Bacterial Persister Cells: The 1-in-10,000 Dormant Survivors of Antibiotics

What persister cells are and where they arise

Persister cells are phenotypic variants within a clonal bacterial population that enter a transient, dormant or slow-growing state and thereby survive concentrations of bactericidal antibiotics that kill the rest of the population. Crucially, they are not resistant: their minimum inhibitory concentration (MIC) is unchanged, and their offspring are as susceptible as the parent strain. What differs is the minimum duration for killing (MDK) — the time the drug needs to sterilize the culture.

• Exponential-phase cultures carry very few persisters (~10⁻⁶ to 10⁻⁵), most arising by stochastic switching.
• Stationary phase, nutrient starvation, and biofilms raise the persister fraction toward 0.1–1%, because stress signals drive many cells into dormancy.

The diagnostic signature is a biphasic time-kill curve: the bulk population dies rapidly over the first hours, then the curve flattens into a slow-dying tail of persisters. Persistence is documented across E. coli, Staphylococcus aureus, Pseudomonas aeruginosa, and Mycobacterium tuberculosis, and it is a leading explanation for why chronic and biofilm-associated infections relapse after apparently successful therapy.

The mechanism, step by step

Persistence is fundamentally about shutting down the targets that antibiotics need active. β-lactams need active cell-wall synthesis; fluoroquinolones need active DNA gyrase and replication; aminoglycosides need active translation and proton-motive force. Dormancy removes these vulnerabilities.

The best-characterized route runs through toxin-antitoxin (TA) modules and the stringent-response alarmone:

• 1. Trigger. Stress or stochastic fluctuation tips the balance in a TA pair. The labile antitoxin (e.g. HipB) is degraded by proteases (Lon, ClpXP), freeing the stable toxin (HipA).
• 2. Toxin acts. HipA is a serine/threonine kinase that phosphorylates GltX (glutamyl-tRNA synthetase) at Ser239, blocking tRNA^Glu charging.
• 3. Stringent response. Uncharged tRNAs stall ribosomes; ribosome-associated RelA (and SpoT) synthesize (p)ppGpp.
• 4. Global shutdown. (p)ppGpp binds RNA polymerase and primase, collapsing rRNA/tRNA synthesis, replication, and translation. The cell becomes metabolically quiescent — a persister.
• 5. Resuscitation. When drug and stress clear, antitoxin is resynthesized, (p)ppGpp is hydrolyzed by SpoT, and the cell resumes growth.

Key molecules and characteristic numbers

The molecular cast is small but well defined:

• HipA / HipB (hipBA operon). The founding TA persistence module in E. coli. The gain-of-function allele hipA7 raises persister frequency ~100–1000-fold (from ~10⁻⁶ to ~10⁻³) without changing MIC.
• (p)ppGpp — guanosine tetra/pentaphosphate, the master alarmone. Cells lacking both synthetases (ΔrelA ΔspoT, "ppGpp⁰") show sharply reduced persistence.
• Lon protease — degrades antitoxins, and its activity is amplified by (p)ppGpp and polyphosphate, creating a positive-feedback loop toward dormancy.
• mRNA endonuclease toxins (MqsR, HicA, and the RelE/MazF families) cleave mRNA or ribosomal targets to halt translation.

Characteristic quantities: exponential persister fraction ≈ 10⁻⁶–10⁻⁵; stationary/biofilm fraction ≈ 10⁻³–10⁻²; dormancy lasts hours to days; and a single dormant cell that resuscitates can reseed an entire infection. Balaban's single-cell work measured switching rates on the order of ~10⁻⁶ per cell per generation into the persister state, with faster switching back out.

How persisters are studied and observed

Because persisters are rare and transient, they resist bulk assays and require single-cell and time-resolved methods:

• Time-kill (biphasic) assays. Expose a culture to a bactericidal drug at ≥10× MIC, plate survivors over time, and read the flattened tail as the persister fraction.
• Microfluidics + microscopy. Nathalie Balaban and colleagues (2004, Science) trapped single E. coli in a "mother machine," tracked lineages through antibiotic exposure, and showed persisters are pre-existing slow/non-growers — defining type I (triggered by passage through stationary phase) and type II (continuously generated stochastically) persisters.
• Fluorescent dilution / reporters. Growth-rate or ribosome reporters and dye-dilution let researchers sort dormant cells by flow cytometry.
• Genetics. The hip mutants (Moyed & Bertrand, 1983) isolated hipA7 by chemical mutagenesis, first tying a gene to "high persistence."

Regulation is probabilistic: a threshold model holds that a cell becomes dormant when free toxin exceeds a set point, and the depth of that excess sets the dormancy duration — turning noisy toxin-antitoxin fluctuations into a bet-hedging switch.

How persistence differs from resistance and tolerance

These three survival strategies are routinely conflated but are mechanistically distinct:

• Resistance lets bacteria grow in the presence of drug; it raises the MIC and is caused by mutations or acquired genes (efflux pumps, β-lactamases, target modification). It is heritable and population-wide.
• Tolerance lets the whole population survive longer without growing; MIC is unchanged but the MDK rises. It is usually a heritable, uniform trait (e.g. slow metabolism).
• Persistence is tolerance restricted to a small subpopulation. MIC is unchanged; only the dormant fraction survives; the trait is not inherited — regrown persisters are again fully susceptible, which is the definitive test distinguishing them from resistant mutants.

Persistence also differs from viable-but-non-culturable (VBNC) cells, an even deeper dormancy that fails to grow on standard media, and from persistence via metabolic dormancy in the stringent response more broadly. The unifying idea: resistance changes the drug's potency (MIC), while persistence and tolerance change the time needed to kill (MDK).

Why it matters: chronic infection and open questions

Persisters are a prime driver of relapsing and chronic infections. In M. tuberculosis, dormant persisters underlie the months-long, multi-drug regimens needed to prevent relapse. In cystic-fibrosis lungs, P. aeruginosa biofilms accumulate high-persistence variants; in S. aureus and E. coli urinary and device-associated infections, persisters seed recurrence after therapy stops. Persistence is also an evolutionary stepping-stone to resistance: tolerant/persistent populations survive long enough to acquire true resistance mutations.

Therapeutic strategies under study include metabolite-potentiated aminoglycosides (adding fructose or fumarate to restart proton-motive force and let the drug enter), the acyldepsipeptide ADEP4 that dysregulates ClpP to self-digest dormant cells, and "anti-persister" pulsed-dosing regimens.

Open questions remain sharp: Is there a single unifying trigger, or many redundant paths? Deleting all ten E. coli mRNA-endonuclease TA modules did not abolish persistence, and several high-profile TA/(p)ppGpp papers were later disputed or retracted — so the field now treats dormancy as multifactorial, with (p)ppGpp, energy depletion, and ATP levels all contributing, and no one gene as the master switch.

Persistence vs. resistance vs. genetically encoded tolerance

Feature	Persistence (persisters)	Tolerance (population-wide)	Resistance
Genetic change	None (phenotypic)	Usually mutation (e.g. metabolic)	Mutation or acquired gene
Fraction of population	Small subpopulation (10⁻⁵–10⁻²)	Whole population	Whole population
Effect on MIC	MIC unchanged	MIC unchanged	MIC increases
What changes	Kill rate / survival time (↑ MDK)	Kill rate (↑ MDK)	Growth in presence of drug
Heritable	No — revert to susceptible	Yes	Yes
Kill curve	Biphasic (fast then slow)	Uniformly slow	No killing at MIC