A Neonatal Unit, a Plasmid and a Cycling Proposal
An outbreak of carbapenem-resistant Enterobacter in a 24-bed neonatal unit involves a strain carrying a ~120 kb IncF plasmid with blaCTX-M-15 and a toxin–antitoxin pair. In drug-free chemostats the plasmid-bearing strain doubles every 42 min while its plasmid-cured isogenic parent doubles every 34 min; in 4 mg/L meropenem the cured strain washes out within 48 h. After 300 generations of serial passage without antibiotic, a derivative retains the plasmid in 99.9% of cells and doubles every 35 min. Mating assays give a transfer frequency of 10⁻⁶ per donor per hour. Whole-genome sequencing shows a single point mutation in the plasmid's replication control region in the derivative. The unit's committee proposes alternating meropenem and gentamicin on a 4-week cycle to 'starve the plasmid of selection', while infection-control staffing has been cut by 30%.
1. Quantify the fitness cost of plasmid carriage and predict the dynamics of resistance with and without meropenem, including the derivative.
Read the worked answer
Doubling times: 42 min (plasmid-bearing) versus 34 min (cured). Extra time per generation = 8 min, i.e. 8/34 ≈ 24% slower growth. In continuous culture the selection coefficient against carriage is approximately s ≈ 1 − 34/42 ≈ 0.19 per generation. In drug-free medium the plasmid-bearing lineage should therefore be displaced over tens of generations.
In 4 mg/L meropenem, the cured strain washes out within 48 h, so the drug imposes very strong positive selection for carriage, far exceeding s = 0.19.
The derivative changes the prediction: it doubles in 35 min, so the cost has fallen to about 1 − 34/35 ≈ 3%, essentially neutral, while retention is 99.9%. Compensation has occurred, here via a mutation in plasmid replication control. Persistence of resistance after drug withdrawal is therefore not paradoxical: the plasmid no longer burdens the host much, so there is little selection to lose it. Expect resistance to decline slowly, if at all, and only where plasmid transfer and re-importation from other patients are also controlled.
2. Critique the 4-week alternating antibiotic cycle using the data. What does the data actually support?
Read the worked answer
Cycling assumes drug-free intervals give the plasmid-free strain an advantage that reverses resistance. The data undermine that assumption twice over.
First, the fitness cost has been compensated: 3% rather than 19%. A drug-free window therefore gives almost no competitive advantage to cured cells, so the plasmid will not be 'selected out' on the timescale of a 4-week block.
Second, transfer at 10⁻⁶ per donor per hour plus ward transmission means the plasmid can re-enter susceptible cells and new patients. The relevant competition is between strains across a hospital, not within one gut, and cycling acts on within-patient selection while doing little to transmission.
Cycling also risks selecting strains resistant to both agents if determinants are linked, and 4-week blocks are long enough for within-host selection to act. The data support a different conclusion: this is primarily a transmission problem. Reinstate infection-control capacity (cohorting, hand hygiene, screening, staff ratios), restrict carbapenems to culture-directed targeted therapy, and use combination therapy only where synergy is demonstrated rather than as a rotation schedule.
3. Explain how the toxin–antitoxin pair and the replication-control mutation each contribute to persistence, and design an experiment to test whether the mutation is sufficient.
Read the worked answer
A toxin–antitoxin pair encodes a stable toxin and a labile antitoxin. When a daughter cell fails to inherit the plasmid, the antitoxin decays faster than the toxin, so the cell is killed or growth-arrested - post-segregational killing. This explains high retention (99.9%) even when carriage is costly, but it does not by itself shorten the 42-min doubling time. The replication-control mutation must therefore do separate work, most plausibly raising copy number or replication efficiency so the host pays less.
Test: cure the host, then introduce in parallel (i) the ancestral plasmid, (ii) the derivative plasmid, and (iii) the ancestral plasmid carrying only the replication-control point mutation, all in the same isogenic cured background. Measure doubling time, plasmid copy number and retention over 200 generations of drug-free serial passage. If the point mutation alone reproduces the 35-min doubling time and 99.9% retention, it is sufficient. If deleting the toxin–antitoxin collapses retention without affecting doubling time, the TA system is required for stability but not for fitness recovery - the two mechanisms are separable.