CFH exerts anti-oxidant effects on retinal pigment epithelial cells independently from protecting against membrane attack complex

Age Related Macular Degeneration (AMD) is the first cause of social blindness in people aged over 65 leading to atrophy of retinal pigment epithelial cells (RPE), photoreceptors and choroids, eventually associated with choroidal neovascularization. Accumulation of undigested cellular debris within RPE cells or under the RPE (Drusen), oxidative stress and inflammatory mediators contribute to the RPE cell death. The major risk to develop AMD is the Y402H polymorphism of complement factor H (CFH). CFH interacting with oxidized phospholipids on the RPE membrane modulates the functions of these cells, but the exact role of CFH in RPE cell death and survival remain poorly understood. The aim of this study was to analyze the potential protective mechanism of CFH on RPE cells submitted to oxidative stress. Upon exposure to oxidized lipids 4-HNE (4-hydroxy-2-nonenal) derived from photoreceptors, both the human RPE cell line ARPE-19 and RPE cells derived from human induced pluripotent stem cells were protected from death only in the presence of the full length human recombinant CFH in the culture medium. This protective effect was independent from the membrane attack complex (MAC) formation. CFH maintained RPE cells tight junctions’ structure and regulated the caspase dependent apoptosis process. These results demonstrated the CFH anti-oxidative stress functions independently of its capacity to inhibit MAC formation.


Results
Only full length CFH protects RPE from oxidative stress-induced cell death. The effects of rec-CFH (300 nM), added at the time of 4-HNE (30 µM) exposure, was evaluated after 6 and 24 hours. The 4-HNE dose of 30 µM was chosen as it induced more than 50% of an ARPE-19 cells death (Supplemental Fig. 1a). A hundred times dose was chosen for recCFH (fragments or full length). Concentration of 4-HNE was chosen to reflect in vivo exposure as it was shown to accumulate in membranes at concentrations ranging from 10 µM to 5 mM in response to oxidative stimuli 33 . We first showed that recCFH or recCFH fragments had no effect on ARPE-19 cells viability in control conditions (Supplemental Fig. 1b,c). Following 6 hours of culture, viable cells were counted using the trypan blue-excluding cell assay. Exposure of ARPE-19 to 4-HNE (30 µM) induced at least 70% ARPE-19 cells death at 6 hours compared to untreated cells (Fig. 1a). Addition of recCFH (300 nM) in the culture medium protected ARPE-19 cells from death by 56% (P < 0.01) as compared to cells treated only with 4-HNE (Fig. 1a). This protection was abolished after 24 hours of culture and was associated with a decrease in the amount of recCFH in the culture medium (Fig. 1b,c). To identify the CCPs domains of recCFH, carrying the antioxidant activity, we tested several recCFH fragments. Because CCPs1-4 domains are essential for the anti-C3 convertase activity of CFH and both CCPs6-7 and CCPs19-20 are important for CFH membrane binding, we decided to test recCFH1-18 (without binding site CCPs19-20), recCFH 8-20 (with only the CCPs19-20 binding site), recCFH 1-7 (contains both anti-C3 convertase CCPs1-4 domains associated to binding site CCP7) and recCFH 7-20 (contains both binding sites without anti-C3 convertase domains). After 6 hours, none of the recCFH fragments significantly protected ARPE-19 cells from death induced by 4-HNE (Fig. 1a), despite their presence in the culture medium (Fig. 1c). Thus, only the full length recCFH was effective to protect RPE cells from 4-HNE-induced cell death. Contrariwise, the co-treatment of 4-HNE and recCFH Y402H , carrying the Y402H polymorphism, did not protect ARPE-19 cells from death, despite its presence in the culture medium (Fig. 1a-c). The protective effect of full length recCFH was investigated in hiPSC-derived RPE (iRPE) cells. iRPE cells grown in monolayers of polygonal pigmented cells (Udry et al., submitted), express most of RPE biomarkers (e.g. RPE65, RLBP1 and BEST) and show phagocytic ability. RecCFH also protected iRPE cells from 4-HNE toxicity (Fig. 1d).
RecCFH added in the culture medium was found, not only, on the ARPE-19 cell membrane upon 4-HNE treatment, but also, in the cytosolic compartment (Fig. 2a). Once in contact with ARPE-19 cells exposed to 4-HNE treatment, recCFH protected C3 from cleavage to its C3 fragments (C3 Frag.), resulting in a higher C3/ C3 Frag. ratio as compared to 4-HNE treated cells (Fig. 2b). Exposure to 4-HNE increased by 19% the deposit of MAC on ARPE-19 cell membranes, identified by C5b9 immunodectection (Fig. 3a,b). Treatment with full length recCFH significantly prevented MAC deposition , but treatment with its polymorphism form recCFH Y402H did not (Fig. 3a,b). Although recCFH 1-7 and 1-18 did not protect from death cells exposed to 4-HNE, they significantly decreased MAC deposit on ARPE19 cells by respectively 86% and 56% (Fig. 3c), suggesting that CFH protection resulted from mechanisms unrelated to MAC deposit (Figs 1a and 3c).
CFH protects RPE tight junctions from oxidative stress-induced disruption. We first used an MTT (3 (4,5-dimethylthiazol-,yl) 2,5 diphenyl tetrazolium bromide) colorimetric assay to investigate the effect of recCFH on the mitochondrial redox potential of ARPE-19 cultures. Co-treatment with 4-HNE and recCFH reduced the decrease previously observed with 4-HNE treatment in the mitochondrial redox potential of these cells (Fig. 4a). Upon 4-HNE treatment, a lower variation of anti-oxidative gene expression (Catalase) was seen, while a significant increase of inos oxidative gene expression in ARPE-19 cells could be observed as compared to untreated cells (Fig. 4a). RecCFH prevented the 4-HNE-induced regulation of pro-and anti-oxidative genes (Fig. 4a). One of RPE functions is to maintain the outer blood-retina barrier by expressing tight and adherence junction proteins, such as ZO-1. 4-HNE treatment altered ZO-1 immunostaining at ARPE-19 (Fig. 4b) and iRPE (Fig. 4c) cell membranes. RecCFH protected RPE cells junction integrity (Fig. 4b,c), as quantified by count the number of ZO-1-immunolabeled fragments according to their length (Fig. 4b,c). The protective effect of recCFH from oxidative stress on the ARPE-19 or iRPE cells structure was confirmed by immunofluorescent experiments using Phalloidin with or without ZO-1 co-labeling (Supplemental Fig. 2). CFH preserves mitochondria and nucleus structure of RPE submitted to oxidative stress. Using electron microscopy, exposure of ARP-19 to 4-HNE induced morphological features of apoptotic cells, including chromatic margination, nuclear condensation and cell fragmentation in apoptotic membrane-bound bodies ( Fig. 5a,b,d,e). Treatment with recCFH preserved the normal cell nucleus morphology (Fig. 5g,h). Dynamic remodeling of mitochondrial morphology is also an important indicator of healthy cells. In 4-HNE-treated ARPE-19 cells, mitochondria showed fractured tubular cristae with a round form compared to untreated ARPE-19 cells, which had many of continuous tubular cristae and an egg-shaped form (Fig. 5c,f). Co-treatment with 4-HNE/recCFH preserved the framework of mitochondria (Fig. 5c,i). All together, these data demonstrated a protective effect of CFH on oxidative stress-induced cellular organites. Because the cellular volume of ARPE-19 cells co-treated with 4-HNE and recCFH was reduced compared to 4-HNE treatment cells (Fig. 5e,h), the expression profile of genes implicated in osmotic flow was investigated. 4-HNE treatment up-regulated Kir7.1 and Kir4.1potassium channel and aquaporin 1 (Aqp1) gene expression, while recCFH reduced significantly these gene expression (Fig. 5j). www.nature.com/scientificreports www.nature.com/scientificreports/ CFH protects RPE from caspase-dependent apoptosis. Mechanisms of CFH on 4-HNE-induced cell death were studied. RecCFH reduced the number of TUNEL positive cells by 60% (vs. 4-HNE treatment, p < 0.001) (Fig. 6a). To explore whether apoptosis was caspase dependent, we measured the levels of capsase3 activation using immunohistochemistry. In untreated ARPE-19 cells, pro-caspase3 immunolabelling was observed in contrast to active caspase3 and caspase9 (Fig. 6b). Exposure to 4-HNE increased the immunolabelling signal of caspase9 and active-caspase3 (Fig. 6b). Co-treatment with recCFH reduced caspase9 and active caspase3, as shown by semi-quantified immunostaining (Fig. 6b). Pro-caspase3 is activated in the apoptotic cell death both by extrinsic (death ligand cascade involving caspase 8) and intrinsic (mitochondrial cascade implicating caspase9) pathways. In this study, expression of caspase 8 mRNA was also reduced 25 times (p < 0.05) on RT-qPCR compared to 4-HNE ARPE-19 cells treatment (Fig. 6c). These data show that CFH regulated both extrinsic and intrinsic apoptosis pathways by modulating caspases expression.

CFH protects RPE from necrosis.
Necrosis is a type of cell death morphologically characterized by swelling, rupture of intracellular organelles, and cell membrane permeabilization, measured by the release of lactate dehydrogenase (LDH). Compared to untreated ARPE-19 cells, exposure to 4-HNE showed a 430% increase of LDH levels in culture medium (Fig. 7a). Treatment with recCFH reduced the LDH increased by 56% (P < 0.01) (Fig. 7a). On western-blot, the levels of receptor-interacting protein kinase 3 (RIP3), identified as a crucial regulator of death receptor-induced necrosis, was decreased by 36% (P < 0.01) in ARPE-19 cells in presence of recCFH compared to 4-HNE only (Fig. 7b). In addition, necrotic cells induce pro-inflammatory cytokines. Quantitative mRNA expression measurements revealed a major increase of several interleukins (Il1β, Il6 and Il8) in 4-HNE-exposed cells compared to untreated cells (Fig. 7c). RecCFH treatment significantly reduced the expression of these inflammatory mediators (Fig. 7c).

Discussion
Oxidative stress is a recognized pathogenic factor in the complex and multi-factorial occurrence of AMD. CFH was previously shown to protect RPE from hydrogen peroxide 34 , but the exact mechanisms of CFH on oxidative stress-induced damages in RPE has remained imperfectly understood. In this study, we showed that only the full length CFH protected RPE cells from death, contrariwise, this effect was abolished by the polymorphism CFH Y402H , demonstrating the importance of the CFH-CCP7 domain binding site. The binding CCP7 seems www.nature.com/scientificreports www.nature.com/scientificreports/ mandatory to mediate protection against 4-HNE-induced cell death. Factor H, the main alternative complement pathway (AP) regulatory protein that circulates in the plasma, controls AP activation and MAC formation on the surface of host cells through its interaction with GAGs, anionic molecules and complement C3 fragment displayed on the cell membrane 35 . In this study, we showed that full length recCFH reduced C3 cleavage and C5b9 deposit on ARP19 cell surface upon 4-HNE treatment, demonstrating a CFH functional activity on AP activation and MAC formation. The CFH domains (CCPs1-4), that has regulatory effect on the AP activation but do not bind on cell membrane, did not protect RPE from oxidative stress death. On the other hand, reduction of C5b9 deposit did not seem to be the major mechanism of cell death inhibition as there was no correlation between MAC and cell death inhibition using the different CFH fragments. Indeed, to be active on cell death, CFH needed to have the AP-regulatory domains responsible for the MAC formation and its two binding sites (with an intact CFH-CCP7 domain). Consistent with our data, it has been shown that protection against cell death is not achieved with CFH anti-C3 convertase domains 36 , and that it requires a cooperative bivalent binding of CFH at the cell membrane surface 37 . The CCP7 domain seems to play a major role as the recCFH Y402H , carrying the Y402H polymorphism, did not show any protecting effect, which could be one of the mechanisms of susceptibility to AMD in the population carrying this polymorphic variant.
The identification of CFH ligands at the surface of apoptotic cells remains unclear. Lipids are unlikely to be ligands for CFH on apoptotic cells 36 but the calcium-dependent phospholipids-binding protein Annexin-II, involved in communication between cell membranes and the cytoplasm and in membrane trafficking and remodeling 38 could be a CFH binding partner on apoptotic cells 36,39 . Interestingly, using different constructs of CFH, Leffler and collaborators show that fragments comprising CCPs6-8 and CCPs19-20 bind on apoptotic cells surface with higher affinity than one binding site 36 , consistent with a stronger survival effect of full length CFH.
Despite complement activation by down regulation of membrane bound complement regulatory proteins expression, apoptotic cells do not undergo lysis process 40 , consistent with a MAC independent death of ARPE-19 cells upon 4-HNE treatment. This complement death protection comes from activity of fluid phase complement inhibitors C4b-binding protein and CFH which limit C9 deposition on apoptotic cells membrane 40 . On other hand, the binding of CFH on apoptotic cell surface triggers the activation of classic complement pathway C1 complex which ensures C3 cleavage to C3b responsible for the opsonization and removal of apoptotic cells debris 41 . In late apoptotic cells phase, CFH is internalized and acts as a cofactor for cathepsin L in the cleavage of C3 to opsonin iC3b which thereby facilitates phagocytosis of dying cells 42 . In this current study, we showed that www.nature.com/scientificreports www.nature.com/scientificreports/ ARPE-19 cells were protected from oxidative stress death independently of MAC formation inhibition but by specific functions of CFH. Indeed, CFH maintained RPE tight junctions and decreased caspase activation pathway just after exposure to oxidative stress. It has been demonstrated that activation of Caspase3 and C-Jun-N-terminal kinase (JNK) are observed in many 4-HNE induced apoptosis cell lines 43 . In this study, we showed that CFH down regulated, two pro-activators of Caspase3, Caspase8 and 9 expression previously inducing by exposure to 4-HNE. CFH regulated apoptotic cells death by modulating both extrinsic (caspase8) and intrinsic (caspase 9) apoptotic process. All together these data indicate that exposed to oxidative stress, CFH protected RPE tight junctions and reduced caspase activation pathway in these dying cells, but when damages were so high CFH was internalized and facilitated the removal of apoptotic cells by producing iC3b 42 . CFH binding domains were shown to have stronger attachment to necrotic cells as compared to apoptotic cells 36 , suggesting also a protective effect of CFH against necrosis. In our experiments, full length CFH also protected cells from necrosis as shown by the down regulation of RIP3 expression and LDH measurement.
It is recently shown that CFH, actively internalized by apoptotic RPE cells, forms complexes with nucleosomes which facilitates their phagocytosis by monocytes, ensuring an efficient removal of dying cells 42 . The binding between CFH with nucleosomes also modulates phagocytes cytokines towards an anti-inflammatory profile 42 . In our experiments, CFH also reduced the expression of pro-inflammatory cytokines by RPE cells, in agreement with a demonstrated reduction of Il-8 by CFH on malondialdehyde-acetaldehyde-induced RPE cell death 21 . However, CFH has no effect on Il-8 expression induced by phorbol myristate acetate oxidative stress 21 , suggesting that the CFH inflammatory regulation is dependent on the nature of oxidative stress.
In conclusion, this study showed that only full length CFH protected both human ARPE-19 cell line and iRPE from oxidative stress-induced cell death created by exposure to 4-HNE. Both necrosis and caspase-dependent apoptosis were reduced by CFH. Exposure to 4-HNE increased MAC deposit on RPE cells while full length CFH as well as CCPs1-7 and CCPs1-18 decreased MAC, but only full length CFH protected from oxidative stress-induced cell death, suggesting an effect independent from MAC formation. The Y402H polymorphism www.nature.com/scientificreports www.nature.com/scientificreports/ form of CFH, that is associated with the risk of AMD, lost the protective effect. Taken together, these results suggest that CFH per se exerts antioxidant protective effects on RPE cells and that blocking the alternative complement pathway activation, without restoring the activity of CFH might not be sufficient to exert full preventive and therapeutic effects in AMD.

Culture and treatment of ARPE-19 cells. The human retinal pigment epithelial cells ARPE-19, a no
transformed human RPE line that displays many differentiated properties typical of RPE in vivo, were established and characterized previously 44 . ARPE-19 were grown in 6 flat bottom cell culture dishes to a confluency in a standard incubator (37 °C, 5% CO 2 ) in DMEM: F12 (Invitrogen, France), supplemented with 10% calf serum, 2 mM glutamine, and 15 mM Hepes (complete culture medium). Confluent cells were cultured with medium containing 1% fetal calf serum for 2 weeks and then exposed to 30 μM 4-HNE (Merck, France) for 1, 6 or 24 h. Time zero of the kinetics corresponds to the moment of the stimulation with 4-HNE. To study the influence of CFH, cells were co-exposed to 300 nM of recCFH or one of its fragments (recCFH CCP1  www.nature.com/scientificreports www.nature.com/scientificreports/ donor, 250 to 500 embryonic body-like aggregates were plated and cultured in P60 (60 mm) cell culture dishes coated with a matrigel matrix (Corning). Following 30 days of differentiation, pigmented foci were micro dissected, collected, seeded in matrigel-coated P60 cell culture dishes and grown for an additional 30 days. Mature pigmented RPE patches were micro dissected, purified by removal of non-RPE like cellular structures, dissociated with Trypsin-EDTA and reseeded in 24-well matrigel-coated plates for further expansion and maturation until passage 3 (P3). From cells at passage 1 (P1) to cells at P3, additional 2 to 3 months of cell culture were required. iRPE cells were incubated in a serum-and antibiotic-free retinal differentiation medium containing DMEM (high glucose, GlutaMAX Supplement, HEPES, ThermoFisher, France), Ham's F-12 Nutrient Mix (ThermoFisher) (3:1 ratio) and 2% B-27 supplement minus vitamin A (ThermoFisher, France). Characterization of iRPE cells and experimentations took place at P3 on day 42. iRPE cells cultured on transwell plates were characterized and compared to human fetal RPE and postmortem human RPE controls (Udry et al., submitted). The expression of specific RPE markers was assessed by RT-PCR, RT-qPCR and immunofluorescence. iRPE cells grew in monolayers of polygonal pigmented cells, demonstrated specific RPE markers expression and generated high TER levels (300 Ω • cm 2 ). For experimentations, iRPE cells were seeded and grown at P3 in 6-well (cell viability assay, semi quantitative Western blot analysis and RT-qPCR) or 24-well (immunocytochemistry) cell culture plates for 42 days. One week prior to experimentations, 1% fetal bovine serum was added to cell culture medium. At day 42, iRPE cells were treated for 6 h with 30 μM 4-hydroxy-2-nonenal (4-HNE) (Merck, France) ± 300 nM recCFH (Laboratoire français du fractionnement et des biotechnologies LFB, France). Untreated cells served as control.
Cell viability assays. Cell viability was assessed by counting trypan blue-excluding cells after adding 0.5% trypan blue and by monitoring LDH (lactate dehydrogenase) release into the culture, with a cytotoxicity detection kit (Roche Diagnostics, Meylan, France) according to manufacturer's recommendations. A micro plate reader calibrated with 600 and 490 nm directly measured the absorbance. www.nature.com/scientificreports www.nature.com/scientificreports/ Measurement of mitochondrial redox potential. Mitochondrial redox potential was assessed spectrophotometrically with an MTT assay (Sigma-Aldrich, France). Cells were seeded at 20 000 cells per well in a 12-well plate. At 30 days after the final culture medium change, cells were stimulated with 4-HNE or both 4-HNE and recCFH for 6 hours. After cell stimulation, cells were washed once with PBS pre-warmed to 37 °C and incubated at 37 °C in 5% CO 2 in a solution of MTT (1 mg mL −1 in PBS) pre-warmed to 37 °C. After 1 h, isopropanol (final concentration 50%) was directly added to the MTT solution, and the 12-well plates were slowly rotated for 10 min at room temperature. The absorbance was directly measured at 570 nm in a microplate reader.  www.nature.com/scientificreports www.nature.com/scientificreports/ to the primary antibody for respectively 60 min at room temperature. The slides were then washed, stained for 5-10 min with DAPI, and washed again in PBS. Slides in the plastic labtek were then mounted with Dako solution (Dako, France) and then examined with a Zeiss confocal Imaging system (LSM710, Zeiss). As a control, the primary antibody was omitted: no staining was observed in any control.
Length of ZO-1 labelled membrane fragments was measured with an ImageJ customized macro, that automatized the following steps. The images stacks of ZO-1 labelling were projected using maximum intensity projection. Linear structures were enhanced by computing the smallest eigen values of the hessian tensor thanks to the ImageJ plugin "FeatureJ" (Erik Meijering, Erasmus University Medical Center, Rotterdam, Netherlands. Plugin available from https://imagescience.org/meijering/software/featurej. Images obtained were then binarized and skeletonized. Junction points of the image skeleton, where three or more segments are branched, were then detected by a binary hit or miss operation using the ImageJ plug-in "Morphology" (G. Landini 2008). The junction points were subtracted from the image skeleton and the remaining skeleton branches were counted and measured using the Analyze Particles function of ImageJ. Those branches were considered as representative of the continuous segments of ZO-1 labeling along the cell membranes. From the measurements obtained, histograms of the segments length where build, and the amount of large segments (x µm < length < y µm) was compared between the different cell culture conditions. Western blot. Total protein was extracted from ARPE-19 cells. The cells were homogenized and solubilized in ice-cold PBS containing protease inhibitors. Briefly, electrophoresis was performed by SDS-PAGE 4-12% Tris-gel and the separated proteins were transferred to nitrocellulose membrane (Immobilon; Millipore, France). The blots were blocked with 5% non fat dry milk. Mouse monoclonal anti-β-actin (1:3000, mouse, Abcam, France) or RIP3(1 :2000, rabbit, ThermoFisher, France) were used as primary antibodies overnight at 4 °C, and then the blots were washed with TBS 1X/milk 1% and incubated separately with the corresponding second antibody coupled to horseradish peroxidase (1:3000, Abcam, France). Blots were developed using the enhanced chemiluminescence Western blotting detection system "ECL-Plus" (Amersham Pharmacia Biotech, Arlington Heights, IL, France) according to manufacturer's recommendations. Quantification of RIP3 and β-actin were accomplished by analyzing the intensity of the bands using ImageJ program (National Institute of Health, Bethesda, MD). Quantitative real-time polymerase chain reaction (RT-qPCR). Cellular material from at least three 6-well plate dishes was pooled for each condition. Total RNA from ARPE-19 was isolated with TRIZOL reagent (Invitrogen, France) according to the manufacturer's instructions, and Superscript II Reverse Transcriptase (Invitrogen, France) was used to reverse transcribe 1 μg of mRNA. Amplification reaction assays contained 1 × SYBR Green PCR Mastermix (Applied Biosystems, France). All real-time PCR oligonucleotide primers were previously experimentally validated by QPCR and BLAST. Primers were designed such that amplicon sizes ranged from 50 to 250 bps (Table 1). A hot start at 95 °C for 5 min was followed by 40 cycles at 95 °C for 15 s and 60 °C for 1 min with the 7300 SDS thermal cycler (Applied Biosystems, France). Controls with no reverse transcriptase were run for each assay to confirm the lack of genomic DNA contamination. Control RT-qPCR reactions were performed without cDNA templates. Actin was used as a suitable reference gene. The standard curve method (Prism 7700 Sequence Detection System; ABI User Bulletin number 2) was used for relative quantification of gene expression. At least three experiments were performed for each gene and sample. For all experiments, each individual sample was run in triplicate wells and the Ct of each well was recorded at the end of the reaction. The average and standard deviation of the three Cts was calculated. Gene expression levels were normalized to actin for each cellular material sample and calculated relative to no treated culture (control) with the following equation: relative expression = 2 −(sampleΔCt − controlΔCt) where ΔCt = mean Ct(target) − mean Ct(actin).

Statistical analyses.
Results are presented as the mean ± SEM. Statistical analyses were performed using GraphPAD Prism 5 software. For data related to qPCR and western blot, comparison between two groups was performed using Mann-Whitney test.