An Electrochemical Study on the Effect of Metal Chelation and Reactive Oxygen Species on a Synthetic Neuromelanin Model

Neuromelanin is present in the cathecolaminergic neuron cells of the substantia nigra and locus coeruleus of the midbrain of primates. Neuromelanin plays a role in Parkinson's disease (PD). Literature reports that neuromelanin features, among others, antioxidant properties by metal ion chelation and free radical scavenging. The pigment has been reported to have prooxidant properties too, in certain experimental conditions. We propose an explorative electrochemical study of the effect of the presence of metal ions and reactive oxygen species (ROS) on the cyclic voltammograms of a synthetic model of neuromelanin. Our work improves the current understanding on experimental conditions where neuromelanin plays an antioxidant or prooxidant behavior, thus possibly contributing to shed light on factors promoting the appearance of PD.


INTRODUCTION
Melanins are a family of biopigments ubiquitous in flora and fauna. The black-brown eumelanin, red-yellow pheomelanin and neuromelanin all belong to the melanin family (Bush et al., 2006). Neuromelanin is mainly present in the cathecolaminergic neuron cells of the substantia nigra and locus coeruleus of the midbrain of primates (Marsden, 1961;Zecca et al., 2004). Neuropathological studies report on the loss of pigmented neurons of the substantia nigra in patients affected by Parkinson's disease (Zecca et al., 2002;Double et al., 2003Double et al., , 2008. It has been proposed in the literature that neuromelanin could have a role in neurotransmission (Meredith et al., 2013). Electron microscopy studies revealed that neuromelanin has a core-shell pheomelanin-eumelanin structure (Bush et al., 2006). Consequently, eumelanin can be proposed as a chemical model of neuromelanin to study interfacial processes between neuromelanin and its surroundings. Sepia melanin is the eumelanin extracted from the ink sac of cuttlefish (Schroeder et al., 2015).
Electrochemical studies on neuromelanin samples in media, including transition metal ions or ROS, are expected to contribute to shed light on the nature of the experimental conditions, where neuromelanin plays an antioxidant or a prooxidant behavior. Our groups reported on the electrochemical behavior of synthetic DHI-and DHICA-melanins in the presence of NH + 4 , Na + , K + , and Cu 2+ , at pH 5. We observed that DHICA-melanin voltammograms showed better resolved features than DHI-melanin in NaCH 3 COO (aq) , whereas DHImelanin voltammograms featured quasi box-shaped behavior. In the presence of Cu 2+ , both DHICA-and DHI-melanin featured adsorption peaks . Zareba et al. reported that both precursors, DHI and DHICA, form melanin pigments in presence of • OH (Novellino et al., 1999). Cecchi et al. reported on the quantitative comparison between free radical scavenging and redox properties of eumelanin biopigments as measured by Briggs Rausher and Folin Ciocalteu assays, to study the antioxidant activity of Sepia and synthetic melanin (Cecchi et al., 2019). Kim et al. used spectroelectrochemical reverse engineering to demonstrate that the free radical scavenging properties of Sepia and fungal melanin are affected by the redox state of the melanin (Kim et al., 2017). The same research group, with a similar approach, demonstrated that pheomelanin has higher redox-based prooxidant activity than eumelanin (Kim et al., 2015). They also reported that, with respect to bare polydopamine, the redox properties of polydopamine-Fe 3+ complexes are strongly suppressed while those of polydopamine-Mg 2+ complexes are maintained (Liu et al., 2019).
In this work, using the cyclic voltammetry technique, we studied the effect of metal chelation, as well as the presence of ROS moieties on the behavior of DHI-DHICAmelanin, considered as neuromelanin synthetic model, and of DHICA-and DHI-melanin. Specifically, we considered iron and copper ions (with corresponding metal-modified melanin samples named from now on as Fe/melanin, Cu/melanin, Cu/Fe/melanin) as well as H 2 O 2 and • OH (Schroeder et al., 2015). For our voltammetric studies, we used an electrolyte mimicking the intraneuronal cell solution. The morphology of the Cu/Fe/melanin samples was studied by scanning electron microscopy (SEM), whereas the chemical effect on the surface of the melanin after exposure to metals and ROS was characterized by X-ray photoelectron spectroscopy (XPS).

Preparation of Melanin Samples on Carbon Paper
We synthesized DHI-melanin, DHICA-melanin, DHI-DHICAmelanin (1.3:1 mol:mol) in situ (Pezzella et al., 1997), on carbon paper current collectors (Spectracarb TM 2050A, 10 mils) by solidstate polymerization from the corresponding building blocks (Pezzella et al., 2015). DHI and DHICA building blocks were synthesized as described hereafter (D'Ischia et al., 2005;Pezzella et al., 2015). Ten milligrams per milliliter solutions of DHI and/or DHICA monomers in methanol (99.8%, Sigma Aldrich) were prepared in ambient conditions. For DHI-DHICA-melanin, 10 mg of powder, including 5 mg of DHI monomer powder and 5 mg of DHICA monomer powder, were dissolved in methanol, in ambient conditions, and the solution was used as a precursor. The monomer solutions (5 µl) were drop cast on carbon paper featuring a geometric area of 0.5 cm 2 (therefore the loading of the melanin samples was ca 0.1 mg cm −2 ). After drop casting, the samples were exposed overnight to NH 3 vapors from NH 3(aq) (Sigma Aldrich, 28-30% w/v) to catalyze the polymerization.

X-Ray Photoelectron Spectroscopy (XPS)
The XPS survey scan and high-resolution XPS analysis was carried out with a VG ESCA-LAB 3 MKII instrument under Mg Ka radiation by applying 300 W (15 kV, 20 mA) power. The pressure in the chamber during the analyses was 3.0 × 10 −9 Torr. The high-resolution spectra were acquired with a pass energy of 20 eV and electrons were collected at a 0 • takeoff angle. Peak fitting was performed with symmetrical Gaussian-Lorentzian product functions after Shirley background subtraction. Wagner sensitivity factors were used to normalize the peak intensities for quantification.

Scanning Electron Microscopy (SEM)
SEM images were acquired at an acceleration voltage of 15 kV in the backscattered electron imaging mode using a FEI Quanta 450 Environmental Scanning Electron Microscope (FE-ESEM).

Cyclic Voltammograms of Melanins
We initially collected cyclic voltammograms of DHICA-melanin, DHI-melanin and DHI-DHICA-melanin, in electrolytes featuring different pH, namely pH 5 and 7 (Figure 2). At pH 5, the electrolyte was 0.25 M NaCH 3 COO, selected on the basis of previous cyclic voltammetry studies carried out in our groups (Wünsche et al., 2015;Kumar et al., 2016;Xu et al., 2017). At pH 7, the electrolyte solution (described in Experimental) was selected to mimic the intraneuronal liquid (Atherton et al., 2008;Kortleven et al., 2011). Surprisingly enough, two types of cyclic voltammograms are observable, both for DHICA-melanin and DHI-DHICA-melanin (Type 1 and Type 2, Figure 2), with all the experimental conditions fixed. Cyclic voltammograms of DHICA-melanin show oxidation FIGURE 2 | Cyclic voltammograms at 5 mV/s of (A) Type 1 and Type 2 DHICA-melanin in 0.25 M NaCH 3 COO pH 5, (B) Type 1 and Type 2 DHICA-melanin, at 5 mV/s, each for two cycles, in the simulated neurological fluid electrolyte pH 7, (C) Type 1 and Type 2 DHI-DHICA-melanin in 0.25 M NaCH 3 COO pH 5. DHI-mlanin only has one type of voltammogram. Only the second cycle is shown.
Frontiers in Bioengineering and Biotechnology | www.frontiersin.org peaks at ca 0.15 and 0.3 V vs. Ag/AgCl (3 M NaCl), at pH 5 (from now on named Type 1 DHICA-melanin, Figure 2A) and broad cathodic peaks at ca. 0.25 V and −0.2 V vs. Ag/AgCl. At pH 7, only one anodic peak is detectable at ca 0.2 V vs. Ag/AgCl for Type 1 DHICA-melanin (Figure 2B), whereas a cathodic peak is located at ca 0.2 V vs. Ag/AgCl. Type 2 DHICA-melanin shows additional anodic and cathodic peaks at both pHs. At pH 5, Type 2 DHICA-melanin features an additional sharp anodic peak at ca −0.05 V (Figure 2A) and a broad wave at ca. −0.1 V. At pH 7, Type 2 DHICA-melanin features two additional anodic peaks at ca. at ca −0.05 and 0.05 V and one additional cathodic peak at ca. −0.1 V (Figure 2B), with respect to Type 1. We tentatively explain the possibility to observe different voltammograms, for formally identical DHICA-samples, with differences in the supramolecular structure of DHICA-melanin. Such differences are attributable to the heterogeneity of the carbon paper where melanin is overgrown; carbon paper is made up of fibers and flat regions at fibers' interconnections (see later, SEM images). Results also show that cycling causes the evolution of the redox features in Type 2 DHICA-melanin, at pH 7 ( Figure 2B): the intensity of the oxidation feature at ca 0.2 V decreases, whereas one of the features at ca −0.05 V increases. Voltammograms of DHI-DHICA-melanin can also feature two types of behavior, at pH 5 (from now on indicated as Type 1 and Type 2 DHI-DHICA-melanin, Figure 2C). The voltammogram of Type 1 DHI-DHICA-melanin is quasi boxshaped, whereas Type 2 DHI-DHICA-melanin features an oxidation peak at ca −0.05 V and a broad reduction feature. DHImelanin features a quasi-box-shaped behavior, in agreement with the literature, both at pH 5 and 7 (Figures 3C,F)  .
After the study of bare melanins, we exposed the pigment to iron and copper metal ions and ROS. The concentrations of Fe 3+ and Cu 2+ were chosen based on the reported concentrations of Fe:melanin ratio in the substantia nigra and Cu:melanin ratio in the locus coeruleus. The concentration of H 2 O 2 was chosen based on the concentration reported for endogeneous H 2 O 2 in the neuron cells (Zareba et al., 1995;Chen et al., 2001;Liu et al., 2004;Zecca et al., 2004Zecca et al., , 2008Huang et al., 2005).

Cyclic Voltammograms of Fe/Melanin Prepared by Pre-immersion (Route I)
Fe 3+ has considerably higher physiological concentrations with respect to Cu 2+ in the brain (highest Fe:melanin ratio is ca 0.2 mol:mol in the substantia nigra, whereas the highest Cu:melanin ratio is 0.002 mol:mol in the locus coeruleus). We therefore considered, initially, Fe 3+ for the study of the effect of metal cations on the voltammetric properties of melanin. We initially adopted the pre-immersion route (route (i), Table 1) for the preparation of the samples (Fe/DHI-melanin, Fe/DHICAmelanin and Fe/DHI-DHICA-melanin, Figure 3).
The cyclic voltammograms were obtained in solutions at pH 7 at different iron concentrations and potential sweeping rates (Figure 3). Voltammograms obtained with Fe/DHICAmelanin (Type 1 DHICA-melanin, Fe:DHICA-melanin 0.04 mol:mol) are quite similar to those of bare melanin (Figure 2 and Figures 3A,D). As the molar ratio of Fe:DHICA-melanin increases, the oxidation feature shifts anodically, to reach ca 0.2 V at 0.2 mol:mol (Figures 3A,D; Bard and Faulker, 2001). The anodic shift of the oxidation could suggest that melanin tends to have prooxidant behavior after chelating Fe 3+ (Liu et al., 2019). We wish to remind here that an antioxidant is a substance that significantly delays or inhibits the oxidation of an oxidizable chemical substrate (Halliwell and Gutteridge, 1995). Oxidative stress is the imbalance between oxidants and antioxidants in favor of the oxidants, potentially leading to the damage of the substrate (Sies, 2000). In this context, being prooxidant means to favor the oxidative stress (Kim et al., 2015). In Fe/DHI-DHICAmelanin (Type 1 DHI-DHICA-melanin) and Fe/DHI-melanin voltammograms, no peaks are observable, unlike their bare counterparts (Figures 3B,C,E,F).
We did not conduct cyclic voltammetry of Cu/melanin or Cu/Fe/melanin samples prepared by pre-immersion.
The presence of Cu 2+ in physiological concentration (Cu:melanin 0.002 mol:mol) was not expected to affect the shape of cyclic voltammetry of melanin or Fe/melanin significantly.

SEM Images of Cu/Fe/Melanin Prepared by Pre-immersion (Route I)
Using SEM, we investigated samples prepared by the preimmersion route, to gain insight on the morphology of the samples. DHI-melanin, DHICA-melanin and DHI-DHICAmelanin modified with copper and iron will be indicated from now on as Cu/Fe/DHI-melanin, Cu/Fe/DHICA-melanin and Cu/Fe/DHI-DHICA-melanin. Cu/Fe/DHICA-melanin on carbon paper feature rod-shaped aggregates, mainly located at the junctions of the carbon paper fibers ( Figure 4A). SEM images of Cu/Fe/DHI-DHICA-melanin show granular aggregates ( Figure 4B). Rod-shaped and granular aggregates have sizes in the micrometric scale. Copper and/or iron chelation likely cause the morphological changes of DHICAand DHI-DHICA-melanin with respect to bare melanins, where no such aggregates are observable (Figures 4A,B; Kumar et al., 2016;Xu et al., 2017). No characteristic features are observable in the SEM images of Cu/Fe/DHI-melanin ( Figure 4C); this type of sample is not distinguishable from bare carbon paper, probably due to the low amount of Cu and Fe chelation in DHI-melanin ( Figure 3C in Xu et al., 2017).

XPS to Study the Presence of Metals in Cu/Fe/Melanin Prepared by Pre-immersion (Route I)
After the characterization of their morphology, samples loaded on carbon paper and fused silica, prepared by preimmersion, were studied by XPS, to shed light on the presence of iron and copper. Iron cations are detected on Cu/Fe/melanin samples loaded on fused silica (Figure S7  Table 1). Only the second cycle is shown. and Table S2). Copper cations are barely detected here, probably due to the low concentration of Cu 2+ used during samples' preparation [calculated to be ≤ ca. 0.018 atomic % for Cu/Fe/DHI-melanin, i.e., below the detection limit of XPS (0.1 atomic%)]. Copper and iron are barely detected on samples loaded on carbon paper ( Figure S6 and Table S1), likely due to the three-dimensional, open structure of the carbon paper, which is not ideal for XPS studies. Interestingly, literature reports that iron ion-binding capacity of neuromelanin is 10-fold greater that of synthetic dopa melanin (Double et al., 2003;Costa et al., 2012).

Effect of the Addition of Fe 3+ to the Electrolyte on Cyclic Voltammograms of DHICA-Melanin (Route II)
Based on the more resolved voltammetric features observable with Fe/DHICA-melanin prepared by pre-immersion, with respect to Fe/DHI-and Fe/DHI-DHICA-melanins (Figure 3), we selected DHICA-melanin to study the effect of the presence of Fe 3+ in the electrolyte solution where bare melanin samples are immersed (route ii). In this type of experiment, we added Fe 2 (SO 4 ) 3 in the electrolyte to form Fe/DHICA-melanin (Type 1 DHICA-melanin, Figures S1A, S2). Similar results as produced for pre-immersed Fe/DHICA-melanin, i.e., an anodic shift of the oxidation potential upon increase of Fe 3+ concentration, were observed (Type 1 DHICA-melanin, Figure S1A).

Effect of Cu 2+ Addition in the Electrolyte to Cyclic Voltammograms of DHICA-Melanin (Route II)
We added Cu(CH 3 COO) 2 in the electrolyte to form Cu/DHICAmelanin (route ii, Figure 5A). The presence of Cu 2+ in physiological concentration (Cu:melanin 0.002 mol:mol) does not seem to affect the shape of the cyclic voltammogram of DHICA-melanin but an oxidation peak at ca 0.1 V in the first voltammetric cycle, attributable to the adsorption process of Cu 2+ cations on the Type 1 DHICA-melanin ( Figure 5C). Effect of Cu 2+ Addition in the Electrolyte to Cyclic Voltammograms of Fe/DHICA-Melanin (Route II) In the electrolyte, after adding Fe 2 (SO 4 ) 3 solutions to form Fe/DHICA-melanin, we added Cu(CH 3 COO) 2 (Table 1 and Figure S1B). The presence of Cu 2+ in physiological concentration (Cu:melanin 0.002 mol:mol) does not seem to affect the shape of the cyclic voltammogram of Fe/DHICA melanin (Figure S1B), when copper ions are added after iron ions.

Effect of Fe 3+ Addition in the Electrolyte to Cyclic Voltammograms of Cu/Melanin (Route II)
In the electrolyte, after adding Cu(CH 3 COO) 2 solutions to form Cu/DHICA-melanin, we added Fe 2 (SO 4 ) 3 solution to form Cu/Fe/melanin (Figures 5A,C). Interestingly, after adding Fe 3+ in the electrolyte, an oxidation peak at ca 0.1 V appears in the first cycle, attributable to the adsorption process of the Fe 3+ cations on the melanin (Figure 5). The presence of high concentration of Fe 3+ (corresponding to Fe:melanin of 0.33 mol:mol), in simultaneous presence with Cu 2+ , leads to a more pronounced oxidation between 0/0.4 V and an additional reduction wave between 0.2/−0.3 V for DHICAmelanin ( Figure 5A).

Effect of H 2 O 2 on Melanin
Besides the effect of the presence of iron and copper ions, we studied the effect of ROS on the voltammetric behavior of bare melanin, to gain insight on a different aspect of the antioxidant behavior of melanin. Initially, we conducted experiments on the effect of exposure to H 2 O 2 on melanin samples (Figure 6 and Figure S3).
Literature report that H 2 O 2 can oxidize melanin into pyrrolic acids in neutral and alkaline media (Figure 1; Sarna et al., 1986;Korytowski and Sarna, 1990;Smith et al., 2017). After exposure to H 2 O 2 , Type 2 DHICA-melanin has more intense reduction and oxidation features ( Figure 6A). In the case of Type 1 DHI-DHICA-melanin, H 2 O 2 brings about the appearance of broad oxidation and reduction features at ca 0 V ( Figure 6B). The cyclic voltammogram of DHI-melanin is not affected by H 2 O 2 ( Figure S3). Among other factors, the π-π stacking structure of DHI-melanin is expected to feature lower reactivity toward H 2 O 2 with respect to DHICA-melanin, where H-bonding contributes to the formation of the supramolecular structure (Panzella et al., 2013). The effect of H 2 O 2 on bare carbon paper is negligible ( Figure S4).

XPS to Study the Effect of H 2 O 2 on Melanin
In order to explain the changes induced in melanin by exposure to H 2 O 2 , we used the XPS technique. For the XPS study, we considered DHI-DHICA-melanin samples (Figures S8, S9 and Table S3). Regarding changes on C 1s, ca 37% of the carbon atoms engaged in C=C disappear after exposure to H 2 O 2 (from 26.0 to 16.3 at%). An increase of the presence of C-O (by 27%), C=O (by 43%), and O-C=O (by 17%) functional groups is observable. At the same time, the amount of C-C remains the same ( Figure S9 and Table S3). These results point to the production of catechol, quinone and carboxyl group on the aromatic rings of melanin. Regarding changes involving O 1s, the presence of aliphatic C-OH groups increases by ca 23% (from 7.8 to 10.3 at%) and the presence of aromatic C-OH groups increases by ca 33% (from 5.4 to 7.2 at%).  The increased amount of redox active catechol and quinone can explain the more pronounced redox features observed in DHICA-melanin and DHI-DHICA-melanin, after exposure to H 2 O 2 (Figure 6).

Effect of • OH on Melanin
Based on the more pronounced effect of H 2 O 2 on DHICAmelanin with respect to DHI-DHICA-and DHI-melanin, DHICA-melanin was selected to gain insight on possible effects of neuromelanin exposure to • OH ( Figure S5). • OH was generated by the Fenton reaction (see Experimental part and Table 1). • OH moieties are expected to cause the oxidation or hydroxylation of melanin (Sarna et al., 1986;Korytowski and Sarna, 1990;Huang et al., 2005;Smith et al., 2017) After exposure to • OH, the behavior of Type 1 DHICA-melanin is similar to that observed after exposure to H 2 O 2 , i.e., more pronounced reduction and oxidation features (Figure S5 and Figure 6), suggesting that H 2 O 2 and • OH moieties play a similar effect on melanin. We did not observe the significant decrease of the current that we somehow expected due to melanin degradation by • OH moieties at acidic and neutral pH (Pilas et al., 1988;Zareba et al., 1995;Zecca et al., 2003;Brillas et al., 2009).

CONCLUSIONS AND PERSPECTIVES
In this work, we studied the effect of metal ions and ROS on the redox (cyclic voltammetry) properties of a neuromelanin model, in order to shed light onto possible relationships between the voltammetric properties and the antioxidant vs. prooxidant behavior of neuromelanin. Considering that literature proposes a core-shell pheomelanin-eumelanin structure for neuromelanin, we made the hypothesis that eumelanin, the shell wet by the electrolyte, is a good model to study the interfacial properties of neuromelanin and, among them, redox processes. SEM images showed the presence of changes in the morphology of DHI-DHICA-melanin and DHICA-melanin (obtained from a mix of the building blocks or exclusively from one of building block of eumelanin, DHI, and DHICA) upon the simultaneous presence of iron and copper cations, with respect to bare melanin samples. Both Cu/Fe/DHI-DHICA-and Cu/Fe/DHICA-melanin, prepared by immersing melanin electrodes in solutions of the metals, form aggregates in the micrometric range. XPS showed the presence of iron in Cu/Fe/DHI-DHICA-, Cu/Fe/DHI-, and Cu/Fe/DHICA-melanin, whereas the detection of copper was more elusive.
We observed changes in the voltammetric properties of DHICA-melanin, in the presence of iron ions: an anodic shift of the oxidation feature was observed in the voltammograms (Fe:melanin 0.2 mol:mol), which could suggest that the antioxidant properties of DHICA-melanin tend to be weaker when it chelates iron ions.
We also observed the evolution of the voltammetric properties of DHI-DHICA-and DHICA-melanins upon exposure to H 2 O 2 . The presence of H 2 O 2 renders the voltammetric features for DHICA-melanin and DHI-DHICA-melanin more pronounced with respect to non-exposed melanin, likely due to the increase of the density of catechol and quinone groups after exposure to H 2 O 2 , in agreement with XPS results.
It has been more challenging to detect the effect of metal ions and ROS on DHI-melanin. Among other factors, the π-π stacked structure of DHI-melanin could bring lower reactivity with respect to DHICA-melanin. Indeed, literature reports that the redox properties of DHICA-melanin are attributable to the destabilizing effects of hindered intermolecular conjugation, which lead to non-planar structures with monomer-like behavior (Panzella et al., 2013).
Our results seem to suggest that DHICA is a more reactive component in eumelanin. We propose to consider the reactivity of the DHICA component, with respect to the DHI component, during studies on the loss of pigmented neurons of the substantia nigra in patients affected by Parkinson's disease.
In perspective, we plan to improve our neuromelanin model including a pheomelanin component in the structure and we wish to follow by time-resolved Electron Paramagnetic Resonance the effect of ROS on the charge transfer properties of neuromelanin.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the manuscript/Supplementary Files.

AUTHOR CONTRIBUTIONS
RX designed the experiments and drafted the manuscript. FS helped in the interpretation of the electrochemical measurements.
CS proposed the idea, supervised the work, and helped drafting the manuscript. All the Authors gave their critical contribution on the whole manuscript.

FUNDING
CS and FS acknowledge the Italy-Quebec Mobility program (MRIF and Italian Minister of Foreign Affairs) for financial support. CS was grateful to NSERC (Discovery). ACKNOWLEDGMENTS E. Di Mauro was acknowledged for careful proof-reading. Y. Drolet was acknowledged for technical support.