Complement C3 knockout protects photoreceptors in the sodium iodate model

Abstract

Complement factor 3 (C3) has emerged as a primary therapeutic target in age-related macular degeneration (AMD) supported by genetic, histologic, and clinical trial evidence. Yet, the site(s) of action are unclear. The purpose of this study was to test the effect of C3 knockout on photoreceptors and retinal pigment epithelial cells (RPE) in the sodium iodate (NaIO₃) model, which mirrors some features of AMD. C3^−/− and WT mice, both on a C57Bl/6J background, were injected intraperitoneally with 25 mg/kg NaIO₃. Electroretinography and optical coherence tomography were performed 7 days later to assess retinal function and structure, respectively. Then, mice were euthanized for retinal immunohistochemistry, quantitative real-time PCR and enzyme-linked immunosorbent assays. NaIO₃ increased C3 protein levels in the neural retina but not RPE. WT but not C3^−/− mice showed NaIO₃-induced iC3b deposition on photoreceptor outer segments. C3^−/− mice were partially protected against photoreceptor layer thinning. There was partial preservation of rod and cone function in the C3^−/− group. Neither RPE structure nor function was protected. These results suggest outer segment opsonization contributes to photoreceptor death in this model, and that targeting C3 can protect photoreceptor structure and function when RPE cells are stressed.

1. Introduction

The complement cascade is a major component of the innate immune system and plays a crucial role in tissue homeostasis and host immunosurveillance by coordinating both innate and adaptive immune signaling (Hajishengallis et al., 2017; Mastellos et al., 2019). Histology showing complement proteins in drusen from patients with age-related macular degeneration (AMD) implicated the complement pathway in the disease (Anderson et al., 2002). Complement factor 3 (C3) expressing cells have been found in the photoreceptor layer in postmortem eyes from AMD patients (Natoli et al., 2017). Genetic risk variants in several complement pathway genes also support the role of complement in AMD (Datta et al., 2017).

Among the many complement components, C3 has risen to become a leading therapeutic target in AMD due to its central position within the complement cascade (Mastellos et al., 2019). Regardless of the activation pathway (classical, lectin, or alternative), all complement pathways result in the generation of C3 convertases, which are enzymatic complexes responsible for binding to and cleaving C3 (Kim et al., 2021). In 2023, the US Food and Drug Administration (FDA) approved anti-C3 pegcetacoplan (Syfovre, Apellis Pharmaceuticals) (McNeil, 2023) based on phase 3 trials demonstrating reduction in geographic atrophy (GA) lesion area growth following intravitreal injections of the drug.

Still, the site of action of C3 in AMD is unknown. In a subset of donor eyes diagnosed with GA, C3 and C4 immunolabeling was detected in rods and cones at locations distal to the GA lesion. In contrast, there was no immunoreactivity in the neural retina (NR) of control donor eyes or in the atrophic areas of GA eyes (Katschke et al., 2018). This same study showed anti-C3 labeling of photoreceptor outer segments. Another study also showed C3 expressing cells located in the photoreceptor layer in postmortem eyes from AMD patients (Natoli et al., 2017). Further, a prior study implicated C3 in photoreceptor death following light damage in a mouse model (Natoli et al., 2017). To investigate potential sites of C3 toxicity and provide another model for testing C3 inhibition, we investigated whether C3 knockout would affect RPE or photoreceptor degeneration in the sodium iodate (NaIO₃) model.

NaIO₃ is an oxidizing agent that initially damages the retinal pigment epithelium (RPE) and leads to death of adjacent photoreceptors (Anderson et al., 2024). Prior studies have suggested that oxidative stress and complement dysregulation contribute to AMD (Datta et al., 2017). Complement activation may be due to changes in gene and protein levels or cellular damage caused by oxidative stress (Pujol-Lereis et al., 2016).

The purpose of this study is to test whether knockout of C3 is able to protect the structure and function of photoreceptors and RPE against the damage induced by NaIO₃.

2. Materials and Methods

2.1. Animals

Male C3^−/− mice (strain B6.129S4-C3^tm1Crr/J) and wild-type (WT) mice (C57BL/6J) were purchased from the Jackson Laboratory (Bar Harbor, ME). The animals were fed ad libitum and housed on a 12:12 hour light/dark cycle. They were studied at age 2–3 months. All procedures were approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania and complied with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. As shown in Supplementary Table 1, a total of 48 mice were used and divided into four groups: WT+Saline (n = 18), WT+NaIO3 (n = 18), C3^−/−+Saline (n = 8), and C3^−/−+NaIO3 (n = 13). The genotype and treatment of mice were masked during analysis; only the ID number was used to identify mice.

2.2. Generation of JBWT5 iPS-RPE cells

Induced pluripotent stem cells (iPSCs) from the Penn iPSC Core (University of Pennsylvania) were thawed and cultured in StemMACS iPS-Brew XF medium (Miltenyi Biotec) on Matrigel-coated plates at 37°C, 5% O₂, and 5% CO₂. Flow cytometry confirmed that over 90% of the cells expressed pluripotency markers (SSEA-4 and TRA-1–60). Once the iPSCs reached approximately 60% confluency, the differentiation protocol (Zhang et al., 2023) was initiated. The cells were then incubated at 37°C with 5% CO₂ in a base medium (DMEM/F12, 2% B27, 1% N2, penicillin-streptomycin, and GlutaMAX) for RPE induction over 0–14 days.

On days 0 and 1, the cells were treated with Noggin (50 ng/mL), DKK1 (10 ng/mL), IGF1 (10 ng/mL), and nicotinamide (10 mM). Treatment for days 2 and 3 consisted of Noggin (50 ng/mL), DKK1 (10 ng/mL), IGF1 (10 ng/mL), bFGF2 (5 ng/mL), and nicotinamide (10 mM). For days 4 and 5, activin A (100 ng/mL), IGF1 (10 ng/mL), and DKK1 (10 ng/mL) were applied. From days 6 to 14, the medium was supplemented daily with activin A (100 ng/mL), SU5402 (5 μM), and VIP (1 nM).

Starting from day 15, cells were maintained in RPE medium (DMEM/Ham’s F12 (70:30), 2% B27, GlutaMAX, and penicillin-streptomycin) until RPE cells with cobblestone morphology appeared. After day 35, the cells were dissociated with Accutase/DNase1 for 30 minutes and subcultured onto Matrigel-coated plates in RPE medium containing the ROCK inhibitor thiazovivin (5 μM). Following attachment overnight, thiazovivin was removed, and the medium was refreshed every 2–3 days. The iPS-RPE cells were subcultured and grown on 12-well Transwell inserts for two months. Following their characterization, they were used for NaIO₃ experiments (Fig. S2).

2.3. NaIO₃ administration

NaIO₃ (Sigma-Aldrich, St. Louis, MO) was dissolved in 0.9% Sodium Chloride Injection USP (B. Braun Medical Inc., Irvine, USA) at a concentration of 1% (w/v). A dose of 25 mg/kg was injected intraperitoneally with an insulin syringe (BD, Franklin Lakes, NJ) with 31 gauge, 8 mm length and 300 μl capacity. A right-sided intraperitoneal injection was administered to the mouse. Mice were then returned to the animal facility for 7 days. The dose and sex (male) were chosen based on our previous findings showing consistent NR and RPE damage. Female mice had less, and more variable, ONL thinning at this dose (Anderson et al., 2024).

For NaIO₃ treatment of iPS-RPE cells, NaIO₃ was dissolved in RPE medium (DMEM/Ham’s F12 (70:30) at a concentration of 1.8 mM (which killed approximately 40% of the cells in 24h). The cells were treated for 24 hours and 500 μL of apical and 1000 μL of basal medium.

2.4. In vivo imaging

A week following NaIO₃ administration, the mice underwent in vivo retinal imaging. Spectral-domain optical coherence tomography (SD-OCT; Leica Bioptigen UHR R2200, Wetzlar, Germany) was utilized for imaging. Images were analyzed in ImageJ (NIH, Bethesda, MD) by measuring the distance from the top of the inner nuclear layer (INL) to Bruch’s membrane, following a method described previously (Anderson et al., 2024). In short, the boundaries were manually outlined, and a custom macro measured the distance between them every 50 pixels (approximately 109 μm). Due to the high correlation of damage between the OD and OS eyes, the thickness measurements from the OD horizontal, OD vertical, OS horizontal, and OS vertical images were averaged. These four averages were then combined to produce a single data point per mouse, which was grouped and plotted according to the treatment group.

2.5. In vivo electroretinography (ERG)

One week following NaIO₃ administration, the mice were kept in darkness overnight to prepare for scotopic ERG assessments of a-, b-, and c-wave responses to a series of white light (6500K) flash stimuli. Under dim red lighting, the dark-adapted mice were individually removed from a dark-adaptation chamber prior to anesthesia. To induce pupil dilation, a mixture of Tropicamide (1%, Akorn, Inc.) and Phenylephrine (2.5%, Akorn, Inc.) was applied to both eyes of each mouse. Mice were weighed, and 15 minutes after dilation, a single intraperitoneal dose of Ketamine (95 mg/kg) and Xylazine (11 mg/kg) (Dechra Veterinarian Products) was administered. Once fully anesthetized (around 4 minutes), a topical anesthetic (0.5% Tetracaine HCL, Alcon Laboratories) was applied to the eyes. The mouse was then placed on a 37°C warming platform for the ERG procedure. Platinum wire-embedded contact lenses were used as recording electrodes, with Refresh Tears (Allergan) serving as the contact medium. A platinum wire loop with BSS Sterile Irrigating Solution (Alcon Laboratories) was inserted into the mouth as a reference electrode. ERGs were recorded using a Diagnosys E3 console, a ganzfeld ColorDome, and Espion software (v6.66.1), with the setup housed in a custom-made aluminum Faraday cage. For scotopic a-, b-, and c-wave assessments, flash stimuli ranging from −3.66 to 2 log candela cd·s/m² were presented in a sequence of increasing intensity. For low-intensity flashes (−3.66 to −2 log cd·s/m²), 10 sweeps were averaged, while for medium (−2 to −0.2) and high (0–2) ranges, 5 and 3 sweeps were averaged, respectively.

Amplitudes for scotopic a- and b-waves were derived from the entire range of stimuli, while c-waves were recorded only between 0, 1, and 2 log cd·s/m². The b-wave amplitude was determined by measuring from the a-wave trough to the b-wave peak, which occurred approximately 5–9 ms and 45 ms post-flash, respectively. C-wave peaks were identified between 0.5 and 3 seconds after the stimulus. Once scotopic testing was complete, a steady 30 cd/m² adapting light field was introduced in preparation for photopic measurements. After 7 minutes of continuous adaptation, cone ERG responses were recorded using six flash intensities ranging from 0 to 3 log cd·s/m² against a 30 cd/m² background. The number of trials averaged decreased from 20 for the lowest intensity to 3 for the highest. The cone b-wave amplitude was measured from the pre-flash baseline to the positive peak following the stimulus.

2.6. RPE flatmount tissue processing and immunofluorescence

The globes were enucleated and fixed briefly in 4% paraformaldehyde (PFA) for 15 minutes. The anterior segment was removed to form an eyecup, and the NR was mechanically separated. Samples were blocked for 1 hour in 10% normal donkey serum in TBS-T before being incubated overnight at 4°C with the primary antibody rabbit anti-ZO-1 (1:200, Invitrogen 40–2200). After rinsing in TBS-T, samples were treated with a secondary antibody (Cy3 donkey anti-rabbit, 1:200) for 1 hour. They were then fixed again in 4% PFA, washed with TBS-T, and mounted for imaging. The fluorophore-labeled secondary antibodies enabled visualization of the primary antibody reactivity.

Samples for cross-sectional IF were prepared as previously described (Hadziahmetovic et al., 2011). Primary antibodies used were: C3b/iC3b (1:800, Hycult Bio; #HM1078), rhodopsin (1:800, abcam ab5417). Secondary antibodies used at 1:1000 were: Cy3 donkey anti-rat IgG (Jackson ImmunoResearch Laboratories, Code#712–165-150), 660 goat anti-mouse IgG (Thermo Fisher Scientific, Cat#A-21054). Epifluorescence images were captured using a Nikon DS-QiMc camera and NIS Elements BR 64-bit software.

2.7. NR and RPE/Choroid (RC) isolation and quantitative RT-PCR

Mice were euthanized with CO₂ asphyxiation. Blood perfusion using 0.9% Sodium Chloride Injection USP (B. Braun Medical Inc., Irvine, USA) was performed on all mice to prepare one eye from each mouse for Enzyme-Linked Immunosorbent Assay (ELISA). Following cervical dislocation, eyes were enucleated and placed in cold 1x Hanks’ balanced salt solution without Ca²⁺ and Mg²⁺ (HBSS−), and the cornea, iris, lens, and NR were dissected. The NR and RPE/Choroid (RC) were flash frozen on dry ice and stored at −80 °C. Additionally, iPS-RPE cells exposed to NaIO₃ were used for RNA extraction. The RNeasy mini kit (Qiagen, Valencia, CA) was used for RNA extraction from NR and RC samples. The RNA quantification and reverse transcription were performed as previously described (Anderson et al., 2024). Gapdh and 18S was used as an endogenous control in mouse and iPS-RPE cells samples, respectively. TaqMan probes (ABI, Grand Island, NY, USA) were used as follows: Gapdh (Mm99999915_g1), Rho (Mm00520345_m1), Hmox1 (Mm00516005_m1), Cat (Mm00437992_m1), Gpx4 (Mm00515041_m1), C1q (Mm00432142_m1), Iba1 (Mm00479862_g1), Opn1sw (Mm00432058_m1), CD11b (Mm00434455_m1), Rpe65 (Mm00504133_m1), Hif1α (Mm00468869_m1), C5 (Mm00439275_m1), Cfb (Mm00433918_g1), Opn1mw (Mm00433560_m1), Sod1 (Mm01700393_g1), Sod2 (Mm01313000_m1), Hif1α (Mm00468869_m1), Nrf2 (Mm00477784_m1), 18S (Hs99999901_s1), RPE65 (Hs01071462_m1), NRF2 (Hs00975961_g1), CAT (Hs00156308_m1), SOD1 (Hs00533490_m1), SOD2 (Hs00167309_m1), HIF1α (Hs00153153_m1), HMOX1 (Hs01110250_m1). Data were analyzed using the comparative CT method after each reaction was performed in technical triplicate (Schmittgen and Livak, 2008).

2.8. Enzyme-linked immunosorbent assay (ELISA)

NR and RC tissues were homogenized in 200 μL of 1X Cell Extraction Buffer PTR from a mouse C3 ELISA kit (ab263884, Abcam, UK). Apical and basal media of iPS-RPE cells grown on transwells and exposed to NaIO₃ were processed for C3 detection. Samples were incubated on ice for 20 minutes, then centrifuged at 18,000 × g for 20 minutes at 4°C. The resulting supernatants were transferred into clean tubes, and protein concentration was measured using the BCA Protein Assay Kit (Thermo Scientific, Rockford, IL). The total protein amount was standardized to 55 μg for NR and 2.4 μg for RC samples. The following ELISA kits were used: mouse C3 (ab263884, Abcam, UK), mouse C1q (A303547, Antibodies.com, UK), mouse Hc (OKEH04595, Aviva Systems Biology, USA), mouse C5a (ABIN6955015, antibodies-online.com, USA), and Mouse C5b-9 Terminal Complement Complex/TCC (LS-F22262, LSBio, USA). Standard curves were generated using CurveExpert software (CurveExpert 1.3, Hyams Development, USA).

2.9. Statistical analysis

Data with four groups (WT+Saline, WT+NalO₃, C3^−/−+Saline, C3^−/−+NalO₃) were analyzed using a one-way Analysis of Variance (ANOVA). When the overall p-value for comparison among 4 groups was statistically significant, post-hoc pairwise comparisons between pairs of treatment groups (total of 6 pairs for each outcome measure) were performed and the multiple pairwise comparisons were corrected by using Tukey-Karmer method. Since OCT thickness measures and ERG measures were taken from both eyes, the inter-eye correlation was accounted for by using generalized estimating equations (GEE).

For data with two groups, we employed the student’s two-tailed t-test to compare the difference between the means. These statistical analyses were performed with GraphPad software (GraphPad Prism 9.0; GraphPad Software, Inc., San Diego, CA). A p value ＜ 0.05 was considered to be statistically significant. Mean ± standard deviation (SD) was used for all graphs and significant values were marked as either * (P < 0.05), ** (P < 0.01), *** (P < 0.001), or **** (P < 0.0001).

3. Results

3.1. C3 knockout results in less photoreceptor layer thinning

To determine whether C3 knockout would affect retinal structure in the NaIO₃ model, mice were imaged with OCT seven days after injection. Mice injected with saline served as the control group. Measurements of retinal thickness were taken using OCT images centered on the optic nerve. After NaIO₃ treatment, C3^−/− mice had less retinal thinning than WT mice (Fig. 1A). As the outer nuclear layer (ONL) in NaIO₃ injected mice was poorly defined, the thickness between the inner nuclear layer/inner plexiform layer border to Bruch’s membrane was measured. Retinal thickness in the WT+NaIO₃ group was significantly reduced compared to the WT+Saline group. However, no significant difference in retinal thickness was observed between the C3^−/−+NaIO₃ and C3^−/−+Saline groups. Additionally, the WT+NaIO3 group exhibited significantly thinner retinas compared to the C3^−/−+NaIO₃ group (Fig. 1B). The “spider graphs” provide a more comprehensive depiction of the retinal thinning throughout varying retinal regions, showing the differences across multiple locations in the mouse retina (Fig. 1C and D).

Fig. 1. SD-OCT scans and thickness measurements of mouse retinas protected against NaIO₃ through systemic C3 knockout.

(A) Representative images of horizontal (OCT-H) and vertical (OCT-V) OCT scans of WT+Saline, WT+NaIO₃, C3^−/−+Saline, and C3^−/−+NaIO₃ group mice. The images shown for each condition are from the two eyes of each mouse. (B-D) Quantification of OCT images through measurements taken from the inner nuclear layer (INL)/inner plexiform layer junction to Bruch’s membrane (BrM) where measurements were taken every 50 pixels, which is approximately 109 μm (^† indicates a bracket that shows the INL/inner plexiform layer junction to BrM). Mean ± SD is reported for WT+Saline (n = 15), WT+NaIO₃ (n = 15), C3^−/−+Saline (n = 8) and C3^−/−+NaIO₃ (n = 8) groups. Brackets highlight the INL and outer nuclear layer (ONL) and the arrow indicates BrM. (B) An average of the 12 measurements from each mouse’s four images (OD horizontal, OD vertical, OS horizontal, and OS vertical) was obtained. (C–D) Mean ± SD is reported of the thickness at the indicated distance from the optic nerve among all mice in each group. The thicknesses of the (C) horizontal and (D) vertical OCT images were averaged together to create one data point per treatment group per location. Data was analyzed using a one-way ANOVA. When the overall p-value for comparison among 4 groups was statistically significant, post-hoc pairwise comparisons between pair of treatment groups were performed and the multiple pairwise comparisons were corrected by using Tukey-Karmer method. The inter-eye correlation was accounted for by using generalized estimating equations. Significance only between groups WT+NaIO₃ and C3^−/−+NaIO₃ is indicated in (C) and (D). **** (P < 0.0001).

3.2. C3 knockout does not effectively protect against NaIO₃-induced RPE cell degeneration

The morphology of RPE cells was analyzed to assess the protective effect of C3 knockout against NaIO₃-induced degeneration. En face immunofluorescence of RPE flat mounts revealed that both the WT+NaIO₃ and C3^−/−+NaIO₃ groups exhibited changes in RPE cells (Fig. 2A and B). The central retina had complete RPE atrophy. Only the choroid was visible. At the edge of atrophy, there was a zone with enlarged, irregularly shaped RPE cells. In the peripheral retinas, the RPE appeared normal. As shown in Fig. 2B and C, the distances from the optic nerve to the degeneration edge, and from the degeneration edge to normal RPE cells, were not significantly different between the WT+NaIO₃ and C3^−/−+NaIO₃ groups.

Fig. 2. En face immunofluorescence of RPE flat mounts demonstrates RPE of C3 knockout mice were not protected against NaIO₃.

(A) Representative 10x en face epifluorescence photomicrographs labeling ZO-1 in WT+Saline, WT+NaIO₃, C3^−/−+Saline, and C3^−/−+NaIO₃ groups, taken at 1mm from the optic nerve. (B) Representative 10x en face epifluorescence photomicrographs labeling ZO-1 in WT+NaIO₃ and C3^−/−+NaIO₃ group mice, showing the region from the optic nerve (left side of image) to the degeneration edge, and from the degeneration edge to normal RPE cells. (C) Quantification of the distances from the optic nerve to the degeneration edge, and from the degeneration edge to normal RPE cells. An unpaired two-tailed Student’s t-test was used to analyze the statistical differences between the WT+NaIO₃ and C3^−/−+NaIO₃ groups. ns (P > 0.05).

3.3. ERG analysis demonstrates C3 knockout protects retinal function

In order to determine if C3 knockout provides functional protection, ERG analysis was performed. NaIO₃ treatment significantly reduced rod a-waves (Fig. 3A, 2 log cd.s/m²), rod b-waves (Fig. 3B, 2 log cd.s/m²), c-waves (Fig. 3C, 2 log cd.s/m²), and cone b-waves (Fig. 3D, 3 log cd.s/m²). C3 knockout provided a slight preservation of rod a-waves at 2 log cd.s/m² stimulus intensity (Fig. 3A). C3 knockout provided a moderate yet significant preservation of rod b-waves at higher stimulus intensities (Fig. 3B), with differences between the two NaIO₃-treated groups becoming apparent from −0.19 log cd.s/m² and above. The knockout had a more pronounced protective effect on cone b-waves, particularly at higher intensity levels (Fig. 3D), with the response divergence starting at 0 log cd.s/m². However, consistent with RPE flat mount results, C3 knockout did not protect c-waves in the RPE (Fig. 3C).

Fig. 3. ERG analysis demonstrates C3 knockout mice were partially protected against NaIO₃.

(A–D) Analysis of ERG (A) rod a-wave, (B) rod b-wave, (C) c-wave, and (D) cone b-wave. A comparison was made for each measurement at every light intensity among the four study groups. The mean responses of both eyes from each mouse were measured. Data was analyzed using a one-way ANOVA. When the overall p-value for comparison among 4 groups was statistically significant, post-hoc pairwise comparisons between pair of treatment groups were performed and the multiple pairwise comparisons were corrected by using Tukey-Karmer method. The inter-eye correlation was accounted for by using generalized estimating equations. Significance only between groups WT+NaIO₃ and C3^−/−+NaIO₃ is indicated in the line graphs on the left. * (P < 0.05), ** (P < 0.01), *** (P < 0.001), and **** (P < 0.0001).

3.4. Complement protein levels affected by NaIO₃ and C3 knockout

To assess changes in complement protein levels, we examined NR and RC samples from four groups of mice, as well as the supernatant from iPS-RPE cells, using ELISA. In the NR (Fig. 4A), we observed a significant increase in the levels of C1q, C3, and C5a proteins in the WT+NaIO₃ group compared to the WT+Saline group. In contrast, the C3^−/−+NaIO₃ group did not show a significant upregulation of these proteins compared to the C3^−/−+Saline group. Only the C3^−/−+NaIO₃ group had a significant increase in C5.

Fig. 4. Quantification of proteins using ELISA.

(A) Quantification of the indicated proteins in lysates of mouse NR samples. (B) Quantification of the indicated proteins in lysates of mouse RPE/Choroid samples. Each data point represents results from one eye of a mouse. (C) Quantification of C3 protein in the apical and basal media from the iPS-RPE cell line-JBWT5 grown on transwells following 24 hours of NaIO₃ treatment. A one-way ANOVA followed by Tukey’s multiple comparisons test was used for statistical analyses. * (P < 0.05), ** (P < 0.01), *** (P < 0.001), and **** (P < 0.0001).

In the RC (Fig. 4B), C1q levels were significantly elevated in both NaIO₃-treated groups. As expected, C3 was undetectable in the NR and RC tissues from C3^−/− mice. Secreted C3 was detected in both the apical and basal supernatants of iPS-RPE cells (Fig. 4C), but did not show a significant difference between the 1.8 mM NaIO₃-treated cells and the control.

3.5. NaIO₃ treatment alters levels of opsin, antioxidant, and inflammation related mRNAs

To gain additional understanding of the effects of NaIO₃ and C3 knockout on the retina, RT-qPCR was used to study mRNA levels in isolated NR and RC. In the NR (Fig. S1A), levels of Rho and Opn1sw were significantly reduced in both NaIO₃-treated groups compared to the saline groups; however, the C3 knockout group showed relatively less downregulation, consistent with some protection of rod photoreceptors. Hmox1 levels were significantly elevated in both NaIO₃-treated groups, though the increase was notably smaller in the C3 knockout group, with significantly lower mRNA levels compared to the WT+NaIO₃ group, suggesting less oxidative stress in the C3^−/−+NaIO_3. Consistent with this, Cat levels were significantly higher in the WT+NaIO₃ group than in the WT+Saline group, while the C3^−/−+NaIO₃ group did not exhibit a significant increase relative to the C3^−/−+Saline group. Gpx4 levels followed a similar trend, with a significant rise in the WT+NaIO₃ group but no significant change in the C3^−/−+NaIO₃ group. Iba1 and CD11b levels were significantly increased in both NaIO₃-treated groups, providing no evidence for a difference in mononuclear phagocyte recruitment or activation.

In the RC (Fig. S1B), levels of Hif1α, C1q, and Iba1 were significantly elevated in both NaIO₃-treated groups, with no reduction in upregulation observed in the C3 knockout group. C5 levels were significantly higher in the WT+NaIO₃ group compared to the WT+Saline group, whereas no significant increase was detected in the C3^−/−+NaIO₃ group relative to the C3^−/−+Saline group. Rpe65, and, surprisingly, levels of antioxidants Cat, Sod1, and Sod2 were diminished in the RPE of both groups treated with NaIO₃. The absence of an increase in antioxidant mRNAs may have to do with the timepoint; many of the RPE cells that were injured by NaIO₃ had already died 7 days after the treatment.

In iPS-RPE cells (Fig. S1C), treatment with 1.8 mM NaIO₃ for 24h resulted in significantly lower RPE65 levels. Except for HMOX1, all other oxidative stress markers, including NRF2, CAT, HIF1α, SOD1, and SOD2, showed significantly higher mRNA expression compared to the control group.

3.6. Immunolabeling shows iC3b in photoreceptor outer segments in NaIO₃-treated WT but not C3 knockout mice

To determine whether opsonization with iC3b may contribute to photoreceptor death in mice treated with NaIO₃, we performed IHC with an anti-iC3b antibody (Fig. 5). Results show strong labeling in the ONL, inner, and outer segments. Co-localization is shown with rhodopsin in the photoreceptors (Fig. 5C, F, I). As a control, sections from C3^−/− mice were labeled and showed markedly diminished signal (Fig. 5G-I). Pixel density was quantified in each group (Fig. 5J, n = 3 mice in each).

Fig. 5. Immunohistochemistry for iC3b.

Representative cross-sectional epifluorescence photomicrographs of (A-C) WT+Saline, (D-F) WT+NaIO₃, and (G-I) C3^−/−+NaIO₃ groups. (A, D, G) No primary control stained with DAPI (blue) for each group. (B, E, H) DAPI (blue) and iC3b (red). (C, F, I) The same images as seen in B, E, and H respectively with the addition of Rho (cyan). Scale bar in I represents 50 μm and is applicable for all figures. (J) Densitometry quantification of iC3b (red) signaling across the groups (n=3 per group). A one-way ANOVA followed by Tukey’s multiple comparisons test was used for statistical analyses. * (P < 0.05), ** (P < 0.01).

4. Discussion

In recent decades, systemically administered NaIO₃ has been widely used as a model of oxidative-stress induced retinal degeneration (Reisenhofer et al., 2017). Changes in retinal morphology are dependent on both time and dose, characterized by degeneration in the RPE layer followed by a reduction in the ONL thickness (Yang et al., 2014). Multiple cell death mechanisms, including apoptosis, necrosis, necroptosis, ferroptosis, and pyroptosis have been implicated in RPE and secondary photoreceptor cell death in this model (Upadhyay and Bonilha, 2024).

In this study, C3^−/− mice exhibited less NaIO₃-induced reduction in the thickness of the photoreceptor cell layer compared to WT mice. ERG results showed partial preservation of rod and cone function in C3^−/− mice. Photoreceptor outer segments were decorated with iC3b in NaIO₃-treated WT but not C3^−/− mice. These results strongly implicate C3 in the pathogenesis of photoreceptor but not RPE death in the NaIO₃ model.

Among the ERG measurements, the cone b-wave was most protected in C3^−/− mice. The rod a and b-waves were slightly protected but the RPE derived c-wave was not. These results are consistent with the lack of morphological protection of the RPE. They also indicate that while many rods were morphologically protected, they had minimal function—perhaps because they were lacking the retinal isomerase activity normally provided by RPE65 in the RPE cells, as shown in the qPCR results. In contrast, cones can use an isomerase present in the cones themselves or in Muller cells (Tang et al., 2011), which may explain why their function is better protected than that of rods.

Based on the ELISA results, NaIO₃ induced a significant elevation in complement proteins C1q, C3, and C5a in the NR, while C3 knockout prevented this complement upregulation and activation. Additionally, the mRNA levels of antioxidants Hmox1, Cat, and Gpx4 were less upregulated in the C3 knockout group. These findings suggest that the protective effect of C3 knockout may also occur through diminished oxidative stress.

C3 promotes photoreceptor death by cleaving into C3a and C3b. There are several inflammatory functions of C3a, including chemotactically attracting phagocytes to infection sites. C3 cleavage/activation can also trigger the formation of the membrane attack complex (C5b-9 complex), which can lyse targeted cells by forming pores in cell membranes, or, in the presence of higher levels of membrane bound complement inhibitors, rather than lysing the cells it can activate intracellular signaling pathways (Triantafilou et al., 2013). IHC in Fig. 5 indicates that NaIO₃ induces iC3b opsonization on photoreceptor outer segments, which likely contribute to photoreceptor death. The lack of these complexes on NaIO₃-treated C3^−/− mice may be the main explanation for photoreceptor protection.

C3 has been found in macrophages/microglia in multiple retinal degeneration models, including the light damage model, where C3 knockout was found to protect photoreceptors (Natoli et al., 2017). Conversely, C3 upregulation in the rd10 model was found to be helpful in clearance of dead photoreceptors, and C3 knockout accelerated structural and functional photoreceptor degeneration (Silverman et al., 2019). We hypothesize that the positive and negative effects of C3 knockout or ablation may be related to the kinetics of retinal degeneration and the type of retinal cells damaged first. When RPE toxicity was primary in the NaIO₃ model, C3 knockout preserved photoreceptors.

While preparing this manuscript, we became aware that C3 knockout was also reported to protect photoreceptors and RPE against NaIO₃ by another group (Wang et al., 2022). They provided evidence that C3a receptor inhibition is protective, implicating recruitment or activation of macrophages or microglia as a harmful component of NaIO₃ pathogenesis. Most of the findings reported in their study are consistent with ours. Our study is unique for including OCT and ELISA data, which are important for in vivo disease staging and understanding the levels and location of C3 protein. Putting their finding implicating the importance of mononuclear phagocytes together with ours showing iC3b on photoreceptor outer segments suggests that photoreceptor opsonization followed by ingestion by mononuclear phagocytes may be important in the model. One difference between the two studies is that they show RPE protection by C3 knockout in their model but we did not find this. Our RPE flat mounts show that C3 knockout did not protect the RPE, which is consistent with the appearance of the RPE on OCT, the lack of c-wave protection and no protection against diminished Rpe65 levels in the qPCR. One potential explanation is differences in the dose and route of administration of NaIO₃. They used intravenous NaIO₃ at 35mg/kg while we used intraperitoneal NaIO₃ at 25mg/kg. Sex differences are another possible explanation. We used male mice only, as we recently defined a distinct sex difference in NaIO₃ toxicity (Anderson et al., 2024), while their manuscript did not specify the sex.

Interestingly, Malfaul et al. (Mulfaul et al., 2020) reported that complement upregulation in the NaIO₃ model was mediated by the damage associated molecular patterns receptor Tlr2. Further, they showed that C5 knockout was protective against the mild photoreceptor degeneration observed three days after NaIO₃ injection.

The source of secreted C3 in the NaIO₃ model may be mononuclear phagocytes or gliotic Müller cells, as implicated in prior studies (Enzbrenner et al., 2021). We also found C3 by ELISA in mouse RPE/choroid and both apical and basal medium from iPS-RPE cells grown on transwells, but the amount coming from RPE cells was not changed by NaIO₃. Nevertheless, C3 secreted from any cell type could be activated to produce the iC3b that we observed on photoreceptor outer segments.

Conclusion

In conclusion, this study has provided detailed analysis of photoreceptor, but not RPE protection provided by knockout of C3 in the NaIO₃ model. This model should prove useful for testing complement inhibitors for photoreceptor protection.

• NaIO₃-induced elevation in complement C3 protein was found only in neural retina.

• C3 knockout provides structural and functional photoreceptor protection against NaIO₃.

• Neither RPE structure nor function was protected by C3 knockout.

Abbreviations

AMD

age-related macular degeneration

BrM

Bruch’s membrane

C3

complement factor 3

Cat

Catalase

ERG

electroretinography

FDA

Food and Drug Administration

GA

geographic atrophy

Hmox1

heme oxygenase 1

IHC

immunohistochemistry

INL

inner nuclear layer

NaIO₃

sodium iodate

NR

neural retina

OCT

optical coherence tomography

OCT-H

optical coherence tomography horizontal scans

OCT-V

optical coherence tomography vertical scans

ONL

outer nuclear layer

RPE

retinal pigment epithelium

Rho

rhodopsin

SD

standard deviation

SD-OCT

spectral-domain optical coherence tomography

ARVO

the Association for Research in Vision and Ophthalmology

WT

wild-type

Data availability

Data will be made available on request.