Coordinate and redox interactions of epinephrine with ferric and ferrous iron at physiological pH

Coordinate and redox interactions of epinephrine (Epi) with iron at physiological pH are essential for understanding two very different phenomena – the detrimental effects of chronic stress on the cardiovascular system and the cross-linking of catecholamine-rich biopolymers and frameworks. Here we show that Epi and Fe3+ form stable high-spin complexes in the 1:1 or 3:1 stoichiometry, depending on the Epi/Fe3+ concentration ratio (low or high). Oxygen atoms on the catechol ring represent the sites of coordinate bond formation within physiologically relevant bidentate 1:1 complex. Redox properties of Epi are slightly impacted by Fe3+. On the other hand, Epi and Fe2+ form a complex that acts as a strong reducing agent, which leads to the production of hydrogen peroxide via O2 reduction, and to a facilitated formation of the Epi–Fe3+ complexes. Epi is not oxidized in this process, i.e. Fe2+ is not an electron shuttle, but the electron donor. Epi-catalyzed oxidation of Fe2+ represents a plausible chemical basis of stress-related damage to heart cells. In addition, our results support the previous findings on the interactions of catecholamine moieties in polymers with iron and provide a novel strategy for improving the efficiency of cross-linking.


Results
Structure of Epi-Fe 3+ complexes. No autooxidation of Epi was observed at pH 7.4. A characteristic spectrum of Epi (λ max = 280 nm) remained unaltered for at least 1 h (Fig. 1a). New bands emerged at longer wavelengths upon incubation with Fe 3+ (Fig. 1b). These were attributed to the coloured Epi-Fe 3+ complexes. When Fe 3+ forms coordinate bonds, electrons in d-orbital split into high and low energy orbitals. For many ligands, including catechols, the energy difference corresponds to the wavelengths in the visible range 20 (Fig. 1c), implying that the same complex is formed regardless of Epi concentration. It is important to stress out that HPLC results showed that Fe 3+ did not provoke detectable degradation of Epi (Fig. 1d). For the [Epi]/[Fe 3+ ] = 4 system, the absorbance at 505 nm showed a gradual increase over a period of 15 min (Fig. 1e). For lower [Epi]/[Fe 3+ ] ratios, the 505 nm band was replaced/shifted within 5 min to either 520 nm or 545 nm band.
Using low-T EPR, it is possible to determine the total spin quantum number of Fe 3+ in Epi-Fe complexes 31 . The 100 K EPR spectrum of 0.1 mM Fe 3+ in 10 mM Tris buffer showed only a weak signal of low-spin Fe 3+ (S = 1/2) at g ~ 2 (Fig. 2a) 32 . In the presence of Epi, a strong g = 4.26 signal that arises from high-spin Fe 3+ (S = 5/2) in orthorhombic symmetry was observed. Next, [ (Fig. 2b). This implies that the maximal stoichiometry is 3. Fe 3+ remained in the high-spin state at all concentration ratios. At 100 K, the line-width of g = 4.26 signal was ∼7.4 mT for all investigated ratios. To gain more information about the symmetry of complexes, the spectra were acquired at 20 K (Fig. 2c) ] = 1 are different, and that the former complex shows higher anisotropy. Our results are in accordance with a previous low-T EPR study of interactions of catecholamine-rich peptides with ferric iron 26 .
Raman spectroscopy was conducted in phosphate instead of Tris buffer, because amides show Raman bands that are in the range of interest here. As a reference, the formation of complexes in the phosphate buffer was investigated also using UV/Vis spectroscopy ( Supplementary Fig. S1). Raman spectra of the Epi-Fe 3+ complex showed bands at ~535, 637, 1270, 1342 and 1489 cm −1 (Fig. 3). Band positions corresponded to the previously reported Raman spectra for Fe 3+ -catecholamine-based biopolymers and metal-organic frameworks 28,33 . These bands were (almost) negligible in the absence of Fe 3+ . It has been shown previously that the interactions of catecholamine moieties in biopolymers with Fe 3+ drastically increase the amplitude of Raman bands 25,33 . The appearance of bands in the presence of Fe 3+ is most likely related to the fact that Raman laser wavelength (532 nm) was close to the electronic transition in Epi-Fe 3+ complexes (λ max = 545 nm), and far from Epi absorption (λ max = 280 nm). The signal at 1489 cm −1 showed the most prominent rise in the presence of Fe 3+ . The band has been assigned to the catechol ring vibration 28,33 . Other bands were assigned as follows: 1342 cm −1 , C-H bending; 1270 cm −1 , C-O stretching; 637 cm −1 , Fe-O stretching; ~535 cm −1 , bending/stretching of the complex 25,33 . It appears that the signal at ∼535 cm −1 was composed of two bands. This is the result of the binding of Fe 3+ to two slightly different O atoms within a bidentate complex with catechol ring 27,34 . The 637 cm −1 band most likely reflects the bending of catechol ring away from co-planarity with the O-Fe 3+ -O that has been observed previously via crystallography 35 . The appearance of bands centred at 1270 and 1342 cm −1 in the presence of Fe 3+ further corroborates Fe-O binding 28 .
Redox properties of Epi-Fe 3+ complexes. Redox activity of Epi/Fe 3+ systems in cyclic voltammograms (CV) was ligand-centred (Fig. 4a). CV of Fe 3+ did not show distinctive peaks. This is most probably related to the predominance of the amorphous Fe 3+ complex with OH − ions at physiological pH ( Supplementary Fig. S2). The peaks correspond to Epi oxidation (E pa ∼ 400 mV), and to the following reduction of oxidation product(s) (E pc ∼ −570 mV) 36 . Oxidation and reduction peak current ratios (I pa /I pc ) were substantially higher than 1 (Fig. 4b), which means that the electron transfer was irreversible. This can be attributed to instability and polymerization of products of Epi oxidation 37 Fig. S3). A direct linear relationship between I pa , I pc , and the square root of scan rate implies that the currents mainly depend on two parameters: the rate at which redox species diffuse to electrode surface (D), and the rate constant of electron transfer (k s ). Other interactions, such as adsorption, were negligible 38 . D and k s were calculated using Randles-Sevick equation and Nicholson Shain method ( Supplementary Fig. S3) 39 Supplementary Fig. S3). This may be related to the delocalization of aromatic π electrons by Fe 3+ . As a reference, Epi and Fe 3+ were also investigated in phosphate buffer ( Supplementary Fig. S4). The same complex predominated at both high and low [Epi]/[Fe 3+ ], which is in agreement with UV/Vis results in this buffer. Epi in the complex showed lower E pa than free Epi.
Interactions of Epi with Fe 2+ . The oxidation of Fe 2+ to Fe 3+ at pH 7.4 was drastically promoted by Epi (Fig. 5a,b). A broad band at λ max = 570 nm emerged within 1 min for different initial Fe 2+ concentrations ([Fe 2+ ] i ). The 570 nm band has been observed previously in similar Epi-Fe 2+ systems, and has been attributed to the Epi-Fe 3+ complexes that are formed following Fe 2+ oxidation 41,42 . However, an evident shift of the absorption maximum compared to Epi/Fe 3+ systems ( Fig. 1), implies that the 570 nm band may arise from some other species. Namely, the reduction of O 2 by Fe 2+ gives different by-products (Supplementary Table S1), including hydroxyl radical (HO • ), a very strong oxidant. These products caused Epi degradation, as shown by HPLC ( Supplementary Fig. S5). The rate constant for the reaction Epi + HO • is an order of magnitude higher than Tris + HO • : 2.2 × 10 10 M −1 s −1 vs. 1.1 × 10 9 M −1 s −1 43,44 . Therefore, 10× higher concentration of Tris (100 mM) was applied to employ Tris as an 'antioxidative buffer' . EPR spin-trapping measurements showed that 100 mM Tris has a significantly higher capacity to remove HO • than 10 mM buffer. As expected, Fe 2+ -related Epi degradation was suppressed in 100 mM Tris, being completely prevented in systems with [Fe 2+ ] i ≤ 0.2 mM (Supplementary   ratio was simulated to be the sum of these two. These results confirm that the shift to 570 nm is related to Epi degradation. More importantly, this corroborates that Epi is not a direct reactant in Fe 2+ oxidation, but acts in a catalyst-like fashion. In line with this, one Epi facilitated the oxidation of two Fe 2+ in 100 mM Tris ( Supplementary Fig. S5).
Fast oxidation of Fe 2+ in the presence of Epi was further supported by low-T EPR and cyclic voltammetry. Fe 2+ was 'EPR silent' in the experimental conditions applied here (perpendicular mode EPR). Within 1 min after the addition of Epi, a strong high-spin Fe 3+ signal was observed (Fig. 5d). In addition, lines that are characteristic for slowly tumbling organic radical appeared in the higher field 45 , confirming that partial degradation of Epi took place in 10 mM Tris. Such signal could not be observed in the systems with Fe 3+ (not shown). Further, I pa and I pc in CV of Fe 2+ showed a slow time-dependent decay in the absence of Epi (Fig. 5e). This reflects Fe 2+ oxidation to Fe 3+ which is CV-inactive, as discussed earlier. In contrast, a rapid change took place in the presence of Epi (Fig. 5f). Fe 2+ -related peaks were diminished and CV acquired shape with E p and I p values as in the CV of analogous system with Fe 3+ ([Epi]/[Fe 3+ ] = 1; Fig. 4a).
Next, we examined O 2 consumption by Fe 2+ oxidation in the presence or absence of Epi (Fig. 6a). Epi substantially increased the initial rate of O 2 consumption. In similar experiments with Fe 3+ no changes in [O 2 ] were observed (not shown). The total decrease in [O 2 ] showed a linear dependence of [Fe 2+ ] i with the slope k ∼ 0.25 (Fig. 6b). This means that four Fe 2+ in total were being oxidized to remove one O 2 , which is in accordance with previous results on Fe 2+ oxidation at pH 7-8 46,47 . Fe 2+ is not 'spent' only on the reduction of O 2 , but also on different reactions that neither remove or produce O 2 , such as the reduction of O 2 •− (generates H 2 O 2 ), Fenton reaction, and the reaction with HO • (Supplementary Table S1). Initial rates of O 2 consumption and k value were used to calculate the initial rates of Fe 2+ oxidation. They were as follows: ∼0.  (Fig. 6b). This is in line with previous studies of Fe 2+  . It is noteworthy that we could not detect O 2 •− or HO • in these systems using EPR spin-trapping, probably because Epi-Fe 2+ complex reduced the paramagnetic spin-adducts 48 . The Epi-Fe 2+ complex was further examined by measuring the redox potential (E h ) under aerobic and anaerobic conditions. As a reducing agent, Fe 2+ caused a considerable and relatively stable drop of E h (Fig. 6c). In the presence of Epi, the change in E h was less pronounced and partially reversible. It can be observed that E h for [Epi]/[Fe 2+ ] i = 4 was stabilized at higher values compared to E h for analogous system with Fe 3+ (Fig. 6c). This is probably related to the accumulation of H 2 O 2 (Fig. 6a) (Fig. 6c). This is in line with the absence of H 2 O 2 accumulation (Fig. 6a) ] = 4, which is in accord with the cyclic voltammetry. Fast oxidation hindered the determination of inherent redox properties of the Epi/Fe 2+ system. Therefore, additional measurements were conducted under anaerobic conditions (Fig. 6d). The addition of Fe 2+ provoked an irreversible decrease of E h that was significantly more pronounced in the presence of Epi. Final E h was more than 120 mV lower in the Epi/Fe 2+ systems compared to E h of corresponding Fe 2+ solutions without Epi. This implies that Epi and Fe 2+ form a strong reducing agent.

Discussion
Epi and Fe 3+ build high-spin complexes at pH 7.4, with 1:1 (λ max at 545 nm) or 3:1 (λ max = 505 nm) stoichiometry depending on the [Epi]/[Fe 3+ ] concentration ratio.  with O atoms on the catechol ring ( Supplementary Fig. S6). Electrochemical data showed that Fe 3+ does not drastically affect redox properties of Epi in Tris buffer, whereas Epi in 1:1 complex in the phosphate buffer was more susceptible to oxidation than free Epi. Nevertheless, Epi was stable in the presence of Fe 3+ . This was confirmed by HPLC and UV/Vis. The formation of 3:1 complex preceded the formation of 1:1 complex at low [Epi]/ [Fe 3+ ]. Hence, the 505 nm band is not related to the formation of quinones or some other products. In addition, the production of adrenochrome (λ max = 480 nm), a common derivative of Epi oxidation, was not observed. This appears to be in discord with some previous reports. However, those have been performed in atypical or complex settings, which might be prone to copper impurities, such as highly acidic media 15,16 , biochemical assays [17][18][19] , or long incubation in multi-component buffers 10 .
Epi and Fe 2+ form a complex, most likely in 1:1 stoichiometry 41 , which represents a strong reducing agent. The oxidation of Fe 2+ was facilitated at least 10× by Epi. A modelling study estimated that Fe 2+ transfers 1.3 electrons to the electron-rich catechol ring 49 , which might result in destabilization of the complex. The promotion of Fe 2+ oxidation by Epi might be further explained by the fact that ligands with harder donor sites are better Fe 3+ stabilizers and decrease the redox potential of Fe 3+ /Fe 2+ pair 23 . The stability constants for catechol complexes with Fe 3+ are significantly higher than complexes with Fe 2+ 22 . According to Pearson's Hard and Soft Acids and Bases principle, Fe 3+ is hard, whereas Fe 2+ is borderline Lewis acid. Hydroxyl groups represent hard bases 50 . It has been calculated that Highest Occupied Molecular Orbital in Epi at physiological pH is located on the catechol ring, and that electrons in the ring are redistributed towards C atoms that carry hydroxyl groups 51 . This makes these -OH groups even harder bases than hydroxyl groups on aliphatic chains. Hence, Epi binds stronger to Fe 3+ than to Fe 2+ due to matched hard-hard interaction. Epi-catalyzed oxidation of Fe 2+ by O 2 results in the production of H 2 O 2 and HO • , and in the formation of Epi-Fe 3+ complexes. Epi is not an electron donor. It is degraded only by reactive by-products, which was prevented by HO • -scavenging activity of high-concentration Tris.
The 1:1 complex appears to be more (patho)physiologically relevant species. Iron is the most abundant transition metal in human plasma with a total concentration of 10-30 μM. The amount of labile iron (different redox-active Fe complexes with small ligands) is variable 52 .
[Epi] in human plasma may reach values >50 nM in response to stress. The concentration can be drastically higher locally, as well as in some pathological conditions, such as adrenal gland tumours (up to 3.5 μM), that are also accompanied by cardiovascular complications 53 . Nevertheless, the concentration of labile iron still appears to be higher than Epi. In addition, the 1:1 complex develops in the phosphate buffer even at higher [Epi]/[Fe 3+ ]. Epi may contribute to the labile iron pool in plasma, thus increasing the solubility of iron, and promoting its redox activity, which is a foe of physiological milieu. The 1:1 complex may even act as a distinct entity with functions that are yet to be discovered. Importantly, Epi-catalyzed oxidation of Fe 2+ , the soluble form of iron in human plasma, represents a plausible chemical mechanism of the cardiotoxic effects of stress-related high Epi concentrations. Hydrogen peroxide is known to pass the cell membrane to hit sensitive intracellular targets, whereas HO • induces membrane lipid peroxidation 52 . It is important to point out that Epi-induced oxidative stress requires the reduced form of iron and that Epi cannot reduce Fe 3+ . This implies that reducing agents (i.e. antioxidants) might not be a beneficial prophylaxis for cardiovascular diseases 54,55 .
The dependence of stoichiometry on the concentration ratio that was established here, has been observed previously for the binding of catecholamine moieties in biopolymers to Fe 3+ 56 . This, as well as the analogy in EPR and Raman spectra 26,28,33 , implies that Epi and Fe 3+ might represent a good experimental model for cross-linking in catecholamine-rich polymers. Slow cross-linking reaction with Fe 3+ is the rate-limiting step in the development of adhesion in such polymers. We have shown that Epi-Fe 3+ complexes developed ~10× faster when Fe 2+ , instead of Fe 3+ , was available to Epi. This may be explained by the fact that the highly soluble Fe 2+ is generally more accessible to ligands, whereas Epi competes with OH − ions for Fe 3+ 57 . The pre-binding of catecholamine moieties to Fe 3+ at low pH has been proposed to increase the efficiency of cross-linking that is initiated by pH increase 27 . Our results indicate that the application/pre-binding of Fe 2+ followed by (spontaneous) oxidation at pH >7, may be a simple alternative strategy for cross-linking promotion.

Methods
Chemicals. All chemicals were of analytical grade: Epi (L-adrenaline; Fluka Biochemika, Buchs, Switzerland), FeCl 3 (Analytika Ltd., Prague, Czech Republic), FeSO 4 (Sigma-Aldrich, St. Louis, MO, USA), Tris (Serva, Heidelberg, Germany). All experiments were performed using bidistilled deionized ultrapure (18 MΩ) water. Stock solutions of Epi (0.2 or 0.4 mM) were prepared fresh each day in 10 mM Tris buffer pH 7.4 and stored on ice in the dark. For Raman spectroscopy and reference UV/Vis and cyclic voltammetry experiments Epi stock solutions were prepared in phosphate buffer (10 mM KH 2 PO 4 , pH 7.4). Epi in solution was repeatedly checked for stability using spectrophotometry. Stock solutions of FeCl 3 (40 mM) and FeSO 4 (40 mM) were prepared in water. Incubation and measurements were conducted in the dark at 293 K (except EPR).
UV/VIS spectroscopy. UV-Vis absorption spectra were obtained using 2501 PC Shimadzu spectrophotometer (Kyoto, Japan). Sample volume was 1 mL. Scan time was 50 s. Samples were freshly prepared and immediately scanned at wavelengths from 800 to 200 nm. Changes of spectra were monitored for at least 30 min.
EPR spectroscopy. Low-T EPR spectra of Fe 3+ were recorded on a Bruker Elexsys II E540 spectrometer operating at X-band (9.4 GHz). Measurements at 100 K were performed using the Bruker N 2 Temperature Controller ER4131VT. Measurements at 20 K were conducted using Oxford Instruments ESR900 helium cryostat. The experimental parameters were: microwave power, 3.2 mW; scan time, 80 s; modulation amplitude, 0.5 mT; modulation frequency, 100 kHz; number of accumulations, 4 (at 100 K) and 2 (at 20 K). At both T, signal amplitude vs. power plot was built to determine the maximum power value. Approximately one half of the maximal power was applied to avoid saturation. All spectra were baseline corrected. Samples were placed in quartz cuvettes (Wilmad-LabGlass, Vineland, NJ, USA) after 1 min (Fe 2+ ) or 15 min (Fe 3+ ) incubation period, and quickly frozen in cold isopentane.
EPR spin-trapping experiments were conducted using DEPMPO spin-trap (Enzo Life Sciences, Inc. Farmingdale, NY, USA) at the final concentration of 5 mM. Hydroxyl radical was generated in the Fenton reaction: Fe 2+ (0.4 mM) + H 2 O 2 (1.2 mM; Carlo Erba Reagents, Milano, Italy). Spectra were recorded after 5 min incubation period using a Varian E104-A EPR spectrometer operating at X-band (9.53 GHz) with the following settings: modulation amplitude, 0. Raman spectroscopy. The Raman spectra were recorded using a DXR Raman microscope(Thermo Fisher Scientific, Waltham, MA, USA). Aliquots of 5 μL solution were placed on calcium fluoride glass and measured under the microscope (with objective magnification of 50×), using the 532 nm laser excitation line, with a constant power illumination of 10 mW. The exposure time was 30 s, with 10 exposures. The laser spot diameter was 1 μm. The scattered light was analyzed by the spectrograph equipped with a 900 lines mm −1 grating using 50 μm slit as spectrograph aperture. In the cases with high fluorescence background, automatic fluorescence correction was performed using the OMNIC software (Thermo Fisher Scientific).
Cyclic voltammetry. The voltammetric measurements were performed using a potentiostat/galvanostat CHI 760b (CH Instruments, Inc, Austin, TX, USA). The electrochemical cell was equipped with: a boron-doped diamond electrode (inner diameter of 3 mm; Windsor Scientific LTD, UK) embedded in a polyether ether ketone body with an inner diameter of 3 mm, a resistivity of 0.075 Ω cm, and a boron doping level of 1000 ppm (working electrode); Ag/AgCl (3 M KCl) (reference electrode); and Pt wire (counter electrode).
Oximetry and redox potential measurements. [O 2 ] was determined using a Clark type oxygen electrode (Hansatech Instruments Ltd., King's Lynn, UK) operating with Lab Pro interface and Logger Pro 3 software (Vernier, Beaverton, OR, USA). All systems were stirred and recorded for 2-5 min before Fe 2+ addition to establish the stability of baseline and zero rate of O 2 change. Decrease in [O 2 ] was monitored for 5 min before the addition of CAT (100 IU; Sigma-Aldrich). Redox potentials were recorded by InLab Redox Micro redox electrode operating with Seven Compact S210 pH meter and LabX software (Mettler-Toledo International Inc., Columbus, OH, USA). Measurements under anaerobic conditions were performed in N 2 dry box (Plas-Lab, Lansing, MI, USA).

Statistics.
All experiments were performed in triplicate. Statistical analysis was performed in STATISTICA 8.0 (StatSoft Inc., Tulsa, OK, USA) using nonparametric 2-tailed Mann-Whitney test (P < 0.05) and optimal curve fitting protocols. The goodness of fits was evaluated by R 2 (the adjusted r-square value).