An electron microscopic study of the permeability of intraocular blood vessels using lanthanum as a tracer in vivo

In the present work lanthanum has been used as an electron microscopic tracer to study intraocular blood vessel permeability in vivo. Aqueous solutions of lanthanum were injected intravenously in rabbits. The eyes were enucleated 30 or 60 min later. Lanthanum was detected as irregular electron-dense deposits in the vascular lumen of intraocular blood vessels. In normal eyes, the tracer did not permeate the endothelium of blood vessels in iris, ciliary processes, choroid, or retina. Conjunctival application of prostaglandin E₁ (25 μg) 15 min before lanthanum injection, induced leakage of the tracer from blood vessels in the anterior parts of the ciliary processes and their iridial extensions. Leakage occurred through gaps between endothelial cells. Blood vessels showing increased permeability to lanthanum were readily detected, as heavy deposits of tracer material had accumulated in their walls. In the present work, lanthanum has been found to be a suitable tracer to study intraocular blood vessel permeability in the experimental animal. The technique is direct and simple, and the tracer is easily visualized in the electron microscope.     

Electron microscopic studies on the blood-aqueous barrier of prostaglandin-treated rabbit eyes. I. Iridial and ciliary processes.

Prostaglandins E1 and E2 were applied topically to rabbit eyes. Structures related to the blood-aqueous barrier in the iridial and ciliary processes, as well as the permeability of the ciliary epithelium to the protein tracer horseradish peroxidase, were studied with the electron microscope. Marked morphological changes, including dilations of the intercellular spaces and separation of the two epithelial cell layers, were found in the epithelium of the iridial processes. Only minor structural changes were found in the epithelium of the ciliary processes. Leakage of peroxidase through the intercellular spaces of the epithelium was demonstrated in the iridial processes and in the anterior parts of the ciliary processes. In the ciliary vessels of the same regions, opening of interendothelial gaps, platelet aggregations, microthrombi, and haemorrhages were found. In a previous in vitro study on the effects of prostaglandins on the movement of peroxidase in the ciliary epithelium, no structural changes of the epithelium were found, and the epithelial diffusion barrier to peroxidase was found to be intact. It is assumed that the breakdown of this barrier in vivo is secondary to vascular changes.     

The source of proteins in the aqueous humor of the normal rabbit

Aqueous fluorophotometric, tracer localization and modeling methods were combined to document the existence of a pathway in the normal rabbit for the diffusion of proteins from the ciliary and iridial process stromas through the iris stroma into the aqueous humor of the anterior chamber. A new custom-conjugated tracer, fluoresceinated horseradish peroxidase (F-HRP), was used. Anesthetized rabbits were injected intravenously with F-HRP (250 mg/kg). In some animals, aqueous fluorophotometric and tracer localization studies were performed on the same eyes. Anterior chamber fluorescence was detected 2-10 min post-injection and rose to concentrations of 0.01-0.05 mg/ml 60 min post-injection. Subsequent tracer localization studies of these eyes revealed that the morphologic components of the blood-aqueous barrier were intact, that is, no leakage of F-HRP from the iris vasculature or across the ciliary epithelium was observed. Separate tracer localization studies were performed to examine the time course of the route(s) by which tracer entered the anterior chamber. These studies revealed a "wave" of tracer that migrated from the ciliary and iridial process stromas, through the iris, and arrived at the anterior iris surface approximately 8 min post-injection. A pharmacokinetic model based on the diffusional pathway was developed to describe the time course of the concentration of plasma macromolecules in the ciliary body, iris and anterior chamber. Model predictions were consistent with aqueous fluorophotometric and tracer localization results. The diffusion model can account for a major fraction of protein entering the aqueous humor of normal rabbit eyes.     

Fluorescent angiographic evaluation of the iris microcirculation in cataract extraction by Nd:YAG laser with wavelength of 1.44 mcm

The purpose of the present study is to evaluate the iridial microcirculation by fluorescent iridial angiography after laser cataract extraction by using a new extraction device, i.e. "RAKOT", based on the Nd:YAG-laser with wave length of 1.44 mcm. 25 eyes (21 patients) aged 71 to 87 with mature senile cataract were examined. The method of laser cataract extraction was used in 14 cases. These eyes made up the main examination group. As for the controls (11 eyes), the manual technic of cataract extraction was used for them. The results in the main group show that the degree of the response of the iridial microvessels to surgery is insignificant and the microcirculation restores completely in 1.5-2 months after surgery provided there are no initial pathological changes in the iridial microcirculation (3 eyes). A short-time intensification of permeability of the iridial microcirculation is possible in eyes with a pronounced initial pathology in the iris (11 cases), however it restores to the original level during 3-3.5 months after surgery. The iridial microcirculation restores in the controls during 3 months after surgery. The described data demonstrate that the laser cataract extraction does not lead to pronounced iridial vascular disorders. A comparison with the controls is indicative of a lower trauma in case of laser method versus the manual technic of lens extraction.     

Subthreshold argon-laser irradiation elicits a pronounced vitreal prostaglandin E2 response

Chorioretinal production of prostaglandin type E₂ (PGE₂) as well as changes in vitreal protein and PGE₂ concentrations were studied following retinal irradiation with low-power laser energy that caused ophthalmoscopically invisible (subthreshold) laser burns. This was compared with changes following ophthalmoscopically visible ( suprathreshold ) laser burns. Our results demonstrate that with in eyes exposed to the lower-power levels, the enhancement in vitreal PGE₂ concentration persisted for a longer period and was more pronounced than in eyes exposed to the suprathreshold levels (a 3-fold and a 2-fold increase above baseline values, respectively). Protein leakage into the vitreous was noted only in the suprathreshold group, indicating a blood-retinal barrier (BRB) disruption. The findings of persistent, excessive PGEZ vitreal levels with no protein leakage in the subthreshold group suggest a possible anti-inflammatory role for PGEZ following low-power laser exposure.This work was partially supported by United States Army Medical Research and Development Command contract 17-85-G-5013 Offprint requests to: N. Naveh     

Influence of prostaglandins on cation movement in the lens

Rabbit lenses incubated at 36°C for 20 hr in a medium containing 10^?3m prostaglandin (PG) E₁, E₂ or E_2^α exhibited a marked increase in sodium and decrease in potassium content; 10^?3m PGF_1^α was without effect. 10^?3m PGE₁ caused a significant increase in cation permeability as evidenced by increased rubidium-86 efflux. Reduced rubidium-86 uptake was observed in lenses incubated in 10^?3m PGE₁ and F_α. 10^?4m prostaglandins (E₁, E₂ and F_2^α) appeared to stimulate the lens cation transport mechanism. Recovery from a cold induced cation shift was enhanced, as was rubidium-86 uptake. Lenses subjected to reduced calcium level in the incubation medium survived better in the presence of 10^?4m PG. It is suggested that the effects of 10^?4m PG are mediated via the adenyl cyclase system. The effects of 10^?3m PG might be due to a large non-specific increase in lens cation permeability.     

The DC-ERG as a highly sensitive measure of effects of prostaglandins

A few microliters of a prostaglandin preparation (PGE₂ or PGF₂) were injected via the ora serrata into the posterior vitreous of one of the eyes in albino rabbits. The fellow eye received an equal volume of saline intravitreally and served as control. The DC electroretinogram (ERG) and the standing potential of the eye (SP) were recorded dkectly with corneal contact lenses, very stable calomel electrodes, and under very constant anesthesia. The b- and c-wave amplitudes increased in response to 0.1 and 1.0 microgram of PGE₂ and PGF₂, respectively. At medium doses there was an increase in the b- and c-wave amplitudes followed by a long-lasting reduction. At very high, nonphysiological doses of PGE₂, b- and c-wave amplitudes decreased as compared with the control eye. Prostaglandins modify inflammatory reactions, influence ion transport across membranes, modulate synaptic transmission, and regulate blood flow to various organs. Effects of extremely low doses on the retina and pigment epithelium might indicate a transmitter-like nature of prostaglandins. The present experimental model might be of use in studies of inflammatory eye disease, prostaglandin inhibitors, and characteristics of the pigment epithelial membranes.     

Inactivation of prostaglandin E1 by ocular tissues in vitro

We have studied the ability of ocular tissues to metabolize prostaglandin E₁ (PGE₁) in vitro and compared this activity with other tissues such as the lung and kidney under standard conditions of temperature, pH and substrate concentrations. Much greater loss of biological activity of PGE₁ was observed following incubation for 20 min with supernatant fractions derived from the kidney cortex than with similar fractions derived from the iris/ciliary body. Supernatant fractions of ocular tissues (iris/ciliary body, conjunctiva and lens) metabolized [³H]PGE₁ very slowly when compared to the rapid metabolism of [³H]PGE₁ observed with kidney and lung preparations. For example, only 20–30% of [³H]PGE₁ was metabolized following a 60-min incubation with iris/ciliary body supernatant compared with 80% [³H]PGE₁ lost within 20 min of incubation of kidney cortex or lung supernatants. The disappearance of PGE₁ was accompanied by the formation of a compound with the same Rf value as 15-keto-PGE₁. The present results indicate that under standard conditions ocular tissues have comparatively little PG dehydrogenase activity and emphasize the importance of the absorptive transport mechanism for the removal of PGs from intraocular fluids.     

Permeability of iridial blood vessels.

The permeability properties of rat iridial vessels were examined using fluorescein angiography combined with light and electron microscopic marker studies. Results show that while control iridial vessels of some young rats are permeable to fluorescein, a marked permeability to this dye is commonly seen in older animals. Following paracentesis there is a marked increase in permeability of the radially arranged iridial vessels of older animals to fluorescein while angiograms on the younger animals were inconclusive. When carbon is injected intravenously following completion of the angiography study, both light and electron microscopic observations show that iridial vessels of control eyes are impermeable to carbon. However, following paracentesis an inflammatory reaction occurs. Carbon is seen leaving the lumen of iridial vessels in enlarged spaces between adjacent endothelial cells. This inflammatory reaction is age-related; the increase in permeability to carbon is greatest in older animals. There is little or no carbon labeling in the choroid or ciliary body of these experimental eyes.Thus, in the rat, paracentesis causes a breakdown in the blood-aqueous barrier by specifically inducing an inflammatory reaction in iridial blood vessels. This inflammatory reaction is age-related.     

Comparison of iridial pigmentation between latanoprost and isopropyl unoprostone: a long term prospective comparative study

AIM: To compare incidence of iridial pigmentation prospectively induced by long term treatment with latanoprost and isopropyl unoprostone (hereafter, unoprostone) in Japanese patients with glaucoma. METHODS: Patients with glaucoma treated with prostaglandin (PG) related ophthalmic solutions were sequentially enrolled. Patients treated for more than 30 months with PG related ophthalmic solutions were subjected to analysis. The entry criteria were no history of intraocular surgery, laser iridotomy, and/or laser trabeculoplasty within 12 months before and after the enrolment; and no history of uveitis; no changes in antiglaucoma drugs within 6 months before and after the enrolment. Photographs of the irides were taken under the same conditions and three glaucoma specialists evaluated the iridial pigmentation with masking of patient information. The correlation of iridial pigmentation with the background factors and the reduction of intraocular pressure (IOP) before and after the treatment were investigated. RESULTS: 48 eyes in 48 patients satisfied the enrolment criteria (25 eyes in the latanoprost group, 23 eyes in the unoprostone group). At the end of the follow up period, iridial pigmentation was present in 15 patients (60.0%) in the latanoprost group and seven patients (30.4%) in the unoprostone group. The correlation between development of iridial pigmentation and age, sex, concurrent use of other ophthalmic solutions, and IOP reduction was not significant. CONCLUSIONS: The incidence of iridial pigmentation induced by latanoprost or unoprostone is high in the case of long term treatment. Iridial pigmentation did not affect PG related ophthalmic solution induced IOP reduction.     

Prostaglandin E2 Levels of Aqueous and Vitreous Humor in Ketorolac 0.4% and Nepafenac 0.1% Administered Healthy Rabbits

Purpose: To compare the lowering effects of ketorolac 0.4% and nepafenac 0.1% on aqueous and vitreous humor prostaglandin E₂ (PGE₂) levels in rabbits.

Methods: Ketorolac and nepafenac ophthalmic solutions were administered to the right eyes of 24 healthy rabbits after randomized division into two groups. The left eyes of these rabbits were considered as controls for the two groups. On the 4th day of the experiment, the samples were taken from the aqueous and vitreous humors of the rabbits bilaterally, and PGE₂ levels were measured by an enzyme immune assay kit.

Results: Ketorolac and nepafenac achieved a statistically significant decrease (p<0.001, for each) in PGE₂ levels in the aqueous (11.75 ± 6.15 and 14.75 ± 7.60 pg/mL, respectively) and the vitreous humor (6.58 ± 4.62 and 9.83 ± 4.55 pg/mL, respectively).

Conclusions: Both ketorolac and nepafenac inhibited PGE₂ levels in both the aqueous and vitreous humors of rabbits. Although PGE₂-lowering effects were similar in the aqueous humor, nepafenac seemed to be more potent than ketorolac in the vitreous humor.     

Comparison of ultrasound biomicroscopy and ultrasonographic parameters in eyes with phacomorphic glaucoma and eyes with mature cataract

Purpose

To compare data on anatomical and biometric parameters, and their combination, obtained by using ultrasound biomicroscopy (UBM) and A-scan in eyes with phacomorphic glaucoma (PG) and eyes with mature cataract and to determine differences that may predispose to development of PG.

Methods

Eighty patients (80 eyes) were enrolled in this cross-sectional study. Anterior chamber parameters, namely anterior chamber depth (ACD), angle-opening distance (AOD), iris–lens contact distance (ILCD), and trabecular-ciliary process distance (TCPD), among other parameters, were assessed by UBM (35 MHz), while axial length (AL) and lenticular thickness (LT) were determined by A-scan biometry. Absolute lenticular position (ALP) and relative lenticular position (RLP) were also compared.

Results

The mean AL of eyes with PG was less (P = 0.64) than the mean AL of eyes with mature cataract. The mean ILCD value in eyes with PG (1.30 ± 0.78 mm) was significantly higher (P = 0.0015) than that in eyes with mature cataract (0.86 ± 0.32 mm). The mean AOD value in eyes with PG (0.15 ± 0.10 mm) was significantly lower (P < 0.0001) than that in eyes with mature cataract (0.23 ± 0.06 mm); similarly, the mean RLP value in eyes with PG (0.20 ± 0.02) was significantly lower (P < 0.0001) than that in eyes with mature cataract (0.17 ± 0.02). TCPD showed negative correlation with LT r = ?0.514; P = 0.017, r ² = 0.264, in eyes with phacomorphic glaucoma and mature cataracts showed significant negative correlation of ILCD with AL (r = ?0.575; P = 0.025, r ² = 0.330).

Conclusion

The results of UBM examination suggest that there are multiple mechanisms underlying the pathophysiology of PG, namely crowding of the anterior segment, increased iris–lens contact distance, and a more pronounced anterior shift of the lens. Simultaneous evaluation of anatomic and biometric parameters can improve diagnostic accuracy in predicting cases at risk for PG.

     

A comparison of the miotic and inflammatory effects of biologically active polypeptides and prostaglandin E2 on the rabbit eye

The ocular effects of some biologically active peptides were studied and compared to those of prostaglandin E₂ (PGE₂) to determine whether the responses of the eye to trauma, characterized by increased intraocular pressure (IOP), the development of anterior chamber flare and partial miosis, might be mediated by such peptides. One to 2 hr after intravitreal injection of 10 μg of PGE₂ into rabbit eyes, ocular hypertension, flare and iridial hyperemia, but only minimal miosis were observed. Maximum miosis developed within 2–3 hr after intravitreal injection of 1·0–100 μg of substance P (SP), SP-octapeptide (SP-8), coherin or eledoisin-related peptide (EDR), while 10–100 μg of vasoactive intestinal polypeptide (VIP), somatostatin or bradykinin (BK) yielded only submaximal miosis and angiotensin II, α-MSH and poly-dl-alanine had little or no miotic effect. None of these peptides caused iridial hyperemia or a cellular invasion of the anterior chamber and only high doses (100 μg) of VIP or BK caused significant increases in the protein concentration of the aqueous humor. Miotic doses of SP, SP-8 or EDR caused a significant increase in IOP in some, but not all, experiments. Thus, PGE₂ can be regarded as a mediator of the ocular irritative response although it may not account for the miosis that is associated with chemical or surgical trauma. However, this autacoid should not be regarded as the mediator of ocular inflammation since it does not elicit a cellular response. In contrast, some polypeptides, particularly SP, SP-8 and EDR are strong miotics and, at least under some circumstances, can act as effective ocular hypertensives, but these peptides do not reproduce any other signs of ocular irritation or inflammation. It is therefore concluded that none of the peptides studied here could, by itself, be the sole mediator of the initial ocular irritative response although some of them may account for the miosis and contribute to the ocular hypertension associated with this response. A combination of some of these peptides together with PGE₂ and/or other PGs may account for all aspects of the ocular irritative response and for most aspects of the ocular inflammatory response.     

Increase in choroidal blood flow in rabbits with endothelin-1 induced transient complete obstruction of retinal vessels

Background: In a previous paper, we reported that retinal blood flow (RBF) ceased immediately after injection of 1 nmol endothelin-1 (ET-1) and no recovery of RBF was detected for at least 50 min. In this study, we confirmed the same duration of RBF cessation and measured choroidal blood flow (CBF) for 180 min. Methods: We measured CBF in a rabbit model of transient complete obstruction of retinal vessels induced by intravitreal injection of a high dose of ET-1, using the hydrogen clearance method. We also investigated the effects of intravitreal injection of ET-1 on intraocular pressure (IOP), blood pressure, pulse rate and blood gases. Results: CBF was significantly greater in the ET-1-injected eyes than in the control eyes 40–130 min after injection of ET-1 (P < 0.05). The maximal CBF ratio in the ET-1-injected eyes was 128 ± 7.4% at 40 min. CBF decreased to the pre-injection level at 140 min after the injection of ET-1. There was no significant change in blood pressure, pulse rate and blood gases throughout this experiment, and there was no significant difference in IOP between ET-1-injected eyes and control eyes. Conclusion: It seems likely that the increase in CBF resulted from some local mechanisms of control that compensated for the decrease in RBF induced by intravitreal injection of ET-1. This model may be useful for investigation of the regulatory system of intraocular circulation, including endothelin receptors.     

Effects of acetylcholine and norepinephrine on incorporation of [32P]orthophosphate into phospholipids of rabbit iridial processes and iris smooth muscle

We have investigated: (a) phospholipid composition, inclusive of higher inositides, of rabbit iridial processes and iris smooth muscles; (b) ³²P_i incorporation into their respective phospholipids, and (c) the effects of muscarinic cholinergic and adrenergic agonists and antagonists on ³²P labelling of phospholipids of the iridial processes and iris smooth muscle. (1) Phosphatidylcholine, phosphatidylethanolamine, their respective plasmalogens and sphingomyelin, were found to be the major phospholipids in the iridial processes and iris smooth muscle. They constituted about 85% of the total phospholipids of these ocular tissues. (2) Both iridial processes and iris smooth muscles rapidly incorporated ³²P_i and [1-¹⁴C]-arachidonic acid into their respective phospholipids, however this incorporation amounted to only 20% of that found for the whole iris-ciliary body. This could suggest a metabolic interrelationship between the iridial processes and the smooth muscle of the iris. (3) Addition of acetylcholine and norepinephrine to the iridial processes and iris smooth muscle increased ³²P labelling of phosphatidic acid and phosphatidylinositol of the tissues. The increase in phospholipid labelling was higher in the iridial processes as compared to the iris smooth muscle. The effect of acetylcholine was blocked by atropine and that of norepinephrine was blocked by phentolamine and prazosin but not by yohimbine. This suggests that the observed effects of these neurotransmitters on phospholipid phosphorylation in the iridial processes and iris smooth muscle are mediated through muscarinic cholinergic and α₁-adrenergic receptors, respectively. The data presented provide additional support for the concept that in the iris-ciliary body the neurotransmitter-induced ³²P labelling of phosphoinositides is probably linked to the functional activities of this tissue.     

The mechanism of aqueous outflow in birds. II. An ultrastructural study of perfused eyes

In order to detect the route of flow of the large molecules of the aqueous humour, a colloidally suspended electron-dense tracer (Thorotrast—particle size approx. 10 nm) was injected in vivo into the anterior chambers of chicken eyes at a controlled intraocular pressure. After about 20 min, the eyes were fixed and processed for electron microscopy. It was observed that almost all vacuoles in the endothelial lining of the aqueous sinus were filled with the tracer element from their basal openings but a leakage of this material into the lumen of the aqueous sinus was seen only from a small proportion of vacuoles which have openings on the luminal aspect. No leakage of the tracer element occurred, however, across the endothelial cell junctions and the majority of micropinocytotic vesicles were devoid of tracer particles. Variable concentrations of the tracer particles were also seen in the lumen of the aqueous sinus and the intrascleral collector channels. It is concluded that the porosity of the endothelial lining of the aqueous sinus is dependent upon the dynamic process of endothelial vacuolation and that the bulk transfer of the aqueous humour takes place via the transient transcellular channels created during this vacuolation cycle. Reference has been made to the factors which could possibly influence this process. A close similarity of this mechanism of aqueous outflow in the eyes of birds to that of primates (Tripathi, 1969a, Tripathi, 1969b, Tripathi, 1971a) and lower mammals (Tripathi and Tripathi, 1972) indicates that the dynamic process of endothelial vacuolation is a fundamental biological process, probably common to the eyes of many vertebrates.     

Enzymatic digestion increases permeability of the outer blood-retinal barrier for high-molecular-weight substances

Background: The purpose of the study was to investigate whether lysosomal enzymes can participate in damaging the outer blood-retinal barrier and to examine the role of glycosaminoglycans in maintaining the barrier function for high-molecular-weight substances. Methods: The ciliary artery was cannulated in freshly enucleated pig eyes. Perfusion was performed with buffer (controls), with heparinase (substrate: heparan sulfate), or with lysosomal enzymes freshly prepared from pig retinal pigment epithelium at 36° C, followed by perfusion with the tracer native ferritin (NF) or the marker cationized ferritin (CF). The eyes were examined by electron microscopy. Results: In controls treated with buffer alone, NF was found in high concentration in the lumina of the choroidal capillaries; however, little NF was found in Bruch's membrane (BsM). The tracer did not penetrate to any extent beyond BsM. In eyes digested with heparinase or lysosomal enzymes, significantly higher numbers of tracer molecules were found in BsM. Furthermore, NF penetrated BsM and was apparent in the subretinal space and also inside retinal pigment epithelial cells, probably due to pinocytosis. Conclusions: The results indicate that heparan sulfate proteoglycan is important for the maintenance of the outer blood-retinal barrier and that lysosomal proteases may participate in damaging this barrier, causing increased permeability to high-molecular-weight substances.Presented in part at the Association for Research in Vision and Ophthalmology Annual Meeting, Sarasota, Florida, 28 April 1991     

Structural and functional assessment by hemispheric asymmetry testing of the macular region in preperimetric glaucoma

Purpose

To investigate structural measurements of the macular area in preperimetric glaucoma (PG) patients using spectral domain optical coherence tomography with two functional measurements [10-2 Humphrey visual field (HFA) and 10-2 Microperimeter-1 (MP-1)] and by macular symmetry testing.

Methods

Fifteen eyes of 15 PG subjects with a retinal nerve fiber layer defect in the inferior hemisphere and 15 eyes of 15 normal control subjects were enrolled. Macular symmetry testing was performed between the superior and inferior zones by comparing zone thickness in each hemisphere. Perimetric sensitivity asymmetry was calculated with two functional measurements. Structure–function relationships between macular symmetry testing and the mean retinal sensitivity of the corresponding hemifield or perimetric sensitivity asymmetry were calculated using Spearman’s rank correlation and linear regression.

Results

Macular zone thickness in the abnormal hemispheres was significantly less than that in normal hemispheres in PG eyes and the corresponding hemispheres in control eyes (P < 0.001). Macular symmetry testing was significantly lower in PG eyes compared to control eyes (P < 0.001). HFA (10-2) and MP-1 (10-2) correlated significantly (rs = 0.81, P < 0.0001). Macular symmetry testing values were significantly correlated with perimetric sensitivity and perimetric sensitivity asymmetry with two functional measurements (rs = 0.61, P = 0.02; HFA and rs = 0.68, P = 0.006; MP-1).

Conclusions

Our results suggest that macular asymmetry analysis can reveal the structure–functional relationship in PG eyes.     

Disruption of iridial blood-aqueous barrier in experimental diabetic rats

The permeability pattern of iridial blood-aqueous barrier in streptozotocin-diabetic rats, maintained for 1–6 months, was studied with the light and electron microscopes, using horseradish peroxidase as a tracer. A reaction product of horseradish peroxidase was confined to the vascular lumina and to a small number of vesicles on the luminal portion of the endothelial cells in control rats and diabetic rats maintained for 1 month. In diabetic rats maintained for 2–6 months, endothelial cells contained a large number of horseradish peroxidase-labeled vesicles which opened into the basement membrane, and intercellular junctions of the endothelial cells were filled with horseradish peroxidase. Horseradish peroxidase was consistently found in the stroma around the vessels, and seemed to reach the anterior chamber through the intercellular spaces between the fibroblasts. However, the passage of horseradish peroxidase was blocked by the posterior epithelial cells.