**4. Discussion**

The RPE is exposed to high-energy light, and it conducts the phagocytosis of oxidized photoreceptor outer segments. Both of these processes are accompanied by a rapid release of reactive oxygen species [6,48,49]. Reactive oxygen species, including H2O2, are on the one hand major cellular stressors [6,50] and on the other hand cellular survival factors [3,51]. Antioxidants are decreased in light-exposed retinae, allowing the intraocular accumulation of H2O2 [52]. We used H2O2 treatment to mimic physiological oxidative stress in serum-free cultivated ARPE-19 cells to investigate the endogenous complement response in ARPE-19 cells independent of external complement sources [53,54].

Oxidative stress increased the concentration of the complement regulators CFH and properdin and the central complement protein C3 in ARPE-19 cells in a time-dependent manner, without access to any extracellular complement source. Previous studies have mostly reported a reduced expression of *cfh* mRNA in RPE cells exposed to oxidative stress [28–31], but these studies did not include further CFH protein analysis. Our reported CFH protein accumulation after H2O2 treatment in polarized, monolayer ARPE-19 cells (using immunohistochemistry) is in contrast to reduced CFH protein detection results in Western blots of non-in vivo-like cultivated ARPE-19 cells following H2O2 treatment [30].

However, it is known that intracellular CFH can enhance the cleavage of endogenously expressed C3 through a cathepsin L (CTSL)-mediated mechanism [55]. The concentrations of lysosomal protease CTSL and the central complement protein C3 were both enhanced under oxidative stress conditions in ARPE-19 cells. Previous studies of RPE cell-derived complement components only focused on *c3* mRNA expression, which was not changed under low H2O2 concentrations [56]. We went a step further and showed that C3 was accumulated in the ARPE-19 cells following oxidative stress. This ARPE-19 cell-dependent local accumulation of C3 was also shown for ARPE-19 cells treated with cigarette smoke [27].

If C3 is activated in the blood, CFH serves as a negative regulator and properdin as a positive regulator. We showed for the first time that oxidative stress increased *properdin* mRNA expression in ARPE-19 cells. This resulted in a higher properdin protein concentration in these cells, which may promote cellular C3 cleavage. In summary, our data described a local production of complement proteins in ARPE-19 cells and an enhanced cellular storage of complement proteins in the cells after H2O2 treatment. This cellular accumulation suggests an autocrine, cellular function of complement proteins in ARPE-19 cells following oxidative stress that is independent of external complement protein sources.

Our studies revealed a colocalization of accumulated, endogenous C3 with complement receptor 3 (CR3, subunit CD11b) in ARPE-19 cells exposed to oxidative stress and an increase in CD11b after 4 h. CR3 expression has been associated with inflammasome activation as a reaction to complement components and/or oxidative stress in white blood cells and RPE cells [57,58]. In agreemen<sup>t</sup> with this association, the addition of H2O2 to ARPE-19 cells increased the time-dependent expression of *nlrp3* and *il-1*β mRNA and subsequently enhanced the secretion of proinflammatory cytokines.

Inflammasome activation can be triggered by reactive oxygen species and has been associated with lipid peroxidation end products and phototoxicity in RPE cells [59,60]. The involvement of the complement components in this oxidative stress response of RPE cells has only been described in relation to extracellularly added anaphylatoxins so far [24], but an endogenous complement of RPE cells has not been suggested as a potential priming factor for the inflammasome. We detected activated C3 cleavage products in ARPE-19 cells, and previous studies have shown that activated C3a can be intracellularly generated in RPE cells independent of the systemic canonical complement system [32–35]. Further, C3 receptors were expressed (CR3, C3aR) and regulated (CR3) under oxidative stress in ARPE-19 cells, indicating a role for endogenous complement components in stressed ARPE-19 cells.

Cellular C3 is cleaved by lysosomal CTSL [18,55], and NLRP3 inflammasome activation depends on this CTSL activity [60]. It has been reported that CTSL inhibition reduces inflammasome activity in ARPE-19 cells exposed to oxidative stress [61]. These findings show the interaction between cell-specific complement component cleavage and inflammasome activity. It is already known that endogenous

C3-driven complement activation is required for IL-1β and IL-6 secretion, as well as for inflammasome activation in immune cells [62]. Our data sugges<sup>t</sup> that proinflammatory cytokine secretion may also be an autocrine mechanism in ARPE-19 cells associated with complement components and receptors.

In addition to C3, C5 has been identified as a key player in cell homeostasis [24]. The *c5aR1* receptor is expressed in RPE cells [63,64] and was increased after oxidative stress treatment. *C5* mRNA expression was not changed, and the biologically highly active C5a fragment, a ligand for C5aR1 with a very low biological half-life (approximately 1 min [65,66]), was not detected in our study. The rapid C5a–C5aR interaction might have interfered with our detection schedule. However, C5aR1 stimulation is associated with IL-8 and VEGF-α secretion in ARPE-19 cells [63,64]. The increased secretion of these proangiogenic factors was also observed following the H2O2 stimuli. The signaling pathway is not exactly known so far, but exclusive C5aR1 activation by non-ARPE-19 cell components can be excluded. In regulatory T-cells, the transcription factor FOXP3 promotes the expression of IL-8 [22], and in bladder cancer cells, a knock-down of *foxp3* has resulted in the reduced expression of *vegf-*α [23]. *Foxp3* mRNA was expressed in ARPE-19 cells and increased under oxidative stress conditions. Previous studies have shown that extracellular C5a can activate FOXP3 in ARPE-19 cells, which is associated with increased IL-8 secretion [25]. We showed that this could also be due to the endogenous activation of C5aR1 following oxidative stress in RPE cells.

These changes in expression and cellular complement protein accumulation following oxidative stress were time-dependent (Supplementary Materials, Figure S6). The first changes in complement receptor (CD11b, C5aR1) and component (CFH, C3) levels in the ARPE-19 cells occurred after 4 h and were accompanied by changes in *nlrp3* and *foxp3* mRNA expression. Downstream alterations in properdin expression, intracellular proteases, and an increase in the epithelial–mesenchymal transition marker as well as a loss of tight junctions were described. This indicates that complement receptor signaling may be involved in the early response of ARPE-19 cells to H2O2 treatment.

Oxidative stress-related cell damage in ARPE-19 cells and retinal degeneration in mouse models of RPE degeneration, as well as hereditary retinal degeneration, were successfully ameliorated using olaparib in previous studies [45–47]. Olaparib is a clinically developed poly(ADP-ribose) polymerase inhibitor that was developed for cancer treatment by blocking the DNA repair mechanism. ARPE-19 cells were resistant to H2O2-induced mitochondrial dysfunction and to energy failure when olaparib was added [45]. We asked the following question: can olaparib also normalize complement-associated proinflammatory expression profiles in H2O2-treated cells? Surprisingly, olaparib accelerated the effect of oxidative stress in ARPE-19 cells and enhanced the expression of complement receptors, complement components, and *nlrp3* mRNA. This shows that the endogenous complement-related, proinflammatory response of ARPE-19 cells could be correlated with defective DNA repair mechanisms.

Finally, it needs to be pointed out that this analysis of endogenous complement components and oxidative stress reaction was primarily set up to generate the first data describing an RPE cell-dependent complement reaction. We took advantage of the most commonly used in vitro RPE model (the ARPE-19 cell line), which expresses well-characterized RPE-specific markers [67,68] and provides an unlimited genetic and environmentally identical availability without any risk of contamination with other or undifferentiated cell types. However, it needs to be considered that this model system bears a higher risk of undergoing an epithelial–mesenchymal transition because of long-term cultivation, showing limitations in measuring transepithelial resistance and less expressed RPE-specific markers compared to primary or stem-cell-derived RPE cells [68]. To follow up with this intriguing line of thinking, future studies are needed to verify these cell-associated complement functions in primary or stem-cell-derived RPE cells with different genetic and environmental backgrounds.
