Introduction

Age-related macular degeneration (AMD) is the most common cause of visual loss in the elderly in developed countries1,2,3. Oxidative stress is commonly attributed to age-associated pathology in many diseases with damage from oxidative stress accumulating during ageing4. In addition, the antioxidant capacity diminishes and the efficiency of repair systems is compromised with age5,6. Oxidative stress also has an important role in the initiation and progression of AMD7,8. Moreover, treatments with antioxidants, such as vitamin C, vitamin E, β-carotene, and zinc, have been found to ameliorate AMD7. Various stimuli can induce different forms of oxidative stress in the eye9. Proportionally compared to the rest of the body, the retina consumes the most oxygen and therefore has several adaptations to deal with oxidative stress10. Dysregulation of pathways regulating oxidative stress are therefore likely to play a role in AMD pathology.

Retinal pigment epithelium (RPE) cell death is characteristic of late stage AMD11. The RPE has many important functions in the retina, including light absorption, transepithelial transport, buffers ions in the subretinal space, participates in the visual cycle, phagocytoses photoreceptor outer segments (POS), and secretes important factors to neighbouring cells12. In the retina, the RPE cells are continuously exposed to oxidative stress as RPE metabolises and recycles POS12,13. RPE cells also have large oxygen fluxes near the plasma membrane12,14 and are exposed to light10,12. Moreover, aged RPE have been found to have a reduced metabolic capacity, leading to increased susceptibility to oxidative stress15. Therefore, RPE need to be well adapted to handle oxidative stress.

An important function of RPE is to phagocytose POS16,17. POS are enriched with PUFAs, to the point that POS possess the highest relative concentration of PUFAs compared to all other body tissues18. The high amount of double bonds in PUFAs make them susceptible to lipid peroxidation by reactive oxygen species (ROS). Docosahexaenoic acid (DHA; 22:6) is an omega-3 PUFA that is relatively rare in human tissue. However, DHA accounts for 60% of the PUFAs in the retina, with its highest concentration in POS18. DHA is an essential structural component of the cell membrane in POS19 and DHA is known to be able to enhance RPE survival20. However, DHA is also the most oxidizable fatty acid in humans18. RPE are constantly exposed to high concentrations of DHA through phagocytosis of POS and through transport of DHA from the choriocapillaris to the photoreceptors for photoreceptor membrane biogenesis21. Oxidative stress can therefore potentially cause lipid peroxidation of DHA in RPE, leading to RPE damage and possibly cell death.

Oxidative stress caused by hydrogen peroxide (H2O2) has been shown to lead to necroptosis, and not apoptosis, in RPE22. Necroptosis is a regulated form of necrosis23. Moreover, oxidative stress induced by tert-butyl hydroperoxide (tBHP) or by glutathione depletion were recently shown to cause ferroptosis in human cells, including RPE24,25,26. Ferroptosis is an iron- and PUFA-dependent form of regulated cell death, which is distinct from apoptosis and necrosis27. DHA is also known to be able to induce ferroptosis28. However, it is not yet known if DHA can contribute to RPE cell death through ferroptosis during oxidative stress.

Dysregulation of antioxidant defensive mechanisms in the retina of AMD patients is not well understood. Equally, it is not well understood how dysregulation of lipid metabolism under such conditions contributes to RPE cell death and ultimately leads to vision loss.

Here, we show that increased intracellular POS or DHA concentration can increase RPE susceptibility to oxidative stress injury, leading to both ferroptosis and necroptosis of RPE cells.

Materials and methods

Cell culture

ARPE-19 (ATCC, Manassas, VA) and hTERT RPE-1 (ATCC) cell lines were cultured and maintained in DMEM/F12 (Gibco, Loughborough, UK) supplemented with 10% foetal calf serum (FCS; Gibco) and l-glutamine (Gibco) at 37 °C and 5% CO2. For experiments, cells were split and seeded in DMEM/F12 supplemented with 1% FCS and l-glutamine unless otherwise stated.

Fatty acid-BSA conjugation

DHA (Sigma-Aldrich, Dorset, UK) and palmitic acid (PA; Sigma-Aldrich) were conjugated to fatty acid-free bovine serum albumin (BSA; Sigma-Aldrich) as previously described29, but with some modifications. Briefly, DHA and PA were dissolved in 50% ethanol to 150 mM at 70 °C and 90 °C respectively. The dissolved fatty acids were then immediately diluted, using prewarmed tips, in prewarmed 10% fatty acid-free BSA solution to 7.5 mM whilst stirring. This was incubated at 37 °C for 1 h to achieve BSA-fatty acid conjugation. The conjugated BSA-fatty acids were 0.22 µm filter-sterilised, aliquoted, and stored at − 20 °C. Conjugated BSA-fatty acids were used in cell culture medium at a final concentration of 500 µM fatty acid and 0.67% BSA unless otherwise stated.

POS-FITC conjugation

Bovine rod POS (InVision BioResources, Seattle, USA) were prepared and conjugated with fluorescein-5-Isothiocyanate (FITC; Invitrogen) as previously described30, with some modifications. POS were washed three times with and resuspended in wash solution, consisting of 10% sucrose, 20 mM sodium phosphate buffer pH 7.2, and 5 mM taurine. To obtain a 2 mg/ml working solution of FITC, 10 mg of FITC was dissolved in 0.1 M sodium-carbonate buffer pH 9.5. To conjugate POS with FITC, 1.5 ml of FITC working solution was added to 5 ml of POS suspension and incubated at 21 °C for 1 h whilst rotating in the dark. The POS-FITC conjugate was then washed twice with and resuspended in DMEM with 2.5% sucrose.

Inhibitor treatment, oxidative stress treatment, and immunocytochemistry

RPE cell lines were seeded at 7.5 × 103 cells/well in a 96-well plate. The following day the cells were pre-treated for 24 h with 40 µM Z-VAD-FMK (Merck Millipore, Nottingham, UK), 40 µM Necrostatin-1 (Nec-1; Cayman Chemical, Ann Arbor, MI), 1 µM GSK2791840B (GSK’840B; GlaxoSmithKline, Brentford, UK), or 5 µM Ferrostatin-1 (Fer-1; Sigma-Aldrich). Pre-treatment with 500 µM deferoxamine (DFO; Sigma-Aldrich) was 1 h before injury.

Medium with inhibitors was transferred to a new plate and kept at 37 °C and 5% CO2. The cells were subsequently stained with 3 µM (4,4-difluoro-5-(4-phenyl-1,3-butadienyl)-4-bora-3a,4a-diaza-s-indacene-3-undecanoic acid; Invitrogen) in culture medium for 1 h at 37 °C and 5% CO2 to assess lipid peroxidation in RPE cell line membranes. C11-BODIPY581/591 incorporates into membranes and the excitation and emission fluorescence spectra of this lipid analogue both shift to shorter wavelengths when oxidised31. The cells were washed once with PBS and the medium with inhibitors was transferred back to the cells.

The cells were then treated with a final concentration of 500 µM H2O2 (Sigma-Aldrich) for 6 h or 500 µM DHA or 50 POS-FITC/cell for 18 h at 37 °C and 5% CO2.

After treatment, the cells were washed with PBS and subsequently stained with 6 µg/ml Hoechst 33342 (Sigma-Aldrich). The cells were then fixed with prewarmed 4% PFA for 10 min at 21 °C and washed with PBS.

For pMLKL and PI staining, the cells were washed with PBS after H2O2 treatment and subsequently stained with 3 µg/ml PI (Sigma-Aldrich) for 20 min at 21 °C. The cells were then fixed with prewarmed 4% PFA for 10 min and permeabilised with 0.1% Triton™ X-100 (Sigma-Aldrich) for 15 min at 21 °C. The cells were then incubated with 1:100 phospho S358 MLKL antibody (Abcam, Cambridge, UK; ab187091) in 3% goat serum for 1 h at 4 °C. After two washes with PBS, the cells were incubated with 1:250 Alexa Fluor 647 antibody (Molecular Probes; A21244) for 30 min at 21 °C covered from light. After another two washes with PBS, the nuclei were counterstained with 300 nM DAPI (Biotium, Fremont, CA) for 10 min at 21 °C covered from light, followed by another three washes with PBS.

Stained cells were imaged with a Leica DMi8 inverted microscope (Leica Microsystems, Milton Keynes, UK).

Flow cytometry for lipid peroxidation

RPE cell lines were seeded and treated with inhibitors in 96-well plates as described above. Medium with inhibitors was transferred to a new plate and kept at 37 °C and 5% CO2. The cells were then incubated for 1 h at 37 °C and 5% CO2 with a final concentration of 3 µM C11-BODIPY581/591 in DMEM/F12 without phenol red (Gibco) and no FCS. The cells were washed once with PBS and the medium with inhibitors was transferred back to the cells. The cells were then treated with a final concentration of 500 µM H2O2 for 3 h at 37 °C and 5% CO2. After H2O2 treatment, the cells were washed once with PBS, trypsinised, and transferred to Eppendorf’s containing medium with 10% FCS to inactivate the trypsin. The cells were then washed once with flow buffer, resuspended in 250 µl of flow buffer containing a final concentration of 2.4 µM DRAQ7 (Biostatus, Shepshed, UK). The samples were kept on ice until analysis with a BD Accuri C6 flowcytometer (BD Biosciences, Franklin Lakes, NJ).

Flow cytometry for Annexin V, PI, MLKL, and pMLKL

RPE cell lines were seeded and when stated treated with inhibitors in 96-well plates as described above. The cells were then treated with a final concentration of 500 µM H2O2 for 3 h at 37 °C and 5% CO2. After H2O2 treatment, the cells were washed once with PBS, trypsinised, and transferred to Eppendorf’s containing medium with 10% FCS to inactivate the trypsin.

For Annexin V and PI staining, the cells were stained using the Annexin V-FITC Apoptosis Detection Kit (Affymetrix, Santa Clara, CA) according to the manufacturer’s instructions.

For MLKL and pMLKL staining, the cells were fixed in prewarmed 4% paraformaldehyde for 10 min at room temperature. After washing with flow buffer, consisting of PBS with 1% BSA and 0.05% NaN3, the cells were permeabilised with 0.2% Triton X-100 for 5 min at room temperature. After washing with flow buffer the cells were incubated with 1:100 MLKL antibody (Abcam; ab184718) or 1:100 phospho S358 MLKL antibody (Abcam; ab187091)32 in 10% normal goat serum in flow buffer for 1 h on ice. After washing the cells were incubated with 1:250 Alexa Fluor 488 antibody (Cell Signaling Technology, Danvers, MA; 4412S) or 1:250 Alexa Fluor 647 antibody (Molecular Probes, Eugene, OR; A21244) for 30 min on ice. The cells were then washed once with flow buffer, resuspended in 250 µl of flow buffer. The samples were kept on ice until analysis with a BD Accuri C6 flowcytometer.

Cell viability assay

RPE cell lines were seeded at 104 cells per well in a 96-well plate. After 24 h, cells were then treated with 500 µM H2O2 or 500 µM BSA-PA or 500 µM BSA-DHA or H2O2 with BSA-PA or H2O2 with BSA-DHA for 6 or 18 h. Respective wells were pre-treated with 40 µM Nec-1 or 5 µM Fer-1 for 1 h before the treatment. Cells treated with 0.67% BSA and 0.17% ethanol were considered as the control (vector control). After treatment cellular viability was analysed using MTT. Briefly, cells were incubated with 0.5 mg/ml MTT (Sigma-Aldrich) for 4 h, and the formed formazan crystals were solubilised using 10% SDS-(0.01N)HCl solution. Absorbance at 570 nm and 690 nm was measured, and the 570/690 ratio was used for the comparison. The absorbance ratio of each group was compared to the control and expressed as a percentage.

qPCR

RPE cell lines were seeded at 3 × 105 cells/well in a 6-well plate. The following day the cells were treated with 300 µM H2O2 and 300 µM BSA-DHA for 24 h. RNA was extracted using TRIzol® Reagent (Ambion, Waltham, MA) according to manufacturer’s instructions. cDNA was made using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA) according to manufacturer’s instructions and primers were designed by Primerdesign Ltd. (Southampton, UK), see Table S1 for sequences. Power SYBR® Green PCR Master Mix (Applied Biosystems) was used in combination with a StepOnePlus™ Real-Time PCR System (Applied Biosystems). The relative gene expression data was analysed using the 2−∆∆Ct method as described by Livak and Schmittgen33,34.

Western blotting

RPE cell lines were seeded at 2.5 × 105 cells/well in a 6-well plate. The following day the cells were treated with 500 µM H2O2, 500 µM BSA-PA, and 500 µM BSA-DHA for 6 h. The cells were then lysed in RIPA buffer supplemented with Complete Mini protease inhibitor cocktail (Sigma-Aldrich) and PhosSTOP phosphatase inhibitor (Roche, Basel, Switzerland) and passed through a 21-gauge needle (Sterican, Hessen, Germany) to shear the DNA. 20 µg of cell lysate in Laemmli sample buffer (Bio-Rad, Hercules, CA) supplemented with β-mercaptoethanol was heated to 98 °C for 5 min and subsequently loaded and ran on a 12% SDS-PAGE gel. The gel was then blotted onto a nitrocellulose membrane (Bio-Rad), blocked with TBS + 5% Marvel for 30 min at room temperature, and incubated with 1:500 ACSL4 antibody (Santa Cruz Biotechnology, Santa Cruz, CA; sc-271800) overnight at 4 °C. The following day the membrane was incubated with 1:4,000 HRP-conjugated secondary antibody (Merck Millipore; 12–349) for 30 min at room temperature. After incubation with Amersham ECL reagent (GE Healthcare, Little Chalfont, UK), an Amersham Hyperfilm (GE Healthcare) was exposed to the membrane. The same membrane was reprobed with 1:5,000 β-actin antibody (Abcam; ab6276) for 1 h at room temperature followed by 1:4,000 HRP-conjugated secondary antibody (Merck Millipore; 12-349) for 30 min at room temperature to check for loading control. Exposed films were developed in a Mi-5 X-ray developer machine (Jet Xray, London, UK).

Statistics

Experiments were repeated at least three times. Statistical analysis was performed using Microsoft Excel version 1905 and GraphPad Prism 6 software. P-values of less than 0.05 were considered to be statistically significant.

Results

Increased lipid ROS in RPE after H2O2, DHA, or POS exposure and is inhibited by Fer-1 and DFO

RPE were exposed to H2O2 in order to investigate the role of apoptosis and necrosis in RPE cell death under oxidative stress conditions. H2O2-exposure did not induce apoptosis in RPE, although it did result in more necrotic cells (Fig. S1). This is in agreement with published findings22. However, other forms of cell death could also play a role in addition to necrosis.

Ferroptosis is an iron- and PUFA-dependent form of cell death and is characterised by peroxidation of PUFAs27. Phagocytosis of POS, containing high concentrations of the PUFA DHA, could become problematic for RPE under oxidative stress conditions. Involvement of ferroptosis in RPE cell death was therefore investigated by stressing RPE cells with H2O2, DHA, or POS. Treatment with iron chelator DFO or the ferroptosis inhibitor Fer-1 reduced the amount of lipid ROS caused by H2O2, DHA, or POS injury (Fig. 1a and b). The necroptosis inhibitor Nec-1 also reduced the amount of lipid ROS but to a more limited extent (Fig. 1b). Necroptosis is a regulated form of necrosis22. However, Nec-1 is also known to protect against ferroptosis35. In contrast, the apoptosis inhibitor Z-VAD did not reduce the amount of lipid ROS (Fig. 1a). These results show that ferroptosis, and possibly necroptosis, could occur when RPE are stressed with H2O2, DHA, or POS.

Since DHA is known to enhance RPE survival when oxidatively stressed, the effect of H2O2 in combination with DHA or POS on RPE were studied by analysing the amount of lipid ROS formed. H2O2 in combination with DHA showed increased lipid ROS and a similar effect was observed in combination with POS (Fig. 1c). These results indicate that DHA and POS can increase cellular damage during oxidative stress in RPE.

Figure 1
figure 1

H2O2-, DHA-, and POS-induced lipid ROS in RPE are reduced by Fer-1 and DFO36. ARPE-19 cells were treated with 40 µM Z-VAD, 40 µM Nec-1, or 5 µM Fer-1 24 h or 500 µM DFO 1 h before injury. Subsequent 3-h treatment with 500 µM H2O2, 500 µM DHA, or 50 POS/cell shows increase of lipid ROS (BODIPY) (a and b). Pre-treatment with Fer-1 and DFO reduces the amount of lipid ROS after injury. Injury with DHA or POS in addition to H2O2 increases the amount of lipid ROS compared to H2O2 alone (c). Scale bar equals 200 µm.

Full size image

MLKL is phosphorylated after H2O2 or POS exposure in RPE

Next, the role of necroptosis was further investigated in RPE stressed with H2O2, DHA, or POS. MLKL is the effector protein of necroptosis when activated through phosphorylation32. RPE were found to express MLKL (Fig. 2a). Moreover, treatment with H2O2 increased the amount of pMLKL (Fig. 2b) and therefore caused necroptosis. DHA or POS treatment did not increase the amount of pMLKL in RPE and therefore likely do not cause necroptosis without H2O2. These results indicate that only H2O2 can induce necroptosis in RPE and that the beneficial effect of Nec-1 on DHA- and POS-induced stress in Fig. 1b seems to be related to the protective effect of Nec-1 on ferroptosis.

Figure 2
figure 2

Increase of pMLKL in RPE after H2O2 or POS exposure36. Treatment of ARPE-19 cells with 500 µM H2O2, but not 500 µM DHA or 50 POS/cell, for 3 h increased phosphorylation of MLKL.

Full size image

Decreased RPE viability caused by fatty acids during oxidative stress is ameliorated with necroptosis and ferroptosis inhibitors

To further investigate the role of fatty acids in inducing cellular damage under oxidative stress conditions, RPE were treated with H2O2 and either PA or DHA. A minimum concentration of H2O2 was used for inducing oxidative stress (Fig. S2). A 6-h exposure of DHA or PA alone did not reduce cellular viability. However, when combined with H2O2, RPE cell viability was significantly reduced, with DHA having a more pronounced effect (Fig. 3a). Treatment with either Nec-1 or Fer-1 improved RPE cell viability under these conditions, underlining the role of necroptosis and ferroptosis in RPE cell death during oxidative stress.

When exposed for 18 h, DHA alone significantly reduced RPE cell viability (Fig. 3b). Moreover, a stronger reduction in cell viability was observed when combined with H2O2. Treatment with Nec-1 and Fer-1 again improved the viability of RPE cells exposed to DHA and H2O2 for the 18-h exposure. In contrast, RPE cells exposed to PA did not show similar improvements in cell viability with these inhibitors. Notably, 18-h exposure to PA alone showed a marked increase in cell viability, which is likely due to high concentrations of PA can change RPE metabolism (see discussion). Nonetheless, 18-h exposure to PA with H2O2 lowered the cell viability to the same level as H2O2 alone and the inhibitors had no positive effect. Together, these results indicate an important role for lipids, and especially PUFAs, during oxidative stress in RPE.

Figure 3
figure 3

Improved RPE viability with necroptosis and ferroptosis inhibitors after H2O2 with lipid exposure. The viability of ARPE-19 cells was tested following a 6-h (a) or 18-h (b) exposure to 500 µM H2O2, 500 µM PA, 500 µM DHA, the combinations H2O2 and PA or DHA, or the BSA vector control. Pre-treatment with Nec-1 or Fer-1 helped to improve the viability. Error bars represent SD of three independent experiments. *P ≤ 0.05, determined by Tukey’s test following a one-way ANOVA. BSA = bovine serum albumin; PA = palmitic acid; DHA = docosahexaenoic acid.

Full size image

Transcriptional changes after H2O2 or DHA exposure in RPE

In order to investigate the molecular pathways involved, the transcriptional changes between RPE cells injured with H2O2 or DHA were studied. First, differential expression of common cellular stress markers was observed. SQSTM1, also known as p62, is a multifunctional protein involved in cellular processes including autophagy and stress. SQSTM1 expression in RPE was significantly downregulated upon H2O2 exposure (Fig. 4a). In contrast, DHA caused a significant upregulation of SQSTM1 expression. HSPA1B, also known as Hsp70, is involved in a wide range of cellular processes, including protection of the proteome against stress. HSPA1B expression was significantly upregulated by both H2O2 and DHA exposure (Fig. 4a). CDKN2A, also known as p16, is involved in regulating the cell cycle and is regarded as a marker of cellular senescence. CDKN2A expression was significantly downregulated by DHA exposure, but not H2O2 exposure (Fig. 4a). These results highlight the markedly different response of RPE towards different forms of stress.

Cells can respond to stress in various ways, from activating pathways promoting survival, to triggering specific cell death pathways. To investigate if apoptotic signalling was induced, the expression of certain apoptosis markers was measured following H2O2 or DHA exposure. BCL2, an important apoptosis regulator preventing apoptotic cell death, was significantly downregulated in RPE exposed to H2O2 (Fig. 4b). However, the expression of proapoptotic marker BAX did not significantly change and moreover the expression of proapoptotic marker BAK1 was significantly downregulated (Fig. 4b). BCL2, BAX and BAK1 expression did not significantly change in RPE exposed to DHA (Fig. 4b). These results again suggest that apoptosis does not play a role in H2O2-induced cell death and that high DHA-exposure also does not lead to apoptosis in RPE.

Expression differences of ferroptosis-related genes was also assessed in H2O2- or DHA-mediated cell death in RPE. Interestingly, the expression of acyl-CoA synthetase long-chain family member 4 (ACSL4) and lysophosphatidylcholine acyltransferase 3 (LPCAT3) were significantly downregulated in RPE treated with H2O2 (Fig. 4c). In contrast, in DHA treated RPE only LPCAT3 expression was significantly downregulated (Fig. 4c). ACSL4 and LPCAT3 are important regulators of ferroptosis. Other important genes involved in ferroptosis such as GPX4 and GCLM did not significantly change in expression in RPE injured with H2O2 or DHA. Together, these results indicate that when exposed to H2O2 or DHA, RPE cells modulate the expression of genes important to induce ferroptosis, in order to prevent more ferroptotic cell death.

Figure 4
figure 4

Differential gene expression in RPE exposed to H2O2 or DHA36. ARPE-19 cells were exposed to 300 µM H2O2 or 300 µM DHA for 24 h. The fold change in mRNA expression was determined of genes related to cellular stress (a), apoptosis (b), and ferroptosis (c). Error bars represent SD of three independent experiments. *P ≤ 0.05 and **P ≤ 0.001, determined by t-tests on ΔCt values relative to corresponding controls.

Full size image

Decreased expression of ACSL4 after H2O2 and DHA exposure in RPE

To further investigate the observed downregulation of the ACSL4 gene in H2O2-treated RPE cells, the protein expression was analyzed. Again, H2O2 significantly reduced the expression of ACSL4 (Fig. 5a compare lanes 1 and 4 and Fig. 5b compare bars 1 and 2) and DHA did not significantly reduce the expression of ACSL4 (Fig. 5a compare lanes 1 and 3 and Fig. 5b compare bars 1 and 5), consistent with the findings in Fig. 4c. Exposure of the RPE cells to the most common saturated fatty acid in POS, PA, showed a similar result as DHA (Fig. 5a compare lanes 1 and 2 and Fig. 5b compare bars 1 and 3).

RPE cells were also exposed to a combination of H2O2 and either PA or DHA to determine how the combined exposure of fatty acids and H2O2 could affect ACSL4 expression. H2O2 significantly reduced the ACSL4 protein levels and the addition of DHA, but not PA, exacerbated this effect significantly more (Fig. 5a compare lanes 4, 5, and 6, and Fig. 5b compare bars 2, 4, and 6). These results indicate that DHA has a specifically negative effect on the expression of ACSL4 under oxidative stress conditions.

Figure 5
figure 5

ACSL4 is downregulated in RPE after H2O2 and DHA exposure36. ARPE-19 cells were exposed to 500 µM H2O2, 500 µM PA, 500 µM DHA or the BSA vector control for 6 h. Representative blot out of three depicting ACSL4 expression changes to H2O2 and fatty acids (a). Densitometric analysis (b). Error bars represent SD of three independent experiments. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, determined by Šidák’s multiple comparisons test following a two-way ANOVA. BSA = bovine serum albumin; PA = palmitic acid; DHA = docosahexaenoic acid. Original blots are presented in Fig. S3.

Full size image

Discussion

AMD is a multifactorial disease and oxidative stress is known to play an important role in the initiation and progression of this disease. Still much is unknown about how oxidative stress contributes to AMD pathology. Elucidating how oxidative stress affects RPE, the cell type severely affected in AMD, could contribute to our understanding of AMD initiation and progression. Here, we show that oxidative stress causes both necroptosis and ferroptosis, while addition of the PUFA DHA can worsen this effect.

Several studies have reported numerous health benefits of dietary DHA, such as supporting fetal development, preventing preterm birth, cardiovascular disease prevention, and improving cognitive function37. Although DHA is essential for normal functioning of the retina and is known for its neuroprotective functions, DHA can also play an important role in mediating cell death. Approximately 50% of the phospholipids in the vertebrate rod photoreceptors contain DHA38. RPE phagocytose POS containing this high concentration of DHA. DHA-containing phospholipids are reservoirs for docosanoids, potent bioactive mediators. The docosanoid neuroprotection D1 (NPD1) forms after oxidative stress and promotes RPE survival20. However, increasing dietary DHA uptake has shown to improve intermediate AMD, but not advanced AMD39. Moreover, with age, phagocytosis of POS can lead to lipid accumulation in RPE, leading to ROS production40. This is likely caused by carboxyethylpyrrole (CEP) protein adducts, which are formed by an oxidation product unique to DHA41. CEP adducts are found to be more abundant in AMD retinal tissue, specifically in the RPE, Bruch’s membrane, and choroid42. In addition, CEP adducts are also found in AMD drusen42 and in lipofuscin43. This indicates that DHA peroxidation is increased and that ferroptosis could be a common form of cell death in AMD. CEP adducts have been shown to stimulate angiogenesis44,45, thus possibly contributing to neovascularization in geographic atrophy AMD. Moreover, mice immunised with CEP adducts develop a geographic atrophy AMD-like phenotype, indicating a role for CEP adducts in the aetiology of AMD46. CEP adducts were also found to be elevated in the plasma of patients with AMD41. CEP adducts could therefore play a role in AMD pathology and possibly function as a biomarker for AMD. It is therefore possible that in a disease setting like AMD, with prolonged oxidative stress in combination with high concentrations of DHA, the negative consequences of the formation of CEP adducts outweigh the positive benefits of the formation of NPD1, leading to an overall negative impact on cell viability by promoting necroptotic and ferroptotic cell death.

Ferroptosis is characterised by peroxidised PUFAs and is iron-dependent. DHA has been shown to efficiently promote ferroptosis before when incorporated into cellular membranes28. The amount of lipid ROS increased in RPE injured with H2O2, DHA, and POS (Fig. 1). Lipid ROS generated in this way was reduced with Fer-1, DFO, and Nec-1, further confirming a role for ferroptosis in these types of RPE injuries. Nec-1 is known to also have a protective effect against ferroptosis, but the mechanism is not yet known35. Increased iron accumulation is found within the retina of AMD patients47,48, which is thought to promote oxidative stress through the Fenton reaction. Iron was also recently found to promote oxidative cell death caused by bisretinoids in the retina49, the precursor of lipofuscin. Therefore, iron accumulation could play an important role in the pathology of AMD. Moreover, it is known that iron chelation can protect RPE50. It is therefore possible that in a disease setting like AMD, DFO could at least reduce RPE ferroptosis. Furthermore, addition of DHA or POS to H2O2-mediated oxidative injury increased the amount of lipid ROS production in RPE (Fig. 1c). This indicates that DHA can increase the amount of ferroptotic cell death of RPE during oxidative stress.

Injury of RPE with H2O2 also caused an increased phosphorylation of S358 of MLKL (Fig. 2b). This specific phosphorylation of MLKL is necessary for necroptosis32. However, this was not found in RPE treated with only DHA or POS. These results show that necroptosis plays an important role in H2O2-induced oxidative stress in RPE, in addition to ferroptosis, but necroptosis does not seem to play a role in PUFA-induced stress.

RPE treated with fatty acids and subsequently injured with H2O2 resulted in decreased cell viability compared to injury with H2O2 alone (Fig. 3). When exposed for 6 h to H2O2, Fer-1 and Nec-1 could rescue additional stress caused by PA and DHA (Fig. 3a). However, under prolonged oxidative stress, when the cells are exposed for 18 h to H2O2, Fer-1 and Nec-1 could not rescue the cells additionally stressed with PA (Fig. 3b). Prolonged PA exposure does increase the cell viability compared to the BSA control, but not in the presence of H2O2. This indicates that at least in the case of prolonged PA exposure there are other cellular processes counteracting the cell death mechanisms. For example, PA has been reported to induce epithelial-mesenchymal transition (EMT) in ARPE-19 cells51. Additionally, RPE cells are reported to have the ability to oxidize PA, generating acetyl CoA and β-hydroxybutyrate, a process that increases the mitochondrial oxygen consumption rate52. This increased mitochondrial metabolic activity or proliferation, possibly due to EMT, could be responsible for the elevated cellular viability observed with PA at 18 h. However, 18-h treatment with the PUFA DHA showed a significantly more decreased cell viability, indicating that peroxidation of DHA could exacerbate oxidative stress in RPE. Importantly, subsequent treatment with necroptosis or ferroptosis inhibitors significantly alleviated this effect. This indicates that oxidative stress in RPE can lead to both necroptosis and ferroptosis, and that this is influenced by DHA.

A differential transcriptomic response was found in RPE injured with H2O2 or DHA (Fig. 4), with a differential expression of common cellular stress markers (Fig. 4a). SQSTM1, also known as p62, was upregulated in RPE exposed to DHA. This is consistent with previous findings where DHA induces SQSTM1 expression and thereby increases autophagy53. Important genes involved in apoptosis did not change expression necessary for apoptosis in RPE exposed to H2O2 or DHA (Fig. 4b), suggesting that RPE use an alternate mode of cell death when oxidatively stressed. The levels of the antioxidant GSH decreases with age and this has been linked to AMD6. The important ferroptosis inhibitor GPX4 is GSH-dependent. However, oxidative stress caused by H2O2 or DHA injury did not change the expression of GPX4 or GCLM in RPE (Fig. 4c). Together, these results show a differential transcriptomic response between RPE injured with H2O2 compared to DHA in terms of the expression of stress-related genes, but a similar response in terms of apoptosis-related genes and ferroptosis-related genes.

Interestingly, ACSL4 was found to be downregulated in expression in RPE injured with H2O2 (Fig. 4c). ACSL4 was also found to be downregulated in protein expression in RPE injured with H2O2 and this was exacerbated when DHA was added under oxidative stress (Fig. 5). The most common saturated fatty acid in POS, PA, did not show the same effect. Additionally, LPCAT3 was found to be downregulated in RPE injured with H2O2 or DHA (Fig. 4c). ACSL4 and LPCAT3 are involved in the execution of ferroptosis54. ACSL4 catalyses the activation of PUFAs by esterification, producing acyl-CoA. Acyl-CoA is an intermediate in various metabolic pathways, including phospholipid production for membrane biogenesis. ACSL4 is found to change lipid composition which affects ferroptosis sensitivity28,55. ACSL4 promotes the incorporation of omega-6 PUFAs in cellular membranes, which makes cells more sensitive to ferroptosis. ACSL4 has been proposed as a biomarker for ferroptosis and inhibition of ACSL4 also promotes necroptosis56. Downregulation of ACSL4 in response to H2O2 and downregulation of LPCAT3 in response to H2O2 or DHA could indicate that RPE have a defensive response towards oxidative stress and PUFAs to prevent ferroptosis and induce necroptosis. However, the concentrations that were used in this study could have been too high or too prolonged to prevent ferroptosis from occurring. Similarly, this defence mechanism could also be overwhelmed in a disease setting like AMD.

It is thought that necroptosis and ferroptosis can compensate each other when either one is inhibited56. H2O2-induced oxidative stress in RPE in this study caused both increased phosphorylation of MLKL and increased lipid ROS formation. However, the important regulators ACSL4 and LPCAT3 were found to be downregulated after H2O2 injury and the additional presence of excess DHA decreased the expression of ACSL4 even more. Inhibition of ACSL4 is known to promote necroptosis56. This shows that there is a complex regulation of the different cell death pathways in response to oxidative stress that may be interconnected. The proclivity of RPE cells to undergo cell death through necroptosis and ferroptosis after oxidative injury have both been found before22,24 but a role for DHA in this process has not been fully established in RPE. Ferroptosis in particular seems a logical form of cell death in RPE during AMD because of the accumulation of DHA through phagocytosis of POS and the increased oxidative stress. Here, we showed that the detrimental effects of H2O2-induced oxidative stress in RPE was exacerbated with DHA which led to both necroptosis and ferroptosis.