CD40 Upregulation in the Retina of Patients With Diabetic Retinopathy: Association With TRAF2/TRAF6 Upregulation and Inflammatory Molecule Expression

Abstract

Purpose: CD40 is upregulated in the retinas of diabetic mice, drives pro-inflammatory molecule expression, and promotes diabetic retinopathy. The role of CD40 in diabetic retinopathy in humans is unknown. Upregulation of CD40 and its downstream signaling molecules TNF receptor associated factors (TRAFs) is a key feature of CD40-driven inflammatory disorders. We examined the expression of CD40, TRAF2, and TRAF6 as well as pro-inflammatory molecules in retinas from patients with diabetic retinopathy.

Methods: Posterior poles from patients with diabetic retinopathy and non-diabetic controls were stained with antibodies against von Willebrand factor (labels endothelial cells), cellular retinaldehyde-binding protein (CRALBP), or vimentin (both label Müller cells) plus antibodies against CD40, TRAF2, TRAF6, ICAM-1, CCL2, TNF-α, and/or phospho-Tyr783 phospholipase Cγ1 (PLCγ1). Sections were analyzed by confocal microscopy.

Results: CD40 expression was increased in endothelial and Müller cells from patients with diabetic retinopathy. CD40 was co-expressed with ICAM-1 in endothelial cells and with CCL2 in Müller cells. TNF-α was detected in retinal cells from these patients, but these cells lacked endothelial/Müller cell markers. CD40 in Müller cells from patients with diabetic retinopathy co-expressed activated phospholipase Cγ1, a molecule that induces TNF-α expression in myeloid cells in mice. CD40 upregulation in endothelial cells and Müller cells from patients with diabetic retinopathy was accompanied by TRAF2 and TRAF6 upregulation.

Conclusions: CD40, TRAF2, and TRAF6 are upregulated in patients with diabetic retinopathy. CD40 associates with expression of pro-inflammatory molecules. These findings suggest that CD40-TRAF signaling may promote pro-inflammatory responses in the retinas of patients with diabetic retinopathy.

It is estimated that there were 451 million patients with diabetes worldwide in 2017, and the prevalence of diabetes will continue to increase to 693 million patients by 2040.¹ Approximately 35% of these patients develop diabetic retinopathy,² a complication that is the most common cause of blindness in working-age adults in developed countries.³ In the early nonproliferative stages of diabetic retinopathy, patients exhibit retinal changes that include microaneurysms and capillary degeneration, leading to areas of nonperfusion.⁴ In proliferative diabetic retinopathy, the subsequent stage of the disease, the ischemia-driven expression of angiogenic factors, such as vascular endothelial growth factor (VEGF) leads to retinal neovascularization.⁵ Furthermore, patients with proliferative diabetic retinopathy can also develop a fibrovascular epiretinal membrane that can result in retinal detachment and visual loss. 

Whereas the pathogenesis of diabetic retinopathy is multifactorial, low-grade chronic inflammation plays an important role in the development of this disease.^6–8 Increased retinal expression of inflammatory molecules, including ICAM-1, TNF-α, and CCL2, is associated with diabetic retinopathy in humans.^9–14 Moreover, experimental evidence in animals revealed that low-grade chronic inflammation contributes to the development of diabetic retinopathy.^6–8 Upregulation of ICAM-1 in retinal endothelial cells promotes leukocyte adherence, and the disruption of ICAM-1 – CD18 interaction in diabetic mice reduces the development of degenerate capillaries.¹⁵ Retinal microglia/macrophages in diabetic mice express TNF-α,¹⁶ a cytokine that promotes capillary degeneration.¹⁷ Moreover, administration of an anti-TNF-α agent reduced leukocyte adhesion in the retina and breakdown of the blood-retinal barrier in diabetic rats.¹⁸ CCL2 promotes vascular permeability and monocyte/macrophage recruitment into the retina of diabetic mice.¹⁹ This is further supported by experimental evidence that administration of a dual inhibitor of CCR2 (receptor for CCL2) and CCR5 reduced both phenomena in the retina of diabetic mice.²⁰ 

Therapeutic approaches for diabetic retinopathy include intravitreal administration of anti-VEGF agents, laser photocoagulation, and steroids.^21,22 However, a large proportion of patients have an inadequate response to anti-VEGF agents, and the current treatment options have significant side effects.^23–25 As such, the development of novel approaches to the treatment of diabetic retinopathy is a critical area of research. Identification of the upstream events that trigger inflammatory responses in diabetic retinopathy may lead to development of such novel approaches. 

CD40 is a central driver of retinal inflammation and the development of retinopathy in diabetic mice.²⁶ CD40 is a member of the TNF receptor superfamily that is upregulated in retinal endothelial cells, Müller cells, and microglia/macrophages in mice with experimental diabetic retinopathy.²⁶ Diabetic CD40^−/− mice do not upregulate ICAM-1, TNF-α, IL-1β, and CCL2 in the retina and do not develop diabetic retinopathy.^26–28 Studies in transgenic CD40^−/− mice with rescue of CD40 restricted to Müller cells or endothelial cells revealed that expression of CD40 isolated to these cells is sufficient to drive inflammatory responses in diabetic mice.^28–30 Diabetic mice that express CD40 restricted to endothelial cells exhibit ICAM-1 upregulation in these cells and develop leukostasis.²⁹ This finding is consistent with the demonstration that ligation of CD40 triggers upregulation of ICAM-1 in retinal endothelial cells.²⁷ Diabetic mice with expression of CD40 restricted to Müller cells upregulate CCL2 in these cells^28,30 in agreement with the ability of CD40 stimulation to induce CCL2 production in retinal Müller cells.²⁷ Importantly, the presence of CD40 in Müller cells also triggers pro-inflammatory molecule expression in bystander microglia/macrophages.^28,30 CD40 ligation in Müller cells induces phospholipase C γ1 (PLCγ1)-dependent release of extracellular ATP that engages the purinergic receptor P2X₇ expressed in bystander microglia/macrophages, enabling these cells to secrete TNF-α and IL-1β.²⁸ Moreover, expression of CD40 restricted to Müller cells in diabetic mice is sufficient for development of early diabetic retinopathy in mice.^28,30 

CD40 functions by recruiting TNF receptor associated factors (TRAFs), of which TRAF2 and TRAF6 are major mediators of the effects of CD40.³¹ Indeed, disruption of CD40-TRAF2 or CD40-TRAF6 signaling is sufficient to markedly inhibit in vitro pro-inflammatory responses in retinal cells.²⁷ Moreover, in contrast to diabetic CD40^−/− mice rescued to express wild type (WT) CD40 in Müller cells, diabetic mice rescued with CD40 that cannot recruit TRAF2 or TRAF6 do not upregulate ICAM-1, TNF-α, IL-1β, CCL2, or P2X₇ in the retina.³⁰ Disruption of CD40-TRAF2 signaling in Müller cells prevents the development of diabetic retinopathy.³⁰ Importantly, intravitreal administration of a cell-permeable peptide that blocks CD40-TRAF2 signaling markedly impaired upregulation of ICAM-1, TNF-α, IL-1β, CCL2, and P2X₇ in the retina as well as reduced retinal leukostasis in diabetic B6 mice.³⁰ These findings support that CD40 is a therapeutic target against diabetic retinopathy. 

The levels of CD40 expression are low under basal conditions.^32,33 However, induction or upregulation of CD40 expression is a key feature of inflammatory disorders driven by CD40.^33,34 Furthermore, increased expression of TRAF appears to correlate with the TRAF pathways that mediate pro-inflammatory responses induced by CD40.³⁵ Thus, identification of the cell types that upregulate CD40 and TRAFs in diabetic retinopathy can provide an indication of the cells in which CD40-TRAF signaling is likely activated. 

Whereas studies in mice support the central role of CD40 in the development of diabetic retinopathy, little is known about the relevance of retinal CD40 in patients with this disease. The studies herein were conducted to determine whether the expression of CD40, TRAF2, and/or TRAF6 are increased in the retinas of patients with diabetic retinopathy and whether CD40 co-localizes with pro-inflammatory molecules in the retinas of these patients. 

Materials and Methods

Human Subjects

Eyes from eight subjects with documented diabetic retinopathy and three non-diabetic control individuals (1 eye per donor) were obtained postmortem through Eversight (Cleveland, OH, USA). Subjects did not have any other known retinal disease. The stage of diabetic retinopathy was determined by reviewing available clinical information. Four diabetic subjects carried a diagnosis of proliferative diabetic retinopathy. No information about the stage of the disease was available in the remaining four subjects. Eyes were fixed in 4% paraformaldehyde within 16 hours after death. Anterior segments and the vitreous were removed from the eye cups. Posterior poles were maintained in paraformaldehyde for more than 24 hours. The use of human material was in accordance with the Declaration of Helsinki on the use of human material for research. 

Immunohistochemistry

Tissues were placed in 30% sucrose followed by embedding in OCT in a mold. Tissues were flash frozen with liquid nitrogen and stored at −80°C. Frozen tissues were sectioned (10 µm) on a Leica cryostat, and the sections were mounted on SuperPlus Slides. Sections of peripheral retinas were incubated with the antibodies listed in Table 1. Fluorescent secondary antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA, USA). Retinas were analyzed blindly using Olympus FV1200 IX-83 confocal microscope (Oberkochen, Germany). Images were obtained at a total magnification of 400 times. Microscope settings were kept constant for all samples. Images were processed in Photoshop CC 19.1.1. using similar linear adjustments for all samples. Semiquantitative assessment of CD40, TRAF2, and TRAF6 expression in retinal endothelial cells and Müller cells was performed using MetaMorph (Nashville, TN, USA). Briefly, images were thresholded to identify von Willebrand factor⁺ (endothelial cells), CRALBP⁺, and vimentin⁺ (Müller cells) areas. Pixel intensity for CD40, TRAF2, and TRAF6 in the selected areas were measured. 

Table 1.

 

Antibodies Used for Immunofluorescence Microscopy

 

Cells

The human Müller cell line MIO M1 (gift from Dr. Gloria Limb; University College London, London, England, UK) transduced with a retroviral vector that encodes human CD40 or an empty vector were described previously.^26,28 Primary human retinal endothelial cells were obtained from Cell Systems (Kirkland, WA, USA). In certain experiments, cells were treated with multimeric human CD154 (CD40 ligand; gift from Dr. Richard Kornbluth, Multimeric Biotherapeutics Inc., La Jolla, CA, USA) or a nonfunctional CD154 mutant (T147N) as control.¹² The human monocytic cell line Monomac6 cells (gift from Rene de Waal Malefyt, DNAX Research Institute, Palo Alto, CA, USA) were incubated with or without IFN- γ (100 IU/mL; PeproTech, Rocky Hill, NJ, USA) plus LPS (100 ng/mL; Sigma Aldrich, St. Louis, MO, USA). MonoMac6 cells were stained with anti-TNF-α antibody in the presence or absence of a specific blocking peptide (NBP1-19532PEP; Novus, Littleton, CO, USA). Cells were transfected with non-targeting siRNA or siRNA against human TRAF2 or TRAF6 obtained from Horizon Discovery (Cambridge, UK). Cells were also transfected with siRNA against ICAM-1 (AGCGGAAGAUCAAGAAAUA) or CCL2 (CCAUGGACCACCUGGACAA). Transfections were performed using 50 nM of siRNA and TransIT-X2 (Mirus, Madison, WI, USA). 

Immunoblot

Membranes were probed with antibodies to TRAF2, TRAF6, ICAM-1, CCL2 (see Table 1), or actin (sc47778; Santa Cruz Biotechnologies, Santa Cruz, CA, USA), followed by incubation with secondary antibody conjugated to horseradish peroxidase (Santa Cruz Biotechnologies). 

Statistics

Results were expressed as the mean ± SEM. Statistical significance was analyzed by ANOVA. Although multiple measurements were obtained in each sample, statistical comparison among patient samples was done using the mean values for these measurements for each patient. Differences were considered statistically significant at P < 0.05. 

Results

Expression of CD40 is Increased in Retinal Endothelial Cells and Müller Cells From Subjects With Diabetic Retinopathy

CD40 can be expressed in a broad range of cells that include retinal endothelial cells, Müller cells, and microglia/macrophages.^26,36 CD40 is also detected in various neurons,³⁷ including ganglion cells in the retina.³⁶ Whereas CD40 is expressed at low levels in the retina of normal, non-diabetic mice,^26,36 diabetes causes upregulation of CD40 in retinal Müller cells and endothelial cells in mice.²⁶ Expression of CD40 restricted to either Müller cells or endothelial cells is sufficient to induce inflammatory responses in the retina of diabetic mice.^28–30 Thus, we centered on those cells for our studies in human retinas. We evaluated posterior poles from eight subjects with diabetic retinopathy, four of whom had a history of proliferative diabetic retinopathy. There was no information about the stage of disease in the remaining four subjects and therefore are labelled as simply having diabetic retinopathy. We examined three non-diabetic subjects without a history of retinal disease as controls. Table 2 summarizes the characteristics of the individuals examined. Retinas from these individuals were examined by immunohistochemistry using antibodies that were validated as shown in Supplementary Figure S1. In this regard, anti-CD40 antibodies only stained Müller cells that expressed CD40 but not CD40⁻ Müller cells; staining with antibodies against TRAF2, TRAF6, ICAM-1, or CCL2 was ablated in cells made deficient in the molecules by transfection with siRNA; and staining with anti-TNF-α antibody was markedly inhibited by incubation with specific blocking peptide (see Supplementary Fig. S1). Control subjects exhibited little immunoreactivity for CD40 in the retina consistent with the low expression of this molecule under basal conditions (Fig. 1). In contrast, CD40 staining was more intense in the retinas from patients with diabetic retinopathy (see Fig. 1). No staining was observed in retinal sections where primary antibodies were omitted (secondary antibodies alone; Supplementary Fig. S2). CD40 expression in patients with diabetic retinopathy was somewhat widespread suggesting expression of CD40 in various retinal cells. Retinal sections were incubated with antibodies against von Willebrand factor to identify retinal endothelial cells and antibodies against CRALBP or Vimentin to identify Müller cells. Retinas from control subjects exhibited little immunoreactivity for CD40 in areas that co-expressed von Willebrand factor (retinal endothelial cells; see Fig. 1A). In contrast, CD40 staining was more intense in areas that co-expressed this endothelial cell marker in retinas from patients with diabetic retinopathy (see Fig. 1A). Similarly, compared to retinas from control subjects, there were areas of more intense CD40 staining that co-expressed the Müller cell markers CRALBP or Vimentin in retinas from patients with diabetic retinopathy (see Fig. 1B). These results indicate that CD40 is upregulated in endothelial cells and Müller cells from patients with diabetic retinopathy. 

Table 2.

 

Summary of Clinical Data

 

Figure 1.

 

CD40 is upregulated in retinal endothelial cells and Müller cells in patients with diabetic retinopathy. Posterior poles from patients with diabetic retinopathy (DR, no available information on disease stage), proliferative diabetic retinopathy (PDR), and non-diabetic controls were incubated with anti-CD40 mAb plus antibodies against either: (A) von Willebrand factor (vWF, marker of endothelial cells); (B) CRALBP; (C) Vimentin (both expressed in Müller cells). Arrowheads show some of the areas where von Willebrand factor, CRALBP, or Vimentin co-express with CD40. Original magnification times 400. Scale bar, 20 µm (A) or 50 µm (B, C). Graphs show pixel intensity for CD40 (arbitrary units [AUs]) in von Willebrand factor⁺, CRALBP⁺, and Vimentin⁺ cells. Six to eight fields per subject were analyzed. Statistical comparison among patient samples was done using the mean values for these measurements. **P < 0.01 by ANOVA.

CD40 was Co-Expressed With ICAM-1 and CCL2 in Retinas of Subjects With Diabetic Retinopathy

We examined whether CD40 in retinal cells from subjects with diabetic retinopathy was co-expressed with pro-inflammatory molecules key to the development of diabetic retinopathy: ICAM-1, CCL2, and TNF-α. As reported,¹⁰ retinal endothelial cells from patients with diabetic retinopathy exhibited increased expression of ICAM-1 (Fig. 2A). Co-staining with anti-CD40 mAb revealed that ICAM-1 was co-expressed with CD40 in retinal endothelial cells from patients with diabetic retinopathy (see Fig. 2A). Compared to Müller cells from control subjects, Müller cells from patients with diabetic retinopathy expressed CCL2 (Fig. 2B). Retinas from these patients revealed areas where this chemokine was co-expressed with CD40 (see Fig. 2B). In addition, we examined expression of TNF-α in retinal endothelial cells and Müller cells. TNF-α was not detected in retinal endothelial cells or Müller cells from control subjects or patients with diabetic retinopathy (Figs. 3A, 3B). Rather, TNF-α was detected in cells that lacked expression of von Willebrand factor and vimentin (see Fig. 3). Commercial antibodies against markers of microglia/macrophages that could be used in combination with the anti-TNF-α antibody were not adequate for identification of microglia/macrophages in the human retinas. Altogether, CD40 was co-expressed with ICAM-1 in retinal endothelial cells and with CCL2 in Müller cells in patients with diabetic retinopathy, whereas these cells did not appear to express TNF-α. 

Figure 2.

 

Co-expression of CD40 and pro-inflammatory molecules in the retinas of patients with diabetic retinopathy. Posterior poles from patients with DR, PDR, and non-diabetic controls were incubated with anti-CD40 mAb with either anti-von Willebrand factor plus anti-ICAM-1 antibodies (A) or anti-CRALBP plus anti-CCL2 antibodies (B). Arrowheads show some of the areas where von Willebrand factor or CRALBP co-express with CD40 and ICAM-1 or CCL2. Original magnification times 400. Scale bar, 20 µm (A) or 50 µm (B).

Figure 3.

 

TNF-α does not appear to be expressed in endothelial cells and Müller cells from patients with diabetic retinopathy. Posterior poles from patients with DR, PDR, and non-diabetic controls were incubated with anti-TNF-α antibody plus either anti-von Willebrand factor (A) or anti-vimentin antibody (B). TNF-α-expressing cells are shown within boxes. Set of images including von Willebrand factor alone or vimentin alone indicate that TNF-α did not associate with von Willebrand factor or vimentin expression. Original magnification times 400. Scale bar, 20 µm (A) or 50 µm (B).

Müller Cells From Patients With Diabetic Retinopathy Expressed Activated PLCγ1

Relevant to the current studies, we previously reported that Müller cells from diabetic mice do not express TNF-α by immunohistochemistry, although this cytokine was detected in retinal microglia/macrophages.²⁸ Importantly, CD40 present in Müller cells induced protein expression of TNF-α in microglia/macrophages through CD40-driven secretion of ATP by Müller cells and purinergic-dependent production of TNF-α by microglia/macrophages.²⁸ CD40-dependent expression of activated PLCγ1 in Müller cells is the upstream event that triggers this purinergic pathway in diabetic mice.²⁸ Thus, we examined whether Müller cells from patients with diabetic retinopathy express activated PLCγ1 that is associated with CD40 expression. Indeed, patients with diabetic retinopathy exhibited detectable immunostaining for phosphor-Tyr783 PLCγ1 (marker of PLCγ1 activation) in Müller cells that co-expressed CD40, whereas those from control subjects did not (Fig. 4). 

Figure. 4.

 

Patients with diabetic retinopathy expressed activated PLC γ 1 in Müller cells. Posterior poles from patients with DR, PDR, and non-diabetic controls were incubated with anti-phospho-Tyr783 PLCγ1 (p-PLC) plus anti-vimentin and anti-CD40 antibodies. Arrowheads show some of the areas where phospho-Tyr783 PLCγ1 is co-expressed with vimentin and CD40. Original magnification times 400. Scale bar, 50 µm.

Retinal Endothelial Cells and Müller Cells From Patients With Diabetic Retinopathy Exhibit Increased Expression of TRAF2 and TRAF6

Increased in vivo expression of not only CD40 but also TRAFs is associated with CD40-driven inflammation.³⁵ Control subjects exhibited low immunoreactivity for TRAF2 and TRAF6 in retinal endothelial cells (Fig. 5) and Müller cells (Fig. 6). In contrast, retinal endothelial cells and Müller cells from patients with diabetic retinopathy displayed more intense staining for TRAF2 and TRAF6, and these proteins were co-expressed with CD40 (see Figs. 5, 6). 

Figure. 5.

 

Patients with diabetic retinopathy exhibit increased expression of TRAF2 and TRAF6 in retinal endothelial cells. Posterior poles from patients with DR, PDR, and non-diabetic controls were incubated with anti-TRAF2 or anti-TRAF6 Abs plus anti-von Willebrand factor antibody. Arrowheads show areas where von Willebrand factor is co-expressed with CD40 and TRAF2 or TRAF6. Original magnification times 400. Scale bar, 20 µm. Graphs show pixel intensity (AU) for TRAF 2 or TRAF6 in von Willebrand factor⁺ cells. Six to eight fields per subject were analyzed. Statistical comparison among patient samples was done using the mean values for these measurements. **P < 0.01 by ANOVA.

Figure. 6.

 

Patients with diabetic retinopathy exhibit increased expression of TRAF2 and TRAF6 in Müller cells. Posterior poles from patients with DR, PDR, and non-diabetic controls were incubated with anti-TRAF2 or anti-TRAF6 Abs plus anti-vimentin antibody. Arrowheads show some of the areas where vimentin is co-expressed with CD40 and TRAF2 or TRAF6. Graphs show signal intensity for TRAF 2 or TRAF6 in vimentin⁺ cells. Original magnification times 400. Scale bar, 50 µm. Graphs show pixel intensity (AU) for TRAF 2 or TRAF6 in vimentin⁺ cells. Six to eight fields per subject were analyzed. Statistical comparison among patient samples was done using the mean values for these measurements. ***P < 0.001 by ANOVA.

Discussion

We report that the expression of CD40 in retinal endothelial cells and Müller cells was increased in patients with diabetic retinopathy (see Fig. 7 for graphic summary). These findings are significant because modest upregulation of CD40 in these cells is known to enhance CD40-driven pro-inflammatory molecule expression.²⁶ Indeed, CD40 was co-expressed with ICAM-1 in endothelial cells and CCL2 in Müller cells from patients with diabetic retinopathy. In addition, these patients expressed activated PLCγ1 in Müller cells. This molecule is a key component of the purinergic pathway that drives expression of pro-inflammatory molecules in the retina of diabetic mice.²⁸ Finally, retinal endothelial cells and Müller cells from patients with diabetic retinopathy exhibited increased expression of TRAF2 and TRAF6, major mediators of CD40 signaling. These findings suggest that the CD40-TRAF pathway plays an important role in the development of diabetic retinopathy in humans because: (i) upregulation of CD40 is a key feature of CD40-driven inflammatory disorders.^33,34 (ii) CD40 co-localized with ICAM-1 and CCL2, pro-inflammatory molecules that promote diabetic retinopathy and are known to be induced in vitro by CD40 ligation in human cells.²⁷ (iii) activation of PLCγ1 is the upstream event that drives purinergic-dependent production of pro-inflammatory cytokines, a pathway that CD40 activates in human cells and that promotes diabetic retinopathy in vivo in mice.²⁸ (iv) TRAF upregulation is associated with CD40-TRAF signaling in vivo,³⁵ and CD40-TRAF2 as well as CD40-TRAF6 signaling are required for pro-inflammatory molecule upregulation in the retina of diabetic mice.³⁰ 

Figure 7.

 

Model of the changes in expression of CD40, TRAF2, TRAF6, and inflammatory molecules in diabetic retinopathy in humans. Expression of CD40 and its downstream signaling molecules TRAF2 and TRAF6 are increased in retinal Müller cells and endothelial cells from patients with diabetic retinopathy. This is accompanied by increased expression of ICAM-1 in retinal endothelial cells, upregulation of CCL2 in Müller cells, and expression of activated PLCγ1 in Müller cells. TNF-α does not appear to be expressed in Müller or endothelial cells whereas it is detected in cells that likely represent microglia/macrophages. Prior in vitro studies revealed that CD40-TRAF signaling upregulates ICAM-1 in retinal endothelial cells and CCL2 production by Müller cells (Ref. 27), as well as activates PLCγ1 in Müller cells, triggering release of extracellular ATP that in turn promotes TNF-α production by myeloid cells (Ref. 28). Created with BioRender.

Relevant to our studies, CD40 expression is increased in renal tubules and infiltrating cells in the kidneys of patients with diabetic nephropathy.³⁸ In addition, peripheral blood mononuclear cells from patients with poorly controlled type I diabetes exhibit increased mRNA levels of the functional type I isoform of CD40.³⁹ Outside of diabetes mellitus, CD40 is upregulated in various disorders, including inflammatory bowel disease, systemic lupus erythematosus, rheumatoid arthritis, multiple sclerosis, graft rejection, and atherosclerosis.^{33,34,40–45} CD40 promotes inflammation in these diseases, a phenomenon that is likely potentiated by CD40 upregulation because increased levels of CD40 markedly enhances CD154-driven inflammatory responses in vitro.²⁶ The co-expression of CD40 with pro-inflammatory molecules known to be directly upregulated by CD40 suggests that CD40 promotes inflammatory responses in the retina of patients with diabetic retinopathy. 

The expression of CD154 (CD40 ligand) is also altered in diabetes. In addition to membrane CD154 present on activated CD4⁺ T cells and platelets, there is also a biologically active soluble form of CD154 present in circulation.⁴⁶ The levels of soluble CD154 are elevated in the blood from patients with diabetes, particularly those with microangiopathy.^47–49 The increased levels of circulating plasma CD154 positively correlated with the presence of diabetic retinopathy and the severity of disease.⁵⁰ Importantly, CD154 upregulation is of functional relevance because serum CD154 from patients with diabetes induces pro-inflammatory responses in endothelial cells and monocytes.⁴⁹ A local increase in CD154 expression may also take place in the retina of diabetics. Infiltrating T cells in the retina and vitreous, including activated CD4⁺ T cells, have been reported in the retinas of patients with diabetic retinopathy.^51–53 Finally, the microthrombosis that occurs in diabetic retinopathy may also contribute to increased retinal levels of CD154 given that activated platelets express this molecule.⁵⁴ 

The expression of TRAFs is modulated during CD40-TRAF signaling. Although CD40 ligation can cause initial TRAF degradation,⁵⁵ studies performed 4 to 16 hours after addition of CD154 showed that CD40 ligation upregulates TRAFs.³⁵ Furthermore, the expression of TRAF2 and TRAF6 is increased in vivo in CD40-driven inflammatory disorders, including atherosclerosis, arterial injury, inflammatory bowel disease, and lupus nephritis.^{34,35,56–59} Studies in a mouse model of arterial injury revealed that TRAF upregulation was dependent on the presence of CD40.⁵⁷ Thus, the observation that retinal cells from patients with diabetic retinopathy not only upregulate CD40 but also TRAF2 and TRAF6 suggests that CD40-TRAF signaling is activated in this disease in humans. Finally, given that TRAF2 and TRAF6 signaling mediate CD40-induced pro-inflammatory molecule upregulation in retinal cells,²⁷ the co-expression of CD40 with pro-inflammatory molecules raises the possibility that CD40-TRAF promotes pro-inflammatory molecule expression in patients with diabetic retinopathy. Of potential relevance, increased expression of TRAF2 correlated with the risk for relapse in patients with inflammatory bowel disease, a CD40-driven disorder.⁵⁸ 

In summary, we report increased retinal expression of CD40 and its downstream signaling molecules TRAF2 and TRAF6 in patients with diabetic retinopathy as well as co-expression of CD40 with pro-inflammatory molecules key to the development of diabetic retinopathy. These findings are important because they may have therapeutic implications. Inhibition of CD40-TRAF signaling has been studied as a novel approach to control CD40-diven inflammatory disorders.^60,61 The upregulation of CD40 and TRAF2 in different retinal cell types and the key role of CD40-TRAF2 signaling as a driver of pro-inflammatory responses in numerous types of cells would support that blocking this pathway would effectively reduce retinal inflammation in diabetes. Indeed, intravitreal administration of a cell-permeable CD40-TRAF2,3 blocking peptide to mice with diabetic retinopathy ablated upregulation of a broad range of pro-inflammatory molecules as well as recruitment of leukocytes to the retina.³⁰ Our study raises the possibility that pharmacologic inhibition of CD40-TRAF2,3 signaling may become a novel approach to treat diabetic retinopathy in humans. Given that retinal pigment epithelial cells have basal expression of CD40,⁶² future studies should examine whether CD40 expression is also increased in these cells in the diabetic retina. In addition, the in vivo role of CD40 in VEGF upregulation should be examined because both Müller cells and retinal pigment epithelial cells are important sources of VEGF in the diabetic retina. 

Acknowledgments

The authors thank Alyssa Hubal for siRNA design, Catherine Doller for expert processing of tissue for histopathology, and Scott Howell for assistance with image analysis. 

Supported by National Institutes of Health (NIH) Grant EY019250 (C.S.S.) and NIH Grant P30 EY11373. 

Disclosure: S. Vos, None; R. Aaron, None; M. Weng, None; J. Daw, None; E. Rodriguez-Rivera, None; C.S. Subauste, None