High Levels of Serum Prolactin Protect Against Diabetic Retinopathy by Increasing Ocular Vasoinhibins

The Institutional Review Board of the Hospital “Dr. Luis Sánchez Bulnes” approved the protocol for blood sample collection. All subjects were recruited in this hospital, provided written informed consent before collection of samples, and were treated in accord with the tenets of the Declaration of Helsinki. The cohort consisted of 181 male mestizo patients with type 1 or type 2 diabetes and an estimated mean time from disease onset of 13.6 ± 0.7 years. Forty healthy male, mestizo volunteers without diabetes served as control subjects. The mean age of the diabetic patients was 61.3 ± 1.2 years, and that of the control group was 57.0 ± 1.7 years. All participants underwent clinical evaluation and a comprehensive ophthalmologic examination including visual field testing, intraocular pressure evaluation, slit-lamp biomicroscopy, and indirect ophthalmoscopy. Also, fluorescein angiography was performed in all patients with DR. Patients with diabetes were under glycemic control and receiving either insulin or oral antidiabetic agents and an appropriate diet. Exclusion criteria included the treatment with common medications causing hyperprolactinemia (major tranquilizers and antipsychotics [chlorpromazine, haloperidol, risperidone, and amisulpride], prokinetics [metoclopramide and domperidone], or antihypertensive drugs [α-methyldopa, reserpine, and verapamil]), and having renal dysfunction (serum creatinine >1.5 mg/dl). Based on the international clinical DR severity scale (8), patients with diabetes were classified as having no DR (NDR) (n = 37), nonproliferative DR (NPDR) (n = 92), or proliferative DR (PDR) (n = 72). Table 1 summarizes the demographics and attributes of all patients studied. Blood samples were obtained by venipuncture from a peripheral vein after overnight fasting and prior to any treatment. Serum was collected and kept at −70°C until use.

Male Wistar rats (250–300 g) and PRL receptor–null and age-matched wild-type mice (9 weeks, 129SvJ background) were maintained and treated in accord with the guidelines of the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. The Bioethics Committee of the Institute of Neurobiology from the National University of Mexico (UNAM) approved all animal experiments. Hyperprolactinemia was induced by the implantation of two anterior pituitary glands (APs) under the kidney capsule as previously described (9), and sham rats were subjected to similar surgery without implantation. Ten days after surgery, AP-implanted and nonimplanted rats received a daily, intraperitoneal (i.p.) injection of vehicle or bromocriptine (5 mg/kg body wt); 5 days later, all animals were killed and the sera and retinas collected to evaluate PRL and vasoinhibin levels, respectively. In other experiments, rats with or without AP implants for 15 days were anesthetized with 70% ketamine and 30% xylazine (1 μl/g body wt, i.p.) and injected intravitreously with PBS or 9 μmol/l recombinant human VEGF (rhVEGF₁₆₅; gift from Genentech, South San Francisco, CA) as described (5). After 24 h, retinal vasopermeability was evaluated by fluoroangiography and the Evans blue method. Finally, to induce a diabetic state, rats with or without AP implants for 15 days were injected or not with a single i.p. dose of streptozotocin (60 mg/kg in 10 mmol/l citrate buffer, pH 4.5) (Sigma-Aldrich, St. Louis, MO) after overnight fasting. Rats with a blood glucose concentration ≥250 mg/dl were considered diabetic. After 75 days, animals were killed to determine the levels of serum PRL and retinal vasoinhibins or to evaluate the level of retinal vasopermeability by the Evans blue method. Some groups of diabetic and nondiabetic rats, with or without AP implants, were injected daily with bromocriptine 5 days before being killed.

Serum PRL was measured in human subjects by a commercial ELISA kit (Genzyme Diagnostics, San Carlos, CA) or in rats and mice by the Nb2 cell bioassay, a standard procedure based on the proliferative response of the Nb2 lymphoma cells to PRL (10).

Pools of rat or mouse retinas (four or six, respectively) were homogenized in 0.5% Nonidet P-40, 0.1% SDS, 50 mmol/l Tris, 150 mmol/l NaCl, 100 μg/ml phenylmethylsulnonyl fluoride, and 1 μg/ml aprotinin (pH 7) and centrifuged (9,600g for 10 min). Supernatant protein (50 μg) was resolved by 15% SDS-polyacrylamide gels under reducing conditions, transferred to nitrocellulose membranes, and probed with 1:500 monoclonal antibody INN-1 to rat PRL that reacts with the NH₂-terminus of PRL (6), or with 1:500 antiserum to mouse PRL (AFP-131078; National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD). Secondary antibodies conjugated to alkaline phosphatase (Bio-Rad Laboratories, Hercules, CA) were used. Optical density values were determined using the Quantity One, 1-D analysis software (Bio-Rad, Hercules, CA).

Rat eyes were fixed in 10% formalin and embedded in paraffin. Twelve-micrometer paraffin sections were then processed for immunohistochemistry using 1:100 monoclonal anti-PRL receptor U-5 antibody (Novus Biologicals, Littleton, CO) and an avidin-biotin-peroxidase reaction (Vector Laboratories, Burlingame, CA).

Fluorescein isothiocyanate (FITC) angiography was carried out as reported (11). Blood vessel network was delineated and its area quantified using the Image Pro-Plus software (Media Cybernetics, Silver Spring, MD).

Retinal vasopermeability was measured using a modification of the Evans blue assay as previously described (5).

Data are given as means ± SEM. The unpaired Student t test was used for all comparisons except for data with multiple groups using ANOVA.

All diabetic patients showed higher levels of PRL than the control subjects (32.8 ± 4.9 vs. 16.3 ± 1.7 ng/ml, P < 0.005). However, patients with PDR had a reduced concentration of PRL (26.7 ± 2.7 ng/ml, P < 0.05) compared with diabetic patients without retinopathy (34.1 ± 3.6 ng/ml) (Fig. 1,A). The pattern of circulating PRL levels (higher in diabetic patients with no retinopathy than in those with PDR) did not depend on the type of diabetes (Fig. 1,B) or on other systemic complications associated with diabetes (hypertension and nephropathy). An inverse relationship between PRL levels and the progression of DR was observed when comparing only individuals having prehypertension (≥120/80 mmHg) or hypertension (≥140/90 mmHg) (Fig. 1,C) and when a subset of all diabetic patients were matched for diabetes duration, glycemia, glycosylated hemoglobin, cholesterol, creatinine, and blood pressure (Fig. 1 D).

Serum PRL is incorporated into the eye and converted to vasoinhibins. To investigate whether serum PRL could serve as a source of retinal vasoinhibins we used a rat model of hyperprolactinemia induced by placing two AP grafts under the kidney capsule for 15 days (9). As demonstrated by the Nb2 cell bioassay and Western blot, AP-grafted rats showed a sixfold increase of serum PRL (Fig. 2,A, n = 20) and threefold higher levels of vasoinhibins in the retina (Fig. 2,B and C, n = 3) compared with the nongrafted controls. Injection of grafted rats with the dopamine D2 receptor agonist bromocriptine, an inhibitor of PRL secretion (12), reduced both PRL in the circulation and vasoinhibins in the retina to basal levels (Fig. 2 A–C).

The ciliary body is responsible for the active transport of plasma proteins to intraocular fluids (13), and we detected that the PRL receptor is localized in this structure (Fig. 2,D, n = 3). To examine if the PRL receptor in the ciliary body could mediate uptake of circulating PRL and its transfer into the vitreous, we analyzed retinal vasoinhibins in PRL receptor–null mice, which are known to be hyperprolactinemic (12). The mice displayed 700-fold higher levels of circulating PRL (Fig. 2,E, n = 8), but the levels of retinal vasoinhibins were similar to those of their wild-type littermates (Fig. 2 F and G, n = 3).

We next investigated whether hyperprolactinemia, by raising intraocular vasoinhibins, could lower diabetes-induced retinal hypervasopermeability. To examine this issue, AP-grafted rats were challenged or not with intravitreously injected VEGF, which is one of the prominent vasopermeability factors in DR (1). VEGF caused multiple hemorrhage areas (Fig. 3,A and B, n = 3) and increased vasopermeability (Fig. 3,C, n = 3) in the retina. These VEGF-induced vascular alterations in the retina were greatly diminished in hyperprolactinemic animals (Fig. 3). Next, retinal vasopermeability was assessed in nongrafted and AP-grafted rats made diabetic by streptozotocin injection. Serum PRL levels were similar in diabetic rats compared with nondiabetic controls (Fig. 3,D, n = 4). Hyperprolactinemia was higher in nondiabetic than in diabetic animals (13-fold vs. 8.5-fold, respectively) and was blocked by treatment with bromocriptine (Fig. 3,D). Consistent with these changes, the levels of vasoinhibins were elevated in the retina of grafted rats, and bromocriptine reduced this increase (Fig. 3,E), supporting that hyperprolactinemia raises vasoinhibins in the retina. Then, we confirmed that retinal vasopermeability increases in diabetic animals (P ≤ 0.011) and found that the magnitude of this increase was significantly smaller (P ≤ 0.05) in hyperprolactinemic diabetic rats compared with the normoprolactinemic counterparts (Fig. 3,F, n = 4). Reduced retinal vasopermeability is likely due to the hyperprolactinemia-induced elevation of intraocular vasoinhibins because it was prevented by bromocriptine (Fig. 3,F). In the absence of diabetes, neither hyperprolactinemia nor administration of bromocriptine affected retinal vasopermeability compared with untreated animals (Fig. 3 F).

The association between circulating PRL levels and DR has long been controversial. Studies performed over 2 decades ago reported increased (14), decreased (15), or normal (16,17) PRL levels in patients with DR. Here, we show that the circulating concentration of PRL is higher in diabetic patients without retinopathy than in those with PDR. These findings indicate an inverse relationship between systemic PRL and the severity of DR and suggest that previous contradictory findings are due to the lack of DR grading and the small number of patients reported (14,,–17).

The inverse correlation between PRL and DR suggests that systemic PRL, after its conversion to vasoinhibins, influences the progression of DR. It was reported recently that patients with DR have lower levels of circulating vasoinhibins than nondiabetic patients (18). This result, together with our finding of higher circulating levels of PRL in diabetic patients without retinopathy, suggests that one way to upregulate intraocular vasoinhibins would be to further increase the systemic levels of PRL, thereby favoring its ocular incorporation and cleavage. Supporting this notion, hyperprolactinemia led to the accumulation of vasoinhibins within the retina of rodents. Notably, genetic deletion of the PRL receptor prevented this effect, indicating receptor-mediated incorporation of systemic PRL into the eye in a manner similar to that described for PRL transport across the choroid plexus to the cerebral spinal fluid (19) or across the mammary epithelium to the milk (20). These results suggest that the PRL receptor mediates the ocular incorporation of systemic PRL, which can then be cleaved to vasoinhibins.

Furthermore, the observation that higher levels of serum PRL mitigate excessive retinal vasopermeability in diabetic and VEGF-injected rats is consistent with hyperprolactinemia producing higher levels of ocular vasoinhibins and suggests hyperprolactinemia as a therapeutic strategy against diabetes-induced retinal hypervasopermeability. A broadly protective role of PRL in diabetes is supported by studies showing that PRL stimulates β-cell proliferation, insulin gene transcription, and insulin secretion in normal physiology and in pregnancy (21,22). Notably, the risk of DR development and progression increases during pregnancy, followed by a high regression rate during the postpartum period (23). The pathogenic mechanisms of DR in pregnancy are not fully understood, but PRL may play a protective role. There is evidence that PRL levels are lower in diabetic than in healthy pregnant women (24) and high, sustained PRL levels occur during lactation, when DR improves significantly (25). However, serum PRL levels showed no correlation with DR during pregnancy (24), and larger studies analyzing various stages of retinopathy are required to evaluate this relationship conclusively.

Here, we reveal PRL to be an important systemic inhibitor of diabetes-induced retinal hypervasopermeability after its intraocular conversion to vasoinhibins, which act directly on endothelial cells to block blood vessel growth, dilation, and permeability and to promote apoptosis-mediated vascular regression (5). Thus, we propose that PRL/vasoinhibins are endogenous regulators of the development and progression of DR and that current medications known to induce hyperprolactinemia constitute novel therapeutic options to treat DR and other vasoproliferative retinopathies.

This study was supported by the National Council of Science and Technology of Mexico Grant SALUD-2008-C01-87015.

No potential conflicts of interest relevant to this article were reported.

E.A. performed most experiments and contributed to postulating the hypothesis, designing the experiments, and writing the manuscript. J.C.R. performed measurements on human serum, participated in animal experiments, and contributed to postulating the hypothesis. S.T. analyzed patient data and contributed to designing the experiments and writing the manuscript. D.M.-P. and H.Q.-M. enrolled patients, collected human serum samples, and carried out the ophthalmologic examination. A.Q.-S. provided hyperprolactinemic rats. N.B. provided PRL receptor–null mice. G.M.E. gave scientific advice, supervised analysis, and edited the manuscript. C.C. conceived, coordinated, and supervised the study and contributed to postulating the hypothesis, designing the experiments, and writing the manuscript.

We thank C. García, F. López-Barrera, G. Nava, D. Mondragón, A. Prado, and M. García for their technical assistance, and D.D. Pless, Neurobiology Institute, UNAM, for critically editing the manuscript.