Abstract
To swim and navigate, motile bacteria synthesize a complex motility machinery involving flagella, motors, and a sensory system. A myriad of studies has elucidated the molecular processes involved, but less is known about the coordination of motility expression with cellular physiology: In Escherichia coli, motility genes are strongly upregulated in nutrient-poor conditions compared to nutrient-replete conditions; yet a quantitative link to cellular motility has not been developed. Here, we systematically investigate gene expression, swimming behavior, and cell growth across a broad spectrum of exponential growth condition. We establish that E. coli up-regulates the expression of motility genes at slow growth to compensate for reduction in cell size, such that the number of flagella per cell is maintained across conditions. The observed 4-5 flagella per cell is the minimum number needed to keep the majority of cells motile. This simple regulatory objective allows E. coli cells to remain motile across a broad range of growth conditions while keeping the biosynthetic and energetic demands to establish and drive the motility machinery at the minimum needed. Given the strong reduction in flagella synthesis resulting from cell size increases at fast growth, our findings also provide a novel physiological perspective on bacterial cell size control: A larger cell-size at fast growth is an efficient strategy to increase the allocation of cellular resources to the synthesis of those proteins required for fast growth, while maintaining processes such as motility which are only needed on a per-cell basis.
To thrive in different environments, bacteria must efficiently allocate their limited resources towards different cellular processes in accordance to what is most needed for their growth and survival (1). Flagella driven motility is one of the most distinct processes of bacterial life which provides cells with novel ways to respond to the conditions they encounter (2). The active movement towards more favorable conditions and away from detrimental ones has been studied in detail on the molecular level (3–5) and can give rise to strong fitness advantages (6–9). But flagella driven motility is also a resource demanding process. For growing E.coli cells, the synthesis of the motility proteins alone ties up a substantial portion of the protein synthesis resources (10, 11), and the assembly and rotation of flagella also demand energy (12–14). Accordingly, motility expression constitutes a burden on cell growth, such that cells with attenuated motility can grow up to 20 % faster and reach about 10 % higher biomass yields (15–17), a strong difference readily affecting the outcome of (laboratory) evolution (18–21). Given this burden, the expression of motility is expected to be highly controlled, coordinated with other cellular processes and demands.
Notably, the expression of motility genes varies strongly with the nutrient conditions cells encounter and more resources are allocated to motility expression in nutrient poor than in replete conditions (22–26). These observations have been taken as support for the idea that motility is a response expressed to search for alternative nutrient sources when local nutrient sources are depleted (22, 25, 26). However, swimming speeds observed during balanced growth do not vary much with the growth rate or the carbon source provided (9). Furthermore, bacterial population exhibit a chemotaxis-driven range expansion (6, 8, 27–29) with expansion speed which is markedly faster in nutrients providing faster growth (9). These latter observations suggest that motility is a phenotype broadly expressed by growing cells, rather than being merely being a foraging response to starvation. But then why are motility genes expressed more in poor growth conditions and how does their degree of expression quantitatively affect swimming? To resolve this puzzle, we systematically investigated the link between gene expression and swimming in different balanced growth conditions. We found that E. coli cells maintain motility by regulating gene expression in coordination with cell size; upregulation of motility genes at slower growth is a necessary compensation to adjust for growth-related changes in cell size such that the number of flagella per cell remains constant. This simple regulatory objective provides an example of how cells maintain function while keeping resource demands minimal. Our findings also provide a new perspective on the relation between cell size control and proteome resource allocation, giving a physiological rationale for the ubiquitously observed positive relation between cell growth and cell size.
To study the relation between swimming behavior and motility gene expression, we first examined gene expression during balanced growth across a broad range of growth conditions, using a physiologically well-characterized strain (WT strain E. coli K-12 HE204, SI Text 1.1). Motility genes are hierarchically regulated and have been assigned into three different classes with the master regulator flhDC being the class-I genes as illustrated in Fig. S1A (30–32). We first studied the expression of fliA, a class-II gene which is expressed coordinately with other class-II genes (including those encoding flagella hook-basal body components,Fig. S1BC) and encodes the sigma factor σF required for the expression of flagella components (class-III genes). Using a LacZ reporter, we quantified the expression level of the fliA promoter (in unit of LacZ activity per biomass; see SI Text 1.4) during balanced growth, with a range of growth rates obtained by supplementing minimal medium with different carbon sources or rich media components (detailed growth conditions described in SI Text 1.2). Consistent with previous reports (22, 24–26), fliA expression was higher at slower growth rates (Fig. 1A, circles): Expression levels change approximately exponentially with growth rate (dashed line), with a ~4.4 fold increase when growth rates change from fast (1.60 1/h for rich defined medium with glucose) to slow (0.28 1/h for aspartate). In contrast, a constitutively active promoter, reported by Ptet-lacZ expression, exhibits only a ~1.3 fold change (Fig. 1A, diamond). The coordinated expression of fliA and other motility genes provides a substantial growth-rate dependent burden for the cell: a deletion of the master regulator flhD, resulting in the complete suppression of motility gene expression, increased growth rate by up to 18% compared to the WT strain, with larger increases realized in slower growth conditions where motility expression in the WT is higher (Fig. S1D).
To understand the consequences of this costly expression on swimming, we next characterized the swimming behavior across growth conditions. Extending a previous approach combining phase contrast microscopy and tracking (9), we quantified the movement of hundreds of cells and analyzed the distributions of observed swimming speeds {vi}during run events (see Fig. 1B, Fig. S2 and SI Text 1.3 for methods). We then extracted the average swimming speed and the fraction of motile cells with swimming velocities vi > 5 μm/s. Notably, despite the ~4.4 fold change of gene expression (Fig. 1A), swimming characteristics varied only weakly: the fraction of swimming cells (αm) remained close to 90 % for all growth conditions (Fig. 1C), and the average swimming speed, , changed only ~1.3 fold from fast (rich defined medium with glucose) to slow growth (aspartate) (Fig. 1D).
One possible explanation for this combined observation of minor changes in motile behavior and the large changes in expression of motility genes would be an adjustment to a possible decrease in flagella motor activity at slow growth: The E. coli flagella motor is driven by the proton motive force (PMF) and the motor rotation frequency is proportional to the PMF (12, 13). Given that the PMF is a result of the metabolic state which might change with growth condition, the cell might compensate for slower rotation in poor growth conditions by increasing the expression level of motility genes. To probe this possibility, we measured the motor activity by tracking the rotation of beads attached to flagella filaments (33, 34). However, the rotation frequency is found to be almost independent of growth (Fig. 1E; a drop of 13 % from growth rate 0.87 1/h to 0.39 1/h).
Why then are motility genes expressed more in slow growth conditions? To investigate this question further, we next performed experiments with a synthetic construct which allows for the smooth titration of motility gene expression in a given growth condition, so that we can separately assess the effect of changing motility expression and growth. We replaced the native promoter of the master regulator flhDC by the Ptet promoter, enabling an inducer-dependent control. Additionally, the construct also carries the above-mentioned PfliA-lacZ as a reporter for a class-II gene expression (see Fig. 2A and SI Text 1.1.3 for cartoon and details). We first grew this strain in fructose minimal medium with different concentrations of the inducer chlortetracycline (cTc). PfliA-lacZ expression decreased smoothly from wild-type levels towards zero when reducing the inducer concentration in the media (Fig. 2A blue points). Decreasing the inducer concentration similarly shifted the distribution of swimming speeds (Fig. 2B). with falling average swimming speeds and motile fractions (Fig. 2CD, blue points). Similar results were obtained by growing cells in other carbon sources that provide faster and slower growth rates (Fig. 2, glucose and mannose as green and magenta points). Overall, these results show that motility gene expression has a strong influence on cellular swimming behaviors at each growth condition, as can also be seen by directly plotting swimming speed and motility fraction against fliA expression (Fig. S3).
To further study the role of growth-rate dependent regulation of motility genes on swimming behavior, we next compared how the swimming phenotypes change at given levels of motility gene expression (independent of the growth rate) using the titratable construct and selected inducer levels, as indicated by the dotted and dashed lines in Fig. 3A. Comparing the swimming behavior at these fixed expression levels, we found a gradual reduction of swimming speed as the growth rate slows down (Fig. 3BC). The decreases in swimming speed at fixed expression levels can be largely accounted for by reduced fraction of motile cells (Fig. 3D), while the average swimming speed of those motile cells did not change significantly (Fig. 3B, grey vertical lines). Together, these observations establish that the upregulation of motility genes at slower growth is necessary to keep the population motile but not to increase the swimming speed of the motile cells.
To better understand the regulation of motility genes and its connection to swimming behavior, we next considered the abundance of motility gene products per cell: Gene expression levels, as those determined via a LacZ reporter, are typically quantified per biomass (e.g., the commonly used “Miller Unit” (35) quantifies LacZ activity per OD, U/ ml/OD600, with OD600 having a constant relation with biomass across growth condition (36); see SI Text 1.4 & SI Text 2). Since biomass itself is proportional to cell volume due to the constancy of biomass density (37, 11), the measurements with the class-II gene reporter PfliA-lacZ reflect the concentration of class-II gene products (flagella hook & basal body; Fig. S1, Fig. 4A, top row). As previously discussed, this concentration is higher when cells grow slower (Fig. 1A). However, bacterial cells also have different cell sizes at different growth rates. In fact, the biomass per cell exhibits an approximate exponential dependence on the growth rate (Fig. 4B), known as the Schaechter-Maaloe-Kjeldgaard relation (38–40). Accordingly, the abundance of class-II gene products per cell is expected to exhibit less change with growth rate than what is observed for the concentration (Fig. 4A, bottom row). Confirming this idea, the PfliA-lacZ expression per cell (unit: U /cell), taken as the product of expression per biomass (unit: U / ml/OD600) and the biomass per cell (unit: ml · OD600/cell), is nearly independent of growth rate (Fig. 4C, filled red points). Remarkably, the exponential relations observed for cell size (Fig. 4B, dashed line) and the expression level per biomass (Fig. 4C, dashed black line) show similar absolute rates (1.18 h−1 and 1.17 h−1), leading to the abundance per cell being independent of growth rate (Fig. 4C, dotted red line).
The above analysis suggests that cells maintain their number of flagella across growth conditions and that the large change of gene expression with growth rate is necessary to keep this number constant as the cell size changes. To more directly confirm this idea, we counted the number of flagella filaments attached to the cells using a staining assay (Fig. S4 and SI Text 1.5). We confirmed that the average number of flagella filaments per WT cell remains within a narrow range across growth conditions (4-5, within the measurement error), see Fig. S4D. As an example, two cells of different sizes but similar flagella numbers are shown in Fig. 4D. Looking at the distribution of filament numbers across the population, we see that very few cells possess only one or zero filaments (Fig. S4B), consistent with a high fraction of motile cells (Fig. 1C). In contrast, the average number of filaments varied strongly for the titratable flhDC strain as the provided inducer concentration was varied (Fig. S5). Particularly, the fraction of cells with zero or one filament clearly increased at lower inducer concentrations (Fig. S5AB) which coincides with the increase in the fraction of non-motile cells at lower inducer concentrations (Fig. 2D). We further confirmed that the class-II gene reporter expression reflects the change of filament number (Fig. 4E): reducing PfliA-lacZ level by titrating flhDC expression led to a linear drop of the average number of filaments in different growth conditions (Fig. 4E, open symbols). In contrast, the WT strain exhibited little variation in either the filament number or gene expression per cell (Fig. 4E, filled circles). In combination, our findings reveal that the regulated adjustment of motility gene expression in different growth conditions compensates for the changes in cell size seen in these conditions, such that a similar number of flagella is maintained for each cell across conditions.
To see how efficiently the motility genes are regulated, consider the relation between the number of flagella per cell, and the fraction of motile cells (Fig. 4F): When the gene expression is low such that there are < 4 flagella/cell (flhDC titration with low inducer levels, diamonds), the motile fraction is proportional to the flagella number (Fig. 4F, grey region: limited motility). In contrast, when expression levels reach close to those of WT such that there are > 4 flagella/cell, almost all cells are motile (Fig. 4F, circle points and yellow region: full motility). An even higher expression level per cell would only increase the costs to express extra flagella and is not observed (Fig. 4F, blue region: non-efficient expression). E. coli thus appears to regulate its motility expression levels such that the associated resource demands to synthesize and rotate flagella are at the minimum necessary to keep most cells motile. While the requirement for ~4 flagella per cell ensures most cells to be motile (yellow region in Fig. 4F), this number is also close to what is minimally required to allow uninterrupted motility when cells half the number of their flagella during cell division.
In this study, we analyzed the regulation of motility genes by E. coli for different balanced growth conditions and found that the fold-change in gene expression per biomass compensates for the variation in cell size, resulting in an approximately constant flagella number per cell. This simple regulatory scheme ensures a fully motile population while keeping resource demands to synthesize and rotate flagella to a minimum. The findings reported here have implications for bacterial motility from the ecological perspective, particularly concerning its role in promoting fitness across different environments. Previous works have highlighted the upregulation as a fingerprint of a specific starvation response with motility triggered when nutrients run out (22, 24–26). In contrast, we here propose that at least a part of the upregulation of swimming in poorer growth conditions is not a starvation response per se but an obligatory regulation to maintain flagella numbers and swimming in diverse growth-supporting conditions as cell size changes. This picture is in line with observations that bacterial cells quickly stop swimming (9), actively brake motor rotation (41, 42), and even release their flagella upon entering starvation (43, 44). Notably, the maintenance of cellular motility in growth-supporting conditions enables cell population to rapidly expand into unoccupied nutrient rich territories, boosting overall population growth (9). The growth advantage of such a navigated range expansion relies on cells being motile across conditions, and a delayed onset of motility only in response to starvation would nullify the fitness advantage (9). Therefore, the efficient regulation of motility genes described here does not only minimizes the resources required to build and fuel the motility machinery, but it also supports fast navigated range expansion which further boost fitness (9, 21, 29).
The findings further provide a new perspective on the relation between cell-size and growth itself. Throughout the text, we have referred to the change in motility gene expression as an up-regulation in poor nutrient conditions. But this change can also be viewed as a down-regulation in nutrient replete conditions when cells grow fast. Given that the goal of the flagella regulatory system is to maintain the number of flagella per cell, we can view the decreased flagella expression at fast growth also as a consequence of increased cell size at fast growth. This view leads us to suggest a physiological rationale for E. coli’s choice of cell size at different growth rates. It is generally preferrable for bacterial cells to keep a small biomass (i.e., cell size) as it promotes efficient diffusive transport, fast nutrient uptake, and strong dispersal (45, 46). However, in favorable conditions allowing for rapid growth, the translational machinery per biomass is the most growth-limiting factor (47, 48) and making cell-size larger can be beneficial to alleviate this bottleneck: By increasing its size at fast growth, the cell effectively reduces the amount of flagella proteins that need to be synthesized, thus allowing more proteomic resources to be allocated towards ribosomes and other components of the translation machinery. Quantitatively, flagella proteins comprise ~3.0 % of the total protein mass in slow carbon-limited conditions and ~0.7 % in rich-defined medium (11). Thus, by increasing its cell size, E. coli manages to “save” 2.3 % of the proteome that would have otherwise been tied up in flagella synthesis. To put this amount in perspective, the entire set of biosynthesis enzymes saved when cells are provided with all amino acids and nucleotides is only ~11 % of the proteome (comparing the proteome composition of cells grown in rich-defined medium supplemented with glucose to those grown in glucose minimal medium). This saving accounts for a large share of the increase of growth rate from 1.0 1/h in glucose minimal medium to 1.8 1/h in rich-defined medium (11), based on the well-established linear relation between the ribosome content and growth rate, where every percent-of-proteome added to the protein synthesis machinery results in an ~0.06 1/h increase in growth rate (11, 23, 47). Thus a 2.3 % saving in proteome allocation to flagella synthesis would amount to a gain of ~0.14 1/h for growth in rich medium. In other words, had E. coli kept its size at that in poor nutrient condition, then it would suffer a 0.14 1/h reduction in growth rate (from the observed growth rate of 1.8 1/h) in rich medium just due to motility expression alone. This proteome resource savings by a change of cell size should be similarly applicable to other cellular processes which demand protein expression on a per-cell basis, including cell division and cell pole maintenance. Therefore, increasing cell size at fast growth is a simple and effective strategy to reduce competition for proteome resources at fast growth, for E. coli and likely many other fast growing bacterial species.
Supplementary Figures
Acknowledgement
We thank Matteo Mori, Chenhao Wu and Christina Ludwig for providing proteomic data, and Angela Dawson and Ekaterina Krasnopeeva for providing the pTOF24 plasmids carrying S219C and sticky fliC. Tomoya Honda acknowledges the JASSO long-term graduate fellowship award. Leonardo Mancini and Teuta Pilizota acknowledge the support of Cunningham Trust award ACC/KWF/PhD1. Work in the Hwa lab is supported by the NIH through grant R01GM109069 and by the NSF through grant MCB 1818384.