Genomics‐Driven Discovery of NO‐Donating Diazeniumdiolate Siderophores in Diverse Plant‐Associated Bacteria

Abstract Siderophores are key players in bacteria–host interactions, with the main function to provide soluble iron for their producers. Gramibactin from rhizosphere bacteria expands siderophore function and diversity as it delivers iron to the host plant and features an unusual diazeniumdiolate moiety for iron chelation. By mutational analysis of the grb gene cluster, we identified genes (grbD and grbE) necessary for diazeniumdiolate formation. Genome mining using a GrbD‐based network revealed a broad range of orthologous gene clusters in mainly plant‐associated Burkholderia/Paraburkholderia species. Two new types of diazeniumdiolate siderophores, megapolibactins and plantaribactin were fully characterized. In vitro assays and in vivo monitoring experiments revealed that the iron chelators also liberate nitric oxide (NO) in plant roots. This finding is important since NO donors are considered as biofertilizers that maintain iron homeostasis and increase overall plant fitness.


General Cultivation Procedures
The medium used for production of gramibactin, plantaribactin, and megapolibactins was MM9. [1] For small cultures (2 mL), a pipette tip of cells from plate was added to the respective medium in culture tubes. The cultures were shaken at 150 rpm and at 30 °C for the selected time. For larger cultures (100 mL to 200 mL), a liquid pre-culture was prepared and the OD600 measured. Cultures were inoculated with an OD600 of 0.5 to 0.8. For large scale cultivations (4 L), pre-cultures were grown overnight and 2% v/v of the preculture was added to the medium. The cultures were shaken at 120 rpm and 30 °C for 3 days. Cultures of the knockout-strains Paraburkholderia graminis ΔgrbD and ΔgrbE and Burkholderia plantarii ΔplbD were always prepared with addition of 0.1% v/v of chloramphenicol (50 mM) in DMSO.

General extraction procedure
For extraction of bacterial culture supernatants, Amberlite® XAD16N resin was used. First, cultures were centrifuged at 8000 rpm for 2-30 min at 20 °C and the supernatant was collected. XAD16N resin was prepared by washing it with acetone and methanol followed by equilibration with distilled water. The equilibrated resin (5% w/v) was added to the culture supernatant. The solution was stirred at room temperature for 2 h. The resin was filtered off, washed with water and eluted with methanol. From the methanol fraction, volatiles were removed and the residual aqueous solution was freeze-dried. The extract was dissolved in MeOH for LC-MS analysis or purified further.

Preparation of Knock-out Strains of P. graminis C4D1M
Genomic DNA from P. graminis C4D1M was isolated as described previously. [1] A gene fragment containing grbD was amplified by PCR with the primer pairs Bg-UK-fw/Bg-UK-PacI and Bg-UK-KpnI/Bg-UK-NheI using DeepVent Polymerase (New England Biolabs) followed by Taq DNA Polymerase (New England Biolabs), respectively. The amplicons were purified using the illustra GFX PCR DNA and Gel Band Purification Kit (GE Healthcare Life Sciences) and cloned into pGEM T-easy vector (Promega), resulting in pGEM-Bg-UK-KO1 and pGEM-Bg-UK-KO2, respectively. The PCR product containing the chloramphenicol resistance gene, which was amplified from pACYC184 [4] with the primers Cm-fw-KpnI and Cm-rv-PacI using DeepVent Polymerase, was purified using the abovementioned procedure. The amplicon was cloned into pGEM T-easy vector and the resulting plasmid was restricted with PacI and KpnI. This chloramphenicol resistance cassette gene was cloned into the PacI/SpeI restricted pGEM-Bg-UK-KO1 and KpnI/NheI restricted pGEM-Bg-UK-KO2, generating pGEM-ΔgrbD. A gene fragment containing grbE was amplified by PCR with the primer pairs grbE-fw/grbE-PacI and grbE-KpnI/grbE-NheI using DeepVent Polymerase followed by Taq DNA Polymerase, respectively. The amplicons were purified using the above-mentioned procedure and then cloned into pGEM T-easy vector, resulting pGEM-grbE-KO1 and pGEM-grbE-KO2, respectively. The abovementioned chloramphenicol resistance cassette gene was cloned into the PacI/SpeI restricted pGEM-grbE-KO1 and KpnI/NheI restricted pGEM-grbE-KO2, generating pGEM-ΔgrbE. P. graminis C4D1M was pre-cultured in LB medium at 30 °C. Overnight cultured cells were inoculated in LB medium (1/100 and 1/50 dilution) and cultured up to OD600=0.4 to 0.6 at 30 °C. The culture broth was centrifuged at 2,500 × g and the supernatant was discarded. The cell pellet was resuspended in 300 mM sucrose solution and centrifuged again. After repeating this washing step twice, the washed cells were resuspended in 300 mM sucrose solution and subjected to electroporation (200 kV) with knockout plasmids (ca. 1−5 µg). Transformed cells were pre-cultured in LB broth (1 mL) for 4−6 h at 30 °C with shaking and then plated on either LB or nutrient agar plates with chloramphenicol (25 µg mL -1 ). After 3 to 5 days, several positive colonies were observed and confirmed by colony PCR. Final PCR confirmation of knock-out mutants was performed using genomic DNA as a template, which was purified with The Wizard ® Genomic DNA Purification Kit (Promega).

Chemical complementation of P. graminis knock out strains
In order to supplement cultures of the knockout-strains P. graminis ΔgrbD and ΔgrbE with synthetic graminine, the respective amount (final concentration 0.1 to 0.5 mg mL -1 ) was dissolved 1 mL of MM9 medium. The solution was sterilized by filtration using 33 mm Ezee Syringe Filters (0.22 µm, PVDF) from Elkay and the filter was washed with 1 mL of MM9 medium. The sterile solution was mixed with 0.1% v/v of chloramphenicol (50 mM) in DMSO and added to the bacterial culture.

Creation of structure similarity networks (SSN)
The web tool EFI-EST [5] was used to create the sequence similarity network based on the amino acid sequence of GrbD using the standard settings. 210 was used as the alignment score to create the final networks. EFI-GNT was used to retrieve the genome neighbourhoods of each node in the network. Every node without a homologue of grbE, the query node, and nodes that had the respective cluster separated on different contigs were removed from the networks.
Identification of gramibactin as a metabolite of Paraburkholderia caledonica P. caledonica DSM17062 cultures were prepared and extracted as described above. Gramibactin isolated from P. graminis was used as a standard for comparison ( Figure S2). The hydrolysed and thus linear congener of gramibactin was identified using HRMSMS experiments ( Figures S3-4).

Marfeys Analysis
For analysis of the stereochemistry of synthesized amino acids or amino acids from isolated natural products, 1-fluoro-2,4dinitrophenyl-5-L-alaninamide (L-FDAA; Marfey's reagent) was used. The amino acid of interest (0.5 to 1 mg) was dissolved in water (100 µL) and 50 µL of NaHCO3 (1M) were added. L-FDAA was prepared as a 10 mg mL -1 solution in acetone shortly before use and 10 µL of this solution were added to the reaction mixture. At 40 °C, the solution was stirred for 1 h. Subsequently, 25 µL of 2 N HCl were added to quench the reaction. 25 µL methanol were added to the reaction mixture and an aliquot was analysed using LC-MS with a Thermo Accucore C18 column (100 × 2.1 mm; 2.6 µm) and an elution gradient [solvent A: H2O + 0.1% HCOOH, solvent B: acetonitrile + 0.1% HCOOH, gradient: 10% B to 20% B in 10 min, 20% to 30% B in 20 min, flow rate: 0.2 mL min -1 , injection volume: 5 µL]. To cope with chromatographic instabilities, chromatograms were aligned to Rt of L-FDAA (16.88 min). D,L-threo-3-hydroxy aspartic acid and D,L-erythro-3-hydroxy aspartic acid were prepared according to literature. [6][7] Their elution order was already shown to be D→L. [8] In order to analyse the amino acids of isolated natural products, the compounds (0.5 to 1 mg) were hydrolyzed in 20% DCl in D2O (450 µL). The reaction mixture was stirred at 105 °C for 16 h. The solvents were removed in vacuo. And the residue was derivatized as described above. Synthetic L-graminine was hydrolysed in the same way before derivatization with L-FDAA.

Elucidation of absolute configuration of 3-hydroxy fatty acids in megapolibactins
Megapolibactin C and F were hydrolysed as described above. The obtained hydrolysate was extracted 4 times with 1 mL chloroform. The combined organic extracts were dried with sodium sulfate and evaporated to dryness. The residue was dissolved in 400 µL dry dichloromethane containing 0.2 mM dimethylaminopyridine. 5 µL S-MTPA-Cl was added and the reaction mixture was stirred for 4 h at room temperature. 1 mg R-3-hydroxy myristic acid and 1 mg R,S-3-hydroxy myristic acid were derivatized in the same way as reference compounds. Reaction mixtures were quenched with 500 µL water and the organic layer was separated. The aqueous phase was extracted 3 times with 800 µL dichloromethane and the combined organic phases were dried with sodium sulfate and evaporated to dryness. The residue was dissolved in methanol and analysed with LC-HRMS (Thermo Accucore C18 column (100 × 2.1 mm; 2.6 µm); elution gradient [solvent A: H2O + 0.1% HCOOH, solvent B: acetonitrile + 0.1% HCOOH, gradient: 73% B for 20 min, flow rate: 0.2 mL min -1 , injection volume: 5 µL]). Both hydroxy fatty acids in megapolibactin C and F were found to be R configured, we suggest the same configuration for the remaining megapolibactins.

Elucidation of serine absolute configurations in plantaribactin
Marfey's analysis revealed the presence of two serine residues with different absolute configurations. Hydrolysis in DCl/D2O did not lead to mass shifts in either of the peaks, ruling out that one of the configurations is an artefact of the hydrolysis. Bioinformatic analysis of the identified gene cluster suggests the second serine to be D-configured, due to the prediction of an epimerase domain within this module. To chemically proof this, we supplemented a 100 mL B. plantarii culture with19.8 mg 2,3,3-L-serine-d3 and cultured it at 30°C for 48 h and extracted the culture as described above. The crude extract was submitted to LC-MSMS analysis. If an epimerization to D-serine takes place, the deuterium label in α-position would be exchanged by a proton, leading to a mass shift of 1. Parent ions with M+2 (deuterated D-serine), M+3 (deuterated L-serine), M+5 (deuterated D-and L-serine) were selected and fragmented. Fragment ions with 0, 1, and 2 serines were compared ( Figure S6) with special interest in the isotopes of a fragment with only serine 2 remaining (m/z 645). The occurrence of a strong m/z 647 and a weak m/z 648 (caused by 13 C) obtained from a parent ion with two labelled serines incorporated indicates that serine 2 lost one deuterium due to epimerization. We conclude that serine 1 is L-configured and serine 2 D-configured. This is in accordance with the bioinformatic analysis of the gene cluster. Figure S6. Elucidation of absolute configuration of serine residues in plantaribactin. Fragments with 0, 1, and 2 serines (columns) were obtained from parent ions indicating incorporation of 0, 1, or 2 deuterated serines (rows).

Identification of fatty acid in plantaribactin
Plantaribactin (1.8 mg) was hydrolysed as described above. The hydrolysate was extracted twice with 0.5 mL chloroform. Combined organic phases were dried with sodium sulfate and evaporated to dryness. The residue was dissolved in 0.5 mL anhydrous methanol and 45 µL TMS-diazomethane (2 M in hexane) was added and the resulting yellow solution was stirred 10 min at room temperature. 3 µL formic acid were added and the colourless solution evaporated to dryness after adding 1 mL toluene. The obtained residue was dissolved in chloroform and submitted to GC-MS analyses. Reference fatty acids (octanoic acid, decanoic acid, undecanoic acid, dodecanoic acid, tridecanoic acid, tetradecanoic acid) were derivatized in the same way. The fatty acid incorporated in plantaribactin was identified to be dodecanoic acid ( Figure S7).   Modified pGL42a with T251A [10] pGL42a_T251A-ΔPlaNRPS pGL42a_T251A containing plbD insertional Cm R cassette This study  [12] DSM9509 ΔPlaNRPS Cm R cassette inserted into plbD This study P. caledonica DSM17062 Wild type; environmental isolate [13] B. glumae DSM9512 Wild type; environmental isolate [14] B. gladioli pv. agaricicola Wild type; environmental isolate [15] pv. cocovenenans Wild type; environmental isolate [16] HKI0739 Wild type; environmental isolate [17]

Detection of plantaribactin in B. glumae cultures
B. glumae DSM9512 cultures were prepared as mentioned before and the obtained supernatants were extracted as described above. EIC traces of crude extracts from B. plantarii and B. glumae cultures were compared ( Figure S10A).

Detection of gladiobactin in Burkholderia gladioli strains
To test whether B. gladioli strains harboring a homologue of the plb gene cluster produce plantaribactin-like siderophores, we cultured B. gladioli pv. agaricicola, B. gladioli pv. cocovenenans, and B. gladioli HKI0739 in MM9 medium (as described above). The  (Table AS1), and the predicted substrate is in agreement with the results from HR-MS/MS analysis ( Figure S10C).

Corn Cultivation Medium
The solutions listed below were prepared and autoclaved separately. Distilled water was also autoclaved and the sterilized solutions were added. Rice cultivation medium [18] The solutions were combined as listed below, pH was adjusted to 5.7 and autoclaved.

Cultivation of Rice Plants
Rice seeds (Oryza sativa 'Arborio Bianco', magicgardenseeds) were peeled and surface sterilized in 70% EtOH for 60 seconds and subsequently in 3% NaOCl solution for 15 min. After removing the liquid, seed were washed thoroughly with sterile water. Sterile seeds were then germinated in distilled water for 3 days in the dark. Germinated seeds were transferred to culture tubes filled with glass rings as solid support and 3 mL rice cultivation medium. Tubes were kept at room temperature under fluorescent light (16 h light, 8 h darkness).

In vitro NO-Release Assay
To study the NO release in a corn root extract, the fluorescent probe 2,3-diaminonaphtalene (DAN, dissolved in DMF) was used. For the in vitro assay, approximately one week old maize plants were harvested and the roots were rinsed with distilled water, separated from their shoot and frozen in liquid nitrogen. They were then ground with mortar and pestle and the plant material was suspended in 66 mM NaH2PO4 buffer (pH 6; 1 mL g -1 root). The extract was centrifuged at 4 °C and max. speed for 10 min. 20 µL of the supernatant were used for the in vitro assay. Assays were carried out in 1.5 mL Eppendorf tubes with the following final concentrations: 2 mM of the test-compound (4 mM for amino acids test-compounds), 1.2 mM H2O2, 0.2 mM DAN. Buffer was added to a final volume of 51.5 µL and the reactions were mixed. After incubation for 20 min in the dark, the reactions were stopped by addition of 250 µL 10 mM NaOH and 280 µL of the solutions were transferred to a black 96-well plate. Fluorescence was measured using a Varioskan LUX microplate reader (Thermo Fisher) (λEx. = 375 nm, λEm. = 415 nm). Every assay was performed in triplicates and every replicate was measured three times. Each experiment (same root extract; measured in one session) was normalized to the maximal mean of all replicates (set to 100). Bar diagrams in figure 4 show means of these normalized replicates with standard deviation between the normalized replicates.

In planta NO Release Assay
Young corn seedlings (2-3 days old, germinated as above) were placed in a vial containing 4 mL standard corn cultivation medium with 10 µM 4,5-diaminofluoresceine-2 diacetate (DAF-2 DA) and incubated at room temperature for 30 min. Seedlings were then removed, and the root was rinsed with deionized water before placing them in nutrient solution containing 100 µM gramibactin (n=3) or no additive (control, n=2). After 1.5-2 h, roots were placed on styrofoam as a solid support and were cut with a wet razorblade. Upon NO release formed triazolofluorescein was visualized using a Zeiss CLSM 710 confocal laser-scanning microscope (Jena, Germany), and Zen software (Zeiss) has been used to generate the images with λEx = 485 nm and λEm = 538 nm. The exact same parameters have been used for all images. Rice seedlings (germinated as described above) were cultured in rice culture medium (containing 0.8% acetonitrile and 100 µM plantaribactin; n=3) or rice medium (containing 0.8% acetonitrile, as control, n=3). After 7 days, plants were removed from the medium and roots were washed with 20 mM HEPES (pH 7.5). Staining solution was prepared freshly (10 µM DAF-FM DA in 20 mM HEPES (pH 7.5)) and roots were incubated in the staining solution for 1 h in the dark. Afterwards roots were washed again with HEPES buffer and analysed using a Zeiss Axio Observer 7 Spinning Disk Confocal Microscope (SDCM, ZEISS, Jena, Germany) with λEx = 493 nm and λEm = 517 nm. Z-Stacks were recorded through the root tip and images were processed using Fiji. Contrast was adjusted in all images to same levels and intensities over each Z-stack was summed.
Subsequently, Pd/C was filtered off and washed with MeOH and water. The flow-through was filtered with a Whatman® 0.45µ syringe filter and then concentrated and freeze-dried to give formyl graminine (14)      NMR (600 MHz, DMSO-d6):    NMR (600 MHz, DMSO-d6):   Figure S29. Putative biosynthetic gene cluster found in the genome of P. caledonica.

P. megapolitana
In the publicly available genome sequence of P. megapolitana DSM23488 (NCBI Accession number NZ_FOQU00000000) misses parts of megI and megJ in the gene cluster. They are however located on a distinct contig and contain the genetic information for 2 NRPS modules. PCR experiments led to the assumption, that this contig was not correctly assembled and is actually part of the meg cluster. We therefore decided to resequence the genome of this strain. The following data originates from this sequencing approach and the predicted NRPS architecture is in accordance with the identified megapolibactins in term of collinearity.

Stachelhaus Code
predicted amino acid NRPS analysis [22] Antismash found A1 D I L X L G V I Gly Gly Gly  Table S19. Predicted A-domain substrate specificity together with incorporated amino acids from the plb NRPS in the B. plantarii genome. Gra: graminine.

Stachelhaus Code
predicted amino acid NRPS analysis [22] Antismash found  Table S20. Predicted A-domain substrate specificity together with incorporated amino acids from the plb homologous NRPS in the B. gladioli pv. agaricicola genome. Differences to the codes extracted from modules in the plb cluster are highlighted in red. Gra: graminine.