Source
Bhattacharyya, Bhattarai, Pfannenstiel, Wilkins, Singh and Harshey (2023), LIT-853, read in NOTE-657.
What was actually shown
- Inheritance. Single planktonic E. coli MG1655 cells, isolated by dilution to extinction, give swarms whose diameters vary more than those from 100- or 10,000-cell inocula. The single-cell daughters of one mother, harvested at a calibrated fourth generation, swarm alike; at the seventh, sibling variability returns. Cells taken from a swarm’s edge are uniform good swarmers and keep it to the fourth generation. This could have come out otherwise: siblings could have been as variable as unrelated cells at G4.
- Iron. Among environmental perturbations and ten plasmid-expressed swarming genes, only fepA and fur change the variability. An iron chelator makes cells better swarmers and holds that to G7; added iron or fepA overexpression makes them worse and holds that to G12. A fepA-promoter sfGFP reporter, bright when Fur is derepressed by low iron, correlates with sorted single cells’ swarm diameter (Spearman r = 0.7067, p = 8.66 × 10⁻¹⁶), not in a constitutive-promoter control (r = 0.15) or in ΔfepA (r = 0.05). In five sorted mothers, G4 daughters follow the mother’s reporter level.
- Not a constant-rate switch. A three-class switching model (XS, M, L) reproduces the lineage fractions when each founding cell keeps its class for a fixed delay (τ ≈ 3 generations for XS and M, ≈ 4 for L) and then switches at constant rates toward a 52/35/13% steady state. The authors rule out bistable and toggle switches and epigenetic inheritance because the state lasts only a few generations.
What this does not say
- Not that iron has been shown to be the carrier. Iron is never measured. The reporter reads Fur-repressed transcription from a multicopy plasmid; the evidence is correlation plus perturbation of the iron pathway. “Iron status, as the Fur regulon reports it” is the most the evidence supports.
- Not that bacteria learn, decide or are conditioned. The paper’s “memory”, “conditioning” and “decision-making” name the persistence of a physiological state across divisions. No association between stimulus and response is tested, and nothing here supports a claim of cognition in bacteria beyond that persistence.
- Not that the state is an alternative stable state. It relaxes back to the population distribution; persistence for a few generations is what a slow variable produces (compare THEORY-136).
- Not a precise timescale. Only G0, G4, G7 and G12 were sampled, with generations inferred from timing, so “four” and “seven” are the time points tested. The fixed-delay model has five parameters fitted to two time points, and no delay-free fit is reported, so the claim that switching is not memoryless is weak.
- Not that the same state governs biofilm and antibiotic survival across generations. Those correlations were measured in the founding cells only. The bet-hedging reading and the generality of the mechanism are the authors’ conjectures.