Evaluation of the Effects of Fucoidans from Fucus Species and Laminaria hyperborea against Oxidative Stress and Iron-Dependent Cell Death

Fucoidans are algal polysaccharides that exhibit protective properties against oxidative stress. The aim of this study was to investigate different fucoidans from brown seaweeds for their ability to protect against iron-dependent oxidative stress (ferroptosis), a main hallmark of retinal and brain diseases, including hemorrhage. We investigated five new high-molecular weight fucoidan extracts from Fucus vesiculosus, F. serratus, and F. distichus subsp. evanescens, a previously published Laminaria hyperborean extract, and commercially available extracts from F. vesiculosus and Undaria pinnatifida. We induced oxidative stress by glutathione depletion (erastin) and H2O2 in four retinal and neuronal cell lines as well as primary cortical neurons. Only extracts from F. serratus, F. distichus subsp. evanescens, and Laminaria hyperborea were partially protective against erastin-induced cell death in ARPE-19 and OMM-1 cells, while none of the extracts showed beneficial effects in neuronal cells. Protective fucoidans also attenuated the decrease in protein levels of the antioxidant enzyme GPX4, a key regulator of ferroptosis. This comprehensive analysis demonstrates that the antioxidant abilities of fucoidans may be cell type-specific, besides depending on the algal species and extraction method. Future studies are needed to further characterize the health-benefiting effects of fucoidans and to determine the exact mechanism underlying their antioxidative abilities.


Introduction
Fucoidans, or sulfated fucans, are polysaccharides of algae cell walls containing fucose sugar and sulfate ester groups. Naturally, they are important for the integrity of the algae cell wall and fend off pathogens and other harsh environmental effects of the ocean and prevent dehydration [1,2]. They exhibit many different biological activities that depend on the origin of the extract, harvest time, the extraction process, chemical composition, structure, and size [3][4][5][6]. High-molecular weight fucoidans reduce inflammatory cell Table 1. Composition of neutral monosaccharides, degree of sulfation, and protein content in the extracted fucoidans.

Monosaccharide Composition (mol %) a Degree of Sulfation b Protein Content (%) c Fuc Xyl Man Gal Glc
The molecular mass (M) and size characteristics of the new fucoidan samples were analyzed by size-exclusion chromatography with triple detection, i.e., multi-angle light scattering (MALS), viscometry and refractive index (RI). MALS/RI provided the weightaverage and number-average molecular weights (MW, Mn) and the root-mean square radius (rg), and the viscometer allowed us to measure the hydrodynamic radius (rh) and intrinsic viscosity [ŋ] of the samples. The MW ranged from 52-1548 kDa ( Table 2). The different extraction methods used resulted in a large variation in MW (FV1: 261 kDa and FV2: 1438 kDa) among fucoidans from the same alga (F. vesiculosus). The fucoidan from Sigma-Aldrich (FVs) was the smallest one (52 kDa), whereas Fuc1 was the largest (1548 kDa). Table 2. Molecular mass and size characteristics of the fucoidans with number-average molar mass (Mn), molecular weight (MW), polydispersity (PD; MW/Mn), rms radius (rg), hydrodynamic radius (rh), and intrinsic viscosity ([ŋ]).

MW (kDa)
Mn ( [4]. According to the MW-and rh-versus-retention time plots, the fucoidans were heterogeneously composed and differed from each other ( Figure 1A,B). Fuc1 was previously described in  as a large random coil shape molecule [4]. The MW and rh of the samples in this study showed a very broad distribution. After a regular decrease, both MW and rh rose slightly towards the higher elution time, which is quite common for branched as well as charged molecules [28].
In addition, the overall shape of the fucoidan samples was determined by Mark-Houwink-Sakurada (MHS) analysis, where [ŋ] is plotted against MW. The MHS slope (αŋ) provides information about the conformation of macromolecules, whereby a value of 0.0 corresponds to an ideal solid sphere, 0.5-0.8 to random coils, and 0.8-1.8 to rigid chains [29]. The MHS slope values of the main fraction of all fucoidans (except FV2) aligned with the theoretical value of random coil conformation (αŋ = 0.77-0.89, Figure 2A,B). Additionally, Fuc1 was described to have a random coil structured main chain with highly branched short side chains [4]. In contrast, the lower MHS slope (αŋ = 0. 48

) of FV2
] of the samples. The M W ranged from 52-1548 kDa ( Table 2). The different extraction methods used resulted in a large variation in M W (FV1: 261 kDa and FV2: 1438 kDa) among fucoidans from the same alga (F. vesiculosus). The fucoidan from Sigma-Aldrich (FVs) was the smallest one (52 kDa), whereas Fuc1 was the largest (1548 kDa). The molecular mass (M) and size characteristics of the new fucoidan samples w analyzed by size-exclusion chromatography with triple detection, i.e., multi-angle l scattering (MALS), viscometry and refractive index (RI). MALS/RI provided the wei average and number-average molecular weights (MW, Mn) and the root-mean squ radius (rg), and the viscometer allowed us to measure the hydrodynamic radius (rh) intrinsic viscosity [ŋ] of the samples. The MW ranged from 52-1548 kDa ( Table 2). different extraction methods used resulted in a large variation in MW (FV1: 261 kDa FV2: 1438 kDa) among fucoidans from the same alga (F. vesiculosus). The fucoidan f Sigma-Aldrich (FVs) was the smallest one (52 kDa), whereas Fuc1 was the largest (1 kDa). According to the MW-and rh-versus-retention time plots, the fucoidans w heterogeneously composed and differed from each other ( Figure 1A,B). Fuc1 previously described in Dörschmann et al. (2020) as a large random coil shape mole [4]. The MW and rh of the samples in this study showed a very broad distribution. Aft regular decrease, both MW and rh rose slightly towards the higher elution time, whic quite common for branched as well as charged molecules [28].
In addition, the overall shape of the fucoidan samples was determined by Ma Houwink-Sakurada (MHS) analysis, where [ŋ] is plotted against MW.
The MHS slope ( provides information about the conformation of macromolecules, whereby a value of corresponds to an ideal solid sphere, 0.5-0.8 to random coils, and 0.8-1.8 to rigid ch [29]. The MHS slope values of the main fraction of all fucoidans (except FV2) aligned w the theoretical value of random coil conformation (αŋ = 0.77-0.89, Figure 2A Additionally, Fuc1 was described to have a random coil structured main chain with hig branched short side chains [4]. In contrast, the lower MHS slope (αŋ = 0.48) of According to the M W -and rh-versus-retention time plots, the fucoidans were heterogeneously composed and differed from each other ( Figure 1A,B). Fuc1 was previously described in Dörschmann et al. (2020) as a large random coil shape molecule [4]. The M W and rh of the samples in this study showed a very broad distribution. After a regular decrease, both M W and rh rose slightly towards the higher elution time, which is quite common for branched as well as charged molecules [28].
In addition, the overall shape of the fucoidan samples was determined by The molecular mass (M) and size characteristics of the new fucoid analyzed by size-exclusion chromatography with triple detection, i.e., scattering (MALS), viscometry and refractive index (RI). MALS/RI prov average and number-average molecular weights (MW, Mn) and the r radius (rg), and the viscometer allowed us to measure the hydrodynam intrinsic viscosity [ŋ] of the samples. The MW ranged from 52-1548 kD different extraction methods used resulted in a large variation in MW (F FV2: 1438 kDa) among fucoidans from the same alga (F. vesiculosus). Th Sigma-Aldrich (FVs) was the smallest one (52 kDa), whereas Fuc1 was kDa). ) provides information about the conformation of macromolecules, whereby a value of 0.0 corresponds to an ideal solid sphere, 0.5-0.8 to random coils, and 0.8-1.8 to rigid chains [29]. The MHS slope values of the main fraction of all fucoidans (except FV2) aligned with the theoretical value of random coil conformation ( The molecular mass (M) and size characteristics of the new fuc analyzed by size-exclusion chromatography with triple detection, i. scattering (MALS), viscometry and refractive index (RI). MALS/RI p average and number-average molecular weights (MW, Mn) and th radius (rg), and the viscometer allowed us to measure the hydrodyna intrinsic viscosity [ŋ] of the samples. The MW ranged from 52-1548 different extraction methods used resulted in a large variation in MW FV2: 1438 kDa) among fucoidans from the same alga (F. vesiculosus) Sigma-Aldrich (FVs) was the smallest one (52 kDa), whereas Fuc1 w kDa).  Figure 2A,B). Additionally, Fuc1 was described to have a random coil structured main chain with highly branched short side chains [4]. In contrast, the lower MHS slope ( The molecular mass (M) and size characteristics of the analyzed by size-exclusion chromatography with triple dete scattering (MALS), viscometry and refractive index (RI). MA average and number-average molecular weights (MW, Mn) radius (rg), and the viscometer allowed us to measure the hy intrinsic viscosity [ŋ] of the samples. The MW ranged from different extraction methods used resulted in a large variatio FV2: 1438 kDa) among fucoidans from the same alga (F. vesi Sigma-Aldrich (FVs) was the smallest one (52 kDa), whereas kDa). Table 2. Molecular mass and size characteristics of the fucoidans wi mass (Mn), molecular weight (MW), polydispersity (PD; MW/Mn), rm radius (rh), and intrinsic viscosity ([ŋ]).

MW
(kDa) Mn (kDa) PD (MW/Mn) rg (nm = 0.48) of FV2 indicated a more compact overall structure, i.e., a random coil tending towards a spherical shape.  To note, it can be observed in the Mark-Houwink-Sakurada plot that the slope of FV2 is much smaller than that of FV3 (Figure 2A), representing a lower intrinsic viscosity due to the different degree of branching present in these two fucoidans. In general, branched molecules have lower viscosity in comparison to linear molecules [30,31]. In addition, the slope bends downwards at the high-MW region, which is due to the lower viscosity of the high-MW molecules. FV3 represents a random coil structure while FV2 is a comparatively compact structure. This is the reason why the smaller Mw fucoidans in this study had a higher average intrinsic viscosity.

Protective Effect of Fucoidans against Erastin-and H 2 O 2 -Induced Cell Death
We first determined the half-maximal lethal dose of erastin and H 2 O 2 in the human RPE cell line ARPE-19, the uveal melanoma cell line OMM-1, the mouse hippocampal neuronal cell line HT-22, the human neuroblastoma cell line SH-SY5Y, and primary mouse cortical neurons ( Figure 3). We used primary mouse cortical neurons because they more closely resemble neurons in vivo compared to cell lines, including their inability to prolifer-ate, which may be of relevance to their resistance to oxidative stress and iron-dependent cell death.
While   Table S1. Next, we assessed the different fucoidan extracts at the half-maximal lethal dose in the different cell types. Fucoidan extracts were added 30 min prior to erastin or H 2 O 2 treatment. For all cell lines, fucoidan concentrations between 0 and 50 µg/mL were applied. For primary cortical neurons, fucoidan concentrations of 0-5 µg/mL were used, because higher concentrations led to cell detachment due to the negative charge of fucoidans that interacted with the poly-l-lysine coating required for primary cortical neurons. Ferrostatin-1 was used as a positive control for erastin-induced oxidative stress in neurons.
Next, we assessed the neuronal cells HT-22 ( Figure 6), SH-SY5Y (Figure 7), and primary cortical neurons ( Figure 8) after treatment with the nine fucoidan extracts and erastin at the respective half-maximal lethal dose for 24 h. None of the extracts was protective against erastin-induced toxicity, which was also confirmed by calcein AM/propidium iodide staining. They also did not induce any toxicity when incubated without erastin. However, it has to be noted that primary cortical neurons were only treated with up to 5 µg/mL due to cell detachment at higher concentrations. In contrast, the positive control, ferrostatin-1, which prevents lipid peroxidation [32], was able to significantly increase survival of HT-22 cells (median: 101.43%, p = 0.011, Figure 6C) and primary cortical neurons (median: 123.59%, p = 0.011, Figure 8C), but not of SH-SY5Y cells (median: 45.74%, p = 0.465, Figure 7C). The latter inability may be explained by the much higher concentration of erastin needed to induce cell death in SH-SY5Y cells.
Taken together, the most promising fucoidan extracts were those from F. serratus (FS), F. distichus subsp. evanescens (FE), and L. hyperborea (Fuc1), which increased cell viability in erastin-induced cell death in ARPE-19 and OMM-1 cells. A summary of the effects of all the extracts at 50 µg/mL (5 µg/mL for primary cortical neurons) depending on the cell types is presented in Table 3.

Fucoidans FS, FE, and Fuc1 Abrogated the Decrease in GPX4 Protein Expression Induced by Erastin
Next, we sought to determine whether the protective effect of FS, FE, and Fuc1 in ARPE-19 and OMM-1 cells was due to their ability to increase the protein expression of GPX4. GPX4 is an antioxidant defense enzyme that reduces lipid peroxides into lipid alcohols and is known to play a crucial role in ferroptosis [33]. As expected, erastin reduced GPX4 expression (vehicle normalized to 1.0) in both cell lines (Figure 9, median: 0.5 for ARPE-19 cells, 0.4 for OMM-1 cells). The different protective fucoidan extracts concentration-dependently attenuated the erastin-induced decrease. In ARPE-19 cells ( Figure 9A), FS completely abrogated the GPX4 protein decrease starting at 10 µg/mL (median: 1.0), while Fuc1 treatment led to a partial recovery at 10-50 µg/mL (median: 0.7 for 10 µg/mL, 0.8 for 50 µg/mL). Concerning OMM-1 cells (Figure 9B), 50 µg/mL FE and Fuc1 increased GPX4 levels after erastin treatment (median: 0.7 for FE, 0.8 for Fuc1). However, the small sample size and large variation of the data were limitations of this study, and therefore, we did not perform statistical analyses. The effect of fucoidans on GPX4 should be confirmed in future studies.

Discussion
In this study, we characterized novel fucoidan extracts from different brown algae species and with different chemical properties. We demonstrated that they exert differential protective effects in ARPE-19 and OMM-1 cells exposed to iron-dependent oxidative stress induced by erastin (ferroptosis), while none of the assessed extracts was protective in neuronal ferroptosis. We further showed that this effect may be mediated by abrogating the decrease in GPX4 induced by erastin.
Here, we assessed iron-dependent oxidative stress (as induced by erastin leading to glutathione depletion, i.e., ferroptosis) because of its relevance in subretinal and brain hemorrhage as well as neurodegeneration. The overall protective effects of the tested fucoidans were limited. Only FS, FE, and Fuc1 showed partial protective effects against erastin-induced cell death in ARPE-19 ( Figure 4) and OMM-1 cells ( Figure 5). To note, we have previously used Fuc1, a high-molecular weight fucoidan that, besides a middle-sized and low-molecular weight L. hyperborea fucoidan, showed protective effects in OMM-1 cells against H 2 O 2 after 24 h, but not in ARPE-19 cells against TBHP treatment for 24 h [4]. We also demonstrated that F. vesiculosus fucoidan from Sigma-Aldrich had protective properties in several uveal melanoma cell lines [39], which may depend on the fucoidan batch used. Another previous study from our group suggested that fucoidans from brown algae species Saccharina latissima, L. digitate, F. distichus subsp. evanescens, F. serratus and F. vesiculosus also seem promising for oxidative stress protection in OMM-1 (24 h) [38].
The overall biological activities are known to depend on the algae species and extraction procedure, but it is in fact their structural composition that determines their activities. Accordingly, the pronounced differences between the structural characteristics of the various F. vesiculosus fucoidans tested in this study could explain why all F. vesiculosus fucoidans in this study did not show any real antioxidative properties. We already showed before that the extraction procedure and purification steps, and thus the resulting structural characteristics, can influence the protective ability of the fucoidans [5]. Importantly, in the previous studies, higher H 2 O 2 concentrations (1000 µM) were needed to exert 50% toxicity after 24 h in OMM-1 cells. We here decided to assess OMM-1 cells at 48 h using the half-maximal lethal dose of 250 µM H 2 O 2 , and hence the time of stimulation may have had an effect on the protective ability of the fucoidans as well. Future studies should also investigate whether repeated application of fucoidans may recover their protective effects.
What may be the reason for the difference in protective effects among the different extracts? Fucoidans are known for their heterogeneity, and many aspects could contribute to these differences: (1) All fucoidans contain mainly fucose, but the antioxidative properties seemed to be independent of the level of fucose [5]. Similarly, the levels of other monosaccharides were not indicative of the antioxidative potential of the fucoidan extracts.
(2) The level of protein also did not seem to make an impact, as both Fuc1 and FVs lack protein, while Fuc1 was protective and FVs was not protective against erastin-induced cell death. (3) Except for the very high degree of sulfation of Fuc1 (1.7), this parameter was quite heterogenous for all other fucoidans, and thus we were not able to observe a direct correlation to oxidative defense capabilities. Sulfate groups contribute to scavenging free radicals, but a steric hindrance of the polymer chains results in long, coil-like molecule structures. In addition, low-molecular weight fucoidans are more capable of scavenging radicals because the sulfate and hydroxyl groups are more exposed due to the compact molecules [40]. (4) Regarding size, we have previously demonstrated that higher molecular weight fucoidans are more promising in terms of their ability to protect against H 2 O 2 -induced death [4]. In this study, the middle-sized FE (148 kDa) and FS (245 kDa) as well as the larger Fuc1 (1548 kDa) showed protective effects against erastin in ARPE-19 and OMM-1 cells, while other fucoidans of similar size did not. We also showed before that high-molecular weight fucoidans are more promising in terms of oxidative stress protection and may interact in cellular pathways and not reactive oxygen species scavenging [4,5,38]. In addition, the polydispersity of the extracts investigated in this study ranged from 1.5 (Fuc1) to 7.5 (FE). Because the extracts on both ends were slightly protective, polydispersity likely did not affect the protective potential in this model. Overall, we were not able to find any plausible correlation to the structure data, extraction procedure, and chemical composition. As these are extracts and not pure compounds, the observed biological effect cannot be ascribed to a single structure. Each extract contains several fragments that may work synergistically or lead to compoundcompound interactions. Future studies should assess different fractions of the extracts to give insight into the relationship between structure and biological activity.
Furthermore, it was remarkable that the tested fucoidan species were only partially protective in the ocular cell lines, while their protective effects were absent in neuronal cell lines and primary neurons. Ocular and neuronal cells show similar antioxidative defense mechanisms such as superoxide dismutase, glyoxalase glutathione reductase, glutathione peroxidase, catalase and nuclear factor E2-related factor 2 (Nrf2) [41,42]. However, RPE cells have developed strong antioxidant defense mechanisms to withstand the high amounts of reactive oxidative species due to light exposure and high metabolism rates [43]. Hence, higher concentrations of erastin and H 2 O 2 may be needed to induce cell death in these cells. HT-22 cells are also known to be more sensitive to oxidative stress [41]. The main difference with uveal and RPE cells is the presence of melanin, which has antioxidative properties and can bind iron [44].
The possible interactions of fucoidans also depend on whether they are taken up by cells or only affect the cell from outside via scavenging oxidative reagents (such as H 2 O 2 ) or interacting with membrane receptors (Figure 10). Fucoidans have been suggested to activate toll-like receptors [45]. Measurements of fucoidans intracellularly to demonstrate that they can enter cells remain technically challenging [46]. Of note, a fluorescently labeled fucoidan from F. vesiculosus was recently shown to be taken up into Caco-2 colorectal carcinoma cells via clathrin-mediated endocytosis [47]. Furthermore, FV fucoidan from Sigma-Aldrich has been demonstrated to increase the activity of the antioxidant enzyme superoxide dismutase 1 [48], and fucoidans (undefined source) augmented nuclear factor E2-related factor 2, a master regulator of gene expression induced by oxidative stress [49][50][51][52]. In a rat model of acetaminophen-induced liver injury, orally administered fucoidan from F. vesiculosus attenuated the decrease in glutathione, superoxide dismutase, and GPX, while blocking the increase in malondialdehyde [53], a marker of lipid peroxidation.
In the present study, we further add first evidence that fucoidans can abrogate the decrease in the protein levels of the antioxidant enzyme GPX4 that is crucial for ferroptosis. We plan to examine other ferroptosis markers such as reactive oxygen species production, lipid peroxidation, and glutathione levels in the future. Further studies are needed to determine the exact mechanisms by which fucoidans exert their protective effects against oxidative stress.

Fucoidan Extraction
Three commercially available fucoidans were used. FVs was purchased from Sigma-Aldrich, and FVm and UPm were provided by Marinova (Cambridge TAS, Australia).
All algae extracts obtained newly for this study were washed from saprophytes before drying. High-molecular weight L. hyperborea fucoidan Fuc1 (1548.6 kDa), identified and obtained from Alginor ASA and described in [4,56], was used. The extraction method and chemical data for molecular weight, sulfate and monosaccharide content as well as structure were described by Dörschmann et al. 2019 [4]. Three Fucus species, F. vesiculosus (FV), F. serratus (FS), and F. evanescens (FE), were identified and harvested by Coastal Research & Management in Kiel Fjord in Germany. The algae were soaked in an ethanol solution (85%) for 15 h. After soaking, the algae were carefully washed with acetone and left to dry for 4 h. Once dried, the algae were ground into 1 mm particles and the fucoidans were prepared as follows: For FV1, F. vesiculosus (harvested in July 2017) was submerged in 100 mM hydrochloric acid (ambient temperature) for 24 h, decanting the acid and subsequently neutralizing it with 1 M sodium hydroxide. The neutralized extraction solvent was transferred to a tube and aqueous calcium chloride (35%) was added in amounts corresponding to 1% calcium chloride in the extraction solvent. The solution was centrifuged for 30 min and the resulting supernatant was recovered. Ethanol was added to the supernatant in amounts corresponding to a 40% ethanol concentration. The solution was centrifuged for another 30 min, and the supernatant was recovered again. Ethanol was added to a final concentration of 70% ethanol, and the solution was centrifuged again. The pellet was collected and washed with ethanol and acetone to dry. Next, the fucoidan pellet was solubilized in deionized water and dialyzed (MWCO = 12-14 kDa), until the conductivity of the deionized water remained constant. The fucoidan was subsequently freeze-dried prior to chemical and biological analysis.
FV2 fucoidan was prepared from the same algae used for FV1 fucoidan, using similar extraction conditions; however, the algae were submerged in hydrochloric acid four times for 24 h, with a decanting and fresh addition of hydrochloric acid after every 24 h. After 96 h, the acid was decanted and neutralized with 1 M sodium hydroxide and subjected to the same cleanup procedure as FV1.
The fucoidans FV3, FS and FE (harvested in October 2017) were prepared by adding 1.5 g of algae material and 100 mM hydrochloric acid to a microwave extraction vessel, which was heated at 80 • C for 30 min in a microwave digestion system (Multiwave GO, Anton Paar). After cooling, the extraction solvent was neutralized with 1 M sodium hydroxide. All fucoidans were recovered and purified by calcium chloride and ethanol, as described for FV1.
Endotoxin levels were not measured in the extracts used in this study. However, microwave irradiation and exposure to acids degrade endotoxins [57,58]. In addition, FVs had been previously shown to be ultra-pure with respect to endotoxins (below 100 EU/mL) [59]. In general, we considered all extracts to be endotoxin-free.
All fucoidans were dissolved in aqua bidest, stored at −20 • C and sterile filtrated before use.

Size Exclusion Chromatography with Multiple Detection
We determined the molecular weight, size, and chain conformation of the fucoidan samples using size-exclusion chromatography coupled with multiple detection. The instrument setup consisted of an Agilent 1200 series HPLC (Agilent Technologies, Waldbronn, Germany) connected to two OHPak LB-806M size-exclusion chromatography columns (8.0 mmID × 300 mmL) and a guard column (Shodex, Munich, Germany). The chromatographically separated samples were detected using an Agilent 1200 series UV detector, a DAWN™ MALS photometer, a ViscoStar ® on-line differential viscometer and an Optilab ® differential refractometer (Wyatt Technology, Dernbach, Germany) connected in series. The samples were eluted using a Na 2 HPO 4 -NaH 2 PO 4 (50 mM) buffer solution (Ph 7.0) containing 150 mM NaCl. The samples (2 mg/mL) were filtered using a 0.45 µm filter before injection to remove any large particles. The used dn/dc was 0.150 mL/g for each fucoidan. Data acquisition and analysis were performed using ASTRA ® 8.0 software (Wyatt Technology, Dernbach, Germany).

Elemental Analysis
The sulfur and nitrogen contents of the fucoidan samples were determined by elemental analysis as previously described [60]. The percentage of sulfate groups (calculated as -SO 3 Na) was used to calculate the degree of sulfation. The total protein content was estimated by multiplying the nitrogen content (%) by 6.25.

Monosaccharide Composition by Gas-Liquid Chromatography of Alditol Acetates
We determined the neutral monosaccharide composition of the fucoidans by acetylation analysis as previously described [60]. Briefly, the fucoidans were hydrolyzed using trifluoroacetic acid [61], reduced and acetylated to obtain alditol acetate derivatives (AA) [62]. The AA were separated using gas-liquid chromatography, and the percentage of the respective AA were calculated as previously reported [37].

Treatments
First, different concentrations of erastin (0.1-30 µM, Cayman Chemical, Ann Arbor, MI, USA) and H 2 O 2 (100-1000 µM, Sigma-Aldrich) diluted in cell culture medium were tested to determine the half-maximal lethal dose at 24 or 48 h, depending on the cell type. Next, the protective effect of fucoidan on erastin-or H 2 O 2 -induced oxidative stress at halfmaximal lethal dose was assessed. Therefore, we added erastin or H 2 O 2 30 min after fucoidan treatment and incubated the cells for 24 or 48 h. For primary cortical neurons, fucoidan concentrations of 0-5 µg/mL were used because higher concentrations led to cell detachment. For all cell lines, fucoidan concentrations between 0 and 50 µg/mL were applied. Ferrostatin-1 (Sigma-Aldrich) was used as a positive control for erastin-induced oxidative stress in neurons and was incubated as described for fucoidan.

Cell Viability
We determined the cell viability 24 h after erastin or H 2 O 2 exposure, except for OMM-1 cells that were assessed at 48 h because 24 h were not sufficient to kill the cells with erastin. We used the 1-(4,5-dimethylthiazol-2-yl)-3,5-diphenylformazan (MTT, Carl Roth GmbH + Co. KG, Karlsruhe, Germany) assay, a colorimetric assay of cell metabolic activity [63]. The water-soluble tetrazolium salt is reduced by metabolically active cells to water-insoluble formazan. After 2-4 h of incubation depending on the cell type, MTT is removed and the cells are permeabilized with dimethylsulfoxide to visualize the formazan crystals. We measured the plates with a CLARIOstar Microplate Reader using CLARIOstar v.5.20 R5/MARS v.3.10 R5 (all BMG LABTECH, Ortenberg, Germany) or Elx800 microplate reader from BioTek (Bad Friedrichshall, Germany). The maximum extinction of formazan is at 550 nm, but at the same wavelength, the dissolvent dimethyl sulfoxide has an extinction of 0.03 OD. Therefore, the wavelength at 655 nm was also measured and subsequently subtracted from the OD value at 550 nm. The mean of four technical replicates was calculated for all conditions and subsequently normalized to the mean of the vehicle treatment without addition of erastin (representing one biological replicate).
The results of the population, quantitative assays of cell viability (MTT) were verified qualitatively using calcein acetoxymethyl (calcein AM, 2 µM, Santa Cruz Biotechnology, Dallas, TX, USA) and propidium iodide (3 µM, Carl Roth) incubated for 15-30 min directly in the cell culture medium without medium change. Calcein AM is cell-permeable and is cleaved by esterases in intact cells. The calcein can then bind to calcium ions, and this binding results in green fluorescence staining [64]. Propidium iodide is a nucleic staining dye that is not able to pass through a viable cell membrane, but can only pass disturbed cell membranes of damaged or dead cells. It interacts with the DNA double helix and emits red fluorescence [65]. Pictures were taken with an Axiovert 200 M or Axiovert 100 microscope (Carl Zeiss AG, Oberkochen, Germany).

Statistical Analysis
Normality was evaluated by the Kolmogorov-Smirnov test and variance homogeneity using Levené test. Because data were not normally distributed or variance homogeneity was not met, Kruskal-Wallis test was performed followed by the post hoc Mann-Whitney U test with α-correction according to Bonferroni-Holm to adjust for the inflation of type I error due to multiple testing. p < 0.05 was considered statistically significant. Vehicle was compared to erastin or H 2 O 2 and the individual fucoidan concentrations with erastin or H 2 O 2 were compared to erastin or H 2 O 2 alone. Each biological replicate was displayed around the regression (concentration-response curves) or median (ferrostatin-1). All statistical analyses were performed with IBM SPSS version 23.

Conclusions
The aim of this study was to investigate different high-molecular weight fucoidan extracts from brown seaweed of different origins for application against iron-dependent oxidative stress, a main hallmark of retinal and brain diseases, including retinal and brain hemorrhage. Only extracts from F. serratus, F. distichus subsp. evanescens and Laminaria hyperborea were partially protective against erastin-induced cell death in ARPE-19 and OMM-1 cells, while none of the extracts showed beneficial effects in neuronal cells. Preliminary data suggest that the protective effects may be linked to increasing GPX4 protein levels, a key antioxidant enzyme involved in iron-dependent oxidative stress, i.e., ferroptosis. Our comprehensive analysis suggests that the antioxidant abilities of fucoidans may be cell type-specific in addition to depending on the algal species and extraction method. Future studies are needed to further characterize the health-benefiting effects of fucoidans and to determine the exact mechanism underlying their antioxidative abilities.