Background
Age-related macular degeneration (AMD) is the leading cause of blindness in developed countries. It has been estimated that 288 million people will be affected by this condition by 2040 [
1,
2]. Neovascular AMD (nAMD) causes severe vision loss and is characterized by the macular invasion of abnormal blood vessels from the choroid (e.g., choroidal neovascularization, CNV, or polypoidal choroidal vasculopathy, PCV) or the retina (i.e. retinal angiomatous proliferation, RAP). These new vessels can subsequently lead to the formation of subretinal fibrotic plaques in the macula, known as macular fibrosis [
3‐
5]. Anti-VEGF therapy is the mainstay of nAMD treatment [
6]. Although anti-VEGF therapy can stabilize and even improve visual function in nAMD, nearly half of patients suffer from poor prognosis, largely due to the development of macular fibrosis [
5]. Currently, there are no medications to prevent or treat macular fibrosis. Therefore, novel strategies based on a better understanding of disease pathogenesis are urgently needed.
Macular fibrosis originates from new blood vessels and is a fibro-vascular membrane [
7]. The conversion of the diseased blood vessels into a fibro-vascular lesion is due to excess deposition of extracellular matrix (ECM) from activated fibroblasts (myofibroblasts). The retina, including the macula, is absent of fibroblasts. Evidence suggests that myofibroblasts can be recruited from the choroid and blood circulation. They may also be transdifferentiated from other cells such as Müller cells, endothelial cells, retinal pigment epithelial (RPE) cells and infiltrating macrophages [
7,
8]. For example, RPE cells can transdifferentiate into myofibroblasts through epithelial-to-mesenchymal transition (EMT) and this process is believed to play a pivotal role in AMD [
3,
5,
7,
9,
10]. The characteristics of EMT in RPE cells include the loss of epithelial features such as cell-to-cell contact, decreased expression of adherence and tight junction proteins. Additionally, the gain of mesenchymal features such as increased cell migration and invasiveness, higher levels of contractility and upregulation of ECM proteins such as collagen type I (Col-I) and fibronectin (FN) and α-smooth muscle actin (α-SMA). The molecular cues governing the transdifferentiation of myofibroblasts from other cells in the context of macular fibrosis remain poorly defined, although sustained inflammation is believed to be the main driver of pathogenic fibrosis.
The complement system is involved in chronic inflammation of many degenerative diseases, including AMD [
11,
12]. In recent years, emerging evidence suggests that the complement system also plays a critical role in organ fibrosis such as the lung [
13,
14], kidney [
15‐
19] and liver [
20]. The complement fragments C3a and C5a are cleavage products of C3 and C5 during complement activation and are involved in diverse immune responses. C3a and C5a can bind their cognate receptors C3aR and C5aR initiating pro-inflammatory, pro-angiogenic and pro-fibrotic responses in both immune and tissue cells. Therefore, they are critically involved in tissue regeneration, remodelling and fibrosis. We previously reported higher plasma levels of C3a and C5a in nAMD patients with subretinal fibrosis [
21]. We further found that C3a but not C5a could induce macrophage-to-myofibroblasts transition [
8]. This study aimed to understand whether C3a and C5a could induce EMT in RPE cells in the context of subretinal fibrosis and whether blocking complement activation could prevent or reduce retinal fibrosis.
Materials and methods
Human eyes
Human eye samples with nAMD were obtained from the San Diego Eye Bank. This study was carried out within the parameters of the Declaration of Helsinki, and tissues were stored in accordance with the UK Human Tissue Act (2004). The research was approved by the Ethical Review Boards of Queen’s University Belfast. The eyes were maintained in formalin. Upon arrival, the eyes were dissected and embedded in paraffin and sectioned at 6 µm thickness.
Animals
C57BL/6J mice aged between 2 and 4 months were used in this study. All animals were housed and bred in a standard pathogen-free experimental facility and exposed to a 12-h light/dark cycle with free access to food and water. All procedures were conducted under the regulation of the UK Home Office Animals (Scientific Procedures) Act 1986. This study was approved by the Animal Welfare and Ethical Review body (AWERB) of Queen's University Belfast and conducted in compliance with the Association for Research in Vision & Ophthalmology Statement for the Use of Animals in Ophthalmology and Vision Research.
Two-stage laser-induced subretinal fibrosis
Subretinal fibrosis was induced using a two-stage laser protocol previously described by our group [
8,
22]. Briefly, CNV was induced by using the laser photocoagulator (HGM Medical Laser System Inc. Salt Lake City, USA). The settings for the laser were as follows: laser power, 250 mv; duration, 0.1 s; and spot size, 100 μm. Four laser spots were delivered per eye. Seven days later, a second laser burn was applied to each CNV lesion using the same laser configuration.
Inhibition of C5 or C5aR in vivo in subretinal fibrosis
To inhibit complement activation, mice were injected intraperitoneally with 250 µg per animal of C5 blocking antibody, BB5.1 (Cat. HM1073, Hycult, Uden, Netherland) 1 h before the second laser injury (day 0). A second injection was performed 5 days later (125 µg/animal). The doses were chosen based on a previous study of this antibody in experimental autoimmune uveitis [
23]. Control mice were injected with the same amount of isotype control (mouse IgG, Cat. MAB002, R&D Systems, Minneapolis, MN). Peripheral blood was collected before the first laser, 24 h after each injection (day 1 and 6) and at the endpoint (day 10) and the serum was isolated and used for the complement activation assays.
To block C5aR, mice were subcutaneously injected daily with the C5aR antagonist PMX53 (1 mg/kg, Cat. 5473, Tocris, Bio-techne, Minneapolis, MN) starting 1 h after the second laser (day 0) until the end point. Control mice were injected with the same volume of vehicle (saline).
Fundus fluorescence angiography
Fundus images and fundus fluorescein angiography (FFA) were conducted on day 10 post second laser, using the Micron IV system and the Discover 2.2 Programme (Phoenix Technology Group, Pleasanton, CA). FFA was carried out 5 min after intra-peritoneal injection of 100 μL of 10% sodium fluorescein (Sigma-Aldrich, Gillingham, UK, Cat. F6377). Exposure level was kept consistent between animals. The area of fluorescein leakage from each lesion were analysed using ImageJ (NIH, Bethesda, MD) by two independent researchers in a masked fashion. Following retinal fundus (Micron IV) examination, mice were killed by CO2 and eyes were collected and fixed in 2% paraformaldehyde for 2 h (Sigma-Aldrich, Cat. 158127) and processed for RPE/choroidal flatmount staining.
Complement activity assay
Classic complement system activation in serum from mice treated with BB5.1 C5 antibody or control mouse IgG was determined using the Hycult Mouse Classical Complement Pathway assay (HIT420, Hycult Biotech, Uden, Netherlands) following manufacturer’s instructions. Besides negative and positive controls provided in the kit, a reference of total complement activity was created by stimulating a pool of the serum samples from the control animals with 2 µg/mL LPS for 2 h 37 °C. Unstimulated plasma was considered as reference negative control. Complement activity (%) of each sample was calculated as follows: (sample – reference negative control)/(total complement activation − reference negative control) × 100.
Primary culture of RPE cells
Primary mouse RPE cells were cultured using the protocol previously described [
24,
25]. Briefly, eyes were collected from 2- to 3-month-old C57BL/6J mice. Anterior segment of the eye (cornea, lens, iris and ciliary body) was removed. The retina was carefully peeled off of the RPE/choroidal eyecup. The RPE/choroidal eyecups were incubated with pre-heated 0.05% trypsin (Gibco, Cat. 10779413) for 45 min at 37 °C. The RPE cells were flushed from the eyecups and cultured in Dulbecco’s modified Eagle medium: nutrient mixture F-12 (DMEM/F12, Cat. 11320033, Gibco, Waltham, MA) supplemented with 15% FCS (Gibco™, Cat. 10270106) and 1% penicillin–streptomycin (Gibco, Cat. 15140122). The phenotype of RPE cells was confirmed by RPE65 staining. Cells from passages 3–5 were used in the study. For experiments, media was changed to lower serum (1% FCS) when fully confluent to facilitate cell quiescence. 24 h later, the different treatments (C3a, C5a or TGF-β2, concentrations ranging from 10 to 100 ng/mL [
8,
16] or 50 nM of PMX53, all dissolved in PBS) were added in DMEM-F12 supplemented with 1% FCS for different lengths of time. The sources of these recombinant proteins are detailed in Table
1.
Immunostaining
Cells were fixed in 2% paraformaldehyde for 20 min, rinsed in PBS, and blocked with 10% BSA (Sigma-Aldrich) and permeabilized with 0.1% Triton X-100 (Sigma-Aldrich). Cells were then incubated overnight (4 °C) with primary antibodies (Table
1) diluted in PBS. Following incubation, cells were incubated with fluorophore-conjugated secondary antibodies (Table
2) at room temperature for 1 h. Cells were counterstained using DAPI-Vectashield (Vector Labs, Burlingame, CA) and examined under Leica DMi8 epifluorescence microscope.
Table 1
Recombinant proteins and primary antibodies used in the study
Recombinant Mouse Complement C3a, CF | 8085-C3 | R&D Systems |
Recombinant Mouse Complement C5a | 2150-C5-025 | R&D Systems |
Recombinant Mouse TGF-β2 | 7346-B2-005 | R&D Systems |
PMX 53 (C5a receptor antagonist) | Cat. No. 5473 | Tocris-Biotechne |
Fibronectin | ab2413 | Abcam | Rabbit |
E Cadherin antibody | orb213706 | Biorbyt | Rabbit |
α-SMA (-Cy3 conjugated) | C6198 | Sigma-Aldrich | Mouse |
Rabbit anti-α-SMA | ab5694 | Abcam | Rabbit |
C5b-9 | ab65811 | Abcam | Rabbit |
C5aR | ab59390 | Abcam | Rabbit |
C5aR | ab117579 | Abcam | Rat |
C5R1 Biotin | ab54378-100 | Abcam | Biotin |
C5a antibody (FITC) | orb360860 | Biorbyt | Rat |
C5a antibody | MAB21501-100 | R&D Systems | Goat |
Collagen 1 Rabbit pAb | ab34710 | Abcam | Rabbit |
Gt X Collagen Type I | AB758 | CHEMICON | Goat |
Vimentin polyclonal antibody | Orb304659 | Biorbyt | Rabbit |
Slug antibody | PSI-3957 | ProSci | Rabbit |
MHC Class II (I-A/I-E) Monoclonal Antibody | 14–5321-82 | eBioscience | Rat |
GAPDH | G8795 | Sigma | Mouse |
AKT/MAPK Pathway Antibody Cocktail | ab151279 | Abcam | Rabbit |
pSmad2/3 | 8828 | Cell Signaling Technologies | Rabbit |
Table 2
Secondary antibodies and ELISA kits used in the study
Secondary antibodies | | |
Alexa Fluor® 594 AffiniPure Donkey Anti-Rabbit IgG (H + L) | 711–585-152 | Jackson Immunoresearch |
Alexa Fluor® 594-AffiniPure Donkey Anti-Goat IgG (H + L) | 705–585-147 | Jackson Immunoresearch |
Alexa Fluor® 488-AffiniPure Donkey Anti-Goat IgG (H + L) | 705–545-147 | Jackson Immunoresearch |
Alexa Fluor® 488-AffiniPure Donkey Anti-Rabbit IgG (H + L) | 711–545-152 | Jackson Immunoresearch |
Alexa Fluor® 488-AffiniPure Fab Donkey Anti-Rat IgG (H + L) | 712–547-003 | Jackson Immunoresearch |
Goat Anti-Rabbit IgG H&L (HRP) | ab6721 | Abcam |
Rabbit Anti-Mouse IgG H&L (HRP) | ab6728 | Abcam |
Rabbit Anti-Goat IgG H&L (HRP) | ab6741 | Abcam |
ELISA kits |
IL-6 | BMS603-2 | Thermo Fisher Scientific |
TNF-α | BMS607-3 | Thermo Fisher Scientific |
Mouse TGF-beta 2 DuoSet ELISA | DY7346-05 | R&D |
Mouse VEGF DuoSet | DY493-05 | R&D |
Mouse Complement Component C5a DuoSet ELISA | DY2150 | R&D |
Mouse TGF-beta 1 DuoSet ELISA | DY1679-05 | R&D |
C5 ELISA | ORB565566 | Biorbyt |
RPE flatmounts were stained as previously described [
22]. The primary and secondary antibodies used in the study are shown in Tables
1 and
2, respectively. Human eye paraffin sections were stained using the protocol previously described [
8]. Briefly, antigen retrieval was carried out by boiling slides in antigen retrieval buffer (0.05% citraconic acid, pH 7.4) (Sigma-Aldrich, Cat. C82604) for 30 min and blocked 1 h at room temperature in 10% donkey serum (Sigma-Aldrich, Cat. D9663). Samples were incubated with the primary antibody (Table
1) overnight at 4 °C and 1 h room temperature in the dark with the corresponding secondary antibody (Table
2).
The samples were cover-slipped with DAPI-Vectashield (Vector Labs) and examined by Leica DMi8 epifluorescence microscope or confocal microscope (Leica TCS SP5, Leica Microsystems Ltd., Wetzlar, Germany). The fibrotic lesion was measured in RPE/choroid flatmount using a protocol described previously [
22]. The measurements were conducted double-blinded by two independent researchers.
Collagen matrix contraction assay
Primary RPE cells (1.5 × 105 cells/mL, 500 μL per well) were suspended in 2 mg/mL rat tail type I collagen (Collagen I Rat Protein, Tail A1048301 Thermo Fisher Scientific, Waltham, MA) which was dissolved in 6 μL of 0.1 M NaOH and seeded in a 24-well plate. After incubation at 37 °C for 3 days in DMEM-F12 + 10% FCS, media was changed to DMEM-F12 containing 1% FCS with different treatments and immediately after the cell collagen gels were detached from the bottom of the wells. Pictures of the surface area of each matrix were taken in Syngene G-Box imaging system (Syngene, Cambridge, UK) at 0 h after the detachment and every 24 h thereafter.
Wound healing assay
Primary RPE cells were seeded in 6-well plates (1.5 × 105 cells per well) and cultured until confluence. The medium was then changed to 1% FCS for 24 h. The cell monolayer was scratched with a 200-µL tip to inflict a wound ∼1 mm in width, washed several times and then treated with C5a in the presence or absence of C5aR antagonist PMX53 in DMEM-F12 with 1% FCS. The wound was photographed immediately and 24 h after the scratch. Images were analysed using ImageJ software (NIH, Bethesda, MD).
Western blot
Samples were homogenized in RIPA buffer containing protease inhibitor cocktail. Protein concentration was determined using a Pierce BCA protein assay kit (Thermo Fisher Scientific. Cat. 23225) or Bradford Assay (Cat. ab119216, Abcam). The blot was performed using 15 to 20 µg of protein according to previously described methods [
26,
27]. Primary and secondary antibodies used are detailed in Tables
1 and
2. Membranes were visualized with enhanced chemiluminescence (Clarity Western ECL Blotting Substrates; Bio-Rad Laboratories) and bands detected using Syngene G-Box imaging system (Syngene). Western Blot analyses were performed using ImageJ software and densitometry normalized to loading control GAPDH or Rab11.
Enzyme-linked immunosorbent assay (ELISA)
Supernatants of primary murine RPE cells treated with C5a or TGF-β2 for different lengths of time was used for determine the concentration of C5a, C5, TGF-β1, TGF-β2 and VEGF. ELISA kits for C5a, C5, TGF-β1, TGF-β2 and VEGF (R&D Systems), IL-6 and TNF-α (Thermo Fisher Scientific-Invitrogen) (Table
2) were used according to the manufacturers’ instructions. The concentration obtained in pg/mL was then normalized by the total protein concentration of each sample measured with Bradford assay (Cat. ab119216, Abcam).
Data analysis
We used the Spectacle platform [
28] with the dataset from the study of Voigt et al. [
29] to analyse the gene expression of epithelial marker E-cadherin (
CDH1), mesenchymal markers FN (
FN1) and αSMA (
ACTA2), C5aR1 (
C5AR1) and C3aR1 (
C3AR1) in human RPE/choroidal cells. The dataset in the study by Voigt et al. includes single cell RNA sequencing (ScRNAseq) information of RPE/choroidal tissue from the macular and peripheral areas of healthy donors. The analysis allows identification of specific cell types that express the genes of our interest and the graph is generated by the Spectacle platform.
The Graph Pad Prism (V6, GraphPad Software, San Diego, CA) was used to create graphs and conduct statistical analyses on our laboratory data. Statistical differences between two groups were assessed via an independent Student’s t test. On larger datasets, one-way or two-way ANOVA was used where appropriate. Bonferroni correction was used for multiple comparison testing.
Discussion
The development of subretinal fibrosis following CNV is associated with multiple biological processes including infiltration of inflammatory cells such as macrophages [
8], EMT of RPE cells, EndoMT and recruitment of fibrocytes from blood and choroid [
7]. The molecular cues that are involved in the conversion of CNV into a fibro-vascular lesion remain poorly defined. In this study, we show that complement activation critically contributed to the development of subretinal fibrosis and blocking complement activation with BB5.1 significantly reduced subretinal fibrosis. Mechanistically, the complement system may promote subretinal fibrosis, at least partially, through C5a-induced EMT in RPE cells.
Previously, we reported higher plasma levels of C3a and C5a in patients with macular fibrosis. The circulating complement fragments (C3a and C5a) may be recruited to the diseased macula and participate in fibrosis development. C3a and C5a can also be released locally from cleavage of C3 and C5 during complement activation in AMD. We found that RPE cells constitutively produced C5/C5a and the production was enhanced by TGF-β2 (Fig.
2C, D), suggesting that RPE-derived C5 may contribute to complement activation in retinal fibrosis. C5a is a potent pro-inflammatory peptide and can act as an anaphylatoxin as well as a chemoattractant for various immune cells including neutrophils, eosinophils, monocytes, and T lymphocytes [
37]. C5a interacts with its receptor C5aR (CD88), a G-protein-coupled receptor capable of modulating cell function and behaviour. The C5a/C5aR pathway is known to be involved in the pathogenesis of various retinal diseases, including uveoretinitis [
23] and AMD [
38,
39] through (a) C5a-mediated recruitment of circulating immune cells [
39] and (b) C5a-induced inflammatory responses in macrophages [
23] and RPE cells [
40]. Here, we show that C5aR, but not C3aR is expressed in RPE cells and activation of the C5a/C5aR pathway led to EMT in RPE cells, a phenomenon that has not been reported before. C5a dose-dependently upregulated the expression of myofibroblast marker FN and down-regulated epithelial marker E-cadherin in RPE cells. The mesenchymal phenotype of C5a-treated RPE cells was further confirmed by their expression of α-SMA, vimentin, the transcription factor Slug (Fig.
3) and their high migration and contraction activities (Fig.
5). It is well-known that TGF-β can induce EMT in RPE cells. Interestingly, we found that the production of C5/C5a in RPE cells was enhanced by TGF-β, and the treatment of RPE with C5a also stimulated the release of TGF-β1 and TGF-β2, suggesting a positive feedback loop between C5a- and TGF-β-induced EMT in RPE cells.
The complement system is known to be involved in tissue/organ fibrosis including the lung [
13,
14], kidney [
15‐
18], liver [
20] and retina [
8]. C3a and C5a can induce mesenchymal transition in macrophages [
8], pericytes [
17], epithelial cells [
15,
16,
41] and endothelial cells [
18]. Complement activation is critically involved in the development of choroidal neovascularization (CNV) [
12,
42]. Parsons et al. suggested that persistent complement activation may be required to maintain the fibrotic scar in CNV, since blocking the alternative pathway of complement activation accelerates retinal repair beyond the normal rate [
42]. Here, we show for the first time, that blocking complement activation using a C5 neutralizing antibody significantly reduced subretinal fibrosis secondary to CNV. Apart from RPE cells, C5aR is also expressed in choroidal macrophages and fibroblasts (Figs.
8 and
10). It has been reported that C5a can stimulate macrophage polarization towards alternatively activated pro-fibrotic M2 phenotype [
43]. C3aR, on the other hand, is expressed predominately in choroidal macrophages (Fig.
10A). We previous showed that C3a but not C5a induced macrophages-to-myofibroblast transition (MMT) [
8]. It is possible that C3a-induced MMT, and C5a-induced M2 macrophage polarization and EMT in RPE cells shown here, may all contribute to complement-mediated subretinal fibrosis. In line with previous reports [
17,
18], we found that C5a-induced EMT in RPE involves the canonical TGF-β pathway (Smad2/3) as well as the non-canonical, the GCPR receptor C5aR pathway (ERK1/2).
A sustained low-grade inflammation in the vascular lesion site is known to be a crucial driver of macular fibrosis [
44]. In addition to MMT and EMT, the complement proteins may also promote macular fibrosis through the induction of pro-inflammatory and pro-fibrotic factors in macrophages and RPE cells. A previous study reported increased expression of IL-8, IL-1β, IL-6, GM-CSF, and CCL2 (MCP-1) in C5a-treated RPE cells [
40]. Here, we found the production of TGF-β1/2 and IL-6 was increased in C5a-treated RPE cells, in line with previous observations in renal tubular epithelial cells and macrophages [
16,
45]. C5a may drive RPE cells into a pro-inflammatory state in the process of transdifferentiating from epithelial to mesenchymal phenotype. These pro-inflammatory and pro-fibrotic mediators can initiate a cascade of cellular changes and induce myofibroblast transition from other cells such as macrophages [
8], Müller cells [
46] or endothelial cells [
7] to further contribute to subretinal fibrosis.
Deletion or blockade of C5aR has been found effective in alleviating tubulointerstitial [
15,
45], pancreatic [
47], glomerular [
18] and pulmonary [
13,
48] fibrosis. In our study, the administration of the C5aR antagonist PMX53 significantly reduced subretinal fibrosis in clinical examination, but less so in immunohistochemical examination. Instead, blocking the overall complement activation (with BB5.1) strongly suppressed subretinal fibrosis. The lack of efficacy of PMX53 in in vivo fibrosis (but not in in vitro EMT) has been observed in arthritis [
49,
50], and possible explanations may include: (1) PMX53/Mas-related gene 2 (MrgX2) mediated mast cell degranulation [
32]. Mast cell degranulation and/or the associated histamine release has been shown to be involved in the pathogenesis of various organ fibrosis [
51,
52]; (2) rapid elimination of the drug or its poor tissue penetrance [
31,
33,
50]; (3) C5a–C5aR2 (C5L2) mediated inflammatory response. C5aR2 is another C5a receptor that is thought to regulate the C5a–C5aR effects although C5aR2 signalling function (anti- or pro-inflammatory properties) is contradictory [
53].
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.