Mammalian intestinal epithelial cells were shown in 2016 to produce an 'AI-2 mimic' - a host metabolite detected by the bacterial quorum-sensing receptor LuxP that activates bacterial autoinducer-2 (AI-2) signaling - in response to bacterial exposure, the cytolytic toxin aerolysin, tight-junction disruption, or nutrient stress, but its chemical identity, biosynthesis, and physiological roles were unknown. In 2026 the Bassler and Seyedsayamdost laboratories identified L-xylosone (a reactive dicarbonyl sugar newly recognized in the human metabolome) and L-xylulose as epithelial AI-2 mimics that agonize LuxP, resolving the identity question for the nutrient-stress condition. What remains open is the mechanistic and physiological completion of this interkingdom signaling system: the non-ascorbate biosynthetic route by which epithelial cells make L-xylosone/L-xylulose and its coupling to tight-junction damage and bacterial stimulation, confirmation of which mimic mediates bacterially elicited activity, the additional bacterial factors that stimulate host mimic production beyond aerolysin, the mechanism by which bacterial transketolase (Tkt) and heptose-biosynthesis enzyme HldE enable LuxP-mediated detection, and the in vivo roles of host-derived AI-2 mimic signaling in host-microbial associations.

Background

Quorum sensing (QS) is the process by which bacteria communicate using extracellular signal molecules called autoinducers to orchestrate collective behaviors such as bioluminescence, biofilm formation, and virulence. One autoinducer, AI-2, is produced by the widely conserved enzyme LuxS from S-ribosylhomocysteine and is unusual in being shared across species, so it is treated as a universal interspecies signal, including in the mammalian gut. Chemically, AI-2 is not one molecule but a family of interconverting compounds derived from 4,5-dihydroxy-2,3-pentanedione (DPD): in boron-rich marine environments DPD cyclizes with borate to S-THMF-borate (a boron-containing cyclic derivative of DPD), the ligand recognized by the periplasmic binding protein LuxP in Vibrio species, whereas in enteric bacteria such as Salmonella and Escherichia coli the receptor LsrB recognizes a boron-free cyclic form (Chen et al. 2002; Ismail et al. 2016).

Ismail, Valastyan and Bassler (2016) reported that the communication runs in the opposite direction to what was expected: mammalian epithelial cells themselves produce an AI-2-mimic activity. Using Caco-2 human intestinal epithelial cells (a standard colonic epithelial cell line) and a Vibrio harveyi reporter strain that cannot synthesize AI-2, they showed that epithelial cells release a heat-stable, small, polar molecule upon direct or indirect bacterial exposure, upon treatment with aerolysin (a pore-forming cytolytic toxin from Aeromonas that disrupts epithelial tight junctions, the seals between adjacent epithelial cells), upon culture in phosphate-buffered saline, which deprives the cells of nutrients and also increases tight-junction permeability, upon tight-junction disruption by dextran sulfate sodium, or in response to other bacterial cytotoxins. The mimic requires boric acid for maximal activity, is detected through the canonical LuxP/LuxQ pathway (LuxP, the periplasmic binding protein, delivering the signal to its partner membrane sensor kinase LuxQ), and activates QS-controlled gene expression in V. harveyi and in the enteric pathogen Salmonella Typhimurium. Eukaryotic cells lack the bacterial AI-2 biosynthetic enzyme LuxS and cannot convert the bacterial intermediate S-ribosylhomocysteine into the mimic, so an independent host pathway must produce it. Bacterial genetics in V. harveyi showed that the aerolysin gene apt and its regulator VIBHAR_02470 contribute to stimulating host mimic production but do not account for all stimulatory activity, while the metabolic genes tkt (transketolase, a pentose-phosphate-pathway enzyme) and hldE (a bifunctional heptose 1-phosphate adenyltransferase in lipopolysaccharide inner-core heptose biosynthesis) are required for the bacterium to detect the mimic even though LuxP is the receptor - a mechanistic puzzle the authors left open. They also left open the mimic's chemical identity, its biosynthesis in epithelial cells, and its in vivo physiological roles. A subsequent study showed that interdomain LuxP mimicry is not unique to mammals: the yeast Saccharomyces cerevisiae produces 4-hydroxy-5-methylfuran-3(2H)-one (MHF), a LuxP agonist whose synthesis requires the enzyme Cff1p (Valastyan et al. 2021).

The identity question was answered in 2026. Shine, Valastyan, Ying, Huang, Seyedsayamdost and Bassler developed a reactivity-based metabolomics strategy that derivatizes the alpha-diketone (two adjacent carbonyl) moiety of DPD-like molecules with an o-diaminobenzene reagent, which abolished the AI-2 activity of both DPD and the Caco-2 mimic preparation, implying the mimic contains an alpha-diketone. Liquid chromatography-mass spectrometry (LC-MS) analysis of the trapped species, combined with chemical synthesis of both enantiomers, identified L-xylosone - a rare, highly reactive oxidation product of xylose metabolism, previously unknown in the human metabolome - as the mimic in nutrient-stressed Caco-2 preparations, with L-xylulose, an intermediate of the human glucuronate-xylulose pathway, as an additional active mimic. L-xylosone activates the V. harveyi reporter with a half-maximal effective concentration (EC50) of about 2 micromolar (the D-enantiomer is inactive), binds LuxP with micromolar affinity, and activates a V. cholerae AI-2 reporter; epithelial production titers (tens of micromolar) exceed these EC50 values and are far above measured luminal DPD concentrations in mice (0.07-0.21 micromolar). Ascorbic acid, the classical precursor of xylosone via spontaneous oxidative degradation, is absent from the media, so L-xylosone must arise by an unknown non-ascorbate route - candidate mammalian enzymes of the glucose-methanol-choline (GMC) oxidoreductase family (the choline dehydrogenase CHDH and the uncharacterized CHDH-like protein B4DMQ4) or spontaneous oxidation of monosaccharides were only hypothesized. The authors explicitly did not confirm that L-xylosone/L-xylulose are the active compounds under aerolysin exposure or tight-junction disruption, could not exclude further mimics lacking alpha-diketones, and did not revisit the Tkt/HldE detection requirement, the remaining bacterial stimulatory factors, or any in vivo function.

Problem Statement

The 2016 discovery that mammalian epithelial cells produce an AI-2 mimic, and the 2026 identification of L-xylosone and L-xylulose as LuxP-agonizing epithelial mimics, define a new host-to-bacteria signaling channel whose mechanism and function remain unexplained. The open problem is to complete the mechanistic and physiological account of this channel - that is, to explain how epithelial cells make the mimic, how bacteria elicit its production, how bacteria detect it, and what it does in vivo. Concretely, the following interlocking questions, all posed or acknowledged by the source work, constitute the surviving core: (1) Biosynthesis and regulation - what is the mammalian, non-ascorbate biosynthetic route to L-xylosone (and L-xylulose) in epithelial cells, how is production triggered by tight-junction disruption and by bacterial exposure, and is L-xylosone/L-xylulose confirmed to be the active mimic under aerolysin- or bacteria-stimulated conditions, not only nutrient stress? (2) Bacterial elicitation - which bacterial factors beyond aerolysin (apt) and its regulator VIBHAR_02470 stimulate host mimic production, given that aerolysin-independent stimulation demonstrably exists? (3) Detection mechanism - why do Vibrio harveyi mutants lacking transketolase (Tkt) or the heptose-biosynthesis enzyme HldE fail to detect the mimic although LuxP is the receptor; what Tkt- or HldE-dependent processing or bacterial physiology enables LuxP-mediated recognition of the mimic? (4) In vivo function - what physiological roles, if any, does host-derived AI-2 mimic signaling play in vivo, for example during epithelial repair, in shaping commensal versus pathogenic associations, or in modulating gut community composition? The four parts of this biosynthetic, elicitation, detection, and in vivo account are jointly posed and together address how and why mammalian epithelia speak to bacterial quorum-sensing systems: the biosynthetic pathway and its regulation (including which mimic mediates activity under aerolysin- and bacteria-stimulated conditions, not only nutrient stress), the remaining bacterial elicitors, the Tkt/HldE-LuxP detection relationship, and the in vivo physiological role.

References

  1. Ismail AS, Valastyan JS, Bassler BL. A Host-Produced Autoinducer-2 Mimic Activates Bacterial Quorum Sensing. Cell Host & Microbe 2016;19(4):470-480. DOI: 10.1016/j.chom.2016.02.020. https://doi.org/10.1016/j.chom.2016.02.020
  2. Shine EE, Valastyan JS, Ying VY, Huang JZ, Seyedsayamdost MR, Bassler BL. Discovery of a Human Metabolite That Mimics the Bacterial Quorum-Sensing Autoinducer AI-2. Journal of the American Chemical Society 2026;148(6):6258-6269. DOI: 10.1021/jacs.5c18527. https://doi.org/10.1021/jacs.5c18527
  3. Valastyan JS, Kraml CM, Pelczer I, Ferrante T, Bassler BL. Saccharomyces cerevisiae Requires CFF1 To Produce 4-Hydroxy-5-methylfuran-3(2H)-One, a Mimic of the Bacterial Quorum-Sensing Autoinducer AI-2. mBio 2021;12(2):e03303-20. DOI: 10.1128/mBio.03303-20. https://doi.org/10.1128/mBio.03303-20
  4. Zhang C, Jin X, Shi Y, Yu F, Wu Y, Call DR, Zhao Z. ArcB initiates quorum sensing to regulate T3SS in Vibrio alginolyticus by recognizing bacterial and host-derived autoinducer-2 as a kinase. Cell Communication and Signaling 2025;23:245. DOI: 10.1186/s12964-025-02258-0. https://doi.org/10.1186/s12964-025-02258-0
  5. Chen X, Schauder S, Potier N, Van Dorsselaer A, Pelczer I, Bassler BL, Hughson FM. Structural identification of a bacterial quorum-sensing signal containing boron. Nature 2002;415(6871):545-549. DOI: 10.1038/415545a. https://doi.org/10.1038/415545a
  6. Thompson JA, Oliveira RA, Djukovic A, Ubeda C, Xavier KB. Manipulation of the Quorum Sensing Signal AI-2 Affects the Antibiotic-Treated Gut Microbiota. Cell Reports 2015;10(11):1861-1871. DOI: 10.1016/j.celrep.2015.02.049. https://doi.org/10.1016/j.celrep.2015.02.049
  7. Oliveira RA, Cabral V, Torcato I, Xavier KB. Deciphering the quorum-sensing lexicon of the gut microbiota. Cell Host & Microbe 2023;31(4):500-512. DOI: 10.1016/j.chom.2023.03.015. https://doi.org/10.1016/j.chom.2023.03.015