Pathogenic role of human C-reactive protein in diabetic retinopathy

Diabetic retinopathy (DR) is the leading cause of vision loss in working-age people in developed countries [1]. Non-proliferative DR (NPDR) presents with progressive intra-retinal leukocyte adherence, retinal neuron degeneration, capillary degeneration, and retinal vascular leakage [2,3]. Without proper treatment, DR can progress to its advanced stage, proliferative DR (PDR), characterized by pathological angiogenesis in the retina [4]. Accumulating evidence demonstrated that impairment of the neurovascular unit due to elevated oxidative stress and inflammation in the retina plays a key role in DR [5–7].

C-reactive protein (CRP) is a prototypical acute-phase protein and predominantly expressed in hepatocytes in response to infection, inflammation, or tissue injury [8]. Previous studies showed that CRP is over-produced in both type 1 and type 2 diabetic patients [9,10]. Accumulating evidence suggests that high levels of CRP are associated with the risk of diabetic microvascular complications, such as nephropathy, neuropathy, and retinopathy [11–13]. At the baseline visit of the Diabetes Control and Complications Trial (DCCT), participants with DR showed significantly higher serum CRP levels compared with those without retinopathy [11]. Furthermore, another two studies demonstrated higher serum CRP levels in patients with PDR compared with those with NPDR [12,13]. These epidemiological studies suggested that CRP can be used as a biomarker of diabetic complications. However, the underlying mechanism for the positive association between CRP and diabetic complications, especially DR, remains unknown.

It has been reported that CRP induces the generation of reactive oxygen species (ROS) and inflammation by binding with Fcγ receptors, including Fcγ receptor I (CD64) and Fcγ receptor II (CD32) [14–16]. CRP can activate the NF-κB signaling pathway, resulting in the generation of ROS and up-regulation of pro-inflammatory factors, such as tumor necrosis factor-α (TNF-α), and cell adhesion molecules, such as intercellular adhesion molecule-1 (ICAM-1) [17]. Liu et al. reported that elevated CRP levels in CRP transgenic mice promoted inflammation and fibrosis in the kidney in the context of diabetes [18]. Also, a recent study showed that CRP exerts a potent angiogenic effect on microvascular endothelial cells [19]. Furthermore, there is evidence that CRP plays a role in the apoptosis process [20]. While CRP has been indicated as the mediator of disease in non-ocular tissues, the pathogenic role of CRP in DR remains unknown.

In the present study, we tested the hypothesis that elevated levels of CRP may contribute to neurovascular impairment in DR by increasing retinal oxidative stress and inflammation. To investigate the pathogenic role of CRP in the context of DR, we used streptozotocin (STZ)-induced type 1 diabetic model in human CRP transgenic (hCRP-Tg) rats. To further investigate the role of CRP in retinal neovascularization (NV) and inflammation, we used the rat model of oxygen-induced retinopathy (OIR). We also measured the CRP-induced oxidative stress and apoptosis in immortalized rat retinal Müller cells (rMC-1) and mouse retinal precursor-like cells (661W) in vitro.

hCRP-Tg Sprague–Dawley rats were generated as described previously [21,22]. hCRP transgene was under the control of mouse albumin enhancer/promoter. The transgene construct of hCRP-Tg rats and genotyping of hCRP-Tg rats were described previously [21]. Wild-type (WT) littermate rats were used as control. All rats were housed in a specific pathogen-free (SPF) environment at 25°C under a 12-h/12-h light–dark cycle at animal facility of University of Oklahoma Health Sciences Center (OUHSC). In all procedures, rats were anesthetized with an intraperitoneal injection of ketamine–xylazine mix (K, 50 mg/kg; X, 5 mg/kg). All procedures using experimental animals were approved by the Institute Animal Care and Use Committee (IACUC) at the OUHSC, and all experiments were performed in compliance with the guidelines of the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Visual Research.

Diabetes was induced in 2-month-old hCRP-Tg rats and WT Sprague–Dawley rats by an STZ injection (55 mg/kg body weight) as described previously [23]. Levels of blood glucose were monitored 3 days after the STZ injection and measured weekly thereafter. Rats with blood glucose constantly higher than 350 mg/dl were considered diabetes [23].

OIR was generated as described previously [24]. Briefly, newborn pups were placed in an oxygen chamber with daily cycling of oxygen concentrations between 80 and 21% from the age postnatal days 0 (P0) to 13 (P13). The animals were maintained in the 12-h light and dark cycle at standard room temperature and humidity. At P14 and P18, fluorescein retinal angiography was performed to quantify the non-perfused area and NV, respectively. Another group of OIR animals were used for retinal histologic and Western blot analysis. Age-matched rats maintained at constant room air were used as normal controls.

WT and hCRP-Tg rats were anesthetized, and pupils were dilated with topical application of cyclopentolate (Akorn, Lake Forest, IL, U.S.A.). Fundus images were captured using Micron IV (Phoenix Research Laboratories, Pleasanton, CA, U.S.A.).

Electroretinography (ERG) was recorded using the Espion Visual Electrophysiology System (Diagnosys, Lowell, MA). Rats were dark-adapted for 16 h for scotopic ERG. The full ERG responses of both eyes were simultaneously recorded at the flash intensity of 600 cd.s/m², and the data in both eyes were recorded and the average of left and right eyes were used for the analysis.

The retinal thickness was measured using a spectral-domain optical coherence tomography (SD-OCT) device (Bioptigen Inc. Durham, NC) following a previously established method [23]. Images were captured with the rectangular scan at 1000 A-scans per B scan, and 200 B-scans per frame. Total retinal thickness was recorded and averaged at 3 × 3 grid positions, 500 µm from the center of the optic nerve head using the InVivo Vue driver software (Bioptigen). The researchers performing the OCT measurement were blinded to the animal group information.

Adherent leukocytes in the retinal vasculature were stained by perfusion with FITC-conjugated concanavalin-A (200 µg/ml), viewed in flat-mounted retinas, and quantified as described previously [25,26].

The eyeball was fixed in 4% paraformaldehyde in PBS solution for 3 days. The retina was dissected and digested with 5% trypsin for 1 h. Then, the retina vessels were gently isolated and stained with Periodic acid–Schiff staining system (Sigma–Aldrich, St. Louis, MO). Avascular capillaries in five random fields of each flat-mounted retinal vasculature were quantified according to a documented protocol [27].

Retinal vascular permeability was measured according to the previous publications [28,29]. Briefly, Evans Blue (30 mg/ml) was carefully injected through the femoral vein (20 mg/kg body weight). Two hours after the Evans Blue injection, the rats were perfused via the left ventricle with pre-warmed PBS solution. Immediately after the perfusion, the retinas were harvested. Evans Blue dye in the retinas was extracted with formamide. The concentrations of Evans Blue in the extracts were measured and normalized by total retinal protein concentrations.

Western blot analysis was performed as described previously [30]. Antibodies for human CRP, TNF-α, total NF-κB p65, and CD64 were purchased from Abcam (Cambridge, MA). Antibodies for phosphorylated NF-κB p65 (Ser⁵³⁶), phosphorylated IKKα/β (Ser^176/180), total IKKα, and total IKKβ were obtained from Cell Signaling (Danvers, MA). Antibodies for total β-catenin and anti-ICAM-1 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). An antibody for CD32 was obtained from R&D Systems (Minneapolis, MN). An antibody for β-actin was purchased from Sigma–Aldrich (St. Louis, MO). Details of primary and secondary antibodies used in the present study are shown in Supplementary Table S1.

For paraffin-embedded sections, rat eyeballs were enucleated and fixed in Davidson’s fixative for 48 h. The eyeballs were embedded in paraffin and 5-μm sections were cut. The deparaffinization and antigen retrieval were performed as described previously [31]. The sections were immunostained with antibodies for 3-Nitrotyrosine (3-NT, Abcam, Cambridge, MA), CD32, human CRP, and rhodopsin (1D4, Abcam, Cambridge, MA). The eyeball sections were stained with peanut agglutinin (PNA; Vector Laboratories, Burlingame, CA) for labeling cone cells. For TUNEL staining, the section slides were incubated in TUNEL reaction mixture containing the enzyme and label solution for 1 h at 37°C in the dark according to the protocols of In Situ Cell Death Kit (Roche Molecular Biochemicals, Indianapolis, IN). In addition, the retina section treated with DNase1 was used as a positive control of TUNEL staining. The slides were mounted with mounting buffer containing DAPI (blue) and then photographed with a laser scanning confocal microscope (FV1000) (Olympus, PA, U.S.A.). For the quantification, the 3-NT intensity was measured by ImageJ software (Bethesda, MD, U.S.A.) according to a documented protocol [32]. Briefly, the whole area of retinal section was selected. Then, ‘mean gray value’ was measured in the selected areas. The values were normalized to the average value of non-diabetic WT rats. For counting TUNEL-positive cells or retinal neuronal cells, cells were counted in three random fields of the central regions (from immediately adjacent to the optic nerve to approximately 1000 µm from optic nerve) of both of the superior and inferior directions in each retinal section. In each group, a minimum of six retinal sections taken from six different rats were used for analysis.

The immortalized rMC-1 were kindly provided by Dr. Vijay Sarthy at Northwestern University, and cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) containing 1 g/l glucose and 10% fetal bovine serum (FBS, Invitrogen, Carlsbad, CA) [33]. The 661W cell line was a generous gift from Dr. Muayyad R. Al-Ubaidi at University of Houston, and cultured as described previously [34]. A human retinal pigment epithelia cell line (ARPE-19) was purchased from American Type Culture Collection (ATCC) and cultured in DMEM/F12 medium containing 1 g/l glucose and 10% FBS. For hCRP treatment, hCRP concentration of 10 µg/ml was selected according to previous publications [14,35–37].

DNA fragmentation was measured in the retinal tissue and cell lines using the cell death ELISA kit (Roche Molecular Biochemicals, Indianapolis, IN) as described previously [25]. Histone-associated DNA fragments were measured at 405/490 nm absorbance. The relative readout was normalized by total protein concentration.

Cell viability was measured by Trypan Blue exclusion assay (Thermo Fisher, Grand Island, NY) as described previously [38]. Briefly, cell pellets were collected and re-suspended in PBS to reach the density of approximately 10⁵–10⁶/ml. Then, Trypan Blue dye solution was added into cell suspension at 1:1 ratio (v/v) to reach the final concentration of 0.2%. Viable and dead cells were quantified using the Cellometer (Nexcelom, Lawrence, MA).

661W cells and rMC-1 cells were starved in serum-free DMEM for 16 h. The cells were then treated with 10 µg/ml hCRP for 3 h. Untreated cells and BSA-treated cells were served as control group. For hCRP treatment, the purified native CRP from human ascites (Sigma–Aldrich, St. Louis, MO) was used. The cells were incubated with fluorescent dye 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA) (Invitrogen, Carlsbad, CA) for 30 min. Then, the DCF fluorescence signal was measured by the Wallac 1420 manager system.

To measure β-catenin transcriptional activity, luciferase-based promoter assay was performed in rMC-1 cells and ARPE-19 cells as described previously [39]. Cells were transfected with 0.30 µg TOPFLASH plasmid containing the luciferase reporter gene under the control of a promoter containing β-catenin/TCF-binding sites and 0.20 µg pRL-TK plasmid for each well of the 24-well plate using Lipofectamine 2000. Then, cells were treated with 10 µg/ml hCRP for 16 h in the presence of 20% L cell conditioned medium or Wnt3A conditioned medium. Untreated cells and BSA-treated cells were used as control of hCRP-treated cells. Then, the cells were lysed for measuring luciferase activity using a dual luciferase assay kit (Promega, Madison, WI) following the Promega’s protocol.

Experiments were performed at least three times independently for in vitro experiments, and at least six rats per group were used for animal experiments. Results were presented as the mean ± standard error of the mean (SEM). Statistical analyses were performed using a two-tailed Student’s t test for comparison of two groups, or using ANOVA followed by the Student–Newman–Keuls test when more than two groups were compared. P<0.05 was set as statistically significant.

As shown by Western blot analysis, hCRP was detected in the liver, kidney, heart, and retina of hCRP-Tg rats (Supplementary Figure S1A,B). Total serum hCRP levels in hCRP-Tg rats were 28.5 mg/l at 2 months of age, and increased to 38.7 mg/l at 5 months, and 43.9 mg/l by the end of 9 months, indicating that the transgene showed increasing expression with age (Supplementary Figure S1C). Moreover, these hCRP-Tg rats fed standard chow ad libitum showed normal levels of fasting blood glucose levels and slightly higher body weights as measured at 3 and 12 months of age compared with WT rats, suggesting that hCRP overexpression does not cause hyperglycemia in rats under normal conditions (Supplementary Figure S1D,E).

To examine retinal function, ERG analysis was performed on rats at the ages of 3, 6, and 9 months. The amplitudes of scotopic ERG response to a series of light intensity were shown in Supplementary Figure S2A,B. For recording the maximal responses of ERG, maximum light flash stimulation (600 cd.s/m²) was selected for comparison. At the age of 3 months, hCRP-Tg rats did not display any significant reduction in amplitudes of scotopic a-wave compared with WT rats (Figure 1A). At the age of 6 and 9 months, hCRP-Tg rats displayed significantly reduced amplitudes of scotopic a-wave compared with WT rats (Figure 1A). At all age points, there was a significant decrease in scotopic b-wave amplitudes in hCRP-Tg rats compared with those in age-matched WT rats (Figure 1B).

As measured by OCT, hCRP-Tg rats showed significant decreases in total retinal thickness at 3, 4, 6, and 9 months of age compared with age-matched WT rats (Figure 1C,D). The thickness of individual retinal layers was analyzed in 6-month-old rats. There was a significant decrease in the outer nuclear layer (ONL) thickness in hCRP-Tg rats compared with WT rats (Figure 1E). To test whether hCRP leads to retinal degeneration by inducing retinal apoptotic death, TUNEL staining and cell death ELISA were performed. There were increased TUNEL-positive cells primarily in the ONL of the retinas of hCRP-Tg rats compared with those in WT rats (Figure 2A). hCRP-Tg showed a four-fold increase in TUNEL-positive cells over that in WT controls (Figure 2B). Consistently, cell death ELISA showed increased apoptosis in the retinas of hCRP-Tg rats (Figure 2C).

In fundus images, hCRP-Tg rats showed a similar density and branch of retinal blood vessels compared with their age-matched WT controls (Supplementary Figure S3A). There was no significant difference between WT rats and hCRP-Tg rats in retinal vascular permeability as shown by the permeability assay using Evans Blue as a tracer (Supplementary Figure S3B). Furthermore, albumin levels in the perfused retinas of hCRP-Tg were similar to those in WT rats (Supplementary Figure S3C,D). Finally, hCRP-Tg rats demonstrated similar numbers of adherent leukocytes per flat-mounted retina compared with those in the controls as shown by leukostasis assay (Supplementary Figure S3E,F). Thus, elevated hCRP levels resulted in retinal neurodegeneration while not affecting retinal vasculature in non-diabetic conditions.

As oxidative stress and inflammation in the retina are known to play key pathogenic roles in DR, we measured immunostained 3-NT and pro-inflammatory markers in the retina of hCRP-Tg rats. Immunostaining results showed that there was an increase in 3-NT signal in the retinal section of 6-month-old hCRP-Tg rats compared with those in age-matched WT rats, suggesting that hCRP induces oxidative stress in the retina (Supplementary Figure S4A,B). As shown by Western blot analysis, there were increased levels of pro-inflammatory factor (TNF-α) and cell adhesion molecule (ICAM-1) in the retinas of 6-month-old hCRP-Tg rats compared with those in age-matched WT rats (Figure 2D,E). As hCRP is known to bind to its receptors, CD32 and CD64, to confer its pro-inflammatory effect [20], we measured CD32 and CD64 in the retina. As shown by Western blot analysis, both of CD32 and CD64 were expressed in the retinas (Figure 2D and Supplementary Figure S5A). CD32 was significantly up-regulated (Figure 2D,E), while CD64 levels showed no significant changes in the retinas of 6-month-old hCRP-Tg rats relative to those in WT rats (Supplementary Figure S5A,C), suggesting that hCRP induced overexpression of CD32 and, subsequently, oxidative stress and inflammation in the retina.

We further investigated the role of hCRP in cell death and oxidative stress in cultured retinal cells. As measured by cell death ELISA, purified hCRP increased apoptosis events by approximately two-fold in 661W and rMC-1 cells (Figure 3A). As measured by Trypan Blue exclusion assay, purified hCRP significantly decreased the viability of these cells compared with the untreated group or BSA-treated group (Figure 3B). In addition, as shown by the DCF method, hCRP also increased the production of ROS by approximately four-fold in 661W and rMC-1 cells (Figure 3C), consistent with the result of 3-NT staining in the retina of hCRP-Tg rats. These results in cultured cells suggest that the effects of hCRP on oxidative stress are through direct interactions with retinal cells.

To evaluate the impact of high levels of hCRP on retinal function under diabetic conditions, diabetes was induced in WT rats and hCRP-Tg rats at the age of 2 months. To rule out the potential impact of diabetes on the expression of hCRP, hCRP levels in retinas were measured by Western blot analysis in non-diabetic hCRP-Tg rats and diabetic hCRP-Tg rats. There was no significant difference of hCRP levels in STZ-induced diabetic hCRP-Tg rats relative to those in non-diabetic hCRP-Tg rats (Supplementary Figure S5A,B). Scotopic ERG was recorded at 1, 4, and 7 months after the onset of diabetes. At 1 month after the onset of diabetes, diabetic hCRP-Tg rats showed no significant decline in the scotopic a-wave and b-wave compared with diabetic WT rats (Figure 4A,B). However, significant declines of scotopic a-wave and b-wave were observed in diabetic hCRP-Tg rats compared with diabetic WT rats at 4 and 7 months of diabetes (Figure 4A,B). Moreover, the implicit time of scotopic a-wave and b-wave was recorded and analyzed in 6-month-old rats. There was no significant increase in implicit time of scotopic a-wave and b-wave in non-diabetic or diabetic hCRP-Tg rats compared with those in WT rats (Supplementary Figure S2C,D). To determine if progressive declines of retinal function were associated with increased retinal cell death in diabetic hCRP-Tg rats, we measured apoptosis in the retina. As measured by TUNEL, there was an increase in TUNEL-positive cells in the retina in diabetic hCRP-Tg rats compared with those in diabetic WT rats (Figure 4C). The retinas of diabetic hCRP-Tg rats showed approximately two-fold increase in TUNEL-positive cells over those in diabetic WT rats (Figure 4D). Similarly, cell death ELISA demonstrated that retinal apoptotic cells were increased by approximately 1.5-fold in diabetic hCRP-Tg rats compared with diabetic WT rats (Figure 4E). To identify the cell types that were affected in retinal degeneration, the densities of retinal photoreceptor cells were quantified. Rhodopsin and PNA were used as markers for labeling rod outer segments and cone photoreceptors, respectively [40,41]. There was a significant decrease in the intensity of rhodopsin staining in diabetic hCRP-Tg rats relative to diabetic WT rats (Supplementary Figure S7A,B). There was a significant decrease in the density of PNA-positive cells in diabetic hCRP-Tg rats vs. diabetic WT rats (Supplementary Figure S8A,B). These results indicated that hCRP promoted retinal photoreceptor degeneration.

To study whether hCRP can induce inflammation in the retina in diabetic condition, we measured the expression of inflammatory factors. Western blot analysis demonstrated a significant up-regulation of TNF-α and ICAM-1 levels in the retinas of diabetic hCRP-Tg rats compared with those of diabetic WT rats (Figure 5A–D). To address the mechanism underlying the pro-oxidative stress and pro-inflammatory effects of hCRP in diabetes, the activation of the NF-κB pathway was measured. As shown by Western blot analysis, there was a significant increase in phosphorylated IKKα (Ser¹⁸⁰) and IKKβ (Ser¹⁷⁶), as well as elevated levels of phosphorylated NF-κB p65 (Ser⁵³⁶) in the retinas of diabetic hCRP-Tg rats compared with those in diabetic WT rats (Figure 5E–G and Supplementary Figure S5A,D). There were similar levels of total IKKα, IKKβ, and total NF-κB p65 between these two groups (Figure 5G). Meanwhile, to test the hypothesis whether Wnt signaling was regulated by hCRP, the activation of Wnt signaling was measured. As shown by Western blot analysis, there was no significant increase in total β-catenin in the retinas of hCRP-Tg rats compared with those in WT rats in non-diabetic condition or in diabetic conditions (Supplementary Figure S5E,F), suggesting that the pro-inflammatory effect of hCRP was not mediated through Wnt signaling. This finding was consistent with luciferase-based promoter activity assay in vitro. hCRP-treated rMC-1 cells or ARPE-19 cells did not show increased transcriptional activities of β-catenin compared with those in untreated group (Supplementary Figure S5G,H). Furthermore, hCRP did not promote Wnt3A ligand-induced Wnt signaling (Supplementary Figure S5G,H). Overall, these results suggested that hCRP-induced oxidative stress and inflammation likely through the NF-κB signaling, rather than the Wnt pathway.

To investigate if retinal vasculature was affected by inflammation, retinal leukostasis and acellular capillary formation were analyzed in 6-month-old diabetic WT rats and diabetic hCRP-Tg rats. We found that there was a significant increase in leukocytes adherent to the retinal vessels in diabetic hCRP-Tg rats compared with those in diabetic WT rats (Figure 6A,B). There was a significant increase in acellular capillaries in diabetic hCRP-Tg rats vs. WT rats (Figure 6C,D). To see if the CRP receptor was expressed in retinal vessels, the CD32 staining was performed in retinal sections of diabetic hCRP-Tg rats. We found that intense CD32 signals in the retinal vasculature as well as in the layer of photoreceptor outer segments (Supplementary Figure S6A). These results indicated that hCRP may promote retinal vascular cell death via CD32 in diabetes.

As OIR is a well-accepted model of ischemia-induced retinal NV, we further investigated the role of hCRP in retinal NV using the OIR model. Under normoxic conditions, retinal vasculature reached the edge of the retina in both WT control rats and hCRP rats at P14, with no significant difference in the morphology, density, and distribution of retinal vasculature between genotypes at P14 (Figure 7A). In contrast, the hCRP-Tg OIR rats showed enlarged non-perfusion areas in the retina relative to WT rats with OIR, indicating that hCRP may enhance the retinal vaso-obliteration induced by hyperoxia (Figure 7B). At P18, hCRP-Tg rats with OIR demonstrated an approximate 2.5-fold increase in the NV area compared with that in WT OIR rats (Figure 7C). Meanwhile, Western blotting demonstrated that VEGF was significantly overexpressed in the retina of hCRP-Tg rats compared with WT rats, suggesting a pro-angiogenic effect of hCRP in the context of OIR (Figure 8A,B).

The retinal inflammation was measured in the OIR model. Pro-inflammatory factors, including TNF-α and ICAM-1, were up-regulated in the retinas of hCRP-Tg rats compared with those in WT rats in the OIR model (Figure 8A,B). Interestingly, different from the normoxic condition, OIR hCRP-Tg rats showed an approximate two-fold increase in leukocytes adherent to retinal vessels compared with WT rats with OIR (Figure 8C,D). Overall, these results suggested that high hCRP levels exacerbated retinal NV as well as inflammation in the context of OIR.

CRP is a well-accepted biomarker of acute inflammation and is also associated with chronic low-grade inflammation and neurodegenerative diseases [42]. Prospective clinical studies have demonstrated a positive correlation between elevated CRP levels and Type 1 and Type 2 diabetes, as well as increased risks for diabetic complications, such as DR [9–11]. However, the pathogenic role of CRP in DR has never been investigated. The present study provides the evidence that elevated hCRP levels resulted in retinal dysfunction by causing retinal neural degeneration under non-diabetic conditions. Under diabetes conditions, elevated hCRP levels exacerbated retinal dysfunction, neural degeneration, inflammation, and vascular dysfunction. hCRP also exacerbated ischemia-induced retinal NV in the OIR model. Concerning the mechanism for the role of CRP in DR pathogenesis, we found that hCRP induced over-production of ROS and overexpression of pro-inflammatory factors under non-diabetic conditions, diabetic conditions, and in the OIR model, likely through CD32/NF-κB signaling, which is responsible at least partially for hCRP-induced retinal cell apoptosis and consequent retinal dysfunction. Taken together, our study has revealed that the elevated hCRP level in diabetes is a novel pathogenic mechanism and a potential therapeutic target for DR.

The normal range of serum CRP in healthy adults varies from 0.8 to 3.0 mg/l as measured by well-accepted high-sensitivity immunoassay [43]. Laursen et al. reported that 90th percentile of serum CRP was 21.4 mg/l in type 1 diabetic patients with retinopathy [44]. Our transgenic rat model showed that mean serum hCRP levels were from 28.5 mg/l in hCRP-Tg rats aged at 3 months, suggesting that levels of hCRP in this model are physiologically relevant to diabetic patients. Moreover, STZ-induced diabetes did not change the retinal levels of hCRP in hCRP-Tg rats, indicating that our transgenic rats are an appropriate model for specifically investigating the role of hCRP in diabetes. In addition, unlike hCRP, endogenous rat CRP (rCRP) is not an acute-phase protein, which is not able to activate complement [45–47]. Thus, we applied this transgenic model with hCRP overexpression.

DR was traditionally known as a retinal vascular complication [48]. However, later studies showed that retinal neuron degeneration also plays a role in the vision loss in patients with DR [5,49]. The present study demonstrated that high levels of hCRP induced declines of ERG a-wave and b-wave, increased apoptosis in the retina, and decreased retinal thickness in non-diabetic condition, suggesting that overexpressed hCRP itself can cause retinal neurodegeneration. To further identify the direct effect of hCRP on retinal cells, we applied purified hCRP to cultured cell lines derived from retinal Müller cells and from photoreceptor cells. The results showed that hCRP induced apoptosis in these cultured retinal cells. These in vivo and in vitro results confirmed that hCRP directly acted on retinal cells and induced apoptotic cell death. Here, we showed for the first time that hCRP was implicated in pro-apoptotic effect in ocular disorders, which was consistent with previous findings in the cardiovascular system and immune system [14,20,50]. Ryu et al. described the apoptotic changes of vascular smooth muscle cells in response to hCRP [14]. In addition, Kim et al. reported that hCRP stimulated apoptosis by inducing G₂/M phase arrest in the cell cycle in monocytes [37].

To study the impact of elevated hCRP levels on retinal neurons and vasculature in diabetes, we induced diabetes in WT rats and hCRP-Tg rats. As demonstrated by TUNEL staining, most of apoptotic cells were located in the ONL, indicating hCRP accelerated death of retinal photoreceptors in diabetes. We confirmed this observation by quantifying the rod and cone photoreceptor cells. We found that hCRP promoted diabetes-induced loss of retinal photoreceptors. This observation can be explained by more intense expression of hCRP and CD32 in the layer of photoreceptor outer segments (POSs), which may mediate the direct detrimental effect in photoreceptor cells. Moreover, photoreceptors generate high levels of ROS and they are also highly vulnerable to oxidative stress [2]. hCRP in the local microenvironment may worsen the photoreceptor apoptosis induced by oxidative stress under diabetic conditions. Meanwhile, we also observed that hCRP worsened the retinal vascular inflammation and vascular cell death in diabetes. We found that hCRP overexpression resulted in more leukocytes attached to retinal vasculature and more acellular capillaries in the retina in diabetes. Given that high hCRP level in the circulation and intense expression of CRP receptor, CD32, in retinal vasculature, hCRP may accelerate diabetes-induced retinal vascular damage via CD32. These findings suggest that elevated hCRP levels in diabetes exacerbate diabetes-induced retinal inflammation, neuronal degeneration, and vascular damage.

The OIR model, one of the well-accepted models for ischemia-induced retinal NV, was used to investigate the pathogenic role of hCRP in retinal NV. In the OIR model, high levels of hCRP exacerbated ischemia-induced retinal NV. Interestingly, elevated hCRP can enlarge the non-perfusion area in the retinas of pups exposed to hyperoxic conditions, as well as promote the later pathological vessel formation in a relative hypoxic environment. This may, presumably, result from more severe vaso-obliteration in hCRP-Tg pups likely via more severe leukostasis. Consistently, we observed enhanced ischemia-induced VEGF overexpression in the retinas of hCRP-Tg pups with OIR. This pro-angiogenic effect of hCRP was consistent with a previous study, which reported that there was an increased VEGF level in the conditioned medium of CRP-treated adipose-derived stem cells, resulting in increased tube formation in human umbilical vein endothelial cells (HUVECs) [51]. However, some authors have observed that hCRP inhibited VEGF production and angiogenesis in non-ocular tissues [52,53]. The disparity regarding the role of hCRP in angiogenesis suggests that CRP may play distinct roles under different conditions.

There are three potential mechanisms for the pathogenic role of hCRP in DR. First, we found that elevated hCRP levels up-regulated the expression of pro-inflammatory factors in both physiological and pathological conditions, including diabetic condition and hypoxic condition in the OIR model. However, hCRP-Tg rats showed a significant increase in retinal leukostasis only in pathological condition, suggesting that elevated hCRP may play a deleterious role in high glucose-induced or hypoxia-induced up-regulation of pro-inflammatory cytokines in the retinas. Second, hCRP may promote diabetes-induced oxidative stress. Immunostaining of 3-NT demonstrated enhanced 3-NT signals in retinal sections of diabetic hCRP-Tg rats compared with diabetic WT rats, supporting that hCRP led to more severe oxidative stress in the retinas. Third, hCRP may contribute to DR by enhancing retinal NV. As we know, hyperglycemia and metabolic disorders in diabetes result in alterations of retinal vasculature, leading to relative hypoxic condition in the retina [54]. Among various angiogenic factors, VEGF plays a key role in the ischemia-induced NV in DR. We found that overexpression of hCRP can up-regulate the VEGF level in the retina of OIR model. The elevated VEGF level in hCRP-Tg of OIR model may occur indirectly as a result of increased vaso-obliteration. Taken together, these results suggest that the hCRP-induced inflammation, oxidative stress, and angiogenesis may contribute to the impairment of the retinal neurovascular unit under diabetic conditions.

To investigate the signaling pathway underlying hCRP-induced pathological change, we first measured retinal levels of the hCRP receptor, CD32. Previous studies have identified CD32 as the major CRP receptor in various tissues. Du Clos et al. identified CD32 as a major CRP receptor in monocytes and neutrophils based on the interaction between CRP and CD32 transfected cells with flow cytometry [20,55]. Moreover, CD32 was shown to be one of the CRP receptors in human aortic endothelial cells (HAECs) and kidney tubular epithelial cells using a neutralizing antibody to CD32 [18,56]. In addition, CRP-treated endothelial cells and the kidney of hCRP-Tg mice showed up-regulation of CD32 [18,56]. Our results first demonstrated that overexpressed hCRP was associated with the up-regulation of CD32 in the retina. This is consistent with the findings in the non-ocular tissues. Second, since the NF-κB pathway was downstream of CD32, and aberrant activation of the NF-κB pathway resulted in the up-regulation of pro-inflammatory factors, we tested whether hCRP-induced DR pathogenesis was associated with the up-regulation of this signaling pathway [57,58]. Our results demonstrated that overexpression of hCRP resulted in the activation of the NF-κB pathway, which may further enhance inflammation, ROS generation, and NV. These pathological alterations may contribute to the apoptotic cell death in the retina, resulting in retinal degeneration and impaired visual function. This notion was supported by observations reported in previous studies [18,57,59,60]. Moreover, since Wnt pathway was also a well-accepted pro-inflammatory signaling in DR [61], we also studied the cross-talk between hCRP and Wnt signaling. However, we found that hCRP did not enhance Wnt pathway activation based on both luciferase-base β-catenin transcriptional activity assay or Western blot analysis for β-catenin levels, suggesting that hCRP effect is unlikely mediated by Wnt signaling. Taken together, these results indicated that hCRP drove the inflammation and NV, at least partially, through NF-κB signaling.

Our study has several limitations. First, we only tested the hCRP role in DR pathogenesis in a type 1 diabetic model. The question of whether hCRP plays a pathogenic role in DR of type 2 diabetes warrants further investigation. Second, increased circulation levels of hCRP may induce systemic inflammatory effect, which may result in secondary pathological changes in ocular tissues. Third, hCRP may drive the inflammation through multiple signaling pathways. We cannot rule out potential signaling pathways in addition to NF-κB signaling.

In summary, our study showed for the first time that hCRP is not only a biomarker, but also a pathogenic factor in DR. hCRP impaired the retinal neurovascular unit at least partly by promoting inflammation, ROS generation, and NV in the retinas. The present study suggests that the up-regulated hCRP-CD32-NF-κB pathway may represent a new pathogenic mechanism for DR and that hCRP could be a potential therapeutic target for DR.

• Background: Elevated blood levels of hCRP are associated with DR. However, the pathogenic role of hCRP in DR has not been fully investigated.

• Summary of results: Overexpression of hCRP resulted in the up-regulation of NF-κB signaling, contributing to the retinal dysfunction and degeneration in diabetes. Moreover, elevated hCRP levels promoted the retinal NV and inflammation in the model of OIR.

• Significance: The present study suggests that the up-regulated hCRP-CD32-NF-κB pathway may represent a new pathogenic mechanism for DR and that hCRP could be a potential therapeutic target for DR.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by the National Institutes of Health (NIH) [grant numbers EY018659, EY019309, EY012231, EY028949, GM122744]; the Juvenile Diabetes Research Foundation (JDRF) [grant number 2-SRA-2019-711-S-B]; and the Oklahoma Center for the Advancement of Science and Technology (OCAST) [grant number HR16-041].

F.Q. and X.M. performed experiments, acquired and analyzed data, and wrote the manuscript. J.C., Q.C., W.W., W.L., and K.Z. conducted experiments and acquired data. Y.-H.S. and Y.W. performed the literature search, wrote the manuscript. J.-x.M. and Q. S. designed the research, analyzed data, wrote and edited the manuscript. J.-X.M. is the guarantor of this work and, as such, had full access to all the data in the present study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors would like to thank the Diabetic Animal Core facility and the Histology Core facility of Diabetes COBRE for their support and assistance in the present study.

CRP

C-reactive protein

DCF

Dichlorodihydrofluorescein

DMEM

Dulbecco’s Modified Eagle’s Medium

DR

Diabetic retinopathy

ERG

Electroretinography

FBS

Fetal bovine serum

hCRP-Tg

Human CRP transgenic

ICAM-1

Intercellular adhesion molecule-1

IKK

Inhibitor of nuclear factor kappa-B kinase

NPDR

Non-proliferative DR

NV

Neovascularization

OCT

Optical coherence tomography

OIR

Oxygen-induced retinopathy

ONL

Outer nuclear layer

PDR

Proliferative DR

PNA

Peanut agglutinin

rMC-1

Rat retinal Müller cell

ROS

Reactive oxygen species

STZ

Streptozotocin

TCF

T cell factor

TNF-α

Tumor necrosis factor-α

TUNEL

Terminal deoxynucleotidyl transferase dUTP nick end labeling

VEGF

Vascular endothelial growth factor

WT

Wild-type

3-NT

3-Nitrotyrosine