Plasminogen Activator Inhibitor (PAI)-1 Overexpression in Retinal Microvessels of PAI-1 Transgenic Mice

Abstract

purpose. Previous studies have suggested that disturbances in plasminogen activator inhibitor (PAI)-1 may be relevant to the development of diabetic microvascular complications. To determine whether overexpression of PAI-1 in cells of retinal microvasculature would result in a disease similar to that observed in diabetes, ocular tissue from transgenic mice that overexpress human PAI-1 were examined.

methods. Transgenic mice were administered ZnSO₄ (25 mM) in their water for up to 49 weeks to activate the metallothionein promoter and stimulate human PAI-1. Colloidal gold immunocytochemistry was used to quantify the human PAI-1 antigen at 7, 20, 34, and 49 weeks of ZnSO₄ administration. Cross sections of retinal microvessels were examined by electron microscopy for changes in basement membrane (BM) thickness. Retinal digest preparations were examined by light microscopy for possible microangiopathy, including changes in endothelial cell-to-pericyte ratios.

results. Human PAI-1 immunoreactivity was detected throughout the retinal capillaries of transgenic mice receiving zinc and increased significantly (P < 0.001) after 20 to 49 weeks of ZnSO₄ administration compared with age-matched transgenic control mice. At 20 and 49 weeks, retinal capillaries of transgenic mice that received zinc showed significantly thickened BMs compared with control animals (P < 0.001). Moreover, wholemounts of the retinal vasculature from PAI-1 transgenic mice demonstrated an increased endothelial cell-to-pericyte ratio.

conclusions. PAI-1 overexpression in retinal microvasculature leads to retinal disease similar to that observed in diabetic retinopathy.

Plasminogen activators (PAs) are serine proteases produced by the endothelium that activate plasminogen to plasmin. Plasmin degrades fibrin as well as basement membrane (BM) components (laminin and fibronectin) and activates matrix metalloproteinases (MMPs). ¹ One of the two PAs, the tissue-type plasminogen activator (t-PA) is associated with fibrinolysis, whereas the other PA, the urokinase-type (u-PA) is implicated in extracellular matrix proteolysis. ²  

PA activity is regulated by plasminogen activator inhibitor (PAI)-1, the primary regulator of fibrinolysis and a modulator of extracellular matrix proteolysis. ^{1 2} PAI-1 is detected in increased quantities in serum, vitreous, subretinal fluid, ³ and retinal microvasculature of humans ⁴ and monkeys ⁵ with diabetes. In vitro studies have shown that PAI-1 is secreted in large amounts by retinal endothelial cells and pericytes compared with the same cells of other vascular beds. ^{6 7} Increased synthesis of extracellular matrix by endothelial cells is believed to be responsible in part for the BM thickening observed in diabetic retinal microvasculature. ^{4 5 8 9 10 11} However, decreased proteolysis as a result of PAI-1 overexpression in diabetes may also be involved, because PAI-1 can protect extracellular matrix from proteolytic degradation. ¹²  

BM thickening and pericyte loss, both of which are retinal changes associated with diabetic retinopathy, relate to capillary wall integrity. ¹³ PAI-1 overexpression, by inhibiting fibrinolysis, ² may facilitate the formation of microthrombi resulting in transient capillary occlusion. Retinal pericytes not only provide support for the capillary wall but may also be involved in contractile actions controlling the diameter of the lumen and regulating the dynamics of microcirculation. Microthrombi and loss of pericytes could contribute to the development of more advanced features of diabetic retinopathy, such as microaneurysms and hemorrhage. ¹⁴  

In an attempt to elucidate what role PAI-1 plays in thrombosis, transgenic mice were developed. ¹⁵ In the creation of PAI-1 transgenic mice, a cDNA construct was designed containing the murine metallothionein I promoter, human endothelial cell PAI-1 cDNA, and the bovine growth hormone polyadenylation signal sequence. Subcutaneous hemorrhages develop in these mice 3 days after birth. By day 12 the tips of the tails are necrotic, and the hind feet are swollen due to venous occlusions. By 2 weeks the tails completely slough off, and there is resolution of the hind limb edema. The metallothionein promoter is constitutively expressed for only the first month of life. However, stimulation with ZnSO₄ included in the drinking water allows for continued expression. 

To determine whether PAI-1 overexpression by retinal endothelial cells could result in disease similar to that observed in diabetes, changes in PAI-1 in retinal capillaries of transgenic mice receiving ZnSO₄ in their drinking water were measured by colloidal gold immunocytochemistry. Retinal microvessels were examined for ultrastructural changes in BM thickness, and elastase digests of the retinal vascular bed were examined for possible pericyte loss. 

Methods

Animals

Breeding pairs of mice transgenic for the PAI-1 gene (B6,SJL-TgN) and of the parent strain (B6,SJL-F₁J) were kindly provided by L. Erickson (Upjohn, Kalamazoo, MI). 

For immunocytochemical localization studies, 48 mice were used. Transgenic mice (n = 12) homozygous for PAI-1 (determined by short stubby tail morphology) and age-matched nontransgenic control animals (n = 12; wild type with normal tail morphology) were given 25 mM ZnSO₄ in their drinking water for up to 49 weeks. Age-matched transgenic (n = 12) and nontransgenic mice (n = 12) received normal water to control for nonspecific effects of ZnSO₄. Three mice from each group were killed by overdose with pentobarbital at 7, 20, 34, and 49 weeks, and the ocular tissue was processed for immunocytochemical localization of human PAI-1 in transgenic mice as well as of mouse PAI-1. 

For BM measurements, 12 mice were used. Transgenic mice were given either water (n = 6) or 25 mM ZnSO₄ (n = 6) and killed at either 20 weeks (n = 3 from each group) or 49 weeks (n = 3 from each group). 

For retinal digests, 12 transgenic mice were also used. These mice were divided into two groups and were given either normal water (n= 6) or 25 mM ZnSO₄ (n = 6) for 20 weeks. 

Mice were cared for in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and with the National Institutes of Health Guide for Care and Use of Laboratory Animals. 

Immunocytochemistry

For immunocytochemical localization of PAI-1, whole eyes were fixed in cold 5% acrolein in 0.1 M sodium cacodylate-HCl buffer (pH 7.4) plus 0.1 M glycine for 1 hour, washed in buffer 4 × 15 minutes, postfixed in 1% osmium tetroxide in 0.1 M sodium cacodylate-HCl (pH 7.4), dehydrated in an ethanol series, infiltrated, and embedded in epoxy resin. Ultrathin sections were taken from the posterior retina within 1.5 mm of the optic nerve head, and sections on nickel grids were oxidized for 5 minutes with 10% periodic acid, followed by two 5-minute washes with deionized water. Grids were treated with 5% urea for 15 minutes and then washed with phosphate-buffered saline (PBS) twice for 5 minutes and once for 15 minutes with PBS blocker containing 2% bovine serum albumin (BSA) and 2% fetal bovine serum (FBS), followed by overnight incubation at 4°C with goat anti-human PAI-1 antibody (diluted 1:200 in PBS blocker; American Diagnostica, Greenwich, CT). After four 5-minute washes with PBS blocker, the grids were incubated for 1 hour at room temperature in donkey anti-goat IgG (Jackson ImmunoResearch, West Grove, PA) secondary antibody labeled with 18 nm colloidal gold. Grids were washed twice for 5 minutes with PBS followed by three 5-minute washes with deionized water. Capillaries from the inner nuclear and plexiform layers of the retina were examined and photographed by electron microscopy. Similarly, endogenous mouse PAI-1 was immunolocalized using sheep anti-mouse PAI-1 IgG (American Diagnostica; diluted 1:250 in PBS blocker) and donkey anti-sheep IgG (Jackson ImmunoResearch) secondary antibody labeled with 12 nm colloidal gold. 

Controls for specificity of labeling with PAI-1 consisted of reaction with primary antibodies preabsorbed with an excess of PAI-1. Controls for nonspecific binding of secondary antibody consisted of reaction of secondary antibody with nonimmune serum. 

Quantitative Analysis of PAI-1 Localization

Analyses were limited to cross sections of capillaries from the inner nuclear and plexiform layers of the retina. A minimum of 15 negatives of electron micrographs (magnification, ×10,000) were obtained by one masked investigator from randomly selected cross sections of retinal capillaries from each ZnSO₄-treated transgenic and nontransgenic mouse (n = 6 per group) and each deionized water–treated aged-matched transgenic and nontransgenic control mouse (n = 6 per group). Data were extracted from negatives by another masked investigator, as described. ¹⁶ After evaluation of the surface area (Sa) occupied by a capillary in a defined compartment, the number of gold particles (Ni) per unit area (expressed in micrometers) per negative was counted and the density of label (Ns) calculated according to the formula Ns = Ni/Sa. If occasional nonspecific binding of colloidal gold particles was seen in negatives of electron micrographs in controls for nonspecific binding, it was subtracted from the total colloidal gold counts of negatives of electron micrographs taken from sections of retinal capillaries from ZnSO₄-treated and control mice. 

Evaluation of BM Thickness

BM thickness of retinal capillaries of eyes (n = 6) from transgenic mice administered ZnSO₄ in their drinking water for 20 and 49 weeks was compared with BM thickness of retinal capillaries of eyes (n = 6) from age-matched transgenic control eyes. At least 10 capillaries per eye from the inner nuclear and plexiform layers were photographed at a magnification of× 10,000. Exact magnification was determined for each set of negatives with a 28,800-line/in. calibration grid (Ernest F. Fullam, Latham, NY). Negatives were printed with a ×3 enlargement. Measurements, to the nearest 0.25 mm, were made of the BM surrounding the endothelial cell and were taken perpendicular to the plane of the BM. ^{17 18} A minimum of 20 measurements were taken for each capillary, and the BM thickness was expressed as the average of 20 measurements. 

Elastase Digests

Retinal digests were prepared using elastase on retinas, as described in detail. ¹⁹ Eyes were removed from recently killed 20-week-old ZnSO₄-treated transgenic mice (n = 3) and age-matched transgenic control animals (n= 3). The retinas were fixed at room temperature by immersing the whole eye (slit at the limbus) in 4% (wt/vol) paraformaldehyde in 50 mM NaK phosphate buffer with 8% sucrose. The fixed retinas were rinsed in deionized water and incubated for 3 minutes in a 37°C agitating water bath in 40 U/ml elastase in NaK phosphate buffer with 150 mM NaCl and 5 mM EDTA (pH 6.5). The tissues were washed overnight in 100 mM Tris-HCl (pH 8.5) and then transferred to deionized water for removal of the loosened vitreous and digested neural elements by gentle agitation using the sides of closed forceps and the sides and ends of very fine brushes (e.g., Series 101, Sceptre 5181731, 4 × 0; Winsor & Newton, UK) as needed. After all loose tissue was removed, the retinas were incubated once more in fresh enzyme for 3 minutes and then subjected to a second overnight wash at room temperature in Tris-HCl buffer. On the third day, the retinas were again transferred to deionized water for additional removal of digested neural elements. The vascular network that was completely free of nonvascular elements was mounted flat by flotation in Ca²⁺- and Mg²⁺-free Dulbecco’s PBS on siliconized slides (S1308; Oncor, Gaithersburg, MD). After they were air dried in a dust-free environment, the mounts of the retinal microvasculature were stained using periodic acid–Schiff reaction and hematoxylin counterstaining, as described. ²⁰ The preparations were then examined by light microscopy and photographed. 

Endothelial–Pericyte Ratios

The stained and intact retinal wholemounts were coded, and subsequent counting was performed masked, as described. ²¹ Ten fields at ×100 magnification were counted for endothelial and pericyte cells, according to previously described morphologic criteria. ²² In every sample, at least 200 cells were counted from the midzone of the retina. Mean values for endothelial cell–pericyte (E/P) ratios were calculated for six retinas from 20-week ZnSO₄-treated transgenic mice and six retinas from transgenic control mice. 

Statistical Analysis

Statistical analysis for comparison among groups was performed using one way analysis of variance and Student’s t-test. Significance was defined as a value of P < 0.05. Values are reported as mean ± SEM. 

Results

PAI-1 Localization in Transgenic Mouse Ocular Tissue

The immunogold microscopic technique allowed visualization of PAI-1 on the basis of reaction of specific antibodies with colloidal gold–conjugated secondary antibodies. Increased human PAI-1 immunoreactivity was localized in endothelial cells of retinal capillaries of transgenic mice accompanied by BM thickening after 20 to 49 weeks of ZnSO₄ administration compared with age-matched transgenic control eyes (Figs. 1A 1B 1C and 2A 2B 2C , respectively). Retinal capillaries processed as controls for specificity of labeling with PAI-1 (reaction of primary antibody preabsorbed with an excess of PAI-1) and nonspecific secondary antibody labeling (exposure of the secondary antibody to nonimmune serum), did not depict any colloidal gold–labeled immunoreactivity (not shown). 

Quantitative Analysis of PAI-1 Localization

Quantitative analysis of the changes in human PAI-1 in retinal capillaries of ZnSO₄-treated transgenic mice was performed. PAI-1 colloidal gold immunoreactivity was localized in endothelial cells of retinal capillaries and was elevated significantly (P < 0.001) at 20, 34, and 49 weeks of ZnSO₄ administration compared with transgenic aged-matched control mice that received water. The human PAI-1 levels in retinal microvasculature of transgenic control animals did not change at the various time points examined (Fig. 3A ). Retinal capillaries of nontransgenic mice receiving ZnSO₄ for up to 49 weeks (expressing only mouse PAI-1), did not depict any changes in mouse PAI-1 immunoreactivity at the various time points examined compared with control animals that received water (Fig. 3B) . Low levels of mouse PAI-1 were detected and were not influenced by zinc treatment. 

Capillary BM Thickness

The mean thickness ± SEM of retinal capillary BMs at 20 weeks was significantly higher (2.35-fold; P < 0.001) in the ZnSO₄-treated transgenic mice (182.82 ± 25.0 nm) than in the control mice (77.63 ± 10.74 nm). Similarly, in the zinc-treated transgenic mice at 49 weeks the mean BM thickness was significantly higher (3.53-fold, 338.60 ± 72.9 nm) than in control animals (95.76 ± 15.55 nm, P < 0.001; Fig. 4 ). 

Elastase Digest Preparations and E/P Ratios

In intact wholemounts of retinal digests the endothelial cell nuclei, seen medially within the vessel wall, were large, oval, and weakly stained and protruded luminally. Pericyte nuclei, seen more laterally, were darkly stained, small, and round and protruded prominently away from the vessel wall. E/P counts were taken from midzones of the retinas. The mean E/P ratio of retinal microvessels from ZnSO₄-treated transgenic mice was significantly higher (3.2 ± 0.6) than the mean E/P ratio from control transgenic retinal microvessels (1.9 ± 0.5; P < 0.05). 

Discussion

Immunocytochemical studies described here provide semiquantitative data demonstrating that retinal microvessels from PAI-1 transgenic mice that received ZnSO₄ in their drinking water to stimulate overexpression of the transgene contained increased amounts of PAI-1 compared with retinal microvessels of aged-matched transgenic control animals. The microvascular endothelial cells appeared to be the primary source of the increased amounts of PAI-1 rather than pericytes. At 20 and 49 weeks of ZnSO₄ administration in transgenic mice, the BM was thickened significantly. Such increased thickening may affect the anatomic and functional relationship between endothelial cells and pericytes and could be implicated in the pericyte loss presently observed, a sign of retinal microangiopathy also seen in early diabetic retinopathy. ²³  

In a recent study, we reported that PAI-1 was found to be elevated significantly in capillaries of monkeys with non–insulin-dependent diabetes mellitus, and the increased PAI-1 in retinal capillaries correlated with BM thickening. ⁵ An earlier immunocytochemical study revealed that retinal microvessels of diabetic individuals contained significantly greater quantities of PAI-1 than retinal microvessels from age-matched nondiabetic control subjects. Moreover, elevated levels of PAI-1 were detected in serum of diabetic individuals ^{3 24} and increased PAI-1 mRNA has been shown in retinal microvessels from ocular tissues of diabetic donors. ²⁵ Decreased t-PA activity is seen in the retina of diabetic individuals. ²⁶ Similarly, nephritic glomeruli express decreased PA activity and increased PAI-1 synthesis, preceding the accumulation of BM in nephritic glomeruli. ²⁷ This abnormality is reversed with correction of the nephritis. 

In virus-induced diabetes mellitus in mice, excessive accumulation of BM is also found in retinal and renal microvessels. ²⁸ BM thickening has been commonly reported in long-term insulin-dependent diabetes, and it appears to be related to glucose levels in the blood. ²⁹ Similar BM thickening was found in retinal capillaries of rats fed a diet containing 50% galactose for 44 weeks. ³⁰ The enzyme aldose reductase, which converts sugars to polyols, has been implicated in BM thickening. ³¹ Inhibition of the enzyme can prevent BM thickening in streptozocin-induced diabetic rats ^{32 33} or galactose-fed rats. ^{18 30 31 34} However, the underlying mechanism of BM thickening is not well understood. 

In vitro studies have shown that supplementation of insulin, proinsulin, insulin-like growth factor (IGF)-I, or glucose at concentrations seen in plasma of diabetic individuals enhance PAI-1protein expression in endothelial cultures. ^{35 36 37} Moreover, we have previously shown that rabbit retinas exposed to IGF-I overexpress PAI-1. ⁴ These agents can enhance transcription of the PAI-1 gene and stabilize PAI-1 mRNA. ³⁸  

Together with our earlier findings, the present data suggest that PAI-1 overexpression by retinal endothelial cells could influence the amount of BM surrounding a vessel. PAI-1 regulates both t-PA and u-PA and the subsequent generation of plasmin. ² When the generation of plasmin is decreased, the activation of MMPs is diminished. ¹ As a result, there is decreased proteolysis influencing the amount of BM surrounding a vessel. Our studies provide a possible mechanism for BM thickening observed in diabetes by supporting a role for increased PAI-1 expression and decreased matrix proteolysis. 

In the present study, thromboses developed in the transgenic mice that overexpress the PAI-1 gene. ¹⁵ PAI-1 overexpression could facilitate microthrombus formation by inhibiting fibrinolysis, stabilizing fibrin, and inducing increased platelet aggregation. ^{39 40} However, these effects are difficult to detect, because they are transient. Repeated capillary thromboses induce endothelial cell damage. Foci of nonperfused capillaries appear in diabetic retinopathy, and these acellular tubes, composed of remnants of BM material, surround the empty pericyte cell region. ⁴¹  

PAI-1 overexpression by the retinal vasculature could contribute to the pathologic changes observed, such as BM thickening and pericyte loss. The present study showing ZnSO₄-induced overexpression of PAI-1 in transgenic mice provides a good model for further studies of PAI-1–induced changes in retinal microvessels—changes that appear similar to those described in diabetic retinopathy. 

³  Deceased September 6, 1997.

 Supported in part by Grant EY-07739 from the National Institutes of Health.

 Submitted for publication September 14, 1999; revised December 20, 1999; accepted January 5, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Profiles of retinal capillaries from transgenic mice that received ZnSO₄ in their drinking water for (A) 20, (B) 34, and (C) 49 weeks. Note the numerous colloidal gold particles (arrows; 18 nm) showing positive immunoreactivity to PAI-1, accompanied by thickened BM (arrowheads). Scale, 1 μm.

Figure 2.

 

Profiles of retinal capillaries from control transgenic mice that received water alone for (A) 20, (B) 34, and (C) 49 weeks. Note some colloidal gold particles (arrows; 18 nm) showing positive immunoreactivity to PAI-1, accompanied by a less thickened BM (arrowheads) compared with Figures 1A 1B and 1C . RBC, red blood cell. Scale, 1μ m.

Figure 3.

 

PAI-1 immunoreactivity in retinal capillaries of transgenic mice after 20, 34, and 49 weeks of ZnSO₄ administration, compared with retinal capillaries in transgenic mice that received water without zinc. The number of colloidal gold particles (positive PAI-1 immunoreactivity) per unit retinal capillary area is shown. Results are expressed as means ± SEM of the number of particles in a minimum of 15 negatives of electron micrographs per group, obtained from random triplicate samples in three independent experiments (n = 3 mice per group). (A) Human PAI-1 immunoreactivity in retinal capillaries showing an increase in colloidal gold in the retinal capillaries of transgenic mice after 20, 34, and 49 weeks of ZnSO₄ administration over that seen in transgenic control mice (*P < 0.001). (B) Mouse PAI-1 immunoreactivity in retinal capillaries showing no change in the number of colloidal particles (positive mouse PAI-1 immunoreactivity) in the retinal capillaries of transgenic mice at 20, 34, and 49 weeks of ZnSO₄ treatment compared with nontransgenic control mice.

Figure 4.

 

Mean retinal capillary BM thickness (in nanometers) of 20-week and 49-week ZnSO₄-treated transgenic mice compared with transgenic control mice. Results are expressed as means ± SEM (n = 3 mice in each group). The mean BM thickness in the 20- and 49-week ZnSO₄-treated mice was significantly greater than in control animals (*P < 0.001).