Structures and broad-spectrum growth-inhibiting activity of formomarinobactin, formylated marinobactin analogs from the Pseudomonas lutea group

Abstract

Pseudomonas graminis LMG 21661^T, an environmental strain of the P. lutea group, produces the siderophore formomarinobactin, a novel marinobactin-like siderophore. Mass spectrometry revealed that formomarinobactin shares the same six-residue peptide backbone as marinobactin but contains formylated rather than acetylated N-hydroxyornithines. Alike marinobactins, formomarinobactins are produced as a suite of siderophores with a conserved hexapeptide core but varying lipid tail lengths (C10–C14), shorter than the C12–C18 characteristic of marinobactins. Both the biosynthesis and cognate receptor genes of the formomarinobactin system in P. graminis are iron regulated but unaffected by zinc or nickel underscoring their role in iron homeostasis. Genome mining combined with mass analyses demonstrated that formomarinobactin production is a conserved trait across the P. lutea group, with one exception which appears to represent an intraspecific cheater that has lost siderophore production. Beyond the producing strains themselves, we identified a widespread distribution of putative formomarinobactin receptors among diverse Pseudomonas species, revealing a substantial capacity within the P. fluorescens lineage to pirate formomarinobactin as an iron source. Putative receptors were also found in genera outside the Pseudomonas genus. Growth stimulation assays confirmed functional formomarinobactin uptake in several Pseudomonas spp. and a Phytopseudomonas strain, with genetic validation in Pseudomonas rhodesiae. Importantly, formomarinobactin production confers more than a nutritional advantage. Members of the P. lutea group producing formomarinobactin display pronounced growth inhibiting activity against a broad spectrum of clinical and environmental Gram-positive and with lower efficacy against Gram-negative bacteria. Purified formomarinobactin was able to inhibit growth under iron-limiting conditions and to a lesser extent in iron-rich conditions, highlighting a dual role for this molecule in both iron acquisition and microbial growth inhibition.

1. Introduction

Pseudomonas constitutes a group of ubiquitous Gram-negative bacteria renowned for their metabolic versatility and capability to colonize a wide range of environments (Silby et al., 2011). Species have been isolated from diverse terrestrial (plants, insects, soil, fruits), aquatic (seawater, freshwater) and host-associated (human tissues) environments (Butaitė et al., 2017; Girard et al., 2021, 2023; Mulet et al., 2021; Silby et al., 2011). The genus increased exponentially with the description of 70 novel species in 10 years and over 400 species are validly published in the ICNP to date, excluding synonyms and misspellings (Freese et al., 2025, accessed November 20th, 2025).

To adapt to different natural environments in iron-limited conditions, Pseudomonas notably rely on specialized metabolites known as siderophores (Cornelis, 2010). Indeed, iron availability is scarce due to low solubility of natural Fe³⁺ under aerobic conditions and neutral pH (Andrews et al., 2003). The most extensively studied siderophores of the genus are pyoverdines, high iron affinity molecules produced by fluorescent Pseudomonas species that play a key role in iron acquisition as well as virulence, and microbial competition (Cornelis et al., 2023; Meyer et al., 1996; Meyer, 2000). These Non-Ribosomal Peptide Synthetases (NRPS) regulated molecules chelate ferric iron (Fe³⁺) with strong specificity and are transported into the cell via dedicated outer-membrane receptors, enabling bacteria to thrive in iron-limited environments (Cornelis, 2010; Ghssein and Ezzeddine, 2022; Ravel and Cornelis, 2003).

However, non-fluorescent Pseudomonas species have also developed iron chelating strategies by producing alternative, usually lower affinity, siderophores (Cornelis and Matthijs, 2002). Although these systems have been less extensively characterized, they also play significant roles in ecological fitness, metal homeostasis, and interactions with plants and other microbes.

Pseudomonas graminis, a member of the non-fluorescent P. lutea group (G), was initially isolated from the phyllosphere of grasses (Behrendt et al., 1999; Peix et al., 2004) and later from apples (Alegre et al., 2013a) and cloud water (Besaury et al., 2017; Wirgot et al., 2019). Strains CPA-7 and 49 M have been consistently studied for their use in the preservation of fresh cut fruits, i.e., apples, pears and melon, or in planta, where they have shown varying growth inhibition activity against Escherichia coli O157: H7, Salmonella Michigan, Listeria innocua and Erwinia amylovora (Abadias et al., 2014; Alegre et al., 2013a; Alegre et al., 2013b; Collazo et al., 2018; Iglesias et al., 2018; Mikiciński et al., 2016). Investigating the mechanisms involved in their antagonism raised hypotheses concerning the involvement of siderophores (Mikiciński et al., 2016) or the reliance on mere carbon competition (Collazo et al., 2017). Further genome sequencing of strain UASWS1507 isolated from an apple tree in Switzerland (Crovadore et al., 2016), PDD-13b-3 from cloud water in France (Besaury et al., 2017), and CPA-7 from apples (Collazo et al., 2025), and analysis of the latter two genomes, highlighted the presence of a siderophore cluster which was not successfully annotated or elucidated by the authors. Nonetheless, Besaury et al. suggested toward pyoverdine production, however P. graminis is non-fluorescent so likely produces a non-fluorescent type siderophore (Meyer et al., 2002).

The aim of our research was to characterize, both in structure and biological activity, iron-regulated molecules produced by Pseudomonas graminis. We were able to identify a NRPS biosynthetic gene cluster in silico and determine the structure of the suite of siderophores produced by LC-HRMS. Our comparative genome and LC–MS analyses with other strains of the P. lutea G indicated its presence in these strains, associated with a functional production and a dedicated receptor. The suite of siderophores produced was shown to correspond to novel lipopeptides differing in lipid tail length, named formomarinobactins. They exhibited both iron chelating and growth inhibiting activities against Gram-positive bacteria and, to a lesser extent, Gram-negative environmental and clinical strains. These findings suggest that P. graminis and members of the P. lutea G, though non-fluorescent, produce a high affinity siderophore involved in their colonization of environmental habitats.

2. Materials and methods

2.1. Strains and growth conditions

All the strains used in this study are reported in Table 1. Pseudomonas strains were cultured in iron-poor Casamino Acids medium (CAA) at 28 °C. The medium was prepared with 5 g/L Bacto Casamino Acids (Difco Laboratories), 0.9 g/L K₂HPO₄, 0.25 g/L MgSO₄.7H₂O. Target organism strains were cultured in a rich (not iron depleted) in-house medium 856 at 37 °C (Matthijs et al., 2013) for the in vitro antagonism assays. To obtain solid medium for both CAA and 856, agar was added at a final concentration of 15 g/L. For the soft agar overlay, 856 medium was supplemented with agar at 7.5 g/L. For 2,2′-bipyridine plates, 2,2′-bipyridine was directly added to melted CAA cooled down to 60 °C at optimal concentrations according to each strain studied.

Table 1.

List of strains and plasmids used in this study.

Strain/plasmid	Biological origin, characteristics	Reference strain
Strains of the P. lutea G
P. graminis LMG 21661T	Grasses, phyllosphere, produces formomarinobactin	BCCM/LMG bacteria collection
P. graminis P200/02	Grasses, phyllosphere, produces formomarinobactin	Behrendt et al. (1999)
P. graminis P265/09	Grasses, phyllosphere, produces formomarinobactin	Behrendt et al. (1999)
P. graminis P294/05	Grasses, phyllosphere, produces formomarinobactin	Behrendt et al. (1999)
P. graminis P334/03	Grasses, phyllosphere, produces formomarinobactin	Behrendt et al. (1999)
Pseudomonas sp. W2Oct36	River water, produces formomarinobactin	Matthijs et al. (2013)
Pseudomonas sp. W2-17	River water, produces formomarinobactin	This study
Pseudomonas sp. W5-01	River water, produces formomarinobactin	This study
Pseudomonas sp. W5-36	River water, produces formomarinobactin	This study
P. lutea LMG 21974T	Grasses, rhizospheric soil, produces formomarinobactin	BCCM/LMG bacteria collection
P. bohemica LMG 30182T	Engraver beetle (Ips acuminatus), does not produce formomarinobactin but has a putative formomarinobactin receptor	BCCM/LMG bacteria collection
P. abietaniphila LMG 20220T	Aerated lagoon of bleached kraft pulp mill effluent, produces formomarinobactin	BCCM/LMG bacteria collection
In vitro antagonism target organism
Staphyloccocus capitis I	Isolated from the skin of a young and healthy individual	This study, gift from S. Flahaut, ULB
Bacillus zanthoxylo RSN02	Isolated from a cave	This study, gift from R. Benayad, ULB
Enterococcus hirae LMG 10274	Unknown source	BCCM/LMG bacteria collection
Pseudomonas aeruginosa LMG 1242T	Type strain. Unknown source	BCCM/LMG bacteria collection
Enterobacteriaceae sp. DV 4951	Isolated from patient with diarrhea	This study, gift from J. P. Hernalsteens, VUB
Candida albicans IHEM 3731	Yeast. Unknown source	BCCM/IHEM collection
Formomarinobactin growth stimulation test organisms (have a predicted receptor)
P. rhodesiae LMG 17764T	Wild-type, produces pyoverdine PYOrho and triabactin	BCCM/LMG bacteria collection
PrhoTΔpvdL-3	Pyoverdine-negative mutant of LMG 17764T, in frame deletion of the pvdL gene	This study
PrhoTΔpvdLΔtrbABC-5 (= PrhoT-KO)	Triabactin-negative mutant of PrhoTΔpvdL-3, in frame deletion of the trbABC operon	This study
PrhoTΔpvdLΔtrbABCΔTWR49273-159 (=PrhoT-KOΔTWR49273-159)	TWR49273-negative mutant of PrhoT-KO, in frame deletion of the formomarinobactin receptor TWR49273	This study
P. simiae WCS417r-M634	Pyoverdine-negative Tn5 mutant of spontaneous rifampicin-resistant mutant of WCS417r. Predicted to produce mupirochelin	Duijff et al. (1993), Grosse et al. (2023)
P. capeferrum WCS358r-JM213	Pyoverdine-negative Tn5 mutant of spontaneous rifampicin-resistant mutant of WCS358r	Marugg et al. (1985)
Pseudomonas sp. WCS374r-BT1	Pyoverdine-negative Tn5 mutant of mutant WCS374r4B1 (pmsB disrupted, blocking salicylic acid and pseudomonine biosynthesis). Has also an aerobactin-like cluster	De Vleesschauwer et al. (2008), Djavaheri et al. (2012)
P. fluorescens SBW25-15F3	Ornicorrugatin-negative Tn5 mutant of the pyoverdine negative mutant PBR840 of SBW25, GmR	Moon et al. (2008)
P. kuykendallii LMG 26364ᵀ	Wild-type	BCCM/LMG bacteria collection
Phytopseudomonas argentinensis LMG 22563ᵀ	Wild-type	BCCM/LMG bacteria collection
Plasmids
pUK21	Cloning vector; lacZα, KmR	Vieira and Messing (1991)
pME3087	Suicide vector, ColE1 replicon, RK2-Mob, TcR	Voisard et al. (1994)
pME-pvdL	pME3087 containing the 5′ and 3′ end of pvdL of LMG 17764T, TcR	This study
pME-trbABC	pME3087 containing the 5′ and 3′ flanking regions of trbABC of LMG 17764T, TcR	This study
pME-TWR49273-7	pME3087 containing the 5′ and 3′ flanking regions of TWR49273.1 of LMG 17764T, TcR	This study

BCCM/LMG Belgian co-ordinated collections of micro-organisms; IHEM Fungi, Human and Animal Health Catalogs. DV 4951 was received as E. coli; however, 16S rRNA gene sequencing identified it only as a member of the family Enterobacteriaceae, with highest similarity to sequences from Enterobacter and Salmonella species, and thus it could not be reliably assigned to a single genus.

2.2. In silico analysis of Pseudomonas graminis LMG 21661^T siderophore gene cluster

Siderophore gene cluster analyses were conducted using BLAST,¹ PKS/NRPS analysis,² Interpro³ and the NRPS/PKS module analysis feature implemented in antiSMASH.⁴

2.3. RT-qPCR assay of Pseudomonas graminis LMG 21661^T grown in different iron conditions

RT-qPCR analyses were performed to determine whether the gene cluster identified in silico was expressed and regulated by iron. To do so, P. graminis LMG 21661^T was grown overnight in liquid CAA medium and 20 mL of the preculture, adjusted to OD₆₀₀ = 0.2, was inoculated in 180 mL of liquid CAA and incubated at 28°C under agitation at 200 rpm, till the OD₆₀₀ reached 0.2. The culture was divided into six subcultures to study the effect of increasing iron concentrations. Consequently, the subcultures were supplemented with FeCl₃ (Merck), ZnCl₂ (Fisher Chemical) or NiCl₂ (Fluka) to a final concentration of 0, 0.5, 1, 2, 4, and 8 μM. A refined concentration gradient of 0, 0.05, 0.1, 0.2, 0.5, and 0.7 μM FeCl₃ was tested. These supplemented subcultures were then incubated for 1 h at 28 °C (200 rpm). Thereafter, the subcultures were centrifuged, resuspended in fresh liquid CAA and RNAprotect Bacteria reagent (Qiagen) was added. The total RNA was extracted with a RNeasy Mini kit (Qiagen). The samples were treated with RNase-Free DNase Set (Qiagen) to eliminate presumptive genomic DNA contamination. The integrity and purity of RNA were assessed by electrophoresis on a 1% Agarose gel and using a NanoDrop DeNovix DS-11 spectrophotometer (DeNovix Inc., Wilmington, DE, United States). Reverse transcription was done with 1 μg of RNA for each sample with RevertAid H Minus First Strand cDNA synthesis kit with oligo(dt) primers (Thermo Scientific). The cDNA was purified by means of the QIAquick PCR Purification kit (Qiagen). Quantification of the cDNA was conducted using Fast start Essential DNA Green Master (Roche Applied Science) and the appropriate primers (Supplementary Table S1) on a LightCycler 96 (Roche). The studied genes were fmbD, the first NRPS gene, and fmbG, the receptor gene. The quantity of DNA was normalized against the expression of nadB (L-aspartate oxidase), chosen as reference gene. Genomic DNA was extracted from LMG 21661^T using the DNeasy Blood and Tissue Kit (Qiagen) of a culture grown overnight in 856 media. A standard curve was generated using five different dilutions of genomic DNA and was used to quantify samples. RT-qPCR analyses were conducted at least in triplicate.

2.4. LC–MS analysis of a siderophore suite produced by Pseudomonas graminis LMG 21661^T

Siderophore production was evaluated by CAS assay (Schwyn and Neilands, 1987). P. graminis LMG 21661^T was analyzed by LC–MS to determine the nature of the siderophore. The strain was grown for 48 h at 28 °C in 100 mL of CAA. Subsequently, the cultures were centrifuged for 15 min at 5861 g (Sorvall Lynx 6000, Thermoscientific). The filtered supernatant (0.2 μM, Millisart Satorius filter, Stedim Biotech) was applied to a CHROMABOND C18 ecf column (Macherey-Nagel). Elution of the compounds of interest was performed with acetonitrile/H₂O (70:30 v/v). The purified fractions were analyzed by LC–MS using an Alliance e2695 separation module (Waters) coupled to an Altus SQ mass detector (Perkin Elmer). The method used consisted of solvents A and B, water and acetonitrile, respectively, both containing 0.1% formic acid. Separation was achieved on a C18 Altima column (250 × 4.6 mm, 10 μm, Grace), at a flow rate of 0.5 mL/ min, with a gradient of 10% B to 70% B in 30 min followed by 5 min at 100% B. This was followed by equilibration for 5 min at 10% B. The samples were analyzed by ESI in positive mode. The MS parameters were cone voltage 49 V, capillary voltage 3.1 kV, source temperature 150 °C, gas desolvation temperature 600 °C. Full scan m/z was acquired between 400 and 1,800 with a scan rate of 1,200 Da/s.

2.5. Large scale purification of siderophores produced by Pseudomonas graminis LMG 21661^T

The siderophore variants observed by LC–MS were further purified on a large scale. Thirty milliliters (10 × 3 mL) of a P. graminis LMG 21661^T overnight preculture grown in 100 mL CAA were used to inoculate 10 L (10 × 1 L) of Minimal Succinate Medium (Vindeirinho et al., 2021). The flasks were then incubated for 72 h at 28 °C under agitation (180 rpm). The cultures were centrifuged at 4,500 rpm for 30 min (Sorvall RC 12BP Plus, Thermo Scientific). The supernatants were pooled and passed through a CHROMABOND C18 ecf column (Macherey-Nagel) (2.5 × 4 cm). Salts were removed by washing the column with 200 mL of mQ water. The siderophores were then eluted with 100 mL of 70% acetonitrile. The fraction obtained was filtered (0.2 μM, Millisart), evaporated then lyophilized. The purification step was repeated 4 times. The semi-purified fraction was further purified using a SunFire Prep C18 column (19 × 250 mm, 5 μM, Waters) connected to an LC Prep 150 system (Waters) equipped with a UV/Vis 2489 detector (Waters). The solvents used were A, water, and solvent B, methanol. The method was a gradient set at a flow rate of 12 mL/min from 40% B to 90% B in 12 min, followed by 4 min at 100% B then column re-equilibration at 40% B for 5 min. UV detection was performed at 220 nm. Automated fraction collection using a Waters Fraction Collector III Collection was set at 80% tube volume between 3 and 15 min. Fractions were analyzed by LC–MS according to an adapted version of the method described in section (LCMS analysis of siderophores) and pooled together according to similarities in chromatogram profiles (same m/z ion and retention time).

2.6. Structural determination of purified siderophore of Pseudomonas graminis LMG 21661^T by LC-ESI-qTOF

The structure of the purified siderophores was further investigated using an LC 1200 (Agilent) chain coupled to a QTOF-6520 (Agilent). Chromatographic separation was conducted on an C18 Altima Grace Column (250 × 4.6 mm, 10 μm, Grace) maintained at 30 °C. Solvents A and B consisted of water and acetonitrile, respectively, with 0.1% formic acid for both and set to a flow rate of 0.5 mL/min. The method started at 30% B held for 2 min, followed by a gradient of 30% B to 90% B in 25 min, 100% B was passed through the column for 4 min, then the column was brought back to initial conditions at 30% B in 1 min and maintained for 5 min. The full method duration was 37 min. The analysis was performed in positive ion mode with electrospray inlet parameters set to 325 °C, 7 L/min and 45 psig for gas temperature, drying gas flow and nebulizer, respectively. Capillary voltage was set to 4.5 kV, fragmentor to 175 V and skimmer to 65 V. Full scan HRMS spectra were recorded between m/z 100 and 1,000 and MS/MS from m/z 50 to 1,000. Targeted MS/MS fragmentation was performed by CID at 55 V for m/z 874.4177, 876.4379, 902.4542, 904.4655 and at 65 V for m/z 930.4878 and 932.5002, with an isolation window of 4 m/z. A second untargeted MS/MS analysis was conducted at 35 eV. MS raw data were converted to .mzML using MSConvert (ProteoWizard) and processed using MZMine 4.

2.7. Acid methanolysis of siderophores and GC–MS analysis of fatty acid methyl esters

Fatty acid methyl ester (FAME) analysis was performed by gas chromatography coupled with mass spectrometry (GC–MS). Rapid sample preparation by acid-catalyzed hydrolysis and methylation was adapted from the method described by Williams et al. (1975). Briefly, to 5 mg of siderophore (2.5 mg for molecule 1), 2.5 mL of acidified MeOH with 2% H₂SO₄ were added. The mixture was heated in a water bath at 80 °C for 4 h. After cooling, 2 mL of H₂O were added, followed by 1 mL of n-hexane (97%). The samples were vortexed and subsequently centrifuged at 4,000 rpm for 1 min. The upper hexane layer was used for GC–MS analysis and lower aqueous layer was used for LC–MS analysis. Molecule 6 (Formomarinobactin F) was not analyzed for lack of sufficient starting material.

Optimization and validation of the chromatographic program to achieve optimal separation of the various compounds were conducted using an SGE BPX-70 capillary column (30 m × 0.25 mm × 0.25 μm) and the Supelco® 37 Component FAME Mix standard solution. Analyses were carried out on an Agilent GC 7820A gas chromatograph equipped with a 5977E mass selective detector and an automatic sampler, all controlled by Agilent MassHunter software. The chromatographic conditions were as follows: injector temperature 250 °C (split ratio 50:1); injection volume 0.5 μL; helium as carrier gas at a constant flow rate of 0.8 mL min⁻¹. The oven temperature program was: 2 min hold at 100 °C, ramp at 5 °C min⁻¹ to 160 °C (hold 10 min), then ramp at 5 °C min⁻¹ to 200 °C and hold for 5 min, resulting in a total run time of 37 min. The MS detector operated in full scan mode (m/z 35–400). Compound identification was based on automated spectral matching with the NIST 2014 mass spectral database. Each proposed identification was subsequently verified to ensure consistent and reliable compound assignment, especially for analytes not present in the reference mixture.

2.8. Creation of a custom reference genome dataset for the Pseudomonas lutea G

Genome sequence data for Pseudomonas strains belonging to the P. lutea G were retrieved from the NCBI RefSeq database (query date: 5 December 2025) (Sayers et al., 2024). Genome quality was evaluated using CheckM v1.2.5 (Parks et al., 2015), retaining only genomes with ≥95% completeness and ≤5% contamination. In total, 46 genomes were selected, comprising 6 complete and 40 draft assemblies, and imported into a custom “PluteaG” database using BioNumerics v7.6.3 (Applied Maths, bioMérieux).

2.9. Genome sequencing of river isolates

The genomes of Pseudomonas sp. W2-17, Pseudomonas sp. W5-01, Pseudomonas sp. W5-36 and Pseudomonas sp. W2Oct36 were sequenced. Therefore, genomic DNA was isolated using a Maxwell® RSC Cultured Cells DNA Kit (cat.# AS1620) and a Maxwell® RSC Instrument (cat.# AS1620), after a prior enzymatic lysis step. The integrity and purity of the DNA was evaluated on a 1.0% (w/v) agarose gel and by spectrophotometric measurements at 234, 260 and 280 nm. A Quantus™ fluorimeter and a QuantiFluor™ ONE dsDNA System (Promega Corporation, Madison, WI, United States) were used to estimate the DNA concentration. Library preparation and genome sequencing were performed by SeqCoast Genomics (Portsmouth, United States). Library preparation was performed using an Illumina DNA Prep tagmentation kit and unique dual indexes. Paired- end sequence reads (2 × 150 bp) were generated using a NextSeq 2000 platform (Illumina Inc., San Diego, CA, United States). Quality check, trimming of the raw sequence reads and genome assembly (de novo) were performed using the Shovill v1.1.0 pipeline,⁵ which uses SPAdes v3.15.4 (Bankevich et al., 2012) as its core and aims at a sequencing depth of 150x. Contigs shorter than 500 bp were excluded from the final assembly. Quality check of the assembly was performed using QUAST v5.2.0 (Gurevich et al., 2013) and checkM v1.2.2⁶ (Parks et al., 2015). The draft genome sequences of Pseudomonas sp. W2-17, Pseudomonas sp. W5-01, Pseudomonas sp. W5-36 and Pseudomonas sp. W2Oct36 have been deposited at DDBJ/EMBL/GenBank under accession numbers JAUFNS000000000.1, JAXCIP000000000.1, JAXCIQ000000000.1 and JAUERW000000000.1, respectively. Annotation of open reading frames (ORF) was performed with the NCBI Prokaryotic Automatic Annotation Pipeline (PGAAP).

2.10. Phylogenomic and receptor phylogeny analyses

A phylogenomic tree of 46 genomes belonging to the P. lutea G was constructed using the Type (Strain) Genome Server TYGS.⁷

Phylogenetic analysis of formomarinobactin and marinobactin receptors from 134 selected strains was conducted using MEGA v12 (Kumar et al., 2024). Although over 500 receptor sequences were initially identified by BLAST, a selected set of representative sequences from reference strains was used to provide a tractable subset for analysis. The protein-based tree was inferred by Maximum Likelihood with the LG(+Freq) model (Le and Gascuel, 2008), selecting the highest log-likelihood topology (−31242.43) from Neighbor-Joining and Maximum Parsimony starting trees; branch support reflects 1,000 bootstrap replicates (Felsenstein, 1985). Rate heterogeneity across 874 aligned amino acid sites was modeled with a 5-category discrete Gamma distribution (+G, parameter = 0.6174) and 17.39% invariant sites (+I). A complementary genome-based tree was constructed via Neighbor-Joining on a GGDC v3.0 formula 2 distance matrix (Meier-Kolthoff et al., 2022) derived from NCBI RefSeq assemblies (O’Leary et al., 2016).

2.11. Selection and siderophore production of Pseudomonas lutea G strains

From an updated Pseudomonas collection (~1,000 strains) based on the one described in Matthijs et al. (2013) which is sequenced for the partial rpoD gene, strains belonging to the P. lutea G were identified based on clustering with type strains of this group. In addition to four P. graminis strains (P200/02, P265/09, P294/05 and P334/03) isolated concomitantly with the P. graminis type strain (Behrendt et al., 1999), four strains were obtained: Pseudomonas sp. W2-17, Pseudomonas sp. W5-01, Pseudomonas sp. W5-36 and Pseudomonas sp. W2Oct36. The latter four strains were isolated from a small Belgian river in year 2001 (W2Oct36, Matthijs et al., 2013) and 2006 (W2-17, W5-01 and W5-36, unpublished). Strains were characterized by CAS assay, and siderophores were purified and analyzed by LC–MS to determine the presence of formomarinobactins.

2.12. In vitro growth inhibition assay against gram-positive and negative bacteria

The strains of the P. lutea G were screened for their potential to produce iron repressed molecules capable of inhibiting the growth of a panel of Gram-positive and Gram-negative bacterial strains (Table 1). Along with P. graminis LMG 21661^T, 11 strains were used to evaluate the conservation of the novel siderophore within the group. Therefore, each Pseudomonas strain was grown overnight in 6 mL liquid CAA. The cultures were adjusted to 5 × 10⁹ cells, and 10 μL was inoculated in the center of 20 mL CAA plates and CAA plates supplemented with 50 μM FeCl₃. The plates were then incubated at 28 °C for 48 h after which the cells were killed using chloroform vapors for 20 min. A soft agar overlay (7.5 g/L agar) mixed with a target strain adjusted at 5 × 10⁶ cells in 856 media was prepared. After evaporation of the chloroform, six mL of the overlay was uniformly poured on each plate and incubated overnight at 37 °C. The assay was performed in triplicate for each condition.

The antimicrobial properties of the purified siderophores were evaluated against the same batch of test strains according to a slightly modified version of the previous assay. A soft agar overlay of each strain in 856 was prepared as described above and overlaid on 20 mL CAA and CAA + 50 μM iron plates. After solidification, a sterile cork borer was used to cut wells in the agar, which were then filled to the top of the well with 75 μL of siderophore fractions at 15 mg/mL in 70% MeOH. As control 70% MeOH was used. The plates were incubated overnight at 37 °C, depending on the strain. The assay was performed in triplicate.

2.13. Formomarinobactin growth stimulation assays of non-producer and mutant strains

To identify strains with a putative formomarinobactin receptor, BLASTP searches were performed using the marinobactin receptor sequence of P. graminis as query against genomes of the genera Pseudomonas, Ectopseudomonas, Stutzerimonas, Phytopseudomonas and Marinobacter. Representative receptors with an E-value of 0 and a query coverage of ≥ 90% were retained if they exhibited an amino acid identity of ≥ 60% (or ≥ 48% for Marinobacter). A representative of the clusteredNR results was chosen randomly for strains outside the P. lutea G. Strains of interest, such as strains available in our in-house collection, were added to the selected candidates.

Siderophore-negative mutants derived from five of these identified strains, P. simiae WCS417, P. capeferrum WCS358, Pseudomonas sp. WCS374, P. fluorescens SBW25 and P. rhodesiae LMG 17764^T were used for growth stimulation assays, in addition to P. bohemica LMG 30182^T, P. kuykendallii LMG 26364ᵀ and Phytopseudomonas argentinensis LMG 22563ᵀ (Table 1).

For the growth stimulation assays, pre-cultures of the strains grown overnight in CAA liquid medium were adjusted to 5 × 10⁶ cells/mL, then inoculated on 20 mL CAA plates supplemented with 2,2′-bipyridine. The optimal concentration of 2,2′-bipyridine was determined by testing a range from 100 to 600 μM, causing minimal background growth. A Whatman paper disk (Ø = 5 mm) was placed at the center of the plates, and 2 × 5 μL of purified siderophore was loaded onto it. Growth stimulation was assessed by recording growth halos of the mutant strains around the siderophore-loaded disks. Experiments were done twice.

2.14. Construction of a siderophore receptor mutant TWR49273 in Pseudomonas rhodesiae LMG 17764^T

To confirm that the siderophore receptors identified by BLAST analyses are formomarinobactin receptors, the putative receptor gene was deleted in the P. rhodesiae LMG 17764^T siderophore-negative mutant Prho^TΔpvdLΔtrbABC-44, hereafter referred to as Prho^TKO-44 (Brandt and Matthijs, unpublished results), which already carried clean deletions of biosynthetic siderophore genes, following a previously described method (Matthijs et al., 2016). In short, a markerless genomic deletion method by homologous recombination was used to achieve an in-frame deletion of the 11 kb internal fragment of the pvdL gene involved in biosynthesis of the chromophore of pyoverdine and of the almost complete trbABC operon of the siderophore triabactin (5 kb). Loss of pyoverdine and triabactin production were verified by LC–MS or CAS assay.

To delete the putative formomarinobactin TonB-dependent receptor, the 5′ and 3′ flanking regions were amplified and cloned in pUK21. After confirmation by sequencing the fragment was recloned in suicide pME3087, resulting in pME-TWR49273, and was integrated into the chromosome of Prho^T-KO-44 by triparental mating using E. coli HB101/pME497 as the mobilizing strain, with selection for tetracycline and chloramphenicol resistant recombinants. Excision of the vector through a second crossing-over event took place following the enrichment of tetracycline-sensitive cells (Schnider-Keel et al., 2000). The mutant obtained was confirmed by PCR amplification and subsequent sequencing of the fragments (using verification primers, Supplementary Table S2). The list of the primers used to amplify the fragments is given in Supplementary Table S2, the constructed plasmids are given in Table 1.

2.15. Statistical analysis

One-Way ANOVA followed by Dunnett’s multiple comparison test for growth inhibition assays of the P. lutea G strains and followed by post hoc Tukey correction for multiple comparisons were performed on growth inhibition assays of the purified formomarinobactins. A Brown-Forsythe and Welch ANOVA followed by Dunnett’s T3 test for multiple comparisons was done for RT-qPCR assays using Graphpad Prism version 8.0.1.

3. Results

3.1. In silico identification of an NRPS siderophore-mediated iron acquisition system in Pseudomonas graminis LMG 21661^T

Although P. graminis spp. do not produce pyoverdine (Behrendt et al., 1999; Butaitė et al., 2017; Peix et al., 2004), colonies of P. graminis LMG 21661ᵀ exhibited a yellow halo on CAS medium, suggesting the synthesis of an alternative siderophore. Genome analysis identified a putative gene cluster for a NRPS siderophore which resided on a 37.5 kb fragment and included 11 open reading frames (ORFs) (Figure 1). Predictions showed that the necessary information for the regulation, biosynthesis and transport of a NRPS siderophore were present in the gene cluster.

Figure 1.

Schematic presentation of the gene cluster of formomarinobactin in strains of the P. lutea G (A), the predicted marinobactin cluster of Ectopseudomonas spp. (B) and the marinobactin cluster of Marinobacter spp. (C). Gene names are indicated above the arrows. The organization of the modules and domains of the two NRPS, FmbD and FmbE, is shown on the arrows. Formo/marinobactin biosynthesis is predicted to begin with a N-terminal acyl AMP ligase starter domain, which activates a fatty acid and initiates biosynthesis and to end with a thioesterase domain. C: condensation domain, A: adenylation domain, T: thiolation, E: epimerization domain. Predicted genes functions obtained by BLASTp and InterPro analyses: fmbA: RNA polymerase sigma-70 factor, ECF subfamily, fmbB: putative ATP-binding cassette transporter, fmbC: lysine N(6)-hydroxylase/L-ornithine N(5)-oxygenase family protein, fmbD and fmbE: non-ribosomal peptide synthases, fmbF: diaminobutyrate-2-oxoglutarate transaminase, fmbG: TonB-dependent receptor, fmbH: MbtH family NRPS accessory protein, fmbI: TauD/TfdA family dioxygenase, fmbJ: N(5)-hydroxyornithine transformylase PvdF and fmbK: siderophore-iron reductase FhuF. The homologue of FmbJ in Ectopseudomonas and Marinobacter spp. is a GNAT-family N-acetyltransferase.

The two NRPSs were predicted to assemble the peptide backbone. The organization of the modules and domains is shown in Figure 1. The length and composition of the peptide were inferred based on the domains of the NRPS. Six classical NRPS modules (C-A-T) were found, two in FmbD and four in FmbE, indicative of a 6-residues long peptide backbone. Adenylation domain specificity and the stereochemistry (D and L-forms, based on epimerization domains) were predicted (Supplementary Table S3), resulting in an Asp – Dab – Ser – (Fo-)OH-Orn – Ser – (Fo-)OH-Orn backbone (D-forms are underlined). In addition, a C-terminal thioesterase (TE) domain which catalyzes the release of the NRPS product was identified in the last protein FmbE. The gene cluster also encodes a MbtH-like protein (FmbH) which is thought to assist many NRPSs by promoting proper folding, stability, and activity of the synthetases (Felnagle et al., 2010).

The first domain of the predicted FmbD protein presented homology with members of the fatty acyl-AMP ligase family (FAAL, cd05931, e = 0) (Figure 1). This domain may activate and introduce long chain fatty acids in both polyketide and NRPS peptide biosynthesis by activation and loading of fatty acids to ACP domains (Patil et al., 2021). The presence of this domain led to the hypothesis that the siderophore is amphiphilic.

Two non-proteinogenic amino acids, 2,4-diaminobutanoic acid (Dab) and (formylated) N5-hydroxyornithine [(Fo-)HO-Orn], were proposed as components of the peptide of the siderophore. L-Dab is probably synthesized from aspartate ß-semialdehyde by diaminobutyrate-2-oxoglutarate transaminase (FmbF). L-ornithine N5-oxygenase (FmbC) converts L-ornithine to N₅-hydroxyornithine. Although the software had difficulty differentiating between non-formylated and formylated hydroxyornithine, the presence of an N(5)-hydroxyornithine transformylase PvdF (FmbJ) in the cluster suggests the hydroxyornithine is likely in the formylated state (McMorran et al., 2001; Ravel and Cornelis, 2003). These modifications generate hydroxamate groups that chelate Fe(III) via oxygen atoms, creating stable iron complexes. Formylated N⁵-hydroxyornithines are characteristic for pyoverdines and contribute strongly to iron chelation (Meyer, 2000). An additional coordination site is probably a hydroxy-carboxylic group formed by the hydroxylation of aspartate, catalyzed by the aspartyl-β-hydroxylating tailoring enzyme FmbI. The last gene, fmbK, coded for a protein that is homologous to FhuF, the putative iron-ferrioxamine B reductase in E. coli, which is proposed to reductively release ferrous iron from ferric-siderophore complexes. FmbK is thus likely involved in the release of iron from internalized ferric-siderophore that then becomes available for cellular use. Finally, the uptake of the ferric-siderophore complex probably occurs through the TonB-dependent receptor FmbG.

Overall, the gene cluster indicates production of a marinobactin-like siderophore, with an N(5)-hydroxyornithine transformylase replacing a GNAT-family N-acetyltransferase (Figure 1), predicting a formylated rather than acetylated siderophore.

3.2. Siderophore gene cluster transcription is regulated by iron availability

To evaluate if the fmb genes from P. graminis LMG 21661^T are functional and regulated by iron, expression of the first NRPS gene fmbD and the TonB-dependent receptor fmbG was verified by means of RT-qPCR under iron limited and iron-rich conditions. A preliminary assay of six bacterial cultures supplemented with increasing iron concentrations (0, 0.5, 1, 2, 4, and 8 μM FeCl₃) showed that the fmbD and fmbG genes were expressed in the absence of iron and repressed by low amounts of Fe³⁺ in the medium. Indeed, exposure to 0.5 μM resulted in complete repression of fmbD expression and an approximately one-third reduction in fmbG expression (data not shown). Iron concentrations were refined (0, 0.05, 0.1, 0.3, 0.5, 0.7 μM FeCl₃) to assess more precisely the iron-dependent regulation of gene expression (Figure 2). A concentration of 0.3 μM FeCl₃ significantly reduced fmbD transcription (p-value < 0.05), the first NRPS gene while the receptor transcription decreased by approximately one third, with complete suppression observed at 0.7 μM FeCl₃ (p-value < 0.01). These results indicate that the cluster is iron regulated. The effect of zinc and nickel on the expression of fmbD and fmbG was also tested, no effect was observed (Supplementary Figure S1) excluding a role as zincophore or nickelophore for formomarinobactin.

Figure 2.

Relative expression of fmbD and fmbG genes with increasing Fe concentrations normalized against reference gene nadB. Experiments were repeated six times. Values correspond to mean ± SEM. A Brown-Forsythe and Welch ANOVA analysis followed by Dunnett’s T3 test for multiple comparisons was done. * p <0.05, ** p <0.01.

3.3. Structure of marinobactin-like siderophore suite produced by Pseudomonas graminis LMG 21661^T

Six peaks were successfully purified as separate molecules by preparative HPLC, and their purity was assessed by LC–MS.

A LC-HRMS full scan analysis was conducted to obtain the precise mass of each molecule. These were determined as m/z 874.4177 (F_A), 876.4379 (F_B), 902.4542 (F_C), 904.4655 (F_D), 930.4878 (F_E) and 932.5002 (F_F) for the [M + H]⁺ ions (Figure 3). According to their m/z values and close retention times, they were hypothesized to be a suite of siderophores. Furthermore, mass differences of 2 Da between m/z 874 and 876, 902 and 904 and, 930 and 932, seemed to suggest 2H corresponding to an unsaturation while the mass differential between m/z 876 and 902 as well as 904 and 930 of 26 Da, are probably the result of C₂H₂, so methylene groups. The mass spectrum consisted of the [M + H]⁺ monoisotopic ion as well as the doubly charged [M + 2H]²⁺ ion (Supplementary Figure S2). Finally, the molecular formulae of the siderophores were also determined by considering isotopic distribution.

Figure 3.

LC-HRMS chromatogram of semi-purified formomarinobactin-rich fraction in positive mode. Six peaks represent m/z values of each formomarinobactin variant.

The LC–MS/MS analysis showed similar fragmentation patterns (neutral losses) for all siderophores at both 35 eV and 55 eV (Supplementary Figure S3) indicating similar structures. Moreover, fragmentation at 55–65 eV, involved the loss of a water molecule (18 Da), then CO₂ (44 Da) characteristics of loss of the carboxylic group on the C-terminal. This is followed by the loss of a fragment between 152 Da (for m/z 874 [M + H]⁺) and 210 Da (for m/z 932 [M + H]⁺) from the N-terminal resulting in a common m/z 660 Da fragment. Also, the same fragmentation peaks were observed below m/z 264/265 denoting a common core (Supplementary Figure S3) and enabling the elucidation of the amino acids present in the peptide sequence. Many of these peaks are shared with marinobactins, namely m/z 86, 150, 170, 211, 219, 237, 255 et 265 (Reitz et al., 2025; Zhang, 1999). In fact, main differences with the fragmentation of marinobactins concerned modifications of the ornithine residue, instead of a loss of 190 and 172 Da for N-acetyl N-hydroxyornithine and N-acetyl N-hydroxyornithyl, respectively, losses of 176 and 158 Da were observed corresponding to N-formyl N-hydroxyornithine and an N-formyl N-hydroxyornithyl residue, respectively (Figure 4). The formylated ornithine residues were alternated with serine (seryl groups), observed by a loss of 87 Da, similarly to marinobactins, followed by the loss of diaminobutyric acid (100/101 Da). As described by Zhang (1999), fragments corresponding to the loss of Dab and Asp/Asp(OH) as well as the fatty acid tail are less evident, especially in the formomarinobactin variants of higher masses. For this reason, GC–MS analyses were also performed on the lipid chains (Supplementary Figure S4). Nonetheless, fragmentation of m/z 904 [M + H]⁺ at 55 eV was strongly similar to that recorded for marinobactin A according to Reitz et al. (2025), suggesting that the molecules have highly similar structures. Marinobactin A has a m/z of 932 [M + H]⁺ in positive mode, the modification of its Ac(OH)Orn groups to Fo(OH)Orn, would lead to a m/z of 904. The results are consistent with the in silico analysis, predicting a marinobactin-like siderophore composed of a common peptidic backbone and a fatty acid chain of varying length (Figure 4). The presence of a hydroxamate-type siderophore was further confirmed by TLC analysis (Supplementary Figure S5).

Figure 4.

Formomarinobactin structure with superimposed GC-MS chromatograms of the lipid tails (A) and fragmentation of peptide chain from COOH terminal (Example with 874 m/z) (B). *Impurity corresponds to (1,1-dimethylethyl) benzene.

In light of the sole modification on the ornithine residue, we propose naming our series of siderophores formomarinobactins: m/z 874 Formomarinobactin A, m/z 876 Formomarinobactin B, m/z 902 Formomarinobactin C, m/z 904 Formomarinobactin D, m/z 930 Formomarinobactin E and m/z 932 Formomarinobactin F. GC–MS data combined with LC–MS/MS analyses confirmed the nature of the lipid tail (Supplementary Figure S4), formomarinobactin D (F_D) thus has a lauryl group (C12:0) for fatty acid tail, as also observed from the common fragment at 660 Da. F_B has a decanoic acyl group (C10:0) and F_F a tetradecanoic acyl group (C14:0). F_E has a myristoleic acyl group (C14:1 Δ9). F_A and F_C both carry an unsaturation as (C10:1 Δ4) and (C12:1 Δ5), respectively. All unsaturated lipid tails are in -cis form.

3.4. Formomarinobactins are the characteristic siderophores of the Pseudomonas lutea G of the Pseudomomas genus

The studied P. graminis strain belongs to the P. lutea G (P. fluorescens lineage), a taxonomic group that currently includes six described species: P. abietaniphila, P. bambusae (not validly publ.), P. bohemica, P. graminis, P. lutea and P. petrae (Figure 5). Phylogenomic analysis resolved the P. lutea G into two well-supported subclades, A and B, containing 13 and 11 (putative) species, respectively, suggesting 18 predicted novel species in addition to the six described. dDDH analysis indicated that P. bambusae may represent two subspecies (Figure 5); its dDDH value with Pseudomonas sp. SLFW was 72.3%, near the lower boundary of the commonly accepted species threshold.

Figure 5.

Phylogram generated using the TYGS platform (https://tygs.dsmz.de) (Meier-Kolthoff and Göker, 2019). Type strains are shown in red, genome-sequenced strains analyzed in this study are shown in blue, and orange triangles indicate experimentally confirmed formomarinobactin production.

Although P. huanghezhanensis shows its highest genomic similarity to members of the P. lutea G (Ge et al., 2024), the dDDH values are very low (<24.1%), and it forms a long-branched, genomically distant lineage sister to the P. lutea G (Figure 5). Due to this uncertain taxonomic position, P. huanghezhanensis was excluded from downstream analyses.

Genome mining revealed that all strains of the P. lutea G, apart from the type strain of P. bohemica (Figure 1), harbor a formomarinobactin biosynthetic gene cluster. In some genomes, this cluster is split, with the fmbH-K genes located distantly from the core biosynthesis genes, as observed, for example, in the P. abietaniphila type strain (Figure 1).

LC–MS analysis of the available type strains of the P. lutea G confirmed that P. graminis, P. lutea and P. abietaniphila exhibited six peaks with m/z 874.7, 876.7, 902.7, 904.7, 930.7 and 932.7. In addition, formomarinobactin production was detected in four additional P. graminis strains whose genomes have not been sequenced (P. graminis P200/02, P294/05, P265/09 and P334/03), as well as in four strains isolated from a small river (Pseudomonas sp. W2Oct36, W2-17, W5-01, and W5-36) (Table 1). These results, combined with genome mining, indicate that formomarinobactin production is conserved across the P. lutea G.

Outside the P. lutea G, a formomarinobactin gene cluster was identified in P. huanghezhanensis BSw22131, a pyoverdine-negative strain. The cluster is flanked by an IS3 family transposase in the intergenic region between the receptor and accessory genes, with a site-specific integrase, phage tail tape measure protein, and IS4-family transposase located upstream. These genomic features indicate recent mobility and suggest horizontal acquisition and genomic instability, which may facilitate biosynthetic gene loss or the generation of variants.

Putative marinobactin gene clusters, identified by the presence of a GNAT-family N-acetyltransferase and absence of N(5)-hydroxyornithine transformylase, are rare within the genus Pseudomonas. Only a single cluster was identified, in P. oryzihabitans USDA-ARSUSMAR, a pyoverdine non producer. In contrast, marinobactin clusters were detected outside the genus, including in Phytopseudomonas argentinensis SA190 and multiple Ectopseudomonas strains (Figure 1). Similar to the formomarinobactin cluster, and unlike Marinobacter, these gene clusters are self-sufficient for 2,4-diaminobutyrate (Dab) biosynthesis.

The marinobactin cluster has previously been investigated in Ectopseudomonas mendocina ymp (formerly Pseudomonas mendocina), where it was shown to contribute to iron acquisition; however, the chemical structure of the corresponding siderophore has not yet been elucidated (Awaya and DuBois, 2008).

3.5. Growth stimulation assays of putative formomarinobactin TonB-dependent receptors

BLAST analysis revealed that homologs of formomarinobactin/marinobactin TonB-dependent receptors, in the absence of the corresponding biosynthetic gene cluster, are widely distributed across the P. fluorescens lineage. Over 1,000 receptor homologs were identified over multiple genera. Because a full phylogenetic reconstruction was beyond the scope of this study, a representative set of 134 genomes was analyzed (Figure 6). Gene- and genome-based phylogenies were constructed to compare receptor relationships with overall genomic relatedness and tree congruence was qualitatively assessed.

Figure 6.

Comparative phylogeny (topology-only) of formomarinobactin and marinobactin receptors and host genomes. (A) Maximum Likelihood tree of 134 representative receptors (subset of >1,000 BLAST hits), inferred from 874 aligned amino acid sites using the LG(+F+G+I) model in MEGA v12 (1,000 bootstrap replicates; values >50% shown). (B) Neighbor-Joining tree of the same 134 strains based on GGDC v3.0 formula 2 intergenomic distances from NCBI RefSeq assemblies (Cellvibrio japonicus outgroup); no branch supports shown due to distance-matrix method. Color coding: formomarinobactin producers (green), marinobactin producers (blue), receptor-only strains (black). Yellow highlight: growth stimulation assay strains. Orange squares: topologically incongruent strains. Cell-surface signaling upstream of receptors in nonproducers: FecIR (Fec system σ⁷⁰+FecR), ECFσ–DUF4880 (ECFσ⁷⁰+DUF4880). Parentheses: partial genes; TR: siderophore transport genes.

The Maximum Likelihood receptor phylogeny resolved three producer-associated subclades (P1–P3) and two receptor-only subclades (R1–R2), each supported by bootstrap values >70% (Figure 6A). Subclade P1 comprised formomarinobactin producing Pseudomonas strains, whereas P2 and P3 contained marinobactin producers from Ectopseudomonas and Marinobacter, respectively. Within producer-associated subclades, several smaller receptor-only lineages were present. Clade R2 was composed of Pseudomonas spp., while R1 included Stutzerimonas, Aquipseudomonas and orphan Pseudomonas strains. Receptors in R1 and R2 were preceded by cell-surface signaling (CSS) genes (fecIR loci) or, in Stutzerimonas, by an ECFσ–DUF4880 module, whereas incongruent, or marinobactin-producing strains, lacked identifiable CSS systems.

The complementary GGDC-based Neighbor-Joining genomic tree revealed several taxonomic inconsistencies relative to the receptor phylogeny. For example, within the formomarinobactin producing P1 clade, nested the receptor-only P. baltica type strain (P. rhizosphaerae G) and the predicted formomarinobactin producing P. huanghezhanensis BSw22131^T (taxonomic position not clear) (Figure 6A). Adjacent to the P. lutea G were receptors from P. japonica 66C3, Pseudomonas sp. NPDC007930, P. typographi CA3A^T, and P. oryzihabitans USDA-ARS-USMAR, which were distantly related to the P. lutea G in the genome-based phylogeny (Figure 6B).

In the P. lutea G the core formomarinobactin biosynthetic gene cluster was located immediately upstream of the receptor gene, except in P. bohemica, which harbored only the receptor preceded by a truncated 3′ end of fmbE. Growth stimulation assays showed that P. bohemica grew in the presence of purified formomarinobactin (Figure 7A), confirming receptor functionality and substrate specificity.

Figure 7.

Formomarinobactin growth stimulation assays on CAA medium supplemented with the iron chelator 2,2′-bipyridine. (A) P. bohemica LMG 30182^T (500 µM), (B) Phytopseudomonas argentinensis LMG 22563^T (formerly classified as Pseudomonas argentinensis) (200 µM), (C) Siderophore-negative mutant Prho^TKO-159 of P. rhodesiae LMG 17764^T (400 µM) and (D) Prho^TKOΔTWR49273-159 (lacking the formomarinobactin receptor) (400 µM).

To test whether receptor-only strains outside the P. lutea G could utilize formomarinobactin as an iron source, growth stimulation assays were performed with selected receptor-only strains (highlighted in yellow in Figure 6) from the subclades P2, R1, and R2. Siderophore-deficient mutants of P. fluorescens SBW25, P. simiae WCS417, P. rhodesiae LMG 17764ᵀ, Pseudomonas sp. WCS374 (all P. fluorescens SG) and P. capeferrum WCS358 (P. putida G) (Supplementary Figures S6A–E), in addition to P. kuykendallii LMG 26364ᵀ (orphan) (Figure 7B) all grew around formomarinobactin-loaded disks, confirming active, functional receptors.

To directly demonstrate that the predicted TonB-dependent receptors mediate formomarinobactin uptake, the receptor gene was deleted in the siderophore-deficient mutant Prho^T-KO-44, yielding Prho^T-KOΔTWR49273-159. The receptor mutant failed to grow in the presence of formomarinobactin (Figures 7C,D), demonstrating that the receptor is essential for formomarinobactin-dependent iron uptake.

3.6. Iron-mediated growth inhibition of Pseudomonas lutea G strains against target microorganisms

The strains of the P. lutea G showed clear antagonistic activity against two Gram positive bacteria, S. capitis I and B. zanthoxylo RSN02 (Figures 8, 9; Supplementary Figure S7). The antibacterial activity observed was iron dependent as its supplementation at 50 μM in the medium resulted in the loss of inhibition zones. Two Gram negative strains, P. aeruginosa LMG 1242^T and Enterobacteriaceae sp. DV4951, were partially inhibited by the strains, visible as a reduced lawn density (Figures 8, 9; Supplementary Figure S7). This indicates that both target strains are stressed but can still grow in the presence of the exometabolites under the tested conditions. No inhibition was once again observed in iron-rich conditions. Conversely, P. bohemica LMG 30182^T, which does not produce formomarinobactin, was the sole exception in these growth inhibition assays; it showed no inhibition against any of the Gram-positive and Gram-negative bacterial strains, whether in iron-rich or iron-limited media (Figures 8, 9). The iron-dependent repression of the growth inhibition, together with its absence in P. bohemica supports the involvement of formomarinobactins in the observed growth inhibition.

Figure 8.

Antagonism of P. lutea G strains against clinical and environmental strains. Experiments were done in triplicate. Values correspond to mean ± SD. One-Way ANOVA followed by Dunnett’s multiple comparisons test. Significance level at p- value < 0.05 compared to LMG 21661^T for all P. lutea G strains against antagonistic bacterial strains except LMG 21974^T against E. coli (ns = non-significant).

Figure 9.

Growth inhibition of P. graminis LMG 21661^T (left column) and P. bohemica LMG 30182^T (right column) against the Gram-positive B. zanthoxylo RSN02 and the Gram-negative P. aeruginosa LMG 1242^T. Images are linearly adjusted for contrast (+20%) for presentation. The results of the growth inhibition against the other strains can be found in Supplementary Figure S7.

Finally, E. hirae was the only clinical strain against which none of the P. lutea G strains were active. Interestingly, none of the strains showed anti-yeast activity against Candida albicans, except for P. bohemica LMG 30182^T (7.5 mm).

Furthermore, in each assay, the inhibition zones of the wild-type, P. graminis LMG 21661^T, were significantly (p-value < 0.05) smaller than those of the other producer strains, highlighting either a lower yield of siderophore or the secretion of other iron-suppressed antibacterial substances by the P. lutea G strains (Figure 8).

3.7. Growth inhibiting potential of Pseudomonas lutea G strains is mediated by amphiphilic siderophores, formomarinobactins

The results obtained for the isolated molecules mirrored those of the strains with a clear growth inhibition against B. zanthoxylo RSN02 and S. capitis I (Figure 10). All 6 molecules showed a similar range of inhibition. The inhibition zones were greater in iron-poor than iron-rich media, however they were distinctly observed in both conditions. These findings further imply that the growth inhibition of the strains was mediated by iron-dependent molecules. The only exception was Enterobacteriaceae sp. DV 4951 which was not inhibited by the purified molecules, perhaps suggesting the presence of other iron-mediated anti-microbial substances or that their activity against DV 4951 was only significant in combination rather than as individual molecules.

Figure 10.

Growth inhibition of purified formomarinobactins F_A (m/z 874), F_B (m/z 876), F_C (m/z 902), F_D (m/z 904), F_E (m/z 930) and F_F (m/z 932) against clinical and environmental strains in iron-limited (left) and iron-supplemented (right) CAA medium. H = halo, for reduced lawn density; C = clear antagonism, for radius devoid of growth. Experiments were done in triplicate. Values correspond to mean ± SD. One-Way ANOVA followed by posthoc Tukey correction for multiple comparisons was applied and significance levels were defined as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Candida albicans and E. hirae were not inhibited by the molecules tested, confirming that the lack of growth inhibition observed in the presence of the P. lutea G strains was due to the lack of activity of the siderophores against these two strains.

Pseudomonas aeruginosa showed a unique inhibition pattern. Indeed, consistent with results from the growth inhibition of the strains, a reduced lawn density halo was observed in the presence of the different siderophores suggesting partial growth inhibition. Additionally, clear growth inhibition was observed for F_C – F_E (Supplementary Figure S8) with a small inhibition zone devoid of growth. This inhibition zone was also observed in the presence of iron, without the presence of partial growth. Such results highlight that the growth inhibition of formomarinobactins against P. aeruginosa follows a concentration gradient. Highly concentrated formomarinobactins are necessary to completely inhibit the growth of P. aeruginosa. Furthermore, in iron-rich environments, the siderophores with shorter acyl chains (F_A-F_D) were more active than the longer acyl chains. Similarly, in iron-poor media there was an inverted trend between the length of the acyl chain and the inhibition diameter between F_C to F_F. This observation implies that for P. aeruginosa both the chemical structure (acyl chain length) and iron availability influence the siderophore’s growth inhibiting effect. The dependence on acyl chain length could reflect differences in membrane interaction, uptake efficiency, or diffusion.

4. Discussion

Amphiphilic siderophores have, until recently, been rarely described in Pseudomonas. Corrugatin, the first such siderophore identified nearly three decades ago, consists of an eight amino-acid peptide linked to a short C8 fatty acid chain (Risse et al., 1998). Subsequently, ornicorrugatin and histicorrugatin were discovered (Matthijs et al., 2008, 2016), these differ by only one or two peptide residues, and we propose grouping them as the corrugatin family. These early examples occupy the hydrophilic end of the amphiphilic spectrum, due to their short fatty acid chains and longer peptide moieties. More recently, lactuchelin was reported as a series of seven-amino-acid cores acylated with C8-C16 chains (Chesneau et al., 2025). Biosynthetic gene clusters for lactuchelin were detected in the genomes of only a handful of Pseudomonas species (Chesneau et al., 2025).

Another suite of amphiphilic siderophores is here described for the grass isolate P. graminis LMG 21661^T which produces structural analogs of marinobactin previously described in Marinobacter sp. (Martinez et al., 2000; Martinez and Butler, 2007), to cope with iron limitation. These analogs share the same 6-residue peptide backbone as marinobactin but differ by formylation rather than acetylation of N-hydroxyornithine. In addition, the P. graminis formomarinobactin carries shorter hydrocarbon chains (C10 to C14) compared with the C12-C18 chains characteristic of marinobactin.

Marinobacter species are widely distributed in diverse marine and saline terrestrial environments, including low-temperature hydrothermal areas (Handley and Lloyd, 2013). Their production of marinobactin may establish a hydrophobicity gradient wherein longer, more hydrophobic siderophores remain near the cell surface, while shorter, more hydrophilic variants extend further into the environment (Martinez et al., 2003). In fact, amphiphilic siderophores represent nearly half of known marine siderophores and are thought to reduce siderophore diffusion in oceanic settings (Boiteau et al., 2016).

Formomarinobactin-producing Pseudomonas strains have been isolated from a variety of terrestrial and freshwater sources, including plants (grasses, rhizosphere, seeds, fruits), chemical contaminated soils, insects, cloud water, lakes, and river waters. A single strain was isolated from seawater. The shorter acyl chains of formomarinobactins may therefore reflect an evolutionary adaptation to more terrestrial or freshwater habitats, where reduced hydrophobicity and membrane interactions are likely advantageous. Lactuchelin, which is also produced by strains from terrestrial or freshwater habitats (Chesneau et al., 2025), has a Ser-rich peptide chain (βOHAsp-Ser-Ser-Ser-Orn-Ser-βOHAsp) that is therefore highly hydrophilic, more than the formo/marinobactin core, and this could compensate for longer C8-C16 tails to maintain motility/solubility in soil.

Phylogenomic mapping of the formomarinobactin siderophore gene cluster revealed that the ability to produce formomarinobactin is not unique to P. graminis but is characteristic of strains of the P. lutea G of the P. fluorescens lineage. The P. lutea G is a relatively small G which consists currently of six described species, but phylogenomic analyses in this study revealed a yet uncovered diversity and identified an additional 18 putative species. Genome mining in combination with mass analysis showed that all these strains are predicted or were shown to produce formomarinobactin, except for P. bohemica LMG 30182^T isolated from bark beetle (Saati-Santamaría et al., 2018), which lacks the biosynthetic genes, but does have a putative formomarinobactin receptor. Phylogenetic analysis indicated that the receptor of P. bohemica likely resulted from the loss of the biosynthesis genes. This pattern aligns with the Black Queen Hypothesis (BQH), where costly public goods such as leaky diffusible siderophores are produced by “helpers” (producers) but exploitable by “beneficiaries” (nonproducers or cheaters) that retain uptake machinery, driving adaptive gene loss for biosynthetic pathways while conserving receptors for efficient scavenging (Morris et al., 2012). In fact, P. bohemica seems to be a more efficient scavenger than the other genome-sequenced strains of the P. lutea G, possessing the highest number of TonB-dependent receptors, with an especially high number of predicted siderophore TonB-dependent receptors. However, further analyses are required to confirm this hypothesis.

Pseudomonas species typically produce one to three distinct siderophores and encode multiple TonB-dependent receptors that enable the utilization of xenosiderophores, siderophores produced by other bacterial or fungal species. BLAST analysis identified numerous Pseudomonas strains harboring a putative formomarinobactin receptor, revealing a substantial capacity within the P. fluorescens lineage to exploit formomarinobactins as an iron source. Growth stimulation by formomarinobactin was confirmed for select Pseudomonas strains, with functional validation in P. rhodesiae. In addition, formomarinobactin growth stimulation was demonstrated for a Phytopseudomonas strain. Further testing with the analog marinobactin, along with mutational studies in selected receptor-expressing strains, would provide deeper insights into substrate specificity and physiological relevance.

This growth-promoting effect is consistent with the high iron-chelation capacity reported for marinobactin siderophores. Although marinobactin-E exhibits a higher intrinsic Fe(III) stability constant (log K_ML = 31.8, Zhang et al., 2009) than pyoverdines’ conditional stabilities at pH 7 (log K_pH7 = 24.26–27.14, Ongena et al., 2002), protonation of the hydroxamate ligands reduces its effective affinity, making it roughly comparable (~26.7) under physiological conditions. The predicted high affinity of formomarinobactins could explain their exclusive occurrence in pyoverdine-negative strains, as maintaining two high-affinity siderophore systems would impose excessive biosynthetic and regulatory costs on the cell. A similar pattern is observed with lactuchelin (Chesneau et al., 2025), implying that both siderophores act as primary iron-chelating systems in their native hosts. This is in contrast with siderophores of the corrugatin-family. Except for corrugatin, which is produced by a pyoverdine non producing species, all the reported cases are in pyoverdine producing Pseudomonas strains. These siderophores therefore generally coexist with pyoverdine and likely assume a secondary role rather than replacing pyoverdine as the principal Fe(III) scavenger.

Additionally, growing evidence indicates that siderophores can chelate other metal ions such as Cu, Ni, Zn, Ga, Al, etc. and some heavy metals (Arnold, 2024; Schalk et al., 2011; Gomes et al., 2024). Hydroxamate-type ligands are known to allow, albeit with lesser affinity, binding with other tri-charged cations such as Ga³⁺ or Al³⁺ (Evers et al., 1989) as well as Zn²⁺, Ni²⁺ and Cu²⁺ and Co²⁺ (Hernlem et al., 1996). However, this binding does not equate efficient uptake by receptors. The biosynthetic gene cluster and receptor of formomarinobactins are not regulated by Zn or Ni but only Fe, suggesting they are true siderophores. Nonetheless, they may still be able to chelate other metal ions due to their ligand type. Several P. lutea strains isolated from metal-contaminated soils attest to their proliferation in the presence of toxic metals which could indicate siderophore-based remediation strategies (Arnold, 2024; Schalk et al., 2011; Hesse et al., 2018; Gomes et al., 2024).

Moreover, the growth inhibiting activity of members of the P. lutea G can largely be attributed to their siderophore production. Growth inhibition of S. capitis I, B. zanthoxylo RSN02 and P. aeruginosa LMG 1242^T by the P. lutea G strains was exclusively observed in iron-limited conditions. In direct contact with the purified formomarinobactins, the growth inhibition was present in both iron-limited and iron-rich media, although inhibition zones were smaller in iron-rich conditions.

Similarly, the production of lactuchelin from P. lactucae CFBP 13502 was necessary for the growth inhibition of Xanthomonas campestris pv. campestris 8004 (Xcc8004). Coculture of Xcc8004 with lactuchelin BGC deletion mutant of CFBP13502 and supplementation of the non-mutant co-culture supernatant with 0.39–50 μM FeCl₃, tested independently, resulted in the total and gradual loss of inhibition, respectively (Chesneau et al., 2025). A gene encoding for the manganese exporter protein, mntH, in Xcc8004 was shown to have been significantly altered in the presence of P. lactucae in iron-limited conditions suggesting an impairment in the Mn/Fe regulation of Xcc8004, prompting authors to argue that iron starvation led to the observed antagonism.

Growth inhibition results for Enterobacteriaceae sp. DV 4951 highlighted the presence of other molecules or a siderophore-independent mechanism involved in the inhibition of the strain by P. lutea G strains. The culture, cell-free supernatant and cell suspension of P. graminis CPA-7 were previously shown to have no antagonistic effect in vitro against E. coli O157: H7 or Salmonella, while CPA-7 dose-dependently reduced the log₁₀(cfu) populations of both strains when inoculated on fresh cut apple and peaches (Alegre et al., 2013b). The authors suggested a competitive reduction rather than the secretion of antagonistic molecules. However, iron concentration was not considered and may explain the lack of activity of CPA-7 in medium against E. coli and Salmonella by the absence of siderophore production.

Pseudomonas bohemica exclusively possesses anti-yeast activity contrary to other P. lutea G strains. An iron regulated molecule is likely responsible for this activity because antifungal activity was iron-repressed. P. bohemica has previously been shown to inhibit the growth of Candida humilis (Saati-Santamaría et al., 2018). This selective activity emphasizes P. bohemica’s potential as a targeted antagonist against specific yeasts.

All in all, the iron chelating efficiency combined with the growth inhibiting properties of amphiphilic formomarinobactin produced by P. graminis and other strains of the P. lutea G shows that siderophore production is not only a mechanism of iron acquisition but also a critical ecological trait that structures the competitive dominance and biocontrol potential of the P. lutea G, including P. graminis, especially, though not exclusively, in iron limited conditions.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. CG is a FRIA grantee of the Fonds de la Recherche Scientifique -FNRS, Belgium. CG thanks the asbl Meurice R&D for financial support through the crédit de stimulation de la recherche “BIOMYCETES”. The Analytical Platform is supported by the FNRS and Université libre de Bruxelles.

Data availability statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary material.

Author contributions

CG: Funding acquisition, Writing – review & editing, Data curation, Methodology. KH: Writing – review & editing, Methodology, Investigation, Writing – original draft, Data curation, Formal analysis. ML: Methodology, Writing – review & editing. BC: Writing – review & editing, Methodology. NB: Methodology, Writing – review & editing. SM: Conceptualization, Writing – review & editing, Supervision, Formal analysis, Investigation, Writing – original draft, Data curation, Resources. GP: Methodology, Writing – review & editing.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. Generative AI (ChatGPT, perplexity) was used for language editing and improving clarity; all scientific content, data interpretation, and conclusions are the authors’ own.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb.2026.1788039/full#supplementary-material