Purpose We recently demonstrated that M2 macrophages were involved in the development of fibrovascular membranes (FVM) associated with proliferative diabetic retinopathy (PDR) possibly through the induction of periostin. The purpose of this study was to determine whether macrophage colony-stimulating factor (M-CSF) and interleukin (IL)-13, inducers of the M2 polarisation of macrophages from monocytes, are elevated in the vitreous of patients with PDR, and whether M2-polarised macrophages induce periostin production.
Methods We measured the levels of M-CSF, granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-4, IL-13, soluble (s)CD163, periostin and vascular endothelial growth factor by sandwich ELISA in vitreous samples collected from 61 eyes of 47 patients with PDR, and 39 eyes of 36 patients with non-diabetic ocular diseases (control group). Human monocytes were polarised in vitro with GM-CSF, interferon-γ, and lipopolysaccharide for M1 macrophages, and M-CSF, IL-4, and IL-13 for M2 macrophages. Quantitative real-time PCR was used to determine the mRNA level of periostin.
Results The concentrations of M-CSF and IL-13 in the vitreous were significantly higher in patients with PDR than in non-diabetic controls (p<0.0001). There was a strong positive correlation between the vitreous concentrations of M-CSF and sCD163 and periostin. The mean vitreous level of IL-13 was significantly higher in eyes with FVMs than in those without FVMs (epicentre only). In vitro studies showed that M2-polarlised macrophages significantly increased the expression of the mRNA of periostin.
Conclusions These findings indicate that the M2 polarisation of macrophages is induced by M-CSF and IL-13 in diabetic retinas. The presence of M-CSF and IL-13 would then promote FVM formation by periostin production.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Diabetic retinopathy (DR) is one of the leading causes of blindness in the working-age population worldwide.1 Retinal neovascularisation occurs at the advanced stage of DR leading to proliferative DR (PDR). Blindness can result from abnormal fibrovascular membrane (FVM) formation with subsequent intravitreal haemorrhage and tractional retinal detachment.2 Despite recent advances in vitrectomy techniques, use of retinal photocoagulation and intravitreal antivascular endothelial growth factor (VEGF), the prognosis for patients with DR is poor especially in those with PDR.3
FVMs usually contain different types of cells, such as macrophages/monocytes, hyalocytes, laminocytes, fibroblasts, retinal glial cells and vascular endothelial cells.4 Among these cells, the macrophages have a wide variety of biological functions.5 ,6 We have demonstrated that macrophage-attracting chemokines, for example, CCL2, CCL3 and CCL4, were involved in retinal neovascularisation through the recruiting of macrophages in a mouse model of oxygen-induced retinal neovascularisation.6–8
Accumulating evidence has suggested that macrophages consist of at least two subgroups, classically activated M1 and alternatively activated M2.9 ,10 The M1 macrophages are proinflammatory and play a critical role in driving inflammation, and the M2 macrophages are associated with debris scavenging, angiogenesis and fibrosis.
We have also shown that there was an increase in the expression of CD163 in the vitreous and FVMs obtained from patients with PDR.11 CD163 is a M2 macrophage marker and has a close relationship with periostin, a matricellular protein that plays a role in tissue remodelling. The increased expression of CD163 indicated that the M2 macrophages probably play a role in the formation of FVMs. However, the molecular mechanisms that regulate the polarisation of M2 macrophages in patients with PDR have not been determined.
Granulocyte-macrophage colony-stimulating factor (GM-CSF) and macrophage colony-stimulating factor (M-CSF) induce monocyte-macrophage lineage differentiation both in vivo and in vitro.9 ,10 The GM-CSF, or M-CSF-differentiated macrophages can be further polarised to more specific cell types in response to additional stimuli. For example, when GM-CSF-differentiated human M1-like macrophages are exposed to T helper (Th1) cytokines, such as interferon-γ (IFNγ), they are polarised into more activated M1 macrophages and express the M1 cell-surface marker CD80.12 By contrast, when M-CSF-differentiated human M2-like macrophages are activated by Th2 cytokines, such as interleukin (IL)-4 and/or IL-13, they are polarised into more activated M2 macrophages and express the M2 cell-surface marker CD163.9
Based on our earlier findings and those reported by others, we hypothesised that the elevated M-CSF in the retinas of patients with DR will induce M2-like macrophage differentiation from recruited monocytes in diabetic retinas. The M-CSF-differentiated macrophages will be further polarised by IL-4 or IL-13 in the retina and/or vitreous cavity, leading to an accumulation of activated M2 macrophages that produce periostin. The periostin will then induce the formation of FVMs associated with PDR.
To test this hypothesis, we determined the intravitreal concentrations of M-CSF, GM-CSF, IL-4 and IL-13 in patients with PDR. Additionally, we examined whether there was an in vitro production of periostin by IL-13-treated and M-CSF-treated M2 macrophages.
Subjects and methods
The procedures used in this study were approved by the ethics committees of the Kyushu University Hospital and Fukuoka University Chikushi Hospital, and the surgical specimens were handled in accordance with the Declaration of Helsinki. All patients were informed on the purpose of this study and gave informed consent to undergo the surgery and to use the vitreous samples obtained during the vitrectomy. The inclusion and exclusion criteria for participation in this study are shown in online supplementary table S1.
At the beginning of vitrectomy, samples of undiluted vitreous fluid (0.5–1.0 mL) were aspirated under standardised conditions and were immediately transferred to sterile tubes.13 The samples were centrifuged for 10 min at 3000 rpm (1630 g) at 4°C, and the supernatants were divided into aliquots and stored at −70°C until analysis. Vitreous samples were collected from 61 eyes of 47 patients (age, 61.9±7.6 years; 27 men, 20 women) with PDR during the initial pars plana vitrectomy. For non-diabetic controls, vitreous samples were collected from 39 eyes of 36 patients (age, 64.4±6.2 years; 12 men, 24 women) who were undergoing vitrectomy for epiretinal membrane (ERM) removal or macular hole (MH) closure. None of the controls had diabetes mellitus. The clinical characteristics of these patients are presented as an online supplementary table S2.
The M-CSF, sCD163 and VEGF concentrations were measured with human M-CSF, sCD163 and VEGF immunoassay kits (R&D Systems, Minneapolis, Minnesota, USA), respectively, according to the manufacturer's instructions.13 The concentrations of GM-CSF, IL-4 and IL-13 were also measured using a human Cytometric Bead Array Kit (BD Biosciences, San Jose, California, USA). The periostin concentration was measured with a sandwich ELISA kit as described in detail.14
Human CD14+ monocytes were purchased from Lonza (2W-400C; Walkersville, Maryland, USA). The cells were cultured in Dulbecco's modified Eagle's medium with 2 mM l-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin and 10% heat-inactivated fetal bovine serum. To generate monocyte-derived macrophages, the monocytes were treated for 6 days with human recombinant GM-CSF (215-GM; 10 ng/mL; R&D Systems) for M1-like generation or M-CSF (216-MC; 25 ng/mL; R&D Systems) for M2-like generation. The M1-like cells were activated for 24 h with IFNγ (285-IF; 20 ng/mL; R&D Systems) and lipopolysaccharide (LPS) (L2637; 100 ng/mL; Sigma Aldrich, St Louis, Missouri, USA), and the M2-like cells were activated for 24 h with IL-13 (213-ILB; 20 ng/mL; R&D Systems).
RNA isolation and real-time quantitative reverse transcriptase PCR (qRT-PCR)
We isolated RNA from the samples as described in detail.15 Briefly, 500 ng of isolated total RNA were converted into cDNA with a First Strand cDNA Synthesis Kit for RT-PCR (Roche Applied Science, Penzburg, Germany), according to the manufacturer's protocols. Real-time qRT-PCR for periostin, GAPDH, CD80 and CD163 was performed as described.6–8 The primer sequences were:
5′-GCTATTCTGACGCCTCAAAACT-3′ and 5′-AGCCTCATTACT CGGTGCAAA-3′; for GAPDH,
5′-GAGTCAACGGATTTGGTCGT-3′ and 5′-CTTGATTTTGG AGGGATCTCGC-3′; for CD80,
5′-AAACTCGCATCTACTGGCAAA-3′ and 5′- GGTTCTTGTA CTCGGGCCATA-3′; for CD163,
5′-TTTGTCAACTTGAGTCCCTTCAC-3′ and 5′- TCCCGCTA CACTTGTTTTCAC-3′.
Statistical analyses were performed using a commercial statistical software package (JMP, V.10.0; SAS Institute, Cary, North Carolina, USA). First, the data were examined by Shapiro–Wilk tests to determine the normality of the distribution. Data that were not normally distributed were analysed by non-parametric statistics. The significance of the differences in the M-CSF and IL-13 levels among the different groups was determined by the Mann–Whitney U test. To determine whether a significant correlation existed among M-CSF, sCD163, IL-13 and VEGF, Spearman's rank correlation tests were used. A two-tailed test with p<0.05 was considered statistically significant.
Vitreous concentrations of M-CSF, GM-CSF, IL-4 and IL-13
To validate the association of M2 macrophages with PDR, we measured the concentration of M-CSF in the vitreous fluids obtained from patients with PDR and with non-diabetic ocular diseases (figure 1). The concentration of M-CSF in the vitreous was significantly higher in patients with PDR (1782.73±253.60 ng/mL) than in the non-diabetic control patients (432.66±72.01 ng/mL; p<0.0001, figure 1A). The mean vitreous level of IL-13 was significantly higher in patients with PDR (132.46±33.77 fg/mL) than in patients with non-diabetic controls (23.59±13.76 fg/mL; p<0.0358, figure 1B). The concentrations of both M-CSF and IL-13 in the vitreous were not significantly different among the non-diabetic control patients with MH and ERM. By contrast, the concentrations of IL-4 and GM-CSF were below the level of detection in all the same samples from patients with PDR and non-diabetic controls (data not shown).
Correlations of M-CSF, sCD163, periostin, IL-13 and VEGF concentrations
We have demonstrated that M2 macrophages were probably involved in the development of FVMs associated with PDR possibly through the induction of periostin.11 Because the concentrations of both M-CSF and IL-13 in the vitreous were significantly higher in patients with PDR, we next determined whether there was a significant correlation between the concentrations of M-CSF and IL-13, sCD163 and periostin. Our results showed that there was a highly significant correlation between the vitreous concentrations of M-CSF and sCD163 (r=0.679, p<0.0001; Spearman correlation; figure 2A), and with periostin (r=0.696, p<0.0001; Spearman correlation; figure 2C) in the 61 eyes with PDR. The correlation between the vitreous concentration of M-CSF and IL-13 was not significant (p=0.492; figure 2B).
Because we have demonstrated that the mean concentration of VEGF in the vitreous was significantly higher in patients with PDR than in non-diabetic controls,13 we also determined whether there was a significant correlation between M-CSF and VEGF. A moderate correlation was found between M-CSF and VEGF in the same patients with PDR (r=0.426, p=0.0006; Spearman correlation coefficient: figure 2D). There was no significant correlation between the vitreous concentrations of M-CSF, IL-13, sCD163, periostin and VEGF in the eyes of non-diabetic control patients (data not shown).
Correlation of IL-13 concentrations in eyes with or without FVMs
We have demonstrated that the mean vitreous levels of sCD163 and periostin were significantly higher in eyes with FVMs than in eyes without FVMs (epicentre only) among the patients with PDR.14 Therefore, we asked whether M-CSF and IL-13 will also be higher in eyes with FVMs. We first subdivided patients with PDR into those with FVMs and those without FVMs (epicentre only). Of the 61 eyes with PDR, the mean vitreous level of M-CSF was 1917.38±319.94 ng/mL in the 47 eyes with FVMs and 1330.69±238.65 ng/mL in the 14 eyes without FVMs. This difference was not statistically significant (p=0.864; figure 3A). However, the mean vitreous level of IL-13 was significantly higher (183.16±48.39 fg/mL) in the same 47 eyes with FVMs than that in the 14 eyes without FVMs (30.77±30.77 fg/mL p=0.0485; figure 3B).
Induction of periostin production by M2-polarised macrophages in vitro
Because there was a highly significant correlation between the vitreous concentrations of M-CSF, CD163 and periostin in the eyes with PDR, and because IL-13 was significantly higher in the eyes with FVMs, we next investigated whether M2-polarised macrophages induced the production of periostin in vitro. Cultured monocytes were differentiated into macrophages using either M-CSF or GM-CSF and subsequently polarised by exposure to IFN-γ+LPS (M1) or IL-13 (M2). It was recently reported that M1 macrophages typically appeared round, morphologically, whereas M2 macrophages appeared stretched and spindle-like.16 Consistent with this, the majority of GM-CSF-treated macrophages appeared round, while the M-CSF-treated macrophage cultures contained a higher number of spindle-like cells, indicating that each group of macrophages is morphologically distinctive M1 or M2, respectively (figure 4A).
We next assessed whether the macrophages expressed the specific cell-surface macrophage markers, CD80 (M1) or CD163 (M2), after polarisation with IFN-γ+LPS of GM-CSF-treated M1-like macrophages, or with IL-13 of M-CSF-treated M2-like macrophages, respectively (figure 4B, C). Our results showed that GM-CSF, M-CSF and IL-13 treated macrophages expressed CD80 weakly, but IFN-γ+LPS significantly upregulated the CD80 expression (figure 4B). By contrast, the M-CSF-treated macrophages induced a higher upregulation of CD163 expression compared with resting macrophages. Addition of IL-13 also significantly induced CD163 expression although to a lesser degree than that by M-CSF-treated macrophages. Although we observed an increase of CD163 under M1-like conditions in GM-CSF-treated cells, the increase was not significant as seen in M-CSF-treated macrophages.
Taken together, these findings indicated that macrophages can be differentiated to M1-like or M2-like macrophages with either GM-CSF or M-CSF, respectively, and can be further polarised to M1 macrophages by IFN-γ+LPS or to M2 macrophages by IL-13, as judged by the mutually distinctive expression of CD80 or CD163 in vitro.
Therefore, we next analysed the effect of these different polarisation procedures on periostin production. Macrophages differentiated in the presence of M-CSF did not increase the expression of the mRNA of periostin compared with resting macrophages. However, the addition of IL-13 to M-CSF-treated macrophages significantly increased periostin mRNA levels by approximately 10-fold (figure 5). By contrast, periostin was not induced by M1-polarised cells treated with GM-CSF alone or in combination with IFN-γ+LPS.
Accumulating evidence suggests that M-CSF plays a role in macrophage-associated diabetic complications including DR.17 ,18 Consistent with these findings, our results showed that the concentration of M-CSF, but not GM-CSF, was significantly higher in the vitreous of patients with PDR than in control patients (figure 1). An early upregulation of M-CSF signalling of neurons, microglia and glia in the retinas of diabetic rats has been reported,19 indicating that a higher concentration of M-CSF in the vitreous of patients with PDR is derived from those cells in the diabetic human retina. Additionally, the concentration of M-CSF and sCD163 in the vitreous of patients with PDR was significantly correlated (figure 2). Recently, we found that CD163-positive M2 macrophages were clustered adjacent to neovascular tufts in a mouse model of oxygen-induced retinopathy (Zhou and Yoshida, manuscript in submission). Along with the predominance of M-CSF over GM-CSF in the vitreous of eyes with PDR in this study, these findings indicate that diabetic retinas are a M2 macrophage-dominant microenvironment.
The concentration of IL-13 was significantly higher in the vitreous of patients with PDR than in control patients, but IL-4 was barely detectable (figure 1). IL-13 shares many functional activities with IL-4, because both cytokines exploit the same IL-4Rα/Stat6 signalling pathways.20 However, recent studies have shown a dominant role for IL-13 in the pathogenesis of several fibrotic diseases including pulmonary fibrosis, asthma and systemic sclerosis.21 In parallel with these findings, the concentration of IL-13 was significantly correlated with the presence of FVMs (figure 3) indicating that IL-13 is closely associated with fibro(vascular) membrane formation in PDR.
We recently reported an increased expression of periostin in the vitreous and ERMs obtained from patients with PDR and proliferative vitreoretinopathy.14 ,22 ,23 In the present study, we showed a higher correlation between the vitreous levels of M-CSF, sCD163 and periostin in eyes with PDR. Moreover, the treatment of M-CSF-differentiated human macrophages by IL-13 resulted in a marked induction of CD163 and periostin with very little induction of CD80 (figures 4 and 5). Together with our previous studies showing the colocalisation of CD163 and periostin in eyes with FVMs,11 these results indicate that the recruited monocytes in diabetic retina may differentiate into M2-like macrophages by M-CSF and further polarised to activated M2 macrophages which promote FVM formation via periostin production.
Together with the results of our previous studies and those of others, we suggest the following possible mechanisms for the macrophage-mediated FVM formation in eyes with PDR. First, retinal ischaemia would induce an upregulation in the expression of the CCL2, CCL3, and CCL4 genes which attract monocytes to the diabetic retina.6–8 Second, M-CSF released from diabetic retina transforms the recruited monocytes into M2-like macrophages. Third, the IL-13 released by the Th2 cells in the retina further polarises to activated M2 macrophages. And fourth, the polarised M2 macrophages produce periostin that promotes retinal neovascularisation and fibrosis.9 In parallel, the ischaemia also enhances the production of VEGF by retinal glial cells and retinal vascular endothelial cells.24 These processes are likely to be important in promoting macrophage-related FVM formation in diabetic retinas. These mechanisms further support the idea that ischaemia-associated retinal neovascularisation may be closely related to chronic inflammation.25
In summary, our results show that there is an increased expression of M-CSF and IL-13 in the vitreous of patients with PDR and the induction of periostin by M2 macrophages. These findings further support our earlier suggestion that M2 macrophages play important roles in the formation of FVMs associated with PDR. Therefore, targeting M2 macrophages may be a novel therapeutic option for preventing FVM formation associated with PDR.
Contributors SY was responsible for study concept and design, data collection and analysis and writing the manuscript. YK, TN, YZ, KI and KI contributed to data analysis. RA, SN, MM, YS, YO and TK contributed to data collection. TI contributed to, reviewed the manuscript and approved publishing of the manuscript.
Funding Masayo Eto provided excellent technical help. This work was supported in part by JSPS KAKENHI Grant Numbers 26293374, 24249083 and 26670757 and Takeda Science Foundation.
Competing interests None.
Patient consent Obtained.
Ethics approval The Ethics Committees of the Kyushu University Hospital and Fukuoka University Chikushi Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.