Iron Coordination Properties of Gramibactin as Model for the New Class of Diazeniumdiolate Based Siderophores

Abstract Gramibactin (GBT) is an archetype for the new class of diazeniumdiolate siderophores, produced by Paraburkholderia graminis, a cereal‐associated rhizosphere bacterium, for which a detailed solution thermodynamic study exploring the iron coordination properties is reported. The acid‐base behavior of gramibactin as well as its complexing ability toward Fe3+ was studied over a wide range of pH values (2≤pH≤11). For the latter the ligand‐competition method employing EDTA was used. Only two species are formed: [Fe(GBT)]− (pH 2 to 9) and [Fe(GBT)(OH)2]3− (pH≥9). The formation of [Fe(GBT)]− and its occurrence in real systems was confirmed by LC‐HRESIMS analysis of the bacteria culture broth extract. The sequestering ability of gramibactin was also evaluated in terms of the parameters pFe and pL0.5. Gramibactin exhibits a higher sequestering ability toward Fe3+ than EDTA and of the same order of magnitude as hydroxamate‐type microbial siderophores, but smaller than most of the catecholate‐type siderophores and much higher than the most known phytosiderophores.

Despite these seminal discoveries, the presence of the N-nitroso-N-hydroxylamine grouph as already been reported in other natural products( some examples are reported in Ta ble S1). [9][10][11][12][13][14][15] Interestingly,s everalo ft he reported diazeniumdiolates presenth igh biological activity,a sa ntibacterial, antiviral, antifungal or even antitumoral agents, which seems to be directly related to the chelating ability of these compounds. [9,[12][13][14][15][16][17][18][19][20] This is the case with fragin, for example, for which C. Jenul et al. have demonstrated that metal chelation is the molecular basis for its observed antifungal activity. [15] Likewise, the action of dopastin as ad opamine b-hydroxylase inhibitor is due to its interaction with the copperi on present in the active centero ft he enzyme. [12] In addition, nitrosoxacins inhibit 5-lipoxygenase, an iron-metalloenzyme that is involved in the biosynthesis of lipid mediators in inflammation processes. [13,14,20] The observed strong dependence between biological activity and metal ion interactions for N-nitroso-N-hydroxylamine compounds along with the recent discoveryo fg ramibactin acting as as iderophore generates af undamentali nterest in its complexation properties. Furthermore,p roviding relevant benchmarks for chelation will set the foundations for future interest in N-nitroso-N-hydroxylamine derivatives as startingp oints for the development of new compounds with potentialp harmaceutical application. Remarkably enough,t o our knowledge,t he binding ability and/ort he chemical speciation of bacterial/natural N-nitroso-N-hydroxylamine compoundst owardm etal ions as well as the stability of their complexes have never been investigated in detail.
In this work, we used at hermodynamic approacht og ain insight into the details of iron sequestration by gramibactin. We aim to contribute to ab etter comprehension of the iron acquisitiona nd uptake mechanisms and the role of this N-nitroso-N-hydroxylamine group in siderophore. Along this line, the complete characterization of the chelating ability of grami-bactin toward iron(III) in aqueous solution over aw ide pH range (2.0 pH 11.0) is reported.T he studies were carried out by potentiometric and spectrophotometric titrations in aqueous KCl solution and the presenceo ft he relevant irongramibactin species was confirmed by mass spectrometry.
The thermodynamic data obtainedh ere establish the basis for ad etailed assessment of gramibactin as bacterial siderophore, especially in comparison to other well-known classes. This assessment is carriedo ut on the basis of sequestration parameters such as pL 0.5 and pM, with the latter being well known among al arge scientific community from different fields. Since these parameters, in particularp M, are often used by researchers that are not regularly involved in their determination or who are not alwaysa ware of the correct conditions to be considered in their calculation process,w eh erein report ad etailed explanation of how and in which conditions they should be determined. With the presented data andd etailsi n this paper we aim to promote the correctu se and evaluation of speciation parameters and to avoid future misleading results and comparisons. This description is accompanied with an exhaustived eterminationo fb oth parameters for relevantc ompounds such as different diazeniumdiolates, phytosiderophores,and relevant siderophores, as wellasE DTA, ac ommonly used chelating agent in biological systems.

Results and Discussion
Acid-base properties of gramibactin The acid-base behavior of gramibactin was studied and protonation constants were determined experimentallyb ys ystematic potentiometric titrationsu nder defined conditions of tem-perature, medium, and ionic strength (i.e., I = 0.1mol dm À3 in KCl (aq) and T = 298.15 AE 0.1 K).
Gramibactin carries severalf unctional groups that can be involved in various protonation/deprotonation equilibria, namely two hydroxyl groups, one of which together with ac arboxylic group is part of an a-hydroxocarboxylate moiety,a nd two Nnitroso-N-hydroxylamine groups. Four protonation steps can be observed within the investigated pH range (2 pH 11) and the corresponding protonation constants are summarized in Ta ble 1.
Given the molecular structure and based on the reference of similar systemsr eported in several thermodynamic databases, [21][22][23] the carboxylic group can be expected as the last to be protonated,w ith an observedl og K value of 2.27, whereas the hydroxyl groups should be protonated first. In the case of gramibactin, only the protonation of the hydroxyl group of the a-hydroxocarboxylate moiety could be determined (log K = 10.94), which nicely corresponds to literature values reported for Rhizoferrin (11.3 and 10.05) that also contains a-hydroxocarboxylate moieties. [24] The deprotonationo ft he second hydroxyl group which is not acidified by neighboring functional groups is expected to occur at pH values greater than 11, which is above the investigatedp Hr ange. Consistent with reported protonation constants for reference compounds containing the N-nitroso-N-hydroxylamine moiety, such as dopastin (log K = 5.2) [12] and nitrosofungin (log K = 5.1) [12,13] (see Ta ble S1 for structures), the processes observed at log K values of 5.71 and4 .87 can be assigned to the protonation of the two diazeniumdiolate groups (cf. SchemeS2). The distribution diagram of the gramibactins pecies with varyingp rotonation state based on the determined protonation constants is depicted in Figure 1.
The monoprotonated gramibactin [H(GBT)] 3À is the major speciesa tn eutral to basic pH values (7 < pH < 9), while the fully deprotonateds pecies[ GBT] 4À starts to form at pH values above 9. Thet ri-and tetraprotonated gramibactin species mainly occur at pH valuesb elow 5. The fully protonated, neutral gramibactin[ H 4 (GBT)] only exists at very acidic pH values (with ca. 20 %a tp H% 3). From these results,i ti se vident that gramibactin is mainly present as an egatively charged compound in the soil microenvironment, not only within the normalp Hr ange used for culturing plants (4 pH 10.5), [25] but also in acidic soils (pH 5.5). [26] The data obtained independently by spectrophotometric titrations not only confirmed the protonation constantsd erived from potentiometric titrations( see Table 1), but also allowed the spectrophotometric characterization of the individual species. Actually,t he molar attenuationo fa ll gramibactin species [H r (GBT)] rÀ4 (r = 0-4) could be determinedf rom fittingt he experimental data on the basis of the given equilibrium model using the HypSpec program. [27] The corresponding spectra are depictedi nF igure 2.
The spectrum of the fully protonated form of gramibactin [H 4 (GBT)] is characterizedb yab and with l max at 225 nm (effi1650 m 2 mol À1 ). As anticipated, the spectrum is virtually unchanged for the monod eprotonated form [H 3 (GBT)] À ,s ince this is related to the deprotonationo ft he carboxylic proton of the a-hydroxocarboxylate moiety.H owever,f or the doubly deprotonated species [H 2 (GBT)] 2À as light bathochromic shift of the former band and the appearance of an additional shoulder at about 250 nm is observed. For the triply deprotonated species [H(GBT)] 3À the former band at 225 nm disappears and the   previouss houldern ow appears as an intense band at 246 nm. This observation is consistentw ith the assignment of the latter two (second and third) deprotonation steps to the N-hydroxylamine protons. The fourth deprotonation step assigned to the proton of the hydroxyl group of the a-hydroxocarboxylate moiety leaves the spectrum for the corresponding species [(GBT)] 4À virtually unchanged (l max = 246 nm, effi2260 m 2 mol À1 ). The observed behavior can be rationalized based on the literature-known protonation-associated variation of the optical properties of the N-nitroso-N-hydroxylamine group. The protonatedf orm is characterized by an intense, broad asymmetric UV absorption band at about 229 to 232 nm, which is assigned to a p-p*t ransition, whereas upon deprotonation this absorption band undergoes ab athochromic shift and appearsi nt he range from 244 to 258 nm with as light increase in the molar attenuation coefficient. [19,[28][29][30][31] Consequently,t he present spectra support the deprotonatione quilibria depicted in Scheme S2.

Speciationoft he iron(III)-gramibactin system
For the determination of the stability constants of the iron(III)gramibactin species, two importanta spects need to be considered for an adequate description of the system.
At first, this concerns the sole speciation of iron(III) in aqueous solution, that is, the iron(III) hydrolysis. In fact, for strong Lewis acids such as iron(III), the acid-base properties (i.e. hydrolysis)m ust be taken into account for ac orrect speciation model  4 ] 5 + species, since they seem to be the most reliable and appropriate to describe iron(III) speciation at variousc oncentration levels. [32][33][34] As the formation of the polynuclear species [Fe 12 (OH) 34 ] 2 + could also be relevant even at low concentrations, [21,23,[32][33][34] the effect of its inclusion/exclusion in the speciation model of our system was investigated by performing the calculationsi nb oth ways. The corresponding values used in this work were adapted from refs. [32][33][34] and are summarized in Ta ble S2 of the Supporting Information.
The second aspect, that needs to be considered here, becomes importantf or ligands with as trong chelating ability toward the metal ion under investigation. For iron(III)-siderophore systems, formation constants higher than 10 15 -10 20 can be expected, which resultsi na na lmost complete shift of the formation equilibrium given in Equation (1) toward the product side.
In such cases, the classical procedure to determine stability constantsbyp rotondisplacemente xperiments (acid-base titrations by exploitingt he above-reported equilibrium) utilizing the reactiono ft he ligand with the relevant metal ion cannot be performed properly,s ince practicallys olely the product complexi sp resent as soon as metal ion and ligand solutions are mixed.Asuitable strategy to overcome this problem is to apply the ligand-competition method. [35,36] Following this procedure, the complexation behavior of gramibactin toward iron(III) in aqueous solution was studied by potentiometry using EDTAa sc ompeting ligand.A long this line, severalm easurements were performed at variousm olar ratios between gramibactin, EDTA, and iron(III). The corresponding evaluation of the experimentald ata was based on the EDTAp rotonation and EDTA-iron(III) complex formation constantst aken from literature and summarized in Ta ble S2. [32][33][34]36] The analysiso fafull set of potentiometric data obtained on the basis of the abovementioned aspects clearly evidenced the formation of two [Fe(GBT)H r ]s pecies within the investigated pH range (2 pH 11), namely [Fe(GBT)] À and [Fe(GBT)(OH) 2 ] 3À , whose stability constantsare reported in Ta ble2.Although surprising at first glance, the lack of ad etectable stepwise deprotonation,i .e.,t he detection of a[ Fe(GBT)(OH)] 2À species, is not unexpected, as such multistep protonation processes are not uncommon for cases involving strong multifunctional ligands and/or highly hydrolysable cations. [37,38]    iron(III)-gramibactin system,aspeciation diagram is depicted in Figure 4. The [Fe(GBT)] À speciesi st he only iron species present in the pH range from 2.0 to ca. 9.0. Above this pH value, the formation of the corresponding dihydroxido species[ Fe(GBT)(OH) 2 ] 3À occurs. In any case, all iron(III) is fully complexed by gramibactin within the investigated pH range. Worth mentioning is also the fact that the inclusion of the polynuclear speciesF e 12 (OH) 34 in the modeld id not affect the results. Indeed, taking this species into account in the calculations, we obtainedl og b 110 = 27.64 AE 0.05 and log b 11-2 = 6.37 AE 0.09 for[ Fe(GBT)] À and [Fe(GBT)(OH) 2 ] 3À ,r espectively,w hich is, within the experimental error,c oincident with the values obtained not considering Fe 12 (OH) 34 in the model (see Ta ble2).
The suitability of EDTAa sc ompeting ligand with gramibactin toward iron(III) complexation is demonstrated by the speciation diagram depicted in Figure 5f or the iron(III)/EDTA/gramibactin system considering a1 :1:1 ratio and an iron(III) concentration of c Fe = 1mmol dm À3 .U nder these experimental condi-tions and at ap H% 2, ca. 60 %o ft he iron(III) is coordinated by EDTA, while only ca. 40 %i sc oordinated by gramibactin. This allows to monitor the formation of the [Fe(GBT)] À species, that reachest he maximum concentration at pH % 5.0, and consequently gives the possibility to accurately determine its formation constant using the ligand-competition approach. Moreover,t he speciation diagram in Figure 5c an also be useful to assess the distribution of iron(III) under conditions that are frequently observed in bacterial growth media, where EDTAi s used as chelatinga gent to keep iron in solution. [40] The resultso btained by potentiometric measurements were confirmed by UV/Vis spectrophotometry (see Figure S1). The analysiso ft he experimental data obtainedb ys pectrophotometrict itrationsu sing the HypSpec program [27] for fitting al-  [19,30] Moreover,u pon coordination of the gramibactin ligand,a na dditional band for the [Fe(GBT)] À species is observed at 343 nm ( Figure S2), which can be assigned to a ligand-to-metal charget ransfer between the a-hydroxocarboxylate hydroxy oxygen donor and iron(III) ion. [24] In the case of the dihydroxido species[ Fe(GBT)(OH) 2 ] 3À ,t he latter ligand-tometal charge transfer band disappearsa nd an ew band is found at around 280 nm.   Based on these observations, it is tempting to attributet he observeds pectralc hanges between the two iron(III)g ramibactin species, i.e.,[ Fe(GBT)] À and [Fe(GBT)(OH) 2 ] 3À ,t ot he loss of the a-hydroxocarboxylate moiety from the ferric center,f or which the octahedral coordination sphere is completed by two hydroxido ligands.
Sequestering ability of gramibactin and its pH dependence As emphasizedb efore, ag eneral interesto ft his study is to provideacomparative assessment of the sequestering ability of gramibactin as ar epresentative example of the new class of diazeniumdiolates iderophores based on its ability for the sequestration of iron. This is fundamental to understand the role of gramibactin in the soil microenvironment, particularly in view of the competitive presence of other chelators. Several compounds have been reported as siderophores for mobilizing and/or transporting iron, such as nicotianamine in higher plants. [41] Other specificp hytosiderophores( PS) are secreted from the roots of graminaceousm onocotyledonousp lants during iron deficiency,l ike mugineic acid (MA) and its derivative, deoxymugineic acid (DMA). [42] Furthermore, al arge number of microbial siderophores (MS), ranging from catecholate-type (e.g.,e nterobactin, amonabactin T) to hydroxamatetype (e.g.,d esferrioxamine B, Ea nd desferriferrichrome), are also present in all soil environments. [43] Therefore, it is important to evaluate and comparet he sequesteringa bility of gramibactin and other siderophores toward iron.H owever,t he simple comparison of the iron-siderophore formation constantsi si nm ost cases not sufficient to address this question.I ndeed, other factors need to be considered, such as differences in the denticity of the ligands, in their coordination modes, and in their acid-base properties (protonation reactions are competitive with respectt ot he metal complex formation,s ince hydrogen ions compete for the same binding sites). [44,45] With all this in mind, severalp arameters have been defined in ordert ocompare the relative strength of differentm etal chelating agents. [44,46,47] For the classification of the sequestering ability of gramibactin we will make use of two specific parameters, namely the pM value (defined as pM = Àlog[M],w here [M],i nt he particular case of iron(III), represents the concentration of the free aqueous ion [Fe(H 2 O) 6 ] 3 + when c L /c M = 10, c M = 10 À6 mol dm À3 ,p H7.4) originally introduced by Raymondp articularly for the comparison of ironsiderophore systems [46] and the semiempirical parameter pL 0.5 (representing the total ligand concentration required to sequester 50 %o ft he metal cation under the given conditions of the system), whichi st he result of efforts to establish ap arameter that is easyt ou se and less susceptible to misinterpretations. [44] Although the two parameters are defined by the concentration of different speciesi nt he relevant solution equilibria (i.e.,m etal ion in case of pM and ligand for pL 0.5 ), for both cases al argerv alue indicates as trongers equestering ability of the ligand under investigation.
Before we start to evaluatet he particular numbers, ac aveat has to be placed here. The generally perceived benefit of the pM value, which leads to its widespread use, is mainly related to its intuitive definition. However,t his apparent advantage comesa tt he expense of frequent misuse of this parameter, particularly by neglecting Raymond's originalr ules andd efinition (c L /c M = 10, c M = 10 À6 mol dm À3 ,pH7.4). [46] Therefore, we recalculated the pM values for all relevant chelators, [1,[48][49][50] even if they have been reported in literaturef or some cases,t oa void any comparison with practically inconsistentn umbers. For a detaileda nd educational discussion on this topics ee the Supporting Information and for example refs [44] and [47].A tf irst, we will focus on the pM value, as this is one of the most frequently applied metric parameters that is also used for other biologically and/ore nvironmentally relevantm etal ions such as gallium(III) and copper(II), although it was specifically introduced for the comparison of iron-siderophore systems. [46] In any case, al arger pM value corresponds to al ower concentration of the free metal ion in solution at equilibrium and, in principle, to ah igher affinity of the relevant ligand for the metal ion studied.C onsequently,t he pFe value for the iron(III)gramibactin system has been calculated (c L /c M = 10, c M = 10 À6 mol dm À3 ,p H7.4) and is depicted in Figure7as ag raphical comparison with the pFe values of different other classes of siderophores. The values recalculated in this work are based on the respective protonation and iron(III) complex formation constantst aken from the original publications (see Table 3), using the iron(III) hydrolysis constantss ummarized in Ta bleS2. This is particularly important in order to obtain ac omparable set of pFe values, since such ac omparison is only rigorously valid if all calculations have been performed following the same rules and applying the same conditions. The pFe value for gramibactin (25.0) is of the same order of magnitude as for known hydroxamate-types iderophores such as desferrioxamine E( 27.5), desferrioxamineB(26.5) and desferriferrichrome (25.3).H owever,e nterobactin (34.3),acatecholate-type siderophore, presentsamuch higheri ron(III)c helating efficacy.R egarding the phytosiderophores mugineica cid, deoximugineica cid, or nicotianamine, the pFev alue of gramibactin is 9l og units larger (25 vs. % 16), while in comparison  with some mono-diazeniumdiolate ligands, such as N-nitroso-N-methyl-hydroxylamine or N-nitroso-N-phenyl-hydroxylamine (cupferron), the pFe value of gramibactin is even 18 log units larger.T he observed effects reflect the differences in the coordinating groups as well as the number and type of donor atoms present in the ligand. The former is particularly evident, as both gramibactin and the phytosiderophores are hexadentate ligands. In fact, the diazeniumdiolate moieties presenti n gramibactin show ac onsiderably better chelating affinity for iron(III) than the functional groups of the phytosiderophores (see Scheme S1). On the other hand, when comparing the chelating ability of ligands bearing the same chelating groups (Nnitroso-N-hydroxylamine), it is observed that the denticity of the ligand is now the dominant effect. This is obvious from the comparison of gramibactin containing two diazeniumdiolate moieties, while in N-nitroso-N-methyl-hydroxylamine and N-nitroso-N-phenylhydroxylamine( cupferron) only one chelating group is available to coordinate the metal ion. For the latter cases, 1:3m etal to ligand species are formed, whereas for gramibactin a1 :1 complex is found. Consequently,t he difference in pFe value (25 vs. 7) can be roughly attributed to the lower stabilization of the bidentate ligand due to the loss of the chelate effect. [54,56,57] Finally,i ti sa lso worth comparing the pFe value of gramibactin with that of EDTA, as the latter is an integral part of most growth media used in laboratory studies, which shows that chelation by gramibactin is thermodynamically favored in terms of the pFe value by about 1.5 log units.
In order to overcome the not insignificant drawbacks of the pM value (see Supporting Information), the parameter pL 0.5 has been introduced, whichw ill be used in the followingt oc ompare the sequestering ability of gramibactin with the relevant types of siderophores already mentioned.I nc ontrast to the pM value, the semiempiricalp arameter pL 0.5 does not refer to the concentration of the free metal ion, but rather represents the total ligand concentrationn eededf or sequestration of half of the metal cation (present as trace) under the given conditions of the system investigated. [44] In the way the parameter pL 0.5 was conceived and is being used in the calculations, it represents the effective sequestering ability of al igand and can be used to make all kinds of possible comparisons (for further details see Supporting Information).
The pL 0.5 values for the iron(III) sequestration calculated for gramibactin and the other relevant ligands at pH 7.4 are summarized in Ta ble 3. To address the situation in the soil microenvironment, the pH profiles of the pL 0.5 values for gramibactin, EDTAa nd the two phytosiderophores mugineica cid and nicotianamine were also calculated and depicted in Figure 8( cf. Ta ble S4).
The obtained pL 0.5 values show basically the same trends as observed forthe pFe values (cf. Figure 7and Table 3). Although the numbers of both parameters cannot be directly compared, it is generally observedt hat the numbersc alculated for the pL 0.5 valuesa re about 10 log units smaller than the correspondingp Fe values,w ith the simple bidentate diazeniumiolate ligands being the only exception.T his leads to the rather Table 3. Protonation and iron(III) complex formationc onstantso fg ramibactin, EDTA, selected diazeniumdiolates and siderophores, andcorresponding pFe and pL 0. 5  surprising effect that the pL 0.5 values fort he iron(III) sequestration calculated of these bidentate diazeniumiolate ligands are in the same order of magnitude as those of the examples considered here for the class of phytosiderophores (mugineic acid, deoximugineic acid, and nicotianamine).
The same basic trends are observed for the calculated pL 0.5 values upon variation of the pH when gramibactin is compared with EDTA and the two phytosiderophores mugineic acid and nicotianamine (see Figure 8). For all pH values considered, gramibactin is am arkedlys tronger chelating agent for iron(III) than the investigated phytosiderophores. Interestingly, even EDTAi sastronger iron(III) chelator with respect to the phytosiderophores. Compared to EDTA, however,g ramibactin exhibits am uch higher sequestering ability at mosto ft he pH values under consideration. For example, at ap Ho f7 .4, ap L 0.5 value that is 1.6 log units larger (gramibactin:1 5.5 vs. EDTA: 13.9) means that under the same experimental conditions a 40-fold lower concentration of gramibactin is required fort he chelation of 50 %o ft he iron(III) in solution. In general, the ability of gramibactin to mobilize iron(III) is higher in the pH range from 4.5 and 6.5 (pL 0.5 values of 16.6 and 16.4, respectively) with am aximum pL 0.5 value of 17.0 at pH 5.5 and significantly decreases down to av alue of about8at very high pH values. In addition, the generally observed decrease of sequestering ability of all four chelators analyzed at high pH values is due to competition with hydroxide ions, i.e.,t he iron(III) hydrolysis. Nevertheless, both gramibactin and EDTA can compete in a more efficient way with the hydrolysis than the phytosiderophores,p resenting ap L 0.5 of about 11 at pH 9.5 vs. ap L 0.5 of less than 2f or mugineic acid.
In fact, the pronounced efficiency of iron(III) chelation by gramibactin reinforces the hypothesis that numerousg raminaceous plants can exploit this microbial siderophore as iron source.M oreover, sequestration with gramibactin stabilizes iron(III) and thereby prevents its reduction to iron(II), which is a commonp rocess occurring in soil with low pH due to the presenceo fh igh concentrationso fa naerobic bacteria. This fact is important since it is known that iron(II) acts as an antag-onistic elementonthe uptake of essential nutrients (e.g.,phosphorous, potassium, and zinc) by plants. [26] From all these observations, either considering the pL 0.5 or pFe values, al igand-exchange reaction of gramibactin and, for instance, mugineic acid (MA), is not expected to occur.I ndeed, the formationo ft he complex [Fe(MA)] À can only be relevant for an extremelyl arge excess of the mugineica cid (c MA /c GBT ! 10 8 ). Consequently,[ Fe(GBT)] À remains the major speciesi ns olution in the pH range from 4t o6 .5, even when competing with relevant phytosiderophores ( Figure S3).

Conclusions
The iron sequestration behavior of gramibactin,a na rchetype for the new class of diazeniumdiolates iderophores,h as been investigated. Gramibactin is produced by Paraburkholderia graminis andc ontains two N-nitroso-N-hydroxylamine (diazeniumdiolate) chelatingg roups as wella so ne a-hydroxocarboxylate. Diazeniumdiolates are ap articularly interesting class of compounds due to their reported pharmacological potential, being only recentlyi dentified as siderophores, which together creates additional interest in the speciation and sequestration properties of gramibactin.
To ward this end, the acid-base properties of gramibactin have been investigated by potentiometric measurements and can be described by four protonation steps,where the relevant log K 1r values (cf. Table 1) correspond to one hydroxyl (10.94), two N-nitroso-N-hydroxylamine (5.71 and 4.87), and the carboxylate group (2.27). Gramibactin forms highly stable complexes with iron(III) ions over aw ide range of pH values. The stabilityc onstants of the formed ferric gramibactin species were determined by potentiometric and spectrophotometric methods using the ligand competition approachw ith EDTAa s competing ligand. [Fe(GBT)] À is the only species present in the pH range from 2t oa bout 9, while at higher pH values the formation of the iron-gramibactin dihydroxido species [Fe(GBT)(OH) 2 ] 3À is observed.
The sequestering ability of gramibactin towardi ron(III) ions was evaluated by means of metric parameters such as pFe and pL 0.5 .I nt erms of pFe, we could observe that at pH 7.4 gramibactin shows am uch highers equestering ability than EDTA (25.0 vs.2 3.5), anticipating that, in the presence of EDTAa s competitor ligand, the thermodynamic equilibrium is shifted toward gramibactin by af actor of 1.5 log units.T he same result is obtainedw hen the pL 0.5 parameters are compared at the same pH value,f or which gramibactin has ap L 0.5 value that is 1.6 log units highert han that of EDTA (15.5 vs. 13.9), corresponding to a4 0-fold lower concentration of gramibactin required for chelating 50 %o ft he iron(III) ions in solution,a s compared to EDTA. Furthermore, the pFe value of 25.0 is of the same order of magnitude as observedf or known hydroxamate-type siderophores. Considering the entire investigated pH range, based on the pL 0.5 parameter,f rom very acidic to highly basic conditions, ad ominants equestering ability is found for gramibactin when compared with the phytosiderophores mugineic acid andn icotianamine. In fact, the highest sequestering ability forg ramibactin occurs at moderate acidic conditions, suggesting some activity of gramibactin as as iderophore produced by the bacteria in the rhizosphere of graminaceousp lants to preventt he formation of iron(II), thereby increasingt heir tolerance toward acidic soils.
These findings directly point to the question of the complexing ability of gramibactin toward iron(II), which will be part of future experiments. This is expected to provide ab etter understanding of its role as siderophore and in reducing environments.I na ddition, future efforts will also include studies on the sequestration ability of gramibactin toward zinc(II) and copper(II) ions, aiminga tap ossible application of this outstanding ligand as antibiotic, antifungal or even antitumoral agent, either by inhibition of metalloenzymes or simple metal depletion.

Experimental Section Chemicals
KCl, Na 2 EDTA, and FeCl 3 ·6 H 2 Os olutions were prepared by weighing the corresponding salts, while HCl and KOH solutions were obtained by diluting concentrated ampoules. HCl and KOH were standardized against sodium carbonate and potassium hydrogen phthalate, respectively,p reviously dried in an oven at T = 383 Kf or at least 2h.F eCl 3 solutions were standardized against EDTAs tandard solutions. [58] Gramibactin was isolated as described elsewhere. [2] The purity of gramibactin was determined by HPLC and potentiometric titrations (! 99 %). All solutions were prepared using analytical grade water and grade Ag lassware. All chemicals were purchased from Sigma-Aldrich (and its brands) at the highest available purity.

Apparatus and procedure for potentiometric measurements
Potentiometric titrations were carried out at I = 0.1 mol dm À3 in KCl (aq) and at T = 298.15 AE 0.1 Ki nt hermostatted cells, using aM ettler To ledo DL50 apparatus, equipped with aS chott Instruments N6180 ISE-H + combined glass electrode. The estimated accuracy was AE 0.20 mV and AE 0.02 mL for potential and titrant volume readings, respectively.T he apparatus was connected to aP Ca nd automatic titrations were performed using the LabX light v1.05 software to control titrant delivery,d ata acquisition and to check for potential stability.A ll potentiometric titrations, including electrode calibrations in terms of free proton concentration (i.e.,p H= Àlog [H + ], not activity), were carried out as reported elsewhere. [35] Here, the titrant solutions were prepared by addition of different amounts of gramibactin (7 10 À4 c GBT /mol dm À3 1 10 À3 ), EDTA and Fe 3 + with different GBT:Fe:EDTAr atios, together with the supporting electrolyte (KCl) to obtain the pre-established ionic strength value (I = 0.1 mol dm À3 ). In all samples, known slight excess of strong acid (HCl) was added in the titrant solution, in order to lower the starting pH of measurements. All the measurements were performed by titrating 10 to 20 mL of the titrant solution with standard KOH (aq) up to pH % 11.8 0-100 points were collected for each titration.
Apparatus and procedure for spectrophotometric measurements AS himadzu UV-1800 UV/Vis spectrophotometer was used to perform the spectrophotometric titrations, carried out at I = 0.1 mol dm À3 in KCl (aq) and T = 298.15 AE 0.1 Ki naglass cuvette (1 cm of path length) placed in the spectrometer equipped with a thermostatted cell holder.A ni noLab pH 7110 equipped with aS ci-enceLine Type N6000A combined ISE-H + glass electrode (SI analytics) was used for pH readings, after its calibration (in terms of free concentration, pH Àlog[H + ]) before each experiment. [22] Measurements were performed by titrating 2.2 cm 3 of the titrant solution with standard KOH (aq) solutions up to pH~11.T itrant solutions consisted of different amounts of GBT (3 10 À5 c GBT /mol dm À3 6 10 À5 )a nd Fe 3 + (1.5 10 À5 c Fe /mol dm À3 6 10 À5 ), HCl excess (c H = 5-8 mmol dm À3 ), together with the supporting electrolyte (KCl) in order to adjust the desired ionic strength (I = 0.1 mol dm À3 ). The homogeneity of the solutions during the titration was maintained by magnetic stirring.

Calculations
The BSTAC4 [59] computer program was used for determination of all the parameters of the acid-base potentiometric titrations as well as for determination of the complex formation constants. Spectrophotometric data were evaluated by HypSpec. [27] For the generation of speciation and sequestration diagrams, the Hyss [60] and ES4ECI [59] programs were used. All complex formation constants are expressed considering the overall equilibrium according to Equation (2), p Fe 3þ þ q L 4À þ r H þ Ð Fe p L q H r ð3pÀ4qþrÞ ð2Þ which is also valid for the ligand protonation (with p = 0) or metal hydrolysis constants (when q = 0a nd r < 0). Protonation constants are also expressed as stepwise equilibria according to Equation (3).
The molar concentration scale (c,m ol dm À3 )i su sed to express formation constants, concentrations and ionic strength, while errors are expressed as AE standard deviation. For simplicity,a nd when not relevant, the charges of the various species are omitted. Ligand acronyms (GBT and EDTA) in species and equilibria refer to the fully deprotonated species (GBT) 4À and (EDTA) 4À .