When a bacterial isolate is reported as susceptible to an antibiotic, the result answers an essential clinical question: can the drug inhibit growth at concentrations relevant to treatment? The minimum inhibitory concentration (MIC) is a foundational measure of antimicrobial susceptibility. But inhibition of growth does not describe how quickly a bacterial population dies after antibiotic exposure begins.
That distinction matters because bacteria can survive antibiotic exposure without becoming resistant. Antibiotic tolerance describes a population that takes longer to be killed despite showing no corresponding increase in MIC. Persistence is more specific: a subpopulation survives substantially longer than the majority, producing a characteristic killing curve in which rapid initial killing gives way to a slower phase. If those surviving persister cells are regrown without antibiotic and challenged again, their descendants remain susceptible.
Resistance is reflected in the concentration of antibiotic required to inhibit bacterial growth, while persistence becomes apparent only when survival is followed over time.
The practical difference has been demonstrated in clinical isolates. In a 2025 study of 15 clinical Escherichia coli isolates, high-resolution time-kill assays exposed distinct patterns of killing across several β-lactams and ciprofloxacin that were not apparent from MIC measurements alone. The researchers noted that tolerant populations can retain the same MIC as otherwise comparable susceptible populations and that neither tolerance nor persistence can currently be detected through routine clinical microbiology diagnostics.1
Persistence Lives in the Kill Curve
Time-kill assays follow bacterial survival directly. A culture is exposed to an antibiotic, samples are collected at successive time points, and the remaining viable bacteria are quantified. The resulting curve shows how quickly the population declines throughout treatment.
For persistence, the shape of that curve carries important information. A susceptible population may decline rapidly at first, followed by a much shallower phase as a smaller fraction survives for longer. This multiphasic pattern distinguishes persistence from uniform population-wide tolerance and from resistance, in which bacteria can continue growing at higher drug concentrations.2
The kinetics can also vary by antibiotic. In the 2025 E. coli study, β-lactam exposure produced an initial bacteriostatic period followed by rapid and then slower killing, while ciprofloxacin lacked the same initial bacteriostatic phase. The proportion of bacteria remaining in the slower-killing fraction also differed among drugs.
Conventional time-kill experiments, however, become difficult to scale. Comparing hundreds of genetic backgrounds ordinarily requires separate cultures, repeated sampling, and viable-cell measurements for each condition. Pooled genetic screens can examine many perturbations simultaneously, but a single survivor measurement loses much of the temporal information that defines persistence.3
Regulatory time-kill sequencing (RTK-seq) was developed to combine pooled genetic screening with time-resolved survival measurements.
Turning Hundreds of Kill Curves Into a Pooled Experiment
In September 2026, researchers at Washington University reported RTK-seq, a functional-genomics method that combines pooled CRISPR interference (CRISPRi), antibiotic exposure, repeated sampling, and sequencing to reconstruct survival trajectories for many bacterial regulatory perturbations at once.3
The researchers used a library containing guide RNAs targeting 183 known E. coli transcription factors. Rather than deleting those regulators, CRISPRi suppressed their expression. The entire library could then be challenged with an antibiotic as a pooled population. Sequencing tracked how each knockdown became more or less abundant among survivors over time, while parallel viable-cell measurements established survival of the overall population. Combining those data allowed the researchers to reconstruct individual time-kill curves from the pooled experiment.
Under ampicillin exposure, the approach produced survival trajectories for 174 transcription-factor knockdowns spanning approximately five orders of magnitude in survival. Most remained close to the negative control, while others produced substantial increases or decreases. The researchers reproduced selected trajectories with conventional individual time-kill experiments, providing an independent check on the pooled measurements.
Separating Persister Abundance From Persister Killing
A fixed survivor count can show that one bacterial population outlasts another, but it cannot necessarily explain how that difference arose. A perturbation could change the size of the slow-surviving fraction, alter how quickly those persisters are subsequently killed, or affect both features.
The RTK-seq team fitted the ampicillin trajectories to a biphasic model that separated the estimated persister fraction from the killing rate of the slower-surviving population. Twenty-nine transcription-factor knockdowns significantly altered the estimated persister fraction, while five significantly altered persister-killing kinetics. Only one regulator, acrR, appeared in both groups.
This gives genetic screens greater resolution. Rather than grouping every condition that produces more or fewer survivors into the same category, RTK-seq can distinguish perturbations associated with different portions of the killing trajectory and provide a more specific basis for investigating the mechanisms that shape persistence.
The Persistence Network Changes With the Antibiotic
Extending RTK-seq beyond ampicillin revealed an even broader source of variation.
The researchers repeated the workflow with streptomycin, an aminoglycoside that inhibits protein synthesis, and ofloxacin, a fluoroquinolone that disrupts DNA replication. The sets of regulators whose knockdown altered persistence differed markedly among the three antibiotics. Under streptomycin treatment, 60 transcription-factor knockdowns showed decreased persistence relative to the control, while the ofloxacin experiment identified 10 knockdowns with pronounced persistence changes.
Only a small subset of regulators showed consistent effects across all three antibiotics. Repression of hipB increased persistence in each condition, while sdiA and ihfB knockdowns decreased it. Much of the remaining pattern depended on the drug.
The result complicates models in which persistence arises from a single regulatory program that operates similarly under any lethal antibiotic stress. In this E. coli system, different drugs exposed different combinations of regulatory dependencies.
The study does not establish that broadly conserved persistence mechanisms do not exist. The shared effects observed across the three antibiotics already suggest some common features, and the experiments examined one organism under defined laboratory conditions. What the results show is that a substantial portion of the persistence phenotype identified by RTK-seq depended on antibiotic context.
The clinical-isolate study points in the same general direction from a different experimental perspective.1 Killing patterns and the proportion of cells occupying slower-killing phases differed among β-lactams and ciprofloxacin in clinical E. coli isolates.
Persistence may therefore reflect not only the physiological state of the bacterium but also the particular stress imposed by the antibiotic being used.
New Regulators Provide Starting Points for Mechanistic Work
RTK-seq recovered regulators linked with previously characterized persistence biology, including pathways involving toxin-antitoxin systems, oxidative stress, drug transport, and membrane permeability. It also identified rcsA, narP, and csgD as transcription factors that the authors reported had not previously been associated with persistence.
Follow-up transcriptomic analysis sought to connect those upstream regulators with downstream genes that might contribute to the observed phenotypes. Among several candidates tested under ampicillin, perturbations of csgE, uspE, and carA affected persistence, although only the carA result reached statistical significance.
Those experiments did not reduce each regulatory effect to a single downstream gene. Because transcription factors can influence multiple targets at once, the resulting phenotype may reflect combinations of transcriptional changes. RTK-seq identifies regulatory nodes that can then be examined in more focused mechanistic studies.
What Antibiotic-Specific Persistence Could Mean for Drug Development
Anti-persister drug development rests on the possibility that bacterial survival during antibiotic treatment can be disrupted even when the organism remains conventionally susceptible. RTK-seq provides a systematic way to identify regulatory perturbations associated with smaller persister fractions or faster killing of surviving cells.
Its antibiotic-specific results raise a more difficult development question. If many persistence dependencies change according to the antibiotic, a target that sensitizes persisters during treatment with one drug may not produce the same effect with another.
Researchers could search for conserved dependencies that operate across multiple antibiotic stresses and might support broadly useful adjunctive therapies. Alternatively, anti-persister activity could be evaluated within a defined antibiotic context, pairing an adjunct with the drug or antibiotic class whose associated survival program it can disrupt.
RTK-seq does not establish which approach will prove more productive, and the regulatory hits identified in the study are not themselves validated drug targets, but it provides a way to test whether candidate persistence mechanisms remain important when the accompanying antibiotic changes.
Laboratory evolution experiments have also shown that antibiotic tolerance can arise before resistance and facilitate the subsequent emergence of resistant mutants under repeated exposure, providing another reason to understand survival phenotypes that fall outside conventional resistance measurements.4
What RTK-seq Does Not Yet Capture
The current RTK-seq experiments used exponentially growing E. coli in defined M9 glucose medium. The authors identify stationary-phase populations, nutrient limitation, and host-mimicking environments as important settings for future studies because bacterial physiology can substantially affect persistence.
The CRISPRi approach also has technical limitations. Repression can be incomplete and can vary among targets, potentially obscuring phenotypes that require stronger changes in transcription-factor activity. The library contains a single guide RNA for each transcription factor, leaving some possibility that guide-specific effects or effects on neighboring genes influence individual results.
The method also interrogates one particular layer of persistence biology. By targeting transcription factors, it maps regulatory influences efficiently, while post-transcriptional regulation, metabolic state, membrane physiology, and growth state can influence survival independently of, or downstream from, the transcriptional perturbations captured here.
Nor is RTK-seq a clinical susceptibility assay. High-resolution time-kill experiments can identify tolerance and persistence in clinical isolates, but their labor and time requirements make them difficult to incorporate into routine diagnostic microbiology.1 RTK-seq instead scales time-resolved phenotyping across many genetic perturbations for mechanistic research.
Whether kinetic persistence measurements can eventually be made practical enough for routine clinical microbiology remains an open question.
Following Survival Through Time
Antibiotic susceptibility tells us whether a drug can inhibit bacterial growth. Persistence reveals another layer of the response: some susceptible cells can follow very different survival trajectories once treatment begins.
RTK-seq makes those trajectories accessible at a scale that conventional time-kill experiments cannot readily achieve. Its first application also suggests that the regulatory routes supporting prolonged survival can change substantially with antibiotic context.
That shifts the emphasis away from searching only for a universal persister state. A more useful question may be which survival program emerges under a particular treatment pressure, when it becomes important during killing, and where that trajectory can be interrupted.
