Expression of Erythropoietin and Its Receptor in the Human Retina: A comparative study of diabetic and nondiabetic subjects

Eighteen human postmortem eyes were obtained from nine diabetic donors (aged 64.1 ± 8.1 years) free of funduscopic abnormalities in the ophthalmological examinations performed during the preceding 2 years. Eighteen eye cups obtained from nine nondiabetic donors matched by age (aged 66.9 ± 6.8 years) were used as the control group. The time elapsed from death to eye enucleation was 3.4 ± 1.9 h. After enucleation, one eye of each donor was snap-frozen in liquid nitrogen and stored at −80°C until they were assayed for mRNA analysis. The other eye was fixed in 4% paraformaldehyde and embedded in paraffin for the immunohistochemical study. The clinical features of diabetic and nondiabetic donors included in the study are shown in Table 1 of the online appendix (available at http://dx.doi.org/10.2337/dc07-2075). Three patients who were being treated with diet and/or oral agents were changed to insulin after admission, and the two patients who were being treated with insulin plus metformin were treated with insulin alone. All patients died within 15 days after admission.

All ocular tissues were used in accordance with applicable laws and with the Declaration of Helsinki for research involving human tissue. In addition, this study was approved by the ethics committee of our hospital.

Human neuroretina and retinal pigment epithelium (RPE) were harvested using the microscopic dissection of isolated eye cups from donors. Total RNA was extracted from isolated retinal tissues using the RNeasy Mini Kit with DNase digestion (Quiagen Distributors, IZASA, Barcelona, Spain) according to the manufacturer's instructions. Quantification and quality of total mRNA were determined with a bioanalyzer in a LabChip (Agilent Technologies, Palo Alto, CA). One microgram of total RNA was directly used for reverse transcription, which was performed with TaqMan Reverse Transcription Reagents (Applied Biosystems, Madrid, Spain) in a 50-μl reaction according to the manufacturer's instructions. Quantitative real-time PCR was performed using 60 ng of the reverse transcription reaction as a template with TaqMan Universal MasterMix (Applied Biosystems). All reactions were conducted as follows: 95°C for 10 min and 50 cycles of 15 s at 95°C and 1 min at 60°C in an Applied Biosystems 7000 system. Each sample was assayed in duplicate and control water samples were included in each experiment. Automatic relative quantification data were obtained in an ABI Prism 7000 (SDS software; Applied Biosystems) using β-actin gene as the endogenous expression reference gene (Hs9999903_m1; Applied Biosystems). TaqMan premade gene expression assays (Applied Biosystems) were used to amplify human Epo (Hs00171267_m1, GenBank accession number NM_000799.2) and EpoR (Hs00181092_m1 Exon Boundary 6–7, GenBank accession number NM_000121.2).

HIF-1α and HIF-2α mRNA transcripts were studied in the same samples with the TaqMan premade gene expression assays (Hs00153153_m1 and Hs01026138_m1, respectively; Applied Biosystems).

Glial fibrillar acidic protein (GFAP) mRNA was assessed in neuroretinas using a TaqMan premade gene expression assay (Hs00157674_m1; Applied Biosystems).

Paraffinized eyes were serially cut at 7-μm thickness. Sections were deparaffinized with xylene and rehydrated in ethanol. Sections were placed in antigen-retrieval solution (Dako A/S, Glostrup, Denmark) at 95°C. To eliminate the autofluorescence of RPE due to melanin and lipofuscin we used a method described elsewhere (15). Sections were then incubated for 1 h with 1% BSA in 0.3% Triton X-100 in PBS to block unspecific binding of the antibodies and then incubated overnight at 4°C with a specific primary antibody to human Epo (1:100, N-19; Santa Cruz Biotechnology, Heidelberg, Germany) and EpoR (1:200, M-20; Santa Cruz Biotechnology). After washing, sections were incubated with Alexa Fluor 488 and 594 secondary antibodies, respectively (Molecular Probes, Eugene, OR), at room temperature for 1 h. Slides were cover-slipped with a drop of mounting medium containing 4,6-diamidino-2-phenylindole for visualization of cell nuclei (Vector Laboratories, Burlingame, CA).

Images were acquired with a confocal laser scanning microscope (FV1000; Olympus, Hamburg, Germany), using a 488-nm laser line for Epo and a 594-nm laser for EpoR. Each image was saved at a resolution of 1,024 × 1,024 pixel image size.

To quantify Epo and EpoR immunofluorescence in RPE and the neuroretina, the total fluorescence intensity values corresponding to 10 field frame images (×40, numerical aperture 0.9) of each retina sample were measured. These results were then normalized, taking into account the area analyzed. All these calculations were made using specific software (FluoView ASW 1.4; Olympus).

Tissue sections were incubated overnight at 4°C with the primary antibody, anti-human GFAP (1:200; Sigma, Madrid, Spain). After washing, sections were incubated with Alexa Fluor 488 (Molecular Probes) secondary antibody for 1 h. GFAP immunofluorescence in the neuroretina was quantified using a laser confocal scanning microscope. The procedure was the same as that mentioned above for Epo and EpoR.

Epo was assessed by radioimmunoassay (Incstar/DiaSorin). The lowest limit of detection was 4.4 mU/ml. EpoR was measured by ELISA (R&D Systems, Minneapolis, MN). The lowest limit of detection was 30 pg/ml.

Vitreal proteins were measured by a previously validated microturbidimetric method with an autoanalyzer (Hitachi 917; Boehringer Mannheim). This method, based on the benzethonium chloride reaction, is a highly specific method for the detection of proteins and has a higher sensibility and reproducibility than the classic method of Lowry. The lowest level of proteins detected was 0.02 mg/ml. Intra- and interassay coefficients of variation were 2.9 and 3.7%, respectively.

Student's t test was used to compare the variables. Correlations were examined by Spearman's rank correlation. Levels of statistical significance were set at P < 0.05.

β-Actin mRNA expression was similar in both the RPE and the neuroretina (NS). In addition, no differences were observed in β-actin mRNA expression between diabetic and nondiabetic retinas (NS). Thus, we have calculated mRNA gene expression after normalization with β-actin.

Significantly higher Epo mRNA expression was observed in RPE than in the neuroretina in the whole group (1.13 ± 1.01 vs. 0.26 ± 0.31; P = 0.003). Higher expression was also observed in RPE than in the neuroretina when diabetic donors and nondiabetic donors were analyzed separately (1.63 ± 1.27 vs. 0.43 ± 0.41; P = 0.03 and 0.68 ± 0.38 vs. 0.13 ± 0.08; P = 0.002). In addition, a higher expression of Epo mRNA was detected in the retina from diabetic donors compared with nondiabetic donors (Fig. 1A).

A significantly higher level of mRNA EpoR was observed in RPE than in the neuroretina in the whole group (1.48 ± 1.24 vs. 0.37 ± 0.41; P = 0.002). Higher expression was also detected in RPE than in the neuroretina when diabetic donors and nondiabetic donors were analyzed separately (1.62 ± 1.16 vs. 0.33 ± 0.24; P = 0.01 and 1.35 ± 1.36 vs. 0.41 ± 0.55; P = 0.07). However, we did not find any difference between EpoR mRNA levels in diabetic and nondiabetic donors in either RPE or the neuroretina (Fig. 1B). In the neuroretina but not in RPE, a direct correlation between EPO and EpoR mRNA was detected in diabetic (r = 0.73; P = 0.04) and nondiabetic donors (r = 0.90; P = 0.001)

HIF-1α and HIF-2α mRNA expressions were similar for diabetic and nondiabetic donors in both RPE (HIF-1α 0.44 ± 0.26 vs. 0.42 ± 0.37; P = 0.89 and HIF-2α 0.43 ± 0.26 vs. 0.48 ± 0.41; P = 0.82) and the neuroretina (HIF-1α 0.46 ± 0.24 vs. 0.36 ± 0.31; P = 0.36 and HIF-2α 0.29 ± 0.16 vs. 0.24 ± 0.21; P = 0.58). HIF-2α mRNA correlated with Epo mRNA in diabetic donors (r = 0.62; P = 0.03) but not in nondiabetic donors (r = 0.26; P = 0.33).

We did not find any significant differences in Epo and EpoR mRNA levels between patients who were changed to insulin treatment after admission and patients who continued to receive oral treatment.

Laser scanning confocal images of Epo and EpoR immunofluorescence are displayed in Figs. 2 and 3. Epo immunofluorescence intensity was higher in diabetic donors than in nondiabetic donors in both RPE (5,770 ± 1,491 vs. 3,933 ± 1,530; P = 0.05) and in the neuroretina (5,644 ± 3,671 vs. 3,671 ± 1,401; P = 0.06) (Fig. 2).

EpoR immunofluorescence was higher in RPE than in the neuroretina in both diabetic donors (8,883 ± 2,171 vs. 4,734 ± 2,474; P < 0.001) and nondiabetic donors (9,079 ± 6,260 vs. 4,763 ± 3,680; P = 0.07). However, no differences in EpoR were detected between diabetic and nondiabetic donors either in the RPE or in the neuroretina (Fig. 3).

A direct correlation was observed between Epo and EpoR in diabetic donors (RPE: r = 0.87; P = 0.002; neuroretina: r = 0.86; P = 0.01) and nondiabetic donors (RPE: r = 0.94; P = 0.005; neuroretina: r = 0.24; P = 0.5).

GFAP mRNA expression was higher in the neuroretina from diabetic donors than from nondiabetic donors (0.40 ± 0.21 vs. 0.13 ± 0.14; P = 0.04). Confocal microscopy revealed a higher immunofluorescence intensity in the neuroretinas from diabetic donors (10148 ± 3,909 vs. 6,798 ± 1,544; P = 0.04) (Fig. 4), thus confirming that neuroglial activation does exist.

Intravitreal Epo concentration was higher in diabetic donors than in nondiabetic control subjects (124 ± 92 vs. 36 mU/ml ± 13; P = 0.03). However, no differences in intravitreal EpoR concentration were detected between diabetic donors and nondiabetic control subjects (99 ± 58 vs. 96 ± 86 pg/ml; P = 0.95). Because intravitreal proteins were similar in diabetic and in nondiabetic donors, the results were not influenced by this confounding factor.

In the present study we have found a higher expression of Epo (both mRNA and protein) in the retinas from diabetic donors without clinically detectable diabetic retinopathy in comparison with the retinas from nondiabetic donors. In addition, intravitreal Epo concentrations were significantly higher in diabetic donors than in nondiabetic donors. Therefore, our results suggest that Epo overexpression is an early event in the retina of diabetic patients. It is important to note that in the retinas from diabetic donors, reactive changes in Müller cells such as upregulation of GFAP were present. Therefore, although the retinas from diabetic donors were without overt vascular abnormalities, they were actually already being damaged by the diabetic milieu. In fact, glial activation and neurodegeneration have been described as early events in the pathogenesis of diabetic retinopathy (16).

The complete mechanisms that regulate Epo expression in the human retina remain to be elucidated. The hypoxia-induced upregulation of Epo expression is modulated at the transcriptional level by HIF and in particular by HIF-2α (17). In the present study, no differences between HIF-1α and HIF-2α were detected between diabetic and nondiabetic donors. These findings suggest not only that Epo overexpression is an early event in the retina of diabetic patients but also that at this stage it is unrelated to a hypoxic-ischemic stimulus. In support of this concept, it should be stressed that EpoR expression, which is upregulated by hypoxia in rat brains (18) and in rat retinas (8), was similar in diabetic and in nondiabetic retinas. It should also be noted that EpoR expression has not been described previously in human retina.

Apart from hypoxia, other factors could regulate Epo expression. Watanabe et al. (11) observed an increase in the vitreous Epo levels in patients with inflammatory eye diseases. Given that inflammation has been involved in the pathogenesis of diabetic retinopathy (19), it might be a contributing factor to the high levels of Epo observed in diabetic patients. Hyperglycemia could be another factor that induces Epo production. Although there are no studies evaluating the effect of glucose on Epo expression in retinal cells, a direct relationship has been shown between glucose and Epo concentrations in a Chinese hamster ovary cell line (20).

A reduction in Epo catabolism could also contribute to the higher Epo levels detected in the retina and the vitreous fluid from diabetic donors. In this regard, the glycosylation of Epo reduces its affinity for EpoR (21). Because Epo is degraded only by EpoR-expressing cells and their receptor binding determines the rate of intracellular degradation (22), it is possible that a higher degree of Epo glycosylation is associates with lower clearance of Epo.

The mechanisms causing the increase in Epo in the diabetic retina cannot be identified in the present study. However, there are many reasons for thinking that in these early stages of diabetic retinopathy Epo might have beneficial rather than pathogenic actions. First, it has been demonstrated using an in vitro model of the bovine blood-brain barrier (BBB) that Epo protects against the vascular endothelial growth factor–induced permeability of the BBB and restores the tight junction proteins (23). Epo treatment also prevents an increase in BBB permeability in a rat model of induced seizures (24). Because the blood-retinal barrier is structurally and functionally similar to the BBB, it is possible that Epo could act as an antipermeability factor in the retina. In fact, Epo was able to improve diabetic macular edema when it was administered for treatment of anemia in diabetic patients with renal failure (25). Second, there is growing evidence that Epo is a neurotrophic factor not only in the brain but also in the retina (4,5). Third, Epo exerts an anti-inflammatory effect on the brain (26), and this action might also be extrapolated to the diabetic retina. Finally, Epo is a potent physiological stimulus for the mobilization of endothelial progenitor cells (27), and, therefore, it could play a relevant role in regulating the traffic of circulating endothelial progenitor cells toward injured retinal sites.

We conclude that Epo overexpression is an early event in the retina of diabetic patients and that at this stage it is not mediated by HIF. This finding suggests that other factors beside hypoxia might be crucial in the upregulation of Epo in the initial stages of diabetic retinopathy. EpoR is also expressed in the retina from diabetic patients, but it is similar in nondiabetic subjects. Further studies are needed to investigate not only the precise regulation of Epo/EpoR in the retina in physiological and pathological conditions but also its potential role in the therapeutic armamentarium.

This study was supported by grants from the Instituto de Salud Carlos III (CIBERDEM), the Ministerio de Ciencia y Tecnología (SAF2006-05284), and the Fundación para la Diabetes and the Generalitat de Catalunya (2005SGR-00030).