Molecular Structure and Biosynthesis of Pyoverdines Produced by Pseudomonas fulva

Abstract

This study explored the biosynthetic mechanisms and structural diversity of pyoverdines (PVDs) produced by Pseudomonas fulva. Genomic analysis using antiSMASH identified the PVD biosynthetic gene cluster, although the C-terminal peptide sequence could not be predicted. Subsequent liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis revealed the full peptide structure, including modified residues, such as N-acetylhydroxyornithine and cyclohydroxyornithine, and confirmed the presence of several PVD isoforms with different chromophore side chains. Comparative LC-MS analysis across Pseudomonas species demonstrated that P. fulva produces unique PVD molecular mass patterns. The bioinformatic and structural modeling of non-ribosomal peptide synthetase PvdL open reading frame 3 revealed that the A2 and A3 adenylation domains are lysine selective. Although their sequences differ from known lysine-specific signatures, AlphaFold3-based structural prediction revealed conserved substrate-binding configurations, suggesting that similar substrate-binding features may have arisen independently. Notably, Thr²⁹⁷, a unique residue in the non-ribosomal code, likely plays a key role in lysine recognition. The high degree of sequence similarity between the A2 and A3 domains may reflect domain duplication and could be involved in the diversification of the PVD structure. Further functional and ecological studies are required to assess the physiological significance of P. fulva PVDs in microbial iron acquisition.

1. Introduction

Iron is an essential element for the survival and proliferation of living organisms. In response to iron limitation, many microorganisms have evolved efficient iron acquisition mechanisms, one of which is the synthesis and secretion of siderophores [1,2,3]. Siderophores chelate iron and transport it into microbial cells, enabling survival in iron-deficient environments. Microorganism-produced siderophores, such as enterobactin from Escherichia coli (with a molecular weight (MW) of ~670 Da), desferrioxamine B from Streptomyces spp. (MW ~560 Da), and ferrichrome from Aspergillus spp. (MW ~740 Da), exhibit a wide molecular weight range from 150 to 2000 Da and display significant structural diversity, including catecholates, hydroxamates, and carboxylates.

Among siderophores, pyoverdines (PVDs), produced by fluorescent Pseudomonas species, possess a peptide backbone with sequence diversification that confers a competitive advantage in interspecies survival [4,5,6,7]. The genus Pseudomonas includes various fluorescent species, including P. aeruginosa [5,8] and P. syringae [9], which are pathogenic to animals and plants, respectively. PVDs are key siderophores involved in the establishment and pathogenicity of P. aeruginosa, contributing not only to iron acquisition but also to virulence through biofilm formation and phenazine production [8]. In addition, there are soil bacteria, such as P. fluorescens [10], P. putida [11,12], and P. protegens [13], known for their plant-protective properties [14]. In environments with relatively high iron concentrations, fluorescent Pseudomonas species produce a low-MW siderophore called pyochelin [1,7]. By contrast, in environments lacking iron, they produce PVDs, characterized by a high iron-binding capacity and efficient receptor-mediated uptake. In soil, fluorescent Pseudomonas frequently colonize the rhizospheres of plants. Their efficient siderophore-based iron acquisition systems allow these bacteria to compete with iron-absorbing bacteria and fungi that parasitize plants, thereby suppressing their growth [13,15].

PVDs are peptides comprising three distinct structural components: a dihydroxyquinoline-type chromophore responsible for fluorescence, a peptide chain of 6 to 12 amino acids, and a small dicarboxylic-acid-derived moiety (e.g., succinic acid, malic acid, or their monoamides, glutamic acid, or α-ketoglutaric acid) attached to the chromophore [16]. The length and sequence of the peptide chain vary among strains, correlating with the specificities of the transporters responsible for the uptake of iron-bound PVDs. To date, the amino acid sequences of at least 60 different PVDs have been subjected to chemical characterization [6]. These sequences are not directly encoded by genomic DNA; rather, their core structure is biosynthesized by non-ribosomal peptide synthetases (NRPSs) [17]. These large multi-component enzymatic systems play pivotal roles in the synthesis of diverse complex peptides in bacteria and fungi. Biosynthetic pathways involve the participation of multiple synthetases and auxiliary enzymes, which are responsible for the production of non-proteinogenic amino acid precursors [4,5,18]. It is postulated that certain gene products localize in the periplasm and participate in the maturation of PVDs. However, the structural diversity and evolutionary origins of PVDs remain largely unexplored.

The species Pseudomonas fulva has been isolated from a variety of clinical samples [19,20] and environmental sources [21,22]. Some reports indicate the potential for this organism to cause disease, while others highlight its utility in environmental remediation [23] and plant growth promotion [22]. Although P. fulva is closely related to the well-studied P. putida, they have separate lineages [24,25]. Furthermore, P. parafulva and P. cremoricolorata are considered as closely related yet distinct species [26]. The peptide sequences of PVDs have been shown to be species-specific within the genus Pseudomonas and have historically been used as markers for species identification [11]. Therefore, elucidating the peptide sequence of an unknown PVD can provide valuable insights into the phylogenetic affiliation of the producing Pseudomonas strain. In this study, to explore the biosynthetic diversity of PVDs, the NRPSs involved in synthesizing the PVD peptide backbone in P. fulva, a relatively understudied species, were identified through genomic analyses. The biosynthetic pathways of P. fulva PVDs share similarity with those of other Pseudomonas species. However, the C-terminal sequence could not predicted from the conventional non-ribosomal code. To elucidate the structures of P. fulva PVDs, tandem mass spectrometry (MS/MS) analysis was conducted, revealing the incorporation of a lysine residue. This prompted an evaluation of the non-ribosomal code in predicting the amino acid selectivity of the NRPS adenylation (A)-domain responsible for incorporating this residue.

2. Materials and Methods

2.1. Bacterial Culture

P. fulva NBRC16639 and other Pseudomonas strains were obtained from the National Institute of Technology and Evaluation (NITE), Tokyo, Japan. The strains were maintained in 702 medium and cultured in succinate medium (22 mM KH₂PO₄, 34 mM K₂HPO₄, 0.4 mM MgSO₄, 7.6 mM (NH₄)₂SO₄, and 56 mM Na₂C₄H₄O₄; pH 7.0) for two days at room temperature, with shaking to induce PVD production. The end of the cultivation was determined by visually confirming the fluorescence of the medium under UV light.

2.2. Genomic Analysis

The genome of the P. fulva NBRC 16639 strain was extracted and purified using a DNA Extraction Kit (QIAGEN N.V., Hilden, Germany). Genomic sequencing, utilizing PacBio sequencing technology, was outsourced to Bioengineering Lab Co., Ltd. (Kanagawa, Japan). The quantification of the genomic DNA from Pseudomonas fulva NBRC 16639 was conducted using the QuantiFluor dsDNA System and a Quantus Fluorometer (Promega Co., Madison, WI, USA). The quality of the DNA was assessed by means of capillary electrophoresis, using a 5200 Fragment Analyzer System and an Agilent HS Genomic DNA 50 kb Kit (Agilent Technologies, Inc., Santa Clara, CA, USA). The purification of the DNA was accomplished using DNA Clean Beads (MGI Tech Co., Ltd., Shenzhen, China) at a volumetric ratio of 1.8:1. Shearing was performed with a g-TUBE (Covaris LLC, Woburn, MA, USA) to obtain DNA fragments of approximately 10–20 kbp. For the construction of the library, SMRTbell Prep Kit 3.0 and a SMRTbell gDNA Sample Amplification Kit (Pacific Biosciences of California, Inc. (PacBio), Menlo Park, CA, USA) were used in accordance with the manufacturer’s protocols. The quality of the PCR product was verified by means of electrophoresis using the 5200 Fragment Analyzer System and the Agilent HS Genomic DNA 50 kb Kit. The processing of sequencing libraries was undertaken using a Revio Polymerase Kit (PacBio), and the subsequent execution of the sequencing was conducted on a Revio system (PacBio). The obtained genomic data were annotated using Rapid Annotation and Subsystem Technology (RAST) [27] (https://rast.nmpdr.org/ (accessed on 3 September 2024)).

2.3. Purification of PVDs

The purification of the PVDs from P. fulva NBRC 16639 and other Pseudomonas strains was performed as previously described [28]. Solid-phase extraction was carried out using Strata-X (1 mL, 30 mg) polymeric reverse-phase cartridges (Phenomenex, Torrance, CA, USA). Briefly, the cartridge was washed with 1 mL of methanol and equilibrated with 1 mL of water (H₂O). The 500 µL supernatant was acidified with 5 µL of formic acid and loaded onto the sorbent. The sorbent was washed with 0.6 mL of water, and the PVD was eluted using 0.6 mL of 30% methanol/H₂O containing 0.1% formic acid.

2.4. LC-MS Analysis

Peptide separation was performed using a COSMOSIL ₅C₁₈-AR-II column (2.0 ID × 100 mm; Nacalai Tesque, Inc., Kyoto, Japan) attached to a Vanquish F-Type UHPLC System SII (Thermo Fisher Scientific, Inc., Waltham, MA, USA). The peptides were eluted over a 35 min linear gradient from 10% acetonitrile–H₂O to 45% acetonitrile–H₂O (containing 0.1% formic acid) at a flow rate of 0.2 mL/min and a column temperature of 40 °C. Mass analysis of the separated peptides was performed using a Q-Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific) connected to a high-performance liquid chromatography (HPLC) system. Ionization was achieved under standard conditions with a Heated-Electrospray Ionization-II source (at a spray voltage of 3.5 kV, a sheath gas flow rate of 40 L/min, an auxiliary gas flow rate of 10 L/min, a capillary temperature of 350 °C, a probe heater temperature of 300 °C, and an S-lens RF level of 60). Precursor ions were acquired using the full MS/dd-MS² mode (at a resolution of 70,000, an ITmax value of 200 ms, an AGC target of 1 × 10⁶, and a scan range from 1000 to 3000 m/z). For ions with a high signal intensity, MS/MS fragment peaks were measured by applying stepped collision energies (at normalized collision energies of 20, 25, and 30) to obtain fragmentation data (at a resolution of 35,000, an IT_max value of 100 ms, an AGC target of 1 × 10⁵, and a scan range from 200 to 2000 m/z).

2.5. Amino Acid Analysis

Approximately 5 µg of purified PVD (isoform A) was hydrolyzed with 5.7 M hydrochloric acid (800 µL) at 110 °C for 24 h. After the solution was dried, the residue was dissolved in 200 µL of 50 mM hydrochloric acid. The amino acid composition of the sample was analyzed using a High-Speed Amino Acid Analyzer (LA8080, Hitachi High-Tech Co., Tokyo, Japan).

2.6. Bioinformatic Analysis

The antiSMASH web server [29] was used to annotate secondary metabolite biosynthetic gene clusters (BGCs) in the P. fulva NBRC 16639 genome (https://antismash.secondarymetabolites.org (accessed on 6 May 2024)). We then collected non-ribosomal codes of Lys-selective A-domains and amino acid sequence IDs from the lookup table file (https://dl.secondarymetabolites.org/releases/stachelhaus/1.1/signatures.tsv (accessed on 4 October 2024)), which is used by antiSMASH to predict A-domain substrates. Using the collected sequence IDs, we also collected amino acid sequence data from the NCBI database (https://www.ncbi.nlm.nih.gov (accessed on 7 October 2024)) and extracted the amino acid sequences of Lys-selective A-domains using annotated data obtained from locally installed antiSMASH (https://dl.secondarymetabolites.org/releases/7.0.0/antismash-7.0.0.tar.gz (accessed on 7 October 2024)). We aligned the sequences of the Lys-selective A-domains using MAFFT [30] and estimated the maximum likelihood phylogenetic tree using MEGA 11 [31] with default parameters. The WebLogo sequence logos of the non-ribosomal codes of the Lys-selective A-domains were generated using the WebLogo server [32].

To analyze the interactions between the A2 and A3 domains of the P. fulva NRPS PvdL open reading frame (ORF) 3 and lysine molecules, we predicted their three-dimensional (3D) structures complexed with lysine, ATP, and Mg²⁺ using AlphaFold3 [33] installed on a local computer. Because the 3D structure of the Lys-selective A-domain in ε-poly-L-lysine synthetase (Pls-A, PDB ID: 7wew) is known to have an adenylation conformation [34], we selected the predicted structure that has the lowest Root-Mean-Square-Deviation value compared to 7wew from the five candidate structures predicted by AlphaFold3. Schematic drawings of the interactions with the lysine molecule were generated using Molecular Operating Environment (MOE) 2024.06 (Chemical Computing Group ULC, Montreal, QC, Canada, H3A 2R7; 2024).

3. Results and Discussion

3.1. Characteristics of the P. fulva Genome

The long-read sequencing of the P. fulva NBRC 16639 strain’s genome, using PacBio technology, generated a single contig, completing the genome’s assembly with a high degree of accuracy (

Table 1. Summary of next-generation sequencing results for the P. fulva NBRC16639 genome.

Base	Count	Percentage
A	939,138	19.10%
T	1,499,495	30.50%
G	1,528,630	31.10%
C	950,693	19.30%
GC	3,028,125	61.60%
All	4,917,956	100.0%
Coding regions predicted by RAST
Coding sequences (protein-coding regions)		4400
Misc RNA (other RNA-coding regions)		68
rRNA (ribosomal RNA-coding regions)		68
tmRNA (transfer-messenger RNA)		1
tRNA (transfer RNA-coding regions)		76

). The total genome length was determined to be 4,917,956 base pairs, with a GC content of 61.6%, both consistent with those of other Pseudomonas species. The previously reported P. fulva DSM 17717 strain has a genome size of 4,770,636 base pairs, indicating that the genome of P. fulva NBRC 16639 is 147,320 base pairs longer [35]. Genomic annotation using the RAST server [27] predicted 4541 coding sequences, 144 more than predicted for P. fulva DSM 17717. These results indicate genomic expansion in P. fulva NBRC 16639, possibly reflecting functional diversification.

3.2. Identification of Secondary Metabolite BGCs Using antiSMASH

To analyze the secondary metabolite BGCs in the genome of P. fulva NBRC 16639, antiSMASH analysis [29] was performed, and eight BGCs were identified. Among them, the eighth cluster shared a high degree of homology with the PVD BGC of P. protegens Pf-5 (Figure 1). This PVD biosynthetic gene cluster is located in the genomic region spanning from 4,777,182 to 4,846,237 base pairs. Ten related genes involved in PVD synthesis were identified within this cluster. These results suggest functional conservation in PVD biosynthesis between P. fulva NBRC 16639 and other Pseudomonas species known to produce structurally diverse siderophores.

The PvdL enzyme, divided into three ORFs (ORF1, ORF2, and ORF3), is an NRPS responsible for synthesizing the peptide backbone of the PVD, including its chromophore component [5]. Non-ribosomal peptide synthesis involves a modular structure, with each NRPS module activating a specific amino acid and incorporating it into the growing peptide chain [17]. The key domains of NRPSs include the condensation (C)-domain, which catalyzes peptide bond formation; the thiolation (T)-domain, also known as the peptide carrier protein (PCP); and the adenylation (A)-domain, which activates amino acid substrates. Substrate-recognizing residues in the A-domain are collectively referred to as the non-ribosomal code [36,37]. Using these features, antiSMASH [29] can predict the amino acid sequence of synthesized peptides.

The sequences of the synthesized product were predicted as follows: PvdL(ORF1), Glu-D-Tyr-Dab (2,4-diaminobutyric acid); PvdL(ORF2), Asp-Ser-Ser; PvdL(ORF3), D-Orn-D-Xaa1-D-Xaa2-Xaa3. The N-terminal Glu residue is first modified with a myristic or myristoleic acid chain. After cytoplasmic biosynthesis, the product is transported to the periplasm, where the fatty acid chain is removed. The glutamic acid residue is then converted to α-ketoglutaric acid, succinic acid, malic acid or their monoamide, producing several PVD isoforms [18]. PvdH converts L-aspartate β-semialdehyde to Dab [5,18,38], which is condensed with the D-tyrosine top form of the quinoline-type chromophore. Cyclization of the chromophore is thought to occur in the periplasm and involves the enzymes PvdN, PvdO, and PvdP [5,18]. The four N-terminal residues following the chromophore were predicted to be Asp-Ser-Ser-D-Orn (ornithine). However, the identity of the three C-terminal residues, represented by Xaa1-Xaa2-Xaa3, could not be predicted. PvdY acts as an acetyltransferase and catalyzes the acetylation of Nδ-hydroxyornithine in P. aeruginosa [39]. This modification forms the iron-binding hydroxamate group. If the peptide chain terminates with an Orn residue, cyclohydroxyornithine (cOHOrn) may be formed, increasing the iron-binding efficiency of the siderophore.

3.3. Molecular Structures of the P. fulva PVDs

To determine the structures of the PVDs produced by P. fulva, LC-MS analysis was performed, integrating bioinformatic predictions with analytical chemistry [28,40]. PVD purified from the culture supernatant was analyzed using a Q-Exactive mass spectrometer (Thermo Fisher Scientific, Inc., Waltham, MA, USA), and five molecular ions were detected, corresponding to different PVD isoforms (Figure 2 and

Table 2. Mass values for PVD isoforms A–D.

Peak	[M+H]+
A	1220.56
B	1205.54
B’	1204.57
C	1187.52
D	1131.50

). Each molecular ion was fragmented using MS/MS, producing primarily B ions, Y″ ions (Y ions with two additional hydrogen atoms), and a few A ions [28,40]. These fragments allowed the complete elucidation of the PVD peptide chain structure, confirming the amino acid sequence and modifications.

MS/MS analysis was performed using the peak for the precursor ion at 1220.56 m/z, which was assigned to a PVD peptide with a malic acid (Mala) side chain attached to the carboxylic acid group of the hydroxyquinoline chromophore (Figure 3A and Appendix A 

Table A1. MS/MS fragment ion measurement values and identification results for PVD isoform A.

Fragment Type	Chemical Formula	Observed m/z (Theoretical)
A4	C26H32N7O12	634.21 (634.210)
A5	C33H44N9O15	806.29 (806.295)
B1	C17H17N4O6	373.11 (373.114)
B2	C21H22N5O9	488.14 (488.141)
B3	C24H27N6O11	575.17 (575.173)
B4	C27H32N7O13	662.20 (662.205)
B5	C34H44N9O16	834.29 (834.290)
B6	C40H56N11O17	962.39 (962.385)
B7	C46H68N13O18	1090.48 (1090.480)
Y″2	C11H23N4O3	259.18 (259.176)
Y″5	C27H52N9O9	646.39 (646.388)
Y″6	C30H57N10O11	733.42 (733.420)
Y″7	C34H62N11O14	848.44 (848.447)

). The peptide sequence was assigned as Asp-Ser-Ser-AcOrn(OH)-Lys-Lys-cOrn(OH). Significant B-ion peaks were observed at m/z 373.1 (B1) and 488 (B2), with an additional peak for the B2 fragment at m/z 471, corresponding to a dehydrated form due to OH group loss [28]. These peaks confirmed that the first amino acid was Asp. Successive B-ion peaks appeared at m/z 575, 662, 834, 962, and 1090, completing the peptide sequence. Fragment peaks from C-terminal fragmentation (Y″ ions) were detected at m/z 259, 733, and 848. These peaks corroborated the sequence deduced from the B-ion series, confirming the accurate assignment of the entire peptide structure. The combined ion fragmentation analysis verified the primary structure of the peptide, including specific modifications, such as acetylated hydroxyornithine (AcOrn(OH)) and cyclohydroxyornithine (cOrn(OH)), essential components of the PVD siderophore.

In the MS/MS fragmentation pattern, characteristic peaks for ions specific to the Mala side chain at m/z 302 were observed in addition to the peak of the common fragment at m/z 204, confirming the characteristic fragmentation of the PVD chromophore [16]. These ions are indicative of the hydroxyquinoline-type chromophore attached to Mala, confirming its structural integrity. In addition, the B₁-ion peak at m/z 488 yielded a fragment peak at m/z 429, suggesting the cleavage of the chromophore region rather than the peptide backbone itself [16].

The MS/MS analysis of other detected molecular ion peaks revealed common amino acid sequences but distinct modifications in the side chains of the chromophore (Figure 3B–D and Appendix A 

Table A2. MS/MS fragment ion measurement values and identification results for PVD isoform B.

Fragment Type	Chemical Formula	Observed m/z (Theoretical)
A4	C26H32N7O11	619.20 (619.215)
B1	C17H17N4O5	357.12 (357.119)
B2	C21H22N5O8	472.15 (472.146)
B3	C24H27N6O10	560.16 (559.178)
B4	C27H32N7O12	646.21 (646.210)
B5	C34H44N9O15	818.30 (818.295)
B6	C40H56N11O16	946.39 (946.390)
B7	C46H68N13O17	1075.47 (1074.485)
Y″2	C11H23N4O3	259.18 (259.176)
Y″6	C30H57N10O11	733.42 (733.420)
Y″7	C34H62N11O14	848.44 (848.447)

,

Table A3. MS/MS fragment ion measurement values and identification results for PVD isoform C.

Fragment Type	Chemical Formula	Observed m/z (Theoretical)
A3	C23H25N5O9	514.92 (514.165)
A4	C26H30N6O11	601.19 (601.197)
B1	C17H15N3O5	340.09 (340.101)
B2	C21H20N4O8	455.12 (455.128)
B3	C24H25N5O10	542.15 (542.160)
B4	C27H30N6O12	629.18 (629.192)
B5	C34H42N8O15	801.27 (801.276)
B6	C40H54N10O16	929.36 (929.371)
B7	C46H66N12O17	1057.46 (1057.466)
Y″2	C11H23N4O3	259.18 (259.176)
Y″4	C24H47N8O7	559.18 (559.356)
Y″5	C27H52N9O9	646.21 (646.388)
Y″6	C30H57N10O11	733.42 (733.420)
Y″7	C34H62N11O14	848.45 (848.447)

and

Table A4. MS/MS fragment ion measurement values and identification results for PVD isoform D.

Fragment Type	Chemical Formula	Observed m/z (Theoretical)
A4	C23H26N6O10	545.16 (545.170)
B2	C18H16N4O7	399.09 (399.101)
B3	C21H21N5O9	486.13 (486.133)
B4	C24H26N6O11	573.16 (573.165)
B5	C31H38N8O14	746.25 (746.250)
B6	C37H50N10O15	873.34 (873.345)
B7	C43H62N12O16	1001.43 (1001.440)
Y″6	C30H57N10O11	733.42 (733.420)

), suggesting structural isoforms of PVD: The peak at 1205.54 m/z was identified as the succinamide-type isoform; the peak at 1204.57 m/z was assigned to the succinylated-type isoform. The peak at 1187.52 m/z corresponded to the cyclic succinamide type formed by intramolecular cyclization; the peak at 1131.50 m/z was designated as an azotobactin-like structure with a five-membered ring added to the chromophore. Both azotobactin and PVDs are known to form stable complexes with metal cations [12,41,42].

3.4. Amino Acid Compositions of the P. fulva PVDs

The amino acid composition of P. fulva PVD isoform A (m/z 1220.56) was analyzed (

Table 3. Amino acid analysis results for PVD isoform A.

Amino Acid	Amount (nmol)	Ratio
Asp	2.94	1.01
Ser	5.84	2
Orn	3.28	1.12
Lys	6.22	2.13

). Comparative analysis with a standard sample of L-Orn validated this identification. The quantified amino acid composition was Asp, Ser, Orn, and Lys in a ratio of 1.0:2.0:1.12:2.13. Compared with the expected theoretical ratio of 1:2:2:2, based on the identified sequence, the detected ratio of the Orn was lower. This discrepancy likely reflects incomplete recovery due to structural modifications, specifically, acetylated hydroxyornithine (AcOHOrn) and cyclohydroxyornithine (cOrnOH), which may have undergone partial hydrolysis during the analysis, reducing the measured Orn content.

3.5. Molecular Mass Distribution of PVDs in Pseudomonas Strains

The molecular masses of the secondary metabolites corresponding to the PVDs in the culture supernatant of Pseudomonas NBRC strains were compared under iron-deficient conditions (

Table 4. LC-MS analysis results for culture supernatants from different Pseudomonas NBRC strains.

Strain	Observed Mass	Sequence	Ref.
P. fulva (NBRC:16637)	1115.47 1187.53 1204.56 1220.55 1238.56	Asp-Ser-Ser-AcOHOrn-Lys-Lys-cOHOrn	This study
P. fulva (NBRC:16639)	1131.50 1187.53 1204.56 1220.55	Asp-Ser-Ser-AcOHOrn-Lys-Lys-cOHOrn	This study
P. parafulva (NBRC:16635)	1190.48 1191.49 1206.50 1207.48	Asp-Orn-(OHAsp-Dab)-Gly-Ser-cOHOrn	[11]
P. parafulva (NBRC:16636)	1188.45 1206.47	Not assigned
P. putida (NBRC:3778)	1368.51 1388.57 1404.57	Not assigned	[43]
P. putida (NBRC:12996)	1089.41 1091.42 1107.42 1122.43	Asp-εLys-OHAsp-Ser-Ala-Ser-cOHOrn	[44]
P. putida (NBRC:14164)	1319.50 1336.52 1337.51 1352.52	Ser-Gln-Dab-Asp-Thr-Gly-Asp-Thr-Thr-Gly	[11]
P. putida (NBRC:14671)	1342.58 1358.58 1359.58 1374.59	Not assigned
P. putida (NBRC:14796)	1012.47 1150.42 1166.42 1184.43	Asp-Orn-Dab-Gly-Ser-Ser-OHAsp-Thr	[43]
P. putida (NBRC:15366)	1305.52 1321.51 1323.53 1338.54 1339.52 1352.55 1354.53 1368.55 1369.53	Asp-Lys-OHAsp-Ser-Ala-Thr-Thr-Thr-cOHOrn	[11]
P. putida (NBRC:100650)	1056.40 1073.42 1074.41 1092.42	Asp-Orn-(OHAsp-Dab)-Gly-Ser-cOHOrn	[11]
P. putida (NBRC:109349)	1187.49 1199.49 1217.50 1235.51 1251.52	Asp-εLys-OHAsp-Ser-Gly-aThr-Lys-cOHOrn	[45]
P. putida (NBRC:110474)	1323.53 1338.54 1339.52 1354.53 1356.55 1372.58	Ser-Gln-Dab-Asp-Thr-Gly-Asp-Thr-Thr-Gly	[11]

). Two P. fulva NBRC strains, 16637 and 16639, exhibited similar molecular mass patterns. However, additional signals were detected at m/z 1115.47 (azotobactin type -16) for strain 16637 and at m/z 1238.56 (malic acid amide type +18) for strain 16637, suggesting potential structural variants. These PVDs likely share the same amino acid sequence as that previously identified in strain 16639. An environmental isolate of P. fulva displayed a molecular mass pattern identical to that of NBRC 16639. Similarly, P. parafulva strains NBRC 16635 and 16636 showed molecular mass patterns that differed from that of P. fulva by only two or three mass units, indicating that these closely related species likely produce structurally similar PVDs.

By contrast, one P. cremoricolorata strain and nine P. putida NBRC strains yielded significantly different molecular mass patterns. Most of the molecular masses detected were consistent with the predicted masses of previously reported PVD sequences. These results suggest that while P. fulva appears to produce a limited variety of PVDs with conserved sequences across strains, P. putida exhibits greater molecular diversity, reflecting different evolutionary pathways for PVD biosynthesis in these Pseudomonas species.

3.6. Bioinformatic Analysis of the Lysine-Incorporating Adenylation Domain of P. fulva PVDs

Although the predicted peptide sequence of the product synthesized from PvdL(ORF3) was D-Orn-D-Xaa1-D-Xaa2-Xaa3, MS analysis suggested that both the first and second unknown amino acids (Xaa1 and Xaa2, respectively) were lysine residues. This indicates that the second and third A-domains (A2 and A3) of PvdL(ORF3) must recognize lysine residues. To validate the non-ribosomal code [36,37] of the A2 and A3 domains in P. fulva PvdL(ORF3), we compared them with other known Lys-selective A-domains (

Table 5. Comparison of the non-ribosomal codes of the A-domains of 26 NRPSs that use lysine as a substrate.

Species	Domain	Signature
Pseudomonas fulva	PvdL(ORF3)A2	D A E D H G T V T K
Pseudomonas fulva	PvdL(ORF3)A3	D A E D H G T V T K
Pseudomonas protegens Pf-5	PvdD A3	D A E D N G T V S K
Pseudomonas tolaasii	APU91751 A6	D A E S V G T I I K
Streptomyces albulus NBRC14147	Pls-A	D A E S I G T V V K
Xenorhabdus miraniensis	Xmir_01407 A6	D A E S I G T V I K
Planktothrix agardhii NIVA-CYA 126/8	apnA A2	D A E D I G S V V K
Nostoc punctiforme PCC 73102	Npun_F2460 A2	D A E D I G S V I K
Stigmatella aurantiaca Sg a15	mxcG A1	D A E D I G T V V K
Anabaena sp. 90	aptA1 A2	D T E D I G S V I K
Anabaena sp. 90	aptA2 A2	D T E D I G S V V K
Mycobacterium tuberculosis H37Rv	Rv0101 A1	D I E D V G S V V K
Bacillus licheniformis	bacB A1	D A E S I G S V C K
Paenibacillus larvae subsp. larvae DSM 25430	ERIC2_c18070 A1	D M E D V G S V D K
Paenibacillus sp. OSY-SE	pbtB A2	D V G D V G S I D K
Streptomyces fradiae	lptC A1	D A D D A G T V D K
Streptomyces sp. NRRL F-4415	gobR A2	D A D D G G F V D K
Uncultured bacterium	mlcL A1	D T D D M G Y V D K
Streptomyces albus subsp. chlorinus	FM076_21195 A1	D T E D M G Y V D K
Streptomyces viridosporus ATCC 14672	cip22 A5	D T D D M G F I D K
Burkholderia sp. B8(2020)	necA A1	D T E N I G T I S K
Mycobacterium tuberculosis H37Rv	MbtF A1	D A Q D A G C V E K
Myxococcus virescens	benD A1	D N E S G G T V A K
Brevibacillus laterosporus	BogB A2	D S G P S G A V D K
Brevibacillus laterosporus	BogC A4	D A G P S G A V D K
Streptomyces filamentosus NRRL 15998	SSGG_00679 A2	D V F E S G G V A K

). The A2 and A3 domains of PvdL(ORF3) in P. fulva share highly similar sequences (95% sequence identity). In particular, the amino acid sequences of the N-terminal core domains of the A2 and A3 domains, both 408 residues long, are almost identical (99% sequence identity). The non-ribosomal code of these A-domains was identified as (1)Asp¹⁹¹-(2)Ala¹⁹²-(3)Glu¹⁹⁵-(4)Asp²³³-(5)His²⁶¹-(6)Gly²⁶³-(7)Thr²⁸⁸-(8)Val²⁹⁶-(9)Thr²⁹⁷-(10)Lys⁴⁸⁶. The divergence of the non-ribosomal codes among the known Lys-selective A-domains is visualized as a sequence logo in Figure 4, which summarizes the conserved amino acid sequences at each of the 10 positions in the non-ribosomal code. Notably, positions 1, 6, and 10 are highly conserved, corresponding to Asp, Gly, and Lys, respectively. Meanwhile, positions 5 and 9 are variable sites in the known Lys-selective A-domains, and His and Thr have not yet been observed at these sites. The apparent inconsistencies at positions 5 and 9 in the non-ribosomal code of P. fulva may explain the failure of antiSMASH to identify lysine as the predicted substrate for the A2 and A3 domains in PvdL(ORF3).

Next, the three-dimensional structures of the A2 and A3 domains of P. fulva PvdL(ORF3) were predicted using AlphaFold3 [33], and the substrate interaction mode was compared with that of the Lys-selective A-domain in ε-poly-L-lysine synthetase (Pls-A), which structure has been determined using X-ray crystallography [34]. Although the sequence similarity degree between the A2 domains of PvdL(ORF3) and Pls-A is relatively low (35% sequence identity), the predicted structure of the A2 domain in the PvdL(ORF3) complex with lysine and ATP molecules (Figure 5A) showed good agreement with the relative positions of the amino acid residues comprising the non-ribosomal code in the crystal structure of the Pls-A complex with lysyl-adenylate (Figure 5B,C). In the predicted structures of A2 and A3 in PvdL(ORF3), the α-amino group of the substrate forms hydrogen bonds with the side chain of (1)Asp¹⁹¹ and the main-chain carbonyl oxygens of (8)Val²⁹⁶ and Gly²⁹⁰ (Figure 5D). These residues correspond to (1)Asp²¹³, (8)Val³⁰⁹, and Gly³⁰³, respectively, in Pls-A (Figure 5E). The carboxyl group of the substrate was stabilized by a salt bridge with (10)Lys⁴⁸⁶, located in the C-terminal subdomain, which is conserved in nearly all NRPS A-domains and plays crucial roles in ensuring precise substrate binding and proper alignment at the reaction site [36]. Furthermore, the side chain of the lysine molecule interacts with (3)Glu¹⁹⁵, (4)Asp²³³, (7)Thr²⁸⁸, and (9)Thr²⁹⁷. Of these, (9)Thr²⁹⁷ is a unique amino acid type of the non-ribosomal code for A2 and A3 in PvdL(ORF3), as described above. The hydroxyl group of the Thr side chain forms a hydrogen bond with the α-amino group in the side chain of the lysine molecule. This indicates that the unique (9)Thr²⁹⁷ of A2 and A3 in PvdL(ORF3) may be important for lysine recognition by A2 and A3 in PvdL(ORF3). The (5)His²⁶¹ residue, which is also a unique amino acid type and not prevalent in the known non-ribosomal code in the Lys-selective A-domains, does not bind to lysine or ATP molecules because (5)His²⁶¹ is buried in the interior of the domain core. A similar outcome was obtained using the predicted structure of the Lys-selective A-domain in PvdD of P. protegens Pf-5 (AAY93354), the phylogenetically closest Lys-selective A-domain to A2 and A3 in P. fulva PvdL(ORF3). The (5)Asn²⁶² residue of the Lys-selective A-domain in the PvdD of P. protegens Pf-5 is also not involved in lysine interactions (Appendix B Figure A1). This observation may explain the low sequence conservation degree at position 5 in the non-ribosomal code among Lys-selective A-domains. This trend is further illustrated in the WebLogo plot (Figure 4), which graphically represents the sequence conservation across 10 key residues of Lys-selective A-domains. In this figure, position 5 exhibits significant variability, supporting the structural findings. The non-ribosomal code of Pls-A was determined to be (1)Asp²¹³-(2)Ala²¹⁴-(3)Glu²¹⁷-(4)Ser²⁵⁶-(5)Ile²⁷⁷-(6)Gly²⁷⁹-(7)Thr³⁰¹-(8)Val³⁰⁹-(9)Val³¹⁰-(10)Lys⁴⁹⁵, and the (3)Glu²¹⁷ residue directly forms a salt bridge with the amino group of the lysine substrate’s side chain [34] (Figure 5E). This (3)Glu²¹⁷ residue in the non-ribosomal code is also conserved in the A2 and A3 domains of PvdL(ORF3) in P. fulva, strongly suggesting lysine selectivity in these domains (Table 5). Further biochemical validation is required to confirm the substrate specificity. In particular, the stereochemistry of the incorporated amino acids (L- or D-form) in non-ribosomal peptide synthesis was inferred from the presence or absence of epimerization (E)-domains in each module of the PvdL. However, the stereochemical configuration cannot be determined using LC-MS/MS analysis alone, and this remains an important subject for future investigation.

4. Conclusions

This study characterized the biosynthetic mechanism and structural diversity of PVDs produced by P. fulva. Initially, antiSMASH was used for genomic bioinformatic analysis to predict the PVD structure, but the sequence of the three C-terminal residues could not be determined. The subsequent detailed structural analysis by LC-MS/MS revealed modified residues, such as AcOrn(OH) and cOrn(OH), which are capable of forming Fe^³+-binding sites, as well as PVD isoforms with different chromophore side chains. Furthermore, the LC-MS-based comparison of the PVD molecular diversity among Pseudomonas species showed that the molecular mass patterns of these PVD isoforms are unique to P. fulva.

PVDs are known to play critical roles in iron acquisition competition among microorganisms. Although the underlying causes of their sequence diversity remain to be fully elucidated, horizontal gene transfer has been proposed as a contributing factor, based on patterns of codon usage and tetranucleotide composition [46]. In environments with limited iron availability, producing a siderophore structurally distinct from common types could confer a competitive advantage to bacteria.

Compared with the C-terminal region of the closely related P. putida PVDs, a marked divergence was identified in that of P. fulva PVDs. Since the N-terminal region is essential for receptor binding (e.g., FpvA [47]), the C-terminal region may be under relaxed selective pressure, allowing for species-specific diversification. In a manner similar to that of other PVDs, the presence of a Lys-selective A-domain in P. fulva may reflect genetic exchange events, such as horizontal gene transfer, although direct evidence is lacking [46].

Furthermore, the adjacent A2 and A3 domains in PvdL(ORF3), which introduce a lysine repeat sequence, share a high level of sequence identity (95%), suggesting that they may have arisen through domain duplication. Non-ribosomal peptide synthetases (NRPSs) are multidomain enzymes composed of modules, such as adenylation, peptidyl carrier, and condensation domains (Figure 1), and it has been reported that repeated domains within a single gene can often have different evolutionary origins [48]. This observation implies that domain duplication and the generation of paralogs within the genome have contributed significantly to the structural and functional diversification of NRPSs. Indeed, the evolution of NRPS domains is hypothesized to involve a gene duplication–deletion mechanism [48], and such domain duplication may also play a role in shaping PVD diversity in P. fulva. However, the specific functional and evolutionary contributions of these duplicated domains require further investigation.

This study has two main limitations. First, although the molecular structure of P. fulva PVDs was clarified, their physiological roles in iron acquisition remain to be tested. Second, while sequence diversity may be associated with horizontal gene transfer and gene duplication, the evolutionary mechanisms underlying these observations remain speculative and await experimental validation.