Fig 1.
Growth laws and protein investment.
a, Overview of bacterial growth law model. Growth rate is determined by the balance of nutrient influx catalyzed by the P-sector and nutrient consumption by protein biosynthesis catalyzed by the R-sector. The Q-sector is a constant growth-rate-independent part of the proteome. Nutrient quality κn determines how much nutrient flux is achieved for a given P-sector fraction. For higher nutrient quality (κn large), a smaller P-sector achieves sufficient nutrient flux and frees up proteome for the R-sector for more biosynthesis, resulting in a higher growth. (panel created with Biorender.) b, Protein copy number of transporters (or first metabolic enzyme) of different substrates plotted against growth rate achieved on these respective substrates. Protein copy numbers were calculated for the same slow carbon-limited growth conditions by combining proteomics [6] and ribosome profiling data sets [12].
Fig 2.
a, Summary of the growth law model and parameters. ϕR is the ribosomal fraction of the proteome. ϕ0 is a constant, and κt is a parameter denoting the translational capacity. ϕP is called the P-sector, the proteome fraction that includes ‘catabolic’ enzymes catalyzing this nutrient flux. κn is the ‘nutritional capacity’ or ‘nutrient quality’. The sum of proteome sectors cannot surpass a maximum fraction, denoted by . λmax resembles a maximum growth rate that is related to the growth-rate invariant fraction of the proteome ϕfixed via λmax = (1−ϕfixed)κt. b, Illustration how different nutrient quality results in different growth rates. For a low nutrient quality (left), a high expression level of the P-sector is required to achieve sufficient nutrient flux. This leaves fewer resources available for the ribosomal sector ϕR and overall results in slower growth. Conversely, for high nutrient quality κn (right), a higher nutrient flux is catalyzed by a smaller P-sector, freeing up proteomic resources for higher expression of the ribosomal sector ϕR and resulting in faster growth. c, Illustration of nutrient flux and growth rate as a function of cAMP-mediated C-sector expression (left). The C-sector is one of the major components of the P-sector in the growth theory, as illustrated in the pie chart (right). Another large part being the ppGpp-activated protein sector that we denote as the S-sector, where S stands for stress. Most transporters and substrate-specific metabolic genes are part of the C-sector and increasing the C-sector increases nutrient flux (dashed lines, left top panel). Higher nutrient flux leads to an increase in growth rate, but only up to an optimum level, at which flux for biosynthesis balances nutrient flux (left bottom panel). At even higher C-sector expression, growth rate drops because there are insufficient proteomic resources for the R-sector to process nutrient flux generated by the disproportionate C-sector. Nutrient quality κn is determined by how much nutrient flux is achieved per C-sector induction. A “good” substrate results in a steep increase in nutrient flux (blue dashed line, left top panel), whereas a “poor” substrate results in a much flatter induction of nutrient flux (orange dashed line, left bottom panel). Therefore, “poor” substrates result in higher cAMP levels but slower growth rates than “good” substrates (left bottom panel). The steepness of nutrient flux induction, defining nutrient quality, is determined by the catalytic rates of substrate-specific enzymes, but also by the expression level of substrate-specific transporters and enzymes. We denote the core proteome fraction of substrate-specific transporters and enzymes by C*-sector, which is a fraction f in the much larger P-sector, f = ϕC*/ϕP. We then denote the core nutrient quality based on fundamental biochemical enzymatic properties by
. The effective nutrient quality that emerges in the growth laws can be modulated by changing the expression fraction f,
. We hypothesize that because the core enzyme proteome fraction ϕC* is a small fraction of the P-sector ϕP, even for the costliest substrates in terms of protein cost, nutrient quality can be dialed up or down in response to ecological needs by changing the expression fraction f. We denote the P-sector fraction that is not part of substrate-specific metabolism as the “adaptability” sector ϕAD: ϕAD = ϕP−ϕC*. Components of this sector are not important for supporting growth in the current growth conditions, but instead constitute a preparatory response. d, Illustration how the core catabolic fraction determines nutrient quality and growth rate. Within the co-regulated C-sector, low expression and a weak induction of substrate-specific enzymes will result in a lower effective nutrient quality and a higher expression level of the adaptability sector ϕAD (left). Conversely, a large core catabolic fraction results in fast growth and relatively low expression of the adaptability sector ϕAD (right). By dialing the core catabolic fraction, bacteria can convert information about their environment conveyed by the nutrient present, into resource allocation decisions determining their adaptability and preparedness for changing environments or the onset of stress. (illustrations created with Biorender).
Fig 3.
Plasticity of nutrient quality.
a, We replaced chromosomal promoters of transporters and metabolic enzymes required for mannose metabolism with the promoter of glucose transporter (P-ptsG). To prevent repression due to lack of glucose, we knocked out the glucose-specific transcriptional regulator Mlc. Finally, to ensure that carbon flux from processed mannose enters the glycolysis pathway, we placed mannose-6-phosphate isomerase under a strong constitutive promoter (P-tet). We refer to this strain as the swapped promoter strain (YCE119). (construct map illustration created with Biorender.) b, Growth rates of wildtype and swapped promoter strain (YCE119) on glucose and mannose. Mannose is one of the slowest substrates for the wildtype (orange circles), and glucose is often considered the best substrate of E. coli with the fastest growth rates in minimal medium (blue circles). Yet the swapped promoter strain grows on mannose as fast as the wildtype on glucose (orange triangles). Therefore, the genetic modifications in the swapped promoter strain have changed the nutrient quality of mannose into the nutrient quality of glucose. This means that nutrient quality is not limited by fundamental biochemical constraints. Unpaired t-test was performed, and following P-values were obtained. WT Glucose:WT mannose P-value<0.0001. WT Mannose:YCE 119 Mannose P-value<0.0001, and WT Glucose:YCE119 Mannose: non-significant (ns) c, Data from Towbin et al. [17], who titrated cAMP levels on different substrates. “Good” substrates have a higher peak growth rate at lower levels of cAMP. d, An inverse correlation of endogenous Crp activity on different substrates with growth rate on the respective substrate from Towbin et al. [17]. Poor substrates require higher cAMP levels for maximum growth. A similar relationship is found when plotting CRP activity vs. growth rate for maximum growth rate (S3 Fig).
Fig 4.
Lag and starvation cost of higher nutrient quality.
a, Illustration of resource allocation for high nutrient quality (left) and low nutrient quality (right). Low nutrient quality reflects high expression levels of the adaptability sector ϕAd. (pie chart illustrations created with Biorender.) b, Relative quantification of proteins that have been shown to be related to lag phase in Basan et al. [22]. Proteomics data from Hui et al [6]. (color bar illustration created with Biorender.) c, Relative quantification of proteins that have been shown to be related to improved starvation survival by Schink et al. [23]. Proteomics data from Hui et al [6]. d, Diauxic shifts from glucose or mannose to acetate. The wildtype strain exhibits a lag phase of several hours from glucose to acetate (blue circles), but almost no lag phase from mannose to acetate (orange circles). The swapped promoter strain (YCE 119) grows at a similar growth rate to glucose on mannose (Fig 3B) but has lost the ability to quickly switch to acetate. Upregulation of the C-sector by addition of cAMP (3.5mM) to the growth medium results in a much shorter lag time, however, only at the cost of slower growth (S4 Fig), as predicted by the model (Fig 2). Three biological replicates were used for all conditions. Mean values of 3 biological replicates is plotted, and error bars represent standard deviation. e, Starvation survival after 7 days of starvation relative to initial CFUs. The wildtype survived carbon starvation much better after growing on mannose (bottom bar) than on glucose (top bar). The faster-growing swapped promoter strain (YCE 119) lost this improved survival on mannose (2nd bar from top). With addition of cAMP (3.5mM) to the mannose growth medium but not to the starvation medium, the swapped promoter strain exhibited improved starvation survival (3rd bard from top) at the cost of a slower growth rate (S4 Fig). Three biological replicates were used for YCE119 grown in N+C+Mannose, WT grown in N+C+Glucose, and 4 biological replicated used for YCE 119 grown in N+C+Mannose with 3.5mM cAMP and WT grown in N+C+Mannose. Every biological replicate was plated in triplicate. Mean survival frequency plotted, and error bars represent standard deviation. Unpaired t-test performed, and following P-values obtained: WT Glucose-YCE119 Mannose: non-significant, WT Glucose-YCE119 Mannose+3.5mM cAMP:0.0054, WT Glucose-WT mannose<0.0001.
Fig 5.
Motility cost of higher nutrient quality.
a, Motility assays on soft agar plates. Colonies on fast growth medium are smaller and show smooth edges (top panel). Colonies on slow growth medium are larger with ruffled edges (middle panel). A fraction of colonies on slow growth medium shows a longer-range swarming phenotype (bottom panel). (pie chart illustrations created with Biorender.) b, Relative quantification of proteins related to motility. Proteomics data from Hui et al. [6]. (color bar illustration created with Biorender.) c, Fraction of colonies exhibiting swarming on soft agar. For wildtype on glucose (blue circle), we observed no swarming, whereas on mannose a substantial number of colonies showed swarming (orange circles). The swapped promoter strain (YCE 119) lost the ability to swarm on mannose (orange triangles), but addition of cAMP (3.5mM) resulted in a large fraction of swarming colonies. 24 colonies were analyzed for WT grown in Glucose, 72 colonies were analyzed for WT grown in Mannose, 24 colonies were analyzed for YCE 119 grown in Glucose, 48 colonies were analyzed for YCE119 grown in mannose, and 24 colonies were analyzed for YSE 119 grown in Mannose with 3.5mM cAMP. d, We tracked growth of colony size over 3 days on minimal medium soft agar plates. Colony sizes of the wildtype increased much more on mannose (orange circles) than on glucose (blue circles) and exhibited rough irregular edges indicative of higher motility. The swapped promoter strain lost this enhanced motility on mannose and only showed an increase in colony size similar to the wildtype on glucose (yellow triangles). However, with the addition of cAMP (3.5mM) to mannose medium, the swapped promoter recovered high motility (yellow inverted triangles).24 colonies were analyzed for WT grown in Glucose, 72 colonies were analyzed for WT grown in Mannose, 24 colonies were analyzed for YCE 119 grown in Glucose, 48 colonies were analyzed for YCE119 grown in Mannose, and 24 colonies were analyzed for YSE 119 grown in Mannose with 3.5mM cAMP. Normalized mean area fold change plotted with standard deviation).
Fig 6.
Nutrient quality as a map of environments and mediator of resource allocation strategies.
Ecological niches in which bacteria like E. coli evolved consist of many strikingly different environments, each coming with their own set of payoffs and hazards. The trajectory in which the bacterium transitions between these environments is partly governed by its natural life cycle, but also has a probabilistic component. To maximize their fitness, bacteria need to identify the environment in which they find themselves and adjust proteome resource allocation strategies accordingly. Nutrients allow bacteria to grow, but they also serve as a major signal that allows microbes to infer information about their environment. Nutrient quality encoded in regulatory architecture and enzymatic properties that were shaped by evolution, serves both as a map of the safety and reliability of the environment and as a regulatory mechanism implementing proteome allocation decisions. The conservation of this simple regulatory architecture across environments is what gives rise to the striking bacterial growth laws and their elegant predictions. (Illustration created with Biorender).