Data on metal-chelating, -immobilisation and biosorption properties by Gordonia rubripertincta CWB2 in dependency on rare earth adaptation

Recent studies have shown that the metal adaptation of Actinobacteria offers a rich source of metal inducible environmentally relevant bio-compounds and molecules. These interact through biosorption towards the unique cell walls or via metal chelating activity of metallophors with trace elements, heavy metals and even with lanthanides to overcome limitations and toxic concentrations. Herein, the purpose is to investigate the adaptation potential of Gordonia rubripertincta CWB2 in dependence of the rare earths and to determine if we can utilize promising metallophore metal affinities for metal separation from aquatic solutions. For details on data interpretation and applicability of siderophores we refer to the related article entitled “Cultivation dependent formation of siderophores by Gordonia rubripertincta CWB2” [1]. The respective workflow comprises a metal adaptation method to demonstrate effects on bacterial growth, pH, metallophore production, and metabolic change. All this was evaluated by LC-MS/MS and effects on biosorption of rare earths was verified by ICP-MS. Furthermore, we were able to carry out batch metal adsorption and desorption studies of metallophores entrapped in inorganic polymers of tetramethoxysilane (TMOS) to determine metal chelating capacities and selective enrichment effects from model solutions. The adaptation potential of strain CWB2 at increased erbium and manganese concentrations was verified by increased chelating activity on agar plates, in liquid assays and demonstrated by the successful enrichment of erbium by metallophore-functionalized TMOS-polymers from an aquatic model solution. Furthermore, the number of detected compounds in dependency of rare earths differ in spectral counts and diversity compared to the wild type. Finally, the biosorption of rare earths for the selected adaptation was increased significantly up to 2-fold compared to the wild-type. Overall a holistic approach to metal stress was utilised, integrating a bacterial erbium adaptation, metal chelating, biosorption of lanthanides and immobilization as well as enrichment of metals using metallophore functionalized inorganic TMOS polymers for separation of metals from aquatic model solutions.

we refer to the related article entitled "Cultivation dependent formation of siderophores by Gordonia rubripertincta CWB2" [1] . The respective workflow comprises a metal adaptation method to demonstrate effects on bacterial growth, pH, metallophore production, and metabolic change. All this was evaluated by LC-MS/MS and effects on biosorption of rare earths was verified by ICP-MS. Furthermore, we were able to carry out batch metal adsorption and desorption studies of metallophores entrapped in inorganic polymers of tetramethoxysilane (TMOS) to determine metal chelating capacities and selective enrichment effects from model solutions. The adaptation potential of strain CWB2 at increased erbium and manganese concentrations was verified by increased chelating activity on agar plates, in liquid assays and demonstrated by the successful enrichment of erbium by metallophorefunctionalized TMOS-polymers from an aquatic model solution. Furthermore, the number of detected compounds in dependency of rare earths differ in spectral counts and diversity compared to the wild type. Finally, the biosorption of rare earths for the selected adaptation was increased significantly up to 2-fold compared to the wild-type. Overall a holistic approach to metal stress was utilised, integrating a bacterial erbium adaptation, metal chelating, biosorption of lanthanides and immobilization as well as enrichment of metals using metallophore functionalized inorganic TMOS polymers for separation of metals from aquatic model solutions.  We report on production and nature of siderophores obtained from the Actinobacterium G. rubripertincta CWB2. Production was optimized and an application is highlighted. Type of data Table  Image Graph Figure  How  Provided as EXCEL file. Parameters for data collection We investigated different media compositions (ionic strength, phosphate concentration, additional trace elements and vitamins and metal induced stress) to optimize metallophore production. Further, we investigated adsorption and desorption of metals by metallophore functionalized inorganic polymers of tetramethoxysilane (TMOS) from model solutions at pH 4. We analysed pH profiles during siderophore production from pH 5 to pH 9.
( continued on next page ) Description of data collection Herein we combine a CAS agar assay with additional metals. A metal adsorption screening of TMOS entrapped siderophore XAD extracts was performed. We investigated pH differences in dependence of rare earths, to relate the siderophore production along the pH. Further, we investigated an adaptation method for manganese and rare earths to get more insights on bacterial metal adaptation potential.

Value of the data
• The data presented on siderophore production by Gordonia rubripertincta CWB2 and its optimization provide a strategy towards novel secondary metabolites. For the first time a rare earth element induced secondary profile was related to secondary metabolites for this genus. • Natural product scientists and in general microbiologists using GNPS and related tools will use these data, as the published dataset can be used for cross-referencing and identification of recurring molecules. • An overview on the desferrioxamine diversity allows fingerprinting of desferrioxamineproducing microorganisms. Further, production of certain derivatives enables evaluation of the promiscuity of involved biosynthetic enzymes or can be evaluated with focus on adaptation of the secreted metabolome towards changing environmental conditions. • The LC-MS/MS-based metabolomics datasets might contain a plethora of new and yet uncharacterized molecules, with interesting biological activities.
A siderophore relevant gene cluster was determined and therefore the production of siderophores by strain CWB2 was investigated. Therefore, we used a modifed CAS-assay [1] . To optimize the timescale for measurements a time-resolved decolorisation of chromazurol S (CAS) was recorded. This time-resolved CAS profile analysis at 30, 50, 65 and 80 h in dependency of different medium com positions showed highest siderophore production in M9 medium [3] with decoloration rates > 80% after 65 h ( Fig. 1 ).
Based on the evaluation presented in Fig. 1 we had selected a constant time point (after 4 h) for measurement of the percentage CAS-response. It was calculated on the difference of the absorbance of the full coloured blue Fe-CAS complex and a decolorized (yellow to reddish) sample at 630 nm. In dependency of the concentration the sample was diluted with M9 medium which served also as a blank. Desferrioxamine B (DFOB) served as a positive reference as it was commercially available as a pure siderophore and could be diluted in M9 medium in different concentrations to calibrate the CAS decolourisation to μM DFOB equivalents in the range of 1-500 μM. Data were fitted to an exponential decolourisation model function to obtain μM DFOB equivalents in μM ( Fig. 2 ). For optimisation of the siderophore production in dependency of medium composition CAS activities were measured and calculated to μM DFOB equivalents. Screening was done in 15 ml falcon tubes. Biological triplicates were measured at four time points. In Fig. 3 time point grouped boxplots of 4 different medium compositions are shown in which we applied an amino    [4] . Table 2 Extraction yields of siderophores. The XAD extractions were obtained from 3 different cultivations of 1 L each with different succinate concentrations during the cultivation period of 158 hours at 30 °C in M9 medium given in mM DFOB equivalents derived by CAS activity measurements and mg L −1 .

Production step
Units acid based medium (minimal medium supplemented with 4 g L −1 casamino acids) a minimal medium supplemented with MgSO 4 and CaCl 2 , the M9 medium [3] and finally a minimal medium supplemented with 5 mM FeCl 3 as a negative control with inhibited siderophore production. ANOVA analysis of the data showed that the M9 medium supplemented with vitamin B1 (thiamine; VitB1) and trace elements resulted in the highest DFOB CAS equivalents (85 μM) after 72 h cultivation, measured after 120 min CAS reaction time (see Fig. 3 ). Further optimisation of siderophore production was done in 1 L scale (biological triplicates) in the M9 medium in dependence of different concentrations of succinate. Succinate presented the most promising carbon source ( Table 2 ). OD measurement and CAS assays were performed to estimate DFOB production. Herein we observed 291 μM DFOB equivalents when Gordonia Table 3 Compounds detected in LC-MS/MS positive mode for 3 different purification strategies of siderophores from strain CWB2. The sum of the precursor intensity related to the number of observed parent mass counts for the siderophore extracts from Gordonia rubripertincta CWB2. As a reference the iron complexed desferrioxamine B was used in the LC-MS workflow. Replicates of samples were averaged and compared for FPLC purified, HPLC and C 18 concentrated, raw XAD and lyophilized crude extract were compared rubripertincta CWB2 was cultivated with 8 g L −1 disodium succinate. The yield of the XAD crude extract was 121 mg L −1 .
After having a robust production system for siderophores, the respective purification was established. Different methods of siderophore purification were applied. Collected fractions of HPLC and FPLC were analyzed by means of LC-MS. Therefore, 3 × 1 L cultures were extracted with XAD and eluted with 30 ml methanol, each. Extracts were further purified by C18chromatography, like described in the methods section or by means of FPLC purification [5] . Samples derived of these procedures were analyzed by LC-MS/MS. Table 3 shows the comparison of 3 different purification strategies, which show the increased precursor intensity of fraction 12 after C 18 -chromatopgraphy. FPLC purification can also be combined with desalting without significant loss of XAD crude extract contents, but without the need of further concentration steps. Lyophilization resulted in degradation of the Ferrioxamine E.
Stress often induces the production of secondary metabolites and was therefore investigated. Siderophore production under metal stress was studied for Gordonia rubripertincta CWB2 and Rhodococcus opacus 1CP on Fe-CAS agar plates supplemented with different metals (final conc. 2 mM). The amount of siderophores produced and secreted by a bacterium can be measured on such agar plates on base of halo formation. Therefore, siderophore halo formation was normalized on colony area and were calculated from image analysis derived pixel-areas. Pictures were taken after 6 weeks and were cropped interactively ( Fig. 4 ) for each plate and colony replicate (3) to create a cropping list. After cropping of the colony positions, double thresholding was performed interactively ( Fig. 4 ), firstly for the colonies and secondly for the siderophore produced halos around the colonies. Sum of filtered colony and halo-pixels were divided by the sum of filtered colony-pixels. The given ratio was plotted here in bar plots grouped by strain and metal, which was used to induce metal stress to investigate conditions, which trigger siderophore production. The herein observed results indicated siderophore triggering by Gd and Ge for strain CWB2 and by Ge, Ni and Pb for Rhodococcus opacus 1CP. The latter strain was used as comparative actinobacterium supposed to produce heterobactin siderophores. In addition, R. erythroplois S43 was used which was recently shown to produce heterobactin like structures upon As stress [6] .
In previous CAS assays the influence of manganese on CAS activity was not significant but a decreasing of growth in LB medium supplemented with MnCl 2 was observed. Gordonia rubripertincta CWB2 showed only poor growth on CAS agar plates, with only one colony grown after 6 weeks at RT. Therefore, we applied an adaptation procedure to manganese and erbium on LB and M9 agar plates. The colony, which showed highest growth and CAS activity ( Fig. 5 ) was separated and used for further experiments.
16S-rRNA gene sequencing of the V3-V4 region of the wild-type strain and their Mn/Er adaptation showed pure DNA of Gordonia rubripertincta CWB2 (data not shown). PCA analysis of the metal biosorption of the wild-type and the Er/Mn adaptation from an acidic multi-element solution in dependency of the elements Fe, Ga, V, Al, La Nd, Eu, Gd and Er, demonstrated clearly a significant effect (see Fig. 6 and Table 4 ). For the Er/Mn adaptation Gordonia rubripertincta CWB2   Table 4 ANOVA analysis of the target elements concentration in supernatants of multi-element solutions containing Fe, Al, Ga, V, and the rare earth's La, Nd, Eu, Gd and Er. Biosorption of 100 mg dry biomass of Gordonia rubripertincta CWB2 and its Er/Mn adapted variants were investigated (see Fig. 6 ). Significant differences between means are identified by ANOVA and fishers LSD test. Means with same letters are not statistically significant at α = 0.05.   increased metal concentrations of rare earth's La, Nd, Eu, Gd were measured in the eluates in comparison to the wild-type Gordonia rubripertincta CWB2. The pH in medium variations for cultures of Gordonia rubripertincta CWB2 and their selected adaptations were measured during cultivation in M9 medium in dependency of rare earth's (2 mM La, 2 mM Nd, 2mM Gd, 2 mM Eu, 2 mM Er), casamino acids and iron limitation to relate siderophore production to pH values ( Fig. 7 ). All cultures increased the pH in the media up to pH 9. For cultivation with rare earths and amino acids the changes in pH were observed after 72 h and do not differ significantly between the 2 variants of Gordonia rubripertincta CWB2.

Fig. 8. TMOS entrapment as a metal immobilizing screening for metallophore crude extracts used in a metal desorption study involving batch adsorption of metals onto metallophore functionalized TMOS powders from XAD crude extracts of Gordonia rubripertincta CWB2. Box plots of 4 replicates were visualized and shown for Ga, V, Nd, and Mo concentrations in μg/ml eluate (500 mM EDTA/ DTPA). Significant differences between means are identified by ANOVA and fishers LSD test. Means with
same letters are not statistically significant at α = 0.05.
The reference groups without additional metal stress showed that the adaptation reached earlier higher pH values than the wild type. Including the supplementation with 2 mM rare earths during cultivation the time scale difference reached the maximum. The wild type Gordonia reached the pH maximum pH 9 after 15 h, the adaptation after 19 h. The same effect was observed for the control groups (M9 medium supplemented with iron).
Besides siderophore production we investigated their applicability. Here the main issue is stabilization and reuse. Therefore, the immobilization of siderophore samples was studied. Metallophore containing XAD extracts were entrapped in tetramethoxysilane (TMOS) and grounded to powder. Metal adsorption/desorption on the functionalized carrier was performed in a model solution, which contained V (931.1 ± 74.8 μg mL −1 ), Ga (76.4 ± 7.4 μg mL −1 ), Mo (1123 ± 86.7μg mL −1 ), and Nd (558.3 ± 113.7 μg mL −1 ). After shaking of 20 ml with 1 g functionalized carrier the samples were centrifuged, and the metals adsorbed onto the carrier pellet were eluted with 500 mM EDTA/DTPA solution. The element concentration in the eluate was measured by ICP-MS. With the 100 μM DFOB-TMOS a significant effect for Ga (126.7 ± 12.9 μg mL −1 ) with a concentration factor of about 10 in comparison to the reference blank carrier material (13.1 ± 0.3 μg mL −1 ) was observed. For vanadium, the highest concentration factor was observed for the C 18 -derived concentrate (see above) from Gordonia rubripertincta CWB2. For neodymium, all metallophore-TMOS materials show significant increases. For molybdenum especially the extracts of Gordonia rubripertincta CWB2 show significant enrichment of the target metal in comparison to the reference with about 2-fold increase (see Fig. 8 and Table 5 ).
Siderophores from XAD extracts were enriched by HPLC fractionation and identified by LC-MS/MS measurements in positive and negative mode combined with molecular networking, using the global natural products social molecular networking (GNPS [7] ) platform ( Fig.s 9 and  10 ; Tables 6 and 7 ). The methods were adapted from previous works [ 8 , 9 ]. Parent masses and fragment spectra were compared to online libraries (GNPS [7] , MetFusion), to identify desferriox-  amine and bisucaberin siderophores as well as citrate. Through molecular networking molecules were identified with fragmentation patterns similar to the desferrioxamines, which are likely to be derivatives and analogues [9] . Overall, 28 desferrioxamine-like molecules were found in positive mode and 14 in negative mode. Annotation of fragment spectra onto molecular structures was guided by in silico fragmentation, using MetFrag [10] . Only 5 of the siderophore-like molecules were identified by data mining and comparing with library entries. The others, here with fragmentation spectra similar to desferrioxamines, remain to be characterized, regarding their structure and differences in metal affinity. We investigated the secreted metabolomes of wild-type Gordonia rubripertincta CWB2 (WT) and erbium/manganese preadapted Gordonia rubripertincta CWB2 (M) ( Fig.s 11 and 12 ). Even if no rare earths such as erbium were added to the cultivation media, the secreted metabolome of the wild-type diverged from the adapted variant ( Fig. 11 ).
The addition of 2 mM erbium chloride strongly impacted the secreted metabolome of wildtype Gordonia rubripertincta CWB2 (WT) as well as the adapted variant (M) ( Fig. 12 ). Comparing overall metabolomic profiles (as presented in Fig. 11 ) of positive and negative mode measurements by cosine similarity, profiles of WT against M + Er and WT + Er against M show the greatest differences. WT + Er com pared to M + Er and WT against M are the most similar ones ( Fig. 12 A). Both strains produced a similar diversity of compounds under + /-Er-conditions ( Fig. 12 B), while the addition of erbium seems to induce the production of more compounds measurable in positive mode, and less compounds in negative mode ( Fig. 12 C). Overall the preadaptation to Er/Mn as well as the addition of 2 mM erbium chloride to the cultivation media affected the secreted metabolome, as for example the maximum m/z of secreted compounds was greater in the absence of erbium ( Fig. 12 D/E).

For genome mining
For genome mining we applied the tool antiSMASH 4 and 5 [2] , for the investigation of the existing siderophore biosynthetic gene clusters in the genome sequence of Gordonia  [11] ). The genome to product analysis tool MIBiG [12] was used to obtain a siderophore product prediction.

Production of siderophores from Gordonia rubripertincta CWB2
For production of siderophores from Gordonia rubripertincta CWB2 (DSM 46758) M9 minimal medium was used as described previously [ 4 , 11 , 13 , 14 ]. The minimal medium was supplemented with a trace element solution [3] and 4 g L −1 succinate as carbon source. A sub-cultivation was done with 3 g L −1 freshly grown LB biomass, which was resuspended in 100 mL iron-depleted medium and incubated for 55 h at 120 rpm and at 30 °C. The sub-cultivated biomass was again harvested by centrifugation (yield of 4 g L −1 ) and finally transferred (0.4 g L −1 ) in new 3 L Erlenmeyer flasks comprising 1 L freshly prepared iron-depleted M9 minimal medium. For screening of siderophore production in dependence of different medium compositions, washed cell mass of strain CWB2 was inoculated at a final OD 595 of 0.05 in 7 ml medium which contained the certain medium compounds, like the basal salt solution (further called minimal medium), the trace element solution and CaCl 2 , and MgSO 4 (further called M9 medium). To optimize time scale for the CAS assay and its incubation time, kinetic time profiles were analysed for 4 h with a frequency rate of 1 scan per 4 sec.

Isolation of crude extracts by solid phase extracted with XAD
For isolation of the solid phase extracted XAD crude extracts 3 × 1 L of in M9 medium cultivated bacteria (0.3 g L −1 inoculum, 4 g up to 12 g succinate as a carbon source 158 h at 30 °C) were used. Culture biomass was separated by centrifugation; a solid-phase extraction was applied to the supernatant. For this, a 10 g L −1 mixture of 5 g L −1 XAD4 and 5 g L −1 XAD16 were washed with water, methanol, and another time with water. The activated resin was added to the supernatant and shaken at 100 rpm at 4 °C for 12 h. The supernatant was decanted, and the resin was washed with 1 L ddH 2 O L −1 medium. Elution of the siderophores was done with 100 mL methanol per 10 g of resin. The procedure was repeated three times to ensure optimal yield. The methanolic crude extract (2 L) was concentrated under reduced pressure ( < 30 °C) to a tenth of volume (200 mL). For further purification and application, the methanolic extracts were diluted with 100 ml H 2 O to separate the MeOH from the extracts in a rotary evaporator under reduced pressure until only the water remained.

Mn/Er adaptation of Gordonia rubripertincta CWB2
For Mn/Er adaptation of Gordonia rubripertincta CWB2 a freshly grown colony from a LB-agar plate was incubated at 30 °C and shaken at 150 rpm overnight in 20 ml LB medium supplemented with 2 mM MnCl 2 . A 10 −6 -fold dilution of these precultures was than plated onto 10 LB-agar plates supplemented with 5 mM ErCl 2 . 48 selected colonies, which were observed earliest, were transferred into visual analysable raster positions on a new Er-LB agar plate. From that plate, grown colonies were transferred to a CAS agar plate as described earlier [ 1 , 4 ] (see Fig. 5 ) to select adapted variants with better growth and increased siderophore activity, further called Mn/Er adaptation of Gordonia rubripertincta CWB2.

Investigation of metal biosorption by Gordonia rubripertincta CWB2
For the investigation of metal biosorption by Gordonia rubripertincta CWB2 and Mn/Er adapted cells were cultivated in 250ml LB and M9 medium as described above, harvested by centrifugation and lyophilized. Cell pellets were weighed, aliquoted in 100 mg portions, and supplemented with 2 ml of a multi element solution containing 47 elements each at 2 mM. After vortexing the solution samples were shaken for 2 h at 30 °C and 150 rpm. Cells were harvested by centrifugation. 100 mg of each sample was first mixed in a sample tube with 200 μl ultrapure water and 1.9 ml 65 % nitric acid. The samples were then mixed with 600 μl of a 4.8 % hydrofluoric acid and digested in the microwave (MLS-ETHOS plus) with a temperature program -heating to 200 °C in 25 minutes, temperature maintained for 5 minutes, then continuously cooling to 75 °C within 30 minutes. After heating, the samples were left to cool for about 2 hours. The completely dissolved samples were transferred into 15 ml sample tubes. Element concentrations in diluted samples (1:10) as well as solutions from the supernatant were measured by ICP-MS (xseries, Thermo Scientific) using 10 μg/l rhenium and rhodium as internal standard according to [15] .

Measurement of time dependent pH profile by PresSens
For the investigation of differences in the pH profile of Gordonia rubripertincta CWB2 and its Er/Mn adapted variants we investigated growth during cultivation in a microplate system (24 wells) supplemented with a pH sensor (PreSens system by PreSens Precision Sensing GmbH).

Entrapment of siderophores in tetramethoxysilane (TMOS) sol gel beads
For embedding the siderophores in sol gel beads a nitric acid catalysed hydrolysis and condensation reaction of tetramethoxysilane (TMOS) and H 2 O is applied. For sol gel beads of 2 ml the reaction was set up in 24 well plates. A reactive mixture of 50 ml TMOS and 1.5 ml 10mM HNO 3 (33.3:1 v/v) was stirred at room temperature (23 °C) for 10 minutes in a 100 ml beaker. Afterwards 1.4 ml of this reactive mixture, 300 μl methanolic siderophore extract and 300 μl 100 mM phosphate buffer at pH 7 were mixed in each well. Further the plate was closed with an adhesive foil and pierced at each well with a pin to allow methanol fumes to vent. To reduce reaction speed temperature was set at 20 °C or lower. Therefore, the plate was stored on ice in a polystyrene box for at least 5 days until the beads hardened. For further adsorption and desorption experiments of metals the sol gel beads were produced for each strain and beads without siderophores were produced as blank controls. Here data were collected and provided as supplemental material as supporting data sets (see suppl. material.).

Screening of metal adsorption/desorption studies batch experiments
For screening of metal adsorption/desorption batch experiments were performed in five replicates with the crushed immobilised metallophore powder. After 6 hours of shaking of 1 g carrier material, which refers to 10 mL of crude extract from 1 L of culture supernatant, at 120 °rpm in 20 mL model solution, the metal analysis was carried out. The samples contained the following metals (see Table 8 ).
The carrier material was separated by centrifugation and treated with 1 ml 500 mM EDTA to eluate the metals from the carrier. Each sample and a blank of the materials without biomolecules were acidified with 0.1 M HNO 3 (10% v/v) filtered through a 0.22 μM filterequipped syringe and prepared for ICP-MS measurement to obtain metal concentrations after the extraction with immobilized metallophores.

Modified CAS agar preparation and image analysis of CAS Halos
We used the modified CAS agar procedure, a variant of the original solid CAS agar assay elsewhere [ 1 , 4 ] in order to screen for potential new compounds of metal stress resistant bacteria. In addition to iron, the metals Gd, Mn, Ger, Co, Ni, Pb, are used in final concentrations of 100 μM in order to analyze the resulting CAS agar halo. The resulting CAS performance was determined with standard image processing tools from pictures taken at the end state of experiments after 5 weeks. Specifically, the CAS_area in dependence of the colony_area was determined by all pixels of the halo_area nH and the colony_area related to the pixels of the colony_area ( Figs. 13 and 14 ) and was used as an indicator for the performance of metal chelation under metal stress and iron limited conditions. The areas for halo and colony were determined using 2 interactively adjusted threshold operations on the input grayscale images.
Through visual analysis of the images, it was found that the background, the colony and the halo are located in different parts of the grayscale color range. The region of the colony is determined by all pixels with grayscale value smaller than tC, and the region of the halo is given by all pixels with grayscale value larger than tH. Firstly, the user is asked to crop the image into smaller parts containing single colonies, whereby the cropped image contains a border area with background. Second, for each one cropped sub image a histogram is plotted, which is used to determine the thresholds tC and tH. The procedure is repeated iteratively, and the output was verified visually. At the end of the procedure, the area of the halo and the colony regions are given by the sum of the pixels. We calculated from three colonies the average and the standard error. Significant differences between treatment groups were identified by the Kruskal Wallis test followed by a Dunn-Bonferroni post hoc method and marked with an asterisk ( * ), when they showed significant difference to the standard iron-CAS agar plate without additional metals ( Fig. 4 ).
Molecules were selected for fragmentation through CID with argon in mass spectrometry, when the intensity exceeded 60 0 0 counts/s. Data were evaluated as described earlier [ 8 , 9 ]. In brief, files were converted from waters raw to .mzXML using Proteowizard (ver. 3.0.9490) [16] . Subsequently, data were uploaded to the GNPS servers, the METABOLOMICS-SNETS workflow was applied [7] , and visualized via Cytoscape (ver. 3.3.0). The annotation of fragment spectra was aided by MetFrag [10] . Overall this will help to identify more valuable metal chelating molecules useful in numerous future applications [17] .

Declaration of Competing Interest
The authors declare that there are no conflicts of interest.