Why σE heterogeneity alone cannot explain colistin heteroresistance

The single-switch model that colistin heteroresistance inherited

The baseline assumption was that one inducible system both senses the antibiotic and builds resistance. In Enterobacter cloacae complex, PhoPQ binds the arnB promoter and drives l-Ara4N addition to lipid A, producing cationic modification that blocks colistin binding; PmrAB, the canonical polymyxin resistance system in E. coli and Salmonella, is dispensable in this species [7]. Random mutagenesis in an E. cloacae strain independently pointed to the arn operon, specifically arnA, as the determinant of colistin heteroresistance [8]. Epidemiology reinforced the pattern: among 138 ECC clinical isolates, 38 (27.5%) were colistin-resistant and all showed a heteroresistant phenotype by population analysis profiling, with heteroresistance concentrated in Enterobacter roggenkampii, Enterobacter kobei, Enterobacter chuandaensis, and E. cloacae subspecies rather than Enterobacter hormaechei or Enterobacter ludwigii [6]. In that survey, colistin-induced arnT upregulation was PhoPQ-dependent, and mcr-9 or mcr-10 carriage did not distinguish resistant from susceptible isolates [6].

That body of work established the resistance machinery but left the population structure unexplained. If PhoPQ and arn are required, what makes only a minority of cells resistant? The implicit answer was heterogeneous activation of the same pathway. The new study tested that assumption directly and found it insufficient [1].

σE marks the survivors, but does not arm them

The anchor paper began with a transposon screen of roughly 16,000 colonies and recovered 42 mutants with reduced colistin-resistant subpopulations, including clpP and sspB, which act together in σE activation [1]. Deleting clpP or sspB lowered survival, while deleting the anti-sigma factor rseA or rseB raised σE activity and expanded the resistant fraction; in the ΔrseA background nearly the entire population survived 100 μg/ml colistin [1]. Single-cell reporters showed σE activity was bimodal before any drug exposure, with about 1% of cells in a high-activity peak, and time-lapse tracking showed that only those high-σE cells survived colistin [1]. The arn promoter, by contrast, was unimodally distributed, so arn expression alone could not explain which cells survived [1].

The decisive experiment was the ΔphoPQ mutant. Heterogeneous σE activation persisted, and a rare subpopulation with elevated MIC was still detectable, but no cell survived the clinical breakpoint [1]. In other words, σE supplies the heterogeneity, PhoPQ supplies the resistance, and neither alone yields heteroresistance. This is a mechanistic separation, not a semantic one: the two traits can be measured independently in the same strain [1].

Why this is not just PhoPQ with extra steps

The strongest precursor evidence comes from Pseudomonas aeruginosa, where Mg2+ depletion by Candida albicans drives very high colistin resistance through convergent mutations in lipid A genes, the phoPQ operon promoter, and the putative Mg2+ transporter PA4824 [2]. In that system, resistance required PhoPQ activity, and deleting phoP sharply reduced MIC in evolved clones, while pmrA deletion had only a mild effect [2]. Two trajectories diverged in fitness cost: htrB2 loss-of-function increased membrane permeability and collateral sensitivity to other antibiotics, whereas an htrB2-independent route achieved resistance without the same trade-offs [2]. That work shows PhoPQ-centered remodeling can generate high resistance under a specific nutritional condition, but it does not describe a separate heterogeneity generator, and its endpoint is a largely resistant population rather than a minority subpopulation [2].

The anchor paper's model accommodates that difference. PhoPQ sets the baseline resistance level across the population, while σE activation determines which cells cross the breakpoint [1]. The authors note that genetic variation in PhoPQ or RseA/σE could shift the frequency of the resistant subpopulation by changing that baseline, which is consistent with the observation that different Enterobacter isolates show different heteroresistance frequencies [1]. The model also fits the earlier Enterobacter finding that inactivating PhoPQ/arn eliminates the resistant subpopulation while manipulating PhoPQ changes its frequency without removing baseline heterogeneity [1].

Single-cell measurement is now the contested ground

The anchor paper's claims rest on single-cell reporters and time-lapse tracking, and that methodological choice matters because population-level assays cannot distinguish heterogeneity generation from resistance execution [1]. A competing framing comes from droplet microfluidics work on E. coli W3110, which infers absolute single-cell susceptibility from paired initial-cell-number and growth-outcome measurements and relates those to MIC and MBC [3]. That study explicitly notes that heteroresistance has been linked to gene amplification in 61% of clinical cases while the remainder is unexplained, and it treats persistence and heteroresistance as points on a spectrum defined by whether the subpopulation can replicate at high drug doses [3]. The anchor paper rules out gene amplification in its strain: no sequence variants or copy-number changes distinguished resistant from susceptible cells, and the resistant fraction returned to baseline within one passage off drug [1].

A separate single-cell study in Pseudomonas aeruginosa used multiplexed transcriptomics across 144 genes to show that quorum-sensing hyper-signaling subpopulations arise independently of inoculum history and external autoinducer levels, pointing to internally regulated differentiation [4]. That is a different system and a different phenotype, but it supports the broader idea that subpopulation emergence can be a regulated property rather than simple noise. The anchor paper's σE bimodality is consistent with that framing, though the mechanism producing the high-σE peak remains unresolved [1].

Where the combinatorial model stops

The conclusion is bounded by its system. The evidence comes from Enterobacter isolates, principally E. cloacae complex strain RS, with supporting experiments in E. kobei Mu208 and two additional clinical isolates that could not be genetically manipulated [1]. The authors state that the mechanism by which σE responds to colistin and how it increases arn-mediated lipid A modification remain unknown, and they could not generate a ppGpp-null strain to rule out a stringent-response contribution [1]. They also note that σE activation can elevate arn expression without evidence of direct σE binding to the arn promoter, leaving direct versus indirect regulation open [1].

Clinical heterogeneity adds another boundary. A colistin-heteroresistant E. cloacae ST48 isolate from raw cow milk carried a stop-codon mutation in phoP in its resistant subpopulation, which sits awkwardly with a simple PhoPQ-required model and suggests strain-specific rewiring [5]. The Japanese survey found heteroresistance only in certain ECC lineages and none in E. hormaechei or E. ludwigii [6], and the earlier mechanistic work was itself limited to particular ECC clusters [7]. The anchor paper's combinatorial paradigm is therefore best read as a validated mechanism in specific Enterobacter backgrounds, not yet a general rule for Enterobacterales. Whether σE bimodality, PhoPQ-dependent arn induction, or some other pairing operates in Klebsiella, Acinetobacter, or E. coli heteroresistance remains to be tested [1].

About These Sources

This research page is built on 8 peer-reviewed studies — published from 2019 to 2026, 6 from 2024 or later, collectively cited 116 times — selected as the most relevant from 12 studies that passed quality screening, drawn from 87 papers retrieved from a database of over 500 million.

Sources used in this answer