Background
Age-related macular degeneration (AMD) is the leading cause of blindness in the elderly in the developed world. Patients with AMD may present with degeneration of retinal pigment epithelium (RPE) and choroid (geographic atrophic, GA) or neovascular membrane and retina and/or RPE detachment (neovascular AMD, nAMD) in the macula. Leakage from neovascular membrane causes macular oedema and visual impairment. Currently, nAMD is treated by intravitreal injections of VEGF inhibitors including ranibizumab (Lucentis, Genentech, San Francisco, CA), bevacizumab (Avastin, Roche, Basel, Switzerland) and more recently aflibercept (Eylea, Regeneron, Tarrytown, NY). Although the therapy can stabilize the disease and even improve vision, not all patients respond to the treatment [
1], suggesting that multiple pathways may be involved in the pathogenesis of nAMD.
Epidemiological and genetic studies have shown that ageing, environmental factors (e.g. tobacco smoking, diet, hypertension, cardiovascular disease) [
2,
3] as well as gene polymorphisms (e.g. complement-related genes (CFH, C3), Toll-like receptors (TLRs) and chemokine receptors (CX3CR1)) [
4‐
7] increase the risk of AMD. Many of the genes involved in AMD are associated with the immune system, suggesting that dysregulated immune response or inflammation may contribute to AMD aetiology, and this concept is now well appreciated and supported by multiple observations [
8]. Inflammatory molecules, including complement components, immunoglobulins and C-reactive protein, have been detected in drusen deposits of AMD patients [
9]. Drusen particles or its constituents, such as amyloid beta, can induce inflammasome activation and inflammatory cytokine expression (IL-1β and IL-8) in macrophages and RPE cells [
10]. Additionally, inflammatory cytokines (IL-6 and IL-8) have been detected in the aqueous humour of nAMD patients [
11].
Inflammation is generally considered as a protective response to infection and tissue injury. When the macula is damaged due to old age and environmental risk factor-mediated oxidative insults, tissue resident immune cells including retinal microglia and choroidal macrophages as well as the complement system may be activated to repair damage and maintain macular function. In addition, circulating immune cells may be recruited to participate in macular inflammation [
12]. How the tissue-protective immune response becomes detrimental and induces angiogenesis in nAMD remains unknown. It is believed that the macular microenvironment (high levels of oxidative lipids/proteins and hypoxia) may critically control the level and type of immune response in the ageing eye. We hypothesize that, in addition to macular microenvironment, immune cells in nAMD patients may have a disease-causing phenotype that makes them more pro-inflammatory or angiogenic once recruited to the damaged macula. A recent transcriptome analysis of monocytes has revealed a systemic inflammatory signature in monocytes from nAMD patients [
13]. Monocytes from nAMD patients have been shown to express higher levels of chemokine receptor CCR1, CCR2 and CX3CR1 [
14] and the activation marker HLA-DR and phosphorylated STAT3 [
15]. We have also found that nAMD patients have higher levels of circulating neutrophils [
16].
This study was performed to investigate the cytokine/chemokine expression and secretion profile of peripheral mononuclear cells (i.e. monocytes and lymphocytes) from nAMD patients. Considering the diversity of the AMD phenotype as well as the mixed response to anti-VEGF therapy [
17], it is feasible that different immune mechanisms may be involved in different types of nAMD. Therefore, we further investigated whether the cytokine profile in nAMD patients was associated with the development of macular fibrosis, macular atrophy or the response to anti-VEGF therapy.
Methods
Study participants
The study protocol was approved by the Research Ethics Committee of Queen’s University Belfast and procedures were performed in accordance with the tenets of the Declaration of Helsinki on research into human volunteers. Participants were recruited from the macular disease clinics in Belfast (Belfast Health and Social Care Trust, UK) with written informed consent obtained from each participant. Spouses, relatives or friends who accompanied patients and who were confirmed to be without retinal disease (colour fundus photography (CFP) and optical coherence tomography (OCT)) were recruited as controls. All participants were older than 50 years of age and structured questionnaires were used to ascertain a history of medical conditions, current medication, family history of AMD, smoking habits (current, former smoker, never smoker) and body mass index (BMI). Participants with systemic inflammatory or autoimmune disorders (e.g. patients with active rheumatoid arthritis or active chronic bronchitis) and participants undergoing steroid therapy or chemotherapy were excluded from the study.
The diagnosis of nAMD was by clinical examination and confirmed by multimodal imaging consisting of fundus photography, autofluorescence, OCT, fluorescein angiography and indocyanine green angiography. In this study, most of the participants (158 of 161) were receiving anti-VEGF therapy prior to enrolment. The number of anti-VEGF (ranibizumab, trade name Lucentis, Genentech, San Francisco, CA) injections received by each patient prior to blood collection was ascertained from the medical records.
Participants with nAMD were followed up for 6 months, and we graded colour images and tomographic scans of the recruited participants at their most recent visit prior to closure of the database. Responsiveness to treatment was defined based on the participant achieving a fluid-free macula at any stage during follow-up. In addition, the status of whether a patient was fluid free at the month 3 and month 6 examinations was also recorded. Participants were classified into the following three categories: complete responder: resolution of leakage at any point in time during follow up; partial responder: exhibiting dependence on VEGF inhibitors but a fluid-free macula never achieved; and non-responder: no morphological improvement or worsening. Macular scar identification was based on both colour and OCT characteristics. On colour fundus photographs, macular scar was defined as well-delineated areas of yellowish-white tissue which on OCT corresponded to the presence of linear bands of hyperreflective material that had either obscured or replaced the normal reflectivity and banding of the neurosensory retina and RPE/Bruch's membrane complex. Macular atrophy (MA) was defined as single or multiple areas of hypopigmentation with well-defined borders and visible large choroidal vessels on CFP which corresponded to window defects on angiography and/or to the loss of cellular layers (outer retina, RPE and choriocapillaris) on the accompanying tomograms.
Participant samples were assigned randomly for experimental analysis with some samples being included in more than one analysis. Three experimental analyses were undertaken with the repository of blood samples. Experiment 1 investigated cytokines/chemokines in the serum (133 nAMD and 43 controls samples); experiment 2 investigated secreted cytokines/chemokines by peripheral blood mononuclear cells (PBMCs) (75 nAMD and 28 control samples) and experiment 3 investigated the intracellular cytokine/chemokine expression by PBMCs (28 nAMD and 27 control samples).
Serum collection
Venous blood (5 ml) was collected in tubes containing serum clot activator and centrifuged at 2000 g for 15 min within 3 h of collection. After centrifugation, the serum was aliquoted and stored at −80 °C until analysis.
PBMC isolation and culture
Whole blood collected (20–25 ml) in tubes containing ethylenediaminetetraacetic acid (EDTA) as anticoagulant between 9:00 am and 12:00 noon was processed within 3 h of collection. PBMCs were isolated by Ficoll-Paque (Histopaque 1077; Sigma-Aldrich, Gillingham, UK) density gradient centrifugation (400 g for 30 min at RT with the break turned off) and washed twice with PBS (300 g for 10 min at 4 °C). PBMCs were resuspended at 2.5 × 106 cells/ml in RPMI 1640 medium containing 10% FCS and 1% penicillin-streptomycin and cultured in 24-well plates (500 μl/well) and treated immediately with lipopolysaccharides (LPS) (2.5 μg/ml; Sigma-Aldrich) or exposure to hypoxia (1% oxygen) for 16 h. The supernatants for PBMC cultures were collected, centrifuged for 5 min at 300 g at 4 °C, and aliquoted and stored at −80 °C until analysis.
Cytometric bead array
Cytokines were measured in the serum by cytometric bead array (CBA) using CBA Flex Sets (CD121a, CD121b, MCP-1, VEGF, TGFβ1, GM-CSF and IFNα) and CBA Enhanced Sensitivity Flex Sets (IFNγ, IL-2, IL-4, IL-6, IL-8, IL-10, IL12p70, IL-17A and TNFα) (BD Biosciences, Oxford, UK) according to the manufacturer’s instructions. Cytokines in PMBC supernatant were measured using a CBA Human Th1/Th2/Th17 Cytokine Kit (IL-2, IL-4, IL-6, IL-10, TNFα, IFNγ and IL-17A) and CBA Flex Sets (CD121a, CD121b, MCP-1, VEGF, GM-CSF and IL-8) (BD Biosciences) according to the manufacturer’s instructions and as described previously [
15]. For cell culture supernatants, the total protein concentration was measured using a Pierce BCA protein assay kit (Thermo Scientific, Loughborough, UK) according to the manufacturer’s instructions. The concentrations of the cytokines were normalized to the total protein concentration (pg/mg total protein). TGF-β1 was measured as a single plex assay and serum samples were activated by incubation with 2.5 N acetic acid/8 M urea for 10 min at RT followed by neutralization with 2.7 N NaOH/1 M HEPES prior to CBA assay according to the manufacturer’s instructions.
Intracellular cytokine expression of PBMC by flow cytometry
PBMCs were stimulated for 4 h with phorbol 12-myristate 13-acetate (PMA; 100 ng/ml; Sigma-Aldrich) and ionomycin (1 μg/ml; Sigma-Aldrich) in the presence of 1× monensin (BioLegend, UK). After incubation, cells were stained with fluorochrome-labelled antibodies (anti-human CD3-FITC (BD Biosciences), CD3-PE-Cy7, IL-17A-PE, IL-4-APC, IL-6-APC (eBiosciences, San Diego, USA ), IFNγ-APC-Cy7, IL-8-PE, IL-10-Brilliant violet 421 and CD11b-APC-Cy7 (BioLegend, London, UK)).
Briefly, PBMC were washed twice with FACS buffer (300 g for 5 min at 4 °C) and resuspended at 10 × 106 cells/ml; 20 μl (2 × 105 cells) were dispensed per FACS tube and incubated with 5 μl Human TruStain FcX (Fc Receptor blocking solution; BioLegend) for 5 min at RT. The cells were then incubated with cell surface antibodies in a total volume of 100 μl FACS buffer for 30 min in the dark at 4 °C. After staining, cells were washed twice with FACS buffer and then fixed and permealized using the Foxp3 Transcription factor Staining Buffer Set (eBiosciences) according to the manufacturer’s instructions. Samples were then incubated with 5 μl Human TruStain FcX (Fc Receptor blocking solution; BioLegend) followed by incubation with intracellular cytokine antibodies in a total volume of 100 μl permealization buffer (eBiosciences) for 40 min in the dark at 4 °C. Cells were washed and acquired on the FACSCantoII flow cytometer (BD Biosciences). Data analysis was performed blindly using FlowJo software version 10.07 for Windows (Tree Star, Oregon, USA).
Gating was performed by first dividing PBMCs into CD11b
+CD3
−, CD11b
−CD3
+ and CD11b
−CD3
− cells (Fig.
2). Gates for intracellular cytokines IL-4, IL-6, IL-8, IL-10, IL-17A and IFNγ were set on total live cells based on the unstained control, and the same gates were then applied to CD11b
+CD3
−, CD11b
−CD3
+ and CD11b
−CD3
− cell subsets.
Statistical analysis
Statistical analysis was performed using the Statistical Package for the Social Sciences, Windows version 21 (SPSS Inc, Armonk, NY). Categorical demographic and clinical data were compared using Pearson’s chi-square test. The distribution of continuous variables was assessed for normality using the Kolmogorov-Smirnov test, and logarithmic transformation was performed if necessary to achieve normal distribution. Normally distributed continuous samples were then compared using the independent samples t test or one-way ANOVA. Age was not normally distributed and the difference between controls and nAMD patients was analysed using the Mann-Whitney U test.
For the associations that were significant in the univariate analysis, multinomial logistic regression was performed to adjust for age and gender. All variables were also tested for association with family history of AMD, history of cardiovascular disease, history of hypertension, history of diabetes, smoking habits, BMI, taking of cardiovascular medication, vitamins and low-dose aspirin using the independent samples t test, one-way ANOVA or Pearson’s correlation. If significant associations were identified, adjustments were made in the multinomial logistic regression analysis. Pearson’s correlation was used to assess the correlation between the number of anti-VEGF injections a patient had received prior to blood collection and the different variables analysed (e.g. cytokine serum levels, secreted cytokine levels and intracellular cytokine levels). Data in figures are presented as mean + standard error of the mean (SEM) calculated from untransformed variables even if the statistical analysis was performed on transformed variables. P values <0.05 were considered statistically significant.
Discussion
Inflammation is known to play a critical role in the development of choroidal neovascularisation in AMD although the underlying mechanism is not fully understood [
8,
19,
20]. Inflammatory cells may migrate from surrounding tissues including the neuronal retina (i.e. microglia) and choroid to the macula or be recruited from circulating immune cells [
12]. In this study, we provide evidence that PBMCs, in particular monocytes from nAMD patients, produce increased levels of IL-8, CCL2 and VEGF, especially under inflammatory conditions. Our results suggest that PBMCs in nAMD patients may contribute to the development of macular angiogenic lesions through producing excessive amounts of IL-8, CCL2 and VEGF.
IL-8 (CXCL8), an important inflammatory cytokine and potent angiogenic factor, has previously been linked to the pathogenesis of AMD. Single-nucleotide polymorphisms (SNPs) in the IL-8 regulatory and coding regions have been associated with increased risk of AMD [
21‐
23] and certain SNP genotypes have been linked to increased IL-8 serum levels [
24]. Exactly how IL-8 contributes to AMD pathology is not known. Previous studies have reported elevated levels of IL-6, IL-8 and CCL2 in the aqueous humour from nAMD patients [
25], and the levels correlated with choroidal neovascularisation (CNV) lesion size [
26]. Although retinal cells, including microglia, vascular endothelial cells [
27] and RPE cells [
28], can produce IL-8, our results suggest that infiltrating immune cells, especially CD11b
+CD3
− monocytes may contribute to increased intraocular IL-8 production in nAMD.
IL-8 is known to be involved in both physiological and pathological angiogenesis through its receptors CXCR1 and CXCR2 [
29]. The pro-angiogenic effect of IL-8 on endothelial cells is mediated partially through the induction of VEGFR2 and VEGF-A expression [
30] although a VEGF-independent pathway has also been reported [
31]. IL-8 production has been associated with resistance to anti-VEGF therapy in tumours, further supporting a VEGF-independent pro-angiogenic mechanism for IL-8 [
32]. In this study, we observed increased levels of IL-8 by PBMCs from nAMD patients who are partially resistant to anti-VEGF therapy compared to PBMCs from complete responders (although the difference was not statistically significant, possibly due to insufficient patient numbers). The results suggest that apart from VEGF, IL-8 may be another important pro-angiogenic factor in nAMD. Further studies using larger patient samples are necessary to confirm the role of IL-8 in nAMD and the response to anti-VEGF therapy.
Previously, we reported increased levels of neutrophils in nAMD patients [
16]. IL-8 is a strong chemotactic factor for neutrophils [
33]. Although how neutrophils are involved in nAMD is not known, the high levels of IL-8 by PBMC may contribute to increased levels of circulating neutrophils in nAMD.
CCL2 plays a crucial role in monocyte and lymphocyte trafficking under inflammatory conditions [
34]. Compelling evidence suggests that the CCL2/CCR2 pathway may play a role in AMD pathogenesis. The aqueous humour of nAMD patients contains high levels of CCL2 [
35], and CCL2 expression was detected in atrophic lesions in AMD patients [
36]. Monocytes from nAMD patients were shown to express higher levels of CCR1 and CCR2 [
14]. Importantly, experimental studies have shown that CCL2 can promote CNV formation through enhanced recruitment of myeloid cells [
37]. In this study, we found that PBMCs from nAMD patients produced higher levels of CCL2, although there was no difference in the serum CCL2 concentration between nAMD patients and controls (which is in line with a previous report [
14]). Our results indicate that CCL2 production by infiltrating immune cells might be related to macular lesion development in nAMD.
In a previous study, we reported increased STAT3 activation in circulating monocytes from nAMD patients [
15], which may contribute to increased VEGF production by PBMC observed in the current study. IL-6 is one of the major cytokines that can active the JAK1/STAT3 pathway. Interestingly, in this study, we found an increased percentage of IL-6 producing CD11b
+CD3
− monocytes in nAMD patients compared to that in controls. A previous study has reported an association between circulating levels of IL-6 (~6 pg/ml) and AMD progression [
38]; however, in our study, the serum levels of IL-6 were below the detection limit, i.e. 0.3 pg/ml in both patients and controls. The role of IL-6 in nAMD warrants further investigation.
The development of macular fibrosis results in irreversible vision loss in nAMD [
39], although the underlying mechanisms remain ill-defined and there is no treatment modality for it. Even following the introduction of anti-VEGF therapies, which have improved visual outcomes markedly, fibrosis remains an integral component of the macular lesion and has been shown to be associated with the poorer outcomes compared to eyes without fibrosis [
40‐
42]. In the present study, we observed increased levels of IL-8 and CCL2 in PBMCs from nAMD patients without fibrosis suggesting that IL-8 and CCL2 from PBMCs might reduce the risk. Interestingly, IL-8 and CCL2 have previously been shown to have pro-fibrotic effects in various tissues including the lung [
43,
44], kidney [
45,
46], liver [
47] and skin [
48,
49]. The tissue-specific role of CCL2 and IL-8 in fibrosis has not been reported before. Therefore, it is possible that those chemokines may have anti-fibrotic effects in the retinal microenvironment. Other chemokines such as CXCL10 have previously been shown to mediate tissue-specific pro- (e.g. in the liver) [
50] as well as anti-fibrotic effects (e.g. in the lung) [
51]. Further studies are necessary to investigate the role of CCL2 and IL-8 in fibrosis in nAMD.
Macular atrophy (MA) is due to the degeneration of RPE and overlying photoreceptors although the underlying mechanism remains poorly defined. MA and nAMD co-exist in many patients, particularly following long-term anti-VEGF therapy [
1]. In the present study, 67% of patients exhibited MA during their follow-up. PBMC from the patients who had MA produced increased levels of VEGF. Previous studies have shown that enhanced MA progression in nAMD patients is associated with anti-VEGF injections [
52]. In the present study, a numerically greater proportion of patients with MA had received more anti-VEGF injections than those without but this did not reach statistical significance (
P = 0.056). The reasons for the higher prevalence of MA in eyes undergoing anti-VEGF treatment is not entirely explicable by a higher rate of exposure to VEGF but may also reflect destruction of the macular tissues by a more active CNV lesion and hence the higher number of treatments required in this group [
41].
The strengths of the present study include independent grading of AMD, fibrosis, MA and anti-VEGF responsiveness and systematic and extensive exploration of cytokine production and secretion by PBMCs in nAMD as well as in patients with fibrosis, MA and partially or completely responding to anti-VEGF treatment.
One of the limitations of this study is that patients were recruited to the study at different stages of nAMD, and therefore, some patients classified as not having macular fibrosis or atrophy may do so over time. Secondly, 98% of patients were on anti-VEGF treatment at the time of enrolment which might have an effect on circulating immune cells although we did not find any correlation between variables analysed and the number of anti-VEGF injections received. Despite our efforts to recruit age-matched controls, there was a significant difference in age between nAMD patients and controls in one of our experimental analyses (serum analysis, experiment 1). Finally, some of the analyses were done with small patient numbers (e.g. patients without MA n = 17, partial responders n = 16, complete responders n = 10) and larger confirmatory studies are necessary to verify our results.