β- O -4 type dilignol compounds and their iron complexes for modelling of iron binding to humic acids: synthesis, characterization, electrochemical studies and algal growth experiments

A series of β- O -4 type dilignol compounds and their iron(III) complexes, designed as model compounds for humic acids, were prepared and characterized by 1 H-NMR and 13 C-NMR spectroscopy, elemental analysis, EPR spectroscopy, IR spectroscopy, UV‒Vis spectroscopy and electrospray ionization mass spectrometry (ESI-MS). Properties regarding iron binding, stability, liphopilicity and bioavailability for microorganisms have been evaluated with cyclic voltammetry, stability studies in water and seawater by means of UV‒Vis spectrophotometry and the algae growth assays with seawater algal species Chlorella salina and Prymnesium parvum. Both established ligands and their iron complexes undergo deprotonation processes in seawater wheares no changes in UV‒Vis spectra were observed in distilled water. The iron(III) complex formation constants, p K a values and liphopilicity of the dilignols were in the same range as for the analogous catechol coordination compound. Synthetized dilignols were prone to redox reactions under biological conditions similar to natural aquatic humic acids. Moreover, an increased iron bioavaiability was observed for the presented complexes compared to corresponding catechol complexes and comparable to the bioavaialability of iron bound in humic acid complexes recovered from Craggie Burn river. Those results confirm that β- O -4 type dilignol compounds are excellent model ligands for aquatic humic acids. molecular peaks of


Introduction
Iron chemistry in seawater and its bioavailability for microorganisms has been intensively investigated for a long time. [1][2][3][4] Several experiments have shown that iron is limiting the productivity of marine ecosystems in huge areas of the ocean [5][6][7][8][9] but only little is known about the chemical speciation of iron in seawater and its accessibility for phytoplankton. Iron is not only an important factor for growth of algae and other microorganisms but also selectively supporting the development and expansion of several species (with lower iron demand), which impacts the food chain within the whole ecosystem. 4,10 Analytical methods provide information about the concentration of dissolved and undissolved iron or iron solubility dependent on the presence of organic ligands. However; exact structures of bioavailable iron complexes and uptake mechanisms remain still unclear. There are several factors influencing iron bioavailability for microorganisms. Besides different uptake mechanisms; kinetics and iron requirements for each species, one pressing issue is the chemical form of bioavailable iron. It is known, that inorganic iron displays an extremely low solubility in seawater, which is the result of the low solubility of iron oxide (the general term 'iron oxide' describes various forms of iron oxide, oxyhydroxide and amorphous iron hydroxide). 11,12 Thus, almost all of dissolved iron in the ocean is bound to organic scaffolds. [13][14][15][16] Those ligands have a large range of different structures and origins. One group of them are siderophores, which are produced by bacteria particularly in order to complex iron(III). [17][18][19] Stability constants for iron complexes of these natural chelators are extremely high and special mechanisms have been developed by certain organisms to release the iron from formed complexes. 20 Another big group of important iron chelators are substances which originate from decomposition of organic material, known as humic substances.
Their main sources are rivers, especially derived from peat bog regions which are rich in aquatic humic substances (AHS). It has been shown, that AHS are greatly improving the bioavailability of iron in coastal waters. [21][22][23][24][25] Also algal studies on microalgae Chlorella salina and Diacronema lutheri in batch cultures support the assumption of the positive impact of AHS on iron supply. 26 Algal cultures treated with isolated AHS showed increased growth response compared to samples treated with iron(III) chloride and EDTA (ethylenediaminetetraacetate), a complexation agent used for commercial algal cultivation 27 ). Knowing the importance of those chelators on phytoplankton growth and vitality, which affect huge amount of other processes like for example the global CO 2 cycle, [28][29][30] it is of great interest to clarify the chemistry behind iron complexation, release and uptake as well as the structure of the complexes and ligands. Unfortunately, research on humic substances, especially AHS is very challenging due to their low concentration in seawater 31 and the structural complexity depending on origin, incorporated building blocks and formation processes. [32][33][34][35] In order to understand the chemistry of AHS regarding iron binding properties and the postulated release by photoreduction 36 the utilization of model systems represents a highly promising approach. In our previous studies, we synthesized simple iron complexes with different coordination motifs and ligand scaffolds and investigated them for their suitability as model compounds for humic acids iron complexes. 37 The model compounds were characterized and investigated by various analytical techniques (cyclic voltammetry, EPR, EXAFS, 36 UV-Vis spectroscopy etc.) in addition to algal batch culture studies on chlorophyte and haptophyte unicellular algae species. The results of the latter assays led to the conclusion, that catechol-based ligand systems are excellent scaffolds for modeling of iron-AHS complexes. Therefore catechol-derived ligands and complexes with structural characteristics and molecular weight in range of natural AHS were established. The main components of humic substances are lignin decomposition products, originating from support tissues in plants and guaiacylglycerol-β-guaiacyl ether represents a suitable model for lignin. 38,39 Binding studies of iron with guaiacylglycerol-β-guaiacyl ether precursors (coniferyl alcohol, sinapic acid, ferulic and coumaric acid) and dehydrogenation polymers of coniferyl alcohol have been already carried by Guillon et al. regarding their sorption properties in comparison to lignin. 40,41 Although guaiacylglycerol-β-guaiacyl ether seems to be a good model for AHS, formation of stable iron complexes is unlikely. For this reason, we decided to modify the β-O-4 backbone of guaiacylglycerolβ-guaiacyl ether by introduction of a free catecholic moiety.
In this work we report the synthesis of diastereomerically pure lignols designed as model compounds for AHS, bearing free catecholic groups and their respective iron complexes. The synthesized substances were characterized by elemental analysis, spectroscopic methods (IR, EPR, UV-Vis), NMR spectroscopy (in the case of the ligands), mass spectrometry (ESI-MS) and cyclic voltammetry. The The proton dissociation constants (pK a ), complex formation constants and lipophilicity (as distribution coefficients) were determined. The bioavailability of the respective iron coordination compounds was elucidated in two unicellular algal species, namely chlorophyta Chlorella salina and haptophyte Prymnesium parvum in batch cultures.

Results and discussion
Synthesis -general overview Scheme 1. Synthesis of β-O-4 type lignol compounds.
Diastereomerically pure guaiacylglycerol-β-guaiacyl ether 6a was synthesized according to previously published procedure. 42 Two novel β-O-4 type lignol compounds 6b,c were synthesized in a similar approach (Scheme 1, see SI). Compound 3 was prepared according to the procedure described by Nakatsubo et al. 43 using tert-buthyl chloroacetate. Aldehydes bearing free phenolic groups were protected with benzyl groups to avoid undesired side reactions (2a-c). Aldol condensation of 3 and 2a-c in the presence of lithium diisopropyl amide at -78°C led to diastereoisomeric mixture of 4a-c (1:1 ratio of erythro and threo, respectively). As described in literature, recrystallization of 4a from EtOAc gave rise to pure erythro diastereoisomer. For the compounds 4b and 4c recrystallization was not successful and column chromatography (n-hexane/EtOAc 5:1 to 1:1) was performed. Due to the very similar elution time, it was difficult to separate both diastereoisomers and only the erythro disatereoisomer was purely separated. The isolated compounds were reduced with lithium aluminum hydride and benzyl groups were cleaved by palladium-catalyzed hydrogenation yielding 6a-c. Formation of the desired dilignols was confirmed by NMR spectroscopy and elemental analysis.
The ESI-MS spectra of the ligands were measured in methanol or in acetonitrile. Overall the positively charged Na + adducts, with m/z values of 329 (6a and 6b), were detected. Lignols were isolated in low to moderate yields due to the crucial purification step including the separation of diastereoisomers.
Thus in this step the yield of the pure erythro diastareoisomers was between 7% and 20%. The Fe(III) complexes were synthesized by deprotonation of the ligand with potassium hydroxide and addition of iron(III) chloride in methanolic solution (see SI). The complexes 7b and 7c were isolated in low to moderate yields (7-27%) and characterized by standard analytical methods. We presume that similar to analogous Fe(III) catechol coordination compounds, dinuclear complexes with two lignols bridging two iron centers were obtained. The mass spectra of the complexes display, similar to isolated and characterized catechol complex, one negatively charged fragment with m/z value of 664, which can be assigned to Fe(lignol) 2 -. All detected molecular peaks of the complexes showed the expected isotopic iron pattern. Also elemental analysis, electrochemical studies and complex formation studies support our proposed structure.

Cyclic voltammetry
The electrochemical behavior of the ligands and complexes was studied in order to estimate the possibility of iron reduction under physiological conditions. The proposed mechanism of iron release in natural humic acid complexes includes the photoreduction of Fe(III) to Fe(II). 36 Fe(II) has low affinity to AHS, dissociates from the complex and can be uptaken by the microorganisms. Although Fe(II) undergoes oxidation in oxygenated seawater, the steady state concentration of Fe(II) remains higher due to this reduction process. 44 Thus redox potentials present important information regarding the bioavailability of iron and cyclic voltammetric measurements of 6b and 6c were performed in aqueous solution at various pH values in the absence and in the presence of iron(III) ions. Representative cyclic voltammograms are shown in Figure 1 for 6b and catechol as reference, and electrochemical data are presented in Table S1. A reversible (E 1/2 = 0.00 V) and a quasi-reversible process (E 1/2 = +0.82 V) can be observed at pH 3.7 in the case of 6b. The reversible process might belong to the one-electron transfer between the catechol and the semiquinone radical. 45 On the other hand mostly irreversible oxidation peaks appear at pH > 8.8, which is the pH range where the ligand 6b is present in its monodeprotonated (HL -) form ( Figure S4A). The current maximum of the cathodic peak at +0.65 V is decreasing with increasing pH and a novel oxidation peak appears at +0.50 V and becomes dominant at pH 9.89, but disappears at pH >12. In addition a new reduction peak is observed at pH ≥11 at +0.27 V. In the presence of half-equivalent iron(III) the position of the quasi-reversible peak pair is already shifted at pH 1.8 and E is somewhat lower (+0.33 vs. +0.24 V) ( Fig. 3b), although practically no complex formation takes place under these conditions ( Figure S4B), because peaks belong to the unbound iron are observed at a lower potential range (E 1/2 = -0.15 at pH 2.7, not shown here). The cathodic peak at +0.42 V is disappearing with increasing pH, while the oxidation peak still can be observed shifting from +0.76 to +0.42 V. The anodic peak is most probably related to the electrochemical oxidation of the ligand, which is mostly bound to the metal ion (e.g. 95% of the ligand is coordinated at pH 8.) In addition a novel cathodic peak is seen at ca. -0.1 V at pH > 9 where more and more tris-ligand species are assumed to be formed and the reduction process can be described to the iron(III) / iron(II) transition. Notably, only irreversible processes are observed at pH > 5. Overall, 6c behaves very similarly to 6b; however, formal potentials E 1/2 of the ligand are always lower by 170-180 mV (Table S1). At the same time in the presence of iron(III) ions the observed anodic and cathodic peak positions are just slightly different in the case of the studied ligands.
Comparing the recorded voltammograms of 6b, 6c to those of catechol at pH ~ 8 when iron(III) was added to the solution it can be noted that the oxidation peak of catechol (+0.64 V) appears at a similar potential as in the case of the other two ligands, while a cathodic peak with high intensity is seen at +0.42 V and another smaller reduction peak appears at lower potentials (+0.11 V). In the case of catechol, iron(III) is present in the bis complex predominantly at 1:2 metal-to-ligand ratio at pH 8 due to the formation of the higher stability complexes compared to ligands 6b and 6c (c.f. Figure S4B in the case of 6b) based on the determined stability constants (vide infra). The isolated Fe(III) complexes 7b,c of the respective dilignols ligands 6b,c were investigated for comparison. However, due to the lower aqueous solubility the experiments were performed in DMF ( Figure S1). Under these conditions the two irreversible reduction peaks with values -0.07 V and +0.52 V and an irreversible oxidation peak at +1.2 V were detected for 7b which can be assigned to ligand and iron reduction. Complex 7c showed three irreversible reduction peaks with the values +0.01 V, +0.29 V and +0.44 V and similar to 7b an oxidation peak at +1.2 V.
Overall, the observed reduction potentials of model ligands, complexes and in situ prepared iron(III) complexes are lying in the area accessible to photoreduction as well as for natural occurring reducing agents. 46 Those studies confirm that those dilignols can be seen as suitable model compounds for AHS regarding their redox activity.

EPR spectroscopy
EPR spectral analyses of the ligands 6a-c, revealed a g-factor of 2.0 similar to the free electron ( Figure 2). Thus, all ligands might represent organic free radicals. Santana-Casiano et al. showed that catechol-type structures can be readily oxidized in the presence of oxygen yielding stable radicals. 47 Stable radicals were also detected in humic acids. 48,49 The high stability of the free radicals can be explained by the delocalization of the electron over the phenol ring. The hyperfine splitting of the signal indicates an interaction between the unpaired electron and magnetic hydrogen nuclei. The line pattern of the EPR spectrum suggests that the electron is delocalized over the carbon atoms of the aromatic ring. The EPR spectrum of the ligand, 6a, differs from the other ligands exhibiting a g-factor of approximately 1.9. The methoxy group at the aromatic ring might participate in the delocalization of the electron, thereby inducing the appearance of additional signals. The two iron complexes, 7b and 7c, showed a distinct signal at g = 4.3 ( Figure 2). This isotropic-looking line is characteristic of

Proton dissociation processes and lipophilicity of the ligands
The proton dissociation constants of the studied ligands 6a-c were determined by pH-potentiometric and UV-Vis spectrophotometric titrations in aqueous solution ( Table 1). The first deprotonation process in all cases takes place at pH < 11.5, therefore pK 1 values could be determined accurately by the evaluation of the pH-potentiometric titration curves; on the other hand, these titrations were used to obtain the exact concentration of the ligands stock solutions and to check their stability in aqueous solution. Consecutive titrations showed that no ligand decomposition occurred in the pH range studied under an argon atmosphere and the stock solutions (pH ~1.8) were stable over 48 h.
The proton dissociation processes could be followed by UV-Vis spectrophotometric titrations as well, since the deprotonation of the hydroxyl functional groups is accompanied by characteristic spectral changes. However, the second step occurs only at pH > 11.5 and spectra had to be recorded at high The lipo-hydrophilic character (logD) of the ligands 6a-c was studied at pH 2.5 and 8.3 via the partitioning between n-octanol and water (Table 1, Figure S7A-B All results indicate that the methoxy derivative has a more lipophilic character compared to 6b,c, which have moderate hydrophilic character like the reference compound catechol. At the chosen slightly alkaline pH the ligands are partly deprotonated, which causes the increased hydrophilicity. We have made an attempt at determining the lipophilicity of the iron(III) complexes of the studied compounds, but a precipitate was formed and clogged between the two phases even at lower ligand concentration (100 M), which hindered the quantitative analysis. Comparing the spectra of the

Complex formation processes of ligands
The complex formation processes of the ligand 6b with iron(III) were studied primarily by pHpotentiometry in aqueous solution. However, the complexation was found to be sluggish, especially in the acidic pH-range meaning that the equilibrium was not reached within 10 min as also reported for catecholates. 50 Therefore UV-Vis spectrophotometry was applied to follow the complexation of 6a-c with iron(III) using longer waiting time in the pH range from 2 to 11 with the exclusion of air; and the process was found to be reversible under this condition. Spectra were recorded in the wavelength range 350 -1000 nm where mostly the strong metal-to-ligand charge transfer (CT) bands can be seen and the non-coordinated ligands do not absorb. A representative spectrum series for the iron(III) -6c (1:3) system is shown in Figure 3A, which shows characteristic changes upon increasing pH. The  max values of the main CT band were decreased with increasing pH ( Figure 3B) and a well-  Table 2, and Figure 3C in the case of 6c). In these complexes most probably the completely deprotonated ligands (L 2-) coordinate via an (O -,O -) donor set as it was reported for the reference compound, catechol, and for other catecholate derivatives. 50,53,54 The coordination of the monoprotonated ligand (HL -) in the metal complexes is very rare and observed only in compounds isolated from organic solvents; 55, 56 however, it was also suggested for the iron(III) mono-ligand complexes of catechol by R.C. Hider et al. 54 It is noteworthy that intramolecular redox processes between Fe(III)-catecholate and Fe(II)-semiquinone species are known to occur, mostly in the case of the mono complex. 45,54,57 In addition existence of dimeric species [Fe 2 L 4 ] 2of catechol in the pH range of the formation of the bis-complex was also suggested based on Mössbauer spectroscopy.
This species was suggested to contain two equivalent sites and a bridging hydroxyl group. 54 Spectrophotometry is not an adequate method to distinguish between the formation of [Fe 2 L 4 ] 2and [FeL 2 ]complexes with the same metal-to-ligand ratio. Therefore two kinds of models could be calculated for the studied systems using the titration data which consist of the formation of the mononuclear or the dinuclear complex, but the latter model gave somewhat poorer fits between the experimental and calculated absorbance values. The calculated molar absorbance spectra and  max values (  Since ligand 6a has lower water solubility compared to the other two studied compounds its complexation with iron(III) was studied via the ligand's bands at lower concentrations (~ 200 M).
The measured absorbance values in the absence and the presence of the metal ion (Fig. S3) were fairly similar suggesting the negligible formation of the iron(III) complexes under these conditions.
The coordination of the monodentate phenolate type ligand is very weak similarly to the case of the phenol itself. 58 Direct comparison of stability constants (Table 2) and concentration distribution curves calculated for the iron(III) complexes formed with 6b, 6c ( Figure S5A) and catechol ( Figure S5B) shows that the studied ligands form somewhat lower stability complexes than catechol. (Notably, direct comparison of the log values in this case is possible due to the similar stochiometry of the complexes formed and the similar pK a values of the ligands.) In these cases the predominant formation of the bis complexes is found at neutral and slightly basic pH values. However, the actual speciation at pH 8 depends on the actual metal-to-ligand ratio ( Figure 3D).
In order to compare the metal ion binding abilities of the studied ligands to each other and to catechol or to other iron chelators (such as the catecholate-containing siderophores) pM compounds (see S10-S12). The time dependent spectra of 7b and 6a-c in water at pH ~8 (see S13-S14), showed also an increase in the same signals as observed in seawater. The same behavior was observed for as time period of 21 days (duration of algal growth tests) and only ligand 6a showed an additional band after 14 days (see S15-S16). Those results confirm that the established compounds are sufficiently stable for algal growth experiments. In order to elucidate the bioavailability of the iron complexes 7b,c as well as the free ligands 6a-c to provide bioavailable iron, algal tests on C.
salina and P. parvum have been performed. Those two species were intentionally chosen due to their wide occurrence and different biology. C. salina is a representative for unicellular green algae. The species react very sensitive to iron limitation. It has been recently reported, that cultures cultivated without iron or chelator EDTA showed poor growth. 26,37,60 The second species P. parvum, is a unicellular haptophyte and this algae is able to grow under photoautotrophic and heterotropic conditions. The species is also less sensitive to iron deficiency and can adapt to several environmental limitations. Moreover, P. parvum is well known for its toxicity and is a cause of mass fish deaths in coastal waters. Investigating both species may help to understand the alterations in iron uptake and preferences of each algae species towards different iron sources. All algae experiments were carried out in batch cultures, using three different nutrition setups. As full medium, seawater with micronutrient and vitamin solution as well as solution of NaNO 3 , Na 2 SiO 3 , NaH 2 PO 4 as described by Guillard et al. 27 , were used (see Table S5)    Figure S17). Table S2 shows the end concentrations of the samples as well as the relation between the end-algae concentration of control samples to samples treated with model compounds from both experiments with C. salina.
Additionally to the maximal algae cell concentration, we also calculated the specific growth rate of algae for all tested compounds after 17 days of each experiment (which was the duration of the algal growth experiment for 7c (Table S3). Similar to the maximal algae cell concentrations, highest specific growth rates were observed for samples treated with 6b and 7b.
The model ligands were also tested on P. parvum for they ability to bind and supply iron ( Figure S18).
The culture of P. parvum was not as sensitive as C. salina to iron deficiency and we did not observe such pronounced effects on algal growth as for C. salina. In fact, the control with full medium  Table S4. Similar to the maximal algae cell concentrations, highest specific growth rates were observed for samples treated 6c.
Summarizing the algal studies results, some of our herein presented compounds can be seen as suitable models for humic acids regarding the iron transport and ability to supply microorganisms with iron. In all experiments, the algal growth in samples without iron was strongly inhibited. For C. salina typical growth pattern, known from iron enrichment experiments, 61,62 was observed. 63,64

Conclusions
In this work, the novel dilignols-based models for aquatic humic acids were synthesized and extensively characterized. In order to investigate the redox properties, cyclicvoltammetric measurements were carried out. All of the compounds can be possibly reduced under biological conditions, which is in accordance with the properties of AHS. Various UV-Vis studies were performed in order to elucidate iron(III) complex formation constants, pK a values of the ligands, lipopilicity and the stability of the compounds in aquatic systems, especially in artificial seawater to verify the potential of the developed model to supply marine microorganisms with iron. Regarding the stability of the compounds we observed changes in spectra of the recorded in seawater which were the result of the deprotonation of free catecholic hydroxy groups. Comparison of the pK 1 values of 6b,c to 6a revealed a significantly weaker acidity of 6a due to the electron donating effect of the methyl group, while 6b and 6c have pK a values, which are similar to catechol. The lipophilicity studies of the ligands showed that methoxy derivative 6a has more lipophilic character compared to the catecholate derivatives, which have moderate hydrophilic character due to the partial deprotonation at increased pH. Stability constants for the formation of the iron(III) complexes with model ligands were somewhat lower as for catechol whereas no data could be collected for 6a due to its very low affinity towards iron. At neutral and slightly basic pH values bis-complexes were predominant. The iron(III) binding ability of the studied ligands was found to be weaker as for the known siderophores. The ability to supply algae with iron was investigated in algal batch cultures. Ligands 6b,c moiety showed very good impact on the culture of C. salina and P. parvum comparable with natural AHS and significantly better as for catechol itself. Also experiments on C. salina with isolated iron complexes gave similar results. However, guaiacylglycerol-β-guaiacyl ether did not enhance the growth of C. salina. Those results suggest that catecholic moieties can play an important role by complexation and stabilization of inorganic iron by humic substances.