Hoan Van Ngo, Seung Hyeon Kim, Hongseok Ha, Sangwoo Kang, Donghyuk Shin, Matthias Gunzer, Kiwook Kim, Jinki Yeom (2026), bioRxiv — DOI 10.64898/2026.04.08.717352 (v1, posted 9 April 2026, CC-BY 4.0). An unrefereed preprint.
Key takeaways
- A two-component system is cast as an oxidative-stress sensor. The paper argues that A. baumannii, which lacks RpoS, Nif and Suf homologs, defends against peroxide and nitric-oxide stress through the PmrA/PmrB TCS rather than through the usual regulators. The support is genetic and reporter-based: pmrA/pmrB nulls fail to grow under H2O2, plasmid-borne copies rescue, and iron chelators (DP, DFO) and a hydroxyl-radical scavenger (thiourea) restore the mutant, placing the effect on the Fenton reaction. OxyR inactivation does not phenocopy, so the two pathways are presented as independent.
- PmrA is placed directly upstream of concrete defence genes. Putative PmrA boxes, luciferase reporters induced by H2O2 (but not by colistin), and EMSA binding (competed out by unlabelled promoter) are given for the hscB-hscA-fdx Isc operon, the ferritin ftnA, the peroxidase ahpF1 and the catalase katE; katG is a stated negative control. This is a promoter-binding and reporter argument, not direct in-vivo transcription across the regulon.
- The sensing chemistry is the novel claim: a nickel cofactor on a histidine box. Four conserved periplasmic histidines (His86/89/90/92), specific to Acinetobacter, are required for H2O2 survival but not colistin resistance; Ni2+ (not Co2+/Mn2+/Zn2+) supplementation boosts the response; ICP-MS and a fluorescent dye detect Ni2+ on wild-type but not histidine-substituted purified periplasmic domain; AlphaFold3 (with Fe2+ as a Ni2+ proxy) models the histidines forming a metal pocket. This is presented as the first report of a TCS sensor using a Ni2+ cofactor with multiple histidines for oxidative sensing.
- A redox switch is argued from simulation, not measured conformations. Molecular-dynamics runs (apo, Ni2+, Ni3+, 100 ns) report that oxidising the bound nickel to Ni3+ expands the periplasmic domain (RMSD ~5.5 Å vs ~4 Å) while keeping tight local histidine-box geometry; PCA and differential contact maps are read as a dominant, directional conformational transition. The switch is a computational inference.
- “Response memory” is the headline concept. Priming with sublethal H2O2 (1—10 µM) is reported to raise both oxidative-defence and antimicrobial- peptide-resistance genes and to improve later survival to 30 mM H2O2 and to colistin, PmrA- and histidine-box-dependent and OxyR-independent. The effect is said to last 30—90 min after signal removal, with CFU counts offered to rule out dilution by division as the explanation for its decay.
- Host and clinical claims. Intratracheally infected mice show less lung inflammation, fewer neutrophils and lower bacterial burden with the pmrB mutant and with the histidine-box substitution; the histidine box is present in a hypervirulent carbapenem-resistant clinical isolate (A0062) and absent in a low-virulence one (A0075), and pmrB disruption in A0062 reduces priming-dependent survival.
The record’s reading of what is shown versus asserted, and the reasons for caution, are in NOTE-656. In short: the mechanistic chain (Ni2+ binding → oxidation → conformational switch → signalling) is assembled from binding assays, reporters and simulation rather than observed end to end, and the central “memory” and virulence claims rest on this one unrefereed study.
Standing in the record
Filed on 2026-10-09 at the owner’s request, who asked for this preprint to be registered and read, with no stated context. It is read on its own merits. Because it is an unrefereed bioRxiv preprint, the reading separates what the experiments show from what the authors claim, and treats the mechanism and the host/clinical conclusions as not yet independently established.
The reading bears on THEORY-136, the record’s statement that a lasting change after a temporary disturbance is only weak evidence for a system with alternative stable states unless hysteresis, initial-state dependence or a lasting shift with slow return excluded is shown. This paper’s “response memory” is exactly such a lasting-shift claim, and it supplies the dilution-excluding control THEORY-136 asks for (CFU unchanged over the memory window) while stopping short of demonstrating true bistability. No THEORY is filed from this reading: the account is a single unrefereed study, and the claim it would state is the one THEORY-136 counsels holding at arm’s length.
Not held in the Anthology of the SOTA. The paper uses AlphaFold3, MD and PCA
as tools but makes no machine-learning-practice claim, so it carries no
anthology-candidate flag.