When E. coli is introduced into fresh growth medium, its population follows a remarkably predictable trajectory: a period of adjustment, a burst of rapid multiplication, a leveling off, and finally a decline. These four phases, known as lag, exponential (or log), stationary, and death, form the classic bacterial growth curve that underpins much of microbiology. The pattern looks simple when sketched on a whiteboard, but the biology driving each transition is anything but. What the cells are actually doing at each stage, and why they switch from one to the next, involves a sophisticated interplay of metabolic reprogramming, stress signaling, and even something resembling programmed cell death.
The Lag Phase
Drop E. coli into a flask of fresh medium and nothing seems to happen for a while. The cells are alive, but they are not dividing. This quiet period is the lag phase, and it exists because the bacteria need to retool their internal machinery to match the new environment. If the cells were previously growing on one carbon source and now face a different one, the enzymes required to process the new nutrient must be built from scratch, while enzymes for the old nutrient are dismantled or diluted out. Metabolomic studies have shown that long lag phases result from the depletion of key internal metabolites that occurs when central carbon metabolism has to suddenly reverse direction, as happens when cells shift between sugars that feed into opposing metabolic pathways.
The length of the lag phase is not fixed. It depends on how different the new environment is from the old one, and on the physiological state of the cells when they arrive. Cells transferred from one glucose-rich medium to another glucose-rich medium barely pause. Cells moved from a nutrient-poor environment to a rich one need time to ramp up the protein-synthesis machinery. One key regulator of this adjustment is the stringent response, which involves a signaling molecule called (p)ppGpp. When amino acids become scarce, rising levels of (p)ppGpp trigger the cell to shift its resources away from making ribosomes and toward making amino acid biosynthesis enzymes. Strains that lack the gene responsible for producing this signal (called relA) take significantly longer to adapt to nutrient downshifts, because they cannot efficiently reallocate their protein-making capacity.
This trade-off between growth speed and adaptability turns out to be universal. Cells that grow fastest in a stable environment tend to have the longest lag when conditions change, because their machinery is heavily optimized for one particular set of nutrients. Cells that grow a bit slower maintain more metabolic flexibility and can pivot faster. The pattern holds not just for E. coli but across many bacterial species, and it can be predicted from the opposing enzyme requirements of the pathways involved.
The Exponential Phase
Once the cells have adjusted, they enter the exponential phase, which is the period microbiologists care about most. Here, every cell in the population divides at a roughly constant rate, so the total number doubles at regular intervals. Under ideal lab conditions with rich medium at 37 °C, E. coli can double about every 20 minutes, though the actual doubling time varies with temperature, nutrient quality, and strain.
What makes exponential growth possible is a tight coordination between how fast the cell makes proteins and how fast it needs to grow. E. coli achieves this by scaling its ribosome content to match the growth rate: faster growth means a larger fraction of the cell’s total protein is devoted to ribosomes. This is not accidental. Research on ribosomal protein fractions has shown that the increase in ribosome concentration with growth rate precisely matches the increased protein synthesis demand, lending support to the idea that E. coli is tuned to maximize its growth rate in any environment that supports growth.
Exponential growth cannot last forever, of course. The cells are consuming nutrients and excreting waste products. During aerobic growth on glucose, E. coli excretes acetate through a process called overflow metabolism, even when oxygen is available to run more efficient metabolic pathways. This acetate accumulates in the medium and eventually starts to inhibit growth. At the same time, the preferred carbon source is being depleted. The combination of nutrient exhaustion and waste buildup is what tips the culture out of exponential growth and into the next phase.
The Stationary Phase
When nutrients run low, or when toxic byproducts accumulate to inhibitory levels, the population stops growing. The number of new cells being produced roughly equals the number dying, so the total count plateaus. This is the stationary phase, and it is far from a passive waiting period. The cells undergo a dramatic physiological transformation to survive what is essentially a famine.
The master regulator of this transformation is a protein called RpoS (also known as σS), an alternative sigma factor that redirects the cell’s gene-expression machinery toward stress-resistance genes. Under normal growth conditions, RpoS is continuously made and continuously destroyed. When nutrients become scarce or other stresses appear, the destruction stops and RpoS accumulates. It then commandeers the cell’s RNA-copying machinery and turns on a large set of genes that help the cell survive starvation, oxidative damage, acid stress, and osmotic shock. RpoS-dependent gene expression leads to a broadly stress-resistant cell state that is very different from the fast-growing cell of the exponential phase.
The regulation of RpoS is itself sensitive to the state of the cell’s central metabolism, creating a feedback loop: as the metabolic machinery winds down, RpoS accumulates, and the genes it activates further reshape the cell’s metabolism for survival rather than growth. This is why stationary-phase E. coli cells are physically tougher. They have thicker cell walls, more stable DNA packaging, and enhanced defenses against reactive oxygen species. A culture in late stationary phase can tolerate insults that would kill exponentially growing cells outright.
The Death Phase
Eventually, even the survival programs of stationary phase cannot keep every cell alive. As nutrients become critically depleted and waste products reach toxic concentrations, the viable cell count begins to drop. This is the death phase, and while it was once viewed as simple starvation, the reality is more interesting. E. coli possesses genetic modules that can actively trigger cell death under stress.
The best-studied of these is the mazEF toxin-antitoxin system. MazF is a stable toxin protein, and MazE is a short-lived antitoxin that neutralizes it. Under normal conditions, both are continuously produced, and MazE keeps MazF in check. But when severe stress inhibits new gene expression, the unstable MazE degrades faster than it can be replaced, freeing MazF to kill the cell. Several antibiotics that block transcription or translation can trigger this same pathway. The system is activated in a density-dependent manner through a quorum-sensing signal, a small peptide called Extracellular Death Factor (EDF), meaning that cells in a crowded, stressed population are more likely to trigger death than isolated cells.
Research using microscopy and cell-sorting techniques has revealed that E. coli actually has at least two distinct death pathways. One is the mazEF-mediated pathway described above, and the other is an apoptotic-like death program mediated by the DNA-damage response genes recA and lexA. Under conditions of severe DNA damage, the mazEF pathway can actually suppress this second death pathway by reducing recA mRNA levels. The interpretation is that mazEF-mediated death is an altruistic process: some cells die in a controlled way that benefits the surviving population, while the more chaotic apoptotic-like pathway is kept in reserve.
Diauxie and the Double Growth Curve
The classic four-phase curve assumes a single nutrient source. In reality, E. coli often encounters mixed sugars, and this produces a distinctive wrinkle in the growth curve called diauxie. When both glucose and lactose are present, the bacteria eat the glucose first and ignore the lactose entirely. Only after glucose is exhausted does the population pause, switch on the genes needed to metabolize lactose, and resume growing. The result is a growth curve with two exponential phases separated by a second lag phase.
This sequential sugar use was one of the earliest observations in molecular biology. The phenomenon is driven by catabolite repression: as long as glucose is available, the cell actively suppresses the genes for lactose metabolism. A mutation in the gene controlling this repression can relieve glucose-lactose diauxie, confirming that the preference is genetically hardwired rather than a passive consequence of enzyme kinetics. Computational models of cellular metabolism have since shown that diauxie is actually an optimal growth strategy given the cell’s limited capacity to produce enzymes. Making the full set of enzymes for two sugars simultaneously would slow growth on either one, so sequential use turns out to be the faster overall approach.
What Happens After Death
Textbooks usually stop at the death phase, but the story of an E. coli culture does not. Populations left in spent medium for weeks or months enter a regime sometimes called long-term stationary phase, where something surprising happens: evolution kicks into high gear. Cells that can scavenge energy from the debris of their dead neighbors gain a selective advantage. This phenomenon, called the Growth Advantage in Stationary Phase (GASP) phenotype, has been documented in cultures maintained for years without fresh nutrients.
In one study, researchers tracked the genomes of over a thousand clones isolated from a single batch culture incubated for 1,200 days. They found nearly 700 mutations, including large-scale deletions up to 64,000 base pairs and amplifications up to 500,000 base pairs. Two distinct evolutionary lineages diverged from the original strain and continued evolving throughout the experiment. On at least two occasions, a mutation that had previously swept through the entire population reverted back to the original version, suggesting that what is beneficial at one point in this dynamic environment can become harmful later. Most of the mutated genes encoded proteins involved in metabolism, transport, or gene regulation. A separate long-term culture maintained for ten years without nutrient addition similarly produced mutants with the GASP phenotype, demonstrating that E. coli can persist and adapt in sealed batch culture far longer than the standard growth curve implies.
The Viable but Nonculturable State
Beyond death and long-term survival, E. coli can enter a state that does not fit neatly into the four-phase model at all. Under certain stresses, including heat, UV radiation, and chemical disinfection, some cells lose the ability to grow on standard culture plates but remain alive by other measures. Their membranes stay intact, they maintain metabolic activity, and under the right conditions, they can wake up and start dividing again. This is the viable but nonculturable (VBNC) state, and it has real-world consequences for food safety and water treatment.
Studies on E. coli O157:H7, a dangerous pathogenic strain, have shown that even after boiling for ten minutes, a small fraction of cells can survive in the VBNC state. In one set of experiments, roughly 28 to 87 out of an initial hundred million cells retained both membrane integrity and measurable metabolic activity after boiling. These VBNC cells were then resuscitated back to full growth when exposed to signaling molecules produced by common gut bacteria and the stress hormone norepinephrine. The resuscitated cells expressed virulence genes normally associated with the pathogenic strain.
UV treatment tells a similar story. Low-dose UV exposure can push cells into the VBNC state while actually boosting their metabolic activity, as measured by increased cellular energy reserves. DNA repair genes and antioxidant defense genes ramp up under these conditions. When VBNC cells encounter biofilms, which release signaling compounds, they can recover in as little as four hours in nutrient-rich conditions. The practical implication is that standard culture-based methods for detecting E. coli contamination may undercount the actual number of living cells present.
Measuring the Growth Curve in Practice
The most common way to track E. coli growth in the lab is to measure optical density (OD), which is essentially how cloudy the culture is. More cells means more cloudiness. It is fast and non-destructive, but it has a well-known limitation: above an OD of about 0.1, the relationship between cloudiness and actual cell count starts to break down. At higher densities the instrument underestimates the true number of cells, because light scattered by one cell can be re-scattered by another before reaching the detector. Researchers comparing OD measurements to actual cell counts have confirmed that the proportionality between cells per milliliter and OD falls off above this threshold.
The alternative is direct plate counting, where samples of the culture are diluted and spread on agar plates, and colonies are counted the next day. This gives a true count of viable cells but is slow and labor-intensive. Mathematical models bridge the gap by fitting curves to OD data and extracting growth parameters like lag time, maximum growth rate, and final population density. The two most widely used models are the Baranyi model and the modified Gompertz model, and a simpler three-phase linear model has been shown to produce comparable estimates of growth rate with lower variability between experiments. When these three models were compared against established growth data for E. coli O157:H7, all three showed good agreement, though the Baranyi model provided marginally better overall curve fitting.
Why the Growth Phase Matters for Biotechnology
If you are using E. coli as a factory to produce a protein, when you flip the switch matters enormously. Most recombinant protein production in E. coli involves growing the culture to a target density and then adding a chemical inducer (commonly IPTG) that turns on the gene for the desired protein. Induce too early, when the population is small and still in early exponential growth, and the metabolic burden of making a foreign protein can stall the culture. Induce too late, when cells are entering stationary phase and their protein-synthesis machinery is winding down, and yields drop.
Systematic profiling of induction timing has shown that the biomass concentration needs to reach a distinct threshold to tolerate the stress of protein production. In one set of experiments, cultures induced within the first nine hours of cultivation required very low inducer concentrations, while cultures at higher densities could handle more aggressive induction because the inducer-to-cell ratio was lower. The optimal window depends on the specific protein, the strain, and the growth temperature, which is why automated screening of many conditions in parallel has become standard practice. Some researchers have even proposed using simple visual inspection of culture turbidity as a rough guide to when cultures have reached the right growth phase for induction, exploiting the predictable relationship between OD and cell physiology.
Environmental Boundaries of the Growth Curve
The four-phase curve assumes conditions that actually support growth. Change the temperature, pH, or water activity enough, and E. coli never enters the exponential phase at all. Modeling the boundary between growth and no-growth has practical importance for food safety. One widely used model defines this boundary as a function of temperature, pH, lactic acid concentration, and water activity. An interesting finding from this work is that at moderate to warm temperatures, the growth-versus-no-growth boundary is essentially independent of temperature; the dominant factors become pH and the concentration of undissociated acid. In other words, cooling food slows E. coli growth, but it is the acidity and chemical environment that determine whether growth happens at all.
Growth Rate Versus Growth Yield
There is a widespread assumption that faster-growing bacteria are simply “better.” But laboratory evolution experiments have revealed a genuine trade-off between growth rate and growth yield, the total amount of biomass produced per unit of nutrient consumed. When E. coli populations were evolved for 20,000 generations under constant conditions, the relationship between rate and yield was not straightforward. Comparisons across independently evolved populations did not consistently show a negative correlation between the two. However, when researchers looked within individual populations, isolating different clones from the same culture, a clear trade-off emerged: faster-growing clones produced less biomass per unit of glucose consumed.
Further adaptive evolution experiments have refined this picture into a two-dimensional trade-off. One dimension is the trade-off between growth rate and yield; the other is between glucose uptake rate and yield. Both dimensions share the same underlying metabolic mechanism: a switch between efficient energy production through the full metabolic cycle and the less efficient but faster overflow pathway that excretes acetate. Cells that gulp glucose quickly and grow fast waste more of it as acetate. Cells that process glucose more carefully grow slower but extract more energy per molecule. This trade-off has implications for biotechnology, where you sometimes want maximum biomass (favoring yield) and sometimes want the fastest possible culture (favoring rate).
Antibiotic Tolerance Across Growth Phases
The growth phase of a bacterial culture profoundly affects how it responds to antibiotics, and this is not just an academic observation. Most antibiotics work best against actively growing cells, because they target processes like cell-wall construction, DNA replication, or protein synthesis that are running at full speed during exponential growth. Cells in lag phase or stationary phase, where these processes are slowed or stopped, can tolerate antibiotic concentrations that would kill their exponentially growing counterparts. This phenomenon, called antibiotic tolerance, is distinct from genetic resistance. The cells do not carry resistance genes; they simply are not doing the things the antibiotic is designed to disrupt.
A subset of cells in any population, called persisters, can survive antibiotic treatment even during exponential growth. Persister cells are thought to be in a dormant or slow-growing state that makes them phenotypically tolerant. Tracking the formation and disappearance of these tolerant cells throughout all growth phases, from lag through exponential into early and late stationary phase, has become a focus of research aimed at understanding why bacterial infections sometimes relapse after apparently successful antibiotic treatment.