of gene products to daughter cells is determined by the age of the mother in single Escherichia coli cells

    Proceedings of the Royal Society B 287:20200569.

    https://doi.org/10.1098/rspb.2020.0569

    • Google Scholar
47. Book

    1. Sokal RR
    2. Rohlf FJ

    (1995)

    Biometry

    New York: WH Freeman and Company.

    • Google Scholar
48. 1. Stewart EJ
    2. Madden R
    3. Paul G
    4. Taddei F
    5. Burland V

    (2005) Aging and death in an organism that reproduces by morphologically symmetric division

    PLOS Biology 3:e45.

    https://doi.org/10.1371/journal.pbio.0030045

    • PubMed
    • Google Scholar
49. 1. Swulius MT
    2. Jensen GJ

    (2012) The helical MreB cytoskeleton in Escherichia coli MC1000/pLE7 is an artifact of the N-Terminal yellow fluorescent protein tag

    Journal of Bacteriology 194:6382–6386.

    https://doi.org/10.1128/JB.00505-12

    • PubMed
    • Google Scholar
50. 1. Tashiro Y
    2. Eida H
    3. Ishii S
    4. Futamata H
    5. Okabe S

    (2017) Generation of small colony variants in biofilms by Escherichia coli Harboring a Conjugative F Plasmid

    Microbes and Environments 32:40–46.

    https://doi.org/10.1264/jsme2.ME16121

    • PubMed
    • Google Scholar
51. 1. Valbuena FM
    2. Fitzgerald I
    3. Strack RL
    4. Andruska N
    5. Smith L
    6. Glick BS

    (2020) A photostable monomeric superfolder green fluorescent protein

    Traffic 21:534–544.

    https://doi.org/10.1111/tra.12737

    • PubMed
    • Google Scholar
52. 1. Vedel S
    2. Nunns H
    3. Košmrlj A
    4. Semsey S
    5. Trusina A

    (2016) Asymmetric damage segregation constitutes an emergent population-level stress response

    Cell Systems 3:187–198.

    https://doi.org/10.1016/j.cels.2016.06.008

    • PubMed
    • Google Scholar
53. 1. Wang P
    2. Robert L
    3. Pelletier J
    4. Dang WL
    5. Taddei F
    6. Wright A
    7. Jun S

    (2010) Robust growth of Escherichia coli

    Current Biology 20:1099–1103.

    https://doi.org/10.1016/j.cub.2010.04.045

    • PubMed
    • Google Scholar
54. 1. Winkler J
    2. Seybert A
    3. König L
    4. Pruggnaller S
    5. Haselmann U
    6. Sourjik V
    7. Weiss M
    8. Frangakis AS
    9. Mogk A
    10. Bukau B

    (2010) Quantitative and spatio-temporal features of protein aggregation in Escherichia coli and consequences on protein quality control and cellular ageing

    The EMBO Journal 29:910–923.

    https://doi.org/10.1038/emboj.2009.412

    • PubMed
    • Google Scholar
55. Book

    1. Woldringh CL
    2. Nanninga N

    (1985)

    Structure of nucleoid and cytoplasm in the intact cell

    In: Nanninga N, editors. Molecular cytology of Escherichia coli. London, New York: Academic Press. pp. 161–197.

    • Google Scholar
56. 1. Zamenhof S
    2. Greer S

    (1958) Heat as an agent producing high frequency of mutations and unstable genes in Escherichia coli

    Nature 182:611–613.

    https://doi.org/10.1038/182611a0

    • PubMed
    • Google Scholar

Author details

1. Lin Chao

   Department of Ecology, Behavior and Evolution, School of Biological Sciences, University of California San Diego, La Jolla, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   Contributed equally with

   Chun Kuen Chan

   For correspondence

   lchao@ucsd.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0009-0005-4114-5841

2. Chun Kuen Chan

   Department of Ecology, Behavior and Evolution, School of Biological Sciences, University of California San Diego, La Jolla, United States

   Contribution

   Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   Contributed equally with

   Lin Chao

   Competing interests

   No competing interests declared

3. Chao Shi

   Department of Ecology, Behavior and Evolution, School of Biological Sciences, University of California San Diego, La Jolla, United States

   Contribution

   Software, Formal analysis, Supervision, Validation, Investigation, Visualization, Methodology, Project administration, Writing – review and editing

   Competing interests

   No competing interests declared

4. Ulla Camilla Rang

   Department of Ecology, Behavior and Evolution, School of Biological Sciences, University of California San Diego, La Jolla, United States

   Contribution

   Conceptualization, Resources, Software, Supervision, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-3589-7327

• 665

  views

• 37

  downloads

• 3

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.